Merge remote branch 'origin/master' into arb_geometry_shader4arb_geometry_shader4

Conflicts:
	src/gallium/auxiliary/tgsi/tgsi_exec.c
	src/mesa/glapi/glapitemp.h
	src/mesa/glapi/glprocs.h
	src/mesa/main/enums.c
	src/mesa/main/remap_helper.h
	src/mesa/state_tracker/st_atom_shader.c
	src/mesa/state_tracker/st_program.c
	src/mesa/state_tracker/st_program.h

585 files changed, 39438 insertions, 8464 deletions

diff --git a/SConstruct b/SConstruct
index d53f4401e5..f43c10cecb 100644
--- a/SConstruct
+++ b/SConstruct
@@ -32,10 +32,10 @@ import common
 default_statetrackers = 'mesa'
 
 if common.default_platform in ('linux', 'freebsd', 'darwin'):
-	default_drivers = 'softpipe,failover,i915,trace,identity,llvmpipe'
+	default_drivers = 'softpipe,failover,svga,i915,trace,identity,llvmpipe'
 	default_winsys = 'xlib'
 elif common.default_platform in ('winddk',):
-	default_drivers = 'softpipe,i915,trace,identity'
+	default_drivers = 'softpipe,svga,i915,trace,identity'
 	default_winsys = 'all'
 else:
 	default_drivers = 'all'
@@ -46,9 +46,9 @@ common.AddOptions(opts)
 opts.Add(ListVariable('statetrackers', 'state trackers to build', default_statetrackers,
                      ['mesa', 'python', 'xorg']))
 opts.Add(ListVariable('drivers', 'pipe drivers to build', default_drivers,
-                     ['softpipe', 'failover', 'i915', 'cell', 'trace', 'r300', 'identity', 'llvmpipe']))
+                     ['softpipe', 'failover', 'svga', 'i915', 'cell', 'trace', 'r300', 'identity', 'llvmpipe']))
 opts.Add(ListVariable('winsys', 'winsys drivers to build', default_winsys,
-                     ['xlib', 'intel', 'gdi', 'radeon']))
+                     ['xlib', 'vmware', 'intel', 'gdi', 'radeon']))
 
 opts.Add(EnumVariable('MSVS_VERSION', 'MS Visual C++ version', None, allowed_values=('7.1', '8.0', '9.0')))
 
diff --git a/configs/default b/configs/default
index c3bb47e70d..15f9285dbf 100644
--- a/configs/default
+++ b/configs/default
@@ -9,7 +9,7 @@ CONFIG_NAME = default
 
 # Version info
 MESA_MAJOR=7
-MESA_MINOR=7
+MESA_MINOR=8
 MESA_TINY=0
 MESA_VERSION = $(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY)
 
@@ -96,7 +96,7 @@ EGL_DRIVERS_DIRS = demo
 GALLIUM_DIRS = auxiliary drivers state_trackers
 GALLIUM_AUXILIARY_DIRS = rbug draw translate cso_cache pipebuffer tgsi sct rtasm util indices vl
 GALLIUM_AUXILIARIES = $(foreach DIR,$(GALLIUM_AUXILIARY_DIRS),$(TOP)/src/gallium/auxiliary/$(DIR)/lib$(DIR).a)
-GALLIUM_DRIVERS_DIRS = softpipe i915 failover trace identity
+GALLIUM_DRIVERS_DIRS = softpipe failover svga i915 trace identity
 GALLIUM_DRIVERS = $(foreach DIR,$(GALLIUM_DRIVERS_DIRS),$(TOP)/src/gallium/drivers/$(DIR)/lib$(DIR).a)
 GALLIUM_WINSYS_DIRS = xlib egl_xlib
 GALLIUM_WINSYS_DRM_DIRS =
diff --git a/configs/linux-debug b/configs/linux-debug
index bc6d2336fc..8b0b3126db 100644
--- a/configs/linux-debug
+++ b/configs/linux-debug
@@ -4,5 +4,6 @@ include $(TOP)/configs/linux
 
 CONFIG_NAME = linux-debug
 
-OPT_FLAGS = -g -ansi -pedantic
+OPT_FLAGS = -g
+CFLAGS += -ansi -pedantic
 DEFINES += -DDEBUG -DDEBUG_MATH
diff --git a/configs/linux-dri b/configs/linux-dri
index 6c3c0ab921..0802543347 100644
--- a/configs/linux-dri
+++ b/configs/linux-dri
@@ -60,7 +60,7 @@ EGL_DRIVERS_DIRS = demo glx
 DRIVER_DIRS = dri
 WINDOW_SYSTEM = dri
 GALLIUM_WINSYS_DIRS = drm
-GALLIUM_WINSYS_DRM_DIRS = intel
+GALLIUM_WINSYS_DRM_DIRS = vmware intel
 GALLIUM_STATE_TRACKERS_DIRS = egl
 
 DRI_DIRS = i810 i915 i965 mach64 mga r128 r200 r300 radeon \
diff --git a/configure.ac b/configure.ac
index cc588d5fab..f9476a46dd 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1190,6 +1190,19 @@ AC_ARG_WITH([max-height],
 )
 
 dnl
+dnl Gallium SVGA configuration
+dnl
+AC_ARG_ENABLE([gallium-svga],
+    [AS_HELP_STRING([--disable-gallium-svga],
+        [build gallium SVGA @<:@default=enabled@:>@])],
+    [enable_gallium_svga="$enableval"],
+    [enable_gallium_svga=yes])
+if test "x$enable_gallium_svga" = xyes; then
+    GALLIUM_WINSYS_DRM_DIRS="$GALLIUM_WINSYS_DRM_DIRS vmware"
+    GALLIUM_DRIVERS_DIRS="$GALLIUM_DRIVERS_DIRS svga"
+fi
+
+dnl
 dnl Gallium Intel configuration
 dnl
 AC_ARG_ENABLE([gallium-intel],
diff --git a/docs/libGL.txt b/docs/libGL.txt
new file mode 100644
index 0000000000..cb98840437
--- /dev/null
+++ b/docs/libGL.txt
@@ -0,0 +1,197 @@
+
+
+
+Introduction
+------------
+
+This document describes the implementation of the XFree86 4.0 libGL.so
+library defined by the Linux/OpenGL Base specification found at
+http://reality.sgi.com/opengl/linux/linuxbase.html.
+
+The documentation is divided into two sections:
+    User's Guide
+    Driver Developer's Guide
+
+Author:  Brian Paul     (brian@precisioninsight.com)
+Date:    February 2000
+
+
+
+User's Guide
+------------
+
+Using libGL.so
+
+The libGL.so library defines the gl- and glX-prefixed functions needed to
+run OpenGL programs.  OpenGL client applications should link with the
+-lGL option to use it.
+
+libGL.so serves two primary functions: GLX protocol generation for indirect
+rendering and loading/management of hardware drivers for direct rendering.
+
+When libGL.so initializes itself it uses the DRI to determine the
+appropriate hardware driver for each screen on the local X display.
+The hardware drivers are expected to be in the /usr/X11R6/lib/modules/dri/
+directory.  Drivers are named with the convention <name>_dri.so where
+<name> is a driver such as "tdfx", "i810", "gamma", etc.
+
+The LIBGL_DRIVERS_DIR environment variable may be used to specify a
+different DRI modules directory, overriding /usr/X11R6/lib/modules/dri/.
+This environment variable is ignored in setuid programs for security
+reasons.
+
+When libGL.so is unable to locate appropriate hardware drivers it will
+fall back to using indirect GLX rendering.
+
+To aid in solving problems, libGL.so will print diagnostic messages to
+stderr if the LIBGL_DEBUG environment variable is defined.
+
+libGL.so is thread safe.  The overhead of thread safety for common,
+single-thread clients is negligible.  However, the overhead of thread
+safety for multi-threaded clients is significant.  Each GL API call
+requires two calls to pthread_get_specific() which can noticably
+impact performance.  Warning:  libGL.so is thread safe but individual
+DRI drivers may not be.  Please consult the documentation for a driver
+to learn if it is thread safe.
+
+
+
+Indirect Rendering
+
+You can force indirect rendering mode by setting the LIBGL_ALWAYS_INDIRECT
+environment variable.  Hardware acceleration will not be used.
+
+
+
+libGL.so Extensibility
+
+libGL.so is designed to be extended without upgrading.  That is,
+drivers may install new OpenGL extension functions into libGL.so
+without requiring libGL.so to be replaced.  Clients of libGL.so should
+use the glXGetProcAddressEXT() function to obtain the address of
+functions by name.  For more details of GLX_ARB_get_proc_address see
+http://oss.sgi.com/projects/ogl-sample/registry/ARB/get_proc_address.spec
+
+libGL.so is also designed with flexibility such that it may be used
+with many generations of hardware drivers to come.
+
+
+
+
+Driver Developer's Guide
+------------------------
+
+This section describes the requirements to make an XFree86 4.0
+libGL.so-compatible hardware driver.  It is not intended for end
+users of libGL.so.
+
+
+XFree86 source files
+
+libGL.so is built inside XFree86 with sources found in xc/lib/GL/.
+Specifically, libGL.so is built from:
+
+	xc/lib/GL/glx/*.c
+	xc/lib/dri/XF86dri.c
+	xc/lib/dri/dri_glx.c
+	xc/lib/GL/mesa/src/glapi.c
+	xc/lib/GL/mesa/src/glapitemp.h
+	xc/lib/GL/mesa/src/glapitable.h
+	xc/lib/GL/mesa/src/glapioffsets.h
+	xc/lib/GL/mesa/src/glapinoop.c
+	xc/lib/GL/mesa/src/glheader.h
+	xc/lib/GL/mesa/src/glthread.c
+	xc/lib/GL/mesa/src/glthread.h
+	xc/lib/GL/mesa/src/X86/glapi_x86.S
+	xc/lib/GL/mesa/src/X86/assyntax.h
+
+Understand that the mesa/src/gl*.[ch] files are not tied to Mesa.  They
+have no dependencies on the rest of Mesa and are designed to be reusable
+in a number of projects.
+
+The glapi_x86.X and assyntax.h files implement x86-optimized dispatch
+of GL functions.  They are not required; C-based dispatch can be used
+instead, with a slight performance penalty.
+
+
+
+Driver loading and binding
+
+When libGL.so initializes itself (via the __glXInitialize function) a
+call is made to driCreateDisplay().  This function uses DRI facilities
+to determine the driver file appropriate for each screen on the local
+display.  Each screen's driver is then opened with dlopen() and asked
+for its __driCreateScreen() function.  The pointers to the __driCreateScreen()
+functions are kept in an array, indexed by screen number, in the
+__DRIdisplayRec struct.
+
+When a driver's __driCreateScreen() function is called, it must initialize
+a __DRIscreenRec struct.  This struct acts as the root of a tree of
+function pointers which are called to create and destroy contexts and
+drawables and perform all the operations needed by the GLX interface.
+See the xc/lib/GL/glx/glxclient.h file for details.
+
+
+
+Dynamic Extension Function Registration
+
+In order to provide forward compatibility with future drivers, libGL.so
+allows drivers to register new OpenGL extension functions which weren't
+known when libGL.so was built.
+
+The register_extensions() function in xc/lib/GL/dri/dri_glx.c is called
+as soon as libGL.so is loaded.  This is done with gcc's constructor
+attribute.  This mechanism will likely have to be changed for other compilers.
+
+register_extensions() loops over all local displays and screens, determines
+the DRI driver for each, and calls the driver's __driRegisterExtensions()
+function, if present.
+
+The __driRegisterExtensions() function can add new entrypoints to libGL
+by calling:
+
+    GLboolean _glapi_add_entrypoint(const char *funcName, GLuint offset)
+
+The parameters are the name of the function (such as "glFoobarEXT") and the
+offset of the dispatch slot in the API dispatch table.  The return value
+indicates success (GL_TRUE) or failure (GL_FALSE).
+
+_glapi_add_entrypoint() will synthesize entrypoint code in assembly
+language.  Assembly languages is required since parameter passing
+can't be handled correctly using a C-based solution.
+
+The address of the new entrypoint is obtained by calling the
+glXGetProcAddressARB() function.
+
+The dispatch offset number MUST be a number allocated by SGI in the same
+manner in which new GL_* constants are allocated.  Using an arbitrary
+offset number will result in many problems.
+
+
+
+Dispatch Management
+
+When a GL context is made current, the driver must install its dispatch
+table as the current dispatch table.  This is done by calling
+
+	void _glapi_set_dispatch(struct _glapi_table *dispatch);
+
+This will install the named dispatch table for the calling thread.
+The current dispatch table for a thread can be obtained by calling
+
+	struct _glapi_table *_glapi_get_dispatch(void);
+
+For higher performance in the common single-thread case, the global
+variable _glapi_Dispatch will point to the current dispatch table.
+This variable will be NULL when in multi-thread mode.
+
+
+
+Context Management
+
+libGL.so uses the XFree86 xthreads package to manage a thread-specific
+current context pointer.  See __glXGet/SetCurrentContext() in glext.c
+
+Drivers may use the _glapi_set/get_context() functions to maintain
+a private thread-specific context pointer.
+
diff --git a/docs/relnotes-7.7.html b/docs/relnotes-7.7.html
index 7ab278bd08..8c8f763b6f 100644
--- a/docs/relnotes-7.7.html
+++ b/docs/relnotes-7.7.html
@@ -34,6 +34,9 @@ tbd
 
 <h2>New features</h2>
 <ul>
+<li>VMware "SVGA" Gallium driver.  This is a Gallium3D driver which targets the
+    VMware virtual graphics device.  It allows Linux OpenGL guest applications
+    to utilize the 3D graphics hardware of the host operating system.
 <li>GL_ARB_draw_elements_base_vertex (supported in Intel i965 and software drivers)</li>
 <li>GL_ARB_depth_clamp (supported in Intel i965 DRI and software drivers)</li>
 <li>GL_NV_depth_clamp (supported in Intel i965 DRI and software drivers)</li>
diff --git a/docs/relnotes-7.8.html b/docs/relnotes-7.8.html
new file mode 100644
index 0000000000..dc003566c7
--- /dev/null
+++ b/docs/relnotes-7.8.html
@@ -0,0 +1,53 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.8 Release Notes / date TBD</H1>
+
+<p>
+Mesa 7.8 is a new development release.
+People who are concerned with stability and reliability should stick
+with a previous release or wait for Mesa 7.8.1.
+</p>
+<p>
+Mesa 7.8 implements the OpenGL 2.1 API, but the version reported by
+glGetString(GL_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 2.1.
+</p>
+<p>
+See the <a href="install.html">Compiling/Installing page</a> for prerequisites
+for DRI hardware acceleration.
+</p>
+
+
+<h2>MD5 checksums</h2>
+<pre>
+tbd
+</pre>
+
+
+<h2>New features</h2>
+<ul>
+<li>TBD
+</ul>
+
+
+<h2>Bug fixes</h2>
+<ul>
+<li>TBD
+</ul>
+
+
+<h2>Changes</h2>
+<ul>
+<li>TBD
+</ul>
+
+</body>
+</html>
diff --git a/docs/relnotes.html b/docs/relnotes.html
index b301ee046a..d0d9b6e5b9 100644
--- a/docs/relnotes.html
+++ b/docs/relnotes.html
@@ -13,6 +13,7 @@ The release notes summarize what's new or changed in each Mesa release.
 </p>
 
 <UL>
+<LI><A HREF="relnotes-7.8.html">7.8 release notes</A>
 <LI><A HREF="relnotes-7.7.html">7.7 release notes</A>
 <LI><A HREF="relnotes-7.6.1.html">7.6.1 release notes</A>
 <LI><A HREF="relnotes-7.6.html">7.6 release notes</A>
diff --git a/progs/demos/ipers.c b/progs/demos/ipers.c
index 5d82b0dc92..ed03673cb6 100644
--- a/progs/demos/ipers.c
+++ b/progs/demos/ipers.c
@@ -133,6 +133,8 @@ initdlists(void)
 
       glEndList();
    }
+
+   gluDeleteQuadric(obj);
 }
 
 static void
diff --git a/progs/demos/morph3d.c b/progs/demos/morph3d.c
index 6aca8270ff..01a06aba0f 100644
--- a/progs/demos/morph3d.c
+++ b/progs/demos/morph3d.c
@@ -887,5 +887,6 @@ int main(int argc, char **argv)
   glutIdleFunc( idle_ );
   glutDisplayFunc( draw );
   glutMainLoop();
-  
+
+  return 0;
 }
diff --git a/progs/demos/projtex.c b/progs/demos/projtex.c
index 99154d7bdc..ad205c7413 100644
--- a/progs/demos/projtex.c
+++ b/progs/demos/projtex.c
@@ -245,6 +245,9 @@ loadImageTextures(void)
        glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
      }
      glTexParameterfv(GL_TEXTURE_2D, GL_TEXTURE_BORDER_COLOR, borderColor);
+
+     free(texData3);
+     free(texData4);
   }
 }
 
diff --git a/progs/demos/ray.c b/progs/demos/ray.c
index c2d8e4f545..e9211aa339 100644
--- a/progs/demos/ray.c
+++ b/progs/demos/ray.c
@@ -834,6 +834,8 @@ initdlists(void)
    gluQuadricTexture(obj, GL_TRUE);
    gluSphere(obj, SPHERE_RADIUS, 16, 16);
    glEndList();
+
+   gluDeleteQuadric(obj);
 }
 
 int
diff --git a/progs/demos/textures.c b/progs/demos/textures.c
index b7bf135d21..d94154a200 100644
--- a/progs/demos/textures.c
+++ b/progs/demos/textures.c
@@ -57,7 +57,7 @@ Idle(void)
 {
    Xrot = glutGet(GLUT_ELAPSED_TIME) * 0.02;
    Yrot = glutGet(GLUT_ELAPSED_TIME) * 0.04;
-   //Zrot += 2.0;
+   /*Zrot += 2.0;*/
    glutPostRedisplay();
 }
 
diff --git a/progs/fp/fp-tri.c b/progs/fp/fp-tri.c
index 52a8fcfc22..26af66ad84 100644
--- a/progs/fp/fp-tri.c
+++ b/progs/fp/fp-tri.c
@@ -197,7 +197,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/point-position.c b/progs/fp/point-position.c
index c0963d7a0b..1ae753c1d0 100644
--- a/progs/fp/point-position.c
+++ b/progs/fp/point-position.c
@@ -55,7 +55,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-depth.c b/progs/fp/tri-depth.c
index 5488469e80..a9f3a6a5be 100644
--- a/progs/fp/tri-depth.c
+++ b/progs/fp/tri-depth.c
@@ -57,7 +57,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-depth2.c b/progs/fp/tri-depth2.c
index 6ed2307115..8c4336817b 100644
--- a/progs/fp/tri-depth2.c
+++ b/progs/fp/tri-depth2.c
@@ -59,7 +59,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-depthwrite.c b/progs/fp/tri-depthwrite.c
index 8e4f3e6245..7b9d70f292 100644
--- a/progs/fp/tri-depthwrite.c
+++ b/progs/fp/tri-depthwrite.c
@@ -55,7 +55,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-depthwrite2.c b/progs/fp/tri-depthwrite2.c
index 3c0b4d30c9..599949551d 100644
--- a/progs/fp/tri-depthwrite2.c
+++ b/progs/fp/tri-depthwrite2.c
@@ -55,7 +55,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-param.c b/progs/fp/tri-param.c
index 57443d71bd..26a804d4b3 100644
--- a/progs/fp/tri-param.c
+++ b/progs/fp/tri-param.c
@@ -54,7 +54,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/fp/tri-tex.c b/progs/fp/tri-tex.c
index 1dbbb201ce..64299e9453 100644
--- a/progs/fp/tri-tex.c
+++ b/progs/fp/tri-tex.c
@@ -76,7 +76,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/glsl/CH18-mandel.frag b/progs/glsl/CH18-mandel.frag
index a472d81252..a972d68bcf 100644
--- a/progs/glsl/CH18-mandel.frag
+++ b/progs/glsl/CH18-mandel.frag
@@ -31,7 +31,7 @@ void main()
     float iter;
 
 //    for (iter = 0.0; iter < MaxIterations && r2 < 4.0; ++iter)
-    for (iter = 0.0; iter < 12 && r2 < 4.0; ++iter)
+    for (iter = 0.0; iter < 12.0 && r2 < 4.0; ++iter)
     {
         float tempreal = real;
 
diff --git a/progs/glsl/bump.c b/progs/glsl/bump.c
index 87669aec73..50a0900f1c 100644
--- a/progs/glsl/bump.c
+++ b/progs/glsl/bump.c
@@ -36,7 +36,7 @@ static GLint win = 0;
 
 static GLfloat xRot = 20.0f, yRot = 0.0f, zRot = 0.0f;
 
-static GLuint tangentAttrib;
+static GLint tangentAttrib;
 
 static GLboolean Anim = GL_FALSE;
 
diff --git a/progs/glsl/convolutions.c b/progs/glsl/convolutions.c
index ee330cf064..d3d2477169 100644
--- a/progs/glsl/convolutions.c
+++ b/progs/glsl/convolutions.c
@@ -382,7 +382,7 @@ static void keyPress(unsigned char key, int x, int y)
    case 27:
       exit(0);
    default:
-      return;
+      break;
    }
    glutPostRedisplay();
 }
diff --git a/progs/glsl/deriv.c b/progs/glsl/deriv.c
index 265a515715..30f2b75fef 100644
--- a/progs/glsl/deriv.c
+++ b/progs/glsl/deriv.c
@@ -140,6 +140,7 @@ MakeSphere(void)
    glNewList(SphereList, GL_COMPILE);
    gluSphere(obj, 2.0f, 30, 15);
    glEndList();
+   gluDeleteQuadric(obj);
 }
 
 
diff --git a/progs/glsl/multinoise.c b/progs/glsl/multinoise.c
index 06207f78b5..d504ba1cc4 100644
--- a/progs/glsl/multinoise.c
+++ b/progs/glsl/multinoise.c
@@ -22,22 +22,22 @@ static const char *FragShaderText[ 4 ] = {
    "void main()\n"
    "{\n"
    "   gl_FragColor.rgb = noise3( gl_TexCoord[ 0 ].w ) * 0.5 + 0.5;\n"
-   "   gl_FragColor.a = 1;\n"
+   "   gl_FragColor.a = 1.0;\n"
    "}\n",
    "void main()\n"
    "{\n"
    "   gl_FragColor.rgb = noise3( gl_TexCoord[ 0 ].xw ) * 0.5 + 0.5;\n"
-   "   gl_FragColor.a = 1;\n"
+   "   gl_FragColor.a = 1.0;\n"
    "}\n",
    "void main()\n"
    "{\n"
    "   gl_FragColor.rgb = noise3( gl_TexCoord[ 0 ].xyw ) * 0.5 + 0.5;\n"
-   "   gl_FragColor.a = 1;\n"
+   "   gl_FragColor.a = 1.0;\n"
    "}\n",
    "void main()\n"
    "{\n"
    "   gl_FragColor.rgb = noise3( gl_TexCoord[ 0 ].xyzw ) * 0.5 + 0.5;\n"
-   "   gl_FragColor.a = 1;\n"
+   "   gl_FragColor.a = 1.0;\n"
    "}\n"
 };
     
@@ -194,7 +194,7 @@ LoadAndCompileShader(GLuint shader, const char *text)
       GLchar log[1000];
       GLsizei len;
       glGetShaderInfoLog(shader, 1000, &len, log);
-      fprintf(stderr, "noise: problem compiling shader: %s\n", log);
+      fprintf(stderr, "multinoise: problem compiling shader: %s\n", log);
       exit(1);
    }
    else {
diff --git a/progs/glsl/multitex.c b/progs/glsl/multitex.c
index bf46fd5210..49b32253ee 100644
--- a/progs/glsl/multitex.c
+++ b/progs/glsl/multitex.c
@@ -369,12 +369,9 @@ InitPrograms(void)
 static void
 InitGL(void)
 {
-   const char *version = (const char *) glGetString(GL_VERSION);
+   if (!ShadersSupported())
+      exit(1);
 
-   if (version[0] != '2' || version[1] != '.') {
-      printf("Warning: this program expects OpenGL 2.0\n");
-      /*exit(1);*/
-   }
    printf("GL_RENDERER = %s\n",(const char *) glGetString(GL_RENDERER));
    printf("Usage:\n");
    printf("  a     - toggle arrays vs. immediate mode rendering\n");
diff --git a/progs/glsl/noise.c b/progs/glsl/noise.c
index bb024b5012..1148580ff4 100644
--- a/progs/glsl/noise.c
+++ b/progs/glsl/noise.c
@@ -28,7 +28,7 @@ static const char *FragShaderText =
    "   vec4 p;\n"
    "   p.xy = gl_TexCoord[0].xy;\n"
    "   p.z = Slice;\n"
-   "   p.w = 0;\n"
+   "   p.w = 0.0;\n"
    "   vec4 n = noise4(p * scale);\n"
    "   gl_FragColor = n * Scale + Bias;\n"
    "}\n";
diff --git a/progs/glsl/shtest.c b/progs/glsl/shtest.c
index 628a7dd5b9..9d76e018e3 100644
--- a/progs/glsl/shtest.c
+++ b/progs/glsl/shtest.c
@@ -627,7 +627,7 @@ Init(void)
    NumAttribs = GetAttribs(Program, Attribs);
    PrintAttribs(Attribs);
 
-   //assert(glGetError() == 0);
+   /*assert(glGetError() == 0);*/
 
    glClearColor(0.4f, 0.4f, 0.8f, 0.0f);
 
diff --git a/progs/glsl/texaaline.c b/progs/glsl/texaaline.c
index 2e4f932ad6..4ad5a761c6 100644
--- a/progs/glsl/texaaline.c
+++ b/progs/glsl/texaaline.c
@@ -310,8 +310,10 @@ MakeMipmap(void)
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_LINEAR);
    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
-   //glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LOD, 4);
-   ////glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 5);
+   /*
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LOD, 4);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 5);
+   */
 
    glTexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE);
 #undef SZ
diff --git a/progs/glsl/texdemo1.c b/progs/glsl/texdemo1.c
index 5b1913a722..2076e6aef9 100644
--- a/progs/glsl/texdemo1.c
+++ b/progs/glsl/texdemo1.c
@@ -400,11 +400,8 @@ InitPrograms(void)
 static void
 Init(GLboolean useImageFiles)
 {
-   const char *version = (const char *) glGetString(GL_VERSION);
-
-   if (version[0] != '2' || version[1] != '.') {
-      printf("Warning: this program expects OpenGL 2.0\n");
-      /*exit(1);*/
+   if (!ShadersSupported()) {
+      exit(1);
    }
    printf("GL_RENDERER = %s\n",(const char *) glGetString(GL_RENDERER));
 
diff --git a/progs/glsl/trirast.c b/progs/glsl/trirast.c
index 53bd91ef97..857342636d 100644
--- a/progs/glsl/trirast.c
+++ b/progs/glsl/trirast.c
@@ -179,9 +179,9 @@ Init(void)
       "\n"
       "void main() {\n"
       "   vec2 p = gl_FragCoord.xy; \n"
-      "   if (crs(v1 - v0, p - v0) >= 0 && \n"
-      "       crs(v2 - v1, p - v1) >= 0 && \n"
-      "       crs(v0 - v2, p - v2) >= 0) \n"
+      "   if (crs(v1 - v0, p - v0) >= 0.0 && \n"
+      "       crs(v2 - v1, p - v1) >= 0.0 && \n"
+      "       crs(v0 - v2, p - v2) >= 0.0) \n"
       "      gl_FragColor = vec4(1.0); \n"
       "   else \n"
       "      gl_FragColor = vec4(0.5); \n"
diff --git a/progs/redbook/quadric.c b/progs/redbook/quadric.c
index 4e46c85f82..7e99098304 100644
--- a/progs/redbook/quadric.c
+++ b/progs/redbook/quadric.c
@@ -116,6 +116,8 @@ void init(void)
    glNewList(startList+3, GL_COMPILE);
       gluPartialDisk(qobj, 0.0, 1.0, 20, 4, 0.0, 225.0);
    glEndList();
+
+   gluDeleteQuadric(qobj);
 }
 
 void display(void)
diff --git a/progs/slang/cltest.c b/progs/slang/cltest.c
index 2753f59f85..dfff6cb226 100644
--- a/progs/slang/cltest.c
+++ b/progs/slang/cltest.c
@@ -237,7 +237,7 @@ void RenderScene (void)
       if (p != NULL)
          p++;
    }
-   if (*p != '\0')
+   if (p && *p != '\0')
       printf ("*** %s\n", "I/O error");
 
    nextprogram = program->next;
diff --git a/progs/slang/vstest.c b/progs/slang/vstest.c
index 472ea9b62e..33c6f09dc5 100644
--- a/progs/slang/vstest.c
+++ b/progs/slang/vstest.c
@@ -220,6 +220,12 @@ static void load_test_file (const char *filename)
 
    fseek (f, 0, SEEK_END);
    size = ftell (f);
+
+   if (size == -1) {
+      fclose (f);
+      return;
+   }
+
    fseek (f, 0, SEEK_SET);
 
    code = (char *) (malloc (size));
diff --git a/progs/tests/arbgpuprog.c b/progs/tests/arbgpuprog.c
index 23aa899d96..6098dca787 100644
--- a/progs/tests/arbgpuprog.c
+++ b/progs/tests/arbgpuprog.c
@@ -134,6 +134,8 @@ static void Init( const char *vertProgFile,
       }
 
       len = fread(buf, 1, 10*1000,f);
+      fclose(f);
+
       glProgramStringARB_func(GL_VERTEX_PROGRAM_ARB,
                               GL_PROGRAM_FORMAT_ASCII_ARB,
                               len,
@@ -170,6 +172,8 @@ static void Init( const char *vertProgFile,
       }
 
       len = fread(buf, 1, 10*1000,f);
+      fclose(f);
+
       glProgramStringARB_func(GL_FRAGMENT_PROGRAM_ARB,
                               GL_PROGRAM_FORMAT_ASCII_ARB,
                               len,
diff --git a/progs/tests/getprocaddress.c b/progs/tests/getprocaddress.c
index a09ea58e1d..b905eeaf81 100644
--- a/progs/tests/getprocaddress.c
+++ b/progs/tests/getprocaddress.c
@@ -3516,7 +3516,7 @@ check_functions( const char *extensions )
    struct name_test_pair *entry;
    int failures = 0, passes = 0, untested = 0;
    int totalFail = 0, totalPass = 0, totalUntested = 0, totalUnsupported = 0;
-   int doTests;
+   int doTests = 0;
    const char *version = (const char *) glGetString(GL_VERSION);
 
    /* The functions list will have "real" entries (consisting of
diff --git a/progs/tests/sharedtex.c b/progs/tests/sharedtex.c
index c07ebd719c..2337b88d3f 100644
--- a/progs/tests/sharedtex.c
+++ b/progs/tests/sharedtex.c
@@ -424,13 +424,13 @@ main(int argc, char *argv[])
 {
    const char *dpyName = XDisplayName(NULL);
 
-   struct window *h0, *h1, *h2, *h3;
+   struct window *h0;
 
    /* four windows and contexts sharing display lists and texture objects */
    h0 = AddWindow(dpyName,  10,  10, NULL);
-   h1 = AddWindow(dpyName, 330,  10, h0);
-   h2 = AddWindow(dpyName,  10, 350, h0);
-   h3 = AddWindow(dpyName, 330, 350, h0);
+   (void) AddWindow(dpyName, 330,  10, h0);
+   (void) AddWindow(dpyName,  10, 350, h0);
+   (void) AddWindow(dpyName, 330, 350, h0);
 
    InitGLstuff(h0);
 
diff --git a/progs/tests/texdown.c b/progs/tests/texdown.c
index 7e46045832..e6881d39a0 100644
--- a/progs/tests/texdown.c
+++ b/progs/tests/texdown.c
@@ -176,6 +176,8 @@ MeasureDownloadRate(void)
    orig_getImage = (GLubyte *) malloc(image_bytes + ALIGN);
    if (!orig_texImage || !orig_getImage) {
       DownloadRate = 0.0;
+      free(orig_texImage);
+      free(orig_getImage);
       return;
    }
 
diff --git a/progs/trivial/clear-fbo-tex.c b/progs/trivial/clear-fbo-tex.c
index a206676e48..238f634bf5 100644
--- a/progs/trivial/clear-fbo-tex.c
+++ b/progs/trivial/clear-fbo-tex.c
@@ -88,7 +88,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/clear-fbo.c b/progs/trivial/clear-fbo.c
index 0aeb45489f..6435c901ad 100644
--- a/progs/trivial/clear-fbo.c
+++ b/progs/trivial/clear-fbo.c
@@ -73,7 +73,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/clear-random.c b/progs/trivial/clear-random.c
index e3da23a8f5..ab67f84518 100644
--- a/progs/trivial/clear-random.c
+++ b/progs/trivial/clear-random.c
@@ -61,7 +61,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(0);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/clear-scissor.c b/progs/trivial/clear-scissor.c
index 0173532748..2b6dee77a9 100644
--- a/progs/trivial/clear-scissor.c
+++ b/progs/trivial/clear-scissor.c
@@ -38,7 +38,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
     glutPostRedisplay();
 }
diff --git a/progs/trivial/clear.c b/progs/trivial/clear.c
index 03857b4b89..56c3ea52e6 100644
--- a/progs/trivial/clear.c
+++ b/progs/trivial/clear.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(0);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/createwin.c b/progs/trivial/createwin.c
index f2cc6f1cff..04a088642b 100644
--- a/progs/trivial/createwin.c
+++ b/progs/trivial/createwin.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-begin-call-end.c b/progs/trivial/dlist-begin-call-end.c
index 0d0aed7c72..da3864a02a 100644
--- a/progs/trivial/dlist-begin-call-end.c
+++ b/progs/trivial/dlist-begin-call-end.c
@@ -87,7 +87,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-dangling.c b/progs/trivial/dlist-dangling.c
index de10628009..756ab93f87 100644
--- a/progs/trivial/dlist-dangling.c
+++ b/progs/trivial/dlist-dangling.c
@@ -74,7 +74,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-edgeflag-dangling.c b/progs/trivial/dlist-edgeflag-dangling.c
index 3d3aaeb694..51504471e2 100644
--- a/progs/trivial/dlist-edgeflag-dangling.c
+++ b/progs/trivial/dlist-edgeflag-dangling.c
@@ -76,7 +76,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-edgeflag.c b/progs/trivial/dlist-edgeflag.c
index 8002129ed1..9456b96473 100644
--- a/progs/trivial/dlist-edgeflag.c
+++ b/progs/trivial/dlist-edgeflag.c
@@ -81,7 +81,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-flat-tri.c b/progs/trivial/dlist-flat-tri.c
index c3dd7921e3..3ee9b283a4 100644
--- a/progs/trivial/dlist-flat-tri.c
+++ b/progs/trivial/dlist-flat-tri.c
@@ -93,7 +93,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-mat-tri.c b/progs/trivial/dlist-mat-tri.c
index ed3a4c5981..d17c4b913b 100644
--- a/progs/trivial/dlist-mat-tri.c
+++ b/progs/trivial/dlist-mat-tri.c
@@ -103,7 +103,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-recursive-call.c b/progs/trivial/dlist-recursive-call.c
index fe06b2bbd7..90224f63cb 100644
--- a/progs/trivial/dlist-recursive-call.c
+++ b/progs/trivial/dlist-recursive-call.c
@@ -118,7 +118,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-tri-flat-tri.c b/progs/trivial/dlist-tri-flat-tri.c
index 4dbb788486..3265a4d6dc 100644
--- a/progs/trivial/dlist-tri-flat-tri.c
+++ b/progs/trivial/dlist-tri-flat-tri.c
@@ -99,7 +99,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/dlist-tri-mat-tri.c b/progs/trivial/dlist-tri-mat-tri.c
index f69854ae58..053bb124a2 100644
--- a/progs/trivial/dlist-tri-mat-tri.c
+++ b/progs/trivial/dlist-tri-mat-tri.c
@@ -102,7 +102,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-clip.c b/progs/trivial/line-clip.c
index 5276baffd5..e4e388ed5c 100644
--- a/progs/trivial/line-clip.c
+++ b/progs/trivial/line-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-cull.c b/progs/trivial/line-cull.c
index 1e1b77a942..c531ff132f 100644
--- a/progs/trivial/line-cull.c
+++ b/progs/trivial/line-cull.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-flat.c b/progs/trivial/line-flat.c
index 14f0ac0782..e2ddb87b9e 100644
--- a/progs/trivial/line-flat.c
+++ b/progs/trivial/line-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-stipple-wide.c b/progs/trivial/line-stipple-wide.c
index 1804ffad3f..767583bbea 100644
--- a/progs/trivial/line-stipple-wide.c
+++ b/progs/trivial/line-stipple-wide.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-userclip-clip.c b/progs/trivial/line-userclip-clip.c
index 8e030b47ce..3265b2c3bf 100644
--- a/progs/trivial/line-userclip-clip.c
+++ b/progs/trivial/line-userclip-clip.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-userclip-nop-clip.c b/progs/trivial/line-userclip-nop-clip.c
index 6fcd4bcfe7..0198e27807 100644
--- a/progs/trivial/line-userclip-nop-clip.c
+++ b/progs/trivial/line-userclip-nop-clip.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-userclip-nop.c b/progs/trivial/line-userclip-nop.c
index e59fd133a5..6c863a3c82 100644
--- a/progs/trivial/line-userclip-nop.c
+++ b/progs/trivial/line-userclip-nop.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-userclip.c b/progs/trivial/line-userclip.c
index e30be5580b..6cfcb6fc73 100644
--- a/progs/trivial/line-userclip.c
+++ b/progs/trivial/line-userclip.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line-wide.c b/progs/trivial/line-wide.c
index b74021dea7..1945712c5d 100644
--- a/progs/trivial/line-wide.c
+++ b/progs/trivial/line-wide.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/line.c b/progs/trivial/line.c
index e1d73280bf..58ab96836f 100644
--- a/progs/trivial/line.c
+++ b/progs/trivial/line.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/lineloop-clip.c b/progs/trivial/lineloop-clip.c
index 45fa47491f..5e6b6217a1 100644
--- a/progs/trivial/lineloop-clip.c
+++ b/progs/trivial/lineloop-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/lineloop.c b/progs/trivial/lineloop.c
index c290dbd8cb..82eccd24f8 100644
--- a/progs/trivial/lineloop.c
+++ b/progs/trivial/lineloop.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/linestrip-clip.c b/progs/trivial/linestrip-clip.c
index f252822921..5e90ea2eb0 100644
--- a/progs/trivial/linestrip-clip.c
+++ b/progs/trivial/linestrip-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/linestrip-flat-stipple.c b/progs/trivial/linestrip-flat-stipple.c
index 5caa724423..bae5c3a1c3 100644
--- a/progs/trivial/linestrip-flat-stipple.c
+++ b/progs/trivial/linestrip-flat-stipple.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/linestrip-stipple-wide.c b/progs/trivial/linestrip-stipple-wide.c
index 701c82c266..0296941a0d 100644
--- a/progs/trivial/linestrip-stipple-wide.c
+++ b/progs/trivial/linestrip-stipple-wide.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/linestrip-stipple.c b/progs/trivial/linestrip-stipple.c
index df2eef96b5..a847d47dee 100644
--- a/progs/trivial/linestrip-stipple.c
+++ b/progs/trivial/linestrip-stipple.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/linestrip.c b/progs/trivial/linestrip.c
index 0219b1ab70..b6d6882146 100644
--- a/progs/trivial/linestrip.c
+++ b/progs/trivial/linestrip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/point-clip.c b/progs/trivial/point-clip.c
index 4c89ba598d..23cfd77863 100644
--- a/progs/trivial/point-clip.c
+++ b/progs/trivial/point-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/point-param.c b/progs/trivial/point-param.c
index 6f43720a89..46bd773265 100644
--- a/progs/trivial/point-param.c
+++ b/progs/trivial/point-param.c
@@ -59,7 +59,7 @@ static void Key(unsigned char key, int x, int y)
     case 27:
 	exit(1);
     default:
-	return;
+	break;
     }
     glutPostRedisplay();
 }
diff --git a/progs/trivial/point-sprite.c b/progs/trivial/point-sprite.c
index 5d29a6a3cf..16e6771514 100644
--- a/progs/trivial/point-sprite.c
+++ b/progs/trivial/point-sprite.c
@@ -96,7 +96,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/point-wide-smooth.c b/progs/trivial/point-wide-smooth.c
index f6e9b8df5f..752cb8aee3 100644
--- a/progs/trivial/point-wide-smooth.c
+++ b/progs/trivial/point-wide-smooth.c
@@ -63,7 +63,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/point-wide.c b/progs/trivial/point-wide.c
index 8abd64c6a9..3b00ee9e50 100644
--- a/progs/trivial/point-wide.c
+++ b/progs/trivial/point-wide.c
@@ -63,7 +63,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/point.c b/progs/trivial/point.c
index 49959dcc48..8f5299cf6b 100644
--- a/progs/trivial/point.c
+++ b/progs/trivial/point.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/poly-flat-clip.c b/progs/trivial/poly-flat-clip.c
index 5490068b08..2a968bed40 100644
--- a/progs/trivial/poly-flat-clip.c
+++ b/progs/trivial/poly-flat-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/poly-flat-unfilled-clip.c b/progs/trivial/poly-flat-unfilled-clip.c
index 26b90ef964..89f4e0656a 100644
--- a/progs/trivial/poly-flat-unfilled-clip.c
+++ b/progs/trivial/poly-flat-unfilled-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/poly-flat.c b/progs/trivial/poly-flat.c
index a4e3cdb633..33c2e04e0f 100644
--- a/progs/trivial/poly-flat.c
+++ b/progs/trivial/poly-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/poly-unfilled.c b/progs/trivial/poly-unfilled.c
index 2ad443dc15..c8c0d7a9e3 100644
--- a/progs/trivial/poly-unfilled.c
+++ b/progs/trivial/poly-unfilled.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/poly.c b/progs/trivial/poly.c
index e5b788ea5b..964eb42d78 100644
--- a/progs/trivial/poly.c
+++ b/progs/trivial/poly.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-clip-all-vertices.c b/progs/trivial/quad-clip-all-vertices.c
index 60c87fc9ce..7712d8ca5b 100644
--- a/progs/trivial/quad-clip-all-vertices.c
+++ b/progs/trivial/quad-clip-all-vertices.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-clip.c b/progs/trivial/quad-clip.c
index 063de6106a..2847c9f960 100644
--- a/progs/trivial/quad-clip.c
+++ b/progs/trivial/quad-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-degenerate.c b/progs/trivial/quad-degenerate.c
index fdc142bcd6..85e129d4be 100644
--- a/progs/trivial/quad-degenerate.c
+++ b/progs/trivial/quad-degenerate.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-flat.c b/progs/trivial/quad-flat.c
index e3147b3b3f..bed98e31a4 100644
--- a/progs/trivial/quad-flat.c
+++ b/progs/trivial/quad-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-offset-factor.c b/progs/trivial/quad-offset-factor.c
index dfe99bbae6..1fc57a54ae 100644
--- a/progs/trivial/quad-offset-factor.c
+++ b/progs/trivial/quad-offset-factor.c
@@ -55,7 +55,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-offset-unfilled.c b/progs/trivial/quad-offset-unfilled.c
index 06590021fe..62f50fb98c 100644
--- a/progs/trivial/quad-offset-unfilled.c
+++ b/progs/trivial/quad-offset-unfilled.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-offset-units.c b/progs/trivial/quad-offset-units.c
index 922529d977..1ac2338692 100644
--- a/progs/trivial/quad-offset-units.c
+++ b/progs/trivial/quad-offset-units.c
@@ -55,7 +55,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-tex-alpha.c b/progs/trivial/quad-tex-alpha.c
index eebaf9170e..9db6792fad 100644
--- a/progs/trivial/quad-tex-alpha.c
+++ b/progs/trivial/quad-tex-alpha.c
@@ -88,7 +88,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-tex-pbo.c b/progs/trivial/quad-tex-pbo.c
index ad41a9a22e..b7aa1db436 100644
--- a/progs/trivial/quad-tex-pbo.c
+++ b/progs/trivial/quad-tex-pbo.c
@@ -105,7 +105,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-unfilled-clip.c b/progs/trivial/quad-unfilled-clip.c
index 761878bd4b..d2e8718775 100644
--- a/progs/trivial/quad-unfilled-clip.c
+++ b/progs/trivial/quad-unfilled-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-unfilled-stipple.c b/progs/trivial/quad-unfilled-stipple.c
index cd7d276928..8c1737e675 100644
--- a/progs/trivial/quad-unfilled-stipple.c
+++ b/progs/trivial/quad-unfilled-stipple.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad-unfilled.c b/progs/trivial/quad-unfilled.c
index d64f17fdf9..b756449d7a 100644
--- a/progs/trivial/quad-unfilled.c
+++ b/progs/trivial/quad-unfilled.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quad.c b/progs/trivial/quad.c
index d360e309d3..582de783d5 100644
--- a/progs/trivial/quad.c
+++ b/progs/trivial/quad.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quads.c b/progs/trivial/quads.c
index fe11fef207..de7854a255 100644
--- a/progs/trivial/quads.c
+++ b/progs/trivial/quads.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quadstrip-cont.c b/progs/trivial/quadstrip-cont.c
index 329523531a..44412698dc 100644
--- a/progs/trivial/quadstrip-cont.c
+++ b/progs/trivial/quadstrip-cont.c
@@ -57,7 +57,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quadstrip-flat.c b/progs/trivial/quadstrip-flat.c
index 228c6c255e..0e0b3b49fa 100644
--- a/progs/trivial/quadstrip-flat.c
+++ b/progs/trivial/quadstrip-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/quadstrip.c b/progs/trivial/quadstrip.c
index d49a9a5302..1be815fb8c 100644
--- a/progs/trivial/quadstrip.c
+++ b/progs/trivial/quadstrip.c
@@ -57,7 +57,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/readpixels.c b/progs/trivial/readpixels.c
index 5671618446..b063f17956 100644
--- a/progs/trivial/readpixels.c
+++ b/progs/trivial/readpixels.c
@@ -39,7 +39,7 @@ static void Key(unsigned char key, int x, int y)
    case 27:
       exit(0);
    default:
-      return;
+      break;
    }
    glutPostRedisplay();
 }
diff --git a/progs/trivial/tri-alpha-tex.c b/progs/trivial/tri-alpha-tex.c
index 780ebe6be5..853d564ae7 100644
--- a/progs/trivial/tri-alpha-tex.c
+++ b/progs/trivial/tri-alpha-tex.c
@@ -88,7 +88,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-alpha.c b/progs/trivial/tri-alpha.c
index aec1cbb377..bd1c88ecf8 100644
--- a/progs/trivial/tri-alpha.c
+++ b/progs/trivial/tri-alpha.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-blend-color.c b/progs/trivial/tri-blend-color.c
index 92f019259e..629b35c4ab 100644
--- a/progs/trivial/tri-blend-color.c
+++ b/progs/trivial/tri-blend-color.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-clear.c b/progs/trivial/tri-clear.c
index f49186bd8a..e52ed81bf4 100644
--- a/progs/trivial/tri-clear.c
+++ b/progs/trivial/tri-clear.c
@@ -63,7 +63,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-clip.c b/progs/trivial/tri-clip.c
index e1deca1bdc..3e3879c985 100644
--- a/progs/trivial/tri-clip.c
+++ b/progs/trivial/tri-clip.c
@@ -56,7 +56,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
     glutPostRedisplay();
 }
diff --git a/progs/trivial/tri-cull-both.c b/progs/trivial/tri-cull-both.c
index 864be710c2..809599d4a6 100644
--- a/progs/trivial/tri-cull-both.c
+++ b/progs/trivial/tri-cull-both.c
@@ -65,7 +65,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-dlist.c b/progs/trivial/tri-dlist.c
index c410be221a..62d0a965d8 100644
--- a/progs/trivial/tri-dlist.c
+++ b/progs/trivial/tri-dlist.c
@@ -75,7 +75,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-fbo.c b/progs/trivial/tri-fbo.c
index 1ed177ffdf..089180f97a 100644
--- a/progs/trivial/tri-fbo.c
+++ b/progs/trivial/tri-fbo.c
@@ -75,7 +75,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-flat-clip.c b/progs/trivial/tri-flat-clip.c
index 2aab5ba00a..a23f8a382b 100644
--- a/progs/trivial/tri-flat-clip.c
+++ b/progs/trivial/tri-flat-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-flat.c b/progs/trivial/tri-flat.c
index ea703ec6f3..0614321844 100644
--- a/progs/trivial/tri-flat.c
+++ b/progs/trivial/tri-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-fog.c b/progs/trivial/tri-fog.c
index 0cea3d3258..7fc254a467 100644
--- a/progs/trivial/tri-fog.c
+++ b/progs/trivial/tri-fog.c
@@ -72,7 +72,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-fp-const-imm.c b/progs/trivial/tri-fp-const-imm.c
index d2df442abf..accea62042 100644
--- a/progs/trivial/tri-fp-const-imm.c
+++ b/progs/trivial/tri-fp-const-imm.c
@@ -92,7 +92,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-fp.c b/progs/trivial/tri-fp.c
index 4d1508120e..8f933befd3 100644
--- a/progs/trivial/tri-fp.c
+++ b/progs/trivial/tri-fp.c
@@ -90,7 +90,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-lit-material.c b/progs/trivial/tri-lit-material.c
index ff9fb2c4dd..16cebb2b98 100644
--- a/progs/trivial/tri-lit-material.c
+++ b/progs/trivial/tri-lit-material.c
@@ -65,7 +65,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-lit.c b/progs/trivial/tri-lit.c
index 15a7ad24c5..26ebd99d91 100644
--- a/progs/trivial/tri-lit.c
+++ b/progs/trivial/tri-lit.c
@@ -65,7 +65,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-multitex-vbo.c b/progs/trivial/tri-multitex-vbo.c
index e319447ac1..ca4edaa5dd 100644
--- a/progs/trivial/tri-multitex-vbo.c
+++ b/progs/trivial/tri-multitex-vbo.c
@@ -192,7 +192,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-orig.c b/progs/trivial/tri-orig.c
index e7cfee3a36..d86d34c39d 100644
--- a/progs/trivial/tri-orig.c
+++ b/progs/trivial/tri-orig.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-query.c b/progs/trivial/tri-query.c
index 94956a86f3..8898f182c7 100644
--- a/progs/trivial/tri-query.c
+++ b/progs/trivial/tri-query.c
@@ -72,7 +72,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-scissor-tri.c b/progs/trivial/tri-scissor-tri.c
index 608ebf29cf..d65502d91b 100644
--- a/progs/trivial/tri-scissor-tri.c
+++ b/progs/trivial/tri-scissor-tri.c
@@ -63,7 +63,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-square.c b/progs/trivial/tri-square.c
index 0b82a1dd8e..2014bd489c 100644
--- a/progs/trivial/tri-square.c
+++ b/progs/trivial/tri-square.c
@@ -63,7 +63,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-stipple.c b/progs/trivial/tri-stipple.c
index aa94fa224b..15a08fec7c 100644
--- a/progs/trivial/tri-stipple.c
+++ b/progs/trivial/tri-stipple.c
@@ -77,7 +77,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-tex-3d.c b/progs/trivial/tri-tex-3d.c
index 3ccbe12510..073a080d7a 100644
--- a/progs/trivial/tri-tex-3d.c
+++ b/progs/trivial/tri-tex-3d.c
@@ -97,7 +97,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-tex.c b/progs/trivial/tri-tex.c
index 56afb4748d..244e154504 100644
--- a/progs/trivial/tri-tex.c
+++ b/progs/trivial/tri-tex.c
@@ -94,7 +94,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-tri.c b/progs/trivial/tri-tri.c
index f996bd01a1..7bf400793e 100644
--- a/progs/trivial/tri-tri.c
+++ b/progs/trivial/tri-tri.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-clip.c b/progs/trivial/tri-unfilled-clip.c
index 2fd894a49a..585919e16d 100644
--- a/progs/trivial/tri-unfilled-clip.c
+++ b/progs/trivial/tri-unfilled-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-edgeflag.c b/progs/trivial/tri-unfilled-edgeflag.c
index 11c21d1bf6..f536f64229 100644
--- a/progs/trivial/tri-unfilled-edgeflag.c
+++ b/progs/trivial/tri-unfilled-edgeflag.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-point.c b/progs/trivial/tri-unfilled-point.c
index 750a254669..11e276e1a5 100644
--- a/progs/trivial/tri-unfilled-point.c
+++ b/progs/trivial/tri-unfilled-point.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-smooth.c b/progs/trivial/tri-unfilled-smooth.c
index eddae176e5..fa31667e65 100644
--- a/progs/trivial/tri-unfilled-smooth.c
+++ b/progs/trivial/tri-unfilled-smooth.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-tri-lit.c b/progs/trivial/tri-unfilled-tri-lit.c
index 1d42b40b71..d6baeb4697 100644
--- a/progs/trivial/tri-unfilled-tri-lit.c
+++ b/progs/trivial/tri-unfilled-tri-lit.c
@@ -65,7 +65,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-tri.c b/progs/trivial/tri-unfilled-tri.c
index 695cc89095..75a70b8a41 100644
--- a/progs/trivial/tri-unfilled-tri.c
+++ b/progs/trivial/tri-unfilled-tri.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-userclip-stip.c b/progs/trivial/tri-unfilled-userclip-stip.c
index ddc0dffd4f..02bfa92f83 100644
--- a/progs/trivial/tri-unfilled-userclip-stip.c
+++ b/progs/trivial/tri-unfilled-userclip-stip.c
@@ -67,7 +67,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled-userclip.c b/progs/trivial/tri-unfilled-userclip.c
index 0dec0bfc9b..11e53fc7cc 100644
--- a/progs/trivial/tri-unfilled-userclip.c
+++ b/progs/trivial/tri-unfilled-userclip.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-unfilled.c b/progs/trivial/tri-unfilled.c
index b98cb9a842..afa1058dad 100644
--- a/progs/trivial/tri-unfilled.c
+++ b/progs/trivial/tri-unfilled.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-userclip.c b/progs/trivial/tri-userclip.c
index 8f37e0fae2..5651c73832 100644
--- a/progs/trivial/tri-userclip.c
+++ b/progs/trivial/tri-userclip.c
@@ -66,7 +66,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-z-9.c b/progs/trivial/tri-z-9.c
index 099e89f6f4..37b9e2f873 100644
--- a/progs/trivial/tri-z-9.c
+++ b/progs/trivial/tri-z-9.c
@@ -61,7 +61,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tri-z-eq.c b/progs/trivial/tri-z-eq.c
index b81c992f7d..6bdac47419 100644
--- a/progs/trivial/tri-z-eq.c
+++ b/progs/trivial/tri-z-eq.c
@@ -61,7 +61,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/trifan-flat-clip.c b/progs/trivial/trifan-flat-clip.c
index 199f91a637..89bc471191 100644
--- a/progs/trivial/trifan-flat-clip.c
+++ b/progs/trivial/trifan-flat-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/trifan-flat-unfilled-clip.c b/progs/trivial/trifan-flat-unfilled-clip.c
index ea3e155387..3a1a226e54 100644
--- a/progs/trivial/trifan-flat-unfilled-clip.c
+++ b/progs/trivial/trifan-flat-unfilled-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/trifan-flat.c b/progs/trivial/trifan-flat.c
index d69b4887e3..d25abdf49e 100644
--- a/progs/trivial/trifan-flat.c
+++ b/progs/trivial/trifan-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/trifan-unfilled.c b/progs/trivial/trifan-unfilled.c
index 91447e4e44..2e3ff8218f 100644
--- a/progs/trivial/trifan-unfilled.c
+++ b/progs/trivial/trifan-unfilled.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/trifan.c b/progs/trivial/trifan.c
index 84eb4172de..b8a7615114 100644
--- a/progs/trivial/trifan.c
+++ b/progs/trivial/trifan.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tristrip-clip.c b/progs/trivial/tristrip-clip.c
index 343e293804..4e6ee1fbca 100644
--- a/progs/trivial/tristrip-clip.c
+++ b/progs/trivial/tristrip-clip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tristrip-flat.c b/progs/trivial/tristrip-flat.c
index 02da97efce..a90b8b6cb6 100644
--- a/progs/trivial/tristrip-flat.c
+++ b/progs/trivial/tristrip-flat.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/tristrip.c b/progs/trivial/tristrip.c
index 77bf2fad28..29915ff111 100644
--- a/progs/trivial/tristrip.c
+++ b/progs/trivial/tristrip.c
@@ -62,7 +62,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/vp-tri-cb-pos.c b/progs/trivial/vp-tri-cb-pos.c
index 42bf9806b1..9cbc4d1193 100644
--- a/progs/trivial/vp-tri-cb-pos.c
+++ b/progs/trivial/vp-tri-cb-pos.c
@@ -81,7 +81,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/vp-tri-cb-tex.c b/progs/trivial/vp-tri-cb-tex.c
index 8290226675..e543e2ec3f 100644
--- a/progs/trivial/vp-tri-cb-tex.c
+++ b/progs/trivial/vp-tri-cb-tex.c
@@ -114,7 +114,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/trivial/vp-tri-invariant.c b/progs/trivial/vp-tri-invariant.c
index ff24139365..4dbe95eb9a 100644
--- a/progs/trivial/vp-tri-invariant.c
+++ b/progs/trivial/vp-tri-invariant.c
@@ -69,7 +69,7 @@ static void Key(unsigned char key, int x, int y)
       case 27:
 	exit(1);
       default:
-	return;
+	break;
     }
 
     glutPostRedisplay();
diff --git a/progs/util/readtex.c b/progs/util/readtex.c
index 8e923b6eb4..81cb626e91 100644
--- a/progs/util/readtex.c
+++ b/progs/util/readtex.c
@@ -109,6 +109,7 @@ static rawImageRec *RawImageOpen(const char *fileName)
          raw->file = fopen(baseName + 1, "rb");
       if(raw->file == NULL) {
          perror(fileName);
+         free(raw);
          return NULL;
       }
    }
@@ -129,6 +130,12 @@ static rawImageRec *RawImageOpen(const char *fileName)
    if (raw->tmp == NULL || raw->tmpR == NULL || raw->tmpG == NULL ||
        raw->tmpB == NULL) {
       fprintf(stderr, "Out of memory!\n");
+      free(raw->tmp);
+      free(raw->tmpR);
+      free(raw->tmpG);
+      free(raw->tmpB);
+      free(raw->tmpA);
+      free(raw);
       return NULL;
    }
 
@@ -138,6 +145,14 @@ static rawImageRec *RawImageOpen(const char *fileName)
       raw->rowSize = (GLint *)malloc(x);
       if (raw->rowStart == NULL || raw->rowSize == NULL) {
          fprintf(stderr, "Out of memory!\n");
+         free(raw->tmp);
+         free(raw->tmpR);
+         free(raw->tmpG);
+         free(raw->tmpB);
+         free(raw->tmpA);
+         free(raw->rowStart);
+         free(raw->rowSize);
+         free(raw);
          return NULL;
       }
       raw->rleEnd = 512 + (2 * x);
@@ -215,6 +230,7 @@ static void RawImageGetData(rawImageRec *raw, TK_RGBImageRec *final)
    final->data = (unsigned char *)malloc((raw->sizeX+1)*(raw->sizeY+1)*4);
    if (final->data == NULL) {
       fprintf(stderr, "Out of memory!\n");
+      return;
    }
 
    ptr = final->data;
@@ -250,6 +266,7 @@ static TK_RGBImageRec *tkRGBImageLoad(const char *fileName)
    final = (TK_RGBImageRec *)malloc(sizeof(TK_RGBImageRec));
    if (final == NULL) {
       fprintf(stderr, "Out of memory!\n");
+      RawImageClose(raw);
       return NULL;
    }
    final->sizeX = raw->sizeX;
@@ -305,6 +322,7 @@ GLboolean LoadRGBMipmaps2( const char *imageFile, GLenum target,
       fprintf(stderr,
               "Error in LoadRGBMipmaps %d-component images not implemented\n",
               image->components );
+      FreeImage(image);
       return GL_FALSE;
    }
 
@@ -356,6 +374,7 @@ GLubyte *LoadRGBImage( const char *imageFile, GLint *width, GLint *height,
       fprintf(stderr,
               "Error in LoadRGBImage %d-component images not implemented\n",
               image->components );
+      FreeImage(image);
       return NULL;
    }
 
@@ -364,8 +383,10 @@ GLubyte *LoadRGBImage( const char *imageFile, GLint *width, GLint *height,
 
    bytes = image->sizeX * image->sizeY * image->components;
    buffer = (GLubyte *) malloc(bytes);
-   if (!buffer)
+   if (!buffer) {
+      FreeImage(image);
       return NULL;
+   }
 
    memcpy( (void *) buffer, (void *) image->data, bytes );
 
@@ -438,6 +459,7 @@ GLushort *LoadYUVImage( const char *imageFile, GLint *width, GLint *height )
       fprintf(stderr,
               "Error in LoadYUVImage %d-component images not implemented\n",
               image->components );
+      FreeImage(image);
       return NULL;
    }
 
diff --git a/progs/util/shaderutil.c b/progs/util/shaderutil.c
index af50d64f51..4fb3199a19 100644
--- a/progs/util/shaderutil.c
+++ b/progs/util/shaderutil.c
@@ -34,7 +34,8 @@ ShadersSupported(void)
       fprintf(stderr, "Warning: Trying ARB GLSL instead of OpenGL 2.x.  This may not work.\n");
       return GL_TRUE;
    }
-   return GL_TRUE;
+   fprintf(stderr, "Sorry, GLSL not supported with this OpenGL.\n");
+   return GL_FALSE;
 }
 
 
@@ -95,6 +96,7 @@ CompileShaderFile(GLenum shaderType, const char *filename)
       shader = CompileShaderText(shaderType, buffer);
    }
    else {
+      fclose(f);
       free(buffer);
       return 0;
    }
diff --git a/progs/vp/vp-tris.c b/progs/vp/vp-tris.c
index 1356242d97..29cd027b00 100644
--- a/progs/vp/vp-tris.c
+++ b/progs/vp/vp-tris.c
@@ -99,9 +99,11 @@ static void Init( void )
    sz = (GLuint) fread(buf, 1, sizeof(buf), f);
    if (!feof(f)) {
       fprintf(stderr, "file too long\n");
+      fclose(f);
       exit(1);
    }
 
+   fclose(f);
    fprintf(stderr, "%.*s\n", sz, buf);
 
    if (strncmp( buf, "!!VP", 4 ) == 0) {
diff --git a/progs/vpglsl/psiz-mul.glsl b/progs/vpglsl/psiz-mul.glsl
index 77f4a46b52..d2a90d8578 100644
--- a/progs/vpglsl/psiz-mul.glsl
+++ b/progs/vpglsl/psiz-mul.glsl
@@ -1,6 +1,6 @@
 
 void main() {
     gl_FrontColor = gl_Color;
-    gl_PointSize = 10 * gl_Color.x;
+    gl_PointSize = 10.0 * gl_Color.x;
     gl_Position = gl_Vertex;
 }
diff --git a/progs/xdemos/Makefile b/progs/xdemos/Makefile
index 53e1c54ef3..77f667978c 100644
--- a/progs/xdemos/Makefile
+++ b/progs/xdemos/Makefile
@@ -8,6 +8,9 @@ INCDIR = $(TOP)/include
 
 LIB_DEP = $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)
 
+# Add X11 and pthread libs to satisfy GNU gold.
+APP_LIB_DEPS += -lX11 -lpthread
+
 LIBS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) $(APP_LIB_DEPS)
 
 PROGS = \
diff --git a/progs/xdemos/corender.c b/progs/xdemos/corender.c
index f2b8145e52..8c70535482 100644
--- a/progs/xdemos/corender.c
+++ b/progs/xdemos/corender.c
@@ -55,6 +55,7 @@ setup_ipc(void)
 
       printf("Waiting for connection from another 'corender'\n");
       Sock = AcceptConnection(k);
+      assert(Sock != -1);
 
       printf("Got connection, sending windowID\n");
 
@@ -186,7 +187,7 @@ redraw(Display *dpy)
        * Without this glClear(), depth buffer for the second process
        * is pretty much broken.
        */
-      //glClear(GL_DEPTH_BUFFER_BIT);
+      /*glClear(GL_DEPTH_BUFFER_BIT);*/
 
       glPushMatrix();
       glTranslatef(1, 0, 0);
diff --git a/progs/xdemos/glxinfo.c b/progs/xdemos/glxinfo.c
index 445d3ea94b..b182a3091d 100644
--- a/progs/xdemos/glxinfo.c
+++ b/progs/xdemos/glxinfo.c
@@ -427,8 +427,6 @@ print_screen_info(Display *dpy, int scrnum, Bool allowDirect, GLboolean limits)
       int nConfigs;
 
       if (!visinfo)
-	 configs = glXChooseFBConfig(dpy, scrnum, fbAttribSingle, &nConfigs);
-      if (!visinfo)
 	 configs = glXChooseFBConfig(dpy, scrnum, fbAttribDouble, &nConfigs);
 
       if (configs) {
@@ -964,8 +962,10 @@ print_fbconfig_info(Display *dpy, int scrnum, InfoMode mode)
    /* get list of all fbconfigs on this screen */
    fbconfigs = glXGetFBConfigs(dpy, scrnum, &numFBConfigs);
 
-   if (numFBConfigs == 0)
+   if (numFBConfigs == 0) {
+      XFree(fbconfigs);
       return;
+   }
 
    printf("%d GLXFBConfigs:\n", numFBConfigs);
    if (mode == Normal)
diff --git a/progs/xdemos/glxpbdemo.c b/progs/xdemos/glxpbdemo.c
index 91fd30dcaa..ecf318ec6a 100644
--- a/progs/xdemos/glxpbdemo.c
+++ b/progs/xdemos/glxpbdemo.c
@@ -106,6 +106,7 @@ MakePbuffer( Display *dpy, int screen, int width, int height )
 
    if (0 == nConfigs || !fbConfigs) {
       printf("Error: glxChooseFBConfig failed\n");
+      XFree(fbConfigs);
       XCloseDisplay(dpy);
       return 0;
    }
diff --git a/progs/xdemos/offset.c b/progs/xdemos/offset.c
index 6c5abf383b..314a4fcdd1 100644
--- a/progs/xdemos/offset.c
+++ b/progs/xdemos/offset.c
@@ -294,6 +294,7 @@ process_input(Display *dpy, Window win) {
 	    default:
 		break;
 	    }
+	    break;
 	case ButtonPress:
 	    prevx = event.xbutton.x;
 	    prevy = event.xbutton.y;
diff --git a/progs/xdemos/pbdemo.c b/progs/xdemos/pbdemo.c
index 2573209336..277df72924 100644
--- a/progs/xdemos/pbdemo.c
+++ b/progs/xdemos/pbdemo.c
@@ -131,6 +131,7 @@ MakePbuffer( Display *dpy, int screen, int width, int height )
       fbConfigs = ChooseFBConfig(dpy, screen, fbAttribs[attempt], &nConfigs);
       if (nConfigs==0 || !fbConfigs) {
          printf("Note: glXChooseFBConfig(%s) failed\n", fbString[attempt]);
+         XFree(fbConfigs);
          continue;
       }
 
diff --git a/progs/xdemos/pbinfo.c b/progs/xdemos/pbinfo.c
index b43adf1bb7..edfa9c1f3b 100644
--- a/progs/xdemos/pbinfo.c
+++ b/progs/xdemos/pbinfo.c
@@ -27,6 +27,7 @@ PrintConfigs(Display *dpy, int screen, Bool horizFormat)
    fbConfigs = GetAllFBConfigs(dpy, screen, &nConfigs);
    if (!nConfigs || !fbConfigs) {
       printf("Error: glxGetFBConfigs failed\n");
+      XFree(fbConfigs);
       return;
    }
 
diff --git a/progs/xdemos/sharedtex_mt.c b/progs/xdemos/sharedtex_mt.c
index 07c1bfcc38..f924448cc4 100644
--- a/progs/xdemos/sharedtex_mt.c
+++ b/progs/xdemos/sharedtex_mt.c
@@ -447,7 +447,7 @@ main(int argc, char *argv[])
    const char *dpyName = XDisplayName(NULL);
    pthread_t t0, t1, t2, t3;
    struct thread_init_arg tia0, tia1, tia2, tia3;
-   struct window *h0, *h1, *h2, *h3;
+   struct window *h0;
 
    XInitThreads();
 
@@ -462,9 +462,9 @@ main(int argc, char *argv[])
 
    /* four windows and contexts sharing display lists and texture objects */
    h0 = AddWindow(gDpy, dpyName,  10,  10, gCtx);
-   h1 = AddWindow(gDpy, dpyName, 330,  10, gCtx);
-   h2 = AddWindow(gDpy, dpyName,  10, 350, gCtx);
-   h3 = AddWindow(gDpy, dpyName, 330, 350, gCtx);
+   (void) AddWindow(gDpy, dpyName, 330,  10, gCtx);
+   (void) AddWindow(gDpy, dpyName,  10, 350, gCtx);
+   (void) AddWindow(gDpy, dpyName, 330, 350, gCtx);
 
    if (!glXMakeCurrent(gDpy, h0->Win, gCtx)) {
       Error(dpyName, "glXMakeCurrent failed for init thread.");
diff --git a/scons/llvm.py b/scons/llvm.py
index 73e9310f71..7b26650290 100644
--- a/scons/llvm.py
+++ b/scons/llvm.py
@@ -58,45 +58,47 @@ def generate(env):
 
         env.PrependENVPath('PATH', llvm_bin_dir)
 
-    if env['msvc']:
+    if env['platform'] == 'windows':
         # XXX: There is no llvm-config on Windows, so assume a standard layout
         if llvm_dir is not None:
             env.Prepend(CPPPATH = [os.path.join(llvm_dir, 'include')])
+            env.AppendUnique(CPPDEFINES = [
+                '__STDC_LIMIT_MACROS', 
+                '__STDC_CONSTANT_MACROS',
+            ])
             env.Prepend(LIBPATH = [os.path.join(llvm_dir, 'lib')])
             env.Prepend(LIBS = [
+                'LLVMX86AsmParser',
+                'LLVMX86AsmPrinter',
+                'LLVMX86CodeGen',
+                'LLVMX86Info',
+                'LLVMLinker',
+                'LLVMipo',
+                'LLVMInterpreter',
+                'LLVMInstrumentation',
+                'LLVMJIT',
+                'LLVMExecutionEngine',
+                'LLVMDebugger',
                 'LLVMBitWriter',
-                'LLVMCore',
-                'LLVMSupport',
-                'LLVMSystem',
-                'LLVMSupport',
-                'LLVMSystem',
-                'LLVMCore',
-                'LLVMCodeGen',
+                'LLVMAsmParser',
+                'LLVMArchive',
+                'LLVMBitReader',
                 'LLVMSelectionDAG',
                 'LLVMAsmPrinter',
-                'LLVMBitReader',
-                'LLVMBitWriter',
-                'LLVMTransformUtils',
-                'LLVMInstrumentation',
+                'LLVMCodeGen',
                 'LLVMScalarOpts',
-                'LLVMipo',
-                'LLVMHello',
-                'LLVMLinker',
-                'LLVMAnalysis',
+                'LLVMTransformUtils',
                 'LLVMipa',
-                'LLVMX86CodeGen',
-                'LLVMX86AsmPrinter',
-                'LLVMExecutionEngine',
-                'LLVMInterpreter',
-                'LLVMJIT',
+                'LLVMAnalysis',
                 'LLVMTarget',
-                'LLVMAsmParser',
-                'LLVMDebugger',
-                'LLVMArchive',
+                'LLVMMC',
+                'LLVMCore',
+                'LLVMSupport',
+                'LLVMSystem',
                 'imagehlp',
                 'psapi',
             ])
-            env['LLVM_VERSION'] = '2.5'
+            env['LLVM_VERSION'] = '2.6'
         return
     elif env.Detect('llvm-config'):
         version = env.backtick('llvm-config --version').rstrip()
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
index d6c16d37b2..eeeed9813f 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aaline.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
@@ -398,9 +398,9 @@ aaline_create_texture(struct aaline_stage *aaline)
    texTemp.target = PIPE_TEXTURE_2D;
    texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */
    texTemp.last_level = MAX_TEXTURE_LEVEL;
-   texTemp.width[0] = 1 << MAX_TEXTURE_LEVEL;
-   texTemp.height[0] = 1 << MAX_TEXTURE_LEVEL;
-   texTemp.depth[0] = 1;
+   texTemp.width0 = 1 << MAX_TEXTURE_LEVEL;
+   texTemp.height0 = 1 << MAX_TEXTURE_LEVEL;
+   texTemp.depth0 = 1;
    pf_get_block(texTemp.format, &texTemp.block);
 
    aaline->texture = screen->texture_create(screen, &texTemp);
@@ -413,11 +413,11 @@ aaline_create_texture(struct aaline_stage *aaline)
     */
    for (level = 0; level <= MAX_TEXTURE_LEVEL; level++) {
       struct pipe_transfer *transfer;
-      const uint size = aaline->texture->width[level];
+      const uint size = u_minify(aaline->texture->width0, level);
       ubyte *data;
       uint i, j;
 
-      assert(aaline->texture->width[level] == aaline->texture->height[level]);
+      assert(aaline->texture->width0 == aaline->texture->height0);
 
       /* This texture is new, no need to flush. 
        */
diff --git a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
index 283502cdf3..27d89721b1 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
@@ -427,9 +427,9 @@ pstip_create_texture(struct pstip_stage *pstip)
    texTemp.target = PIPE_TEXTURE_2D;
    texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */
    texTemp.last_level = 0;
-   texTemp.width[0] = 32;
-   texTemp.height[0] = 32;
-   texTemp.depth[0] = 1;
+   texTemp.width0 = 32;
+   texTemp.height0 = 32;
+   texTemp.depth0 = 1;
    pf_get_block(texTemp.format, &texTemp.block);
 
    pstip->texture = screen->texture_create(screen, &texTemp);
diff --git a/src/gallium/auxiliary/draw/draw_pt.c b/src/gallium/auxiliary/draw/draw_pt.c
index dbb5ac7182..4865a2d854 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/draw/draw_pt.c
@@ -192,7 +192,8 @@ draw_print_arrays(struct draw_context *draw, uint prim, int start, uint count)
                 prim, start, count);
 
    for (i = 0; i < count; i++) {
-      uint ii, j;
+      uint ii = 0;
+      uint j;
 
       if (draw->pt.user.elts) {
          /* indexed arrays */
diff --git a/src/gallium/auxiliary/draw/draw_pt_vcache.c b/src/gallium/auxiliary/draw/draw_pt_vcache.c
index d3f179ced1..757c487454 100644
--- a/src/gallium/auxiliary/draw/draw_pt_vcache.c
+++ b/src/gallium/auxiliary/draw/draw_pt_vcache.c
@@ -346,7 +346,8 @@ vcache_check_run( struct draw_pt_front_end *frontend,
                        vcache->fetch_max,
                        draw_count);
       
-   if (max_index == 0xffffffff ||
+   if (max_index >= DRAW_PIPE_MAX_VERTICES ||
+       fetch_count >= UNDEFINED_VERTEX_ID ||
        fetch_count > draw_count) {
       if (0) debug_printf("fail\n");
       goto fail;
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
index 2ef4293d4d..2f973684f6 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
@@ -1,6 +1,6 @@
 /**************************************************************************
  *
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007-2009 VMware, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -18,7 +18,7 @@
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
@@ -80,11 +80,27 @@ struct fenced_buffer_list
  */
 struct fenced_buffer
 {
+   /* 
+    * Immutable members.
+    */
+
    struct pb_buffer base;
-   
    struct pb_buffer *buffer;
+   struct fenced_buffer_list *list;
+
+   /**
+    * Protected by fenced_buffer_list::mutex
+    */
+   struct list_head head;
 
-   /* FIXME: protect access with mutex */
+   /**
+    * Following members are mutable and protected by this mutex.
+    * 
+    * You may lock this mutex alone, or lock it with fenced_buffer_list::mutex
+    * held, but in order to prevent deadlocks you must never lock 
+    * fenced_buffer_list::mutex with this mutex held.
+    */
+   pipe_mutex mutex;
 
    /**
     * A bitmask of PIPE_BUFFER_USAGE_CPU/GPU_READ/WRITE describing the current
@@ -96,9 +112,6 @@ struct fenced_buffer
    struct pb_validate *vl;
    unsigned validation_flags;
    struct pipe_fence_handle *fence;
-
-   struct list_head head;
-   struct fenced_buffer_list *list;
 };
 
 
@@ -110,15 +123,24 @@ fenced_buffer(struct pb_buffer *buf)
 }
 
 
+/**
+ * Add the buffer to the fenced list.
+ * 
+ * fenced_buffer_list::mutex and fenced_buffer::mutex must be held, in this 
+ * order before calling this function.
+ * 
+ * Reference count should be incremented before calling this function.
+ */
 static INLINE void
-_fenced_buffer_add(struct fenced_buffer *fenced_buf)
+fenced_buffer_add_locked(struct fenced_buffer_list *fenced_list, 
+                         struct fenced_buffer *fenced_buf)
 {
-   struct fenced_buffer_list *fenced_list = fenced_buf->list;
-
    assert(pipe_is_referenced(&fenced_buf->base.base.reference));
    assert(fenced_buf->flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE);
    assert(fenced_buf->fence);
 
+   /* TODO: Move the reference count increment here */
+   
 #ifdef DEBUG
    LIST_DEL(&fenced_buf->head);
    assert(fenced_list->numUnfenced);
@@ -130,32 +152,16 @@ _fenced_buffer_add(struct fenced_buffer *fenced_buf)
 
 
 /**
- * Actually destroy the buffer.
+ * Remove the buffer from the fenced list.
+ * 
+ * fenced_buffer_list::mutex and fenced_buffer::mutex must be held, in this 
+ * order before calling this function.
+ * 
+ * Reference count should be decremented after calling this function.
  */
 static INLINE void
-_fenced_buffer_destroy(struct fenced_buffer *fenced_buf)
-{
-   struct fenced_buffer_list *fenced_list = fenced_buf->list;
-   
-   assert(!pipe_is_referenced(&fenced_buf->base.base.reference));
-   assert(!fenced_buf->fence);
-#ifdef DEBUG
-   assert(fenced_buf->head.prev);
-   assert(fenced_buf->head.next);
-   LIST_DEL(&fenced_buf->head);
-   assert(fenced_list->numUnfenced);
-   --fenced_list->numUnfenced;
-#else
-   (void)fenced_list;
-#endif
-   pb_reference(&fenced_buf->buffer, NULL);
-   FREE(fenced_buf);
-}
-
-
-static INLINE void
-_fenced_buffer_remove(struct fenced_buffer_list *fenced_list,
-                      struct fenced_buffer *fenced_buf)
+fenced_buffer_remove_locked(struct fenced_buffer_list *fenced_list,
+                            struct fenced_buffer *fenced_buf)
 {
    struct pb_fence_ops *ops = fenced_list->ops;
 
@@ -177,37 +183,53 @@ _fenced_buffer_remove(struct fenced_buffer_list *fenced_list,
    ++fenced_list->numUnfenced;
 #endif
    
-   /**
-    * FIXME!!!
-    */
-
-   if(!pipe_is_referenced(&fenced_buf->base.base.reference))
-      _fenced_buffer_destroy(fenced_buf);
+   /* TODO: Move the reference count decrement and destruction here */
 }
 
 
+/**
+ * Wait for the fence to expire, and remove it from the fenced list.
+ * 
+ * fenced_buffer::mutex must be held. fenced_buffer_list::mutex must not be 
+ * held -- it will
+ */
 static INLINE enum pipe_error
-_fenced_buffer_finish(struct fenced_buffer *fenced_buf)
+fenced_buffer_finish_locked(struct fenced_buffer_list *fenced_list,
+                              struct fenced_buffer *fenced_buf)
 {
-   struct fenced_buffer_list *fenced_list = fenced_buf->list;
    struct pb_fence_ops *ops = fenced_list->ops;
+   enum pipe_error ret = PIPE_ERROR;
 
 #if 0
    debug_warning("waiting for GPU");
 #endif
 
+   assert(pipe_is_referenced(&fenced_buf->base.base.reference));
    assert(fenced_buf->fence);
+
+   /* Acquire the global lock */
+   pipe_mutex_unlock(fenced_buf->mutex);
+   pipe_mutex_lock(fenced_list->mutex);
+   pipe_mutex_lock(fenced_buf->mutex);
+
    if(fenced_buf->fence) {
-      if(ops->fence_finish(ops, fenced_buf->fence, 0) != 0) {
-	 return PIPE_ERROR;
+      if(ops->fence_finish(ops, fenced_buf->fence, 0) == 0) {
+         /* Remove from the fenced list */
+         /* TODO: remove consequents */
+         fenced_buffer_remove_locked(fenced_list, fenced_buf);
+         
+         p_atomic_dec(&fenced_buf->base.base.reference.count);
+         assert(pipe_is_referenced(&fenced_buf->base.base.reference));
+
+         fenced_buf->flags &= ~PIPE_BUFFER_USAGE_GPU_READ_WRITE;
+
+         ret = PIPE_OK;
       }
-      /* Remove from the fenced list */
-      /* TODO: remove consequents */
-      _fenced_buffer_remove(fenced_list, fenced_buf);
    }
 
-   fenced_buf->flags &= ~PIPE_BUFFER_USAGE_GPU_READ_WRITE;
-   return PIPE_OK;
+   pipe_mutex_unlock(fenced_list->mutex);
+
+   return ret;
 }
 
 
@@ -215,7 +237,7 @@ _fenced_buffer_finish(struct fenced_buffer *fenced_buf)
  * Free as many fenced buffers from the list head as possible. 
  */
 static void
-_fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list, 
+fenced_buffer_list_check_free_locked(struct fenced_buffer_list *fenced_list, 
                                int wait)
 {
    struct pb_fence_ops *ops = fenced_list->ops;
@@ -228,21 +250,28 @@ _fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list,
    while(curr != &fenced_list->delayed) {
       fenced_buf = LIST_ENTRY(struct fenced_buffer, curr, head);
 
+      pipe_mutex_lock(fenced_buf->mutex);
+
       if(fenced_buf->fence != prev_fence) {
 	 int signaled;
 	 if (wait)
 	    signaled = ops->fence_finish(ops, fenced_buf->fence, 0);
 	 else
 	    signaled = ops->fence_signalled(ops, fenced_buf->fence, 0);
-	 if (signaled != 0)
+	 if (signaled != 0) {
+            pipe_mutex_unlock(fenced_buf->mutex);
 	    break;
+         }
 	 prev_fence = fenced_buf->fence;
       }
       else {
 	 assert(ops->fence_signalled(ops, fenced_buf->fence, 0) == 0);
       }
 
-      _fenced_buffer_remove(fenced_list, fenced_buf);
+      fenced_buffer_remove_locked(fenced_list, fenced_buf);
+      pipe_mutex_unlock(fenced_buf->mutex);
+
+      pb_reference((struct pb_buffer **)&fenced_buf, NULL);
 
       curr = next; 
       next = curr->next;
@@ -256,30 +285,25 @@ fenced_buffer_destroy(struct pb_buffer *buf)
    struct fenced_buffer *fenced_buf = fenced_buffer(buf);   
    struct fenced_buffer_list *fenced_list = fenced_buf->list;
 
-   pipe_mutex_lock(fenced_list->mutex);
    assert(!pipe_is_referenced(&fenced_buf->base.base.reference));
-   if (fenced_buf->fence) {
-      struct pb_fence_ops *ops = fenced_list->ops;
-      if(ops->fence_signalled(ops, fenced_buf->fence, 0) == 0) {
-	 struct list_head *curr, *prev;
-	 curr = &fenced_buf->head;
-	 prev = curr->prev;
-	 do {
-	    fenced_buf = LIST_ENTRY(struct fenced_buffer, curr, head);
-	    assert(ops->fence_signalled(ops, fenced_buf->fence, 0) == 0);
-	    _fenced_buffer_remove(fenced_list, fenced_buf);
-	    curr = prev;
-	    prev = curr->prev;
-	 } while (curr != &fenced_list->delayed);
-      }	  
-      else {
-	 /* delay destruction */
-      }
-   }
-   else {
-      _fenced_buffer_destroy(fenced_buf);
-   }
+   assert(!fenced_buf->fence);
+
+#ifdef DEBUG
+   pipe_mutex_lock(fenced_list->mutex);
+   assert(fenced_buf->head.prev);
+   assert(fenced_buf->head.next);
+   LIST_DEL(&fenced_buf->head);
+   assert(fenced_list->numUnfenced);
+   --fenced_list->numUnfenced;
    pipe_mutex_unlock(fenced_list->mutex);
+#else
+   (void)fenced_list;
+#endif
+
+   pb_reference(&fenced_buf->buffer, NULL);
+
+   pipe_mutex_destroy(fenced_buf->mutex);
+   FREE(fenced_buf);
 }
 
 
@@ -290,24 +314,23 @@ fenced_buffer_map(struct pb_buffer *buf,
    struct fenced_buffer *fenced_buf = fenced_buffer(buf);
    struct fenced_buffer_list *fenced_list = fenced_buf->list;
    struct pb_fence_ops *ops = fenced_list->ops;
-   void *map;
+   void *map = NULL;
+
+   pipe_mutex_lock(fenced_buf->mutex);
 
    assert(!(flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE));
    
    /* Serialize writes */
    if((fenced_buf->flags & PIPE_BUFFER_USAGE_GPU_WRITE) ||
       ((fenced_buf->flags & PIPE_BUFFER_USAGE_GPU_READ) && (flags & PIPE_BUFFER_USAGE_CPU_WRITE))) {
-      if(flags & PIPE_BUFFER_USAGE_DONTBLOCK) {
+      if((flags & PIPE_BUFFER_USAGE_DONTBLOCK) &&
+          ops->fence_signalled(ops, fenced_buf->fence, 0) == 0) {
          /* Don't wait for the GPU to finish writing */
-         if(ops->fence_signalled(ops, fenced_buf->fence, 0) == 0)
-            _fenced_buffer_remove(fenced_list, fenced_buf);
-         else
-            return NULL;
-      }
-      else {
-         /* Wait for the GPU to finish writing */
-         _fenced_buffer_finish(fenced_buf);
+         goto finish;
       }
+
+      /* Wait for the GPU to finish writing */
+      fenced_buffer_finish_locked(fenced_list, fenced_buf);
    }
 
 #if 0
@@ -324,6 +347,9 @@ fenced_buffer_map(struct pb_buffer *buf,
       fenced_buf->flags |= flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE;
    }
 
+finish:
+   pipe_mutex_unlock(fenced_buf->mutex);
+   
    return map;
 }
 
@@ -332,6 +358,9 @@ static void
 fenced_buffer_unmap(struct pb_buffer *buf)
 {
    struct fenced_buffer *fenced_buf = fenced_buffer(buf);
+   
+   pipe_mutex_lock(fenced_buf->mutex);
+   
    assert(fenced_buf->mapcount);
    if(fenced_buf->mapcount) {
       pb_unmap(fenced_buf->buffer);
@@ -339,6 +368,8 @@ fenced_buffer_unmap(struct pb_buffer *buf)
       if(!fenced_buf->mapcount)
 	 fenced_buf->flags &= ~PIPE_BUFFER_USAGE_CPU_READ_WRITE;
    }
+   
+   pipe_mutex_unlock(fenced_buf->mutex);
 }
 
 
@@ -350,11 +381,14 @@ fenced_buffer_validate(struct pb_buffer *buf,
    struct fenced_buffer *fenced_buf = fenced_buffer(buf);
    enum pipe_error ret;
    
+   pipe_mutex_lock(fenced_buf->mutex);
+
    if(!vl) {
       /* invalidate */
       fenced_buf->vl = NULL;
       fenced_buf->validation_flags = 0;
-      return PIPE_OK;
+      ret = PIPE_OK;
+      goto finish;
    }
    
    assert(flags & PIPE_BUFFER_USAGE_GPU_READ_WRITE);
@@ -362,14 +396,17 @@ fenced_buffer_validate(struct pb_buffer *buf,
    flags &= PIPE_BUFFER_USAGE_GPU_READ_WRITE;
 
    /* Buffer cannot be validated in two different lists */ 
-   if(fenced_buf->vl && fenced_buf->vl != vl)
-      return PIPE_ERROR_RETRY;
+   if(fenced_buf->vl && fenced_buf->vl != vl) {
+      ret = PIPE_ERROR_RETRY;
+      goto finish;
+   }
    
 #if 0
    /* Do not validate if buffer is still mapped */
    if(fenced_buf->flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE) {
       /* TODO: wait for the thread that mapped the buffer to unmap it */
-      return PIPE_ERROR_RETRY;
+      ret = PIPE_ERROR_RETRY;
+      goto finish;
    }
    /* Final sanity checking */
    assert(!(fenced_buf->flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE));
@@ -379,17 +416,21 @@ fenced_buffer_validate(struct pb_buffer *buf,
    if(fenced_buf->vl == vl &&
       (fenced_buf->validation_flags & flags) == flags) {
       /* Nothing to do -- buffer already validated */
-      return PIPE_OK;
+      ret = PIPE_OK;
+      goto finish;
    }
    
    ret = pb_validate(fenced_buf->buffer, vl, flags);
    if (ret != PIPE_OK)
-      return ret;
+      goto finish;
    
    fenced_buf->vl = vl;
    fenced_buf->validation_flags |= flags;
    
-   return PIPE_OK;
+finish:
+   pipe_mutex_unlock(fenced_buf->mutex);
+
+   return ret;
 }
 
 
@@ -404,29 +445,36 @@ fenced_buffer_fence(struct pb_buffer *buf,
    fenced_buf = fenced_buffer(buf);
    fenced_list = fenced_buf->list;
    ops = fenced_list->ops;
-   
-   if(fence == fenced_buf->fence) {
-      /* Nothing to do */
-      return;
-   }
 
-   assert(fenced_buf->vl);
-   assert(fenced_buf->validation_flags);
-   
    pipe_mutex_lock(fenced_list->mutex);
-   if (fenced_buf->fence)
-      _fenced_buffer_remove(fenced_list, fenced_buf);
-   if (fence) {
-      ops->fence_reference(ops, &fenced_buf->fence, fence);
-      fenced_buf->flags |= fenced_buf->validation_flags;
-      _fenced_buffer_add(fenced_buf);
-   }
-   pipe_mutex_unlock(fenced_list->mutex);
+   pipe_mutex_lock(fenced_buf->mutex);
+
+   assert(pipe_is_referenced(&fenced_buf->base.base.reference));
+
+   if(fence != fenced_buf->fence) {
+      assert(fenced_buf->vl);
+      assert(fenced_buf->validation_flags);
+      
+      if (fenced_buf->fence) {
+         fenced_buffer_remove_locked(fenced_list, fenced_buf);
+         p_atomic_dec(&fenced_buf->base.base.reference.count);
+         assert(pipe_is_referenced(&fenced_buf->base.base.reference));
+      }
+      if (fence) {
+         ops->fence_reference(ops, &fenced_buf->fence, fence);
+         fenced_buf->flags |= fenced_buf->validation_flags;
+         p_atomic_inc(&fenced_buf->base.base.reference.count);
+         fenced_buffer_add_locked(fenced_list, fenced_buf);
+      }
+
+      pb_fence(fenced_buf->buffer, fence);
    
-   pb_fence(fenced_buf->buffer, fence);
+      fenced_buf->vl = NULL;
+      fenced_buf->validation_flags = 0;
+   }
 
-   fenced_buf->vl = NULL;
-   fenced_buf->validation_flags = 0;
+   pipe_mutex_unlock(fenced_buf->mutex);
+   pipe_mutex_unlock(fenced_list->mutex);
 }
 
 
@@ -436,6 +484,7 @@ fenced_buffer_get_base_buffer(struct pb_buffer *buf,
                               pb_size *offset)
 {
    struct fenced_buffer *fenced_buf = fenced_buffer(buf);
+   /* NOTE: accesses immutable members only -- mutex not necessary */
    pb_get_base_buffer(fenced_buf->buffer, base_buf, offset);
 }
 
@@ -475,6 +524,8 @@ fenced_buffer_create(struct fenced_buffer_list *fenced_list,
    buf->buffer = buffer;
    buf->list = fenced_list;
    
+   pipe_mutex_init(buf->mutex);
+
 #ifdef DEBUG
    pipe_mutex_lock(fenced_list->mutex);
    LIST_ADDTAIL(&buf->head, &fenced_list->unfenced);
@@ -516,7 +567,7 @@ fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list,
                               int wait)
 {
    pipe_mutex_lock(fenced_list->mutex);
-   _fenced_buffer_list_check_free(fenced_list, wait);
+   fenced_buffer_list_check_free_locked(fenced_list, wait);
    pipe_mutex_unlock(fenced_list->mutex);
 }
 
@@ -538,11 +589,13 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list)
    next = curr->next;
    while(curr != &fenced_list->unfenced) {
       fenced_buf = LIST_ENTRY(struct fenced_buffer, curr, head);
+      pipe_mutex_lock(fenced_buf->mutex);
       assert(!fenced_buf->fence);
       debug_printf("%10p %7u %7u\n",
                    (void *) fenced_buf,
                    fenced_buf->base.base.size,
                    p_atomic_read(&fenced_buf->base.base.reference.count));
+      pipe_mutex_unlock(fenced_buf->mutex);
       curr = next; 
       next = curr->next;
    }
@@ -552,6 +605,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list)
    while(curr != &fenced_list->delayed) {
       int signaled;
       fenced_buf = LIST_ENTRY(struct fenced_buffer, curr, head);
+      pipe_mutex_lock(fenced_buf->mutex);
       signaled = ops->fence_signalled(ops, fenced_buf->fence, 0);
       debug_printf("%10p %7u %7u %10p %s\n",
                    (void *) fenced_buf,
@@ -559,6 +613,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list)
                    p_atomic_read(&fenced_buf->base.base.reference.count),
                    (void *) fenced_buf->fence,
                    signaled == 0 ? "y" : "n");
+      pipe_mutex_unlock(fenced_buf->mutex);
       curr = next; 
       next = curr->next;
    }
@@ -579,8 +634,8 @@ fenced_buffer_list_destroy(struct fenced_buffer_list *fenced_list)
 #if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
       sched_yield();
 #endif
-      _fenced_buffer_list_check_free(fenced_list, 1);
       pipe_mutex_lock(fenced_list->mutex);
+      fenced_buffer_list_check_free_locked(fenced_list, 1);
    }
 
 #ifdef DEBUG
@@ -588,6 +643,7 @@ fenced_buffer_list_destroy(struct fenced_buffer_list *fenced_list)
 #endif
       
    pipe_mutex_unlock(fenced_list->mutex);
+   pipe_mutex_destroy(fenced_list->mutex);
    
    fenced_list->ops->destroy(fenced_list->ops);
    
diff --git a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
index eb492076b7..080fd4c731 100644
--- a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
+++ b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
@@ -1129,3 +1129,35 @@ TGSI Instruction Specification
 
   target            Label of target instruction.
 
+
+3  Other tokens
+===============
+
+
+3.1  Declaration Semantic
+-------------------------
+
+
+  Follows Declaration token if Semantic bit is set.
+
+  Since its purpose is to link a shader with other stages of the pipeline,
+  it is valid to follow only those Declaration tokens that declare a register
+  either in INPUT or OUTPUT file.
+
+  SemanticName field contains the semantic name of the register being declared.
+  There is no default value.
+
+  SemanticIndex is an optional subscript that can be used to distinguish
+  different register declarations with the same semantic name. The default value
+  is 0.
+
+  The meanings of the individual semantic names are explained in the following
+  sections.
+
+
+3.1.1  FACE
+
+  Valid only in a fragment shader INPUT declaration.
+
+  FACE.x is negative when the primitive is back facing. FACE.x is positive
+  when the primitive is front facing.
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c
index 4fa10e2f7e..9791e58db3 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.c
@@ -419,6 +419,7 @@ tgsi_default_instruction( void )
    instruction.NrTokens = 1;
    instruction.Opcode = TGSI_OPCODE_MOV;
    instruction.Saturate = TGSI_SAT_NONE;
+   instruction.Predicate = 0;
    instruction.NumDstRegs = 1;
    instruction.NumSrcRegs = 1;
    instruction.Padding  = 0;
@@ -428,12 +429,12 @@ tgsi_default_instruction( void )
 }
 
 struct tgsi_instruction
-tgsi_build_instruction(
-   unsigned opcode,
-   unsigned saturate,
-   unsigned num_dst_regs,
-   unsigned num_src_regs,
-   struct tgsi_header *header )
+tgsi_build_instruction(unsigned opcode,
+                       unsigned saturate,
+                       unsigned predicate,
+                       unsigned num_dst_regs,
+                       unsigned num_src_regs,
+                       struct tgsi_header *header)
 {
    struct tgsi_instruction instruction;
 
@@ -445,6 +446,7 @@ tgsi_build_instruction(
    instruction = tgsi_default_instruction();
    instruction.Opcode = opcode;
    instruction.Saturate = saturate;
+   instruction.Predicate = predicate;
    instruction.NumDstRegs = num_dst_regs;
    instruction.NumSrcRegs = num_src_regs;
 
@@ -472,9 +474,9 @@ tgsi_default_full_instruction( void )
    unsigned i;
 
    full_instruction.Instruction = tgsi_default_instruction();
+   full_instruction.InstructionPredicate = tgsi_default_instruction_predicate();
    full_instruction.InstructionExtLabel = tgsi_default_instruction_ext_label();
    full_instruction.InstructionExtTexture = tgsi_default_instruction_ext_texture();
-   full_instruction.InstructionExtPredicate = tgsi_default_instruction_ext_predicate();
    for( i = 0;  i < TGSI_FULL_MAX_DST_REGISTERS; i++ ) {
       full_instruction.FullDstRegisters[i] = tgsi_default_full_dst_register();
    }
@@ -504,14 +506,34 @@ tgsi_build_full_instruction(
    instruction = (struct tgsi_instruction *) &tokens[size];
    size++;
 
-   *instruction = tgsi_build_instruction(
-      full_inst->Instruction.Opcode,
-      full_inst->Instruction.Saturate,
-      full_inst->Instruction.NumDstRegs,
-      full_inst->Instruction.NumSrcRegs,
-      header );
+   *instruction = tgsi_build_instruction(full_inst->Instruction.Opcode,
+                                         full_inst->Instruction.Saturate,
+                                         full_inst->Instruction.Predicate,
+                                         full_inst->Instruction.NumDstRegs,
+                                         full_inst->Instruction.NumSrcRegs,
+                                         header);
    prev_token = (struct tgsi_token  *) instruction;
 
+   if (full_inst->Instruction.Predicate) {
+      struct tgsi_instruction_predicate *instruction_predicate;
+
+      if (maxsize <= size) {
+         return 0;
+      }
+      instruction_predicate = (struct tgsi_instruction_predicate *)&tokens[size];
+      size++;
+
+      *instruction_predicate =
+         tgsi_build_instruction_predicate(full_inst->InstructionPredicate.Index,
+                                          full_inst->InstructionPredicate.Negate,
+                                          full_inst->InstructionPredicate.SwizzleX,
+                                          full_inst->InstructionPredicate.SwizzleY,
+                                          full_inst->InstructionPredicate.SwizzleZ,
+                                          full_inst->InstructionPredicate.SwizzleW,
+                                          instruction,
+                                          header);
+   }
+
    if( tgsi_compare_instruction_ext_label(
          full_inst->InstructionExtLabel,
          tgsi_default_instruction_ext_label() ) ) {
@@ -550,29 +572,6 @@ tgsi_build_full_instruction(
       prev_token = (struct tgsi_token  *) instruction_ext_texture;
    }
 
-   if (tgsi_compare_instruction_ext_predicate(full_inst->InstructionExtPredicate,
-                                              tgsi_default_instruction_ext_predicate())) {
-      struct tgsi_instruction_ext_predicate *instruction_ext_predicate;
-
-      if (maxsize <= size) {
-         return 0;
-      }
-      instruction_ext_predicate = (struct tgsi_instruction_ext_predicate *)&tokens[size];
-      size++;
-
-      *instruction_ext_predicate =
-         tgsi_build_instruction_ext_predicate(full_inst->InstructionExtPredicate.SrcIndex,
-                                              full_inst->InstructionExtPredicate.Negate,
-                                              full_inst->InstructionExtPredicate.SwizzleX,
-                                              full_inst->InstructionExtPredicate.SwizzleY,
-                                              full_inst->InstructionExtPredicate.SwizzleZ,
-                                              full_inst->InstructionExtPredicate.SwizzleW,
-                                              prev_token,
-                                              instruction,
-                                              header);
-      prev_token = (struct tgsi_token *)instruction_ext_predicate;
-   }
-
    for( i = 0;  i <   full_inst->Instruction.NumDstRegs; i++ ) {
       const struct tgsi_full_dst_register *reg = &full_inst->FullDstRegisters[i];
       struct tgsi_dst_register *dst_register;
@@ -746,6 +745,47 @@ tgsi_build_full_instruction(
    return size;
 }
 
+struct tgsi_instruction_predicate
+tgsi_default_instruction_predicate(void)
+{
+   struct tgsi_instruction_predicate instruction_predicate;
+
+   instruction_predicate.SwizzleX = TGSI_SWIZZLE_X;
+   instruction_predicate.SwizzleY = TGSI_SWIZZLE_Y;
+   instruction_predicate.SwizzleZ = TGSI_SWIZZLE_Z;
+   instruction_predicate.SwizzleW = TGSI_SWIZZLE_W;
+   instruction_predicate.Negate = 0;
+   instruction_predicate.Index = 0;
+   instruction_predicate.Padding = 0;
+
+   return instruction_predicate;
+}
+
+struct tgsi_instruction_predicate
+tgsi_build_instruction_predicate(int index,
+                                 unsigned negate,
+                                 unsigned swizzleX,
+                                 unsigned swizzleY,
+                                 unsigned swizzleZ,
+                                 unsigned swizzleW,
+                                 struct tgsi_instruction *instruction,
+                                 struct tgsi_header *header)
+{
+   struct tgsi_instruction_predicate instruction_predicate;
+
+   instruction_predicate = tgsi_default_instruction_predicate();
+   instruction_predicate.SwizzleX = swizzleX;
+   instruction_predicate.SwizzleY = swizzleY;
+   instruction_predicate.SwizzleZ = swizzleZ;
+   instruction_predicate.SwizzleW = swizzleW;
+   instruction_predicate.Negate = negate;
+   instruction_predicate.Index = index;
+
+   instruction_grow(instruction, header);
+
+   return instruction_predicate;
+}
+
 /** test for inequality of 32-bit values pointed to by a and b */
 static INLINE boolean
 compare32(const void *a, const void *b)
@@ -835,60 +875,6 @@ tgsi_build_instruction_ext_texture(
    return instruction_ext_texture;
 }
 
-struct tgsi_instruction_ext_predicate
-tgsi_default_instruction_ext_predicate(void)
-{
-   struct tgsi_instruction_ext_predicate instruction_ext_predicate;
-
-   instruction_ext_predicate.Type = TGSI_INSTRUCTION_EXT_TYPE_PREDICATE;
-   instruction_ext_predicate.SwizzleX = TGSI_SWIZZLE_X;
-   instruction_ext_predicate.SwizzleY = TGSI_SWIZZLE_Y;
-   instruction_ext_predicate.SwizzleZ = TGSI_SWIZZLE_Z;
-   instruction_ext_predicate.SwizzleW = TGSI_SWIZZLE_W;
-   instruction_ext_predicate.Negate = 0;
-   instruction_ext_predicate.SrcIndex = 0;
-   instruction_ext_predicate.Padding = 0;
-   instruction_ext_predicate.Extended = 0;
-
-   return instruction_ext_predicate;
-}
-
-unsigned
-tgsi_compare_instruction_ext_predicate(struct tgsi_instruction_ext_predicate a,
-                                       struct tgsi_instruction_ext_predicate b)
-{
-   a.Padding = b.Padding = 0;
-   a.Extended = b.Extended = 0;
-   return compare32(&a, &b);
-}
-
-struct tgsi_instruction_ext_predicate
-tgsi_build_instruction_ext_predicate(unsigned index,
-                                     unsigned negate,
-                                     unsigned swizzleX,
-                                     unsigned swizzleY,
-                                     unsigned swizzleZ,
-                                     unsigned swizzleW,
-                                     struct tgsi_token *prev_token,
-                                     struct tgsi_instruction *instruction,
-                                     struct tgsi_header *header)
-{
-   struct tgsi_instruction_ext_predicate instruction_ext_predicate;
-
-   instruction_ext_predicate = tgsi_default_instruction_ext_predicate();
-   instruction_ext_predicate.SwizzleX = swizzleX;
-   instruction_ext_predicate.SwizzleY = swizzleY;
-   instruction_ext_predicate.SwizzleZ = swizzleZ;
-   instruction_ext_predicate.SwizzleW = swizzleW;
-   instruction_ext_predicate.Negate = negate;
-   instruction_ext_predicate.SrcIndex = index;
-
-   prev_token->Extended = 1;
-   instruction_grow(instruction, header);
-
-   return instruction_ext_predicate;
-}
-
 struct tgsi_src_register
 tgsi_default_src_register( void )
 {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.h b/src/gallium/auxiliary/tgsi/tgsi_build.h
index 669712eb8f..0fe5f229d3 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.h
@@ -143,6 +143,7 @@ struct tgsi_instruction
 tgsi_build_instruction(
    unsigned opcode,
    unsigned saturate,
+   unsigned predicate,
    unsigned num_dst_regs,
    unsigned num_src_regs,
    struct tgsi_header *header );
@@ -157,6 +158,19 @@ tgsi_build_full_instruction(
    struct tgsi_header *header,
    unsigned maxsize );
 
+struct tgsi_instruction_predicate
+tgsi_default_instruction_predicate(void);
+
+struct tgsi_instruction_predicate
+tgsi_build_instruction_predicate(int index,
+                                 unsigned negate,
+                                 unsigned swizzleX,
+                                 unsigned swizzleY,
+                                 unsigned swizzleZ,
+                                 unsigned swizzleW,
+                                 struct tgsi_instruction *instruction,
+                                 struct tgsi_header *header);
+
 struct tgsi_instruction_ext_label
 tgsi_default_instruction_ext_label( void );
 
@@ -187,24 +201,6 @@ tgsi_build_instruction_ext_texture(
    struct tgsi_instruction *instruction,
    struct tgsi_header *header );
 
-struct tgsi_instruction_ext_predicate
-tgsi_default_instruction_ext_predicate(void);
-
-unsigned
-tgsi_compare_instruction_ext_predicate(struct tgsi_instruction_ext_predicate a,
-                                       struct tgsi_instruction_ext_predicate b);
-
-struct tgsi_instruction_ext_predicate
-tgsi_build_instruction_ext_predicate(unsigned index,
-                                     unsigned negate,
-                                     unsigned swizzleX,
-                                     unsigned swizzleY,
-                                     unsigned swizzleZ,
-                                     unsigned swizzleW,
-                                     struct tgsi_token *prev_token,
-                                     struct tgsi_instruction *instruction,
-                                     struct tgsi_header *header);
-
 struct tgsi_src_register
 tgsi_default_src_register( void );
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index ade5e7cbbd..2bf367ba50 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -62,9 +62,6 @@
 
 #define FAST_MATH 1
 
-/** for tgsi_full_instruction::Flags */
-#define SOA_DEPENDENCY_FLAG 0x1
-
 #define TILE_TOP_LEFT     0
 #define TILE_TOP_RIGHT    1
 #define TILE_BOTTOM_LEFT  2
@@ -340,20 +337,6 @@ tgsi_exec_machine_bind_shader(
             maxInstructions += 10;
          }
 
-         if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) {
-            uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
-            parse.FullToken.FullInstruction.Flags = SOA_DEPENDENCY_FLAG;
-            /* XXX we only handle SOA dependencies properly for MOV/SWZ
-             * at this time!
-             */
-            if (opcode != TGSI_OPCODE_MOV) {
-               debug_printf("Warning: SOA dependency in instruction"
-                            " is not handled:\n");
-               tgsi_dump_instruction(&parse.FullToken.FullInstruction,
-                                     numInstructions);
-            }
-         }
-
          memcpy(instructions + numInstructions,
                 &parse.FullToken.FullInstruction,
                 sizeof(instructions[0]));
@@ -395,6 +378,7 @@ tgsi_exec_machine_create( void )
 
    mach->Addrs = &mach->Temps[TGSI_EXEC_TEMP_ADDR];
    mach->MaxGeometryShaderOutputs = TGSI_MAX_TOTAL_VERTICES;
+   mach->Predicates = &mach->Temps[TGSI_EXEC_TEMP_P0];
 
    /* Setup constants. */
    for( i = 0; i < 4; i++ ) {
@@ -526,7 +510,7 @@ micro_ddy(
    dst->f[0] =
    dst->f[1] =
    dst->f[2] =
-   dst->f[3] = src->f[TILE_TOP_LEFT] - src->f[TILE_BOTTOM_LEFT];
+   dst->f[3] = src->f[TILE_BOTTOM_LEFT] - src->f[TILE_TOP_LEFT];
 }
 
 static void
@@ -604,6 +588,24 @@ micro_exp2(
    dst->f[2] = util_fast_exp2( src->f[2] );
    dst->f[3] = util_fast_exp2( src->f[3] );
 #else
+
+#if DEBUG
+   /* Inf is okay for this instruction, so clamp it to silence assertions. */
+   uint i;
+   union tgsi_exec_channel clamped;
+
+   for (i = 0; i < 4; i++) {
+      if (src->f[i] > 127.99999f) {
+         clamped.f[i] = 127.99999f;
+      } else if (src->f[i] < -126.99999f) {
+         clamped.f[i] = -126.99999f;
+      } else {
+         clamped.f[i] = src->f[i];
+      }
+   }
+   src = &clamped;
+#endif
+
    dst->f[0] = powf( 2.0f, src->f[0] );
    dst->f[1] = powf( 2.0f, src->f[1] );
    dst->f[2] = powf( 2.0f, src->f[2] );
@@ -1202,10 +1204,10 @@ fetch_src_file_channel(
          assert(index->i[1] < TGSI_EXEC_NUM_PREDS);
          assert(index->i[2] < TGSI_EXEC_NUM_PREDS);
          assert(index->i[3] < TGSI_EXEC_NUM_PREDS);
-         chan->u[0] = mach->Addrs[0].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Addrs[0].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Addrs[0].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Addrs[0].xyzw[swizzle].u[3];
+         chan->u[0] = mach->Predicates[0].xyzw[swizzle].u[0];
+         chan->u[1] = mach->Predicates[0].xyzw[swizzle].u[1];
+         chan->u[2] = mach->Predicates[0].xyzw[swizzle].u[2];
+         chan->u[3] = mach->Predicates[0].xyzw[swizzle].u[3];
          break;
 
       case TGSI_FILE_OUTPUT:
@@ -1496,10 +1498,17 @@ store_dest(
       dst = &mach->Addrs[index].xyzw[chan_index];
       break;
 
+   case TGSI_FILE_LOOP:
+      assert(reg->DstRegister.Index == 0);
+      assert(mach->LoopCounterStackTop > 0);
+      assert(chan_index == CHAN_X);
+      dst = &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[chan_index];
+      break;
+
    case TGSI_FILE_PREDICATE:
       index = reg->DstRegister.Index;
       assert(index < TGSI_EXEC_NUM_PREDS);
-      dst = &mach->Addrs[index].xyzw[chan_index];
+      dst = &mach->Predicates[index].xyzw[chan_index];
       break;
 
    default:
@@ -1507,6 +1516,47 @@ store_dest(
       return;
    }
 
+   if (inst->Instruction.Predicate) {
+      uint swizzle;
+      union tgsi_exec_channel *pred;
+
+      switch (chan_index) {
+      case CHAN_X:
+         swizzle = inst->InstructionPredicate.SwizzleX;
+         break;
+      case CHAN_Y:
+         swizzle = inst->InstructionPredicate.SwizzleY;
+         break;
+      case CHAN_Z:
+         swizzle = inst->InstructionPredicate.SwizzleZ;
+         break;
+      case CHAN_W:
+         swizzle = inst->InstructionPredicate.SwizzleW;
+         break;
+      default:
+         assert(0);
+         return;
+      }
+
+      assert(inst->InstructionPredicate.Index == 0);
+
+      pred = &mach->Predicates[inst->InstructionPredicate.Index].xyzw[swizzle];
+
+      if (inst->InstructionPredicate.Negate) {
+         for (i = 0; i < QUAD_SIZE; i++) {
+            if (pred->u[i]) {
+               execmask &= ~(1 << i);
+            }
+         }
+      } else {
+         for (i = 0; i < QUAD_SIZE; i++) {
+            if (!pred->u[i]) {
+               execmask &= ~(1 << i);
+            }
+         }
+      }
+   }
+
    switch (inst->Instruction.Saturate) {
    case TGSI_SAT_NONE:
       for (i = 0; i < QUAD_SIZE; i++)
@@ -1762,6 +1812,64 @@ exec_tex(struct tgsi_exec_machine *mach,
    }
 }
 
+static void
+exec_txd(struct tgsi_exec_machine *mach,
+         const struct tgsi_full_instruction *inst)
+{
+   const uint unit = inst->FullSrcRegisters[3].SrcRegister.Index;
+   union tgsi_exec_channel r[4];
+   uint chan_index;
+
+   /*
+    * XXX: This is fake TXD -- the derivatives are not taken into account, yet.
+    */
+
+   switch (inst->InstructionExtTexture.Texture) {
+   case TGSI_TEXTURE_1D:
+   case TGSI_TEXTURE_SHADOW1D:
+
+      FETCH(&r[0], 0, CHAN_X);
+
+      fetch_texel(mach->Samplers[unit],
+                  &r[0], &ZeroVec, &ZeroVec, 0.0f,  /* S, T, P, BIAS */
+                  &r[0], &r[1], &r[2], &r[3]);      /* R, G, B, A */
+      break;
+
+   case TGSI_TEXTURE_2D:
+   case TGSI_TEXTURE_RECT:
+   case TGSI_TEXTURE_SHADOW2D:
+   case TGSI_TEXTURE_SHADOWRECT:
+
+      FETCH(&r[0], 0, CHAN_X);
+      FETCH(&r[1], 0, CHAN_Y);
+      FETCH(&r[2], 0, CHAN_Z);
+
+      fetch_texel(mach->Samplers[unit],
+                  &r[0], &r[1], &r[2], 0.0f,    /* inputs */
+                  &r[0], &r[1], &r[2], &r[3]);  /* outputs */
+      break;
+
+   case TGSI_TEXTURE_3D:
+   case TGSI_TEXTURE_CUBE:
+
+      FETCH(&r[0], 0, CHAN_X);
+      FETCH(&r[1], 0, CHAN_Y);
+      FETCH(&r[2], 0, CHAN_Z);
+
+      fetch_texel(mach->Samplers[unit],
+                  &r[0], &r[1], &r[2], 0.0f,
+                  &r[0], &r[1], &r[2], &r[3]);
+      break;
+
+   default:
+      assert(0);
+   }
+
+   FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+      STORE(&r[chan_index], 0, chan_index);
+   }
+}
+
 
 /**
  * Evaluate a constant-valued coefficient at the position of the
@@ -1831,53 +1939,58 @@ typedef void (* eval_coef_func)(
    unsigned chan );
 
 static void
-exec_declaration(
-   struct tgsi_exec_machine *mach,
-   const struct tgsi_full_declaration *decl )
+exec_declaration(struct tgsi_exec_machine *mach,
+                 const struct tgsi_full_declaration *decl)
 {
-   if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) {
-      if( decl->Declaration.File == TGSI_FILE_INPUT ) {
-         unsigned first, last, mask;
-         eval_coef_func eval;
+   if (mach->Processor == TGSI_PROCESSOR_FRAGMENT) {
+      if (decl->Declaration.File == TGSI_FILE_INPUT) {
+         uint first, last, mask;
 
          first = decl->DeclarationRange.First;
          last = decl->DeclarationRange.Last;
          mask = decl->Declaration.UsageMask;
 
-         switch( decl->Declaration.Interpolate ) {
-         case TGSI_INTERPOLATE_CONSTANT:
-            eval = eval_constant_coef;
-            break;
+         if (decl->Semantic.SemanticName == TGSI_SEMANTIC_POSITION) {
+            assert(decl->Semantic.SemanticIndex == 0);
+            assert(first == last);
+            assert(mask = TGSI_WRITEMASK_XYZW);
 
-         case TGSI_INTERPOLATE_LINEAR:
-            eval = eval_linear_coef;
-            break;
+            mach->Inputs[first] = mach->QuadPos;
+         } else if (decl->Semantic.SemanticName == TGSI_SEMANTIC_FACE) {
+            uint i;
 
-         case TGSI_INTERPOLATE_PERSPECTIVE:
-            eval = eval_perspective_coef;
-            break;
+            assert(decl->Semantic.SemanticIndex == 0);
+            assert(first == last);
 
-         default:
-            assert( 0 );
-            return;
-         }
-
-         if( mask == TGSI_WRITEMASK_XYZW ) {
-            unsigned i, j;
-
-            for( i = first; i <= last; i++ ) {
-               for( j = 0; j < NUM_CHANNELS; j++ ) {
-                  eval( mach, i, j );
-               }
+            for (i = 0; i < QUAD_SIZE; i++) {
+               mach->Inputs[first].xyzw[0].f[i] = mach->Face;
+            }
+         } else {
+            eval_coef_func eval;
+            uint i, j;
+
+            switch (decl->Declaration.Interpolate) {
+            case TGSI_INTERPOLATE_CONSTANT:
+               eval = eval_constant_coef;
+               break;
+
+            case TGSI_INTERPOLATE_LINEAR:
+               eval = eval_linear_coef;
+               break;
+
+            case TGSI_INTERPOLATE_PERSPECTIVE:
+               eval = eval_perspective_coef;
+               break;
+
+            default:
+               assert(0);
+               return;
             }
-         }
-         else {
-            unsigned i, j;
 
-            for( j = 0; j < NUM_CHANNELS; j++ ) {
-               if( mask & (1 << j) ) {
-                  for( i = first; i <= last; i++ ) {
-                     eval( mach, i, j );
+            for (j = 0; j < NUM_CHANNELS; j++) {
+               if (mask & (1 << j)) {
+                  for (i = first; i <= last; i++) {
+                     eval(mach, i, j);
                   }
                }
             }
@@ -1894,6 +2007,7 @@ exec_instruction(
 {
    uint chan_index;
    union tgsi_exec_channel r[10];
+   union tgsi_exec_channel d[8];
 
    (*pc)++;
 
@@ -1902,42 +2016,27 @@ exec_instruction(
    case TGSI_OPCODE_FLR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_flr( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_flr(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_MOV:
-      if (inst->Flags & SOA_DEPENDENCY_FLAG) {
-         /* Do all fetches into temp regs, then do all stores to avoid
-          * intermediate/accidental clobbering.  This could be done all the
-          * time for MOV but for other instructions we'll need more temps...
-          */
-         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
-            FETCH( &r[chan_index], 0, chan_index );
-         }
-         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
-            STORE( &r[chan_index], 0, chan_index );
-         }
+      FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+         FETCH(&d[chan_index], 0, chan_index);
       }
-      else {
-         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
-            FETCH( &r[0], 0, chan_index );
-            STORE( &r[0], 0, chan_index );
-         }
+      FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_LIT:
-      if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
-         STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
-      }
-
       if (IS_CHANNEL_ENABLED( *inst, CHAN_Y ) || IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
          FETCH( &r[0], 0, CHAN_X );
          if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
-            micro_max( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-            STORE( &r[0], 0, CHAN_Y );
+            micro_max(&d[CHAN_Y], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
          }
 
          if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
@@ -1948,11 +2047,19 @@ exec_instruction(
             micro_min( &r[2], &r[2], &mach->Temps[TEMP_128_I].xyzw[TEMP_128_C] );
             micro_max( &r[2], &r[2], &mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C] );
             micro_pow( &r[1], &r[1], &r[2] );
-            micro_lt( &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-            STORE( &r[0], 0, CHAN_Z );
+            micro_lt(&d[CHAN_Z], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
          }
-      }
 
+         if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+            STORE(&d[CHAN_Y], 0, CHAN_Y);
+         }
+         if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+            STORE(&d[CHAN_Z], 0, CHAN_Z);
+         }
+      }
+      if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
+         STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
+      }
       if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
          STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W );
       }
@@ -2020,14 +2127,13 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_MUL:
-      FOR_EACH_ENABLED_CHANNEL( *inst, chan_index )
-      {
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
-
-         micro_mul( &r[0], &r[0], &r[1] );
-
-         STORE(&r[0], 0, chan_index);
+         micro_mul(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2035,8 +2141,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_add( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_add(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2092,25 +2200,29 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_DST:
-      if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
-         STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
-      }
-
       if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
          FETCH( &r[0], 0, CHAN_Y );
          FETCH( &r[1], 1, CHAN_Y);
-         micro_mul( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, CHAN_Y );
+         micro_mul(&d[CHAN_Y], &r[0], &r[1]);
       }
-
       if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
-         FETCH( &r[0], 0, CHAN_Z );
-         STORE( &r[0], 0, CHAN_Z );
+         FETCH(&d[CHAN_Z], 0, CHAN_Z);
       }
-
       if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
-         FETCH( &r[0], 1, CHAN_W );
-         STORE( &r[0], 0, CHAN_W );
+         FETCH(&d[CHAN_W], 1, CHAN_W);
+      }
+
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+         STORE(&mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+         STORE(&d[CHAN_Y], 0, CHAN_Y);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+         STORE(&d[CHAN_Z], 0, CHAN_Z);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
+         STORE(&d[CHAN_W], 0, CHAN_W);
       }
       break;
 
@@ -2120,9 +2232,10 @@ exec_instruction(
          FETCH(&r[1], 1, chan_index);
 
          /* XXX use micro_min()?? */
-         micro_lt( &r[0], &r[0], &r[1], &r[0], &r[1] );
-
-         STORE(&r[0], 0, chan_index);
+         micro_lt(&d[chan_index], &r[0], &r[1], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2132,9 +2245,10 @@ exec_instruction(
          FETCH(&r[1], 1, chan_index);
 
          /* XXX use micro_max()?? */
-         micro_lt( &r[0], &r[0], &r[1], &r[1], &r[0] );
-
-         STORE(&r[0], 0, chan_index );
+         micro_lt(&d[chan_index], &r[0], &r[1], &r[1], &r[0] );
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2143,8 +2257,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_lt(&d[chan_index], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2153,9 +2269,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-
-         micro_le( &r[0], &r[1], &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_le(&d[chan_index], &r[1], &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2166,8 +2283,10 @@ exec_instruction(
          FETCH( &r[1], 1, chan_index );
          micro_mul( &r[0], &r[0], &r[1] );
          FETCH( &r[1], 2, chan_index );
-         micro_add( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_add(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2175,10 +2294,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
-
-         micro_sub( &r[0], &r[0], &r[1] );
-
-         STORE(&r[0], 0, chan_index);
+         micro_sub(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2187,12 +2306,12 @@ exec_instruction(
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
          FETCH(&r[2], 2, chan_index);
-
          micro_sub( &r[1], &r[1], &r[2] );
          micro_mul( &r[0], &r[0], &r[1] );
-         micro_add( &r[0], &r[0], &r[2] );
-
-         STORE(&r[0], 0, chan_index);
+         micro_add(&d[chan_index], &r[0], &r[2]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2201,8 +2320,10 @@ exec_instruction(
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
          FETCH(&r[2], 2, chan_index);
-         micro_lt(&r[0], &mach->Temps[TEMP_HALF_I].xyzw[TEMP_HALF_C], &r[2], &r[0], &r[1]);
-         STORE(&r[0], 0, chan_index);
+         micro_lt(&d[chan_index], &mach->Temps[TEMP_HALF_I].xyzw[TEMP_HALF_C], &r[2], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2227,8 +2348,10 @@ exec_instruction(
    case TGSI_OPCODE_FRC:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_frc( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_frc(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2238,8 +2361,10 @@ exec_instruction(
          FETCH(&r[1], 1, chan_index);
          micro_max(&r[0], &r[0], &r[1]);
          FETCH(&r[1], 2, chan_index);
-         micro_min(&r[0], &r[0], &r[1]);
-         STORE(&r[0], 0, chan_index);
+         micro_min(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2247,19 +2372,17 @@ exec_instruction(
    case TGSI_OPCODE_ARR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_rnd( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_rnd(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_EX2:
       FETCH(&r[0], 0, CHAN_X);
 
-#if FAST_MATH
       micro_exp2( &r[0], &r[0] );
-#else
-      micro_pow( &r[0], &mach->Temps[TEMP_2_I].xyzw[TEMP_2_C], &r[0] );
-#endif
 
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( &r[0], 0, chan_index );
@@ -2295,11 +2418,7 @@ exec_instruction(
       FETCH(&r[4], 1, CHAN_Y);
 
       micro_mul( &r[5], &r[3], &r[4] );
-      micro_sub( &r[2], &r[2], &r[5] );
-
-      if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
-         STORE( &r[2], 0, CHAN_X );
-      }
+      micro_sub(&d[CHAN_X], &r[2], &r[5]);
 
       FETCH(&r[2], 1, CHAN_X);
 
@@ -2308,20 +2427,21 @@ exec_instruction(
       FETCH(&r[5], 0, CHAN_X);
 
       micro_mul( &r[1], &r[1], &r[5] );
-      micro_sub( &r[3], &r[3], &r[1] );
-
-      if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
-         STORE( &r[3], 0, CHAN_Y );
-      }
+      micro_sub(&d[CHAN_Y], &r[3], &r[1]);
 
       micro_mul( &r[5], &r[5], &r[4] );
       micro_mul( &r[0], &r[0], &r[2] );
-      micro_sub( &r[5], &r[5], &r[0] );
+      micro_sub(&d[CHAN_Z], &r[5], &r[0]);
 
-      if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
-         STORE( &r[5], 0, CHAN_Z );
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+         STORE(&d[CHAN_X], 0, CHAN_X);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+         STORE(&d[CHAN_Y], 0, CHAN_Y);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+         STORE(&d[CHAN_Z], 0, CHAN_Z);
       }
-
       if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
          STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W );
       }
@@ -2330,11 +2450,11 @@ exec_instruction(
     case TGSI_OPCODE_ABS:
        FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
           FETCH(&r[0], 0, chan_index);
-
-          micro_abs( &r[0], &r[0] );
-
-          STORE(&r[0], 0, chan_index);
+          micro_abs(&d[chan_index], &r[0]);
        }
+       FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
+      }
        break;
 
    case TGSI_OPCODE_RCC:
@@ -2386,16 +2506,20 @@ exec_instruction(
    case TGSI_OPCODE_DDX:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_ddx( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_ddx(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_DDY:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_ddy( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_ddy(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2476,10 +2600,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_eq( &r[0], &r[0], &r[1],
-                   &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C],
-                   &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_eq(&d[chan_index], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2493,8 +2617,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_le(&d[chan_index], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2510,8 +2636,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_le(&d[chan_index], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2519,8 +2647,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_eq( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
-         STORE( &r[0], 0, chan_index );
+         micro_eq(&d[chan_index], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2550,7 +2680,7 @@ exec_instruction(
       /* src[1] = d[strq]/dx */
       /* src[2] = d[strq]/dy */
       /* src[3] = sampler unit */
-      assert (0);
+      exec_txd(mach, inst);
       break;
 
    case TGSI_OPCODE_TXL:
@@ -2594,13 +2724,8 @@ exec_instruction(
          micro_mul(&r[3], &r[3], &r[1]);
          micro_add(&r[2], &r[2], &r[3]);
          FETCH(&r[3], 0, CHAN_X);
-         micro_add(&r[2], &r[2], &r[3]);
-         if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
-            STORE(&r[2], 0, CHAN_X);
-         }
-         if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
-            STORE(&r[2], 0, CHAN_Z);
-         }
+         micro_add(&d[CHAN_X], &r[2], &r[3]);
+         
       }
       if (IS_CHANNEL_ENABLED(*inst, CHAN_Y) ||
           IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
@@ -2610,13 +2735,20 @@ exec_instruction(
          micro_mul(&r[3], &r[3], &r[1]);
          micro_add(&r[2], &r[2], &r[3]);
          FETCH(&r[3], 0, CHAN_Y);
-         micro_add(&r[2], &r[2], &r[3]);
-         if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
-            STORE(&r[2], 0, CHAN_Y);
-         }
-         if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
-            STORE(&r[2], 0, CHAN_W);
-         }
+         micro_add(&d[CHAN_Y], &r[2], &r[3]);
+         
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+         STORE(&d[CHAN_X], 0, CHAN_X);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+         STORE(&d[CHAN_Y], 0, CHAN_Y);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+         STORE(&d[CHAN_X], 0, CHAN_Z);
+      }
+      if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
+         STORE(&d[CHAN_Y], 0, CHAN_W);
       }
       break;
 
@@ -2701,8 +2833,10 @@ exec_instruction(
    /* TGSI_OPCODE_SGN */
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_sgn( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_sgn(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2711,10 +2845,10 @@ exec_instruction(
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
          FETCH(&r[2], 2, chan_index);
-
-         micro_lt( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2] );
-
-         STORE(&r[0], 0, chan_index);
+         micro_lt(&d[chan_index], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2889,32 +3023,40 @@ exec_instruction(
    case TGSI_OPCODE_CEIL:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_ceil( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_ceil(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_I2F:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_i2f( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_i2f(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_NOT:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_not( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_not(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
    case TGSI_OPCODE_TRUNC:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_trunc( &r[0], &r[0] );
-         STORE( &r[0], 0, chan_index );
+         micro_trunc(&d[chan_index], &r[0]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2922,8 +3064,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_shl( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_shl(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2931,8 +3075,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_ishr( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_ishr(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2940,8 +3086,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_and( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_and(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2949,8 +3097,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_or( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_or(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2962,8 +3112,10 @@ exec_instruction(
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
-         micro_xor( &r[0], &r[0], &r[1] );
-         STORE( &r[0], 0, chan_index );
+         micro_xor(&d[chan_index], &r[0], &r[1]);
+      }
+      FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+         STORE(&d[chan_index], 0, chan_index);
       }
       break;
 
@@ -2992,8 +3144,23 @@ exec_instruction(
       for (chan_index = 0; chan_index < 3; chan_index++) {
          FETCH( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[chan_index], 0, chan_index );
       }
-      STORE( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_Y], 0, CHAN_X );
       ++mach->LoopCounterStackTop;
+      STORE(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X], 0, CHAN_X);
+      /* update LoopMask */
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[0] <= 0.0f) {
+         mach->LoopMask &= ~0x1;
+      }
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[1] <= 0.0f) {
+         mach->LoopMask &= ~0x2;
+      }
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[2] <= 0.0f) {
+         mach->LoopMask &= ~0x4;
+      }
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[3] <= 0.0f) {
+         mach->LoopMask &= ~0x8;
+      }
+      /* TODO: if mach->LoopMask == 0, jump to end of loop */
+      UPDATE_EXEC_MASK(mach);
       /* fall-through (for now) */
    case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
@@ -3007,28 +3174,28 @@ exec_instruction(
 
    case TGSI_OPCODE_ENDFOR:
       assert(mach->LoopCounterStackTop > 0);
-      micro_sub( &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X], 
-                 &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X],
-                 &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
+      micro_sub(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y], 
+                &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y],
+                &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]);
       /* update LoopMask */
-      if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[0] <= 0) {
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[0] <= 0.0f) {
          mach->LoopMask &= ~0x1;
       }
-      if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[1] <= 0 ) {
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[1] <= 0.0f) {
          mach->LoopMask &= ~0x2;
       }
-      if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[2] <= 0 ) {
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[2] <= 0.0f) {
          mach->LoopMask &= ~0x4;
       }
-      if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[3] <= 0 ) {
+      if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[3] <= 0.0f) {
          mach->LoopMask &= ~0x8;
       }
-      micro_add( &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y], 
-                 &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y], 
-                 &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Z]);
+      micro_add(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X], 
+                &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X], 
+                &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Z]);
       assert(mach->LoopLabelStackTop > 0);
       inst = mach->Instructions + mach->LoopLabelStack[mach->LoopLabelStackTop - 1];
-      STORE( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_Y], 0, CHAN_X );
+      STORE(&mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_X], 0, CHAN_X);
       /* Restore ContMask, but don't pop */
       assert(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
@@ -3095,7 +3262,28 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_ENDSUB:
-      /* no-op */
+      /*
+       * XXX: This really should be a no-op. We should never reach this opcode.
+       */
+
+      assert(mach->CallStackTop > 0);
+      mach->CallStackTop--;
+
+      mach->CondStackTop = mach->CallStack[mach->CallStackTop].CondStackTop;
+      mach->CondMask = mach->CondStack[mach->CondStackTop];
+
+      mach->LoopStackTop = mach->CallStack[mach->CallStackTop].LoopStackTop;
+      mach->LoopMask = mach->LoopStack[mach->LoopStackTop];
+
+      mach->ContStackTop = mach->CallStack[mach->CallStackTop].ContStackTop;
+      mach->ContMask = mach->ContStack[mach->ContStackTop];
+
+      assert(mach->FuncStackTop > 0);
+      mach->FuncMask = mach->FuncStack[--mach->FuncStackTop];
+
+      *pc = mach->CallStack[mach->CallStackTop].ReturnAddr;
+
+      UPDATE_EXEC_MASK(mach);
       break;
 
    case TGSI_OPCODE_NOP:
@@ -3106,6 +3294,8 @@ exec_instruction(
    }
 }
 
+#define DEBUG_EXECUTION 0
+
 
 /**
  * Run TGSI interpreter.
@@ -3149,10 +3339,67 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
       exec_declaration( mach, mach->Declarations+i );
    }
 
-   /* execute instructions, until pc is set to -1 */
-   while (pc != -1) {
-      assert(pc < (int) mach->NumInstructions);
-      exec_instruction( mach, mach->Instructions + pc, &pc );
+   {
+#if DEBUG_EXECUTION
+      struct tgsi_exec_vector temps[TGSI_EXEC_NUM_TEMPS + TGSI_EXEC_NUM_TEMP_EXTRAS];
+      struct tgsi_exec_vector outputs[PIPE_MAX_ATTRIBS];
+      uint inst = 1;
+
+      memcpy(temps, mach->Temps, sizeof(temps));
+      memcpy(outputs, mach->Outputs, sizeof(outputs));
+#endif
+
+      /* execute instructions, until pc is set to -1 */
+      while (pc != -1) {
+
+#if DEBUG_EXECUTION
+         uint i;
+
+         tgsi_dump_instruction(&mach->Instructions[pc], inst++);
+#endif
+
+         assert(pc < (int) mach->NumInstructions);
+         exec_instruction(mach, mach->Instructions + pc, &pc);
+
+#if DEBUG_EXECUTION
+         for (i = 0; i < TGSI_EXEC_NUM_TEMPS + TGSI_EXEC_NUM_TEMP_EXTRAS; i++) {
+            if (memcmp(&temps[i], &mach->Temps[i], sizeof(temps[i]))) {
+               uint j;
+
+               memcpy(&temps[i], &mach->Temps[i], sizeof(temps[i]));
+               debug_printf("TEMP[%2u] = ", i);
+               for (j = 0; j < 4; j++) {
+                  if (j > 0) {
+                     debug_printf("           ");
+                  }
+                  debug_printf("(%6f, %6f, %6f, %6f)\n",
+                               temps[i].xyzw[0].f[j],
+                               temps[i].xyzw[1].f[j],
+                               temps[i].xyzw[2].f[j],
+                               temps[i].xyzw[3].f[j]);
+               }
+            }
+         }
+         for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+            if (memcmp(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]))) {
+               uint j;
+
+               memcpy(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]));
+               debug_printf("OUT[%2u] =  ", i);
+               for (j = 0; j < 4; j++) {
+                  if (j > 0) {
+                     debug_printf("           ");
+                  }
+                  debug_printf("{%6f, %6f, %6f, %6f}\n",
+                               outputs[i].xyzw[0].f[j],
+                               outputs[i].xyzw[1].f[j],
+                               outputs[i].xyzw[2].f[j],
+                               outputs[i].xyzw[3].f[j]);
+               }
+            }
+         }
+#endif
+      }
    }
 
 #if 0
@@ -3166,5 +3413,10 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
    }
 #endif
 
+   assert(mach->CondStackTop == 0);
+   assert(mach->LoopStackTop == 0);
+   assert(mach->ContStackTop == 0);
+   assert(mach->CallStackTop == 0);
+
    return ~mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
 }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index 71fd300cbb..afaf5c39c4 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -225,6 +225,7 @@ struct tgsi_exec_machine
    struct tgsi_exec_vector       Outputs[TGSI_MAX_TOTAL_VERTICES];
 
    struct tgsi_exec_vector       *Addrs;
+   struct tgsi_exec_vector       *Predicates;
 
    struct tgsi_sampler           **Samplers;
 
@@ -241,6 +242,7 @@ struct tgsi_exec_machine
    /* FRAGMENT processor only. */
    const struct tgsi_interp_coef *InterpCoefs;
    struct tgsi_exec_vector       QuadPos;
+   float                         Face;    /**< +1 if front facing, -1 if back facing */
 
    /* Conditional execution masks */
    uint CondMask;  /**< For IF/ELSE/ENDIF */
@@ -261,7 +263,7 @@ struct tgsi_exec_machine
    uint LoopLabelStack[TGSI_EXEC_MAX_LOOP_NESTING];
    int LoopLabelStackTop;
 
-   /** Loop counter stack (x = count, y = current, z = step) */
+   /** Loop counter stack (x = index, y = counter, z = step) */
    struct tgsi_exec_vector LoopCounterStack[TGSI_EXEC_MAX_LOOP_NESTING];
    int LoopCounterStackTop;
    
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.c b/src/gallium/auxiliary/tgsi/tgsi_parse.c
index 83f9df1183..9ca2993452 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.c
@@ -175,6 +175,10 @@ tgsi_parse_token(
       copy_token(&inst->Instruction, &token);
       extended = inst->Instruction.Extended;
 
+      if (inst->Instruction.Predicate) {
+         next_token(ctx, &inst->InstructionPredicate);
+      }
+
       while( extended ) {
          struct tgsi_src_register_ext token;
 
@@ -189,10 +193,6 @@ tgsi_parse_token(
             copy_token(&inst->InstructionExtTexture, &token);
             break;
 
-         case TGSI_INSTRUCTION_EXT_TYPE_PREDICATE:
-            copy_token(&inst->InstructionExtPredicate, &token);
-            break;
-
          default:
             assert( 0 );
          }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.h b/src/gallium/auxiliary/tgsi/tgsi_parse.h
index 76f1676d85..cb4772ade8 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.h
@@ -80,9 +80,9 @@ struct tgsi_full_immediate
 struct tgsi_full_instruction
 {
    struct tgsi_instruction             Instruction;
+   struct tgsi_instruction_predicate   InstructionPredicate;
    struct tgsi_instruction_ext_label   InstructionExtLabel;
    struct tgsi_instruction_ext_texture InstructionExtTexture;
-   struct tgsi_instruction_ext_predicate InstructionExtPredicate;
    struct tgsi_full_dst_register       FullDstRegisters[TGSI_FULL_MAX_DST_REGISTERS];
    struct tgsi_full_src_register       FullSrcRegisters[TGSI_FULL_MAX_SRC_REGISTERS];
    uint Flags;  /**< user-defined usage */
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
index 80e6ff2b79..557c885a8a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -47,9 +47,9 @@ union tgsi_any_token {
    struct tgsi_immediate imm;
    union  tgsi_immediate_data imm_data;
    struct tgsi_instruction insn;
+   struct tgsi_instruction_predicate insn_predicate;
    struct tgsi_instruction_ext_label insn_ext_label;
    struct tgsi_instruction_ext_texture insn_ext_texture;
-   struct tgsi_instruction_ext_predicate insn_ext_predicate;
    struct tgsi_src_register src;
    struct tgsi_src_register_ext_mod src_ext_mod;
    struct tgsi_dimension dim;
@@ -72,6 +72,7 @@ struct ureg_tokens {
 #define UREG_MAX_IMMEDIATE 32
 #define UREG_MAX_TEMP 256
 #define UREG_MAX_ADDR 2
+#define UREG_MAX_LOOP 1
 #define UREG_MAX_PRED 1
 
 #define DOMAIN_DECL 0
@@ -123,6 +124,7 @@ struct ureg_program
 
    unsigned nr_addrs;
    unsigned nr_preds;
+   unsigned nr_loops;
    unsigned nr_instructions;
 
    struct ureg_tokens domain[2];
@@ -383,6 +385,7 @@ struct ureg_src ureg_DECL_constant(struct ureg_program *ureg,
       i = ureg->nr_constant_ranges++;
       ureg->constant_range[i].first = index;
       ureg->constant_range[i].last = index;
+      goto out;
    }
 
    /* Collapse all ranges down to one:
@@ -450,6 +453,19 @@ struct ureg_dst ureg_DECL_address( struct ureg_program *ureg )
    return ureg_dst_register( TGSI_FILE_ADDRESS, 0 );
 }
 
+/* Allocate a new loop register.
+ */
+struct ureg_dst
+ureg_DECL_loop(struct ureg_program *ureg)
+{
+   if (ureg->nr_loops < UREG_MAX_LOOP) {
+      return ureg_dst_register(TGSI_FILE_LOOP, ureg->nr_loops++);
+   }
+
+   assert(0);
+   return ureg_dst_register(TGSI_FILE_LOOP, 0);
+}
+
 /* Allocate a new predicate register.
  */
 struct ureg_dst
@@ -694,35 +710,27 @@ ureg_emit_insn(struct ureg_program *ureg,
    validate( opcode, num_dst, num_src );
    
    out = get_tokens( ureg, DOMAIN_INSN, count );
-   out[0].value = 0;
-   out[0].insn.Type = TGSI_TOKEN_TYPE_INSTRUCTION;
-   out[0].insn.NrTokens = 0;
+   out[0].insn = tgsi_default_instruction();
    out[0].insn.Opcode = opcode;
    out[0].insn.Saturate = saturate;
    out[0].insn.NumDstRegs = num_dst;
    out[0].insn.NumSrcRegs = num_src;
-   out[0].insn.Padding = 0;
 
    result.insn_token = ureg->domain[DOMAIN_INSN].count - count;
+   result.extended_token = result.insn_token;
 
    if (predicate) {
-      out[0].insn.Extended = 1;
-      out[1].insn_ext_predicate = tgsi_default_instruction_ext_predicate();
-      out[1].insn_ext_predicate.Negate = pred_negate;
-      out[1].insn_ext_predicate.SwizzleX = pred_swizzle_x;
-      out[1].insn_ext_predicate.SwizzleY = pred_swizzle_y;
-      out[1].insn_ext_predicate.SwizzleZ = pred_swizzle_z;
-      out[1].insn_ext_predicate.SwizzleW = pred_swizzle_w;
-
-      result.extended_token = result.insn_token + 1;
-   } else {
-      out[0].insn.Extended = 0;
-
-      result.extended_token = result.insn_token;
+      out[0].insn.Predicate = 1;
+      out[1].insn_predicate = tgsi_default_instruction_predicate();
+      out[1].insn_predicate.Negate = pred_negate;
+      out[1].insn_predicate.SwizzleX = pred_swizzle_x;
+      out[1].insn_predicate.SwizzleY = pred_swizzle_y;
+      out[1].insn_predicate.SwizzleZ = pred_swizzle_z;
+      out[1].insn_predicate.SwizzleW = pred_swizzle_w;
    }
 
    ureg->nr_instructions++;
-   
+
    return result;
 }
 
@@ -1065,6 +1073,13 @@ static void emit_decls( struct ureg_program *ureg )
                        0, ureg->nr_addrs );
    }
 
+   if (ureg->nr_loops) {
+      emit_decl_range(ureg,
+                      TGSI_FILE_LOOP,
+                      0,
+                      ureg->nr_loops);
+   }
+
    if (ureg->nr_preds) {
       emit_decl_range(ureg,
                       TGSI_FILE_PREDICATE,
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
index e6fad1f0f4..b89f55d5a8 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
@@ -163,6 +163,9 @@ struct ureg_dst
 ureg_DECL_address( struct ureg_program * );
 
 struct ureg_dst
+ureg_DECL_loop( struct ureg_program * );
+
+struct ureg_dst
 ureg_DECL_predicate(struct ureg_program *);
 
 /* Supply an index to the sampler declaration as this is the hook to
diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index 5038642599..5372df5735 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -354,9 +354,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
       texTemp.target = PIPE_TEXTURE_2D;
       texTemp.format = src->format;
       texTemp.last_level = 0;
-      texTemp.width[0] = srcW;
-      texTemp.height[0] = srcH;
-      texTemp.depth[0] = 1;
+      texTemp.width0 = srcW;
+      texTemp.height0 = srcH;
+      texTemp.depth0 = 1;
       pf_get_block(src->format, &texTemp.block);
 
       tex = screen->texture_create(screen, &texTemp);
@@ -389,10 +389,10 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    }
    else {
       pipe_texture_reference(&tex, src->texture);
-      s0 = srcX0 / (float)tex->width[0];
-      s1 = srcX1 / (float)tex->width[0];
-      t0 = srcY0 / (float)tex->height[0];
-      t1 = srcY1 / (float)tex->height[0];
+      s0 = srcX0 / (float)tex->width0;
+      s1 = srcX1 / (float)tex->width0;
+      t0 = srcY0 / (float)tex->height0;
+      t1 = srcY1 / (float)tex->height0;
    }
 
 
@@ -518,13 +518,13 @@ util_blit_pixels_tex(struct blit_state *ctx,
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
 
-   assert(tex->width[0] != 0);
-   assert(tex->height[0] != 0);
+   assert(tex->width0 != 0);
+   assert(tex->height0 != 0);
 
-   s0 = srcX0 / (float)tex->width[0];
-   s1 = srcX1 / (float)tex->width[0];
-   t0 = srcY0 / (float)tex->height[0];
-   t1 = srcY1 / (float)tex->height[0];
+   s0 = srcX0 / (float)tex->width0;
+   s1 = srcX1 / (float)tex->width0;
+   t0 = srcY0 / (float)tex->height0;
+   t1 = srcY1 / (float)tex->height0;
 
    assert(ctx->pipe->screen->is_format_supported(ctx->pipe->screen, dst->format,
                                                  PIPE_TEXTURE_2D,
diff --git a/src/gallium/auxiliary/util/u_cpu_detect.c b/src/gallium/auxiliary/util/u_cpu_detect.c
index facbe69173..a08241971c 100644
--- a/src/gallium/auxiliary/util/u_cpu_detect.c
+++ b/src/gallium/auxiliary/util/u_cpu_detect.c
@@ -80,7 +80,7 @@ static int has_cpuid(void);
 #if defined(PIPE_ARCH_X86)
 
 /* The sigill handlers */
-#if defined(PIPE_OS_LINUX) //&& defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC)
+#if defined(PIPE_OS_LINUX) /*&& defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC)*/
 static void
 sigill_handler_sse(int signal, struct sigcontext sc)
 {
@@ -240,7 +240,7 @@ check_os_katmai_support(void)
       __asm __volatile ("xorps %xmm0, %xmm0");
 #elif defined(PIPE_CC_MSVC)
       __asm {
-          xorps xmm0, xmm0        // executing SSE instruction
+          xorps xmm0, xmm0        /* executing SSE instruction */
       }
 #else
 #error Unsupported compiler
@@ -283,7 +283,7 @@ check_os_katmai_support(void)
     * and therefore to be safe I'm going to leave this test in here.
     */
    if (util_cpu_caps.has_sse) {
-      //      test_os_katmai_exception_support();
+      /* test_os_katmai_exception_support(); */
    }
 
    /* Restore the original signal handlers.
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index f1bf94f17d..b9cc2aa716 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -97,3 +97,13 @@ PIPE_FORMAT_B8G8R8A8_SRGB         , arith , 1, 1, u8  , u8  , u8  , u8  , zyxw,
 PIPE_FORMAT_B8G8R8X8_SRGB         , arith , 1, 1, u8  , u8  , u8  , u8  , zyx1, srgb 
 PIPE_FORMAT_X8UB8UG8SR8S_NORM     , arith , 1, 1, sn8 , sn8 , un8 , x8  , 1zyx, rgb
 PIPE_FORMAT_B6UG5SR5S_NORM        , arith , 1, 1, sn5 , sn5 , un6 ,     , xyz1, rgb
+PIPE_FORMAT_YCBCR                 , yuv   , 2, 1, x32 ,     ,     ,     , xyz1, yuv
+PIPE_FORMAT_YCBCR_REV             , yuv   , 2, 1, x32 ,     ,     ,     , xyz1, yuv
+PIPE_FORMAT_DXT1_RGBA             , dxt   , 4, 4, x64 ,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT1_RGB              , dxt   , 4, 4, x64 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_DXT3_RGBA             , dxt   , 4, 4, x128,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT5_RGBA             , dxt   , 4, 4, x128,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT1_SRGBA            , dxt   , 4, 4, x64 ,     ,     ,     , xyzw, srgb
+PIPE_FORMAT_DXT1_SRGB             , dxt   , 4, 4, x64 ,     ,     ,     , xyz1, srgb
+PIPE_FORMAT_DXT3_SRGBA            , dxt   , 4, 4, x128,     ,     ,     , xyzw, srgb
+PIPE_FORMAT_DXT5_SRGBA            , dxt   , 4, 4, x128,     ,     ,     , xyzw, srgb
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c
index aa823aa218..f67f1e458d 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.c
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.c
@@ -45,6 +45,7 @@
 #include "util/u_draw_quad.h"
 #include "util/u_gen_mipmap.h"
 #include "util/u_simple_shaders.h"
+#include "util/u_math.h"
 
 #include "cso_cache/cso_context.h"
 
@@ -1125,12 +1126,12 @@ make_1d_mipmap(struct gen_mipmap_state *ctx,
       
       srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
                                           PIPE_TRANSFER_READ, 0, 0,
-                                          pt->width[srcLevel],
-                                          pt->height[srcLevel]);
+                                          u_minify(pt->width0, srcLevel),
+                                          u_minify(pt->height0, srcLevel));
       dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
                                           PIPE_TRANSFER_WRITE, 0, 0,
-                                          pt->width[dstLevel],
-                                          pt->height[dstLevel]);
+                                          u_minify(pt->width0, dstLevel),
+                                          u_minify(pt->height0, dstLevel));
 
       srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
       dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1168,12 +1169,12 @@ make_2d_mipmap(struct gen_mipmap_state *ctx,
       
       srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
                                           PIPE_TRANSFER_READ, 0, 0,
-                                          pt->width[srcLevel],
-                                          pt->height[srcLevel]);
+                                          u_minify(pt->width0, srcLevel),
+                                          u_minify(pt->height0, srcLevel));
       dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
                                           PIPE_TRANSFER_WRITE, 0, 0,
-                                          pt->width[dstLevel],
-                                          pt->height[dstLevel]);
+                                          u_minify(pt->width0, dstLevel),
+                                          u_minify(pt->height0, dstLevel));
 
       srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
       dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1213,12 +1214,12 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
       
       srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
                                           PIPE_TRANSFER_READ, 0, 0,
-                                          pt->width[srcLevel],
-                                          pt->height[srcLevel]);
+                                          u_minify(pt->width0, srcLevel),
+                                          u_minify(pt->height0, srcLevel));
       dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
                                           PIPE_TRANSFER_WRITE, 0, 0,
-                                          pt->width[dstLevel],
-                                          pt->height[dstLevel]);
+                                          u_minify(pt->width0, dstLevel),
+                                          u_minify(pt->height0, dstLevel));
 
       srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
       dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1575,8 +1576,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
        * Setup framebuffer / dest surface
        */
       fb.cbufs[0] = surf;
-      fb.width = pt->width[dstLevel];
-      fb.height = pt->height[dstLevel];
+      fb.width = u_minify(pt->width0, dstLevel);
+      fb.height = u_minify(pt->height0, dstLevel);
       cso_set_framebuffer(ctx->cso, &fb);
 
       /*
@@ -1597,8 +1598,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
       offset = set_vertex_data(ctx,
                                pt->target,
                                face,
-                               (float) pt->width[dstLevel],
-                               (float) pt->height[dstLevel]);
+                               (float) u_minify(pt->width0, dstLevel),
+                               (float) u_minify(pt->height0, dstLevel));
 
       util_draw_vertex_buffer(ctx->pipe, 
                               ctx->vbuf,
diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index 75b075f160..7e75702701 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -491,6 +491,43 @@ util_next_power_of_two(unsigned x)
 
 
 /**
+ * Return number of bits set in n.
+ */
+static INLINE unsigned
+util_bitcount(unsigned n)
+{
+#if defined(PIPE_CC_GCC)
+   return __builtin_popcount(n);
+#else
+   /* XXX there are more clever ways of doing this */
+   unsigned bits = 0;
+   while (n) {
+      bits += (n & 1);
+      n = n >> 1;
+   }
+   return bits;
+#endif
+}
+
+
+/**
+ * Reverse byte order of a 32 bit word.
+ */
+static INLINE uint32_t
+util_bswap32(uint32_t n)
+{
+#if defined(PIPE_CC_GCC)
+   return __builtin_bswap32(n);
+#else
+   return (n >> 24) |
+          ((n >> 8) & 0x0000ff00) |
+          ((n << 8) & 0x00ff0000) |
+          (n << 24);
+#endif
+}
+
+
+/**
  * Clamp X to [MIN, MAX].
  * This is a macro to allow float, int, uint, etc. types.
  */
@@ -499,6 +536,9 @@ util_next_power_of_two(unsigned x)
 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
 #define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
 
+#define MIN3( A, B, C ) MIN2( MIN2( A, B ), C )
+#define MAX3( A, B, C ) MAX2( MAX2( A, B ), C )
+
 
 static INLINE int
 align(int value, int alignment)
@@ -507,9 +547,9 @@ align(int value, int alignment)
 }
 
 static INLINE unsigned
-minify(unsigned value)
+u_minify(unsigned value, unsigned levels)
 {
-    return MAX2(1, value >> 1);
+    return MAX2(1, value >> levels);
 }
 
 #ifndef COPY_4V
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index eda883b3b9..9dacc6d83d 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -302,7 +302,10 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
 static INLINE void
 util_pack_color(const float rgba[4], enum pipe_format format, void *dest)
 {
-   ubyte r, g, b, a;
+   ubyte r = 0;
+   ubyte g = 0;
+   ubyte b = 0;
+   ubyte a = 0;
 
    if (pf_size_x(format) <= 8) {
       /* format uses 8-bit components or less */
diff --git a/src/gallium/auxiliary/util/u_surface.c b/src/gallium/auxiliary/util/u_surface.c
index 85e443204e..de8c266db8 100644
--- a/src/gallium/auxiliary/util/u_surface.c
+++ b/src/gallium/auxiliary/util/u_surface.c
@@ -79,9 +79,9 @@ util_create_rgba_surface(struct pipe_screen *screen,
    templ.target = target;
    templ.format = format;
    templ.last_level = 0;
-   templ.width[0] = width;
-   templ.height[0] = height;
-   templ.depth[0] = 1;
+   templ.width0 = width;
+   templ.height0 = height;
+   templ.depth0 = 1;
    pf_get_block(format, &templ.block);
    templ.tex_usage = usage;
 
diff --git a/src/gallium/auxiliary/vl/vl_compositor.c b/src/gallium/auxiliary/vl/vl_compositor.c
index cda6dc134a..592dd17421 100644
--- a/src/gallium/auxiliary/vl/vl_compositor.c
+++ b/src/gallium/auxiliary/vl/vl_compositor.c
@@ -455,8 +455,8 @@ void vl_compositor_render(struct vl_compositor          *compositor,
    assert(dst_area);
    assert(picture_type == PIPE_MPEG12_PICTURE_TYPE_FRAME);
 
-   compositor->fb_state.width = dst_surface->width[0];
-   compositor->fb_state.height = dst_surface->height[0];
+   compositor->fb_state.width = dst_surface->width0;
+   compositor->fb_state.height = dst_surface->height0;
    compositor->fb_state.cbufs[0] = compositor->pipe->screen->get_tex_surface
    (
       compositor->pipe->screen,
@@ -504,12 +504,12 @@ void vl_compositor_render(struct vl_compositor          *compositor,
    vs_consts->dst_trans.z = 0;
    vs_consts->dst_trans.w = 0;
 
-   vs_consts->src_scale.x = src_area->w / (float)src_surface->width[0];
-   vs_consts->src_scale.y = src_area->h / (float)src_surface->height[0];
+   vs_consts->src_scale.x = src_area->w / (float)src_surface->width0;
+   vs_consts->src_scale.y = src_area->h / (float)src_surface->height0;
    vs_consts->src_scale.z = 1;
    vs_consts->src_scale.w = 1;
-   vs_consts->src_trans.x = src_area->x / (float)src_surface->width[0];
-   vs_consts->src_trans.y = src_area->y / (float)src_surface->height[0];
+   vs_consts->src_trans.x = src_area->x / (float)src_surface->width0;
+   vs_consts->src_trans.y = src_area->y / (float)src_surface->height0;
    vs_consts->src_trans.z = 0;
    vs_consts->src_trans.w = 0;
 
diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
index c4ba69817f..1934965995 100644
--- a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
+++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
@@ -681,7 +681,7 @@ xfer_buffers_map(struct vl_mpeg12_mc_renderer *r)
       (
          r->pipe->screen, r->textures.all[i],
          0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0,
-         r->textures.all[i]->width[0], r->textures.all[i]->height[0]
+         r->textures.all[i]->width0, r->textures.all[i]->height0
       );
 
       r->texels[i] = r->pipe->screen->transfer_map(r->pipe->screen, r->tex_transfer[i]);
@@ -835,26 +835,26 @@ init_buffers(struct vl_mpeg12_mc_renderer *r)
    /* TODO: Accomodate HW that can't do this and also for cases when this isn't precise enough */
    template.format = PIPE_FORMAT_R16_SNORM;
    template.last_level = 0;
-   template.width[0] = r->pot_buffers ?
+   template.width0 = r->pot_buffers ?
       util_next_power_of_two(r->picture_width) : r->picture_width;
-   template.height[0] = r->pot_buffers ?
+   template.height0 = r->pot_buffers ?
       util_next_power_of_two(r->picture_height) : r->picture_height;
-   template.depth[0] = 1;
+   template.depth0 = 1;
    pf_get_block(template.format, &template.block);
    template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC;
 
    r->textures.individual.y = r->pipe->screen->texture_create(r->pipe->screen, &template);
 
    if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420) {
-      template.width[0] = r->pot_buffers ?
+      template.width0 = r->pot_buffers ?
          util_next_power_of_two(r->picture_width / 2) :
          r->picture_width / 2;
-      template.height[0] = r->pot_buffers ?
+      template.height0 = r->pot_buffers ?
          util_next_power_of_two(r->picture_height / 2) :
          r->picture_height / 2;
    }
    else if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_422)
-      template.height[0] = r->pot_buffers ?
+      template.height0 = r->pot_buffers ?
          util_next_power_of_two(r->picture_height / 2) :
          r->picture_height / 2;
 
@@ -1283,8 +1283,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
       PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
    );
 
-   vs_consts->denorm.x = r->surface->width[0];
-   vs_consts->denorm.y = r->surface->height[0];
+   vs_consts->denorm.x = r->surface->width0;
+   vs_consts->denorm.y = r->surface->height0;
 
    pipe_buffer_unmap(r->pipe->screen, r->vs_const_buf.buffer);
 
@@ -1633,8 +1633,8 @@ vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer
       renderer->past = past;
       renderer->future = future;
       renderer->fence = fence;
-      renderer->surface_tex_inv_size.x = 1.0f / surface->width[0];
-      renderer->surface_tex_inv_size.y = 1.0f / surface->height[0];
+      renderer->surface_tex_inv_size.x = 1.0f / surface->width0;
+      renderer->surface_tex_inv_size.y = 1.0f / surface->height0;
    }
 
    while (num_macroblocks) {
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index d9bae5b641..5b87286d4c 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_emit.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c
@@ -27,6 +27,7 @@
 
 #include "pipe/p_inlines.h"
 #include "util/u_memory.h"
+#include "util/u_math.h"
 #include "cell_context.h"
 #include "cell_gen_fragment.h"
 #include "cell_state.h"
@@ -299,9 +300,9 @@ cell_emit_state(struct cell_context *cell)
                for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
                   texture->start[level] = (ct->mapped +
                                            ct->level_offset[level]);
-                  texture->width[level] = ct->base.width[level];
-                  texture->height[level] = ct->base.height[level];
-                  texture->depth[level] = ct->base.depth[level];
+                  texture->width[level] = u_minify(ct->base.width0, level);
+                  texture->height[level] = u_minify(ct->base.height0, level);
+                  texture->depth[level] = u_minify(ct->base.depth0, level);
                }
                texture->target = ct->base.target;
             }
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index ae4c61efb3..e6b8a87045 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -49,9 +49,9 @@ cell_texture_layout(struct cell_texture *ct)
 {
    struct pipe_texture *pt = &ct->base;
    unsigned level;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
 
    ct->buffer_size = 0;
 
@@ -65,9 +65,6 @@ cell_texture_layout(struct cell_texture *ct)
       w_tile = align(width, TILE_SIZE);
       h_tile = align(height, TILE_SIZE);
 
-      pt->width[level] = width;
-      pt->height[level] = height;
-      pt->depth[level] = depth;
       pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w_tile);  
       pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h_tile);  
 
@@ -83,9 +80,9 @@ cell_texture_layout(struct cell_texture *ct)
 
       ct->buffer_size += size;
 
-      width = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
    }
 }
 
@@ -276,8 +273,8 @@ cell_get_tex_surface(struct pipe_screen *screen,
       pipe_reference_init(&ps->reference, 1);
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
+      ps->width = u_minify(pt->width0, level);
+      ps->height = u_minify(pt->height0, level);
       ps->offset = ct->level_offset[level];
       /* XXX may need to override usage flags (see sp_texture.c) */
       ps->usage = usage;
@@ -386,8 +383,8 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
    struct pipe_texture *pt = transfer->texture;
    struct cell_texture *ct = cell_texture(pt);
    const uint level = ctrans->base.level;
-   const uint texWidth = pt->width[level];
-   const uint texHeight = pt->height[level];
+   const uint texWidth = u_minify(pt->width0, level);
+   const uint texHeight = u_minify(pt->height0, level);
    const uint stride = ct->stride[level];
    unsigned size;
 
@@ -440,8 +437,8 @@ cell_transfer_unmap(struct pipe_screen *screen,
    struct pipe_texture *pt = transfer->texture;
    struct cell_texture *ct = cell_texture(pt);
    const uint level = ctrans->base.level;
-   const uint texWidth = pt->width[level];
-   const uint texHeight = pt->height[level];
+   const uint texWidth = u_minify(pt->width0, level);
+   const uint texHeight = u_minify(pt->height0, level);
    const uint stride = ct->stride[level];
 
    if (!ct->mapped) {
diff --git a/src/gallium/drivers/i915/i915_debug.c b/src/gallium/drivers/i915/i915_debug.c
index e6640e587b..c6e6d6fd31 100644
--- a/src/gallium/drivers/i915/i915_debug.c
+++ b/src/gallium/drivers/i915/i915_debug.c
@@ -851,6 +851,7 @@ static boolean i915_debug_packet( struct debug_stream *stream )
       default:
 	 return debug(stream, "", 0);
       }
+      break;
    default:
       assert(0);
       return 0;
diff --git a/src/gallium/drivers/i915/i915_state.c b/src/gallium/drivers/i915/i915_state.c
index 7d48e6e84d..71f00bc346 100644
--- a/src/gallium/drivers/i915/i915_state.c
+++ b/src/gallium/drivers/i915/i915_state.c
@@ -58,8 +58,10 @@ translate_wrap_mode(unsigned wrap)
       return TEXCOORDMODE_CLAMP_EDGE;
    case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
       return TEXCOORDMODE_CLAMP_BORDER;
-//   case PIPE_TEX_WRAP_MIRRORED_REPEAT:
-//      return TEXCOORDMODE_MIRROR;
+/*
+   case PIPE_TEX_WRAP_MIRRORED_REPEAT:
+      return TEXCOORDMODE_MIRROR;
+*/
    default:
       return TEXCOORDMODE_WRAP;
    }
diff --git a/src/gallium/drivers/i915/i915_state_sampler.c b/src/gallium/drivers/i915/i915_state_sampler.c
index c5e9084d12..cbac4175c8 100644
--- a/src/gallium/drivers/i915/i915_state_sampler.c
+++ b/src/gallium/drivers/i915/i915_state_sampler.c
@@ -231,7 +231,7 @@ i915_update_texture(struct i915_context *i915,
 {
    const struct pipe_texture *pt = &tex->base;
    uint format, pitch;
-   const uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0];
+   const uint width = pt->width0, height = pt->height0, depth = pt->depth0;
    const uint num_levels = pt->last_level;
    unsigned max_lod = num_levels * 4;
    unsigned tiled = MS3_USE_FENCE_REGS;
diff --git a/src/gallium/drivers/i915/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c
index 286c9ace8e..c7b86dd4c5 100644
--- a/src/gallium/drivers/i915/i915_texture.c
+++ b/src/gallium/drivers/i915/i915_texture.c
@@ -105,10 +105,6 @@ i915_miptree_set_level_info(struct i915_texture *tex,
 
    assert(level < PIPE_MAX_TEXTURE_LEVELS);
 
-   pt->width[level] = w;
-   pt->height[level] = h;
-   pt->depth[level] = d;
-   
    pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w);
    pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h);
 
@@ -168,16 +164,16 @@ i915_scanout_layout(struct i915_texture *tex)
       return FALSE;
 
    i915_miptree_set_level_info(tex, 0, 1,
-                               tex->base.width[0],
-                               tex->base.height[0],
+                               tex->base.width0,
+                               tex->base.height0,
                                1);
    i915_miptree_set_image_offset(tex, 0, 0, 0, 0);
 
-   if (tex->base.width[0] >= 240) {
+   if (tex->base.width0 >= 240) {
       tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size);
       tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8);
       tex->hw_tiled = INTEL_TILE_X;
-   } else if (tex->base.width[0] == 64 && tex->base.height[0] == 64) {
+   } else if (tex->base.width0 == 64 && tex->base.height0 == 64) {
       tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size);
       tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8);
    } else {
@@ -185,7 +181,7 @@ i915_scanout_layout(struct i915_texture *tex)
    }
 
    debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
-      tex->base.width[0], tex->base.height[0], pt->block.size,
+      tex->base.width0, tex->base.height0, pt->block.size,
       tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy);
 
    return TRUE;
@@ -203,12 +199,12 @@ i915_display_target_layout(struct i915_texture *tex)
       return FALSE;
 
    /* fallback to normal textures for small textures */
-   if (tex->base.width[0] < 240)
+   if (tex->base.width0 < 240)
       return FALSE;
 
    i915_miptree_set_level_info(tex, 0, 1,
-                               tex->base.width[0],
-                               tex->base.height[0],
+                               tex->base.width0,
+                               tex->base.height0,
                                1);
    i915_miptree_set_image_offset(tex, 0, 0, 0, 0);
 
@@ -217,7 +213,7 @@ i915_display_target_layout(struct i915_texture *tex)
    tex->hw_tiled = INTEL_TILE_X;
 
    debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
-      tex->base.width[0], tex->base.height[0], pt->block.size,
+      tex->base.width0, tex->base.height0, pt->block.size,
       tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy);
 
    return TRUE;
@@ -228,8 +224,8 @@ i915_miptree_layout_2d(struct i915_texture *tex)
 {
    struct pipe_texture *pt = &tex->base;
    unsigned level;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
 
@@ -254,8 +250,8 @@ i915_miptree_layout_2d(struct i915_texture *tex)
 
       tex->total_nblocksy += nblocksy;
 
-      width = minify(width);
-      height = minify(height);
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
       nblocksx = pf_get_nblocksx(&pt->block, width);
       nblocksy = pf_get_nblocksy(&pt->block, height);
    }
@@ -267,9 +263,9 @@ i915_miptree_layout_3d(struct i915_texture *tex)
    struct pipe_texture *pt = &tex->base;
    unsigned level;
 
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
    unsigned stack_nblocksy = 0;
@@ -285,36 +281,34 @@ i915_miptree_layout_3d(struct i915_texture *tex)
 
       stack_nblocksy += MAX2(2, nblocksy);
 
-      width = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
       nblocksx = pf_get_nblocksx(&pt->block, width);
       nblocksy = pf_get_nblocksy(&pt->block, height);
    }
 
    /* Fixup depth image_offsets: 
     */
-   depth = pt->depth[0];
    for (level = 0; level <= pt->last_level; level++) {
       unsigned i;
       for (i = 0; i < depth; i++) 
          i915_miptree_set_image_offset(tex, level, i, 0, i * stack_nblocksy);
 
-      depth = minify(depth);
+      depth = u_minify(depth, 1);
    }
 
    /* Multiply slice size by texture depth for total size.  It's
     * remarkable how wasteful of memory the i915 texture layouts
     * are.  They are largely fixed in the i945.
     */
-   tex->total_nblocksy = stack_nblocksy * pt->depth[0];
+   tex->total_nblocksy = stack_nblocksy * pt->depth0;
 }
 
 static void
 i915_miptree_layout_cube(struct i915_texture *tex)
 {
    struct pipe_texture *pt = &tex->base;
-   unsigned width = pt->width[0], height = pt->height[0];
+   unsigned width = pt->width0, height = pt->height0;
    const unsigned nblocks = pt->nblocksx[0];
    unsigned level;
    unsigned face;
@@ -383,8 +377,8 @@ i945_miptree_layout_2d(struct i915_texture *tex)
    unsigned level;
    unsigned x = 0;
    unsigned y = 0;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
 
@@ -407,8 +401,8 @@ i945_miptree_layout_2d(struct i915_texture *tex)
     */
    if (pt->last_level > 0) {
       unsigned mip1_nblocksx 
-         = align(pf_get_nblocksx(&pt->block, minify(width)), align_x)
-         + pf_get_nblocksx(&pt->block, minify(minify(width)));
+         = align(pf_get_nblocksx(&pt->block, u_minify(width, 1)), align_x)
+         + pf_get_nblocksx(&pt->block, u_minify(width, 2));
 
       if (mip1_nblocksx > nblocksx)
          tex->stride = mip1_nblocksx * pt->block.size;
@@ -439,8 +433,8 @@ i945_miptree_layout_2d(struct i915_texture *tex)
          y += nblocksy;
       }
 
-      width  = minify(width);
-      height = minify(height);
+      width  = u_minify(width, 1);
+      height = u_minify(height, 1);
       nblocksx = pf_get_nblocksx(&pt->block, width);
       nblocksy = pf_get_nblocksy(&pt->block, height);
    }
@@ -450,9 +444,9 @@ static void
 i945_miptree_layout_3d(struct i915_texture *tex)
 {
    struct pipe_texture *pt = &tex->base;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
    unsigned pack_x_pitch, pack_x_nr;
@@ -495,9 +489,9 @@ i945_miptree_layout_3d(struct i915_texture *tex)
          pack_y_pitch >>= 1;
       }
 
-      width = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
       nblocksx = pf_get_nblocksx(&pt->block, width);
       nblocksy = pf_get_nblocksy(&pt->block, height);
    }
@@ -511,11 +505,11 @@ i945_miptree_layout_cube(struct i915_texture *tex)
 
    const unsigned nblocks = pt->nblocksx[0];
    unsigned face;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
 
    /*
-   printf("%s %i, %i\n", __FUNCTION__, pt->width[0], pt->height[0]);
+   printf("%s %i, %i\n", __FUNCTION__, pt->width0, pt->height0);
    */
 
    assert(width == height); /* cubemap images are square */
@@ -651,8 +645,8 @@ i915_texture_create(struct pipe_screen *screen,
    pipe_reference_init(&tex->base.reference, 1);
    tex->base.screen = screen;
 
-   tex->base.nblocksx[0] = pf_get_nblocksx(&tex->base.block, tex->base.width[0]);
-   tex->base.nblocksy[0] = pf_get_nblocksy(&tex->base.block, tex->base.height[0]);
+   tex->base.nblocksx[0] = pf_get_nblocksx(&tex->base.block, tex->base.width0);
+   tex->base.nblocksy[0] = pf_get_nblocksy(&tex->base.block, tex->base.height0);
    
    if (is->is_i945) {
       if (!i945_miptree_layout(tex))
@@ -667,7 +661,7 @@ i915_texture_create(struct pipe_screen *screen,
 
 
    /* for scanouts and cursors, cursors arn't scanouts */
-   if (templat->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY && templat->width[0] != 64)
+   if (templat->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY && templat->width0 != 64)
       buf_usage = INTEL_NEW_SCANOUT;
    else
       buf_usage = INTEL_NEW_TEXTURE;
@@ -710,7 +704,7 @@ i915_texture_blanket(struct pipe_screen * screen,
    /* Only supports one type */
    if (base->target != PIPE_TEXTURE_2D ||
        base->last_level != 0 ||
-       base->depth[0] != 1) {
+       base->depth0 != 1) {
       return NULL;
    }
 
@@ -724,7 +718,7 @@ i915_texture_blanket(struct pipe_screen * screen,
 
    tex->stride = stride[0];
 
-   i915_miptree_set_level_info(tex, 0, 1, base->width[0], base->height[0], 1);
+   i915_miptree_set_level_info(tex, 0, 1, base->width0, base->height0, 1);
    i915_miptree_set_image_offset(tex, 0, 0, 0, 0);
 
    pipe_buffer_reference(&tex->buffer, buffer);
@@ -788,8 +782,8 @@ i915_get_tex_surface(struct pipe_screen *screen,
       pipe_reference_init(&ps->reference, 1);
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
+      ps->width = u_minify(pt->width0, level);
+      ps->height = u_minify(pt->height0, level);
       ps->offset = offset;
       ps->usage = flags;
    }
@@ -919,7 +913,7 @@ i915_texture_blanket_intel(struct pipe_screen *screen,
    /* Only supports one type */
    if (base->target != PIPE_TEXTURE_2D ||
        base->last_level != 0 ||
-       base->depth[0] != 1) {
+       base->depth0 != 1) {
       return NULL;
    }
 
@@ -933,7 +927,7 @@ i915_texture_blanket_intel(struct pipe_screen *screen,
 
    tex->stride = stride;
 
-   i915_miptree_set_level_info(tex, 0, 1, base->width[0], base->height[0], 1);
+   i915_miptree_set_level_info(tex, 0, 1, base->width0, base->height0, 1);
    i915_miptree_set_image_offset(tex, 0, 0, 0, 0);
 
    tex->buffer = buffer;
diff --git a/src/gallium/drivers/i915/intel_winsys.h b/src/gallium/drivers/i915/intel_winsys.h
index 2c8dc63f3f..c6bf6e6f7f 100644
--- a/src/gallium/drivers/i915/intel_winsys.h
+++ b/src/gallium/drivers/i915/intel_winsys.h
@@ -42,21 +42,21 @@ enum intel_buffer_usage
    INTEL_USAGE_2D_TARGET = 0x04,
    INTEL_USAGE_2D_SOURCE = 0x08,
    /* use on vertex */
-   INTEL_USAGE_VERTEX    = 0x10,
+   INTEL_USAGE_VERTEX    = 0x10
 };
 
 enum intel_buffer_type
 {
    INTEL_NEW_TEXTURE,
    INTEL_NEW_SCANOUT, /**< a texture used for scanning out from */
-   INTEL_NEW_VERTEX,
+   INTEL_NEW_VERTEX
 };
 
 enum intel_buffer_tile
 {
    INTEL_TILE_NONE,
    INTEL_TILE_X,
-   INTEL_TILE_Y,
+   INTEL_TILE_Y
 };
 
 struct intel_batchbuffer {
diff --git a/src/gallium/drivers/identity/id_objects.c b/src/gallium/drivers/identity/id_objects.c
index e893e59940..bc9bc7121d 100644
--- a/src/gallium/drivers/identity/id_objects.c
+++ b/src/gallium/drivers/identity/id_objects.c
@@ -180,3 +180,42 @@ identity_transfer_destroy(struct identity_transfer *id_transfer)
    screen->tex_transfer_destroy(id_transfer->transfer);
    FREE(id_transfer);
 }
+
+struct pipe_video_surface *
+identity_video_surface_create(struct identity_screen *id_screen,
+                              struct pipe_video_surface *video_surface)
+{
+   struct identity_video_surface *id_video_surface;
+
+   if (!video_surface) {
+      goto error;
+   }
+
+   assert(video_surface->screen == id_screen->screen);
+
+   id_video_surface = CALLOC_STRUCT(identity_video_surface);
+   if (!id_video_surface) {
+      goto error;
+   }
+
+   memcpy(&id_video_surface->base,
+          video_surface,
+          sizeof(struct pipe_video_surface));
+
+   pipe_reference_init(&id_video_surface->base.reference, 1);
+   id_video_surface->base.screen = &id_screen->base;
+   id_video_surface->video_surface = video_surface;
+
+   return &id_video_surface->base;
+
+error:
+   pipe_video_surface_reference(&video_surface, NULL);
+   return NULL;
+}
+
+void
+identity_video_surface_destroy(struct identity_video_surface *id_video_surface)
+{
+   pipe_video_surface_reference(&id_video_surface->video_surface, NULL);
+   FREE(id_video_surface);
+}
diff --git a/src/gallium/drivers/identity/id_objects.h b/src/gallium/drivers/identity/id_objects.h
index ce58faa3c7..77cc719079 100644
--- a/src/gallium/drivers/identity/id_objects.h
+++ b/src/gallium/drivers/identity/id_objects.h
@@ -31,6 +31,7 @@
 
 #include "pipe/p_compiler.h"
 #include "pipe/p_state.h"
+#include "pipe/p_video_state.h"
 
 #include "id_screen.h"
 
@@ -67,6 +68,14 @@ struct identity_transfer
 };
 
 
+struct identity_video_surface
+{
+   struct pipe_video_surface base;
+
+   struct pipe_video_surface *video_surface;
+};
+
+
 static INLINE struct identity_buffer *
 identity_buffer(struct pipe_buffer *_buffer)
 {
@@ -103,6 +112,15 @@ identity_transfer(struct pipe_transfer *_transfer)
    return (struct identity_transfer *)_transfer;
 }
 
+static INLINE struct identity_video_surface *
+identity_video_surface(struct pipe_video_surface *_video_surface)
+{
+   if (!_video_surface) {
+      return NULL;
+   }
+   (void)identity_screen(_video_surface->screen);
+   return (struct identity_video_surface *)_video_surface;
+}
 
 static INLINE struct pipe_buffer *
 identity_buffer_unwrap(struct pipe_buffer *_buffer)
@@ -165,5 +183,12 @@ identity_transfer_create(struct identity_texture *id_texture,
 void
 identity_transfer_destroy(struct identity_transfer *id_transfer);
 
+struct pipe_video_surface *
+identity_video_surface_create(struct identity_screen *id_screen,
+                              struct pipe_video_surface *video_surface);
+
+void
+identity_video_surface_destroy(struct identity_video_surface *id_video_surface);
+
 
 #endif /* ID_OBJECTS_H */
diff --git a/src/gallium/drivers/identity/id_public.h b/src/gallium/drivers/identity/id_public.h
index cac14cfd60..3d2862eaa0 100644
--- a/src/gallium/drivers/identity/id_public.h
+++ b/src/gallium/drivers/identity/id_public.h
@@ -37,4 +37,4 @@ identity_screen_create(struct pipe_screen *screen);
 struct pipe_context *
 identity_context_create(struct pipe_screen *screen, struct pipe_context *pipe);
 
-#endif /* PT_PUBLIC_H */
+#endif /* ID_PUBLIC_H */
diff --git a/src/gallium/drivers/identity/id_screen.c b/src/gallium/drivers/identity/id_screen.c
index 26439637d0..53eae3ef54 100644
--- a/src/gallium/drivers/identity/id_screen.c
+++ b/src/gallium/drivers/identity/id_screen.c
@@ -379,6 +379,33 @@ identity_screen_buffer_destroy(struct pipe_buffer *_buffer)
    identity_buffer_destroy(identity_buffer(_buffer));
 }
 
+static struct pipe_video_surface *
+identity_screen_video_surface_create(struct pipe_screen *_screen,
+                                     enum pipe_video_chroma_format chroma_format,
+                                     unsigned width,
+                                     unsigned height)
+{
+   struct identity_screen *id_screen = identity_screen(_screen);
+   struct pipe_screen *screen = id_screen->screen;
+   struct pipe_video_surface *result;
+
+   result = screen->video_surface_create(screen,
+                                         chroma_format,
+                                         width,
+                                         height);
+
+   if (result) {
+      return identity_video_surface_create(id_screen, result);
+   }
+   return NULL;
+}
+
+static void
+identity_screen_video_surface_destroy(struct pipe_video_surface *_vsfc)
+{
+   identity_video_surface_destroy(identity_video_surface(_vsfc));
+}
+
 static void
 identity_screen_flush_frontbuffer(struct pipe_screen *_screen,
                                   struct pipe_surface *_surface,
@@ -472,6 +499,12 @@ identity_screen_create(struct pipe_screen *screen)
    if (screen->buffer_unmap)
       id_screen->base.buffer_unmap = identity_screen_buffer_unmap;
    id_screen->base.buffer_destroy = identity_screen_buffer_destroy;
+   if (screen->video_surface_create) {
+      id_screen->base.video_surface_create = identity_screen_video_surface_create;
+   }
+   if (screen->video_surface_destroy) {
+      id_screen->base.video_surface_destroy = identity_screen_video_surface_destroy;
+   }
    id_screen->base.flush_frontbuffer = identity_screen_flush_frontbuffer;
    id_screen->base.fence_reference = identity_screen_fence_reference;
    id_screen->base.fence_signalled = identity_screen_fence_signalled;
diff --git a/src/gallium/drivers/llvmpipe/README b/src/gallium/drivers/llvmpipe/README
index 89d08834a3..0c3f00fd58 100644
--- a/src/gallium/drivers/llvmpipe/README
+++ b/src/gallium/drivers/llvmpipe/README
@@ -51,21 +51,22 @@ Requirements
 
  - Linux
  
- - udis86, http://udis86.sourceforge.net/ . Use my repository, which decodes
-   opcodes not yet supported by upstream.
+ - A x86 or amd64 processor.  64bit mode is preferred.
  
-     git clone git://people.freedesktop.org/~jrfonseca/udis86
-     cd udis86
-     ./configure --with-pic
-     make
-     sudo make install
+   Support for sse2 is strongly encouraged.  Support for ssse3, and sse4.1 will
+   yield the most efficient code.  The less features the CPU has the more
+   likely is that you ran into underperforming, buggy, or incomplete code.  
+   
+   See /proc/cpuinfo to know what your CPU supports.
+ 
+ - LLVM 2.5 or greater. LLVM 2.6 is preferred.
  
- - LLVM 2.5. On Debian based distributions do:
+   On Debian based distributions do:
  
      aptitude install llvm-dev
 
-   There is a typo in one of the llvm-dev 2.5 headers, that causes compilation
-   errors in the debug build:
+   There is a typo in one of the llvm 2.5 headers, that may cause compilation
+   errors. To fix it apply the change:
 
      --- /usr/include/llvm-c/Core.h.orig	2009-08-10 15:38:54.000000000 +0100
      +++ /usr/include/llvm-c/Core.h	2009-08-10 15:38:25.000000000 +0100
@@ -79,12 +80,17 @@ Requirements
           #endif
           return reinterpret_cast<T**>(Vals);
  
- - A x86 or amd64 processor with support for sse2, sse3, and sse4.1 SIMD
-   instructions. This is necessary because we emit several SSE intrinsics for
-   convenience. See /proc/cpuinfo to know what your CPU supports.
- 
- - scons
+ - scons (optional)
 
+ - udis86, http://udis86.sourceforge.net/ (optional):
+ 
+     git clone git://udis86.git.sourceforge.net/gitroot/udis86/udis86
+     cd udis86
+     ./autogen.sh
+     ./configure --with-pic
+     make
+     sudo make install
+ 
 
 Building
 ========
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.c b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
index 98ec1cb1b9..d438c0e63d 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_depth.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
@@ -90,7 +90,7 @@ lp_depth_type(const struct util_format_description *format_desc,
 
    if(format_desc->channel[swizzle].type == UTIL_FORMAT_TYPE_FLOAT) {
       type.floating = TRUE;
-      assert(swizzle = 0);
+      assert(swizzle == 0);
       assert(format_desc->channel[swizzle].size == format_desc->block.bits);
    }
    else if(format_desc->channel[swizzle].type == UTIL_FORMAT_TYPE_UNSIGNED) {
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample.c b/src/gallium/drivers/llvmpipe/lp_bld_sample.c
index 4d272bea87..af70ddc6ab 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_sample.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_sample.c
@@ -59,9 +59,9 @@ lp_sampler_static_state(struct lp_sampler_static_state *state,
 
    state->format            = texture->format;
    state->target            = texture->target;
-   state->pot_width         = util_is_pot(texture->width[0]);
-   state->pot_height        = util_is_pot(texture->height[0]);
-   state->pot_depth         = util_is_pot(texture->depth[0]);
+   state->pot_width         = util_is_pot(texture->width0);
+   state->pot_height        = util_is_pot(texture->height0);
+   state->pot_depth         = util_is_pot(texture->depth0);
 
    state->wrap_s            = sampler->wrap_s;
    state->wrap_t            = sampler->wrap_t;
diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c
index 13535dd638..bce3baec16 100644
--- a/src/gallium/drivers/llvmpipe/lp_jit.c
+++ b/src/gallium/drivers/llvmpipe/lp_jit.c
@@ -154,6 +154,7 @@ lp_jit_screen_init(struct llvmpipe_screen *screen)
 #if 0
    /* For simulating less capable machines */
    util_cpu_caps.has_sse3 = 0;
+   util_cpu_caps.has_ssse3 = 0;
    util_cpu_caps.has_sse4_1 = 0;
 #endif
 
@@ -167,7 +168,7 @@ lp_jit_screen_init(struct llvmpipe_screen *screen)
    if (LLVMCreateJITCompiler(&screen->engine, screen->provider, 1, &error)) {
       _debug_printf("%s\n", error);
       LLVMDisposeMessage(error);
-      abort();
+      assert(0);
    }
 
    screen->target = LLVMGetExecutionEngineTargetData(screen->engine);
diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index 0518927458..0fb133486a 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -27,6 +27,7 @@
 
 
 #include "util/u_memory.h"
+#include "util/u_format.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_screen.h"
 
@@ -131,17 +132,17 @@ llvmpipe_is_format_supported( struct pipe_screen *_screen,
 {
    struct llvmpipe_screen *screen = llvmpipe_screen(_screen);
    struct llvmpipe_winsys *winsys = screen->winsys;
+   const struct util_format_description *format_desc;
+
+   format_desc = util_format_description(format);
+   if(!format_desc)
+      return FALSE;
 
    assert(target == PIPE_TEXTURE_1D ||
           target == PIPE_TEXTURE_2D ||
           target == PIPE_TEXTURE_3D ||
           target == PIPE_TEXTURE_CUBE);
 
-   if(format == PIPE_FORMAT_Z16_UNORM)
-      return FALSE;
-   if(format == PIPE_FORMAT_S8_UNORM)
-      return FALSE;
-
    switch(format) {
    case PIPE_FORMAT_DXT1_RGB:
    case PIPE_FORMAT_DXT1_RGBA:
@@ -152,8 +153,51 @@ llvmpipe_is_format_supported( struct pipe_screen *_screen,
       break;
    }
 
-   if(tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET)
-      return winsys->is_displaytarget_format_supported(winsys, format);
+   if(tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+      if(format_desc->block.width != 1 ||
+         format_desc->block.height != 1)
+         return FALSE;
+
+      if(format_desc->layout != UTIL_FORMAT_LAYOUT_SCALAR &&
+         format_desc->layout != UTIL_FORMAT_LAYOUT_ARITH &&
+         format_desc->layout != UTIL_FORMAT_LAYOUT_ARRAY)
+         return FALSE;
+
+      if(format_desc->colorspace != UTIL_FORMAT_COLORSPACE_RGB &&
+         format_desc->colorspace != UTIL_FORMAT_COLORSPACE_SRGB)
+         return FALSE;
+   }
+
+   if(tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) {
+      if(!winsys->is_displaytarget_format_supported(winsys, format))
+         return FALSE;
+   }
+
+   if(tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL) {
+      if(format_desc->colorspace != UTIL_FORMAT_COLORSPACE_ZS)
+         return FALSE;
+
+      /* FIXME: Temporary restriction. See lp_state_fs.c. */
+      if(format_desc->block.bits != 32)
+         return FALSE;
+   }
+
+   /* FIXME: Temporary restrictions. See lp_bld_sample_soa.c */
+   if(tex_usage & PIPE_TEXTURE_USAGE_SAMPLER) {
+      if(format_desc->block.width != 1 ||
+         format_desc->block.height != 1)
+         return FALSE;
+
+      if(format_desc->layout != UTIL_FORMAT_LAYOUT_SCALAR &&
+         format_desc->layout != UTIL_FORMAT_LAYOUT_ARITH &&
+         format_desc->layout != UTIL_FORMAT_LAYOUT_ARRAY)
+         return FALSE;
+
+      if(format_desc->colorspace != UTIL_FORMAT_COLORSPACE_RGB &&
+         format_desc->colorspace != UTIL_FORMAT_COLORSPACE_SRGB &&
+         format_desc->colorspace != UTIL_FORMAT_COLORSPACE_ZS)
+         return FALSE;
+   }
 
    return TRUE;
 }
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 2e9aa9fffe..ee0f69b2af 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -148,6 +148,20 @@ generate_depth(LLVMBuilderRef builder,
    format_desc = util_format_description(key->zsbuf_format);
    assert(format_desc);
 
+   /*
+    * Depths are expected to be between 0 and 1, even if they are stored in
+    * floats. Setting these bits here will ensure that the lp_build_conv() call
+    * below won't try to unnecessarily clamp the incoming values.
+    */
+   if(src_type.floating) {
+      src_type.sign = FALSE;
+      src_type.norm = TRUE;
+   }
+   else {
+      assert(!src_type.sign);
+      assert(src_type.norm);
+   }
+
    /* Pick the depth type. */
    dst_type = lp_depth_type(format_desc, src_type.width*src_type.length);
 
@@ -155,14 +169,11 @@ generate_depth(LLVMBuilderRef builder,
    assert(dst_type.width == src_type.width);
    assert(dst_type.length == src_type.length);
 
-#if 1
-   src = lp_build_clamped_float_to_unsigned_norm(builder,
-                                                 src_type,
-                                                 dst_type.width,
-                                                 src);
-#else
    lp_build_conv(builder, src_type, dst_type, &src, 1, &src, 1);
-#endif
+
+   dst_ptr = LLVMBuildBitCast(builder,
+                              dst_ptr,
+                              LLVMPointerType(lp_build_vec_type(dst_type), 0), "");
 
    lp_build_depth_test(builder,
                        &key->depth,
@@ -611,10 +622,12 @@ generate_fragment(struct llvmpipe_context *lp,
     * Translate the LLVM IR into machine code.
     */
 
+#ifdef DEBUG
    if(LLVMVerifyFunction(variant->function, LLVMPrintMessageAction)) {
       LLVMDumpValue(variant->function);
-      abort();
+      assert(0);
    }
+#endif
 
    LLVMRunFunctionPassManager(screen->pass, variant->function);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_state_sampler.c b/src/gallium/drivers/llvmpipe/lp_state_sampler.c
index c69d90c723..8333805a3f 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_sampler.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_sampler.c
@@ -102,8 +102,8 @@ llvmpipe_set_sampler_textures(struct pipe_context *pipe,
       if(tex) {
          struct llvmpipe_texture *lp_tex = llvmpipe_texture(tex);
          struct lp_jit_texture *jit_tex = &llvmpipe->jit_context.textures[i];
-         jit_tex->width = tex->width[0];
-         jit_tex->height = tex->height[0];
+         jit_tex->width = tex->width0;
+         jit_tex->height = tex->height0;
          jit_tex->stride = lp_tex->stride[0];
          if(!lp_tex->dt)
             jit_tex->data = lp_tex->data;
diff --git a/src/gallium/drivers/llvmpipe/lp_state_vs.c b/src/gallium/drivers/llvmpipe/lp_state_vs.c
index 15c3029614..8a761648e7 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_vs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_vs.c
@@ -92,5 +92,6 @@ llvmpipe_delete_vs_state(struct pipe_context *pipe, void *vs)
       (struct lp_vertex_shader *)vs;
 
    draw_delete_vertex_shader(llvmpipe->draw, state->draw_data);
+   FREE( (void *)state->shader.tokens );
    FREE( state );
 }
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_cache.c b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
index 773e848242..c7c4143bc6 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_cache.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
@@ -36,6 +36,7 @@
 #include "util/u_memory.h"
 #include "util/u_tile.h"
 #include "util/u_format.h"
+#include "util/u_math.h"
 #include "lp_context.h"
 #include "lp_surface.h"
 #include "lp_texture.h"
@@ -270,8 +271,8 @@ lp_find_cached_tex_tile(struct llvmpipe_tex_tile_cache *tc,
                                      addr.bits.level, 
                                      addr.bits.z, 
                                      PIPE_TRANSFER_READ, 0, 0,
-                                     tc->texture->width[addr.bits.level],
-                                     tc->texture->height[addr.bits.level]);
+                                     u_minify(tc->texture->width0, addr.bits.level),
+                                     u_minify(tc->texture->height0, addr.bits.level));
 
          tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans);
 
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
index a1365a045f..0d01c07fb5 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
@@ -544,7 +544,7 @@ compute_lambda(struct tgsi_sampler *tgsi_sampler,
       float dsdy = s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT];
       dsdx = fabsf(dsdx);
       dsdy = fabsf(dsdy);
-      rho = MAX2(dsdx, dsdy) * texture->width[0];
+      rho = MAX2(dsdx, dsdy) * texture->width0;
    }
    if (t) {
       float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT];
@@ -552,7 +552,7 @@ compute_lambda(struct tgsi_sampler *tgsi_sampler,
       float max;
       dtdx = fabsf(dtdx);
       dtdy = fabsf(dtdy);
-      max = MAX2(dtdx, dtdy) * texture->height[0];
+      max = MAX2(dtdx, dtdy) * texture->height0;
       rho = MAX2(rho, max);
    }
    if (p) {
@@ -561,7 +561,7 @@ compute_lambda(struct tgsi_sampler *tgsi_sampler,
       float max;
       dpdx = fabsf(dpdx);
       dpdy = fabsf(dpdy);
-      max = MAX2(dpdx, dpdy) * texture->depth[0];
+      max = MAX2(dpdx, dpdy) * texture->depth0;
       rho = MAX2(rho, max);
    }
 
@@ -726,9 +726,9 @@ get_texel(const struct tgsi_sampler *tgsi_sampler,
    const struct pipe_texture *texture = samp->texture;
    const struct pipe_sampler_state *sampler = samp->sampler;
 
-   if (x < 0 || x >= (int) texture->width[level] ||
-       y < 0 || y >= (int) texture->height[level] ||
-       z < 0 || z >= (int) texture->depth[level]) {
+   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
+       y < 0 || y >= (int) u_minify(texture->height0, level) ||
+       z < 0 || z >= (int) u_minify(texture->depth0, level)) {
       rgba[0][j] = sampler->border_color[0];
       rgba[1][j] = sampler->border_color[1];
       rgba[2][j] = sampler->border_color[2];
@@ -1093,8 +1093,8 @@ lp_get_samples_2d_common(struct tgsi_sampler *tgsi_sampler,
 
    assert(sampler->normalized_coords);
 
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
 
@@ -1250,9 +1250,9 @@ lp_get_samples_3d(struct tgsi_sampler *tgsi_sampler,
 
    assert(sampler->normalized_coords);
 
-   width = texture->width[level0];
-   height = texture->height[level0];
-   depth = texture->depth[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
+   depth = u_minify(texture->depth0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1394,8 +1394,8 @@ lp_get_samples_rect(struct tgsi_sampler *tgsi_sampler,
    /* texture RECTS cannot be mipmapped */
    assert(level0 == level1);
 
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
 
@@ -1513,8 +1513,8 @@ lp_get_samples(struct tgsi_sampler *tgsi_sampler,
 
    /* Do this elsewhere: 
     */
-   samp->xpot = util_unsigned_logbase2( samp->texture->width[0] );
-   samp->ypot = util_unsigned_logbase2( samp->texture->height[0] );
+   samp->xpot = util_unsigned_logbase2( samp->texture->width0 );
+   samp->ypot = util_unsigned_logbase2( samp->texture->height0 );
 
    /* Try to hook in a faster sampler.  Ultimately we'll have to
     * code-generate these.  Luckily most of this looks like it is
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index a00f2495df..65d62fd072 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -57,9 +57,9 @@ llvmpipe_texture_layout(struct llvmpipe_screen *screen,
 {
    struct pipe_texture *pt = &lpt->base;
    unsigned level;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
 
    unsigned buffer_size = 0;
 
@@ -68,9 +68,6 @@ llvmpipe_texture_layout(struct llvmpipe_screen *screen,
    for (level = 0; level <= pt->last_level; level++) {
       unsigned nblocksx, nblocksy;
 
-      pt->width[level] = width;
-      pt->height[level] = height;
-      pt->depth[level] = depth;
       pt->nblocksx[level] = pf_get_nblocksx(&pt->block, width);  
       pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height);
 
@@ -87,9 +84,9 @@ llvmpipe_texture_layout(struct llvmpipe_screen *screen,
                       ((pt->target == PIPE_TEXTURE_CUBE) ? 6 : depth) *
                       lpt->stride[level]);
 
-      width  = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width  = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
    }
 
    lpt->data = align_malloc(buffer_size, 16);
@@ -104,13 +101,13 @@ llvmpipe_displaytarget_layout(struct llvmpipe_screen *screen,
    struct llvmpipe_winsys *winsys = screen->winsys;
 
    pf_get_block(lpt->base.format, &lpt->base.block);
-   lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width[0]);  
-   lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height[0]);  
+   lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width0);  
+   lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height0);  
 
    lpt->dt = winsys->displaytarget_create(winsys,
                                           lpt->base.format,
-                                          lpt->base.width[0],
-                                          lpt->base.height[0],
+                                          lpt->base.width0,
+                                          lpt->base.height0,
                                           16,
                                           &lpt->stride[0] );
 
@@ -172,7 +169,7 @@ llvmpipe_texture_blanket(struct pipe_screen * screen,
    /* Only supports one type */
    if (base->target != PIPE_TEXTURE_2D ||
        base->last_level != 0 ||
-       base->depth[0] != 1) {
+       base->depth0 != 1) {
       return NULL;
    }
 
@@ -183,8 +180,8 @@ llvmpipe_texture_blanket(struct pipe_screen * screen,
    lpt->base = *base;
    pipe_reference_init(&lpt->base.reference, 1);
    lpt->base.screen = screen;
-   lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width[0]);  
-   lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height[0]);  
+   lpt->base.nblocksx[0] = pf_get_nblocksx(&lpt->base.block, lpt->base.width0);  
+   lpt->base.nblocksy[0] = pf_get_nblocksy(&lpt->base.block, lpt->base.height0);  
    lpt->stride[0] = stride[0];
 
    pipe_buffer_reference(&lpt->buffer, buffer);
@@ -229,8 +226,8 @@ llvmpipe_get_tex_surface(struct pipe_screen *screen,
       pipe_reference_init(&ps->reference, 1);
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
+      ps->width = u_minify(pt->width0, level);
+      ps->height = u_minify(pt->height0, level);
       ps->offset = lpt->level_offset[level];
       ps->usage = usage;
 
diff --git a/src/gallium/drivers/nv04/nv04_fragtex.c b/src/gallium/drivers/nv04/nv04_fragtex.c
index 21f990fd53..0cce71ad1d 100644
--- a/src/gallium/drivers/nv04/nv04_fragtex.c
+++ b/src/gallium/drivers/nv04/nv04_fragtex.c
@@ -57,8 +57,8 @@ nv04_fragtex_build(struct nv04_context *nv04, int unit)
 		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER
 		| nv04_fragtex_format(pt->format)
 		| ( (pt->last_level + 1) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT )
-		| ( log2i(pt->width[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT )
-		| ( log2i(pt->height[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT )
+		| ( log2i(pt->width0) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT )
+		| ( log2i(pt->height0) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT )
 		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE
 		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE
 		;
diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c
index 93f752faec..4fd72c82e6 100644
--- a/src/gallium/drivers/nv04/nv04_miptree.c
+++ b/src/gallium/drivers/nv04/nv04_miptree.c
@@ -1,6 +1,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv04_context.h"
 #include "nv04_screen.h"
@@ -9,31 +10,29 @@ static void
 nv04_miptree_layout(struct nv04_miptree *nv04mt)
 {
 	struct pipe_texture *pt = &nv04mt->base;
-	uint width = pt->width[0], height = pt->height[0];
+	uint width = pt->width0, height = pt->height0;
 	uint offset = 0;
 	int nr_faces, l;
 
 	nr_faces = 1;
 
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
 
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 		
-		nv04mt->level[l].pitch = pt->width[0];
+		nv04mt->level[l].pitch = pt->width0;
 		nv04mt->level[l].pitch = (nv04mt->level[l].pitch + 63) & ~63;
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
+		width  = u_minify(width, 1);
+		height = u_minify(height, 1);
 	}
 
 	for (l = 0; l <= pt->last_level; l++) {
 
 		nv04mt->level[l].image_offset = 
 			CALLOC(nr_faces, sizeof(unsigned));
-		offset += nv04mt->level[l].pitch * pt->height[l];
+		offset += nv04mt->level[l].pitch * u_minify(pt->height0, l);
 	}
 
 	nv04mt->total_size = offset;
@@ -75,7 +74,7 @@ nv04_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv04_miptree);
@@ -120,8 +119,8 @@ nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
diff --git a/src/gallium/drivers/nv04/nv04_transfer.c b/src/gallium/drivers/nv04/nv04_transfer.c
index 6618660743..e6456429f4 100644
--- a/src/gallium/drivers/nv04/nv04_transfer.c
+++ b/src/gallium/drivers/nv04/nv04_transfer.c
@@ -2,6 +2,7 @@
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv04_context.h"
 #include "nv04_screen.h"
@@ -20,9 +21,9 @@ nv04_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
+	template->width0 = u_minify(pt->width0, level);
+	template->height0 = u_minify(pt->height0, level);
+	template->depth0 = 1;
 	template->block = pt->block;
 	template->nblocksx[0] = pt->nblocksx[level];
 	template->nblocksy[0] = pt->nblocksx[level];
diff --git a/src/gallium/drivers/nv10/nv10_fragtex.c b/src/gallium/drivers/nv10/nv10_fragtex.c
index 27f2f87584..906fdfeeb9 100644
--- a/src/gallium/drivers/nv10/nv10_fragtex.c
+++ b/src/gallium/drivers/nv10/nv10_fragtex.c
@@ -62,9 +62,9 @@ nv10_fragtex_build(struct nv10_context *nv10, int unit)
 
 	txf  = tf->format << 8;
 	txf |= (pt->last_level + 1) << 16;
-	txf |= log2i(pt->width[0]) << 20;
-	txf |= log2i(pt->height[0]) << 24;
-	txf |= log2i(pt->depth[0]) << 28;
+	txf |= log2i(pt->width0) << 20;
+	txf |= log2i(pt->height0) << 24;
+	txf |= log2i(pt->depth0) << 28;
 	txf |= 8;
 
 	switch (pt->target) {
@@ -89,7 +89,7 @@ nv10_fragtex_build(struct nv10_context *nv10, int unit)
 	OUT_RING  (0x40000000); /* enable */
 	OUT_RING  (txs);
 	OUT_RING  (ps->filt | 0x2000 /* magic */);
-	OUT_RING  ((pt->width[0] << 16) | pt->height[0]);
+	OUT_RING  ((pt->width0 << 16) | pt->height0);
 	OUT_RING  (ps->bcol);
 #endif
 }
diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c
index 34e3c2ebd7..b2a6c59b74 100644
--- a/src/gallium/drivers/nv10/nv10_miptree.c
+++ b/src/gallium/drivers/nv10/nv10_miptree.c
@@ -1,6 +1,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv10_context.h"
 #include "nv10_screen.h"
@@ -10,7 +11,7 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt)
 {
 	struct pipe_texture *pt = &nv10mt->base;
 	boolean swizzled = FALSE;
-	uint width = pt->width[0], height = pt->height[0];
+	uint width = pt->width0, height = pt->height0;
 	uint offset = 0;
 	int nr_faces, l, f;
 
@@ -21,8 +22,7 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt)
 	}
 	
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
+
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
@@ -35,15 +35,15 @@ nv10_miptree_layout(struct nv10_miptree *nv10mt)
 		nv10mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
+		width  = u_minify(width, 1);
+		height = u_minify(height, 1);
 
 	}
 
 	for (f = 0; f < nr_faces; f++) {
 		for (l = 0; l <= pt->last_level; l++) {
 			nv10mt->level[l].image_offset[f] = offset;
-			offset += nv10mt->level[l].pitch * pt->height[l];
+			offset += nv10mt->level[l].pitch * u_minify(pt->height0, l);
 		}
 	}
 
@@ -58,7 +58,7 @@ nv10_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv10_miptree);
@@ -133,8 +133,8 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
diff --git a/src/gallium/drivers/nv10/nv10_transfer.c b/src/gallium/drivers/nv10/nv10_transfer.c
index 8feb85e4bd..ec54297ab0 100644
--- a/src/gallium/drivers/nv10/nv10_transfer.c
+++ b/src/gallium/drivers/nv10/nv10_transfer.c
@@ -2,6 +2,7 @@
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv10_context.h"
 #include "nv10_screen.h"
@@ -20,9 +21,9 @@ nv10_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
+	template->width0 = u_minify(pt->width0, level);
+	template->height0 = u_minify(pt->height0, level);
+	template->depth0 = 1;
 	template->block = pt->block;
 	template->nblocksx[0] = pt->nblocksx[level];
 	template->nblocksy[0] = pt->nblocksx[level];
diff --git a/src/gallium/drivers/nv20/nv20_fragtex.c b/src/gallium/drivers/nv20/nv20_fragtex.c
index 495a7be912..2db4a4015a 100644
--- a/src/gallium/drivers/nv20/nv20_fragtex.c
+++ b/src/gallium/drivers/nv20/nv20_fragtex.c
@@ -62,9 +62,9 @@ nv20_fragtex_build(struct nv20_context *nv20, int unit)
 
 	txf  = tf->format << 8;
 	txf |= (pt->last_level + 1) << 16;
-	txf |= log2i(pt->width[0]) << 20;
-	txf |= log2i(pt->height[0]) << 24;
-	txf |= log2i(pt->depth[0]) << 28;
+	txf |= log2i(pt->width0) << 20;
+	txf |= log2i(pt->height0) << 24;
+	txf |= log2i(pt->depth0) << 28;
 	txf |= 8;
 
 	switch (pt->target) {
@@ -89,7 +89,7 @@ nv20_fragtex_build(struct nv20_context *nv20, int unit)
 	OUT_RING  (0x40000000); /* enable */
 	OUT_RING  (txs);
 	OUT_RING  (ps->filt | 0x2000 /* magic */);
-	OUT_RING  ((pt->width[0] << 16) | pt->height[0]);
+	OUT_RING  ((pt->width0 << 16) | pt->height0);
 	OUT_RING  (ps->bcol);
 #endif
 }
diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c
index 185fbf53e0..554e28e47d 100644
--- a/src/gallium/drivers/nv20/nv20_miptree.c
+++ b/src/gallium/drivers/nv20/nv20_miptree.c
@@ -1,6 +1,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv20_context.h"
 #include "nv20_screen.h"
@@ -9,7 +10,7 @@ static void
 nv20_miptree_layout(struct nv20_miptree *nv20mt)
 {
 	struct pipe_texture *pt = &nv20mt->base;
-	uint width = pt->width[0], height = pt->height[0];
+	uint width = pt->width0, height = pt->height0;
 	uint offset = 0;
 	int nr_faces, l, f;
 	uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER |
@@ -25,21 +26,19 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt)
 	}
 	
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
 		if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
-			nv20mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64);
+			nv20mt->level[l].pitch = align(pt->width0 * pt->block.size, 64);
 		else
-			nv20mt->level[l].pitch = pt->width[l] * pt->block.size;
+			nv20mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size;
 
 		nv20mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
+		width  = u_minify(width, 1);
+		height = u_minify(height, 1);
 	}
 
 	for (f = 0; f < nr_faces; f++) {
@@ -47,14 +46,14 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt)
 			nv20mt->level[l].image_offset[f] = offset;
 
 			if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) &&
-			    pt->width[l + 1] > 1 && pt->height[l + 1] > 1)
-				offset += align(nv20mt->level[l].pitch * pt->height[l], 64);
+			    u_minify(pt->width0, l + 1) > 1 && u_minify(pt->height0, l + 1) > 1)
+				offset += align(nv20mt->level[l].pitch * u_minify(pt->height0, l), 64);
 			else
-				offset += nv20mt->level[l].pitch * pt->height[l];
+				offset += nv20mt->level[l].pitch * u_minify(pt->height0, l);
 		}
 
 		nv20mt->level[l].image_offset[f] = offset;
-		offset += nv20mt->level[l].pitch * pt->height[l];
+		offset += nv20mt->level[l].pitch * u_minify(pt->height0, l);
 	}
 
 	nv20mt->total_size = offset;
@@ -68,7 +67,7 @@ nv20_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv20_miptree);
@@ -100,8 +99,8 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 	mt->base.screen = screen;
 
 	/* Swizzled textures must be POT */
-	if (pt->width[0] & (pt->width[0] - 1) ||
-	    pt->height[0] & (pt->height[0] - 1))
+	if (pt->width0 & (pt->width0 - 1) ||
+	    pt->height0 & (pt->height0 - 1))
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
@@ -167,8 +166,8 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c
index 81b4f1a917..87b5c14a3c 100644
--- a/src/gallium/drivers/nv20/nv20_transfer.c
+++ b/src/gallium/drivers/nv20/nv20_transfer.c
@@ -2,6 +2,7 @@
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv20_context.h"
 #include "nv20_screen.h"
@@ -20,9 +21,9 @@ nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
+	template->width0 = u_minify(pt->width0, level);
+	template->height0 = u_minify(pt->height0, level);
+	template->depth0 = 1;
 	template->block = pt->block;
 	template->nblocksx[0] = pt->nblocksx[level];
 	template->nblocksy[0] = pt->nblocksx[level];
diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c
index cc0385426c..0ce702d6f8 100644
--- a/src/gallium/drivers/nv30/nv30_fragprog.c
+++ b/src/gallium/drivers/nv30/nv30_fragprog.c
@@ -4,6 +4,7 @@
 #include "pipe/p_inlines.h"
 
 #include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_dump.h"
 #include "tgsi/tgsi_parse.h"
 #include "tgsi/tgsi_util.h"
 
@@ -131,7 +132,7 @@ emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src)
 				sizeof(uint32_t) * 4);
 		}
 
-		sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT);	
+		sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT);
 		break;
 	case NV30SR_NONE:
 		sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT);
@@ -666,7 +667,7 @@ nv30_fragprog_prepare(struct nv30_fpc *fpc)
 		{
 			struct tgsi_full_immediate *imm;
 			float vals[4];
-			
+
 			imm = &p.FullToken.FullImmediate;
 			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
 			assert(fpc->nr_imm < MAX_IMM);
@@ -754,7 +755,7 @@ nv30_fragprog_translate(struct nv30_context *nv30,
 	fp->insn[fpc->inst_offset + 1] = 0x00000000;
 	fp->insn[fpc->inst_offset + 2] = 0x00000000;
 	fp->insn[fpc->inst_offset + 3] = 0x00000000;
-	
+
 	fp->translated = TRUE;
 	fp->on_hw = FALSE;
 out_err:
@@ -838,7 +839,7 @@ nv30_fragprog_validate(struct nv30_context *nv30)
 update_constants:
 	if (fp->nr_consts) {
 		float *map;
-		
+
 		map = pipe_buffer_map(pscreen, constbuf,
 				      PIPE_BUFFER_USAGE_CPU_READ);
 		for (i = 0; i < fp->nr_consts; i++) {
diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c
index dca760cae6..b3293ee700 100644
--- a/src/gallium/drivers/nv30/nv30_fragtex.c
+++ b/src/gallium/drivers/nv30/nv30_fragtex.c
@@ -74,9 +74,9 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit)
 
 	txf  = tf->format;
 	txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0);
-	txf |= log2i(pt->width[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT;
-	txf |= log2i(pt->height[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT;
-	txf |= log2i(pt->depth[0]) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT;
+	txf |= log2i(pt->width0) << NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT;
+	txf |= log2i(pt->height0) << NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT;
+	txf |= log2i(pt->depth0) << NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT;
 	txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000;
 
 	switch (pt->target) {
@@ -115,8 +115,8 @@ nv30_fragtex_build(struct nv30_context *nv30, int unit)
 	so_data  (so, NV34TCL_TX_ENABLE_ENABLE | ps->en);
 	so_data  (so, txs);
 	so_data  (so, ps->filt | 0x2000 /*voodoo*/);
-	so_data  (so, (pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) |
-		       pt->height[0]);
+	so_data  (so, (pt->width0 << NV34TCL_TX_NPOT_SIZE_W_SHIFT) |
+		       pt->height0);
 	so_data  (so, ps->bcol);
 
 	return so;
diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c
index 280696d450..b4c306d127 100644
--- a/src/gallium/drivers/nv30/nv30_miptree.c
+++ b/src/gallium/drivers/nv30/nv30_miptree.c
@@ -1,6 +1,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv30_context.h"
 
@@ -8,7 +9,7 @@ static void
 nv30_miptree_layout(struct nv30_miptree *nv30mt)
 {
 	struct pipe_texture *pt = &nv30mt->base;
-	uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0];
+	uint width = pt->width0, height = pt->height0, depth = pt->depth0;
 	uint offset = 0;
 	int nr_faces, l, f;
 	uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER |
@@ -21,29 +22,26 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt)
 		nr_faces = 6;
 	} else
 	if (pt->target == PIPE_TEXTURE_3D) {
-		nr_faces = pt->depth[0];
+		nr_faces = pt->depth0;
 	} else {
 		nr_faces = 1;
 	}
 
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
-		pt->depth[l] = depth;
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
 		if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
-			nv30mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64);
+			nv30mt->level[l].pitch = align(pt->width0 * pt->block.size, 64);
 		else
-			nv30mt->level[l].pitch = pt->width[l] * pt->block.size;
+			nv30mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size;
 
 		nv30mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
-		depth  = MAX2(1, depth  >> 1);
+		width  = u_minify(width, 1);
+		height = u_minify(height, 1);
+		depth  = u_minify(depth, 1);
 	}
 
 	for (f = 0; f < nr_faces; f++) {
@@ -51,14 +49,14 @@ nv30_miptree_layout(struct nv30_miptree *nv30mt)
 			nv30mt->level[l].image_offset[f] = offset;
 
 			if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) &&
-			    pt->width[l + 1] > 1 && pt->height[l + 1] > 1)
-				offset += align(nv30mt->level[l].pitch * pt->height[l], 64);
+			    u_minify(pt->width0, l + 1) > 1 && u_minify(pt->height0, l + 1) > 1)
+				offset += align(nv30mt->level[l].pitch * u_minify(pt->height0, l), 64);
 			else
-				offset += nv30mt->level[l].pitch * pt->height[l];
+				offset += nv30mt->level[l].pitch * u_minify(pt->height0, l);
 		}
 
 		nv30mt->level[l].image_offset[f] = offset;
-		offset += nv30mt->level[l].pitch * pt->height[l];
+		offset += nv30mt->level[l].pitch * u_minify(pt->height0, l);
 	}
 
 	nv30mt->total_size = offset;
@@ -79,8 +77,8 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 	mt->base.screen = pscreen;
 
 	/* Swizzled textures must be POT */
-	if (pt->width[0] & (pt->width[0] - 1) ||
-	    pt->height[0] & (pt->height[0] - 1))
+	if (pt->width0 & (pt->width0 - 1) ||
+	    pt->height0 & (pt->height0 - 1))
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
@@ -134,7 +132,7 @@ nv30_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv30_miptree);
@@ -182,8 +180,8 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
index 221ae1b5f8..7cd36902eb 100644
--- a/src/gallium/drivers/nv30/nv30_screen.c
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -116,7 +116,10 @@ nv30_screen_surface_format_supported(struct pipe_screen *pscreen,
 		case PIPE_FORMAT_Z24X8_UNORM:
 			return TRUE;
 		case PIPE_FORMAT_Z16_UNORM:
-			return (front->format == PIPE_FORMAT_R5G6B5_UNORM);
+			if (front) {
+				return (front->format == PIPE_FORMAT_R5G6B5_UNORM);
+			}
+			return TRUE;
 		default:
 			break;
 		}
diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c
index 98011decf7..5e429b4d85 100644
--- a/src/gallium/drivers/nv30/nv30_transfer.c
+++ b/src/gallium/drivers/nv30/nv30_transfer.c
@@ -2,6 +2,7 @@
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv30_context.h"
 #include "nv30_screen.h"
@@ -20,9 +21,9 @@ nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
+	template->width0 = u_minify(pt->width0, level);
+	template->height0 = u_minify(pt->height0, level);
+	template->depth0 = 1;
 	template->block = pt->block;
 	template->nblocksx[0] = pt->nblocksx[level];
 	template->nblocksy[0] = pt->nblocksx[level];
diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c
index e2ec57564d..44abc84596 100644
--- a/src/gallium/drivers/nv40/nv40_fragtex.c
+++ b/src/gallium/drivers/nv40/nv40_fragtex.c
@@ -117,11 +117,11 @@ nv40_fragtex_build(struct nv40_context *nv40, int unit)
 	so_data  (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en);
 	so_data  (so, txs);
 	so_data  (so, ps->filt | tf->sign | 0x2000 /*voodoo*/);
-	so_data  (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) |
-		       pt->height[0]);
+	so_data  (so, (pt->width0 << NV40TCL_TEX_SIZE0_W_SHIFT) |
+		       pt->height0);
 	so_data  (so, ps->bcol);
 	so_method(so, nv40->screen->curie, NV40TCL_TEX_SIZE1(unit), 1);
-	so_data  (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp);
+	so_data  (so, (pt->depth0 << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp);
 
 	return so;
 }
diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c
index 465dd3b069..f73bedff6d 100644
--- a/src/gallium/drivers/nv40/nv40_miptree.c
+++ b/src/gallium/drivers/nv40/nv40_miptree.c
@@ -1,6 +1,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv40_context.h"
 
@@ -8,7 +9,7 @@ static void
 nv40_miptree_layout(struct nv40_miptree *mt)
 {
 	struct pipe_texture *pt = &mt->base;
-	uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0];
+	uint width = pt->width0, height = pt->height0, depth = pt->depth0;
 	uint offset = 0;
 	int nr_faces, l, f;
 	uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER |
@@ -21,29 +22,26 @@ nv40_miptree_layout(struct nv40_miptree *mt)
 		nr_faces = 6;
 	} else
 	if (pt->target == PIPE_TEXTURE_3D) {
-		nr_faces = pt->depth[0];
+		nr_faces = pt->depth0;
 	} else {
 		nr_faces = 1;
 	}
 
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
-		pt->depth[l] = depth;
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
 		if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
-			mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64);
+			mt->level[l].pitch = align(pt->width0 * pt->block.size, 64);
 		else
-			mt->level[l].pitch = pt->width[l] * pt->block.size;
+			mt->level[l].pitch = u_minify(pt->width0, l) * pt->block.size;
 
 		mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
-		depth  = MAX2(1, depth  >> 1);
+		width  = u_minify(width, 1);
+		height = u_minify(height, 1);
+		depth  = u_minify(depth, 1);
 	}
 
 	for (f = 0; f < nr_faces; f++) {
@@ -51,14 +49,14 @@ nv40_miptree_layout(struct nv40_miptree *mt)
 			mt->level[l].image_offset[f] = offset;
 
 			if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) &&
-			    pt->width[l + 1] > 1 && pt->height[l + 1] > 1)
-				offset += align(mt->level[l].pitch * pt->height[l], 64);
+			    u_minify(pt->width0, l + 1) > 1 && u_minify(pt->height0, l + 1) > 1)
+				offset += align(mt->level[l].pitch * u_minify(pt->height0, l), 64);
 			else
-				offset += mt->level[l].pitch * pt->height[l];
+				offset += mt->level[l].pitch * u_minify(pt->height0, l);
 		}
 
 		mt->level[l].image_offset[f] = offset;
-		offset += mt->level[l].pitch * pt->height[l];
+		offset += mt->level[l].pitch * u_minify(pt->height0, l);
 	}
 
 	mt->total_size = offset;
@@ -79,8 +77,8 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 	mt->base.screen = pscreen;
 
 	/* Swizzled textures must be POT */
-	if (pt->width[0] & (pt->width[0] - 1) ||
-	    pt->height[0] & (pt->height[0] - 1))
+	if (pt->width0 & (pt->width0 - 1) ||
+	    pt->height0 & (pt->height0 - 1))
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
@@ -128,7 +126,7 @@ nv40_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv40_miptree);
@@ -176,8 +174,8 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c
index 92caee6f38..36e253c96f 100644
--- a/src/gallium/drivers/nv40/nv40_transfer.c
+++ b/src/gallium/drivers/nv40/nv40_transfer.c
@@ -2,6 +2,7 @@
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv40_context.h"
 #include "nv40_screen.h"
@@ -20,9 +21,9 @@ nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
+	template->width0 = u_minify(pt->width0, level);
+	template->height0 = u_minify(pt->height0, level);
+	template->depth0 = 1;
 	template->block = pt->block;
 	template->nblocksx[0] = pt->nblocksx[level];
 	template->nblocksy[0] = pt->nblocksx[level];
diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c
index 9c20c5cc28..3d58746793 100644
--- a/src/gallium/drivers/nv50/nv50_miptree.c
+++ b/src/gallium/drivers/nv50/nv50_miptree.c
@@ -61,8 +61,8 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp)
 	struct nouveau_device *dev = nouveau_screen(pscreen)->device;
 	struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree);
 	struct pipe_texture *pt = &mt->base.base;
-	unsigned width = tmp->width[0], height = tmp->height[0];
-	unsigned depth = tmp->depth[0], image_alignment;
+	unsigned width = tmp->width0, height = tmp->height0;
+	unsigned depth = tmp->depth0, image_alignment;
 	uint32_t tile_flags;
 	int ret, i, l;
 
@@ -92,9 +92,6 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp)
 	for (l = 0; l <= pt->last_level; l++) {
 		struct nv50_miptree_level *lvl = &mt->level[l];
 
-		pt->width[l] = width;
-		pt->height[l] = height;
-		pt->depth[l] = depth;
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
@@ -102,9 +99,9 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp)
 		lvl->pitch = align(pt->nblocksx[l] * pt->block.size, 64);
 		lvl->tile_mode = get_tile_mode(pt->nblocksy[l], depth);
 
-		width = MAX2(1, width >> 1);
-		height = MAX2(1, height >> 1);
-		depth = MAX2(1, depth >> 1);
+		width = u_minify(width, 1);
+		height = u_minify(height, 1);
+		depth = u_minify(depth, 1);
 	}
 
 	image_alignment  = get_tile_height(mt->level[0].tile_mode) * 64;
@@ -122,7 +119,7 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp)
 
 			size  = lvl->pitch;
 			size *= align(pt->nblocksy[l], tile_h);
-			size *= align(pt->depth[l], tile_d);
+			size *= align(u_minify(pt->depth0, l), tile_d);
 
 			lvl->image_offset[i] = mt->total_size;
 
@@ -151,7 +148,7 @@ nv50_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv50_miptree);
@@ -202,8 +199,8 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ps->texture, pt);
 	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
+	ps->width = u_minify(pt->width0, level);
+	ps->height = u_minify(pt->height0, level);
 	ps->usage = flags;
 	pipe_reference_init(&ps->reference, 1);
 	ps->face = face;
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index bf50982dd1..855079f293 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -2914,7 +2914,7 @@ nv50_fp_move_results(struct nv50_pc *pc)
 static void
 nv50_program_fixup_insns(struct nv50_pc *pc)
 {
-	struct nv50_program_exec *e, *prev = NULL, **bra_list;
+	struct nv50_program_exec *e, **bra_list;
 	unsigned i, n, pos;
 
 	bra_list = CALLOC(pc->p->exec_size, sizeof(struct nv50_program_exec *));
@@ -2926,6 +2926,16 @@ nv50_program_fixup_insns(struct nv50_pc *pc)
 		if (e->param.index >= 0 && !e->param.mask)
 			bra_list[n++] = e;
 
+	/* last instruction must be long so it can have the exit bit set */
+	if (!is_long(pc->p->exec_tail))
+		convert_to_long(pc, pc->p->exec_tail);
+	/* set exit bit */
+	pc->p->exec_tail->inst[1] |= 1;
+
+	/* !immd on exit insn simultaneously means !join */
+	assert(!is_immd(pc->p->exec_head));
+	assert(!is_immd(pc->p->exec_tail));
+
 	/* Make sure we don't have any single 32 bit instructions. */
 	for (e = pc->p->exec_head, pos = 0; e; e = e->next) {
 		pos += is_long(e) ? 2 : 1;
@@ -2937,23 +2947,8 @@ nv50_program_fixup_insns(struct nv50_pc *pc)
 			convert_to_long(pc, e);
 			++pos;
 		}
-		if (e->next)
-			prev = e;
 	}
 
-	assert(!is_immd(pc->p->exec_head));
-	assert(!is_immd(pc->p->exec_tail));
-
-	/* last instruction must be long so it can have the end bit set */
-	if (!is_long(pc->p->exec_tail)) {
-		convert_to_long(pc, pc->p->exec_tail);
-		if (prev)
-			convert_to_long(pc, prev);
-	}
-	assert(!(pc->p->exec_tail->inst[1] & 2));
-	/* set the end-bit */
-	pc->p->exec_tail->inst[1] |= 1;
-
 	FREE(bra_list);
 }
 
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index 799d2758fe..19b8ef07b5 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -285,7 +285,7 @@ nv50_state_validate(struct nv50_context *nv50)
 		so = so_new(33, 0);
 		so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_PATTERN(0), 32);
 		for (i = 0; i < 32; i++)
-			so_data(so, nv50->stipple.stipple[i]);
+			so_data(so, util_bswap32(nv50->stipple.stipple[i]));
 		so_ref(so, &nv50->state.stipple);
 		so_ref(NULL, &so);
 	}
diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c
index 2813f54477..417d367942 100644
--- a/src/gallium/drivers/nv50/nv50_tex.c
+++ b/src/gallium/drivers/nv50/nv50_tex.c
@@ -131,9 +131,9 @@ nv50_tex_construct(struct nv50_context *nv50, struct nouveau_stateobj *so,
 		 NOUVEAU_BO_RD, 0, 0);
 	so_data (so, mode);
 	so_data (so, 0x00300000);
-	so_data (so, mt->base.base.width[0] | (1 << 31));
+	so_data (so, mt->base.base.width0 | (1 << 31));
 	so_data (so, (mt->base.base.last_level << 28) |
-		 (mt->base.base.depth[0] << 16) | mt->base.base.height[0]);
+		 (mt->base.base.depth0 << 16) | mt->base.base.height0);
 	so_data (so, 0x03000000);
 	so_data (so, mt->base.base.last_level << 4);
 
diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c
index ea61357aaa..39d65279fc 100644
--- a/src/gallium/drivers/nv50/nv50_transfer.c
+++ b/src/gallium/drivers/nv50/nv50_transfer.c
@@ -1,6 +1,7 @@
 
 #include "pipe/p_context.h"
 #include "pipe/p_inlines.h"
+#include "util/u_math.h"
 
 #include "nv50_context.h"
 
@@ -156,9 +157,9 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 	tx->base.block = pt->block;
 	if (!pt->nblocksx[level]) {
 		tx->base.nblocksx = pf_get_nblocksx(&pt->block,
-						    pt->width[level]);
+						    u_minify(pt->width0, level));
 		tx->base.nblocksy = pf_get_nblocksy(&pt->block,
-						    pt->height[level]);
+						    u_minify(pt->height0, level));
 	} else {
 		tx->base.nblocksx = pt->nblocksx[level];
 		tx->base.nblocksy = pt->nblocksy[level];
@@ -167,9 +168,9 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 	tx->base.usage = usage;
 
 	tx->level_pitch = lvl->pitch;
-	tx->level_width = mt->base.base.width[level];
-	tx->level_height = mt->base.base.height[level];
-	tx->level_depth = mt->base.base.depth[level];
+	tx->level_width = u_minify(mt->base.base.width0, level);
+	tx->level_height = u_minify(mt->base.base.height0, level);
+	tx->level_depth = u_minify(mt->base.base.depth0, level);
 	tx->level_offset = lvl->image_offset[image];
 	tx->level_tiling = lvl->tile_mode;
 	tx->level_x = pf_get_nblocksx(&tx->base.block, x);
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index ae23329b83..769733b6dd 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -69,6 +69,7 @@ static void r300_destroy_context(struct pipe_context* context)
     FREE(r300->blend_color_state);
     FREE(r300->rs_block);
     FREE(r300->scissor_state);
+    FREE(r300->vertex_info);
     FREE(r300->viewport_state);
     FREE(r300);
 }
@@ -123,15 +124,24 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     r300->context.clear = r300_clear;
 
-    if (r300screen->caps->has_tcl)
-    {
+    if (r300screen->caps->has_tcl) {
         r300->context.draw_arrays = r300_draw_arrays;
         r300->context.draw_elements = r300_draw_elements;
         r300->context.draw_range_elements = r300_draw_range_elements;
-    }
-    else
-    {
-        assert(0);
+    } else {
+        r300->context.draw_arrays = r300_swtcl_draw_arrays;
+        r300->context.draw_elements = r300_draw_elements;
+        r300->context.draw_range_elements = r300_swtcl_draw_range_elements;
+
+        /* Create a Draw. This is used for SW TCL. */
+        r300->draw = draw_create();
+        /* Enable our renderer. */
+        draw_set_rasterize_stage(r300->draw, r300_draw_stage(r300));
+        /* Enable Draw's clipping. */
+        draw_set_driver_clipping(r300->draw, FALSE);
+        /* Force Draw to never do viewport transform, since we can do
+         * transform in hardware, always. */
+        draw_set_viewport_state(r300->draw, &r300_viewport_identity);
     }
 
     r300->context.is_texture_referenced = r300_is_texture_referenced;
@@ -143,18 +153,9 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     r300->blend_color_state = CALLOC_STRUCT(r300_blend_color_state);
     r300->rs_block = CALLOC_STRUCT(r300_rs_block);
     r300->scissor_state = CALLOC_STRUCT(r300_scissor_state);
+    r300->vertex_info = CALLOC_STRUCT(r300_vertex_info);
     r300->viewport_state = CALLOC_STRUCT(r300_viewport_state);
 
-    /* Create a Draw. This is used for vert collation and SW TCL. */
-    r300->draw = draw_create();
-    /* Enable our renderer. */
-    draw_set_rasterize_stage(r300->draw, r300_draw_stage(r300));
-    /* Disable Draw's clipping if TCL is present. */
-    draw_set_driver_clipping(r300->draw, r300_screen(screen)->caps->has_tcl);
-    /* Force Draw to never do viewport transform, since (again) we can do
-     * transform in hardware, always. */
-    draw_set_viewport_state(r300->draw, &r300_viewport_identity);
-
     /* Open up the OQ BO. */
     r300->oqbo = screen->buffer_create(screen, 4096,
             PIPE_BUFFER_USAGE_VERTEX, 4096);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index f954ba7f9a..39c0914cff 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -92,6 +92,10 @@ struct r300_sampler_state {
     uint32_t filter0;      /* R300_TX_FILTER0: 0x4400 */
     uint32_t filter1;      /* R300_TX_FILTER1: 0x4440 */
     uint32_t border_color; /* R300_TX_BORDER_COLOR: 0x45c0 */
+
+    /* Min/max LOD must be clamped to [0, last_level], thus
+     * it's dependent on a currently bound texture */
+    unsigned min_lod, max_lod;
 };
 
 struct r300_scissor_state {
@@ -219,11 +223,6 @@ struct r300_texture {
 struct r300_vertex_info {
     /* Parent class */
     struct vertex_info vinfo;
-    /* Map of vertex attributes into PVS memory for HW TCL,
-     * or GA memory for SW TCL. */
-    int vs_tab[16];
-    /* Map of rasterizer attributes from GB through RS to US. */
-    int fs_tab[16];
 
     /* R300_VAP_PROG_STREAK_CNTL_[0-7] */
     uint32_t vap_prog_stream_cntl[8];
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index eeb97a2d37..98a39390bf 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -129,7 +129,9 @@ static const float * get_shader_constant(
     struct rc_constant * constant,
     struct r300_constant_buffer * externals)
 {
-    static const float zero[4] = { 0.0, 0.0, 0.0, 0.0 };
+    static float vec[4] = { 0.0, 0.0, 0.0, 1.0 };
+    struct pipe_texture *tex;
+
     switch(constant->Type) {
         case RC_CONSTANT_EXTERNAL:
             return externals->constants[constant->u.External];
@@ -137,11 +139,31 @@ static const float * get_shader_constant(
         case RC_CONSTANT_IMMEDIATE:
             return constant->u.Immediate;
 
+        case RC_CONSTANT_STATE:
+            switch (constant->u.State[0]) {
+                /* Factor for converting rectangle coords to
+                 * normalized coords. Should only show up on non-r500. */
+                case RC_STATE_R300_TEXRECT_FACTOR:
+                    tex = &r300->textures[constant->u.State[1]]->tex;
+                    vec[0] = 1.0 / tex->width0;
+                    vec[1] = 1.0 / tex->height0;
+                    break;
+
+                default:
+                    debug_printf("r300: Implementation error: "
+                        "Unknown RC_CONSTANT type %d\n", constant->u.State[0]);
+            }
+            break;
+
         default:
-            debug_printf("r300: Implementation error: Unhandled constant type %i\n",
-                constant->Type);
-            return zero;
+            debug_printf("r300: Implementation error: "
+                "Unhandled constant type %d\n", constant->Type);
     }
+
+    /* This should either be (0, 0, 0, 1), which should be a relatively safe
+     * RGBA or STRQ value, or it could be one of the RC_CONSTANT_STATE
+     * state factors. */
+    return vec;
 }
 
 /* Convert a normal single-precision float into the 7.16 format
@@ -561,6 +583,8 @@ void r300_emit_texture(struct r300_context* r300,
                        unsigned offset)
 {
     uint32_t filter0 = sampler->filter0;
+    uint32_t format0 = tex->state.format0;
+    unsigned min_level, max_level;
     CS_LOCALS(r300);
 
     /* to emulate 1D textures through 2D ones correctly */
@@ -569,13 +593,20 @@ void r300_emit_texture(struct r300_context* r300,
         filter0 |= R300_TX_WRAP_T(R300_TX_CLAMP_TO_EDGE);
     }
 
+    /* determine min/max levels */
+    /* the MAX_MIP level is the largest (finest) one */
+    max_level = MIN2(sampler->max_lod, tex->tex.last_level);
+    min_level = MIN2(sampler->min_lod, max_level);
+    format0 |= R300_TX_NUM_LEVELS(max_level);
+    filter0 |= R300_TX_MAX_MIP_LEVEL(min_level);
+
     BEGIN_CS(16);
     OUT_CS_REG(R300_TX_FILTER0_0 + (offset * 4), filter0 |
         (offset << 28));
     OUT_CS_REG(R300_TX_FILTER1_0 + (offset * 4), sampler->filter1);
     OUT_CS_REG(R300_TX_BORDER_COLOR_0 + (offset * 4), sampler->border_color);
 
-    OUT_CS_REG(R300_TX_FORMAT0_0 + (offset * 4), tex->state.format0);
+    OUT_CS_REG(R300_TX_FORMAT0_0 + (offset * 4), format0);
     OUT_CS_REG(R300_TX_FORMAT1_0 + (offset * 4), tex->state.format1);
     OUT_CS_REG(R300_TX_FORMAT2_0 + (offset * 4), tex->state.format2);
     OUT_CS_REG_SEQ(R300_TX_OFFSET_0 + (offset * 4), 1);
@@ -690,12 +721,22 @@ void r300_emit_vertex_format_state(struct r300_context* r300)
     END_CS;
 }
 
+
 void r300_emit_vertex_program_code(struct r300_context* r300,
                                    struct r300_vertex_program_code* code)
 {
     int i;
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
     unsigned instruction_count = code->length / 4;
+
+    int vtx_mem_size = r300screen->caps->is_r500 ? 128 : 72;
+    int input_count = MAX2(util_bitcount(code->InputsRead), 1);
+    int output_count = MAX2(util_bitcount(code->OutputsWritten), 1);
+    int temp_count = MAX2(code->num_temporaries, 1);
+    int pvs_num_slots = MIN3(vtx_mem_size / input_count,
+                             vtx_mem_size / output_count, 10);
+    int pvs_num_controllers = MIN2(vtx_mem_size / temp_count, 6);
+
     CS_LOCALS(r300);
 
     if (!r300screen->caps->has_tcl) {
@@ -708,8 +749,7 @@ void r300_emit_vertex_program_code(struct r300_context* r300,
     /* R300_VAP_PVS_CODE_CNTL_0
      * R300_VAP_PVS_CONST_CNTL
      * R300_VAP_PVS_CODE_CNTL_1
-     * See the r5xx docs for instructions on how to use these.
-     * XXX these could be optimized to select better values... */
+     * See the r5xx docs for instructions on how to use these. */
     OUT_CS_REG_SEQ(R300_VAP_PVS_CODE_CNTL_0, 3);
     OUT_CS(R300_PVS_FIRST_INST(0) |
             R300_PVS_XYZW_VALID_INST(instruction_count - 1) |
@@ -722,10 +762,11 @@ void r300_emit_vertex_program_code(struct r300_context* r300,
     for (i = 0; i < code->length; i++)
         OUT_CS(code->body.d[i]);
 
-    OUT_CS_REG(R300_VAP_CNTL, R300_PVS_NUM_SLOTS(10) |
-            R300_PVS_NUM_CNTLRS(5) |
+    OUT_CS_REG(R300_VAP_CNTL, R300_PVS_NUM_SLOTS(pvs_num_slots) |
+            R300_PVS_NUM_CNTLRS(pvs_num_controllers) |
             R300_PVS_NUM_FPUS(r300screen->caps->num_vert_fpus) |
-            R300_PVS_VF_MAX_VTX_NUM(12));
+            R300_PVS_VF_MAX_VTX_NUM(12) |
+            (r300screen->caps->is_r500 ? R500_TCL_STATE_OPTIMIZATION : 0));
     END_CS;
 }
 
@@ -790,13 +831,22 @@ void r300_emit_viewport_state(struct r300_context* r300,
     END_CS;
 }
 
+void r300_emit_texture_count(struct r300_context* r300)
+{
+    CS_LOCALS(r300);
+
+    BEGIN_CS(2);
+    OUT_CS_REG(R300_TX_ENABLE, (1 << r300->texture_count) - 1);
+    END_CS;
+
+}
+
 void r300_flush_textures(struct r300_context* r300)
 {
     CS_LOCALS(r300);
 
-    BEGIN_CS(4);
+    BEGIN_CS(2);
     OUT_CS_REG(R300_TX_INVALTAGS, 0);
-    OUT_CS_REG(R300_TX_ENABLE, (1 << r300->texture_count) - 1);
     END_CS;
 }
 
@@ -950,6 +1000,8 @@ validate:
     /* Samplers and textures are tracked separately but emitted together. */
     if (r300->dirty_state &
             (R300_ANY_NEW_SAMPLERS | R300_ANY_NEW_TEXTURES)) {
+        r300_emit_texture_count(r300);
+
         for (i = 0; i < MIN2(r300->sampler_count, r300->texture_count); i++) {
   	    if (r300->dirty_state &
 		((R300_NEW_SAMPLER << i) | (R300_NEW_TEXTURE << i))) {
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index 7c83c5166d..3797d3d332 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -92,6 +92,8 @@ void r300_emit_vertex_shader(struct r300_context* r300,
 void r300_emit_viewport_state(struct r300_context* r300,
                               struct r300_viewport_state* viewport);
 
+void r300_emit_texture_count(struct r300_context* r300);
+
 void r300_flush_textures(struct r300_context* r300);
 
 /* Emit all dirty state. */
diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h
index 8ca785cb58..3a419b24b0 100644
--- a/src/gallium/drivers/r300/r300_reg.h
+++ b/src/gallium/drivers/r300/r300_reg.h
@@ -1293,7 +1293,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #        define R500_RS_INST_TEX_ID(x)                  ((x) << 0)
 #define R500_RS_INST_TEX_CN_WRITE			(1 << 4)
 #define R500_RS_INST_TEX_ADDR_SHIFT			5
-#        define R500_RS_INST_TEX_ADDR(x)                ((x) << 0)
+#        define R500_RS_INST_TEX_ADDR(x)                ((x) << 5)
 #define R500_RS_INST_COL_ID_SHIFT			12
 #        define R500_RS_INST_COL_ID(x)                  ((x) << 12)
 #define R500_RS_INST_COL_CN_NO_WRITE			(0 << 16)
@@ -1463,6 +1463,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_MIN_FILTER_MIP_NEAREST    (1 << 13)
 #	define R300_TX_MIN_FILTER_MIP_LINEAR     (2 << 13)
 #	define R300_TX_MIN_FILTER_MIP_MASK       (3 << 13)
+#       define R300_TX_MAX_MIP_LEVEL_SHIFT       17
+#       define R300_TX_MAX_MIP_LEVEL_MASK        (0xf << 17)
 #	define R300_TX_MAX_ANISO_1_TO_1          (0 << 21)
 #	define R300_TX_MAX_ANISO_2_TO_1          (1 << 21)
 #	define R300_TX_MAX_ANISO_4_TO_1          (2 << 21)
@@ -1471,6 +1473,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_MAX_ANISO_MASK            (7 << 21)
 #       define R300_TX_WRAP_S(x)                 ((x) << 0)
 #       define R300_TX_WRAP_T(x)                 ((x) << 3)
+#       define R300_TX_MAX_MIP_LEVEL(x)          ((x) << 17)
 
 #define R300_TX_FILTER1_0                      0x4440
 #	define R300_CHROMA_KEY_MODE_DISABLE    0
@@ -1500,8 +1503,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_TX_HEIGHTMASK_MASK           (2047 << 11)
 #	define R300_TX_DEPTHMASK_SHIFT           22
 #	define R300_TX_DEPTHMASK_MASK            (0xf << 22)
-#       define R300_TX_MAX_MIP_LEVEL_SHIFT       26
-#       define R300_TX_MAX_MIP_LEVEL_MASK        (0xf << 26)
 #       define R300_TX_SIZE_PROJECTED            (1 << 30)
 #       define R300_TX_PITCH_EN                  (1 << 31)
 #       define R300_TX_WIDTH(x)                  ((x) << 0)
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 62e1456ed3..4c5fb405c6 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -183,8 +183,6 @@ boolean r300_draw_range_elements(struct pipe_context* pipe,
         return FALSE;
     }
 
-    setup_vertex_attributes(r300);
-
     setup_index_buffer(r300, indexBuffer, indexSize);
 
     r300_emit_dirty_state(r300);
@@ -226,8 +224,6 @@ boolean r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
         return FALSE;
     }
 
-    setup_vertex_attributes(r300);
-
     r300_emit_dirty_state(r300);
 
     r300_emit_aos(r300, start);
diff --git a/src/gallium/drivers/r300/r300_screen.h b/src/gallium/drivers/r300/r300_screen.h
index 41df31f670..1ce5ff3904 100644
--- a/src/gallium/drivers/r300/r300_screen.h
+++ b/src/gallium/drivers/r300/r300_screen.h
@@ -27,6 +27,8 @@
 
 #include "r300_chipset.h"
 
+struct r300_winsys;
+
 struct r300_screen {
     /* Parent class */
     struct pipe_screen screen;
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index d1eced61db..a88d66db24 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -523,6 +523,11 @@ static void*
                                                    state->mag_img_filter,
                                                    state->min_mip_filter);
 
+    /* Unfortunately, r300-r500 don't support floating-point mipmap lods. */
+    /* We must pass these to the emit function to clamp them properly. */
+    sampler->min_lod = MAX2((unsigned)state->min_lod, 0);
+    sampler->max_lod = MAX2((unsigned)ceilf(state->max_lod), 0);
+
     lod_bias = CLAMP((int)(state->lod_bias * 32), -(1 << 9), (1 << 9) - 1);
 
     sampler->filter1 |= lod_bias << R300_LOD_BIAS_SHIFT;
@@ -571,6 +576,7 @@ static void r300_set_sampler_textures(struct pipe_context* pipe,
                                       struct pipe_texture** texture)
 {
     struct r300_context* r300 = r300_context(pipe);
+    boolean is_r500 = r300_screen(r300->context.screen)->caps->is_r500;
     int i;
 
     /* XXX magic num */
@@ -585,6 +591,13 @@ static void r300_set_sampler_textures(struct pipe_context* pipe,
             pipe_texture_reference((struct pipe_texture**)&r300->textures[i],
                 texture[i]);
             r300->dirty_state |= (R300_NEW_TEXTURE << i);
+
+            /* R300-specific - set the texrect factor in a fragment shader */
+            if (!is_r500 && r300->textures[i]->is_npot) {
+                /* XXX It would be nice to re-emit just 1 constant,
+                 * XXX not all of them */
+                r300->dirty_state |= R300_NEW_FRAGMENT_SHADER_CONSTANTS;
+            }
         }
     }
 
@@ -674,6 +687,8 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
         draw_flush(r300->draw);
         draw_set_vertex_buffers(r300->draw, count, buffers);
     }
+
+    r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
 }
 
 static void r300_set_vertex_elements(struct pipe_context* pipe,
@@ -706,9 +721,6 @@ static void* r300_create_vs_state(struct pipe_context* pipe,
 
         tgsi_scan_shader(shader->tokens, &vs->info);
 
-        /* Appease Draw. */
-        vs->draw = draw_create_vertex_shader(r300->draw, shader);
-
         return (void*)vs;
     } else {
         return draw_create_vertex_shader(r300->draw, shader);
@@ -719,8 +731,6 @@ static void r300_bind_vs_state(struct pipe_context* pipe, void* shader)
 {
     struct r300_context* r300 = r300_context(pipe);
 
-    draw_flush(r300->draw);
-
     if (r300_screen(pipe->screen)->caps->has_tcl) {
         struct r300_vertex_shader* vs = (struct r300_vertex_shader*)shader;
 
@@ -731,10 +741,10 @@ static void r300_bind_vs_state(struct pipe_context* pipe, void* shader)
             r300_translate_vertex_shader(r300, vs);
         }
 
-        draw_bind_vertex_shader(r300->draw, vs->draw);
         r300->vs = vs;
         r300->dirty_state |= R300_NEW_VERTEX_SHADER | R300_NEW_VERTEX_SHADER_CONSTANTS;
     } else {
+        draw_flush(r300->draw);
         draw_bind_vertex_shader(r300->draw,
                 (struct draw_vertex_shader*)shader);
     }
@@ -748,7 +758,6 @@ static void r300_delete_vs_state(struct pipe_context* pipe, void* shader)
         struct r300_vertex_shader* vs = (struct r300_vertex_shader*)shader;
 
         rc_constants_destroy(&vs->code.constants);
-        draw_delete_vertex_shader(r300->draw, vs->draw);
         FREE((void*)vs->state.tokens);
         FREE(shader);
     } else {
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 131006ca60..4315780b4f 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -47,8 +47,8 @@ struct r300_shader_derived_value {
 
 unsigned r300_shader_key_hash(void* key) {
     struct r300_shader_key* shader_key = (struct r300_shader_key*)key;
-    unsigned vs = (unsigned)shader_key->vs;
-    unsigned fs = (unsigned)shader_key->fs;
+    unsigned vs = (intptr_t)shader_key->vs;
+    unsigned fs = (intptr_t)shader_key->fs;
 
     return (vs << 16) | (fs & 0xffff);
 }
@@ -61,88 +61,46 @@ int r300_shader_key_compare(void* key1, void* key2) {
         (shader_key1->fs == shader_key2->fs);
 }
 
-/* Set up the vs_tab and routes. */
-static void r300_vs_tab_routes(struct r300_context* r300,
-                               struct r300_vertex_info* vformat)
+/* Set up the vs_output_tab and routes. */
+static void r300_vs_output_tab_routes(struct r300_context* r300,
+                                      int* vs_output_tab)
 {
-    struct r300_screen* r300screen = r300_screen(r300->context.screen);
-    struct vertex_info* vinfo = &vformat->vinfo;
-    int* tab = vformat->vs_tab;
+    struct vertex_info* vinfo = &r300->vertex_info->vinfo;
     boolean pos = FALSE, psize = FALSE, fog = FALSE;
     int i, texs = 0, cols = 0;
-    struct tgsi_shader_info* info;
-
-    if (r300screen->caps->has_tcl) {
-        /* Use vertex shader to determine required routes. */
-        info = &r300->vs->info;
-    } else {
-        /* Use fragment shader to determine required routes. */
-        info = &r300->fs->info;
-    }
+    struct tgsi_shader_info* info = &r300->fs->info;
 
-    assert(info->num_inputs <= 16);
+    /* XXX One day we should figure out how to handle a different number of
+     * VS outputs and FS inputs, as well as a different number of vertex streams
+     * and VS inputs. It's definitely one of the sources of hardlocks. */
 
-    if (!r300screen->caps->has_tcl || !r300->rs_state->enable_vte)
-    {
-        for (i = 0; i < info->num_inputs; i++) {
-            switch (r300->vs->code.inputs[i]) {
-                case TGSI_SEMANTIC_POSITION:
-                    pos = TRUE;
-                    tab[i] = 0;
-                    break;
-                case TGSI_SEMANTIC_COLOR:
-                    tab[i] = 2 + cols;
-                    cols++;
-                    break;
-                case TGSI_SEMANTIC_PSIZE:
-                    assert(psize == FALSE);
-                    psize = TRUE;
-                    tab[i] = 15;
-                    break;
-                case TGSI_SEMANTIC_FOG:
-                    assert(fog == FALSE);
-                    fog = TRUE;
-                    /* Fall through */
-                case TGSI_SEMANTIC_GENERIC:
-                    tab[i] = 6 + texs;
-                    texs++;
-                    break;
-                default:
-                    debug_printf("r300: Unknown vertex input %d\n",
-                        info->input_semantic_name[i]);
-                    break;
-            }
-        }
-    }
-    else
-    {
-        /* Just copy vert attribs over as-is. */
-        for (i = 0; i < info->num_inputs; i++) {
-            tab[i] = i;
-        }
-
-        for (i = 0; i < info->num_outputs; i++) {
-            switch (info->output_semantic_name[i]) {
-                case TGSI_SEMANTIC_POSITION:
-                    pos = TRUE;
-                    break;
-                case TGSI_SEMANTIC_COLOR:
-                    cols++;
-                    break;
-                case TGSI_SEMANTIC_PSIZE:
-                    psize = TRUE;
-                    break;
-                case TGSI_SEMANTIC_FOG:
-                    fog = TRUE;
-                    /* Fall through */
-                case TGSI_SEMANTIC_GENERIC:
-                    texs++;
-                    break;
-                default:
-                    debug_printf("r300: Unknown vertex output %d\n",
-                        info->output_semantic_name[i]);
-                    break;
-            }
+    for (i = 0; i < info->num_inputs; i++) {
+        switch (info->input_semantic_name[i]) {
+            case TGSI_SEMANTIC_POSITION:
+                pos = TRUE;
+                vs_output_tab[i] = 0;
+                break;
+            case TGSI_SEMANTIC_COLOR:
+                vs_output_tab[i] = 2 + cols;
+                cols++;
+                break;
+            case TGSI_SEMANTIC_PSIZE:
+                assert(psize == FALSE);
+                psize = TRUE;
+                vs_output_tab[i] = 15;
+                break;
+            case TGSI_SEMANTIC_FOG:
+                assert(fog == FALSE);
+                fog = TRUE;
+                /* Fall through */
+            case TGSI_SEMANTIC_GENERIC:
+                vs_output_tab[i] = 6 + texs;
+                texs++;
+                break;
+            default:
+                debug_printf("r300: Unknown vertex input %d\n",
+                    info->input_semantic_name[i]);
+                break;
         }
     }
 
@@ -161,14 +119,13 @@ static void r300_vs_tab_routes(struct r300_context* r300,
 
     /* We need to add vertex position attribute only for SW TCL case,
      * for HW TCL case it could be generated by vertex shader */
-    if (!pos && !r300screen->caps->has_tcl) {
-        debug_printf("r300: Forcing vertex position attribute emit...\n");
+    if (!pos) {
         /* Make room for the position attribute
-         * at the beginning of the tab. */
+         * at the beginning of the vs_output_tab. */
         for (i = 15; i > 0; i--) {
-            tab[i] = tab[i-1];
+            vs_output_tab[i] = vs_output_tab[i-1];
         }
-        tab[0] = 0;
+        vs_output_tab[0] = 0;
     }
 
     /* Position. */
@@ -227,43 +184,78 @@ static void r300_vs_tab_routes(struct r300_context* r300,
 }
 
 /* Update the PSC tables. */
-static void r300_vertex_psc(struct r300_context* r300,
-                            struct r300_vertex_info* vformat)
+static void r300_vertex_psc(struct r300_context* r300)
 {
-    struct r300_screen* r300screen = r300_screen(r300->context.screen);
-    struct vertex_info* vinfo = &vformat->vinfo;
-    int* tab = vformat->vs_tab;
+    struct r300_vertex_info *vformat = r300->vertex_info;
     uint16_t type, swizzle;
     enum pipe_format format;
-    unsigned i, attrib_count;
+    unsigned i;
 
     /* Vertex shaders have no semantics on their inputs,
-     * so PSC should just route stuff based on their info,
+     * so PSC should just route stuff based on the vertex elements,
      * and not on attrib information. */
-    if (r300screen->caps->has_tcl) {
-        attrib_count = r300->vs->info.num_inputs;
-        DBG(r300, DBG_DRAW, "r300: routing %d attribs in psc for vs\n",
-                attrib_count);
-    } else {
-        attrib_count = vinfo->num_attribs;
-        DBG(r300, DBG_DRAW, "r300: attrib count: %d\n", attrib_count);
-        for (i = 0; i < attrib_count; i++) {
-            DBG(r300, DBG_DRAW, "r300: attrib: offset %d, interp %d, size %d,"
-                   " tab %d\n", vinfo->attrib[i].src_index,
-                   vinfo->attrib[i].interp_mode, vinfo->attrib[i].emit,
-                   tab[i]);
+    DBG(r300, DBG_DRAW, "r300: vs expects %d attribs, routing %d elements"
+            " in psc\n",
+            r300->vs->info.num_inputs,
+            r300->vertex_element_count);
+
+    for (i = 0; i < r300->vertex_element_count; i++) {
+        format = r300->vertex_element[i].src_format;
+
+        type = r300_translate_vertex_data_type(format) |
+            (i << R300_DST_VEC_LOC_SHIFT);
+        swizzle = r300_translate_vertex_data_swizzle(format);
+
+        if (i % 2) {
+            vformat->vap_prog_stream_cntl[i >> 1] |= type << 16;
+            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 16;
+        } else {
+            vformat->vap_prog_stream_cntl[i >> 1] |= type;
+            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle;
         }
     }
 
+
+    assert(i <= 15);
+
+    /* Set the last vector in the PSC. */
+    if (i) {
+        i -= 1;
+    }
+    vformat->vap_prog_stream_cntl[i >> 1] |=
+        (R300_LAST_VEC << (i & 1 ? 16 : 0));
+}
+
+/* Update the PSC tables for SW TCL, using Draw. */
+static void r300_swtcl_vertex_psc(struct r300_context* r300,
+                                  int* vs_output_tab)
+{
+    struct r300_vertex_info *vformat = r300->vertex_info;
+    struct vertex_info* vinfo = &vformat->vinfo;
+    uint16_t type, swizzle;
+    enum pipe_format format;
+    unsigned i, attrib_count;
+
+    /* For each Draw attribute, route it to the fragment shader according
+     * to the vs_output_tab. */
+    attrib_count = vinfo->num_attribs;
+    DBG(r300, DBG_DRAW, "r300: attrib count: %d\n", attrib_count);
+    for (i = 0; i < attrib_count; i++) {
+        DBG(r300, DBG_DRAW, "r300: attrib: offset %d, interp %d, size %d,"
+               " vs_output_tab %d\n", vinfo->attrib[i].src_index,
+               vinfo->attrib[i].interp_mode, vinfo->attrib[i].emit,
+               vs_output_tab[i]);
+    }
+
     for (i = 0; i < attrib_count; i++) {
         /* Make sure we have a proper destination for our attribute. */
-        assert(tab[i] != -1);
+        assert(vs_output_tab[i] != -1);
 
         format = draw_translate_vinfo_format(vinfo->attrib[i].emit);
 
         /* Obtain the type of data in this attribute. */
         type = r300_translate_vertex_data_type(format) |
-            tab[i] << R300_DST_VEC_LOC_SHIFT;
+            vs_output_tab[i] << R300_DST_VEC_LOC_SHIFT;
 
         /* Obtain the swizzle for this attribute. Note that the default
          * swizzle in the hardware is not XYZW! */
@@ -272,12 +264,10 @@ static void r300_vertex_psc(struct r300_context* r300,
         /* Add the attribute to the PSC table. */
         if (i & 1) {
             vformat->vap_prog_stream_cntl[i >> 1] |= type << 16;
-
             vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 16;
         } else {
-            vformat->vap_prog_stream_cntl[i >> 1] |= type <<  0;
-
-            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle << 0;
+            vformat->vap_prog_stream_cntl[i >> 1] |= type;
+            vformat->vap_prog_stream_cntl_ext[i >> 1] |= swizzle;
         }
     }
 
@@ -289,74 +279,12 @@ static void r300_vertex_psc(struct r300_context* r300,
         (R300_LAST_VEC << (i & 1 ? 16 : 0));
 }
 
-/* Set up the mappings from GB to US, for RS block. */
-static void r300_update_fs_tab(struct r300_context* r300,
-                               struct r300_vertex_info* vformat)
-{
-    struct tgsi_shader_info* info = &r300->fs->info;
-    int i, cols = 0, texs = 0, cols_emitted = 0;
-    int* tab = vformat->fs_tab;
-
-    for (i = 0; i < 16; i++) {
-        tab[i] = -1;
-    }
-
-    assert(info->num_inputs <= 16);
-    for (i = 0; i < info->num_inputs; i++) {
-        switch (info->input_semantic_name[i]) {
-            case TGSI_SEMANTIC_COLOR:
-                tab[i] = INTERP_LINEAR;
-                cols++;
-                break;
-            case TGSI_SEMANTIC_POSITION:
-            case TGSI_SEMANTIC_PSIZE:
-                debug_printf("r300: Implementation error: Can't use "
-                        "pos attribs in fragshader yet!\n");
-                /* Pass through for now */
-            case TGSI_SEMANTIC_FOG:
-            case TGSI_SEMANTIC_GENERIC:
-                tab[i] = INTERP_PERSPECTIVE;
-                break;
-            default:
-                debug_printf("r300: Unknown vertex input %d\n",
-                    info->input_semantic_name[i]);
-                break;
-        }
-    }
-
-    /* Now that we know where everything is... */
-    DBG(r300, DBG_DRAW, "r300: fp input count: %d\n", info->num_inputs);
-    for (i = 0; i < info->num_inputs; i++) {
-        switch (tab[i]) {
-            case INTERP_LINEAR:
-                DBG(r300, DBG_DRAW, "r300: attrib: "
-                        "stack offset %d, color,    tab %d\n",
-                        i, cols_emitted);
-                tab[i] = cols_emitted;
-                cols_emitted++;
-                break;
-            case INTERP_PERSPECTIVE:
-                DBG(r300, DBG_DRAW, "r300: attrib: "
-                        "stack offset %d, texcoord, tab %d\n",
-                        i, cols + texs);
-                tab[i] = cols + texs;
-                texs++;
-                break;
-            case -1:
-                debug_printf("r300: Implementation error: Bad fp interp!\n");
-            default:
-                break;
-        }
-    }
-
-}
-
 /* Set up the RS block. This is the part of the chipset that actually does
  * the rasterization of vertices into fragments. This is also the part of the
  * chipset that locks up if any part of it is even slightly wrong. */
-static void r300_update_rs_block(struct r300_context* r300,
-                                 struct r300_rs_block* rs)
+static void r300_update_rs_block(struct r300_context* r300)
 {
+    struct r300_rs_block* rs = r300->rs_block;
     struct tgsi_shader_info* info = &r300->fs->info;
     int col_count = 0, fp_offset = 0, i, tex_count = 0;
     int rs_tex_comp = 0;
@@ -390,17 +318,18 @@ static void r300_update_rs_block(struct r300_context* r300,
             col_count++;
         }
 
+        for (i = 0; i < col_count; i++) {
+            rs->inst[i] |= R500_RS_INST_COL_ID(i) |
+                R500_RS_INST_COL_CN_WRITE | R500_RS_INST_COL_ADDR(fp_offset);
+            fp_offset++;
+        }
+
         for (i = 0; i < tex_count; i++) {
             rs->inst[i] |= R500_RS_INST_TEX_ID(i) |
                 R500_RS_INST_TEX_CN_WRITE | R500_RS_INST_TEX_ADDR(fp_offset);
             fp_offset++;
         }
 
-        for (i = 0; i < col_count; i++) {
-            rs->inst[i] |= R500_RS_INST_COL_ID(i) |
-                R500_RS_INST_COL_CN_WRITE | R500_RS_INST_COL_ADDR(fp_offset);
-            fp_offset++;
-        }
     } else {
         for (i = 0; i < info->num_inputs; i++) {
             switch (info->input_semantic_name[i]) {
@@ -425,8 +354,10 @@ static void r300_update_rs_block(struct r300_context* r300,
             }
         }
 
+        /* Rasterize at least one color, or bad things happen. */
         if (col_count == 0) {
             rs->ip[0] |= R300_RS_COL_FMT(R300_RS_COL_FMT_0001);
+            col_count++;
         }
 
         if (tex_count == 0) {
@@ -437,9 +368,10 @@ static void r300_update_rs_block(struct r300_context* r300,
                 R300_RS_SEL_Q(R300_RS_SEL_K1);
         }
 
-        /* Rasterize at least one color, or bad things happen. */
-        if ((col_count == 0) && (tex_count == 0)) {
-            col_count++;
+        for (i = 0; i < col_count; i++) {
+            rs->inst[i] |= R300_RS_INST_COL_ID(i) |
+                R300_RS_INST_COL_CN_WRITE | R300_RS_INST_COL_ADDR(fp_offset);
+            fp_offset++;
         }
 
         for (i = 0; i < tex_count; i++) {
@@ -447,28 +379,22 @@ static void r300_update_rs_block(struct r300_context* r300,
                 R300_RS_INST_TEX_CN_WRITE | R300_RS_INST_TEX_ADDR(fp_offset);
             fp_offset++;
         }
-
-        for (i = 0; i < col_count; i++) {
-            rs->inst[i] |= R300_RS_INST_COL_ID(i) |
-                R300_RS_INST_COL_CN_WRITE | R300_RS_INST_COL_ADDR(fp_offset);
-            fp_offset++;
-        }
     }
 
     rs->count = (rs_tex_comp) | (col_count << R300_IC_COUNT_SHIFT) |
         R300_HIRES_EN;
 
-    rs->inst_count = MAX2(MAX2(col_count - 1, tex_count - 1), 0);
+    rs->inst_count = MAX3(col_count - 1, tex_count - 1, 0);
 }
 
 /* Update the vertex format. */
 static void r300_update_derived_shader_state(struct r300_context* r300)
 {
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
-    struct r300_vertex_info* vformat;
-    struct r300_rs_block* rs_block;
+    int vs_output_tab[16];
     int i;
 
+
     /*
     struct r300_shader_key* key;
     struct r300_shader_derived_value* value;
@@ -495,27 +421,26 @@ static void r300_update_derived_shader_state(struct r300_context* r300)
             (void*)key, (void*)value);
     } */
 
-    /* XXX This will be refactored ASAP. */
-    vformat = CALLOC_STRUCT(r300_vertex_info);
-    rs_block = CALLOC_STRUCT(r300_rs_block);
+    /* Reset structures */
+    memset(r300->rs_block, 0, sizeof(struct r300_rs_block));
+    memset(r300->vertex_info, 0, sizeof(struct r300_vertex_info));
 
     for (i = 0; i < 16; i++) {
-        vformat->vs_tab[i] = -1;
-        vformat->fs_tab[i] = -1;
+        vs_output_tab[i] = -1;
     }
 
-    r300_vs_tab_routes(r300, vformat);
-    r300_vertex_psc(r300, vformat);
-    r300_update_fs_tab(r300, vformat);
+    /* Update states */
+    r300_vs_output_tab_routes(r300, vs_output_tab);
 
-    r300_update_rs_block(r300, rs_block);
+    if (r300screen->caps->has_tcl) {
+        r300_vertex_psc(r300);
+    } else {
+        r300_swtcl_vertex_psc(r300, vs_output_tab);
+    }
 
-    FREE(r300->vertex_info);
-    FREE(r300->rs_block);
+    r300_update_rs_block(r300);
 
-    r300->vertex_info = vformat;
-    r300->rs_block = rs_block;
-    r300->dirty_state |= (R300_NEW_VERTEX_FORMAT | R300_NEW_RS_BLOCK);
+    r300->dirty_state |= R300_NEW_RS_BLOCK;
 }
 
 static void r300_update_ztop(struct r300_context* r300)
@@ -553,9 +478,9 @@ static void r300_update_ztop(struct r300_context* r300)
 
 void r300_update_derived_state(struct r300_context* r300)
 {
-    /* XXX */
-    if (TRUE || r300->dirty_state &
-        (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER)) {
+    if (r300->dirty_state &
+        (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER |
+         R300_NEW_VERTEX_FORMAT)) {
         r300_update_derived_shader_state(r300);
     }
 
diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
index c07e6ae676..46d1cb39b5 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.c
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -84,7 +84,7 @@ void r300_emit_invariant_state(struct r300_context* r300)
     END_CS;
 
     /* XXX unsorted stuff from surface_fill */
-    BEGIN_CS(60 + (caps->has_tcl ? 5 : 0) + (caps->is_r500 ? 4 : 0));
+    BEGIN_CS(56 + (caps->has_tcl ? 5 : 0) + (caps->is_r500 ? 4 : 0));
     /* Flush PVS. */
     OUT_CS_REG(R300_VAP_PVS_STATE_FLUSH_REG, 0x0);
 
@@ -135,8 +135,6 @@ void r300_emit_invariant_state(struct r300_context* r300)
     OUT_CS_REG(R300_ZB_DEPTHCLEARVALUE, 0x00000000);
     OUT_CS_REG(R300_ZB_HIZ_OFFSET, 0x00000000);
     OUT_CS_REG(R300_ZB_HIZ_PITCH, 0x00000000);
-    OUT_CS_REG(R300_VAP_VTX_STATE_CNTL, 0x1);
-    OUT_CS_REG(R300_VAP_VSM_VTX_ASSM, 0x405);
     OUT_CS_REG(R300_SE_VTE_CNTL, 0x0000043F);
 
     /* XXX */
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index aea25cf71d..093a21ebe2 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -34,8 +34,8 @@ static void r300_setup_texture_state(struct r300_texture* tex, boolean is_r500)
     struct r300_texture_state* state = &tex->state;
     struct pipe_texture *pt = &tex->tex;
 
-    state->format0 = R300_TX_WIDTH((pt->width[0] - 1) & 0x7ff) |
-                     R300_TX_HEIGHT((pt->height[0] - 1) & 0x7ff);
+    state->format0 = R300_TX_WIDTH((pt->width0 - 1) & 0x7ff) |
+                     R300_TX_HEIGHT((pt->height0 - 1) & 0x7ff);
 
     if (tex->is_npot) {
         /* rectangles love this */
@@ -43,8 +43,7 @@ static void r300_setup_texture_state(struct r300_texture* tex, boolean is_r500)
         state->format2 = (tex->pitch[0] - 1) & 0x1fff;
     } else {
         /* power of two textures (3D, mipmaps, and no pitch) */
-        state->format0 |= R300_TX_DEPTH(util_logbase2(pt->depth[0]) & 0xf) |
-                          R300_TX_NUM_LEVELS(pt->last_level & 0xf);
+        state->format0 |= R300_TX_DEPTH(util_logbase2(pt->depth0) & 0xf);
     }
 
     state->format1 = r300_translate_texformat(pt->format);
@@ -58,17 +57,17 @@ static void r300_setup_texture_state(struct r300_texture* tex, boolean is_r500)
     /* large textures on r500 */
     if (is_r500)
     {
-        if (pt->width[0] > 2048) {
+        if (pt->width0 > 2048) {
             state->format2 |= R500_TXWIDTH_BIT11;
         }
-        if (pt->height[0] > 2048) {
+        if (pt->height0 > 2048) {
             state->format2 |= R500_TXHEIGHT_BIT11;
         }
     }
-    assert(is_r500 || (pt->width[0] <= 2048 && pt->height[0] <= 2048));
+    assert(is_r500 || (pt->width0 <= 2048 && pt->height0 <= 2048));
 
     debug_printf("r300: Set texture state (%dx%d, %d levels)\n",
-		 pt->width[0], pt->height[0], pt->last_level);
+		 pt->width0, pt->height0, pt->last_level);
 }
 
 unsigned r300_texture_get_offset(struct r300_texture* tex, unsigned level,
@@ -106,7 +105,7 @@ unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level)
         return 0;
     }
 
-    return align(pf_get_stride(&tex->tex.block, tex->tex.width[level]), 32);
+    return align(pf_get_stride(&tex->tex.block, u_minify(tex->tex.width0, level)), 32);
 }
 
 static void r300_setup_miptree(struct r300_texture* tex)
@@ -116,14 +115,8 @@ static void r300_setup_miptree(struct r300_texture* tex)
     int i;
 
     for (i = 0; i <= base->last_level; i++) {
-        if (i > 0) {
-            base->width[i] = minify(base->width[i-1]);
-            base->height[i] = minify(base->height[i-1]);
-            base->depth[i] = minify(base->depth[i-1]);
-        }
-
-        base->nblocksx[i] = pf_get_nblocksx(&base->block, base->width[i]);
-        base->nblocksy[i] = pf_get_nblocksy(&base->block, base->height[i]);
+        base->nblocksx[i] = pf_get_nblocksx(&base->block, u_minify(base->width0, i));
+        base->nblocksy[i] = pf_get_nblocksy(&base->block, u_minify(base->height0, i));
 
         stride = r300_texture_get_stride(tex, i);
         layer_size = stride * base->nblocksy[i];
@@ -131,7 +124,7 @@ static void r300_setup_miptree(struct r300_texture* tex)
         if (base->target == PIPE_TEXTURE_CUBE)
             size = layer_size * 6;
         else
-            size = layer_size * base->depth[i];
+            size = layer_size * u_minify(base->depth0, i);
 
         tex->offset[i] = align(tex->size, 32);
         tex->size = tex->offset[i] + size;
@@ -140,15 +133,15 @@ static void r300_setup_miptree(struct r300_texture* tex)
 
         debug_printf("r300: Texture miptree: Level %d "
                 "(%dx%dx%d px, pitch %d bytes)\n",
-                i, base->width[i], base->height[i], base->depth[i],
-                stride);
+                i, u_minify(base->width0, i), u_minify(base->height0, i),
+                u_minify(base->depth0, i), stride);
     }
 }
 
 static void r300_setup_flags(struct r300_texture* tex)
 {
-    tex->is_npot = !util_is_power_of_two(tex->tex.width[0]) ||
-                   !util_is_power_of_two(tex->tex.height[0]);
+    tex->is_npot = !util_is_power_of_two(tex->tex.width0) ||
+                   !util_is_power_of_two(tex->tex.height0);
 }
 
 /* Create a new texture. */
@@ -208,8 +201,8 @@ static struct pipe_surface* r300_get_tex_surface(struct pipe_screen* screen,
         pipe_reference_init(&surface->reference, 1);
         pipe_texture_reference(&surface->texture, texture);
         surface->format = texture->format;
-        surface->width = texture->width[level];
-        surface->height = texture->height[level];
+        surface->width = u_minify(texture->width0, level);
+        surface->height = u_minify(texture->height0, level);
         surface->offset = offset;
         surface->usage = flags;
         surface->zslice = zslice;
@@ -237,7 +230,7 @@ static struct pipe_texture*
 
     /* Support only 2D textures without mipmaps */
     if (base->target != PIPE_TEXTURE_2D ||
-        base->depth[0] != 1 ||
+        base->depth0 != 1 ||
         base->last_level != 0) {
         return NULL;
     }
@@ -287,9 +280,9 @@ r300_video_surface_create(struct pipe_screen *screen,
     template.target = PIPE_TEXTURE_2D;
     template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
     template.last_level = 0;
-    template.width[0] = util_next_power_of_two(width);
-    template.height[0] = util_next_power_of_two(height);
-    template.depth[0] = 1;
+    template.width0 = util_next_power_of_two(width);
+    template.height0 = util_next_power_of_two(height);
+    template.depth0 = 1;
     pf_get_block(template.format, &template.block);
     template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER |
                          PIPE_TEXTURE_USAGE_RENDER_TARGET;
diff --git a/src/gallium/drivers/r300/r300_vbo.c b/src/gallium/drivers/r300/r300_vbo.c
index a6a159667a..6ebaf715dc 100644
--- a/src/gallium/drivers/r300/r300_vbo.c
+++ b/src/gallium/drivers/r300/r300_vbo.c
@@ -34,52 +34,6 @@
 #include "r300_reg.h"
 #include "r300_winsys.h"
 
-static INLINE void setup_vertex_attribute(struct r300_vertex_info *vinfo,
-                                          struct pipe_vertex_element *vert_elem,
-                                          unsigned attr_num)
-{
-    uint16_t hw_fmt1, hw_fmt2;
-
-    hw_fmt1 = r300_translate_vertex_data_type(vert_elem->src_format) |
-        (attr_num << R300_DST_VEC_LOC_SHIFT);
-    hw_fmt2 = r300_translate_vertex_data_swizzle(vert_elem->src_format);
-
-    if (attr_num % 2 == 0)
-    {
-        vinfo->vap_prog_stream_cntl[attr_num >> 1] = hw_fmt1;
-        vinfo->vap_prog_stream_cntl_ext[attr_num >> 1] = hw_fmt2;
-    }
-    else
-    {
-        vinfo->vap_prog_stream_cntl[attr_num >> 1] |= hw_fmt1 << 16;
-        vinfo->vap_prog_stream_cntl_ext[attr_num >> 1] |= hw_fmt2 << 16;
-    }
-}
-
-static void finish_vertex_attribs_setup(struct r300_vertex_info *vinfo,
-                                        unsigned attribs_num)
-{
-    uint32_t last_vec_bit = (attribs_num % 2 == 0) ?
-        (R300_LAST_VEC << 16) : R300_LAST_VEC;
-
-    assert(attribs_num > 0 && attribs_num <= 16);
-    vinfo->vap_prog_stream_cntl[(attribs_num - 1) >> 1] |= last_vec_bit;
-}
-
-void setup_vertex_attributes(struct r300_context *r300)
-{
-    struct pipe_vertex_element *vert_elem;
-    int i;
-
-    for (i = 0; i < r300->vertex_element_count; i++) {
-        vert_elem = &r300->vertex_element[i];
-        setup_vertex_attribute(r300->vertex_info, vert_elem, i);
-    }
-
-    finish_vertex_attribs_setup(r300->vertex_info,
-        r300->vertex_element_count);
-}
-
 static INLINE int get_buffer_offset(struct r300_context *r300,
                                     unsigned int buf_nr,
                                     unsigned int elem_offset)
diff --git a/src/gallium/drivers/r300/r300_vs.h b/src/gallium/drivers/r300/r300_vs.h
index 2a4ce315e3..00b02bf510 100644
--- a/src/gallium/drivers/r300/r300_vs.h
+++ b/src/gallium/drivers/r300/r300_vs.h
@@ -35,9 +35,6 @@ struct r300_vertex_shader {
     struct pipe_shader_state state;
     struct tgsi_shader_info info;
 
-    /* Fallback shader, because Draw has issues */
-    struct draw_vertex_shader* draw;
-
     /* Has this shader been translated yet? */
     boolean translated;
 
diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index 0e2fc11163..9f1524cc66 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -90,14 +90,15 @@ softpipe_destroy( struct pipe_context *pipe )
    if (softpipe->draw)
       draw_destroy( softpipe->draw );
 
-      softpipe->quad.shade->destroy( softpipe->quad.shade );
-      softpipe->quad.depth_test->destroy( softpipe->quad.depth_test );
-      softpipe->quad.blend->destroy( softpipe->quad.blend );
+   softpipe->quad.shade->destroy( softpipe->quad.shade );
+   softpipe->quad.depth_test->destroy( softpipe->quad.depth_test );
+   softpipe->quad.blend->destroy( softpipe->quad.blend );
 
    for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
       sp_destroy_tile_cache(softpipe->cbuf_cache[i]);
       pipe_surface_reference(&softpipe->framebuffer.cbufs[i], NULL);
    }
+
    sp_destroy_tile_cache(softpipe->zsbuf_cache);
    pipe_surface_reference(&softpipe->framebuffer.zsbuf, NULL);
 
@@ -120,7 +121,7 @@ softpipe_destroy( struct pipe_context *pipe )
  * if (the texture is being used as a framebuffer surface)
  *    return PIPE_REFERENCED_FOR_WRITE
  * else if (the texture is a bound texture source)
- *    return PIPE_REFERENCED_FOR_READ  XXX not done yet
+ *    return PIPE_REFERENCED_FOR_READ
  * else
  *    return PIPE_UNREFERENCED
  */
@@ -132,6 +133,7 @@ softpipe_is_texture_referenced( struct pipe_context *pipe,
    struct softpipe_context *softpipe = softpipe_context( pipe );
    unsigned i;
 
+   /* check if any of the bound drawing surfaces are this texture */
    if (softpipe->dirty_render_cache) {
       for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) {
          if (softpipe->framebuffer.cbufs[i] && 
@@ -145,7 +147,12 @@ softpipe_is_texture_referenced( struct pipe_context *pipe,
       }
    }
    
-   /* FIXME: we also need to do the same for the texture cache */
+   /* check if any of the tex_cache textures are this texture */
+   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
+      if (softpipe->tex_cache[i] &&
+          softpipe->tex_cache[i]->texture == texture)
+         return PIPE_REFERENCED_FOR_READ;
+   }
    
    return PIPE_UNREFERENCED;
 }
@@ -247,9 +254,9 @@ softpipe_create( struct pipe_screen *screen )
 
 
    /* setup quad rendering stages */
-      softpipe->quad.shade = sp_quad_shade_stage(softpipe);
-      softpipe->quad.depth_test = sp_quad_depth_test_stage(softpipe);
-      softpipe->quad.blend = sp_quad_blend_stage(softpipe);
+   softpipe->quad.shade = sp_quad_shade_stage(softpipe);
+   softpipe->quad.depth_test = sp_quad_depth_test_stage(softpipe);
+   softpipe->quad.blend = sp_quad_blend_stage(softpipe);
 
 
    /*
@@ -279,7 +286,6 @@ softpipe_create( struct pipe_screen *screen )
    draw_set_render(softpipe->draw, softpipe->vbuf_backend);
 
 
-
    /* plug in AA line/point stages */
    draw_install_aaline_stage(softpipe->draw, &softpipe->pipe);
    draw_install_aapoint_stage(softpipe->draw, &softpipe->pipe);
@@ -295,4 +301,3 @@ softpipe_create( struct pipe_screen *screen )
    softpipe_destroy(&softpipe->pipe);
    return NULL;
 }
-
diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c
index 4076114d39..a8999ed347 100644
--- a/src/gallium/drivers/softpipe/sp_fs_exec.c
+++ b/src/gallium/drivers/softpipe/sp_fs_exec.c
@@ -126,7 +126,13 @@ exec_run( const struct sp_fragment_shader *base,
    setup_pos_vector(quad->posCoef, 
                     (float)quad->input.x0, (float)quad->input.y0, 
                     &machine->QuadPos);
-   
+
+   if (quad->input.facing) {
+      machine->Face = -1.0f;
+   } else {
+      machine->Face = 1.0f;
+   }
+
    quad->inout.mask &= tgsi_exec_machine_run( machine );
    if (quad->inout.mask == 0)
       return FALSE;
diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c
index bd7306ea67..1d8142311a 100644
--- a/src/gallium/drivers/softpipe/sp_state_fs.c
+++ b/src/gallium/drivers/softpipe/sp_state_fs.c
@@ -143,6 +143,7 @@ softpipe_delete_vs_state(struct pipe_context *pipe, void *vs)
    struct sp_vertex_shader *state = (struct sp_vertex_shader *) vs;
 
    draw_delete_vertex_shader(softpipe->draw, state->draw_data);
+   FREE( (void *)state->shader.tokens );
    FREE( state );
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c
index c22ee86b66..e26153b1d9 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.c
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.c
@@ -521,7 +521,7 @@ compute_lambda_1d(const struct sp_sampler_varient *samp,
    const struct pipe_sampler_state *sampler = samp->sampler;
    float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
    float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
-   float rho = MAX2(dsdx, dsdy) * texture->width[0];
+   float rho = MAX2(dsdx, dsdy) * texture->width0;
    float lambda;
 
    lambda = util_fast_log2(rho);
@@ -545,8 +545,8 @@ compute_lambda_2d(const struct sp_sampler_varient *samp,
    float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
    float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
    float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
-   float maxx = MAX2(dsdx, dsdy) * texture->width[0];
-   float maxy = MAX2(dtdx, dtdy) * texture->height[0];
+   float maxx = MAX2(dsdx, dsdy) * texture->width0;
+   float maxy = MAX2(dtdx, dtdy) * texture->height0;
    float rho  = MAX2(maxx, maxy);
    float lambda;
 
@@ -573,9 +573,9 @@ compute_lambda_3d(const struct sp_sampler_varient *samp,
    float dtdy = fabsf(t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT]);
    float dpdx = fabsf(p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]);
    float dpdy = fabsf(p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT]);
-   float maxx = MAX2(dsdx, dsdy) * texture->width[0];
-   float maxy = MAX2(dtdx, dtdy) * texture->height[0];
-   float maxz = MAX2(dpdx, dpdy) * texture->depth[0];
+   float maxx = MAX2(dsdx, dsdy) * texture->width0;
+   float maxy = MAX2(dtdx, dtdy) * texture->height0;
+   float maxz = MAX2(dpdx, dpdy) * texture->depth0;
    float rho, lambda;
 
    rho = MAX2(maxx, maxy);
@@ -644,8 +644,8 @@ get_texel_2d(const struct sp_sampler_varient *samp,
    const struct pipe_texture *texture = samp->texture;
    unsigned level = addr.bits.level;
 
-   if (x < 0 || x >= (int) texture->width[level] ||
-       y < 0 || y >= (int) texture->height[level]) {
+   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
+       y < 0 || y >= (int) u_minify(texture->height0, level)) {
       return samp->sampler->border_color;
    }
    else {
@@ -737,9 +737,9 @@ get_texel_3d(const struct sp_sampler_varient *samp,
    const struct pipe_texture *texture = samp->texture;
    unsigned level = addr.bits.level;
 
-   if (x < 0 || x >= (int) texture->width[level] ||
-       y < 0 || y >= (int) texture->height[level] ||
-       z < 0 || z >= (int) texture->depth[level]) {
+   if (x < 0 || x >= (int) u_minify(texture->width0, level) ||
+       y < 0 || y >= (int) u_minify(texture->height0, level) ||
+       z < 0 || z >= (int) u_minify(texture->depth0, level)) {
       return samp->sampler->border_color;
    }
    else {
@@ -925,7 +925,7 @@ img_filter_1d_nearest(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
+   width = u_minify(texture->width0, level0);
 
    assert(width > 0);
 
@@ -961,8 +961,8 @@ img_filter_2d_nearest(struct tgsi_sampler *tgsi_sampler,
 
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1008,8 +1008,8 @@ img_filter_cube_nearest(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1046,9 +1046,9 @@ img_filter_3d_nearest(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
-   depth = texture->depth[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
+   depth = u_minify(texture->depth0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1088,7 +1088,7 @@ img_filter_1d_linear(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
+   width = u_minify(texture->width0, level0);
 
    assert(width > 0);
 
@@ -1127,8 +1127,8 @@ img_filter_2d_linear(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1174,8 +1174,8 @@ img_filter_cube_linear(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
 
    assert(width > 0);
    assert(height > 0);
@@ -1221,9 +1221,9 @@ img_filter_3d_linear(struct tgsi_sampler *tgsi_sampler,
    union tex_tile_address addr;
 
    level0 = samp->level;
-   width = texture->width[level0];
-   height = texture->height[level0];
-   depth = texture->depth[level0];
+   width = u_minify(texture->width0, level0);
+   height = u_minify(texture->height0, level0);
+   depth = u_minify(texture->depth0, level0);
 
    addr.value = 0;
    addr.bits.level = level0;
@@ -1778,8 +1778,8 @@ sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp,
 
    samp->texture = texture;
    samp->cache = tex_cache;
-   samp->xpot = util_unsigned_logbase2( texture->width[0] );
-   samp->ypot = util_unsigned_logbase2( texture->height[0] );
+   samp->xpot = util_unsigned_logbase2( texture->width0 );
+   samp->ypot = util_unsigned_logbase2( texture->height0 );
    samp->level = CLAMP((int) sampler->min_lod, 0, (int) texture->last_level);
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
index 407a22a9f4..e50a76a73b 100644
--- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
+++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
@@ -35,6 +35,7 @@
 #include "pipe/p_inlines.h"
 #include "util/u_memory.h"
 #include "util/u_tile.h"
+#include "util/u_math.h"
 #include "sp_context.h"
 #include "sp_surface.h"
 #include "sp_texture.h"
@@ -246,9 +247,9 @@ sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc,
                                      addr.bits.level, 
                                      addr.bits.z, 
                                      PIPE_TRANSFER_READ, 0, 0,
-                                     tc->texture->width[addr.bits.level],
-                                     tc->texture->height[addr.bits.level]);
-
+                                     u_minify(tc->texture->width0, addr.bits.level),
+                                     u_minify(tc->texture->height0, addr.bits.level));
+         
          tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans);
 
          tc->tex_face = addr.bits.face;
diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index 7caf2928b4..ac5f61e46f 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -52,16 +52,17 @@ softpipe_texture_layout(struct pipe_screen *screen,
 {
    struct pipe_texture *pt = &spt->base;
    unsigned level;
-   unsigned width = pt->width[0];
-   unsigned height = pt->height[0];
-   unsigned depth = pt->depth[0];
+   unsigned width = pt->width0;
+   unsigned height = pt->height0;
+   unsigned depth = pt->depth0;
 
    unsigned buffer_size = 0;
 
+   pt->width0 = width;
+   pt->height0 = height;
+   pt->depth0 = depth;
+
    for (level = 0; level <= pt->last_level; level++) {
-      pt->width[level] = width;
-      pt->height[level] = height;
-      pt->depth[level] = depth;
       pt->nblocksx[level] = pf_get_nblocksx(&pt->block, width);  
       pt->nblocksy[level] = pf_get_nblocksy(&pt->block, height);  
       spt->stride[level] = pt->nblocksx[level]*pt->block.size;
@@ -72,9 +73,9 @@ softpipe_texture_layout(struct pipe_screen *screen,
                       ((pt->target == PIPE_TEXTURE_CUBE) ? 6 : depth) *
                       spt->stride[level]);
 
-      width  = minify(width);
-      height = minify(height);
-      depth = minify(depth);
+      width  = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
    }
 
    spt->buffer = screen->buffer_create(screen, 32,
@@ -96,12 +97,12 @@ softpipe_displaytarget_layout(struct pipe_screen *screen,
                      PIPE_BUFFER_USAGE_GPU_READ_WRITE);
    unsigned tex_usage = spt->base.tex_usage;
 
-   spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width[0]);  
-   spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height[0]);  
+   spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width0);  
+   spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height0);  
 
    spt->buffer = screen->surface_buffer_create( screen, 
-                                                spt->base.width[0], 
-                                                spt->base.height[0],
+                                                spt->base.width0, 
+                                                spt->base.height0,
                                                 spt->base.format,
                                                 usage,
                                                 tex_usage,
@@ -126,9 +127,9 @@ softpipe_texture_create(struct pipe_screen *screen,
    pipe_reference_init(&spt->base.reference, 1);
    spt->base.screen = screen;
 
-   spt->pot = (util_is_power_of_two(template->width[0]) &&
-               util_is_power_of_two(template->height[0]) &&
-               util_is_power_of_two(template->depth[0]));
+   spt->pot = (util_is_power_of_two(template->width0) &&
+               util_is_power_of_two(template->height0) &&
+               util_is_power_of_two(template->depth0));
 
    if (spt->base.tex_usage & (PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
                               PIPE_TEXTURE_USAGE_PRIMARY)) {
@@ -163,7 +164,7 @@ softpipe_texture_blanket(struct pipe_screen * screen,
    /* Only supports one type */
    if (base->target != PIPE_TEXTURE_2D ||
        base->last_level != 0 ||
-       base->depth[0] != 1) {
+       base->depth0 != 1) {
       return NULL;
    }
 
@@ -174,8 +175,8 @@ softpipe_texture_blanket(struct pipe_screen * screen,
    spt->base = *base;
    pipe_reference_init(&spt->base.reference, 1);
    spt->base.screen = screen;
-   spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width[0]);  
-   spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height[0]);  
+   spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width0);  
+   spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height0);  
    spt->stride[0] = stride[0];
 
    pipe_buffer_reference(&spt->buffer, buffer);
@@ -213,8 +214,8 @@ softpipe_get_tex_surface(struct pipe_screen *screen,
       pipe_reference_init(&ps->reference, 1);
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
+      ps->width = u_minify(pt->width0, level);
+      ps->height = u_minify(pt->height0, level);
       ps->offset = spt->level_offset[level];
       ps->usage = usage;
 
@@ -434,9 +435,9 @@ softpipe_video_surface_create(struct pipe_screen *screen,
    template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
    template.last_level = 0;
    /* vl_mpeg12_mc_renderer expects this when it's initialized with pot_buffers=true */
-   template.width[0] = util_next_power_of_two(width);
-   template.height[0] = util_next_power_of_two(height);
-   template.depth[0] = 1;
+   template.width0 = util_next_power_of_two(width);
+   template.height0 = util_next_power_of_two(height);
+   template.depth0 = 1;
    pf_get_block(template.format, &template.block);
    template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET;
 
diff --git a/src/gallium/drivers/svga/Makefile b/src/gallium/drivers/svga/Makefile
new file mode 100644
index 0000000000..38b63394e3
--- /dev/null
+++ b/src/gallium/drivers/svga/Makefile
@@ -0,0 +1,57 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = svga
+
+C_SOURCES = \
+	svgadump/svga_shader_dump.c \
+	svgadump/svga_shader_op.c \
+	svgadump/svga_dump.c \
+	svga_cmd.c \
+	svga_context.c \
+	svga_draw.c \
+	svga_draw_arrays.c \
+	svga_draw_elements.c \
+	svga_pipe_blend.c \
+	svga_pipe_blit.c \
+	svga_pipe_clear.c \
+	svga_pipe_constants.c \
+	svga_pipe_depthstencil.c \
+	svga_pipe_draw.c \
+	svga_pipe_flush.c \
+	svga_pipe_fs.c \
+	svga_pipe_misc.c \
+	svga_pipe_query.c \
+	svga_pipe_rasterizer.c \
+	svga_pipe_sampler.c \
+	svga_pipe_vertex.c \
+	svga_pipe_vs.c \
+	svga_screen.c \
+	svga_screen_buffer.c \
+	svga_screen_texture.c \
+	svga_screen_cache.c \
+	svga_state.c \
+	svga_state_need_swtnl.c \
+	svga_state_constants.c \
+	svga_state_framebuffer.c \
+	svga_state_rss.c \
+	svga_state_tss.c \
+	svga_state_vdecl.c \
+	svga_state_fs.c \
+	svga_state_vs.c \
+	svga_swtnl_backend.c \
+	svga_swtnl_draw.c \
+	svga_swtnl_state.c \
+	svga_tgsi.c \
+	svga_tgsi_decl_sm20.c \
+	svga_tgsi_decl_sm30.c \
+	svga_tgsi_insn.c
+
+LIBRARY_INCLUDES = \
+	-I$(TOP)/src/gallium/drivers/svga/include
+
+LIBRARY_DEFINES = \
+	-std=gnu99 -fvisibility=hidden \
+	-DHAVE_STDINT_H -DHAVE_SYS_TYPES_H
+
+include ../../Makefile.template
diff --git a/src/gallium/drivers/svga/SConscript b/src/gallium/drivers/svga/SConscript
new file mode 100644
index 0000000000..737b791ceb
--- /dev/null
+++ b/src/gallium/drivers/svga/SConscript
@@ -0,0 +1,72 @@
+Import('*')
+
+env = env.Clone()
+
+if env['platform'] in ['linux']:
+	env.Append(CCFLAGS = ['-fvisibility=hidden'])
+
+if env['gcc']:
+	env.Append(CPPDEFINES = [
+		'HAVE_STDINT_H', 
+		'HAVE_SYS_TYPES_H',
+	])
+	
+env.Prepend(CPPPATH = [
+	'include',
+])
+
+env.Append(CPPDEFINES = [
+])
+
+sources = [
+    'svga_cmd.c',
+    'svga_context.c',
+    'svga_draw.c',
+    'svga_draw_arrays.c',
+    'svga_draw_elements.c',
+    'svga_pipe_blend.c',
+    'svga_pipe_blit.c',
+    'svga_pipe_clear.c',
+    'svga_pipe_constants.c',
+    'svga_pipe_depthstencil.c',
+    'svga_pipe_draw.c',
+    'svga_pipe_flush.c',
+    'svga_pipe_fs.c',
+    'svga_pipe_misc.c',
+    'svga_pipe_query.c',
+    'svga_pipe_rasterizer.c',
+    'svga_pipe_sampler.c',
+    'svga_pipe_vertex.c',
+    'svga_pipe_vs.c',
+    'svga_screen.c',
+    'svga_screen_buffer.c',
+    'svga_screen_cache.c',
+    'svga_screen_texture.c',
+    'svga_state.c',
+    'svga_state_constants.c',
+    'svga_state_framebuffer.c',
+    'svga_state_need_swtnl.c',
+    'svga_state_rss.c',
+    'svga_state_tss.c',
+    'svga_state_vdecl.c',
+    'svga_state_fs.c',
+    'svga_state_vs.c',
+    'svga_swtnl_backend.c',
+    'svga_swtnl_draw.c',
+    'svga_swtnl_state.c',
+    'svga_tgsi.c',
+    'svga_tgsi_decl_sm20.c',
+    'svga_tgsi_decl_sm30.c',
+    'svga_tgsi_insn.c',
+    
+    'svgadump/svga_dump.c',
+    'svgadump/svga_shader_dump.c',
+    'svgadump/svga_shader_op.c',
+]
+
+svga = env.ConvenienceLibrary(
+	target = 'svga',
+	source = sources,
+)
+
+Export('svga')
diff --git a/src/gallium/drivers/svga/include/README b/src/gallium/drivers/svga/include/README
new file mode 100644
index 0000000000..a0b8916104
--- /dev/null
+++ b/src/gallium/drivers/svga/include/README
@@ -0,0 +1,3 @@
+This directory contains the headers from the VMware SVGA Device Developer Kit:
+
+   https://vmware-svga.svn.sourceforge.net/svnroot/vmware-svga/trunk/lib/vmware/
diff --git a/src/gallium/drivers/svga/include/svga3d_caps.h b/src/gallium/drivers/svga/include/svga3d_caps.h
new file mode 100644
index 0000000000..714ce9f45f
--- /dev/null
+++ b/src/gallium/drivers/svga/include/svga3d_caps.h
@@ -0,0 +1,139 @@
+/**********************************************************
+ * Copyright 2007-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/*
+ * svga3d_caps.h --
+ *
+ *       Definitions for SVGA3D hardware capabilities.  Capabilities
+ *       are used to query for optional rendering features during
+ *       driver initialization. The capability data is stored as very
+ *       basic key/value dictionary within the "FIFO register" memory
+ *       area at the beginning of BAR2.
+ *
+ *       Note that these definitions are only for 3D capabilities.
+ *       The SVGA device also has "device capabilities" and "FIFO
+ *       capabilities", which are non-3D-specific and are stored as
+ *       bitfields rather than key/value pairs.
+ */
+
+#ifndef _SVGA3D_CAPS_H_
+#define _SVGA3D_CAPS_H_
+
+#define SVGA_FIFO_3D_CAPS_SIZE   (SVGA_FIFO_3D_CAPS_LAST - \
+                                  SVGA_FIFO_3D_CAPS + 1)
+
+
+/*
+ * SVGA3dCapsRecordType
+ *
+ *    Record types that can be found in the caps block.
+ *    Related record types are grouped together numerically so that
+ *    SVGA3dCaps_FindRecord() can be applied on a range of record
+ *    types.
+ */
+
+typedef enum {
+   SVGA3DCAPS_RECORD_UNKNOWN        = 0,
+   SVGA3DCAPS_RECORD_DEVCAPS_MIN    = 0x100,
+   SVGA3DCAPS_RECORD_DEVCAPS        = 0x100,
+   SVGA3DCAPS_RECORD_DEVCAPS_MAX    = 0x1ff,
+} SVGA3dCapsRecordType;
+
+
+/*
+ * SVGA3dCapsRecordHeader
+ *
+ *    Header field leading each caps block record. Contains the offset (in
+ *    register words, NOT bytes) to the next caps block record (or the end
+ *    of caps block records which will be a zero word) and the record type
+ *    as defined above.
+ */
+
+typedef
+struct SVGA3dCapsRecordHeader {
+   uint32 length;
+   SVGA3dCapsRecordType type;
+}
+SVGA3dCapsRecordHeader;
+
+
+/*
+ * SVGA3dCapsRecord
+ *
+ *    Caps block record; "data" is a placeholder for the actual data structure
+ *    contained within the record; for example a record containing a FOOBAR
+ *    structure would be of size "sizeof(SVGA3dCapsRecordHeader) +
+ *    sizeof(FOOBAR)".
+ */
+
+typedef
+struct SVGA3dCapsRecord {
+   SVGA3dCapsRecordHeader header;
+   uint32 data[1];
+}
+SVGA3dCapsRecord;
+
+
+typedef uint32 SVGA3dCapPair[2];
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3dCaps_FindRecord
+ *
+ *    Finds the record with the highest-valued type within the given range
+ *    in the caps block.
+ *
+ *    Result: pointer to found record, or NULL if not found.
+ *
+ *----------------------------------------------------------------------
+ */
+
+static INLINE SVGA3dCapsRecord *
+SVGA3dCaps_FindRecord(const uint32 *capsBlock,
+                      SVGA3dCapsRecordType recordTypeMin,
+                      SVGA3dCapsRecordType recordTypeMax)
+{
+   SVGA3dCapsRecord *record, *found = NULL;
+   uint32 offset;
+
+   /*
+    * Search linearly through the caps block records for the specified type.
+    */
+   for (offset = 0; capsBlock[offset] != 0; offset += capsBlock[offset]) {
+      record = (SVGA3dCapsRecord *) (capsBlock + offset);
+      if ((record->header.type >= recordTypeMin) &&
+          (record->header.type <= recordTypeMax) &&
+          (!found || (record->header.type > found->header.type))) {
+         found = record;
+      }
+   }
+
+   return found;
+}
+
+
+#endif // _SVGA3D_CAPS_H_
diff --git a/src/gallium/drivers/svga/include/svga3d_reg.h b/src/gallium/drivers/svga/include/svga3d_reg.h
new file mode 100644
index 0000000000..77cb453310
--- /dev/null
+++ b/src/gallium/drivers/svga/include/svga3d_reg.h
@@ -0,0 +1,1793 @@
+/**********************************************************
+ * Copyright 1998-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/*
+ * svga3d_reg.h --
+ *
+ *       SVGA 3D hardware definitions
+ */
+
+#ifndef _SVGA3D_REG_H_
+#define _SVGA3D_REG_H_
+
+#include "svga_reg.h"
+
+
+/*
+ * 3D Hardware Version
+ *
+ *   The hardware version is stored in the SVGA_FIFO_3D_HWVERSION fifo
+ *   register.   Is set by the host and read by the guest.  This lets
+ *   us make new guest drivers which are backwards-compatible with old
+ *   SVGA hardware revisions.  It does not let us support old guest
+ *   drivers.  Good enough for now.
+ *
+ */
+
+#define SVGA3D_MAKE_HWVERSION(major, minor)      (((major) << 16) | ((minor) & 0xFF))
+#define SVGA3D_MAJOR_HWVERSION(version)          ((version) >> 16)
+#define SVGA3D_MINOR_HWVERSION(version)          ((version) & 0xFF)
+
+typedef enum {
+   SVGA3D_HWVERSION_WS5_RC1   = SVGA3D_MAKE_HWVERSION(0, 1),
+   SVGA3D_HWVERSION_WS5_RC2   = SVGA3D_MAKE_HWVERSION(0, 2),
+   SVGA3D_HWVERSION_WS51_RC1  = SVGA3D_MAKE_HWVERSION(0, 3),
+   SVGA3D_HWVERSION_WS6_B1    = SVGA3D_MAKE_HWVERSION(1, 1),
+   SVGA3D_HWVERSION_FUSION_11 = SVGA3D_MAKE_HWVERSION(1, 4),
+   SVGA3D_HWVERSION_WS65_B1   = SVGA3D_MAKE_HWVERSION(2, 0),
+   SVGA3D_HWVERSION_CURRENT   = SVGA3D_HWVERSION_WS65_B1,
+} SVGA3dHardwareVersion;
+
+/*
+ * Generic Types
+ */
+
+typedef uint32 SVGA3dBool; /* 32-bit Bool definition */
+#define SVGA3D_NUM_CLIPPLANES                   6
+#define SVGA3D_MAX_SIMULTANEOUS_RENDER_TARGETS  8
+
+
+/*
+ * Surface formats.
+ *
+ * If you modify this list, be sure to keep GLUtil.c in sync. It
+ * includes the internal format definition of each surface in
+ * GLUtil_ConvertSurfaceFormat, and it contains a table of
+ * human-readable names in GLUtil_GetFormatName.
+ */
+
+typedef enum SVGA3dSurfaceFormat {
+   SVGA3D_FORMAT_INVALID = 0,
+
+   SVGA3D_X8R8G8B8       = 1,
+   SVGA3D_A8R8G8B8       = 2,
+
+   SVGA3D_R5G6B5         = 3,
+   SVGA3D_X1R5G5B5       = 4,
+   SVGA3D_A1R5G5B5       = 5,
+   SVGA3D_A4R4G4B4       = 6,
+
+   SVGA3D_Z_D32          = 7,
+   SVGA3D_Z_D16          = 8,
+   SVGA3D_Z_D24S8        = 9,
+   SVGA3D_Z_D15S1        = 10,
+
+   SVGA3D_LUMINANCE8            = 11,
+   SVGA3D_LUMINANCE4_ALPHA4     = 12,
+   SVGA3D_LUMINANCE16           = 13,
+   SVGA3D_LUMINANCE8_ALPHA8     = 14,
+
+   SVGA3D_DXT1           = 15,
+   SVGA3D_DXT2           = 16,
+   SVGA3D_DXT3           = 17,
+   SVGA3D_DXT4           = 18,
+   SVGA3D_DXT5           = 19,
+
+   SVGA3D_BUMPU8V8       = 20,
+   SVGA3D_BUMPL6V5U5     = 21,
+   SVGA3D_BUMPX8L8V8U8   = 22,
+   SVGA3D_BUMPL8V8U8     = 23,
+
+   SVGA3D_ARGB_S10E5     = 24,   /* 16-bit floating-point ARGB */
+   SVGA3D_ARGB_S23E8     = 25,   /* 32-bit floating-point ARGB */
+
+   SVGA3D_A2R10G10B10    = 26,
+
+   /* signed formats */
+   SVGA3D_V8U8           = 27,
+   SVGA3D_Q8W8V8U8       = 28,
+   SVGA3D_CxV8U8         = 29,
+
+   /* mixed formats */
+   SVGA3D_X8L8V8U8       = 30,
+   SVGA3D_A2W10V10U10    = 31,
+
+   SVGA3D_ALPHA8         = 32,
+
+   /* Single- and dual-component floating point formats */
+   SVGA3D_R_S10E5        = 33,
+   SVGA3D_R_S23E8        = 34,
+   SVGA3D_RG_S10E5       = 35,
+   SVGA3D_RG_S23E8       = 36,
+
+   /*
+    * Any surface can be used as a buffer object, but SVGA3D_BUFFER is
+    * the most efficient format to use when creating new surfaces
+    * expressly for index or vertex data.
+    */
+   SVGA3D_BUFFER         = 37,
+
+   SVGA3D_Z_D24X8        = 38,
+
+   SVGA3D_V16U16         = 39,
+
+   SVGA3D_G16R16         = 40,
+   SVGA3D_A16B16G16R16   = 41,
+
+   /* Packed Video formats */
+   SVGA3D_UYVY           = 42,
+   SVGA3D_YUY2           = 43,
+
+   SVGA3D_FORMAT_MAX
+} SVGA3dSurfaceFormat;
+
+typedef uint32 SVGA3dColor; /* a, r, g, b */
+
+/*
+ * These match the D3DFORMAT_OP definitions used by Direct3D. We need
+ * them so that we can query the host for what the supported surface
+ * operations are (when we're using the D3D backend, in particular),
+ * and so we can send those operations to the guest.
+ */
+typedef enum {
+   SVGA3DFORMAT_OP_TEXTURE                               = 0x00000001,
+   SVGA3DFORMAT_OP_VOLUMETEXTURE                         = 0x00000002,
+   SVGA3DFORMAT_OP_CUBETEXTURE                           = 0x00000004,
+   SVGA3DFORMAT_OP_OFFSCREEN_RENDERTARGET                = 0x00000008,
+   SVGA3DFORMAT_OP_SAME_FORMAT_RENDERTARGET              = 0x00000010,
+   SVGA3DFORMAT_OP_ZSTENCIL                              = 0x00000040,
+   SVGA3DFORMAT_OP_ZSTENCIL_WITH_ARBITRARY_COLOR_DEPTH   = 0x00000080,
+
+/*
+ * This format can be used as a render target if the current display mode
+ * is the same depth if the alpha channel is ignored. e.g. if the device
+ * can render to A8R8G8B8 when the display mode is X8R8G8B8, then the
+ * format op list entry for A8R8G8B8 should have this cap.
+ */
+   SVGA3DFORMAT_OP_SAME_FORMAT_UP_TO_ALPHA_RENDERTARGET  = 0x00000100,
+
+/*
+ * This format contains DirectDraw support (including Flip).  This flag
+ * should not to be set on alpha formats.
+ */
+   SVGA3DFORMAT_OP_DISPLAYMODE                           = 0x00000400,
+
+/*
+ * The rasterizer can support some level of Direct3D support in this format
+ * and implies that the driver can create a Context in this mode (for some
+ * render target format).  When this flag is set, the SVGA3DFORMAT_OP_DISPLAYMODE
+ * flag must also be set.
+ */
+   SVGA3DFORMAT_OP_3DACCELERATION                        = 0x00000800,
+
+/*
+ * This is set for a private format when the driver has put the bpp in
+ * the structure.
+ */
+   SVGA3DFORMAT_OP_PIXELSIZE                             = 0x00001000,
+
+/*
+ * Indicates that this format can be converted to any RGB format for which
+ * SVGA3DFORMAT_OP_MEMBEROFGROUP_ARGB is specified
+ */
+   SVGA3DFORMAT_OP_CONVERT_TO_ARGB                       = 0x00002000,
+
+/*
+ * Indicates that this format can be used to create offscreen plain surfaces.
+ */
+   SVGA3DFORMAT_OP_OFFSCREENPLAIN                        = 0x00004000,
+
+/*
+ * Indicated that this format can be read as an SRGB texture (meaning that the
+ * sampler will linearize the looked up data)
+ */
+   SVGA3DFORMAT_OP_SRGBREAD                              = 0x00008000,
+
+/*
+ * Indicates that this format can be used in the bumpmap instructions
+ */
+   SVGA3DFORMAT_OP_BUMPMAP                               = 0x00010000,
+
+/*
+ * Indicates that this format can be sampled by the displacement map sampler
+ */
+   SVGA3DFORMAT_OP_DMAP                                  = 0x00020000,
+
+/*
+ * Indicates that this format cannot be used with texture filtering
+ */
+   SVGA3DFORMAT_OP_NOFILTER                              = 0x00040000,
+
+/*
+ * Indicates that format conversions are supported to this RGB format if
+ * SVGA3DFORMAT_OP_CONVERT_TO_ARGB is specified in the source format.
+ */
+   SVGA3DFORMAT_OP_MEMBEROFGROUP_ARGB                    = 0x00080000,
+
+/*
+ * Indicated that this format can be written as an SRGB target (meaning that the
+ * pixel pipe will DE-linearize data on output to format)
+ */
+   SVGA3DFORMAT_OP_SRGBWRITE                             = 0x00100000,
+
+/*
+ * Indicates that this format cannot be used with alpha blending
+ */
+   SVGA3DFORMAT_OP_NOALPHABLEND                          = 0x00200000,
+
+/*
+ * Indicates that the device can auto-generated sublevels for resources
+ * of this format
+ */
+   SVGA3DFORMAT_OP_AUTOGENMIPMAP                         = 0x00400000,
+
+/*
+ * Indicates that this format can be used by vertex texture sampler
+ */
+   SVGA3DFORMAT_OP_VERTEXTEXTURE                         = 0x00800000,
+
+/*
+ * Indicates that this format supports neither texture coordinate wrap
+ * modes, nor mipmapping
+ */
+   SVGA3DFORMAT_OP_NOTEXCOORDWRAPNORMIP                  = 0x01000000
+} SVGA3dFormatOp;
+
+/*
+ * This structure is a conversion of SVGA3DFORMAT_OP_*.
+ * Entries must be located at the same position.
+ */
+typedef union {
+   uint32 value;
+   struct {
+      uint32 texture : 1;
+      uint32 volumeTexture : 1;
+      uint32 cubeTexture : 1;
+      uint32 offscreenRenderTarget : 1;
+      uint32 sameFormatRenderTarget : 1;
+      uint32 unknown1 : 1;
+      uint32 zStencil : 1;
+      uint32 zStencilArbitraryDepth : 1;
+      uint32 sameFormatUpToAlpha : 1;
+      uint32 unknown2 : 1;
+      uint32 displayMode : 1;
+      uint32 acceleration3d : 1;
+      uint32 pixelSize : 1;
+      uint32 convertToARGB : 1;
+      uint32 offscreenPlain : 1;
+      uint32 sRGBRead : 1;
+      uint32 bumpMap : 1;
+      uint32 dmap : 1;
+      uint32 noFilter : 1;
+      uint32 memberOfGroupARGB : 1;
+      uint32 sRGBWrite : 1;
+      uint32 noAlphaBlend : 1;
+      uint32 autoGenMipMap : 1;
+      uint32 vertexTexture : 1;
+      uint32 noTexCoordWrapNorMip : 1;
+   };
+} SVGA3dSurfaceFormatCaps;
+
+/*
+ * SVGA_3D_CMD_SETRENDERSTATE Types.  All value types
+ * must fit in a uint32.
+ */
+
+typedef enum {
+   SVGA3D_RS_INVALID                   = 0,
+   SVGA3D_RS_ZENABLE                   = 1,     /* SVGA3dBool */
+   SVGA3D_RS_ZWRITEENABLE              = 2,     /* SVGA3dBool */
+   SVGA3D_RS_ALPHATESTENABLE           = 3,     /* SVGA3dBool */
+   SVGA3D_RS_DITHERENABLE              = 4,     /* SVGA3dBool */
+   SVGA3D_RS_BLENDENABLE               = 5,     /* SVGA3dBool */
+   SVGA3D_RS_FOGENABLE                 = 6,     /* SVGA3dBool */
+   SVGA3D_RS_SPECULARENABLE            = 7,     /* SVGA3dBool */
+   SVGA3D_RS_STENCILENABLE             = 8,     /* SVGA3dBool */
+   SVGA3D_RS_LIGHTINGENABLE            = 9,     /* SVGA3dBool */
+   SVGA3D_RS_NORMALIZENORMALS          = 10,    /* SVGA3dBool */
+   SVGA3D_RS_POINTSPRITEENABLE         = 11,    /* SVGA3dBool */
+   SVGA3D_RS_POINTSCALEENABLE          = 12,    /* SVGA3dBool */
+   SVGA3D_RS_STENCILREF                = 13,    /* uint32 */
+   SVGA3D_RS_STENCILMASK               = 14,    /* uint32 */
+   SVGA3D_RS_STENCILWRITEMASK          = 15,    /* uint32 */
+   SVGA3D_RS_FOGSTART                  = 16,    /* float */
+   SVGA3D_RS_FOGEND                    = 17,    /* float */
+   SVGA3D_RS_FOGDENSITY                = 18,    /* float */
+   SVGA3D_RS_POINTSIZE                 = 19,    /* float */
+   SVGA3D_RS_POINTSIZEMIN              = 20,    /* float */
+   SVGA3D_RS_POINTSIZEMAX              = 21,    /* float */
+   SVGA3D_RS_POINTSCALE_A              = 22,    /* float */
+   SVGA3D_RS_POINTSCALE_B              = 23,    /* float */
+   SVGA3D_RS_POINTSCALE_C              = 24,    /* float */
+   SVGA3D_RS_FOGCOLOR                  = 25,    /* SVGA3dColor */
+   SVGA3D_RS_AMBIENT                   = 26,    /* SVGA3dColor */
+   SVGA3D_RS_CLIPPLANEENABLE           = 27,    /* SVGA3dClipPlanes */
+   SVGA3D_RS_FOGMODE                   = 28,    /* SVGA3dFogMode */
+   SVGA3D_RS_FILLMODE                  = 29,    /* SVGA3dFillMode */
+   SVGA3D_RS_SHADEMODE                 = 30,    /* SVGA3dShadeMode */
+   SVGA3D_RS_LINEPATTERN               = 31,    /* SVGA3dLinePattern */
+   SVGA3D_RS_SRCBLEND                  = 32,    /* SVGA3dBlendOp */
+   SVGA3D_RS_DSTBLEND                  = 33,    /* SVGA3dBlendOp */
+   SVGA3D_RS_BLENDEQUATION             = 34,    /* SVGA3dBlendEquation */
+   SVGA3D_RS_CULLMODE                  = 35,    /* SVGA3dFace */
+   SVGA3D_RS_ZFUNC                     = 36,    /* SVGA3dCmpFunc */
+   SVGA3D_RS_ALPHAFUNC                 = 37,    /* SVGA3dCmpFunc */
+   SVGA3D_RS_STENCILFUNC               = 38,    /* SVGA3dCmpFunc */
+   SVGA3D_RS_STENCILFAIL               = 39,    /* SVGA3dStencilOp */
+   SVGA3D_RS_STENCILZFAIL              = 40,    /* SVGA3dStencilOp */
+   SVGA3D_RS_STENCILPASS               = 41,    /* SVGA3dStencilOp */
+   SVGA3D_RS_ALPHAREF                  = 42,    /* float (0.0 .. 1.0) */
+   SVGA3D_RS_FRONTWINDING              = 43,    /* SVGA3dFrontWinding */
+   SVGA3D_RS_COORDINATETYPE            = 44,    /* SVGA3dCoordinateType */
+   SVGA3D_RS_ZBIAS                     = 45,    /* float */
+   SVGA3D_RS_RANGEFOGENABLE            = 46,    /* SVGA3dBool */
+   SVGA3D_RS_COLORWRITEENABLE          = 47,    /* SVGA3dColorMask */
+   SVGA3D_RS_VERTEXMATERIALENABLE      = 48,    /* SVGA3dBool */
+   SVGA3D_RS_DIFFUSEMATERIALSOURCE     = 49,    /* SVGA3dVertexMaterial */
+   SVGA3D_RS_SPECULARMATERIALSOURCE    = 50,    /* SVGA3dVertexMaterial */
+   SVGA3D_RS_AMBIENTMATERIALSOURCE     = 51,    /* SVGA3dVertexMaterial */
+   SVGA3D_RS_EMISSIVEMATERIALSOURCE    = 52,    /* SVGA3dVertexMaterial */
+   SVGA3D_RS_TEXTUREFACTOR             = 53,    /* SVGA3dColor */
+   SVGA3D_RS_LOCALVIEWER               = 54,    /* SVGA3dBool */
+   SVGA3D_RS_SCISSORTESTENABLE         = 55,    /* SVGA3dBool */
+   SVGA3D_RS_BLENDCOLOR                = 56,    /* SVGA3dColor */
+   SVGA3D_RS_STENCILENABLE2SIDED       = 57,    /* SVGA3dBool */
+   SVGA3D_RS_CCWSTENCILFUNC            = 58,    /* SVGA3dCmpFunc */
+   SVGA3D_RS_CCWSTENCILFAIL            = 59,    /* SVGA3dStencilOp */
+   SVGA3D_RS_CCWSTENCILZFAIL           = 60,    /* SVGA3dStencilOp */
+   SVGA3D_RS_CCWSTENCILPASS            = 61,    /* SVGA3dStencilOp */
+   SVGA3D_RS_VERTEXBLEND               = 62,    /* SVGA3dVertexBlendFlags */
+   SVGA3D_RS_SLOPESCALEDEPTHBIAS       = 63,    /* float */
+   SVGA3D_RS_DEPTHBIAS                 = 64,    /* float */
+
+
+   /*
+    * Output Gamma Level
+    *
+    * Output gamma effects the gamma curve of colors that are output from the
+    * rendering pipeline.  A value of 1.0 specifies a linear color space. If the
+    * value is <= 0.0, gamma correction is ignored and linear color space is
+    * used.
+    */
+
+   SVGA3D_RS_OUTPUTGAMMA               = 65,    /* float */
+   SVGA3D_RS_ZVISIBLE                  = 66,    /* SVGA3dBool */
+   SVGA3D_RS_LASTPIXEL                 = 67,    /* SVGA3dBool */
+   SVGA3D_RS_CLIPPING                  = 68,    /* SVGA3dBool */
+   SVGA3D_RS_WRAP0                     = 69,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP1                     = 70,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP2                     = 71,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP3                     = 72,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP4                     = 73,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP5                     = 74,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP6                     = 75,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP7                     = 76,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP8                     = 77,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP9                     = 78,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP10                    = 79,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP11                    = 80,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP12                    = 81,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP13                    = 82,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP14                    = 83,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_WRAP15                    = 84,    /* SVGA3dWrapFlags */
+   SVGA3D_RS_MULTISAMPLEANTIALIAS      = 85,    /* SVGA3dBool */
+   SVGA3D_RS_MULTISAMPLEMASK           = 86,    /* uint32 */
+   SVGA3D_RS_INDEXEDVERTEXBLENDENABLE  = 87,    /* SVGA3dBool */
+   SVGA3D_RS_TWEENFACTOR               = 88,    /* float */
+   SVGA3D_RS_ANTIALIASEDLINEENABLE     = 89,    /* SVGA3dBool */
+   SVGA3D_RS_COLORWRITEENABLE1         = 90,    /* SVGA3dColorMask */
+   SVGA3D_RS_COLORWRITEENABLE2         = 91,    /* SVGA3dColorMask */
+   SVGA3D_RS_COLORWRITEENABLE3         = 92,    /* SVGA3dColorMask */
+   SVGA3D_RS_SEPARATEALPHABLENDENABLE  = 93,    /* SVGA3dBool */
+   SVGA3D_RS_SRCBLENDALPHA             = 94,    /* SVGA3dBlendOp */
+   SVGA3D_RS_DSTBLENDALPHA             = 95,    /* SVGA3dBlendOp */
+   SVGA3D_RS_BLENDEQUATIONALPHA        = 96,    /* SVGA3dBlendEquation */
+   SVGA3D_RS_MAX
+} SVGA3dRenderStateName;
+
+typedef enum {
+   SVGA3D_VERTEXMATERIAL_NONE     = 0,    /* Use the value in the current material */
+   SVGA3D_VERTEXMATERIAL_DIFFUSE  = 1,    /* Use the value in the diffuse component */
+   SVGA3D_VERTEXMATERIAL_SPECULAR = 2,    /* Use the value in the specular component */
+} SVGA3dVertexMaterial;
+
+typedef enum {
+   SVGA3D_FILLMODE_INVALID = 0,
+   SVGA3D_FILLMODE_POINT   = 1,
+   SVGA3D_FILLMODE_LINE    = 2,
+   SVGA3D_FILLMODE_FILL    = 3,
+   SVGA3D_FILLMODE_MAX
+} SVGA3dFillModeType;
+
+
+typedef
+union {
+   struct {
+      uint16   mode;       /* SVGA3dFillModeType */
+      uint16   face;       /* SVGA3dFace */
+   };
+   uint32 uintValue;
+} SVGA3dFillMode;
+
+typedef enum {
+   SVGA3D_SHADEMODE_INVALID = 0,
+   SVGA3D_SHADEMODE_FLAT    = 1,
+   SVGA3D_SHADEMODE_SMOOTH  = 2,
+   SVGA3D_SHADEMODE_PHONG   = 3,     /* Not supported */
+   SVGA3D_SHADEMODE_MAX
+} SVGA3dShadeMode;
+
+typedef
+union {
+   struct {
+      uint16 repeat;
+      uint16 pattern;
+   };
+   uint32 uintValue;
+} SVGA3dLinePattern;
+
+typedef enum {
+   SVGA3D_BLENDOP_INVALID            = 0,
+   SVGA3D_BLENDOP_ZERO               = 1,
+   SVGA3D_BLENDOP_ONE                = 2,
+   SVGA3D_BLENDOP_SRCCOLOR           = 3,
+   SVGA3D_BLENDOP_INVSRCCOLOR        = 4,
+   SVGA3D_BLENDOP_SRCALPHA           = 5,
+   SVGA3D_BLENDOP_INVSRCALPHA        = 6,
+   SVGA3D_BLENDOP_DESTALPHA          = 7,
+   SVGA3D_BLENDOP_INVDESTALPHA       = 8,
+   SVGA3D_BLENDOP_DESTCOLOR          = 9,
+   SVGA3D_BLENDOP_INVDESTCOLOR       = 10,
+   SVGA3D_BLENDOP_SRCALPHASAT        = 11,
+   SVGA3D_BLENDOP_BLENDFACTOR        = 12,
+   SVGA3D_BLENDOP_INVBLENDFACTOR     = 13,
+   SVGA3D_BLENDOP_MAX
+} SVGA3dBlendOp;
+
+typedef enum {
+   SVGA3D_BLENDEQ_INVALID            = 0,
+   SVGA3D_BLENDEQ_ADD                = 1,
+   SVGA3D_BLENDEQ_SUBTRACT           = 2,
+   SVGA3D_BLENDEQ_REVSUBTRACT        = 3,
+   SVGA3D_BLENDEQ_MINIMUM            = 4,
+   SVGA3D_BLENDEQ_MAXIMUM            = 5,
+   SVGA3D_BLENDEQ_MAX
+} SVGA3dBlendEquation;
+
+typedef enum {
+   SVGA3D_FRONTWINDING_INVALID = 0,
+   SVGA3D_FRONTWINDING_CW      = 1,
+   SVGA3D_FRONTWINDING_CCW     = 2,
+   SVGA3D_FRONTWINDING_MAX
+} SVGA3dFrontWinding;
+
+typedef enum {
+   SVGA3D_FACE_INVALID  = 0,
+   SVGA3D_FACE_NONE     = 1,
+   SVGA3D_FACE_FRONT    = 2,
+   SVGA3D_FACE_BACK     = 3,
+   SVGA3D_FACE_FRONT_BACK = 4,
+   SVGA3D_FACE_MAX
+} SVGA3dFace;
+
+/*
+ * The order and the values should not be changed
+ */
+
+typedef enum {
+   SVGA3D_CMP_INVALID              = 0,
+   SVGA3D_CMP_NEVER                = 1,
+   SVGA3D_CMP_LESS                 = 2,
+   SVGA3D_CMP_EQUAL                = 3,
+   SVGA3D_CMP_LESSEQUAL            = 4,
+   SVGA3D_CMP_GREATER              = 5,
+   SVGA3D_CMP_NOTEQUAL             = 6,
+   SVGA3D_CMP_GREATEREQUAL         = 7,
+   SVGA3D_CMP_ALWAYS               = 8,
+   SVGA3D_CMP_MAX
+} SVGA3dCmpFunc;
+
+/*
+ * SVGA3D_FOGFUNC_* specifies the fog equation, or PER_VERTEX which allows
+ * the fog factor to be specified in the alpha component of the specular
+ * (a.k.a. secondary) vertex color.
+ */
+typedef enum {
+   SVGA3D_FOGFUNC_INVALID          = 0,
+   SVGA3D_FOGFUNC_EXP              = 1,
+   SVGA3D_FOGFUNC_EXP2             = 2,
+   SVGA3D_FOGFUNC_LINEAR           = 3,
+   SVGA3D_FOGFUNC_PER_VERTEX       = 4
+} SVGA3dFogFunction;
+
+/*
+ * SVGA3D_FOGTYPE_* specifies if fog factors are computed on a per-vertex
+ * or per-pixel basis.
+ */
+typedef enum {
+   SVGA3D_FOGTYPE_INVALID          = 0,
+   SVGA3D_FOGTYPE_VERTEX           = 1,
+   SVGA3D_FOGTYPE_PIXEL            = 2,
+   SVGA3D_FOGTYPE_MAX              = 3
+} SVGA3dFogType;
+
+/*
+ * SVGA3D_FOGBASE_* selects depth or range-based fog. Depth-based fog is
+ * computed using the eye Z value of each pixel (or vertex), whereas range-
+ * based fog is computed using the actual distance (range) to the eye.
+ */
+typedef enum {
+   SVGA3D_FOGBASE_INVALID          = 0,
+   SVGA3D_FOGBASE_DEPTHBASED       = 1,
+   SVGA3D_FOGBASE_RANGEBASED       = 2,
+   SVGA3D_FOGBASE_MAX              = 3
+} SVGA3dFogBase;
+
+typedef enum {
+   SVGA3D_STENCILOP_INVALID        = 0,
+   SVGA3D_STENCILOP_KEEP           = 1,
+   SVGA3D_STENCILOP_ZERO           = 2,
+   SVGA3D_STENCILOP_REPLACE        = 3,
+   SVGA3D_STENCILOP_INCRSAT        = 4,
+   SVGA3D_STENCILOP_DECRSAT        = 5,
+   SVGA3D_STENCILOP_INVERT         = 6,
+   SVGA3D_STENCILOP_INCR           = 7,
+   SVGA3D_STENCILOP_DECR           = 8,
+   SVGA3D_STENCILOP_MAX
+} SVGA3dStencilOp;
+
+typedef enum {
+   SVGA3D_CLIPPLANE_0              = (1 << 0),
+   SVGA3D_CLIPPLANE_1              = (1 << 1),
+   SVGA3D_CLIPPLANE_2              = (1 << 2),
+   SVGA3D_CLIPPLANE_3              = (1 << 3),
+   SVGA3D_CLIPPLANE_4              = (1 << 4),
+   SVGA3D_CLIPPLANE_5              = (1 << 5),
+} SVGA3dClipPlanes;
+
+typedef enum {
+   SVGA3D_CLEAR_COLOR              = 0x1,
+   SVGA3D_CLEAR_DEPTH              = 0x2,
+   SVGA3D_CLEAR_STENCIL            = 0x4
+} SVGA3dClearFlag;
+
+typedef enum {
+   SVGA3D_RT_DEPTH                 = 0,
+   SVGA3D_RT_STENCIL               = 1,
+   SVGA3D_RT_COLOR0                = 2,
+   SVGA3D_RT_COLOR1                = 3,
+   SVGA3D_RT_COLOR2                = 4,
+   SVGA3D_RT_COLOR3                = 5,
+   SVGA3D_RT_COLOR4                = 6,
+   SVGA3D_RT_COLOR5                = 7,
+   SVGA3D_RT_COLOR6                = 8,
+   SVGA3D_RT_COLOR7                = 9,
+   SVGA3D_RT_MAX,
+   SVGA3D_RT_INVALID               = ((uint32)-1),
+} SVGA3dRenderTargetType;
+
+#define SVGA3D_MAX_RT_COLOR (SVGA3D_RT_COLOR7 - SVGA3D_RT_COLOR0 + 1)
+
+typedef
+union {
+   struct {
+      uint32  red   : 1;
+      uint32  green : 1;
+      uint32  blue  : 1;
+      uint32  alpha : 1;
+   };
+   uint32 uintValue;
+} SVGA3dColorMask;
+
+typedef enum {
+   SVGA3D_VBLEND_DISABLE            = 0,
+   SVGA3D_VBLEND_1WEIGHT            = 1,
+   SVGA3D_VBLEND_2WEIGHT            = 2,
+   SVGA3D_VBLEND_3WEIGHT            = 3,
+} SVGA3dVertexBlendFlags;
+
+typedef enum {
+   SVGA3D_WRAPCOORD_0   = 1 << 0,
+   SVGA3D_WRAPCOORD_1   = 1 << 1,
+   SVGA3D_WRAPCOORD_2   = 1 << 2,
+   SVGA3D_WRAPCOORD_3   = 1 << 3,
+   SVGA3D_WRAPCOORD_ALL = 0xF,
+} SVGA3dWrapFlags;
+
+/*
+ * SVGA_3D_CMD_TEXTURESTATE Types.  All value types
+ * must fit in a uint32.
+ */
+
+typedef enum {
+   SVGA3D_TS_INVALID                    = 0,
+   SVGA3D_TS_BIND_TEXTURE               = 1,    /* SVGA3dSurfaceId */
+   SVGA3D_TS_COLOROP                    = 2,    /* SVGA3dTextureCombiner */
+   SVGA3D_TS_COLORARG1                  = 3,    /* SVGA3dTextureArgData */
+   SVGA3D_TS_COLORARG2                  = 4,    /* SVGA3dTextureArgData */
+   SVGA3D_TS_ALPHAOP                    = 5,    /* SVGA3dTextureCombiner */
+   SVGA3D_TS_ALPHAARG1                  = 6,    /* SVGA3dTextureArgData */
+   SVGA3D_TS_ALPHAARG2                  = 7,    /* SVGA3dTextureArgData */
+   SVGA3D_TS_ADDRESSU                   = 8,    /* SVGA3dTextureAddress */
+   SVGA3D_TS_ADDRESSV                   = 9,    /* SVGA3dTextureAddress */
+   SVGA3D_TS_MIPFILTER                  = 10,   /* SVGA3dTextureFilter */
+   SVGA3D_TS_MAGFILTER                  = 11,   /* SVGA3dTextureFilter */
+   SVGA3D_TS_MINFILTER                  = 12,   /* SVGA3dTextureFilter */
+   SVGA3D_TS_BORDERCOLOR                = 13,   /* SVGA3dColor */
+   SVGA3D_TS_TEXCOORDINDEX              = 14,   /* uint32 */
+   SVGA3D_TS_TEXTURETRANSFORMFLAGS      = 15,   /* SVGA3dTexTransformFlags */
+   SVGA3D_TS_TEXCOORDGEN                = 16,   /* SVGA3dTextureCoordGen */
+   SVGA3D_TS_BUMPENVMAT00               = 17,   /* float */
+   SVGA3D_TS_BUMPENVMAT01               = 18,   /* float */
+   SVGA3D_TS_BUMPENVMAT10               = 19,   /* float */
+   SVGA3D_TS_BUMPENVMAT11               = 20,   /* float */
+   SVGA3D_TS_TEXTURE_MIPMAP_LEVEL       = 21,   /* uint32 */
+   SVGA3D_TS_TEXTURE_LOD_BIAS           = 22,   /* float */
+   SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL  = 23,   /* uint32 */
+   SVGA3D_TS_ADDRESSW                   = 24,   /* SVGA3dTextureAddress */
+
+
+   /*
+    * Sampler Gamma Level
+    *
+    * Sampler gamma effects the color of samples taken from the sampler.  A
+    * value of 1.0 will produce linear samples.  If the value is <= 0.0 the
+    * gamma value is ignored and a linear space is used.
+    */
+
+   SVGA3D_TS_GAMMA                      = 25,   /* float */
+   SVGA3D_TS_BUMPENVLSCALE              = 26,   /* float */
+   SVGA3D_TS_BUMPENVLOFFSET             = 27,   /* float */
+   SVGA3D_TS_COLORARG0                  = 28,   /* SVGA3dTextureArgData */
+   SVGA3D_TS_ALPHAARG0                  = 29,   /* SVGA3dTextureArgData */
+   SVGA3D_TS_MAX
+} SVGA3dTextureStateName;
+
+typedef enum {
+   SVGA3D_TC_INVALID                   = 0,
+   SVGA3D_TC_DISABLE                   = 1,
+   SVGA3D_TC_SELECTARG1                = 2,
+   SVGA3D_TC_SELECTARG2                = 3,
+   SVGA3D_TC_MODULATE                  = 4,
+   SVGA3D_TC_ADD                       = 5,
+   SVGA3D_TC_ADDSIGNED                 = 6,
+   SVGA3D_TC_SUBTRACT                  = 7,
+   SVGA3D_TC_BLENDTEXTUREALPHA         = 8,
+   SVGA3D_TC_BLENDDIFFUSEALPHA         = 9,
+   SVGA3D_TC_BLENDCURRENTALPHA         = 10,
+   SVGA3D_TC_BLENDFACTORALPHA          = 11,
+   SVGA3D_TC_MODULATE2X                = 12,
+   SVGA3D_TC_MODULATE4X                = 13,
+   SVGA3D_TC_DSDT                      = 14,
+   SVGA3D_TC_DOTPRODUCT3               = 15,
+   SVGA3D_TC_BLENDTEXTUREALPHAPM       = 16,
+   SVGA3D_TC_ADDSIGNED2X               = 17,
+   SVGA3D_TC_ADDSMOOTH                 = 18,
+   SVGA3D_TC_PREMODULATE               = 19,
+   SVGA3D_TC_MODULATEALPHA_ADDCOLOR    = 20,
+   SVGA3D_TC_MODULATECOLOR_ADDALPHA    = 21,
+   SVGA3D_TC_MODULATEINVALPHA_ADDCOLOR = 22,
+   SVGA3D_TC_MODULATEINVCOLOR_ADDALPHA = 23,
+   SVGA3D_TC_BUMPENVMAPLUMINANCE       = 24,
+   SVGA3D_TC_MULTIPLYADD               = 25,
+   SVGA3D_TC_LERP                      = 26,
+   SVGA3D_TC_MAX
+} SVGA3dTextureCombiner;
+
+#define SVGA3D_TC_CAP_BIT(svga3d_tc_op) (svga3d_tc_op ? (1 << (svga3d_tc_op - 1)) : 0)
+
+typedef enum {
+   SVGA3D_TEX_ADDRESS_INVALID    = 0,
+   SVGA3D_TEX_ADDRESS_WRAP       = 1,
+   SVGA3D_TEX_ADDRESS_MIRROR     = 2,
+   SVGA3D_TEX_ADDRESS_CLAMP      = 3,
+   SVGA3D_TEX_ADDRESS_BORDER     = 4,
+   SVGA3D_TEX_ADDRESS_MIRRORONCE = 5,
+   SVGA3D_TEX_ADDRESS_EDGE       = 6,
+   SVGA3D_TEX_ADDRESS_MAX
+} SVGA3dTextureAddress;
+
+/*
+ * SVGA3D_TEX_FILTER_NONE as the minification filter means mipmapping is
+ * disabled, and the rasterizer should use the magnification filter instead.
+ */
+typedef enum {
+   SVGA3D_TEX_FILTER_NONE           = 0,
+   SVGA3D_TEX_FILTER_NEAREST        = 1,
+   SVGA3D_TEX_FILTER_LINEAR         = 2,
+   SVGA3D_TEX_FILTER_ANISOTROPIC    = 3,
+   SVGA3D_TEX_FILTER_FLATCUBIC      = 4, // Deprecated, not implemented
+   SVGA3D_TEX_FILTER_GAUSSIANCUBIC  = 5, // Deprecated, not implemented
+   SVGA3D_TEX_FILTER_PYRAMIDALQUAD  = 6, // Not currently implemented
+   SVGA3D_TEX_FILTER_GAUSSIANQUAD   = 7, // Not currently implemented
+   SVGA3D_TEX_FILTER_MAX
+} SVGA3dTextureFilter;
+
+typedef enum {
+   SVGA3D_TEX_TRANSFORM_OFF    = 0,
+   SVGA3D_TEX_TRANSFORM_S      = (1 << 0),
+   SVGA3D_TEX_TRANSFORM_T      = (1 << 1),
+   SVGA3D_TEX_TRANSFORM_R      = (1 << 2),
+   SVGA3D_TEX_TRANSFORM_Q      = (1 << 3),
+   SVGA3D_TEX_PROJECTED        = (1 << 15),
+} SVGA3dTexTransformFlags;
+
+typedef enum {
+   SVGA3D_TEXCOORD_GEN_OFF              = 0,
+   SVGA3D_TEXCOORD_GEN_EYE_POSITION     = 1,
+   SVGA3D_TEXCOORD_GEN_EYE_NORMAL       = 2,
+   SVGA3D_TEXCOORD_GEN_REFLECTIONVECTOR = 3,
+   SVGA3D_TEXCOORD_GEN_SPHERE           = 4,
+   SVGA3D_TEXCOORD_GEN_MAX
+} SVGA3dTextureCoordGen;
+
+/*
+ * Texture argument constants for texture combiner
+ */
+typedef enum {
+   SVGA3D_TA_INVALID    = 0,
+   SVGA3D_TA_CONSTANT   = 1,
+   SVGA3D_TA_PREVIOUS   = 2,
+   SVGA3D_TA_DIFFUSE    = 3,
+   SVGA3D_TA_TEXTURE    = 4,
+   SVGA3D_TA_SPECULAR   = 5,
+   SVGA3D_TA_MAX
+} SVGA3dTextureArgData;
+
+#define SVGA3D_TM_MASK_LEN 4
+
+/* Modifiers for texture argument constants defined above. */
+typedef enum {
+   SVGA3D_TM_NONE       = 0,
+   SVGA3D_TM_ALPHA      = (1 << SVGA3D_TM_MASK_LEN),
+   SVGA3D_TM_ONE_MINUS  = (2 << SVGA3D_TM_MASK_LEN),
+} SVGA3dTextureArgModifier;
+
+#define SVGA3D_INVALID_ID         ((uint32)-1)
+#define SVGA3D_MAX_CLIP_PLANES    6
+
+/*
+ * This is the limit to the number of fixed-function texture
+ * transforms and texture coordinates we can support. It does *not*
+ * correspond to the number of texture image units (samplers) we
+ * support!
+ */
+#define SVGA3D_MAX_TEXTURE_COORDS 8
+
+/*
+ * Vertex declarations
+ *
+ * Notes:
+ *
+ * SVGA3D_DECLUSAGE_POSITIONT is for pre-transformed vertices. If you
+ * draw with any POSITIONT vertex arrays, the programmable vertex
+ * pipeline will be implicitly disabled. Drawing will take place as if
+ * no vertex shader was bound.
+ */
+
+typedef enum {
+   SVGA3D_DECLUSAGE_POSITION     = 0,
+   SVGA3D_DECLUSAGE_BLENDWEIGHT,       //  1
+   SVGA3D_DECLUSAGE_BLENDINDICES,      //  2
+   SVGA3D_DECLUSAGE_NORMAL,            //  3
+   SVGA3D_DECLUSAGE_PSIZE,             //  4
+   SVGA3D_DECLUSAGE_TEXCOORD,          //  5
+   SVGA3D_DECLUSAGE_TANGENT,           //  6
+   SVGA3D_DECLUSAGE_BINORMAL,          //  7
+   SVGA3D_DECLUSAGE_TESSFACTOR,        //  8
+   SVGA3D_DECLUSAGE_POSITIONT,         //  9
+   SVGA3D_DECLUSAGE_COLOR,             // 10
+   SVGA3D_DECLUSAGE_FOG,               // 11
+   SVGA3D_DECLUSAGE_DEPTH,             // 12
+   SVGA3D_DECLUSAGE_SAMPLE,            // 13
+   SVGA3D_DECLUSAGE_MAX
+} SVGA3dDeclUsage;
+
+typedef enum {
+   SVGA3D_DECLMETHOD_DEFAULT     = 0,
+   SVGA3D_DECLMETHOD_PARTIALU,
+   SVGA3D_DECLMETHOD_PARTIALV,
+   SVGA3D_DECLMETHOD_CROSSUV,          // Normal
+   SVGA3D_DECLMETHOD_UV,
+   SVGA3D_DECLMETHOD_LOOKUP,           // Lookup a displacement map
+   SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED, // Lookup a pre-sampled displacement map
+} SVGA3dDeclMethod;
+
+typedef enum {
+   SVGA3D_DECLTYPE_FLOAT1        =  0,
+   SVGA3D_DECLTYPE_FLOAT2        =  1,
+   SVGA3D_DECLTYPE_FLOAT3        =  2,
+   SVGA3D_DECLTYPE_FLOAT4        =  3,
+   SVGA3D_DECLTYPE_D3DCOLOR      =  4,
+   SVGA3D_DECLTYPE_UBYTE4        =  5,
+   SVGA3D_DECLTYPE_SHORT2        =  6,
+   SVGA3D_DECLTYPE_SHORT4        =  7,
+   SVGA3D_DECLTYPE_UBYTE4N       =  8,
+   SVGA3D_DECLTYPE_SHORT2N       =  9,
+   SVGA3D_DECLTYPE_SHORT4N       = 10,
+   SVGA3D_DECLTYPE_USHORT2N      = 11,
+   SVGA3D_DECLTYPE_USHORT4N      = 12,
+   SVGA3D_DECLTYPE_UDEC3         = 13,
+   SVGA3D_DECLTYPE_DEC3N         = 14,
+   SVGA3D_DECLTYPE_FLOAT16_2     = 15,
+   SVGA3D_DECLTYPE_FLOAT16_4     = 16,
+   SVGA3D_DECLTYPE_MAX,
+} SVGA3dDeclType;
+
+/*
+ * This structure is used for the divisor for geometry instancing;
+ * it's a direct translation of the Direct3D equivalent.
+ */
+typedef union {
+   struct {
+      /*
+       * For index data, this number represents the number of instances to draw.
+       * For instance data, this number represents the number of
+       * instances/vertex in this stream
+       */
+      uint32 count : 30;
+
+      /*
+       * This is 1 if this is supposed to be the data that is repeated for
+       * every instance.
+       */
+      uint32 indexedData : 1;
+
+      /*
+       * This is 1 if this is supposed to be the per-instance data.
+       */
+      uint32 instanceData : 1;
+   };
+
+   uint32 value;
+} SVGA3dVertexDivisor;
+
+typedef enum {
+   SVGA3D_PRIMITIVE_INVALID                     = 0,
+   SVGA3D_PRIMITIVE_TRIANGLELIST                = 1,
+   SVGA3D_PRIMITIVE_POINTLIST                   = 2,
+   SVGA3D_PRIMITIVE_LINELIST                    = 3,
+   SVGA3D_PRIMITIVE_LINESTRIP                   = 4,
+   SVGA3D_PRIMITIVE_TRIANGLESTRIP               = 5,
+   SVGA3D_PRIMITIVE_TRIANGLEFAN                 = 6,
+   SVGA3D_PRIMITIVE_MAX
+} SVGA3dPrimitiveType;
+
+typedef enum {
+   SVGA3D_COORDINATE_INVALID                   = 0,
+   SVGA3D_COORDINATE_LEFTHANDED                = 1,
+   SVGA3D_COORDINATE_RIGHTHANDED               = 2,
+   SVGA3D_COORDINATE_MAX
+} SVGA3dCoordinateType;
+
+typedef enum {
+   SVGA3D_TRANSFORM_INVALID                     = 0,
+   SVGA3D_TRANSFORM_WORLD                       = 1,
+   SVGA3D_TRANSFORM_VIEW                        = 2,
+   SVGA3D_TRANSFORM_PROJECTION                  = 3,
+   SVGA3D_TRANSFORM_TEXTURE0                    = 4,
+   SVGA3D_TRANSFORM_TEXTURE1                    = 5,
+   SVGA3D_TRANSFORM_TEXTURE2                    = 6,
+   SVGA3D_TRANSFORM_TEXTURE3                    = 7,
+   SVGA3D_TRANSFORM_TEXTURE4                    = 8,
+   SVGA3D_TRANSFORM_TEXTURE5                    = 9,
+   SVGA3D_TRANSFORM_TEXTURE6                    = 10,
+   SVGA3D_TRANSFORM_TEXTURE7                    = 11,
+   SVGA3D_TRANSFORM_WORLD1                      = 12,
+   SVGA3D_TRANSFORM_WORLD2                      = 13,
+   SVGA3D_TRANSFORM_WORLD3                      = 14,
+   SVGA3D_TRANSFORM_MAX
+} SVGA3dTransformType;
+
+typedef enum {
+   SVGA3D_LIGHTTYPE_INVALID                     = 0,
+   SVGA3D_LIGHTTYPE_POINT                       = 1,
+   SVGA3D_LIGHTTYPE_SPOT1                       = 2, /* 1-cone, in degrees */
+   SVGA3D_LIGHTTYPE_SPOT2                       = 3, /* 2-cone, in radians */
+   SVGA3D_LIGHTTYPE_DIRECTIONAL                 = 4,
+   SVGA3D_LIGHTTYPE_MAX
+} SVGA3dLightType;
+
+typedef enum {
+   SVGA3D_CUBEFACE_POSX                         = 0,
+   SVGA3D_CUBEFACE_NEGX                         = 1,
+   SVGA3D_CUBEFACE_POSY                         = 2,
+   SVGA3D_CUBEFACE_NEGY                         = 3,
+   SVGA3D_CUBEFACE_POSZ                         = 4,
+   SVGA3D_CUBEFACE_NEGZ                         = 5,
+} SVGA3dCubeFace;
+
+typedef enum {
+   SVGA3D_SHADERTYPE_COMPILED_DX8               = 0,
+   SVGA3D_SHADERTYPE_VS                         = 1,
+   SVGA3D_SHADERTYPE_PS                         = 2,
+   SVGA3D_SHADERTYPE_MAX
+} SVGA3dShaderType;
+
+typedef enum {
+   SVGA3D_CONST_TYPE_FLOAT                      = 0,
+   SVGA3D_CONST_TYPE_INT                        = 1,
+   SVGA3D_CONST_TYPE_BOOL                       = 2,
+} SVGA3dShaderConstType;
+
+#define SVGA3D_MAX_SURFACE_FACES                6
+
+typedef enum {
+   SVGA3D_STRETCH_BLT_POINT                     = 0,
+   SVGA3D_STRETCH_BLT_LINEAR                    = 1,
+   SVGA3D_STRETCH_BLT_MAX
+} SVGA3dStretchBltMode;
+
+typedef enum {
+   SVGA3D_QUERYTYPE_OCCLUSION                   = 0,
+   SVGA3D_QUERYTYPE_MAX
+} SVGA3dQueryType;
+
+typedef enum {
+   SVGA3D_QUERYSTATE_PENDING     = 0,      /* Waiting on the host (set by guest) */
+   SVGA3D_QUERYSTATE_SUCCEEDED   = 1,      /* Completed successfully (set by host) */
+   SVGA3D_QUERYSTATE_FAILED      = 2,      /* Completed unsuccessfully (set by host) */
+   SVGA3D_QUERYSTATE_NEW         = 3,      /* Never submitted (For guest use only) */
+} SVGA3dQueryState;
+
+typedef enum {
+   SVGA3D_WRITE_HOST_VRAM        = 1,
+   SVGA3D_READ_HOST_VRAM         = 2,
+} SVGA3dTransferType;
+
+/*
+ * The maximum number vertex arrays we're guaranteed to support in
+ * SVGA_3D_CMD_DRAWPRIMITIVES.
+ */
+#define SVGA3D_MAX_VERTEX_ARRAYS   32
+
+/*
+ * Identifiers for commands in the command FIFO.
+ *
+ * IDs between 1000 and 1039 (inclusive) were used by obsolete versions of
+ * the SVGA3D protocol and remain reserved; they should not be used in the
+ * future.
+ *
+ * IDs between 1040 and 1999 (inclusive) are available for use by the
+ * current SVGA3D protocol.
+ *
+ * FIFO clients other than SVGA3D should stay below 1000, or at 2000
+ * and up.
+ */
+
+#define SVGA_3D_CMD_LEGACY_BASE            1000
+#define SVGA_3D_CMD_BASE                   1040
+
+#define SVGA_3D_CMD_SURFACE_DEFINE         SVGA_3D_CMD_BASE + 0
+#define SVGA_3D_CMD_SURFACE_DESTROY        SVGA_3D_CMD_BASE + 1
+#define SVGA_3D_CMD_SURFACE_COPY           SVGA_3D_CMD_BASE + 2
+#define SVGA_3D_CMD_SURFACE_STRETCHBLT     SVGA_3D_CMD_BASE + 3
+#define SVGA_3D_CMD_SURFACE_DMA            SVGA_3D_CMD_BASE + 4
+#define SVGA_3D_CMD_CONTEXT_DEFINE         SVGA_3D_CMD_BASE + 5
+#define SVGA_3D_CMD_CONTEXT_DESTROY        SVGA_3D_CMD_BASE + 6
+#define SVGA_3D_CMD_SETTRANSFORM           SVGA_3D_CMD_BASE + 7
+#define SVGA_3D_CMD_SETZRANGE              SVGA_3D_CMD_BASE + 8
+#define SVGA_3D_CMD_SETRENDERSTATE         SVGA_3D_CMD_BASE + 9
+#define SVGA_3D_CMD_SETRENDERTARGET        SVGA_3D_CMD_BASE + 10
+#define SVGA_3D_CMD_SETTEXTURESTATE        SVGA_3D_CMD_BASE + 11
+#define SVGA_3D_CMD_SETMATERIAL            SVGA_3D_CMD_BASE + 12
+#define SVGA_3D_CMD_SETLIGHTDATA           SVGA_3D_CMD_BASE + 13
+#define SVGA_3D_CMD_SETLIGHTENABLED        SVGA_3D_CMD_BASE + 14
+#define SVGA_3D_CMD_SETVIEWPORT            SVGA_3D_CMD_BASE + 15
+#define SVGA_3D_CMD_SETCLIPPLANE           SVGA_3D_CMD_BASE + 16
+#define SVGA_3D_CMD_CLEAR                  SVGA_3D_CMD_BASE + 17
+#define SVGA_3D_CMD_PRESENT                SVGA_3D_CMD_BASE + 18    // Deprecated
+#define SVGA_3D_CMD_SHADER_DEFINE          SVGA_3D_CMD_BASE + 19
+#define SVGA_3D_CMD_SHADER_DESTROY         SVGA_3D_CMD_BASE + 20
+#define SVGA_3D_CMD_SET_SHADER             SVGA_3D_CMD_BASE + 21
+#define SVGA_3D_CMD_SET_SHADER_CONST       SVGA_3D_CMD_BASE + 22
+#define SVGA_3D_CMD_DRAW_PRIMITIVES        SVGA_3D_CMD_BASE + 23
+#define SVGA_3D_CMD_SETSCISSORRECT         SVGA_3D_CMD_BASE + 24
+#define SVGA_3D_CMD_BEGIN_QUERY            SVGA_3D_CMD_BASE + 25
+#define SVGA_3D_CMD_END_QUERY              SVGA_3D_CMD_BASE + 26
+#define SVGA_3D_CMD_WAIT_FOR_QUERY         SVGA_3D_CMD_BASE + 27
+#define SVGA_3D_CMD_PRESENT_READBACK       SVGA_3D_CMD_BASE + 28    // Deprecated
+#define SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN SVGA_3D_CMD_BASE + 29
+#define SVGA_3D_CMD_MAX                    SVGA_3D_CMD_BASE + 30
+
+#define SVGA_3D_CMD_FUTURE_MAX             2000
+
+/*
+ * Common substructures used in multiple FIFO commands:
+ */
+
+typedef struct {
+   union {
+      struct {
+         uint16  function;       // SVGA3dFogFunction
+         uint8   type;           // SVGA3dFogType
+         uint8   base;           // SVGA3dFogBase
+      };
+      uint32     uintValue;
+   };
+} SVGA3dFogMode;
+
+/*
+ * Uniquely identify one image (a 1D/2D/3D array) from a surface. This
+ * is a surface ID as well as face/mipmap indices.
+ */
+
+typedef
+struct SVGA3dSurfaceImageId {
+   uint32               sid;
+   uint32               face;
+   uint32               mipmap;
+} SVGA3dSurfaceImageId;
+
+typedef
+struct SVGA3dGuestImage {
+   SVGAGuestPtr         ptr;
+
+   /*
+    * A note on interpretation of pitch: This value of pitch is the
+    * number of bytes between vertically adjacent image
+    * blocks. Normally this is the number of bytes between the first
+    * pixel of two adjacent scanlines. With compressed textures,
+    * however, this may represent the number of bytes between
+    * compression blocks rather than between rows of pixels.
+    *
+    * XXX: Compressed textures currently must be tightly packed in guest memory.
+    *
+    * If the image is 1-dimensional, pitch is ignored.
+    *
+    * If 'pitch' is zero, the SVGA3D device calculates a pitch value
+    * assuming each row of blocks is tightly packed.
+    */
+   uint32 pitch;
+} SVGA3dGuestImage;
+
+
+/*
+ * FIFO command format definitions:
+ */
+
+/*
+ * The data size header following cmdNum for every 3d command
+ */
+typedef
+struct {
+   uint32               id;
+   uint32               size;
+} SVGA3dCmdHeader;
+
+/*
+ * A surface is a hierarchy of host VRAM surfaces: 1D, 2D, or 3D, with
+ * optional mipmaps and cube faces.
+ */
+
+typedef
+struct {
+   uint32               width;
+   uint32               height;
+   uint32               depth;
+} SVGA3dSize;
+
+typedef enum {
+   SVGA3D_SURFACE_CUBEMAP              = (1 << 0),
+   SVGA3D_SURFACE_HINT_STATIC          = (1 << 1),
+   SVGA3D_SURFACE_HINT_DYNAMIC         = (1 << 2),
+   SVGA3D_SURFACE_HINT_INDEXBUFFER     = (1 << 3),
+   SVGA3D_SURFACE_HINT_VERTEXBUFFER    = (1 << 4),
+   SVGA3D_SURFACE_HINT_TEXTURE         = (1 << 5),
+   SVGA3D_SURFACE_HINT_RENDERTARGET    = (1 << 6),
+   SVGA3D_SURFACE_HINT_DEPTHSTENCIL    = (1 << 7),
+   SVGA3D_SURFACE_HINT_WRITEONLY       = (1 << 8),
+} SVGA3dSurfaceFlags;
+
+typedef
+struct {
+   uint32               numMipLevels;
+} SVGA3dSurfaceFace;
+
+typedef
+struct {
+   uint32                      sid;
+   SVGA3dSurfaceFlags          surfaceFlags;
+   SVGA3dSurfaceFormat         format;
+   SVGA3dSurfaceFace           face[SVGA3D_MAX_SURFACE_FACES];
+   /*
+    * Followed by an SVGA3dSize structure for each mip level in each face.
+    *
+    * A note on surface sizes: Sizes are always specified in pixels,
+    * even if the true surface size is not a multiple of the minimum
+    * block size of the surface's format. For example, a 3x3x1 DXT1
+    * compressed texture would actually be stored as a 4x4x1 image in
+    * memory.
+    */
+} SVGA3dCmdDefineSurface;       /* SVGA_3D_CMD_SURFACE_DEFINE */
+
+typedef
+struct {
+   uint32               sid;
+} SVGA3dCmdDestroySurface;      /* SVGA_3D_CMD_SURFACE_DESTROY */
+
+typedef
+struct {
+   uint32               cid;
+} SVGA3dCmdDefineContext;       /* SVGA_3D_CMD_CONTEXT_DEFINE */
+
+typedef
+struct {
+   uint32               cid;
+} SVGA3dCmdDestroyContext;      /* SVGA_3D_CMD_CONTEXT_DESTROY */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dClearFlag      clearFlag;
+   uint32               color;
+   float                depth;
+   uint32               stencil;
+   /* Followed by variable number of SVGA3dRect structures */
+} SVGA3dCmdClear;               /* SVGA_3D_CMD_CLEAR */
+
+typedef
+struct SVGA3dCopyRect {
+   uint32               x;
+   uint32               y;
+   uint32               w;
+   uint32               h;
+   uint32               srcx;
+   uint32               srcy;
+} SVGA3dCopyRect;
+
+typedef
+struct SVGA3dCopyBox {
+   uint32               x;
+   uint32               y;
+   uint32               z;
+   uint32               w;
+   uint32               h;
+   uint32               d;
+   uint32               srcx;
+   uint32               srcy;
+   uint32               srcz;
+} SVGA3dCopyBox;
+
+typedef
+struct {
+   uint32               x;
+   uint32               y;
+   uint32               w;
+   uint32               h;
+} SVGA3dRect;
+
+typedef
+struct {
+   uint32               x;
+   uint32               y;
+   uint32               z;
+   uint32               w;
+   uint32               h;
+   uint32               d;
+} SVGA3dBox;
+
+typedef
+struct {
+   uint32               x;
+   uint32               y;
+   uint32               z;
+} SVGA3dPoint;
+
+typedef
+struct {
+   SVGA3dLightType      type;
+   SVGA3dBool           inWorldSpace;
+   float                diffuse[4];
+   float                specular[4];
+   float                ambient[4];
+   float                position[4];
+   float                direction[4];
+   float                range;
+   float                falloff;
+   float                attenuation0;
+   float                attenuation1;
+   float                attenuation2;
+   float                theta;
+   float                phi;
+} SVGA3dLightData;
+
+typedef
+struct {
+   uint32               sid;
+   /* Followed by variable number of SVGA3dCopyRect structures */
+} SVGA3dCmdPresent;             /* SVGA_3D_CMD_PRESENT */
+
+typedef
+struct {
+   SVGA3dRenderStateName   state;
+   union {
+      uint32               uintValue;
+      float                floatValue;
+   };
+} SVGA3dRenderState;
+
+typedef
+struct {
+   uint32               cid;
+   /* Followed by variable number of SVGA3dRenderState structures */
+} SVGA3dCmdSetRenderState;      /* SVGA_3D_CMD_SETRENDERSTATE */
+
+typedef
+struct {
+   uint32                 cid;
+   SVGA3dRenderTargetType type;
+   SVGA3dSurfaceImageId   target;
+} SVGA3dCmdSetRenderTarget;     /* SVGA_3D_CMD_SETRENDERTARGET */
+
+typedef
+struct {
+   SVGA3dSurfaceImageId  src;
+   SVGA3dSurfaceImageId  dest;
+   /* Followed by variable number of SVGA3dCopyBox structures */
+} SVGA3dCmdSurfaceCopy;               /* SVGA_3D_CMD_SURFACE_COPY */
+
+typedef
+struct {
+   SVGA3dSurfaceImageId  src;
+   SVGA3dSurfaceImageId  dest;
+   SVGA3dBox             boxSrc;
+   SVGA3dBox             boxDest;
+   SVGA3dStretchBltMode  mode;
+} SVGA3dCmdSurfaceStretchBlt;         /* SVGA_3D_CMD_SURFACE_STRETCHBLT */
+
+typedef
+struct {
+   /*
+    * If the discard flag is present in a surface DMA operation, the host may
+    * discard the contents of the current mipmap level and face of the target
+    * surface before applying the surface DMA contents.
+    */
+   uint32 discard : 1;
+
+   /*
+    * If the unsynchronized flag is present, the host may perform this upload
+    * without syncing to pending reads on this surface.
+    */
+   uint32 unsynchronized : 1;
+
+   /*
+    * Guests *MUST* set the reserved bits to 0 before submitting the command
+    * suffix as future flags may occupy these bits.
+    */
+   uint32 reserved : 30;
+} SVGA3dSurfaceDMAFlags;
+
+typedef
+struct {
+   SVGA3dGuestImage      guest;
+   SVGA3dSurfaceImageId  host;
+   SVGA3dTransferType    transfer;
+   /*
+    * Followed by variable number of SVGA3dCopyBox structures. For consistency
+    * in all clipping logic and coordinate translation, we define the
+    * "source" in each copyBox as the guest image and the
+    * "destination" as the host image, regardless of transfer
+    * direction.
+    *
+    * For efficiency, the SVGA3D device is free to copy more data than
+    * specified. For example, it may round copy boxes outwards such
+    * that they lie on particular alignment boundaries.
+    */
+} SVGA3dCmdSurfaceDMA;                /* SVGA_3D_CMD_SURFACE_DMA */
+
+/*
+ * SVGA3dCmdSurfaceDMASuffix --
+ *
+ *    This is a command suffix that will appear after a SurfaceDMA command in
+ *    the FIFO.  It contains some extra information that hosts may use to
+ *    optimize performance or protect the guest.  This suffix exists to preserve
+ *    backwards compatibility while also allowing for new functionality to be
+ *    implemented.
+ */
+
+typedef
+struct {
+   uint32 suffixSize;
+
+   /*
+    * The maximum offset is used to determine the maximum offset from the
+    * guestPtr base address that will be accessed or written to during this
+    * surfaceDMA.  If the suffix is supported, the host will respect this
+    * boundary while performing surface DMAs.
+    *
+    * Defaults to MAX_UINT32
+    */
+   uint32 maximumOffset;
+
+   /*
+    * A set of flags that describes optimizations that the host may perform
+    * while performing this surface DMA operation.  The guest should never rely
+    * on behaviour that is different when these flags are set for correctness.
+    *
+    * Defaults to 0
+    */
+   SVGA3dSurfaceDMAFlags flags;
+} SVGA3dCmdSurfaceDMASuffix;
+
+/*
+ * SVGA_3D_CMD_DRAW_PRIMITIVES --
+ *
+ *   This command is the SVGA3D device's generic drawing entry point.
+ *   It can draw multiple ranges of primitives, optionally using an
+ *   index buffer, using an arbitrary collection of vertex buffers.
+ *
+ *   Each SVGA3dVertexDecl defines a distinct vertex array to bind
+ *   during this draw call. The declarations specify which surface
+ *   the vertex data lives in, what that vertex data is used for,
+ *   and how to interpret it.
+ *
+ *   Each SVGA3dPrimitiveRange defines a collection of primitives
+ *   to render using the same vertex arrays. An index buffer is
+ *   optional.
+ */
+
+typedef
+struct {
+   /*
+    * A range hint is an optional specification for the range of indices
+    * in an SVGA3dArray that will be used. If 'last' is zero, it is assumed
+    * that the entire array will be used.
+    *
+    * These are only hints. The SVGA3D device may use them for
+    * performance optimization if possible, but it's also allowed to
+    * ignore these values.
+    */
+   uint32               first;
+   uint32               last;
+} SVGA3dArrayRangeHint;
+
+typedef
+struct {
+   /*
+    * Define the origin and shape of a vertex or index array. Both
+    * 'offset' and 'stride' are in bytes. The provided surface will be
+    * reinterpreted as a flat array of bytes in the same format used
+    * by surface DMA operations. To avoid unnecessary conversions, the
+    * surface should be created with the SVGA3D_BUFFER format.
+    *
+    * Index 0 in the array starts 'offset' bytes into the surface.
+    * Index 1 begins at byte 'offset + stride', etc. Array indices may
+    * not be negative.
+    */
+   uint32               surfaceId;
+   uint32               offset;
+   uint32               stride;
+} SVGA3dArray;
+
+typedef
+struct {
+   /*
+    * Describe a vertex array's data type, and define how it is to be
+    * used by the fixed function pipeline or the vertex shader. It
+    * isn't useful to have two VertexDecls with the same
+    * VertexArrayIdentity in one draw call.
+    */
+   SVGA3dDeclType       type;
+   SVGA3dDeclMethod     method;
+   SVGA3dDeclUsage      usage;
+   uint32               usageIndex;
+} SVGA3dVertexArrayIdentity;
+
+typedef
+struct {
+   SVGA3dVertexArrayIdentity  identity;
+   SVGA3dArray                array;
+   SVGA3dArrayRangeHint       rangeHint;
+} SVGA3dVertexDecl;
+
+typedef
+struct {
+   /*
+    * Define a group of primitives to render, from sequential indices.
+    *
+    * The value of 'primitiveType' and 'primitiveCount' imply the
+    * total number of vertices that will be rendered.
+    */
+   SVGA3dPrimitiveType  primType;
+   uint32               primitiveCount;
+
+   /*
+    * Optional index buffer. If indexArray.surfaceId is
+    * SVGA3D_INVALID_ID, we render without an index buffer. Rendering
+    * without an index buffer is identical to rendering with an index
+    * buffer containing the sequence [0, 1, 2, 3, ...].
+    *
+    * If an index buffer is in use, indexWidth specifies the width in
+    * bytes of each index value. It must be less than or equal to
+    * indexArray.stride.
+    *
+    * (Currently, the SVGA3D device requires index buffers to be tightly
+    * packed. In other words, indexWidth == indexArray.stride)
+    */
+   SVGA3dArray          indexArray;
+   uint32               indexWidth;
+
+   /*
+    * Optional index bias. This number is added to all indices from
+    * indexArray before they are used as vertex array indices. This
+    * can be used in multiple ways:
+    *
+    *  - When not using an indexArray, this bias can be used to
+    *    specify where in the vertex arrays to begin rendering.
+    *
+    *  - A positive number here is equivalent to increasing the
+    *    offset in each vertex array.
+    *
+    *  - A negative number can be used to render using a small
+    *    vertex array and an index buffer that contains large
+    *    values. This may be used by some applications that
+    *    crop a vertex buffer without modifying their index
+    *    buffer.
+    *
+    * Note that rendering with a negative bias value may be slower and
+    * use more memory than rendering with a positive or zero bias.
+    */
+   int32                indexBias;
+} SVGA3dPrimitiveRange;
+
+typedef
+struct {
+   uint32               cid;
+   uint32               numVertexDecls;
+   uint32               numRanges;
+
+   /*
+    * There are two variable size arrays after the
+    * SVGA3dCmdDrawPrimitives structure. In order,
+    * they are:
+    *
+    * 1. SVGA3dVertexDecl, quantity 'numVertexDecls'
+    * 2. SVGA3dPrimitiveRange, quantity 'numRanges'
+    * 3. Optionally, SVGA3dVertexDivisor, quantity 'numVertexDecls' (contains
+    *    the frequency divisor for this the corresponding vertex decl)
+    */
+} SVGA3dCmdDrawPrimitives;      /* SVGA_3D_CMD_DRAWPRIMITIVES */
+
+typedef
+struct {
+   uint32                   stage;
+   SVGA3dTextureStateName   name;
+   union {
+      uint32                value;
+      float                 floatValue;
+   };
+} SVGA3dTextureState;
+
+typedef
+struct {
+   uint32               cid;
+   /* Followed by variable number of SVGA3dTextureState structures */
+} SVGA3dCmdSetTextureState;      /* SVGA_3D_CMD_SETTEXTURESTATE */
+
+typedef
+struct {
+   uint32                   cid;
+   SVGA3dTransformType      type;
+   float                    matrix[16];
+} SVGA3dCmdSetTransform;          /* SVGA_3D_CMD_SETTRANSFORM */
+
+typedef
+struct {
+   float                min;
+   float                max;
+} SVGA3dZRange;
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dZRange         zRange;
+} SVGA3dCmdSetZRange;             /* SVGA_3D_CMD_SETZRANGE */
+
+typedef
+struct {
+   float                diffuse[4];
+   float                ambient[4];
+   float                specular[4];
+   float                emissive[4];
+   float                shininess;
+} SVGA3dMaterial;
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dFace           face;
+   SVGA3dMaterial       material;
+} SVGA3dCmdSetMaterial;           /* SVGA_3D_CMD_SETMATERIAL */
+
+typedef
+struct {
+   uint32               cid;
+   uint32               index;
+   SVGA3dLightData      data;
+} SVGA3dCmdSetLightData;           /* SVGA_3D_CMD_SETLIGHTDATA */
+
+typedef
+struct {
+   uint32               cid;
+   uint32               index;
+   uint32               enabled;
+} SVGA3dCmdSetLightEnabled;      /* SVGA_3D_CMD_SETLIGHTENABLED */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dRect           rect;
+} SVGA3dCmdSetViewport;           /* SVGA_3D_CMD_SETVIEWPORT */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dRect           rect;
+} SVGA3dCmdSetScissorRect;         /* SVGA_3D_CMD_SETSCISSORRECT */
+
+typedef
+struct {
+   uint32               cid;
+   uint32               index;
+   float                plane[4];
+} SVGA3dCmdSetClipPlane;           /* SVGA_3D_CMD_SETCLIPPLANE */
+
+typedef
+struct {
+   uint32               cid;
+   uint32               shid;
+   SVGA3dShaderType     type;
+   /* Followed by variable number of DWORDs for shader bycode */
+} SVGA3dCmdDefineShader;           /* SVGA_3D_CMD_SHADER_DEFINE */
+
+typedef
+struct {
+   uint32               cid;
+   uint32               shid;
+   SVGA3dShaderType     type;
+} SVGA3dCmdDestroyShader;         /* SVGA_3D_CMD_SHADER_DESTROY */
+
+typedef
+struct {
+   uint32                  cid;
+   uint32                  reg;     /* register number */
+   SVGA3dShaderType        type;
+   SVGA3dShaderConstType   ctype;
+   uint32                  values[4];
+} SVGA3dCmdSetShaderConst;        /* SVGA_3D_CMD_SET_SHADER_CONST */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dShaderType     type;
+   uint32               shid;
+} SVGA3dCmdSetShader;             /* SVGA_3D_CMD_SET_SHADER */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dQueryType      type;
+} SVGA3dCmdBeginQuery;           /* SVGA_3D_CMD_BEGIN_QUERY */
+
+typedef
+struct {
+   uint32               cid;
+   SVGA3dQueryType      type;
+   SVGAGuestPtr         guestResult;  /* Points to an SVGA3dQueryResult structure */
+} SVGA3dCmdEndQuery;                  /* SVGA_3D_CMD_END_QUERY */
+
+typedef
+struct {
+   uint32               cid;          /* Same parameters passed to END_QUERY */
+   SVGA3dQueryType      type;
+   SVGAGuestPtr         guestResult;
+} SVGA3dCmdWaitForQuery;              /* SVGA_3D_CMD_WAIT_FOR_QUERY */
+
+typedef
+struct {
+   uint32               totalSize;    /* Set by guest before query is ended. */
+   SVGA3dQueryState     state;        /* Set by host or guest. See SVGA3dQueryState. */
+   union {                            /* Set by host on exit from PENDING state */
+      uint32            result32;
+   };
+} SVGA3dQueryResult;
+
+/*
+ * SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN --
+ *
+ *    This is a blit from an SVGA3D surface to a Screen Object. Just
+ *    like GMR-to-screen blits, this blit may be directed at a
+ *    specific screen or to the virtual coordinate space.
+ *
+ *    The blit copies from a rectangular region of an SVGA3D surface
+ *    image to a rectangular region of a screen or screens.
+ *
+ *    This command takes an optional variable-length list of clipping
+ *    rectangles after the body of the command. If no rectangles are
+ *    specified, there is no clipping region. The entire destRect is
+ *    drawn to. If one or more rectangles are included, they describe
+ *    a clipping region. The clip rectangle coordinates are measured
+ *    relative to the top-left corner of destRect.
+ *
+ *    This clipping region serves multiple purposes:
+ *
+ *      - It can be used to perform an irregularly shaped blit more
+ *        efficiently than by issuing many separate blit commands.
+ *
+ *      - It is equivalent to allowing blits with non-integer
+ *        source coordinates. You could blit just one half-pixel
+ *        of a source, for example, by specifying a larger
+ *        destination rectangle than you need, then removing
+ *        part of it using a clip rectangle.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ *
+ * Limitations:
+ *
+ *    - Currently, no backend supports blits from a mipmap or face
+ *      other than the first one.
+ */
+
+typedef
+struct {
+   SVGA3dSurfaceImageId srcImage;
+   SVGASignedRect       srcRect;
+   uint32               destScreenId; /* Screen ID or SVGA_ID_INVALID for virt. coords */
+   SVGASignedRect       destRect;     /* Supports scaling if src/rest different size */
+   /* Clipping: zero or more SVGASignedRects follow */
+} SVGA3dCmdBlitSurfaceToScreen;         /* SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN */
+
+
+/*
+ * Capability query index.
+ *
+ * Notes:
+ *
+ *   1. SVGA3D_DEVCAP_MAX_TEXTURES reflects the maximum number of
+ *      fixed-function texture units available. Each of these units
+ *      work in both FFP and Shader modes, and they support texture
+ *      transforms and texture coordinates. The host may have additional
+ *      texture image units that are only usable with shaders.
+ *
+ *   2. The BUFFER_FORMAT capabilities are deprecated, and they always
+ *      return TRUE. Even on physical hardware that does not support
+ *      these formats natively, the SVGA3D device will provide an emulation
+ *      which should be invisible to the guest OS.
+ *
+ *      In general, the SVGA3D device should support any operation on
+ *      any surface format, it just may perform some of these
+ *      operations in software depending on the capabilities of the
+ *      available physical hardware.
+ *
+ *      XXX: In the future, we will add capabilities that describe in
+ *      detail what formats are supported in hardware for what kinds
+ *      of operations.
+ */
+
+typedef enum {
+   SVGA3D_DEVCAP_3D                                = 0,
+   SVGA3D_DEVCAP_MAX_LIGHTS                        = 1,
+   SVGA3D_DEVCAP_MAX_TEXTURES                      = 2,  /* See note (1) */
+   SVGA3D_DEVCAP_MAX_CLIP_PLANES                   = 3,
+   SVGA3D_DEVCAP_VERTEX_SHADER_VERSION             = 4,
+   SVGA3D_DEVCAP_VERTEX_SHADER                     = 5,
+   SVGA3D_DEVCAP_FRAGMENT_SHADER_VERSION           = 6,
+   SVGA3D_DEVCAP_FRAGMENT_SHADER                   = 7,
+   SVGA3D_DEVCAP_MAX_RENDER_TARGETS                = 8,
+   SVGA3D_DEVCAP_S23E8_TEXTURES                    = 9,
+   SVGA3D_DEVCAP_S10E5_TEXTURES                    = 10,
+   SVGA3D_DEVCAP_MAX_FIXED_VERTEXBLEND             = 11,
+   SVGA3D_DEVCAP_D16_BUFFER_FORMAT                 = 12, /* See note (2) */
+   SVGA3D_DEVCAP_D24S8_BUFFER_FORMAT               = 13, /* See note (2) */
+   SVGA3D_DEVCAP_D24X8_BUFFER_FORMAT               = 14, /* See note (2) */
+   SVGA3D_DEVCAP_QUERY_TYPES                       = 15,
+   SVGA3D_DEVCAP_TEXTURE_GRADIENT_SAMPLING         = 16,
+   SVGA3D_DEVCAP_MAX_POINT_SIZE                    = 17,
+   SVGA3D_DEVCAP_MAX_SHADER_TEXTURES               = 18,
+   SVGA3D_DEVCAP_MAX_TEXTURE_WIDTH                 = 19,
+   SVGA3D_DEVCAP_MAX_TEXTURE_HEIGHT                = 20,
+   SVGA3D_DEVCAP_MAX_VOLUME_EXTENT                 = 21,
+   SVGA3D_DEVCAP_MAX_TEXTURE_REPEAT                = 22,
+   SVGA3D_DEVCAP_MAX_TEXTURE_ASPECT_RATIO          = 23,
+   SVGA3D_DEVCAP_MAX_TEXTURE_ANISOTROPY            = 24,
+   SVGA3D_DEVCAP_MAX_PRIMITIVE_COUNT               = 25,
+   SVGA3D_DEVCAP_MAX_VERTEX_INDEX                  = 26,
+   SVGA3D_DEVCAP_MAX_VERTEX_SHADER_INSTRUCTIONS    = 27,
+   SVGA3D_DEVCAP_MAX_FRAGMENT_SHADER_INSTRUCTIONS  = 28,
+   SVGA3D_DEVCAP_MAX_VERTEX_SHADER_TEMPS           = 29,
+   SVGA3D_DEVCAP_MAX_FRAGMENT_SHADER_TEMPS         = 30,
+   SVGA3D_DEVCAP_TEXTURE_OPS                       = 31,
+   SVGA3D_DEVCAP_SURFACEFMT_X8R8G8B8               = 32,
+   SVGA3D_DEVCAP_SURFACEFMT_A8R8G8B8               = 33,
+   SVGA3D_DEVCAP_SURFACEFMT_A2R10G10B10            = 34,
+   SVGA3D_DEVCAP_SURFACEFMT_X1R5G5B5               = 35,
+   SVGA3D_DEVCAP_SURFACEFMT_A1R5G5B5               = 36,
+   SVGA3D_DEVCAP_SURFACEFMT_A4R4G4B4               = 37,
+   SVGA3D_DEVCAP_SURFACEFMT_R5G6B5                 = 38,
+   SVGA3D_DEVCAP_SURFACEFMT_LUMINANCE16            = 39,
+   SVGA3D_DEVCAP_SURFACEFMT_LUMINANCE8_ALPHA8      = 40,
+   SVGA3D_DEVCAP_SURFACEFMT_ALPHA8                 = 41,
+   SVGA3D_DEVCAP_SURFACEFMT_LUMINANCE8             = 42,
+   SVGA3D_DEVCAP_SURFACEFMT_Z_D16                  = 43,
+   SVGA3D_DEVCAP_SURFACEFMT_Z_D24S8                = 44,
+   SVGA3D_DEVCAP_SURFACEFMT_Z_D24X8                = 45,
+   SVGA3D_DEVCAP_SURFACEFMT_DXT1                   = 46,
+   SVGA3D_DEVCAP_SURFACEFMT_DXT2                   = 47,
+   SVGA3D_DEVCAP_SURFACEFMT_DXT3                   = 48,
+   SVGA3D_DEVCAP_SURFACEFMT_DXT4                   = 49,
+   SVGA3D_DEVCAP_SURFACEFMT_DXT5                   = 50,
+   SVGA3D_DEVCAP_SURFACEFMT_BUMPX8L8V8U8           = 51,
+   SVGA3D_DEVCAP_SURFACEFMT_A2W10V10U10            = 52,
+   SVGA3D_DEVCAP_SURFACEFMT_BUMPU8V8               = 53,
+   SVGA3D_DEVCAP_SURFACEFMT_Q8W8V8U8               = 54,
+   SVGA3D_DEVCAP_SURFACEFMT_CxV8U8                 = 55,
+   SVGA3D_DEVCAP_SURFACEFMT_R_S10E5                = 56,
+   SVGA3D_DEVCAP_SURFACEFMT_R_S23E8                = 57,
+   SVGA3D_DEVCAP_SURFACEFMT_RG_S10E5               = 58,
+   SVGA3D_DEVCAP_SURFACEFMT_RG_S23E8               = 59,
+   SVGA3D_DEVCAP_SURFACEFMT_ARGB_S10E5             = 60,
+   SVGA3D_DEVCAP_SURFACEFMT_ARGB_S23E8             = 61,
+   SVGA3D_DEVCAP_MAX_VERTEX_SHADER_TEXTURES        = 63,
+
+   /*
+    * Note that MAX_SIMULTANEOUS_RENDER_TARGETS is a maximum count of color
+    * render targets.  This does no include the depth or stencil targets.
+    */
+   SVGA3D_DEVCAP_MAX_SIMULTANEOUS_RENDER_TARGETS   = 64,
+
+   SVGA3D_DEVCAP_SURFACEFMT_V16U16                 = 65,
+   SVGA3D_DEVCAP_SURFACEFMT_G16R16                 = 66,
+   SVGA3D_DEVCAP_SURFACEFMT_A16B16G16R16           = 67,
+   SVGA3D_DEVCAP_SURFACEFMT_UYVY                   = 68,
+   SVGA3D_DEVCAP_SURFACEFMT_YUY2                   = 69,
+
+   /*
+    * Don't add new caps into the previous section; the values in this
+    * enumeration must not change. You can put new values right before
+    * SVGA3D_DEVCAP_MAX.
+    */
+   SVGA3D_DEVCAP_MAX                                  /* This must be the last index. */
+} SVGA3dDevCapIndex;
+
+typedef union {
+   Bool   b;
+   uint32 u;
+   int32  i;
+   float  f;
+} SVGA3dDevCapResult;
+
+#endif /* _SVGA3D_REG_H_ */
diff --git a/src/gallium/drivers/svga/include/svga3d_shaderdefs.h b/src/gallium/drivers/svga/include/svga3d_shaderdefs.h
new file mode 100644
index 0000000000..2078c4a8a4
--- /dev/null
+++ b/src/gallium/drivers/svga/include/svga3d_shaderdefs.h
@@ -0,0 +1,519 @@
+/**********************************************************
+ * Copyright 2007-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/*
+ * svga3d_shaderdefs.h --
+ *
+ * SVGA3D byte code format and limit definitions.
+ *
+ * The format of the byte code directly corresponds to that defined
+ * by Microsoft DirectX SDK 9.0c (file d3d9types.h). The format can
+ * also be extended so that different shader formats can be supported
+ * for example GLSL, ARB vp/fp, NV/ATI shader formats, etc.
+ *
+ */
+
+#ifndef __SVGA3D_SHADER_DEFS__
+#define __SVGA3D_SHADER_DEFS__
+
+/* SVGA3D shader hardware limits. */
+
+#define SVGA3D_INPUTREG_MAX            16
+#define SVGA3D_OUTPUTREG_MAX           12
+#define SVGA3D_VERTEX_SAMPLERREG_MAX   4
+#define SVGA3D_PIXEL_SAMPLERREG_MAX    16
+#define SVGA3D_SAMPLERREG_MAX          (SVGA3D_PIXEL_SAMPLERREG_MAX+\
+                                        SVGA3D_VERTEX_SAMPLERREG_MAX)
+#define SVGA3D_TEMPREG_MAX             32
+#define SVGA3D_CONSTREG_MAX            256
+#define SVGA3D_CONSTINTREG_MAX         16
+#define SVGA3D_CONSTBOOLREG_MAX        16
+#define SVGA3D_ADDRREG_MAX             1
+#define SVGA3D_PREDREG_MAX             1
+
+/* SVGA3D byte code specific limits */
+
+#define SVGA3D_MAX_SRC_REGS      4
+#define SVGA3D_MAX_NESTING_LEVEL 32
+
+/* SVGA3D version information. */
+
+#define SVGA3D_VS_TYPE  0xFFFE
+#define SVGA3D_PS_TYPE  0xFFFF
+
+typedef struct {
+   union {
+      struct {
+         uint32 minor : 8;
+         uint32 major : 8;
+         uint32 type : 16;
+      };
+
+      uint32 value;
+   };
+} SVGA3dShaderVersion;
+
+#define SVGA3D_VS_10 ((SVGA3D_VS_TYPE << 16) | 1 << 8)
+#define SVGA3D_VS_11 (SVGA3D_VS_10 | 1)
+#define SVGA3D_VS_20 ((SVGA3D_VS_TYPE << 16) | 2 << 8)
+#define SVGA3D_VS_30 ((SVGA3D_VS_TYPE << 16) | 3 << 8)
+
+#define SVGA3D_PS_10 ((SVGA3D_PS_TYPE << 16) | 1 << 8)
+#define SVGA3D_PS_11 (SVGA3D_PS_10 | 1)
+#define SVGA3D_PS_12 (SVGA3D_PS_10 | 2)
+#define SVGA3D_PS_13 (SVGA3D_PS_10 | 3)
+#define SVGA3D_PS_14 (SVGA3D_PS_10 | 4)
+#define SVGA3D_PS_20 ((SVGA3D_PS_TYPE << 16) | 2 << 8)
+#define SVGA3D_PS_30 ((SVGA3D_PS_TYPE << 16) | 3 << 8)
+
+/* The *_ENABLED are for backwards compatibility with old drivers */
+typedef enum {
+   SVGA3DPSVERSION_NONE = 0,
+   SVGA3DPSVERSION_ENABLED = 1,
+   SVGA3DPSVERSION_11 = 3,
+   SVGA3DPSVERSION_12 = 5,
+   SVGA3DPSVERSION_13 = 7,
+   SVGA3DPSVERSION_14 = 9,
+   SVGA3DPSVERSION_20 = 11,
+   SVGA3DPSVERSION_30 = 13,
+   SVGA3DPSVERSION_40 = 15,
+   SVGA3DPSVERSION_MAX
+} SVGA3dPixelShaderVersion;
+
+typedef enum {
+   SVGA3DVSVERSION_NONE = 0,
+   SVGA3DVSVERSION_ENABLED = 1,
+   SVGA3DVSVERSION_11 = 3,
+   SVGA3DVSVERSION_20 = 5,
+   SVGA3DVSVERSION_30 = 7,
+   SVGA3DVSVERSION_40 = 9,
+   SVGA3DVSVERSION_MAX
+} SVGA3dVertexShaderVersion;
+
+/* SVGA3D instruction op codes. */
+
+typedef enum {
+   SVGA3DOP_NOP = 0,
+   SVGA3DOP_MOV,
+   SVGA3DOP_ADD,
+   SVGA3DOP_SUB,
+   SVGA3DOP_MAD,
+   SVGA3DOP_MUL,
+   SVGA3DOP_RCP,
+   SVGA3DOP_RSQ,
+   SVGA3DOP_DP3,
+   SVGA3DOP_DP4,
+   SVGA3DOP_MIN,
+   SVGA3DOP_MAX,
+   SVGA3DOP_SLT,
+   SVGA3DOP_SGE,
+   SVGA3DOP_EXP,
+   SVGA3DOP_LOG,
+   SVGA3DOP_LIT,
+   SVGA3DOP_DST,
+   SVGA3DOP_LRP,
+   SVGA3DOP_FRC,
+   SVGA3DOP_M4x4,
+   SVGA3DOP_M4x3,
+   SVGA3DOP_M3x4,
+   SVGA3DOP_M3x3,
+   SVGA3DOP_M3x2,
+   SVGA3DOP_CALL,
+   SVGA3DOP_CALLNZ,
+   SVGA3DOP_LOOP,
+   SVGA3DOP_RET,
+   SVGA3DOP_ENDLOOP,
+   SVGA3DOP_LABEL,
+   SVGA3DOP_DCL,
+   SVGA3DOP_POW,
+   SVGA3DOP_CRS,
+   SVGA3DOP_SGN,
+   SVGA3DOP_ABS,
+   SVGA3DOP_NRM,
+   SVGA3DOP_SINCOS,
+   SVGA3DOP_REP,
+   SVGA3DOP_ENDREP,
+   SVGA3DOP_IF,
+   SVGA3DOP_IFC,
+   SVGA3DOP_ELSE,
+   SVGA3DOP_ENDIF,
+   SVGA3DOP_BREAK,
+   SVGA3DOP_BREAKC,
+   SVGA3DOP_MOVA,
+   SVGA3DOP_DEFB,
+   SVGA3DOP_DEFI,
+   SVGA3DOP_TEXCOORD = 64,
+   SVGA3DOP_TEXKILL,
+   SVGA3DOP_TEX,
+   SVGA3DOP_TEXBEM,
+   SVGA3DOP_TEXBEML,
+   SVGA3DOP_TEXREG2AR,
+   SVGA3DOP_TEXREG2GB = 70,
+   SVGA3DOP_TEXM3x2PAD,
+   SVGA3DOP_TEXM3x2TEX,
+   SVGA3DOP_TEXM3x3PAD,
+   SVGA3DOP_TEXM3x3TEX,
+   SVGA3DOP_RESERVED0,
+   SVGA3DOP_TEXM3x3SPEC,
+   SVGA3DOP_TEXM3x3VSPEC,
+   SVGA3DOP_EXPP,
+   SVGA3DOP_LOGP,
+   SVGA3DOP_CND = 80,
+   SVGA3DOP_DEF,
+   SVGA3DOP_TEXREG2RGB,
+   SVGA3DOP_TEXDP3TEX,
+   SVGA3DOP_TEXM3x2DEPTH,
+   SVGA3DOP_TEXDP3,
+   SVGA3DOP_TEXM3x3,
+   SVGA3DOP_TEXDEPTH,
+   SVGA3DOP_CMP,
+   SVGA3DOP_BEM,
+   SVGA3DOP_DP2ADD = 90,
+   SVGA3DOP_DSX,
+   SVGA3DOP_DSY,
+   SVGA3DOP_TEXLDD,
+   SVGA3DOP_SETP,
+   SVGA3DOP_TEXLDL,
+   SVGA3DOP_BREAKP = 96,
+   SVGA3DOP_LAST_INST,
+   SVGA3DOP_PHASE = 0xFFFD,
+   SVGA3DOP_COMMENT = 0xFFFE,
+   SVGA3DOP_END = 0xFFFF,
+} SVGA3dShaderOpCodeType;
+
+/* SVGA3D operation control/comparison function types */
+
+typedef enum {
+   SVGA3DOPCONT_NONE,
+   SVGA3DOPCONT_PROJECT,   /* Projective texturing */
+   SVGA3DOPCONT_BIAS,      /* Texturing with a LOD bias */
+} SVGA3dShaderOpCodeControlFnType;
+
+typedef enum {
+   SVGA3DOPCOMP_RESERVED0 = 0,
+   SVGA3DOPCOMP_GT,
+   SVGA3DOPCOMP_EQ,
+   SVGA3DOPCOMP_GE,
+   SVGA3DOPCOMP_LT,
+   SVGA3DOPCOMPC_NE,
+   SVGA3DOPCOMP_LE,
+   SVGA3DOPCOMP_RESERVED1
+} SVGA3dShaderOpCodeCompFnType;
+
+/* SVGA3D register types */
+
+typedef enum {
+    SVGA3DREG_TEMP = 0,       /* Temporary register file */
+    SVGA3DREG_INPUT,          /* Input register file */
+    SVGA3DREG_CONST,          /* Constant register file */
+    SVGA3DREG_ADDR,           /* Address register for VS */
+    SVGA3DREG_TEXTURE = 3,    /* Texture register file for PS */
+    SVGA3DREG_RASTOUT,        /* Rasterizer register file */
+    SVGA3DREG_ATTROUT,        /* Attribute output register file */
+    SVGA3DREG_TEXCRDOUT,      /* Texture coordinate output register file */
+    SVGA3DREG_OUTPUT = 6,     /* Output register file for VS 3.0+ */
+    SVGA3DREG_CONSTINT,       /* Constant integer vector register file */
+    SVGA3DREG_COLOROUT,       /* Color output register file */
+    SVGA3DREG_DEPTHOUT,       /* Depth output register file */
+    SVGA3DREG_SAMPLER,        /* Sampler state register file */
+    SVGA3DREG_CONST2,         /* Constant register file 2048 - 4095 */
+    SVGA3DREG_CONST3,         /* Constant register file 4096 - 6143 */
+    SVGA3DREG_CONST4,         /* Constant register file 6144 - 8191 */
+    SVGA3DREG_CONSTBOOL,      /* Constant boolean register file */
+    SVGA3DREG_LOOP,           /* Loop counter register file */
+    SVGA3DREG_TEMPFLOAT16,    /* 16-bit float temp register file */
+    SVGA3DREG_MISCTYPE,       /* Miscellaneous (single) registers */
+    SVGA3DREG_LABEL,          /* Label */
+    SVGA3DREG_PREDICATE,      /* Predicate register */
+} SVGA3dShaderRegType;
+
+/* SVGA3D rasterizer output register types */
+
+typedef enum {
+   SVGA3DRASTOUT_POSITION = 0,
+   SVGA3DRASTOUT_FOG,
+   SVGA3DRASTOUT_PSIZE
+} SVGA3dShaderRastOutRegType;
+
+/* SVGA3D miscellaneous register types */
+
+typedef enum {
+   SVGA3DMISCREG_POSITION = 0,   /* Input position x,y,z,rhw (PS) */
+   SVGA3DMISCREG_FACE            /* Floating point primitive area (PS) */
+} SVGA3DShaderMiscRegType;
+
+/* SVGA3D sampler types */
+
+typedef enum {
+   SVGA3DSAMP_UNKNOWN = 0, /* Uninitialized value */
+   SVGA3DSAMP_2D = 2,      /* dcl_2d s# (for declaring a 2-D texture) */
+   SVGA3DSAMP_CUBE,        /* dcl_cube s# (for declaring a cube texture) */
+   SVGA3DSAMP_VOLUME,      /* dcl_volume s# (for declaring a volume texture) */
+} SVGA3dShaderSamplerType;
+
+/* SVGA3D sampler format classes */
+
+typedef enum {
+   SVGA3DSAMPFORMAT_ARGB,        /* ARGB formats */
+   SVGA3DSAMPFORMAT_V8U8,        /* Sign and normalize (SNORM) V & U */
+   SVGA3DSAMPFORMAT_Q8W8V8U8,    /* SNORM all */
+   SVGA3DSAMPFORMAT_CxV8U8,      /* SNORM V & U, C=SQRT(1-U^2-V^2) */
+   SVGA3DSAMPFORMAT_X8L8V8U8,    /* SNORM V & U */
+   SVGA3DSAMPFORMAT_A2W10V10U10, /* SNORM W, V & U */
+   SVGA3DSAMPFORMAT_DXT_PMA,     /* DXT pre-multiplied alpha */
+   SVGA3DSAMPFORMAT_YUV,         /* YUV video format */
+   SVGA3DSAMPFORMAT_UYVY,        /* UYVY video format */
+   SVGA3DSAMPFORMAT_Rx,          /* R16F/32F */
+   SVGA3DSAMPFORMAT_RxGx,        /* R16FG16F, R32FG32F */
+   SVGA3DSAMPFORMAT_V16U16,      /* SNORM all */
+} SVGA3DShaderSamplerFormatClass;
+
+/* SVGA3D write mask */
+
+#define SVGA3DWRITEMASK_0    1 /* Component 0 (X;Red) */
+#define SVGA3DWRITEMASK_1    2 /* Component 1 (Y;Green) */
+#define SVGA3DWRITEMASK_2    4 /* Component 2 (Z;Blue) */
+#define SVGA3DWRITEMASK_3    8 /* Component 3 (W;Alpha) */
+#define SVGA3DWRITEMASK_ALL 15 /* All components */
+
+/* SVGA3D destination modifiers */
+
+#define SVGA3DDSTMOD_NONE              0 /* nop */
+#define SVGA3DDSTMOD_SATURATE          1 /* clamp to [0, 1] */
+#define SVGA3DDSTMOD_PARTIALPRECISION  2 /* Partial precision hint */
+
+/*
+ * Relevant to multisampling only:
+ * When the pixel center is not covered, sample
+ * attribute or compute gradients/LOD
+ * using multisample "centroid" location.
+ * "Centroid" is some location within the covered
+ * region of the pixel.
+ */
+
+#define SVGA3DDSTMOD_MSAMPCENTROID     4
+
+/* SVGA3D source swizzle */
+
+#define SVGA3DSWIZZLE_REPLICATEX 0x00
+#define SVGA3DSWIZZLE_REPLICATEY 0x55
+#define SVGA3DSWIZZLE_REPLICATEZ 0xAA
+#define SVGA3DSWIZZLE_REPLICATEW 0xFF
+#define SVGA3DSWIZZLE_NONE       0xE4
+#define SVGA3DSWIZZLE_YZXW       0xC9
+#define SVGA3DSWIZZLE_ZXYW       0xD2
+#define SVGA3DSWIZZLE_WXYZ       0x1B
+
+/* SVGA3D source modifiers */
+
+typedef enum {
+    SVGA3DSRCMOD_NONE = 0, /* nop */
+    SVGA3DSRCMOD_NEG,      /* negate */
+    SVGA3DSRCMOD_BIAS,     /* bias */
+    SVGA3DSRCMOD_BIASNEG,  /* bias and negate */
+    SVGA3DSRCMOD_SIGN,     /* sign */
+    SVGA3DSRCMOD_SIGNNEG,  /* sign and negate */
+    SVGA3DSRCMOD_COMP,     /* complement */
+    SVGA3DSRCMOD_X2,       /* x2 */
+    SVGA3DSRCMOD_X2NEG,    /* x2 and negate */
+    SVGA3DSRCMOD_DZ,       /* divide through by z component */
+    SVGA3DSRCMOD_DW,       /* divide through by w component */
+    SVGA3DSRCMOD_ABS,      /* abs() */
+    SVGA3DSRCMOD_ABSNEG,   /* -abs() */
+    SVGA3DSRCMOD_NOT,      /* ! (for predicate register) */
+} SVGA3dShaderSrcModType;
+
+/* SVGA3D instruction token */
+
+typedef struct {
+   union {
+      struct {
+         uint32 comment_op : 16;
+         uint32 comment_size : 16;
+      };
+
+      struct {
+         uint32 op : 16;
+         uint32 control : 3;
+         uint32 reserved2 : 5;
+         uint32 size : 4;
+         uint32 predicated : 1;
+         uint32 reserved1 : 1;
+         uint32 coissue : 1;
+         uint32 reserved0 : 1;
+      };
+
+      uint32 value;
+   };
+} SVGA3dShaderInstToken;
+
+/* SVGA3D destination parameter token */
+
+typedef struct {
+   union {
+      struct {
+         uint32 num : 11;
+         uint32 type_upper : 2;
+         uint32 relAddr : 1;
+         uint32 reserved1 : 2;
+         uint32 mask : 4;
+         uint32 dstMod : 4;
+         uint32 shfScale : 4;
+         uint32 type_lower : 3;
+         uint32 reserved0 : 1;
+      };
+
+      uint32 value;
+   };
+} SVGA3dShaderDestToken;
+
+/* SVGA3D source parameter token */
+
+typedef struct {
+   union {
+      struct {
+         uint32 num : 11;
+         uint32 type_upper : 2;
+         uint32 relAddr : 1;
+         uint32 reserved1 : 2;
+         uint32 swizzle : 8;
+         uint32 srcMod : 4;
+         uint32 type_lower : 3;
+         uint32 reserved0 : 1;
+      };
+
+      uint32 value;
+   };
+} SVGA3dShaderSrcToken;
+
+/* SVGA3DOP_DCL parameter tokens */
+
+typedef struct {
+   union {
+      struct {
+         union {
+            struct {
+               uint32 usage : 5;
+               uint32 reserved1 : 11;
+               uint32 index : 4;
+               uint32 reserved0 : 12;
+            }; /* input / output declaration */
+
+            struct {
+               uint32 reserved3 : 27;
+               uint32 type : 4;
+               uint32 reserved2 : 1;
+            }; /* sampler declaration */
+         };
+
+         SVGA3dShaderDestToken dst;
+      };
+
+      uint32 values[2];
+   };
+} SVGA3DOpDclArgs;
+
+/* SVGA3DOP_DEF parameter tokens */
+
+typedef struct {
+   union {
+      struct {
+         SVGA3dShaderDestToken dst;
+
+         union {
+            float constValues[4];
+            int constIValues[4];
+            Bool constBValue;
+         };
+      };
+
+      uint32 values[5];
+   };
+} SVGA3DOpDefArgs;
+
+/* SVGA3D shader token */
+
+typedef union {
+   uint32 value;
+   SVGA3dShaderInstToken inst;
+   SVGA3dShaderDestToken dest;
+   SVGA3dShaderSrcToken src;
+} SVGA3dShaderToken;
+
+/* SVGA3D shader program */
+
+typedef struct {
+   SVGA3dShaderVersion version;
+   /* SVGA3dShaderToken stream */
+} SVGA3dShaderProgram;
+
+/* SVGA3D version specific register assignments */
+
+static const uint32 SVGA3D_INPUT_REG_POSITION_VS11 = 0;
+static const uint32 SVGA3D_INPUT_REG_PSIZE_VS11 = 1;
+static const uint32 SVGA3D_INPUT_REG_FOG_VS11 = 3;
+static const uint32 SVGA3D_INPUT_REG_FOG_MASK_VS11 = SVGA3DWRITEMASK_3;
+static const uint32 SVGA3D_INPUT_REG_COLOR_BASE_VS11 = 2;
+static const uint32 SVGA3D_INPUT_REG_TEXCOORD_BASE_VS11 = 4;
+
+static const uint32 SVGA3D_INPUT_REG_COLOR_BASE_PS11 = 0;
+static const uint32 SVGA3D_INPUT_REG_TEXCOORD_BASE_PS11 = 2;
+static const uint32 SVGA3D_OUTPUT_REG_DEPTH_PS11 = 0;
+static const uint32 SVGA3D_OUTPUT_REG_COLOR_PS11 = 1;
+
+static const uint32 SVGA3D_INPUT_REG_COLOR_BASE_PS20 = 0;
+static const uint32 SVGA3D_INPUT_REG_COLOR_NUM_PS20 = 2;
+static const uint32 SVGA3D_INPUT_REG_TEXCOORD_BASE_PS20 = 2;
+static const uint32 SVGA3D_INPUT_REG_TEXCOORD_NUM_PS20 = 8;
+static const uint32 SVGA3D_OUTPUT_REG_COLOR_BASE_PS20 = 1;
+static const uint32 SVGA3D_OUTPUT_REG_COLOR_NUM_PS20 = 4;
+static const uint32 SVGA3D_OUTPUT_REG_DEPTH_BASE_PS20 = 0;
+static const uint32 SVGA3D_OUTPUT_REG_DEPTH_NUM_PS20 = 1;
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3dShaderGetRegType --
+ *
+ *      As the register type is split into two non sequential fields,
+ *      this function provides an useful way of accessing the actual
+ *      register type without having to manually concatenate the
+ *      type_upper and type_lower fields.
+ *
+ * Results:
+ *      Returns the register type.
+ *
+ *----------------------------------------------------------------------
+ */
+
+static INLINE SVGA3dShaderRegType
+SVGA3dShaderGetRegType(uint32 token)
+{
+   SVGA3dShaderSrcToken src;
+   src.value = token;
+   return (SVGA3dShaderRegType)(src.type_upper << 3 | src.type_lower);
+}
+
+#endif /* __SVGA3D_SHADER_DEFS__ */
diff --git a/src/gallium/drivers/svga/include/svga_reg.h b/src/gallium/drivers/svga/include/svga_reg.h
new file mode 100644
index 0000000000..1b96c2ec07
--- /dev/null
+++ b/src/gallium/drivers/svga/include/svga_reg.h
@@ -0,0 +1,1346 @@
+/**********************************************************
+ * Copyright 1998-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/*
+ * svga_reg.h --
+ *
+ *    Virtual hardware definitions for the VMware SVGA II device.
+ */
+
+#ifndef _SVGA_REG_H_
+#define _SVGA_REG_H_
+
+/*
+ * PCI device IDs.
+ */
+#define PCI_VENDOR_ID_VMWARE            0x15AD
+#define PCI_DEVICE_ID_VMWARE_SVGA2      0x0405
+
+/*
+ * Legal values for the SVGA_REG_CURSOR_ON register in old-fashioned
+ * cursor bypass mode. This is still supported, but no new guest
+ * drivers should use it.
+ */
+#define SVGA_CURSOR_ON_HIDE            0x0   /* Must be 0 to maintain backward compatibility */
+#define SVGA_CURSOR_ON_SHOW            0x1   /* Must be 1 to maintain backward compatibility */
+#define SVGA_CURSOR_ON_REMOVE_FROM_FB  0x2   /* Remove the cursor from the framebuffer because we need to see what's under it */
+#define SVGA_CURSOR_ON_RESTORE_TO_FB   0x3   /* Put the cursor back in the framebuffer so the user can see it */
+
+/*
+ * The maximum framebuffer size that can traced for e.g. guests in VESA mode.
+ * The changeMap in the monitor is proportional to this number. Therefore, we'd
+ * like to keep it as small as possible to reduce monitor overhead (using
+ * SVGA_VRAM_MAX_SIZE for this increases the size of the shared area by over
+ * 4k!).
+ *
+ * NB: For compatibility reasons, this value must be greater than 0xff0000.
+ *     See bug 335072.
+ */
+#define SVGA_FB_MAX_TRACEABLE_SIZE      0x1000000
+
+#define SVGA_MAX_PSEUDOCOLOR_DEPTH      8
+#define SVGA_MAX_PSEUDOCOLORS           (1 << SVGA_MAX_PSEUDOCOLOR_DEPTH)
+#define SVGA_NUM_PALETTE_REGS           (3 * SVGA_MAX_PSEUDOCOLORS)
+
+#define SVGA_MAGIC         0x900000UL
+#define SVGA_MAKE_ID(ver)  (SVGA_MAGIC << 8 | (ver))
+
+/* Version 2 let the address of the frame buffer be unsigned on Win32 */
+#define SVGA_VERSION_2     2
+#define SVGA_ID_2          SVGA_MAKE_ID(SVGA_VERSION_2)
+
+/* Version 1 has new registers starting with SVGA_REG_CAPABILITIES so
+   PALETTE_BASE has moved */
+#define SVGA_VERSION_1     1
+#define SVGA_ID_1          SVGA_MAKE_ID(SVGA_VERSION_1)
+
+/* Version 0 is the initial version */
+#define SVGA_VERSION_0     0
+#define SVGA_ID_0          SVGA_MAKE_ID(SVGA_VERSION_0)
+
+/* "Invalid" value for all SVGA IDs. (Version ID, screen object ID, surface ID...) */
+#define SVGA_ID_INVALID    0xFFFFFFFF
+
+/* Port offsets, relative to BAR0 */
+#define SVGA_INDEX_PORT         0x0
+#define SVGA_VALUE_PORT         0x1
+#define SVGA_BIOS_PORT          0x2
+#define SVGA_IRQSTATUS_PORT     0x8
+
+/*
+ * Interrupt source flags for IRQSTATUS_PORT and IRQMASK.
+ *
+ * Interrupts are only supported when the
+ * SVGA_CAP_IRQMASK capability is present.
+ */
+#define SVGA_IRQFLAG_ANY_FENCE            0x1    /* Any fence was passed */
+#define SVGA_IRQFLAG_FIFO_PROGRESS        0x2    /* Made forward progress in the FIFO */
+#define SVGA_IRQFLAG_FENCE_GOAL           0x4    /* SVGA_FIFO_FENCE_GOAL reached */
+
+/*
+ * Registers
+ */
+
+enum {
+   SVGA_REG_ID = 0,
+   SVGA_REG_ENABLE = 1,
+   SVGA_REG_WIDTH = 2,
+   SVGA_REG_HEIGHT = 3,
+   SVGA_REG_MAX_WIDTH = 4,
+   SVGA_REG_MAX_HEIGHT = 5,
+   SVGA_REG_DEPTH = 6,
+   SVGA_REG_BITS_PER_PIXEL = 7,       /* Current bpp in the guest */
+   SVGA_REG_PSEUDOCOLOR = 8,
+   SVGA_REG_RED_MASK = 9,
+   SVGA_REG_GREEN_MASK = 10,
+   SVGA_REG_BLUE_MASK = 11,
+   SVGA_REG_BYTES_PER_LINE = 12,
+   SVGA_REG_FB_START = 13,            /* (Deprecated) */
+   SVGA_REG_FB_OFFSET = 14,
+   SVGA_REG_VRAM_SIZE = 15,
+   SVGA_REG_FB_SIZE = 16,
+
+   /* ID 0 implementation only had the above registers, then the palette */
+
+   SVGA_REG_CAPABILITIES = 17,
+   SVGA_REG_MEM_START = 18,           /* (Deprecated) */
+   SVGA_REG_MEM_SIZE = 19,
+   SVGA_REG_CONFIG_DONE = 20,         /* Set when memory area configured */
+   SVGA_REG_SYNC = 21,                /* See "FIFO Synchronization Registers" */
+   SVGA_REG_BUSY = 22,                /* See "FIFO Synchronization Registers" */
+   SVGA_REG_GUEST_ID = 23,            /* Set guest OS identifier */
+   SVGA_REG_CURSOR_ID = 24,           /* (Deprecated) */
+   SVGA_REG_CURSOR_X = 25,            /* (Deprecated) */
+   SVGA_REG_CURSOR_Y = 26,            /* (Deprecated) */
+   SVGA_REG_CURSOR_ON = 27,           /* (Deprecated) */
+   SVGA_REG_HOST_BITS_PER_PIXEL = 28, /* (Deprecated) */
+   SVGA_REG_SCRATCH_SIZE = 29,        /* Number of scratch registers */
+   SVGA_REG_MEM_REGS = 30,            /* Number of FIFO registers */
+   SVGA_REG_NUM_DISPLAYS = 31,        /* (Deprecated) */
+   SVGA_REG_PITCHLOCK = 32,           /* Fixed pitch for all modes */
+   SVGA_REG_IRQMASK = 33,             /* Interrupt mask */
+
+   /* Legacy multi-monitor support */
+   SVGA_REG_NUM_GUEST_DISPLAYS = 34,/* Number of guest displays in X/Y direction */
+   SVGA_REG_DISPLAY_ID = 35,        /* Display ID for the following display attributes */
+   SVGA_REG_DISPLAY_IS_PRIMARY = 36,/* Whether this is a primary display */
+   SVGA_REG_DISPLAY_POSITION_X = 37,/* The display position x */
+   SVGA_REG_DISPLAY_POSITION_Y = 38,/* The display position y */
+   SVGA_REG_DISPLAY_WIDTH = 39,     /* The display's width */
+   SVGA_REG_DISPLAY_HEIGHT = 40,    /* The display's height */
+
+   /* See "Guest memory regions" below. */
+   SVGA_REG_GMR_ID = 41,
+   SVGA_REG_GMR_DESCRIPTOR = 42,
+   SVGA_REG_GMR_MAX_IDS = 43,
+   SVGA_REG_GMR_MAX_DESCRIPTOR_LENGTH = 44,
+
+   SVGA_REG_TRACES = 45,            /* Enable trace-based updates even when FIFO is on */
+   SVGA_REG_TOP = 46,               /* Must be 1 more than the last register */
+
+   SVGA_PALETTE_BASE = 1024,        /* Base of SVGA color map */
+   /* Next 768 (== 256*3) registers exist for colormap */
+
+   SVGA_SCRATCH_BASE = SVGA_PALETTE_BASE + SVGA_NUM_PALETTE_REGS
+                                    /* Base of scratch registers */
+   /* Next reg[SVGA_REG_SCRATCH_SIZE] registers exist for scratch usage:
+      First 4 are reserved for VESA BIOS Extension; any remaining are for
+      the use of the current SVGA driver. */
+};
+
+
+/*
+ * Guest memory regions (GMRs):
+ *
+ * This is a new memory mapping feature available in SVGA devices
+ * which have the SVGA_CAP_GMR bit set. Previously, there were two
+ * fixed memory regions available with which to share data between the
+ * device and the driver: the FIFO ('MEM') and the framebuffer. GMRs
+ * are our name for an extensible way of providing arbitrary DMA
+ * buffers for use between the driver and the SVGA device. They are a
+ * new alternative to framebuffer memory, usable for both 2D and 3D
+ * graphics operations.
+ *
+ * Since GMR mapping must be done synchronously with guest CPU
+ * execution, we use a new pair of SVGA registers:
+ *
+ *   SVGA_REG_GMR_ID --
+ *
+ *     Read/write.
+ *     This register holds the 32-bit ID (a small positive integer)
+ *     of a GMR to create, delete, or redefine. Writing this register
+ *     has no side-effects.
+ *
+ *   SVGA_REG_GMR_DESCRIPTOR --
+ *
+ *     Write-only.
+ *     Writing this register will create, delete, or redefine the GMR
+ *     specified by the above ID register. If this register is zero,
+ *     the GMR is deleted. Any pointers into this GMR (including those
+ *     currently being processed by FIFO commands) will be
+ *     synchronously invalidated.
+ *
+ *     If this register is nonzero, it must be the physical page
+ *     number (PPN) of a data structure which describes the physical
+ *     layout of the memory region this GMR should describe. The
+ *     descriptor structure will be read synchronously by the SVGA
+ *     device when this register is written. The descriptor need not
+ *     remain allocated for the lifetime of the GMR.
+ *
+ *     The guest driver should write SVGA_REG_GMR_ID first, then
+ *     SVGA_REG_GMR_DESCRIPTOR.
+ *
+ *   SVGA_REG_GMR_MAX_IDS --
+ *
+ *     Read-only.
+ *     The SVGA device may choose to support a maximum number of
+ *     user-defined GMR IDs. This register holds the number of supported
+ *     IDs. (The maximum supported ID plus 1)
+ *
+ *   SVGA_REG_GMR_MAX_DESCRIPTOR_LENGTH --
+ *
+ *     Read-only.
+ *     The SVGA device may choose to put a limit on the total number
+ *     of SVGAGuestMemDescriptor structures it will read when defining
+ *     a single GMR.
+ *
+ * The descriptor structure is an array of SVGAGuestMemDescriptor
+ * structures. Each structure may do one of three things:
+ *
+ *   - Terminate the GMR descriptor list.
+ *     (ppn==0, numPages==0)
+ *
+ *   - Add a PPN or range of PPNs to the GMR's virtual address space.
+ *     (ppn != 0, numPages != 0)
+ *
+ *   - Provide the PPN of the next SVGAGuestMemDescriptor, in order to
+ *     support multi-page GMR descriptor tables without forcing the
+ *     driver to allocate physically contiguous memory.
+ *     (ppn != 0, numPages == 0)
+ *
+ * Note that each physical page of SVGAGuestMemDescriptor structures
+ * can describe at least 2MB of guest memory. If the driver needs to
+ * use more than one page of descriptor structures, it must use one of
+ * its SVGAGuestMemDescriptors to point to an additional page.  The
+ * device will never automatically cross a page boundary.
+ *
+ * Once the driver has described a GMR, it is immediately available
+ * for use via any FIFO command that uses an SVGAGuestPtr structure.
+ * These pointers include a GMR identifier plus an offset into that
+ * GMR.
+ *
+ * The driver must check the SVGA_CAP_GMR bit before using the GMR
+ * registers.
+ */
+
+/*
+ * Special GMR IDs, allowing SVGAGuestPtrs to point to framebuffer
+ * memory as well.  In the future, these IDs could even be used to
+ * allow legacy memory regions to be redefined by the guest as GMRs.
+ *
+ * Using the guest framebuffer (GFB) at BAR1 for general purpose DMA
+ * is being phased out. Please try to use user-defined GMRs whenever
+ * possible.
+ */
+#define SVGA_GMR_NULL         ((uint32) -1)
+#define SVGA_GMR_FRAMEBUFFER  ((uint32) -2)  // Guest Framebuffer (GFB)
+
+typedef
+struct SVGAGuestMemDescriptor {
+   uint32 ppn;
+   uint32 numPages;
+} SVGAGuestMemDescriptor;
+
+typedef
+struct SVGAGuestPtr {
+   uint32 gmrId;
+   uint32 offset;
+} SVGAGuestPtr;
+
+
+/*
+ * SVGAGMRImageFormat --
+ *
+ *    This is a packed representation of the source 2D image format
+ *    for a GMR-to-screen blit. Currently it is defined as an encoding
+ *    of the screen's color depth and bits-per-pixel, however, 16 bits
+ *    are reserved for future use to identify other encodings (such as
+ *    RGBA or higher-precision images).
+ *
+ *    Currently supported formats:
+ *
+ *       bpp depth  Format Name
+ *       --- -----  -----------
+ *        32    24  32-bit BGRX
+ *        24    24  24-bit BGR
+ *        16    16  RGB 5-6-5
+ *        16    15  RGB 5-5-5
+ *
+ */
+
+typedef
+struct SVGAGMRImageFormat {
+   union {
+      struct {
+         uint32 bitsPerPixel : 8;
+         uint32 colorDepth   : 8;
+         uint32 reserved     : 16;  // Must be zero
+      };
+
+      uint32 value;
+   };
+} SVGAGMRImageFormat;
+
+/*
+ * SVGAColorBGRX --
+ *
+ *    A 24-bit color format (BGRX), which does not depend on the
+ *    format of the legacy guest framebuffer (GFB) or the current
+ *    GMRFB state.
+ */
+
+typedef
+struct SVGAColorBGRX {
+   union {
+      struct {
+         uint32 b : 8;
+         uint32 g : 8;
+         uint32 r : 8;
+         uint32 x : 8;  // Unused
+      };
+
+      uint32 value;
+   };
+} SVGAColorBGRX;
+
+
+/*
+ * SVGASignedRect --
+ * SVGASignedPoint --
+ *
+ *    Signed rectangle and point primitives. These are used by the new
+ *    2D primitives for drawing to Screen Objects, which can occupy a
+ *    signed virtual coordinate space.
+ *
+ *    SVGASignedRect specifies a half-open interval: the (left, top)
+ *    pixel is part of the rectangle, but the (right, bottom) pixel is
+ *    not.
+ */
+
+typedef
+struct SVGASignedRect {
+   int32  left;
+   int32  top;
+   int32  right;
+   int32  bottom;
+} SVGASignedRect;
+
+typedef
+struct SVGASignedPoint {
+   int32  x;
+   int32  y;
+} SVGASignedPoint;
+
+
+/*
+ *  Capabilities
+ *
+ *  Note the holes in the bitfield. Missing bits have been deprecated,
+ *  and must not be reused. Those capabilities will never be reported
+ *  by new versions of the SVGA device.
+ */
+
+#define SVGA_CAP_NONE               0x00000000
+#define SVGA_CAP_RECT_COPY          0x00000002
+#define SVGA_CAP_CURSOR             0x00000020
+#define SVGA_CAP_CURSOR_BYPASS      0x00000040   // Legacy (Use Cursor Bypass 3 instead)
+#define SVGA_CAP_CURSOR_BYPASS_2    0x00000080   // Legacy (Use Cursor Bypass 3 instead)
+#define SVGA_CAP_8BIT_EMULATION     0x00000100
+#define SVGA_CAP_ALPHA_CURSOR       0x00000200
+#define SVGA_CAP_3D                 0x00004000
+#define SVGA_CAP_EXTENDED_FIFO      0x00008000
+#define SVGA_CAP_MULTIMON           0x00010000   // Legacy multi-monitor support
+#define SVGA_CAP_PITCHLOCK          0x00020000
+#define SVGA_CAP_IRQMASK            0x00040000
+#define SVGA_CAP_DISPLAY_TOPOLOGY   0x00080000   // Legacy multi-monitor support
+#define SVGA_CAP_GMR                0x00100000
+#define SVGA_CAP_TRACES             0x00200000
+
+
+/*
+ * FIFO register indices.
+ *
+ * The FIFO is a chunk of device memory mapped into guest physmem.  It
+ * is always treated as 32-bit words.
+ *
+ * The guest driver gets to decide how to partition it between
+ * - FIFO registers (there are always at least 4, specifying where the
+ *   following data area is and how much data it contains; there may be
+ *   more registers following these, depending on the FIFO protocol
+ *   version in use)
+ * - FIFO data, written by the guest and slurped out by the VMX.
+ * These indices are 32-bit word offsets into the FIFO.
+ */
+
+enum {
+   /*
+    * Block 1 (basic registers): The originally defined FIFO registers.
+    * These exist and are valid for all versions of the FIFO protocol.
+    */
+
+   SVGA_FIFO_MIN = 0,
+   SVGA_FIFO_MAX,       /* The distance from MIN to MAX must be at least 10K */
+   SVGA_FIFO_NEXT_CMD,
+   SVGA_FIFO_STOP,
+
+   /*
+    * Block 2 (extended registers): Mandatory registers for the extended
+    * FIFO.  These exist if the SVGA caps register includes
+    * SVGA_CAP_EXTENDED_FIFO; some of them are valid only if their
+    * associated capability bit is enabled.
+    *
+    * Note that when originally defined, SVGA_CAP_EXTENDED_FIFO implied
+    * support only for (FIFO registers) CAPABILITIES, FLAGS, and FENCE.
+    * This means that the guest has to test individually (in most cases
+    * using FIFO caps) for the presence of registers after this; the VMX
+    * can define "extended FIFO" to mean whatever it wants, and currently
+    * won't enable it unless there's room for that set and much more.
+    */
+
+   SVGA_FIFO_CAPABILITIES = 4,
+   SVGA_FIFO_FLAGS,
+   // Valid with SVGA_FIFO_CAP_FENCE:
+   SVGA_FIFO_FENCE,
+
+   /*
+    * Block 3a (optional extended registers): Additional registers for the
+    * extended FIFO, whose presence isn't actually implied by
+    * SVGA_CAP_EXTENDED_FIFO; these exist if SVGA_FIFO_MIN is high enough to
+    * leave room for them.
+    *
+    * These in block 3a, the VMX currently considers mandatory for the
+    * extended FIFO.
+    */
+
+   // Valid if exists (i.e. if extended FIFO enabled):
+   SVGA_FIFO_3D_HWVERSION,       /* See SVGA3dHardwareVersion in svga3d_reg.h */
+   // Valid with SVGA_FIFO_CAP_PITCHLOCK:
+   SVGA_FIFO_PITCHLOCK,
+
+   // Valid with SVGA_FIFO_CAP_CURSOR_BYPASS_3:
+   SVGA_FIFO_CURSOR_ON,          /* Cursor bypass 3 show/hide register */
+   SVGA_FIFO_CURSOR_X,           /* Cursor bypass 3 x register */
+   SVGA_FIFO_CURSOR_Y,           /* Cursor bypass 3 y register */
+   SVGA_FIFO_CURSOR_COUNT,       /* Incremented when any of the other 3 change */
+   SVGA_FIFO_CURSOR_LAST_UPDATED,/* Last time the host updated the cursor */
+
+   // Valid with SVGA_FIFO_CAP_RESERVE:
+   SVGA_FIFO_RESERVED,           /* Bytes past NEXT_CMD with real contents */
+
+   /*
+    * Valid with SVGA_FIFO_CAP_SCREEN_OBJECT:
+    *
+    * By default this is SVGA_ID_INVALID, to indicate that the cursor
+    * coordinates are specified relative to the virtual root. If this
+    * is set to a specific screen ID, cursor position is reinterpreted
+    * as a signed offset relative to that screen's origin. This is the
+    * only way to place the cursor on a non-rooted screen.
+    */
+   SVGA_FIFO_CURSOR_SCREEN_ID,
+
+   /*
+    * XXX: The gap here, up until SVGA_FIFO_3D_CAPS, can be used for new
+    * registers, but this must be done carefully and with judicious use of
+    * capability bits, since comparisons based on SVGA_FIFO_MIN aren't
+    * enough to tell you whether the register exists: we've shipped drivers
+    * and products that used SVGA_FIFO_3D_CAPS but didn't know about some of
+    * the earlier ones.  The actual order of introduction was:
+    * - PITCHLOCK
+    * - 3D_CAPS
+    * - CURSOR_* (cursor bypass 3)
+    * - RESERVED
+    * So, code that wants to know whether it can use any of the
+    * aforementioned registers, or anything else added after PITCHLOCK and
+    * before 3D_CAPS, needs to reason about something other than
+    * SVGA_FIFO_MIN.
+    */
+
+   /*
+    * 3D caps block space; valid with 3D hardware version >=
+    * SVGA3D_HWVERSION_WS6_B1.
+    */
+   SVGA_FIFO_3D_CAPS      = 32,
+   SVGA_FIFO_3D_CAPS_LAST = 32 + 255,
+
+   /*
+    * End of VMX's current definition of "extended-FIFO registers".
+    * Registers before here are always enabled/disabled as a block; either
+    * the extended FIFO is enabled and includes all preceding registers, or
+    * it's disabled entirely.
+    *
+    * Block 3b (truly optional extended registers): Additional registers for
+    * the extended FIFO, which the VMX already knows how to enable and
+    * disable with correct granularity.
+    *
+    * Registers after here exist if and only if the guest SVGA driver
+    * sets SVGA_FIFO_MIN high enough to leave room for them.
+    */
+
+   // Valid if register exists:
+   SVGA_FIFO_GUEST_3D_HWVERSION, /* Guest driver's 3D version */
+   SVGA_FIFO_FENCE_GOAL,         /* Matching target for SVGA_IRQFLAG_FENCE_GOAL */
+   SVGA_FIFO_BUSY,               /* See "FIFO Synchronization Registers" */
+
+   /*
+    * Always keep this last.  This defines the maximum number of
+    * registers we know about.  At power-on, this value is placed in
+    * the SVGA_REG_MEM_REGS register, and we expect the guest driver
+    * to allocate this much space in FIFO memory for registers.
+    */
+    SVGA_FIFO_NUM_REGS
+};
+
+
+/*
+ * Definition of registers included in extended FIFO support.
+ *
+ * The guest SVGA driver gets to allocate the FIFO between registers
+ * and data.  It must always allocate at least 4 registers, but old
+ * drivers stopped there.
+ *
+ * The VMX will enable extended FIFO support if and only if the guest
+ * left enough room for all registers defined as part of the mandatory
+ * set for the extended FIFO.
+ *
+ * Note that the guest drivers typically allocate the FIFO only at
+ * initialization time, not at mode switches, so it's likely that the
+ * number of FIFO registers won't change without a reboot.
+ *
+ * All registers less than this value are guaranteed to be present if
+ * svgaUser->fifo.extended is set. Any later registers must be tested
+ * individually for compatibility at each use (in the VMX).
+ *
+ * This value is used only by the VMX, so it can change without
+ * affecting driver compatibility; keep it that way?
+ */
+#define SVGA_FIFO_EXTENDED_MANDATORY_REGS  (SVGA_FIFO_3D_CAPS_LAST + 1)
+
+
+/*
+ * FIFO Synchronization Registers
+ *
+ *  This explains the relationship between the various FIFO
+ *  sync-related registers in IOSpace and in FIFO space.
+ *
+ *  SVGA_REG_SYNC --
+ *
+ *       The SYNC register can be used in two different ways by the guest:
+ *
+ *         1. If the guest wishes to fully sync (drain) the FIFO,
+ *            it will write once to SYNC then poll on the BUSY
+ *            register. The FIFO is sync'ed once BUSY is zero.
+ *
+ *         2. If the guest wants to asynchronously wake up the host,
+ *            it will write once to SYNC without polling on BUSY.
+ *            Ideally it will do this after some new commands have
+ *            been placed in the FIFO, and after reading a zero
+ *            from SVGA_FIFO_BUSY.
+ *
+ *       (1) is the original behaviour that SYNC was designed to
+ *       support.  Originally, a write to SYNC would implicitly
+ *       trigger a read from BUSY. This causes us to synchronously
+ *       process the FIFO.
+ *
+ *       This behaviour has since been changed so that writing SYNC
+ *       will *not* implicitly cause a read from BUSY. Instead, it
+ *       makes a channel call which asynchronously wakes up the MKS
+ *       thread.
+ *
+ *       New guests can use this new behaviour to implement (2)
+ *       efficiently. This lets guests get the host's attention
+ *       without waiting for the MKS to poll, which gives us much
+ *       better CPU utilization on SMP hosts and on UP hosts while
+ *       we're blocked on the host GPU.
+ *
+ *       Old guests shouldn't notice the behaviour change. SYNC was
+ *       never guaranteed to process the entire FIFO, since it was
+ *       bounded to a particular number of CPU cycles. Old guests will
+ *       still loop on the BUSY register until the FIFO is empty.
+ *
+ *       Writing to SYNC currently has the following side-effects:
+ *
+ *         - Sets SVGA_REG_BUSY to TRUE (in the monitor)
+ *         - Asynchronously wakes up the MKS thread for FIFO processing
+ *         - The value written to SYNC is recorded as a "reason", for
+ *           stats purposes.
+ *
+ *       If SVGA_FIFO_BUSY is available, drivers are advised to only
+ *       write to SYNC if SVGA_FIFO_BUSY is FALSE. Drivers should set
+ *       SVGA_FIFO_BUSY to TRUE after writing to SYNC. The MKS will
+ *       eventually set SVGA_FIFO_BUSY on its own, but this approach
+ *       lets the driver avoid sending multiple asynchronous wakeup
+ *       messages to the MKS thread.
+ *
+ *  SVGA_REG_BUSY --
+ *
+ *       This register is set to TRUE when SVGA_REG_SYNC is written,
+ *       and it reads as FALSE when the FIFO has been completely
+ *       drained.
+ *
+ *       Every read from this register causes us to synchronously
+ *       process FIFO commands. There is no guarantee as to how many
+ *       commands each read will process.
+ *
+ *       CPU time spent processing FIFO commands will be billed to
+ *       the guest.
+ *
+ *       New drivers should avoid using this register unless they
+ *       need to guarantee that the FIFO is completely drained. It
+ *       is overkill for performing a sync-to-fence. Older drivers
+ *       will use this register for any type of synchronization.
+ *
+ *  SVGA_FIFO_BUSY --
+ *
+ *       This register is a fast way for the guest driver to check
+ *       whether the FIFO is already being processed. It reads and
+ *       writes at normal RAM speeds, with no monitor intervention.
+ *
+ *       If this register reads as TRUE, the host is guaranteeing that
+ *       any new commands written into the FIFO will be noticed before
+ *       the MKS goes back to sleep.
+ *
+ *       If this register reads as FALSE, no such guarantee can be
+ *       made.
+ *
+ *       The guest should use this register to quickly determine
+ *       whether or not it needs to wake up the host. If the guest
+ *       just wrote a command or group of commands that it would like
+ *       the host to begin processing, it should:
+ *
+ *         1. Read SVGA_FIFO_BUSY. If it reads as TRUE, no further
+ *            action is necessary.
+ *
+ *         2. Write TRUE to SVGA_FIFO_BUSY. This informs future guest
+ *            code that we've already sent a SYNC to the host and we
+ *            don't need to send a duplicate.
+ *
+ *         3. Write a reason to SVGA_REG_SYNC. This will send an
+ *            asynchronous wakeup to the MKS thread.
+ */
+
+
+/*
+ * FIFO Capabilities
+ *
+ *      Fence -- Fence register and command are supported
+ *      Accel Front -- Front buffer only commands are supported
+ *      Pitch Lock -- Pitch lock register is supported
+ *      Video -- SVGA Video overlay units are supported
+ *      Escape -- Escape command is supported
+ *
+ * XXX: Add longer descriptions for each capability, including a list
+ *      of the new features that each capability provides.
+ *
+ * SVGA_FIFO_CAP_SCREEN_OBJECT --
+ *
+ *    Provides dynamic multi-screen rendering, for improved Unity and
+ *    multi-monitor modes. With Screen Object, the guest can
+ *    dynamically create and destroy 'screens', which can represent
+ *    Unity windows or virtual monitors. Screen Object also provides
+ *    strong guarantees that DMA operations happen only when
+ *    guest-initiated. Screen Object deprecates the BAR1 guest
+ *    framebuffer (GFB) and all commands that work only with the GFB.
+ *
+ *    New registers:
+ *       FIFO_CURSOR_SCREEN_ID, VIDEO_DATA_GMRID, VIDEO_DST_SCREEN_ID
+ *
+ *    New 2D commands:
+ *       DEFINE_SCREEN, DESTROY_SCREEN, DEFINE_GMRFB, BLIT_GMRFB_TO_SCREEN,
+ *       BLIT_SCREEN_TO_GMRFB, ANNOTATION_FILL, ANNOTATION_COPY
+ *
+ *    New 3D commands:
+ *       BLIT_SURFACE_TO_SCREEN
+ *
+ *    New guarantees:
+ *
+ *       - The host will not read or write guest memory, including the GFB,
+ *         except when explicitly initiated by a DMA command.
+ *
+ *       - All DMA, including legacy DMA like UPDATE and PRESENT_READBACK,
+ *         is guaranteed to complete before any subsequent FENCEs.
+ *
+ *       - All legacy commands which affect a Screen (UPDATE, PRESENT,
+ *         PRESENT_READBACK) as well as new Screen blit commands will
+ *         all behave consistently as blits, and memory will be read
+ *         or written in FIFO order.
+ *
+ *         For example, if you PRESENT from one SVGA3D surface to multiple
+ *         places on the screen, the data copied will always be from the
+ *         SVGA3D surface at the time the PRESENT was issued in the FIFO.
+ *         This was not necessarily true on devices without Screen Object.
+ *
+ *         This means that on devices that support Screen Object, the
+ *         PRESENT_READBACK command should not be necessary unless you
+ *         actually want to read back the results of 3D rendering into
+ *         system memory. (And for that, the BLIT_SCREEN_TO_GMRFB
+ *         command provides a strict superset of functionality.)
+ *
+ *       - When a screen is resized, either using Screen Object commands or
+ *         legacy multimon registers, its contents are preserved.
+ */
+
+#define SVGA_FIFO_CAP_NONE                  0
+#define SVGA_FIFO_CAP_FENCE             (1<<0)
+#define SVGA_FIFO_CAP_ACCELFRONT        (1<<1)
+#define SVGA_FIFO_CAP_PITCHLOCK         (1<<2)
+#define SVGA_FIFO_CAP_VIDEO             (1<<3)
+#define SVGA_FIFO_CAP_CURSOR_BYPASS_3   (1<<4)
+#define SVGA_FIFO_CAP_ESCAPE            (1<<5)
+#define SVGA_FIFO_CAP_RESERVE           (1<<6)
+#define SVGA_FIFO_CAP_SCREEN_OBJECT     (1<<7)
+
+
+/*
+ * FIFO Flags
+ *
+ *      Accel Front -- Driver should use front buffer only commands
+ */
+
+#define SVGA_FIFO_FLAG_NONE                 0
+#define SVGA_FIFO_FLAG_ACCELFRONT       (1<<0)
+#define SVGA_FIFO_FLAG_RESERVED        (1<<31) // Internal use only
+
+/*
+ * FIFO reservation sentinel value
+ */
+
+#define SVGA_FIFO_RESERVED_UNKNOWN      0xffffffff
+
+
+/*
+ * Video overlay support
+ */
+
+#define SVGA_NUM_OVERLAY_UNITS 32
+
+
+/*
+ * Video capabilities that the guest is currently using
+ */
+
+#define SVGA_VIDEO_FLAG_COLORKEY        0x0001
+
+
+/*
+ * Offsets for the video overlay registers
+ */
+
+enum {
+   SVGA_VIDEO_ENABLED = 0,
+   SVGA_VIDEO_FLAGS,
+   SVGA_VIDEO_DATA_OFFSET,
+   SVGA_VIDEO_FORMAT,
+   SVGA_VIDEO_COLORKEY,
+   SVGA_VIDEO_SIZE,          // Deprecated
+   SVGA_VIDEO_WIDTH,
+   SVGA_VIDEO_HEIGHT,
+   SVGA_VIDEO_SRC_X,
+   SVGA_VIDEO_SRC_Y,
+   SVGA_VIDEO_SRC_WIDTH,
+   SVGA_VIDEO_SRC_HEIGHT,
+   SVGA_VIDEO_DST_X,         // Signed int32
+   SVGA_VIDEO_DST_Y,         // Signed int32
+   SVGA_VIDEO_DST_WIDTH,
+   SVGA_VIDEO_DST_HEIGHT,
+   SVGA_VIDEO_PITCH_1,
+   SVGA_VIDEO_PITCH_2,
+   SVGA_VIDEO_PITCH_3,
+   SVGA_VIDEO_DATA_GMRID,    // Optional, defaults to SVGA_GMR_FRAMEBUFFER
+   SVGA_VIDEO_DST_SCREEN_ID, // Optional, defaults to virtual coords (SVGA_ID_INVALID)
+   SVGA_VIDEO_NUM_REGS
+};
+
+
+/*
+ * SVGA Overlay Units
+ *
+ *      width and height relate to the entire source video frame.
+ *      srcX, srcY, srcWidth and srcHeight represent subset of the source
+ *      video frame to be displayed.
+ */
+
+typedef struct SVGAOverlayUnit {
+   uint32 enabled;
+   uint32 flags;
+   uint32 dataOffset;
+   uint32 format;
+   uint32 colorKey;
+   uint32 size;
+   uint32 width;
+   uint32 height;
+   uint32 srcX;
+   uint32 srcY;
+   uint32 srcWidth;
+   uint32 srcHeight;
+   int32  dstX;
+   int32  dstY;
+   uint32 dstWidth;
+   uint32 dstHeight;
+   uint32 pitches[3];
+   uint32 dataGMRId;
+   uint32 dstScreenId;
+} SVGAOverlayUnit;
+
+
+/*
+ * SVGAScreenObject --
+ *
+ *    This is a new way to represent a guest's multi-monitor screen or
+ *    Unity window. Screen objects are only supported if the
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT capability bit is set.
+ *
+ *    If Screen Objects are supported, they can be used to fully
+ *    replace the functionality provided by the framebuffer registers
+ *    (SVGA_REG_WIDTH, HEIGHT, etc.) and by SVGA_CAP_DISPLAY_TOPOLOGY.
+ *
+ *    The screen object is a struct with guaranteed binary
+ *    compatibility. New flags can be added, and the struct may grow,
+ *    but existing fields must retain their meaning.
+ *
+ */
+
+#define SVGA_SCREEN_HAS_ROOT    (1 << 0)  // Screen is present in the virtual coord space
+#define SVGA_SCREEN_IS_PRIMARY  (1 << 1)  // Guest considers this screen to be 'primary'
+#define SVGA_SCREEN_FULLSCREEN_HINT (1 << 2)   // Guest is running a fullscreen app here
+
+typedef
+struct SVGAScreenObject {
+   uint32 structSize;   // sizeof(SVGAScreenObject)
+   uint32 id;
+   uint32 flags;
+   struct {
+      uint32 width;
+      uint32 height;
+   } size;
+   struct {
+      int32 x;
+      int32 y;
+   } root;              // Only used if SVGA_SCREEN_HAS_ROOT is set.
+} SVGAScreenObject;
+
+
+/*
+ *  Commands in the command FIFO:
+ *
+ *  Command IDs defined below are used for the traditional 2D FIFO
+ *  communication (not all commands are available for all versions of the
+ *  SVGA FIFO protocol).
+ *
+ *  Note the holes in the command ID numbers: These commands have been
+ *  deprecated, and the old IDs must not be reused.
+ *
+ *  Command IDs from 1000 to 1999 are reserved for use by the SVGA3D
+ *  protocol.
+ *
+ *  Each command's parameters are described by the comments and
+ *  structs below.
+ */
+
+typedef enum {
+   SVGA_CMD_INVALID_CMD           = 0,
+   SVGA_CMD_UPDATE                = 1,
+   SVGA_CMD_RECT_COPY             = 3,
+   SVGA_CMD_DEFINE_CURSOR         = 19,
+   SVGA_CMD_DEFINE_ALPHA_CURSOR   = 22,
+   SVGA_CMD_UPDATE_VERBOSE        = 25,
+   SVGA_CMD_FRONT_ROP_FILL        = 29,
+   SVGA_CMD_FENCE                 = 30,
+   SVGA_CMD_ESCAPE                = 33,
+   SVGA_CMD_DEFINE_SCREEN         = 34,
+   SVGA_CMD_DESTROY_SCREEN        = 35,
+   SVGA_CMD_DEFINE_GMRFB          = 36,
+   SVGA_CMD_BLIT_GMRFB_TO_SCREEN  = 37,
+   SVGA_CMD_BLIT_SCREEN_TO_GMRFB  = 38,
+   SVGA_CMD_ANNOTATION_FILL       = 39,
+   SVGA_CMD_ANNOTATION_COPY       = 40,
+   SVGA_CMD_MAX
+} SVGAFifoCmdId;
+
+#define SVGA_CMD_MAX_ARGS           64
+
+
+/*
+ * SVGA_CMD_UPDATE --
+ *
+ *    This is a DMA transfer which copies from the Guest Framebuffer
+ *    (GFB) at BAR1 + SVGA_REG_FB_OFFSET to any screens which
+ *    intersect with the provided virtual rectangle.
+ *
+ *    This command does not support using arbitrary guest memory as a
+ *    data source- it only works with the pre-defined GFB memory.
+ *    This command also does not support signed virtual coordinates.
+ *    If you have defined screens (using SVGA_CMD_DEFINE_SCREEN) with
+ *    negative root x/y coordinates, the negative portion of those
+ *    screens will not be reachable by this command.
+ *
+ *    This command is not necessary when using framebuffer
+ *    traces. Traces are automatically enabled if the SVGA FIFO is
+ *    disabled, and you may explicitly enable/disable traces using
+ *    SVGA_REG_TRACES. With traces enabled, any write to the GFB will
+ *    automatically act as if a subsequent SVGA_CMD_UPDATE was issued.
+ *
+ *    Traces and SVGA_CMD_UPDATE are the only supported ways to render
+ *    pseudocolor screen updates. The newer Screen Object commands
+ *    only support true color formats.
+ *
+ * Availability:
+ *    Always available.
+ */
+
+typedef
+struct {
+   uint32 x;
+   uint32 y;
+   uint32 width;
+   uint32 height;
+} SVGAFifoCmdUpdate;
+
+
+/*
+ * SVGA_CMD_RECT_COPY --
+ *
+ *    Perform a rectangular DMA transfer from one area of the GFB to
+ *    another, and copy the result to any screens which intersect it.
+ *
+ * Availability:
+ *    SVGA_CAP_RECT_COPY
+ */
+
+typedef
+struct {
+   uint32 srcX;
+   uint32 srcY;
+   uint32 destX;
+   uint32 destY;
+   uint32 width;
+   uint32 height;
+} SVGAFifoCmdRectCopy;
+
+
+/*
+ * SVGA_CMD_DEFINE_CURSOR --
+ *
+ *    Provide a new cursor image, as an AND/XOR mask.
+ *
+ *    The recommended way to position the cursor overlay is by using
+ *    the SVGA_FIFO_CURSOR_* registers, supported by the
+ *    SVGA_FIFO_CAP_CURSOR_BYPASS_3 capability.
+ *
+ * Availability:
+ *    SVGA_CAP_CURSOR
+ */
+
+typedef
+struct {
+   uint32 id;             // Reserved, must be zero.
+   uint32 hotspotX;
+   uint32 hotspotY;
+   uint32 width;
+   uint32 height;
+   uint32 andMaskDepth;   // Value must be 1 or equal to BITS_PER_PIXEL
+   uint32 xorMaskDepth;   // Value must be 1 or equal to BITS_PER_PIXEL
+   /*
+    * Followed by scanline data for AND mask, then XOR mask.
+    * Each scanline is padded to a 32-bit boundary.
+   */
+} SVGAFifoCmdDefineCursor;
+
+
+/*
+ * SVGA_CMD_DEFINE_ALPHA_CURSOR --
+ *
+ *    Provide a new cursor image, in 32-bit BGRA format.
+ *
+ *    The recommended way to position the cursor overlay is by using
+ *    the SVGA_FIFO_CURSOR_* registers, supported by the
+ *    SVGA_FIFO_CAP_CURSOR_BYPASS_3 capability.
+ *
+ * Availability:
+ *    SVGA_CAP_ALPHA_CURSOR
+ */
+
+typedef
+struct {
+   uint32 id;             // Reserved, must be zero.
+   uint32 hotspotX;
+   uint32 hotspotY;
+   uint32 width;
+   uint32 height;
+   /* Followed by scanline data */
+} SVGAFifoCmdDefineAlphaCursor;
+
+
+/*
+ * SVGA_CMD_UPDATE_VERBOSE --
+ *
+ *    Just like SVGA_CMD_UPDATE, but also provide a per-rectangle
+ *    'reason' value, an opaque cookie which is used by internal
+ *    debugging tools. Third party drivers should not use this
+ *    command.
+ *
+ * Availability:
+ *    SVGA_CAP_EXTENDED_FIFO
+ */
+
+typedef
+struct {
+   uint32 x;
+   uint32 y;
+   uint32 width;
+   uint32 height;
+   uint32 reason;
+} SVGAFifoCmdUpdateVerbose;
+
+
+/*
+ * SVGA_CMD_FRONT_ROP_FILL --
+ *
+ *    This is a hint which tells the SVGA device that the driver has
+ *    just filled a rectangular region of the GFB with a solid
+ *    color. Instead of reading these pixels from the GFB, the device
+ *    can assume that they all equal 'color'. This is primarily used
+ *    for remote desktop protocols.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_ACCELFRONT
+ */
+
+#define  SVGA_ROP_COPY                    0x03
+
+typedef
+struct {
+   uint32 color;     // In the same format as the GFB
+   uint32 x;
+   uint32 y;
+   uint32 width;
+   uint32 height;
+   uint32 rop;       // Must be SVGA_ROP_COPY
+} SVGAFifoCmdFrontRopFill;
+
+
+/*
+ * SVGA_CMD_FENCE --
+ *
+ *    Insert a synchronization fence.  When the SVGA device reaches
+ *    this command, it will copy the 'fence' value into the
+ *    SVGA_FIFO_FENCE register. It will also compare the fence against
+ *    SVGA_FIFO_FENCE_GOAL. If the fence matches the goal and the
+ *    SVGA_IRQFLAG_FENCE_GOAL interrupt is enabled, the device will
+ *    raise this interrupt.
+ *
+ * Availability:
+ *    SVGA_FIFO_FENCE for this command,
+ *    SVGA_CAP_IRQMASK for SVGA_FIFO_FENCE_GOAL.
+ */
+
+typedef
+struct {
+   uint32 fence;
+} SVGAFifoCmdFence;
+
+
+/*
+ * SVGA_CMD_ESCAPE --
+ *
+ *    Send an extended or vendor-specific variable length command.
+ *    This is used for video overlay, third party plugins, and
+ *    internal debugging tools. See svga_escape.h
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_ESCAPE
+ */
+
+typedef
+struct {
+   uint32 nsid;
+   uint32 size;
+   /* followed by 'size' bytes of data */
+} SVGAFifoCmdEscape;
+
+
+/*
+ * SVGA_CMD_DEFINE_SCREEN --
+ *
+ *    Define or redefine an SVGAScreenObject. See the description of
+ *    SVGAScreenObject above.  The video driver is responsible for
+ *    generating new screen IDs. They should be small positive
+ *    integers. The virtual device will have an implementation
+ *    specific upper limit on the number of screen IDs
+ *    supported. Drivers are responsible for recycling IDs. The first
+ *    valid ID is zero.
+ *
+ *    - Interaction with other registers:
+ *
+ *    For backwards compatibility, when the GFB mode registers (WIDTH,
+ *    HEIGHT, PITCHLOCK, BITS_PER_PIXEL) are modified, the SVGA device
+ *    deletes all screens other than screen #0, and redefines screen
+ *    #0 according to the specified mode. Drivers that use
+ *    SVGA_CMD_DEFINE_SCREEN should destroy or redefine screen #0.
+ *
+ *    If you use screen objects, do not use the legacy multi-mon
+ *    registers (SVGA_REG_NUM_GUEST_DISPLAYS, SVGA_REG_DISPLAY_*).
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGAScreenObject screen;   // Variable-length according to version
+} SVGAFifoCmdDefineScreen;
+
+
+/*
+ * SVGA_CMD_DESTROY_SCREEN --
+ *
+ *    Destroy an SVGAScreenObject. Its ID is immediately available for
+ *    re-use.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   uint32 screenId;
+} SVGAFifoCmdDestroyScreen;
+
+
+/*
+ * SVGA_CMD_DEFINE_GMRFB --
+ *
+ *    This command sets a piece of SVGA device state called the
+ *    Guest Memory Region Framebuffer, or GMRFB. The GMRFB is a
+ *    piece of light-weight state which identifies the location and
+ *    format of an image in guest memory or in BAR1. The GMRFB has
+ *    an arbitrary size, and it doesn't need to match the geometry
+ *    of the GFB or any screen object.
+ *
+ *    The GMRFB can be redefined as often as you like. You could
+ *    always use the same GMRFB, you could redefine it before
+ *    rendering from a different guest screen, or you could even
+ *    redefine it before every blit.
+ *
+ *    There are multiple ways to use this command. The simplest way is
+ *    to use it to move the framebuffer either to elsewhere in the GFB
+ *    (BAR1) memory region, or to a user-defined GMR. This lets a
+ *    driver use a framebuffer allocated entirely out of normal system
+ *    memory, which we encourage.
+ *
+ *    Another way to use this command is to set up a ring buffer of
+ *    updates in GFB memory. If a driver wants to ensure that no
+ *    frames are skipped by the SVGA device, it is important that the
+ *    driver not modify the source data for a blit until the device is
+ *    done processing the command. One efficient way to accomplish
+ *    this is to use a ring of small DMA buffers. Each buffer is used
+ *    for one blit, then we move on to the next buffer in the
+ *    ring. The FENCE mechanism is used to protect each buffer from
+ *    re-use until the device is finished with that buffer's
+ *    corresponding blit.
+ *
+ *    This command does not affect the meaning of SVGA_CMD_UPDATE.
+ *    UPDATEs always occur from the legacy GFB memory area. This
+ *    command has no support for pseudocolor GMRFBs. Currently only
+ *    true-color 15, 16, and 24-bit depths are supported. Future
+ *    devices may expose capabilities for additional framebuffer
+ *    formats.
+ *
+ *    The default GMRFB value is undefined. Drivers must always send
+ *    this command at least once before performing any blit from the
+ *    GMRFB.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGAGuestPtr        ptr;
+   uint32              bytesPerLine;
+   SVGAGMRImageFormat  format;
+} SVGAFifoCmdDefineGMRFB;
+
+
+/*
+ * SVGA_CMD_BLIT_GMRFB_TO_SCREEN --
+ *
+ *    This is a guest-to-host blit. It performs a DMA operation to
+ *    copy a rectangular region of pixels from the current GMRFB to
+ *    one or more Screen Objects.
+ *
+ *    The destination coordinate may be specified relative to a
+ *    screen's origin (if a screen ID is specified) or relative to the
+ *    virtual coordinate system's origin (if the screen ID is
+ *    SVGA_ID_INVALID). The actual destination may span zero or more
+ *    screens, in the case of a virtual destination rect or a rect
+ *    which extends off the edge of the specified screen.
+ *
+ *    This command writes to the screen's "base layer": the underlying
+ *    framebuffer which exists below any cursor or video overlays. No
+ *    action is necessary to explicitly hide or update any overlays
+ *    which exist on top of the updated region.
+ *
+ *    The SVGA device is guaranteed to finish reading from the GMRFB
+ *    by the time any subsequent FENCE commands are reached.
+ *
+ *    This command consumes an annotation. See the
+ *    SVGA_CMD_ANNOTATION_* commands for details.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGASignedPoint  srcOrigin;
+   SVGASignedRect   destRect;
+   uint32           destScreenId;
+} SVGAFifoCmdBlitGMRFBToScreen;
+
+
+/*
+ * SVGA_CMD_BLIT_SCREEN_TO_GMRFB --
+ *
+ *    This is a host-to-guest blit. It performs a DMA operation to
+ *    copy a rectangular region of pixels from a single Screen Object
+ *    back to the current GMRFB.
+ *
+ *    Usage note: This command should be used rarely. It will
+ *    typically be inefficient, but it is necessary for some types of
+ *    synchronization between 3D (GPU) and 2D (CPU) rendering into
+ *    overlapping areas of a screen.
+ *
+ *    The source coordinate is specified relative to a screen's
+ *    origin. The provided screen ID must be valid. If any parameters
+ *    are invalid, the resulting pixel values are undefined.
+ *
+ *    This command reads the screen's "base layer". Overlays like
+ *    video and cursor are not included, but any data which was sent
+ *    using a blit-to-screen primitive will be available, no matter
+ *    whether the data's original source was the GMRFB or the 3D
+ *    acceleration hardware.
+ *
+ *    Note that our guest-to-host blits and host-to-guest blits aren't
+ *    symmetric in their current implementation. While the parameters
+ *    are identical, host-to-guest blits are a lot less featureful.
+ *    They do not support clipping: If the source parameters don't
+ *    fully fit within a screen, the blit fails. They must originate
+ *    from exactly one screen. Virtual coordinates are not directly
+ *    supported.
+ *
+ *    Host-to-guest blits do support the same set of GMRFB formats
+ *    offered by guest-to-host blits.
+ *
+ *    The SVGA device is guaranteed to finish writing to the GMRFB by
+ *    the time any subsequent FENCE commands are reached.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGASignedPoint  destOrigin;
+   SVGASignedRect   srcRect;
+   uint32           srcScreenId;
+} SVGAFifoCmdBlitScreenToGMRFB;
+
+
+/*
+ * SVGA_CMD_ANNOTATION_FILL --
+ *
+ *    This is a blit annotation. This command stores a small piece of
+ *    device state which is consumed by the next blit-to-screen
+ *    command. The state is only cleared by commands which are
+ *    specifically documented as consuming an annotation. Other
+ *    commands (such as ESCAPEs for debugging) may intervene between
+ *    the annotation and its associated blit.
+ *
+ *    This annotation is a promise about the contents of the next
+ *    blit: The video driver is guaranteeing that all pixels in that
+ *    blit will have the same value, specified here as a color in
+ *    SVGAColorBGRX format.
+ *
+ *    The SVGA device can still render the blit correctly even if it
+ *    ignores this annotation, but the annotation may allow it to
+ *    perform the blit more efficiently, for example by ignoring the
+ *    source data and performing a fill in hardware.
+ *
+ *    This annotation is most important for performance when the
+ *    user's display is being remoted over a network connection.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGAColorBGRX  color;
+} SVGAFifoCmdAnnotationFill;
+
+
+/*
+ * SVGA_CMD_ANNOTATION_COPY --
+ *
+ *    This is a blit annotation. See SVGA_CMD_ANNOTATION_FILL for more
+ *    information about annotations.
+ *
+ *    This annotation is a promise about the contents of the next
+ *    blit: The video driver is guaranteeing that all pixels in that
+ *    blit will have the same value as those which already exist at an
+ *    identically-sized region on the same or a different screen.
+ *
+ *    Note that the source pixels for the COPY in this annotation are
+ *    sampled before applying the anqnotation's associated blit. They
+ *    are allowed to overlap with the blit's destination pixels.
+ *
+ *    The copy source rectangle is specified the same way as the blit
+ *    destination: it can be a rectangle which spans zero or more
+ *    screens, specified relative to either a screen or to the virtual
+ *    coordinate system's origin. If the source rectangle includes
+ *    pixels which are not from exactly one screen, the results are
+ *    undefined.
+ *
+ * Availability:
+ *    SVGA_FIFO_CAP_SCREEN_OBJECT
+ */
+
+typedef
+struct {
+   SVGASignedPoint  srcOrigin;
+   uint32           srcScreenId;
+} SVGAFifoCmdAnnotationCopy;
+
+#endif
diff --git a/src/gallium/drivers/svga/include/svga_types.h b/src/gallium/drivers/svga/include/svga_types.h
new file mode 100644
index 0000000000..7fd9bab03a
--- /dev/null
+++ b/src/gallium/drivers/svga/include/svga_types.h
@@ -0,0 +1,46 @@
+/**********************************************************
+ * Copyright 1998-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef _SVGA_TYPES_H_
+#define _SVGA_TYPES_H_
+
+#include "pipe/p_compiler.h"
+
+typedef int64_t int64;
+typedef uint64_t uint64;
+
+typedef int32_t int32;
+typedef uint32_t uint32;
+
+typedef int16_t int16;
+typedef uint16_t uint16;
+
+typedef int8_t int8;
+typedef uint8_t uint8;
+
+typedef uint8_t Bool;
+
+#endif /* _SVGA_TYPES_H_ */
+
diff --git a/src/gallium/drivers/svga/svga_cmd.c b/src/gallium/drivers/svga/svga_cmd.c
new file mode 100644
index 0000000000..a0da7d7e5d
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_cmd.c
@@ -0,0 +1,1427 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * svga_cmd.c --
+ *
+ *      Command construction utility for the SVGA3D protocol used by
+ *      the VMware SVGA device, based on the svgautil library.
+ */
+
+#include "svga_winsys.h"
+#include "svga_screen_buffer.h"
+#include "svga_screen_texture.h"
+#include "svga_cmd.h"
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * surface_to_surfaceid --
+ *
+ *      Utility function for surface ids.
+ *      Can handle null surface. Does a surface_reallocation so you need
+ *      to have allocated the fifo space before converting.
+ *
+ * Results:
+ *      id is filld out.
+ *
+ * Side effects:
+ *      One surface relocation is preformed for texture handle.
+ *
+ *----------------------------------------------------------------------
+ */
+
+static INLINE
+void surface_to_surfaceid(struct svga_winsys_context *swc, // IN
+                          struct pipe_surface *surface,    // IN
+                          SVGA3dSurfaceImageId *id,        // OUT
+                          unsigned flags)                  // IN
+{
+   if(surface) {
+      struct svga_surface *s = svga_surface(surface);
+      swc->surface_relocation(swc, &id->sid, s->handle, flags);
+      id->face = s->real_face; /* faces have the same order */
+      id->mipmap = s->real_level;
+   }
+   else {
+      id->sid = SVGA3D_INVALID_ID;
+      id->face = 0;
+      id->mipmap = 0;
+   }
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_FIFOReserve --
+ *
+ *      Reserve space for an SVGA3D FIFO command.
+ *
+ *      The 2D SVGA commands have been around for a while, so they
+ *      have a rather asymmetric structure. The SVGA3D protocol is
+ *      more uniform: each command begins with a header containing the
+ *      command number and the full size.
+ *
+ *      This is a convenience wrapper around SVGA_FIFOReserve. We
+ *      reserve space for the whole command, and write the header.
+ *
+ *      This function must be paired with SVGA_FIFOCommitAll().
+ *
+ * Results:
+ *      Returns a pointer to the space reserved for command-specific
+ *      data. It must be 'cmdSize' bytes long.
+ *
+ * Side effects:
+ *      Begins a FIFO reservation.
+ *
+ *----------------------------------------------------------------------
+ */
+
+void *
+SVGA3D_FIFOReserve(struct svga_winsys_context *swc,
+                   uint32 cmd,       // IN
+                   uint32 cmdSize,   // IN
+                   uint32 nr_relocs) // IN
+{
+   SVGA3dCmdHeader *header;
+
+   header = swc->reserve(swc, sizeof *header + cmdSize, nr_relocs);
+   if(!header)
+      return NULL;
+
+   header->id = cmd;
+   header->size = cmdSize;
+
+   return &header[1];
+}
+
+
+void
+SVGA_FIFOCommitAll(struct svga_winsys_context *swc)
+{
+   swc->commit(swc);
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DefineContext --
+ *
+ *      Create a new context, to be referred to with the provided ID.
+ *
+ *      Context objects encapsulate all render state, and shader
+ *      objects are per-context.
+ *
+ *      Surfaces are not per-context. The same surface can be shared
+ *      between multiple contexts, and surface operations can occur
+ *      without a context.
+ *
+ *      If the provided context ID already existed, it is redefined.
+ *
+ *      Context IDs are arbitrary small non-negative integers,
+ *      global to the entire SVGA device.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DefineContext(struct svga_winsys_context *swc)  // IN
+{
+   SVGA3dCmdDefineContext *cmd;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_CONTEXT_DEFINE, sizeof *cmd, 0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+
+   swc->commit(swc);
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DestroyContext --
+ *
+ *      Delete a context created with SVGA3D_DefineContext.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DestroyContext(struct svga_winsys_context *swc)  // IN
+{
+   SVGA3dCmdDestroyContext *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_CONTEXT_DESTROY, sizeof *cmd, 0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+   
+   cmd->cid = swc->cid;
+   
+   swc->commit(swc);
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginDefineSurface --
+ *
+ *      Begin a SURFACE_DEFINE command. This reserves space for it in
+ *      the FIFO, and returns pointers to the command's faces and
+ *      mipsizes arrays.
+ *
+ *      This function must be paired with SVGA_FIFOCommitAll().
+ *      The faces and mipSizes arrays are initialized to zero.
+ *
+ *      This creates a "surface" object in the SVGA3D device,
+ *      with the provided surface ID (sid). Surfaces are generic
+ *      containers for host VRAM objects like textures, vertex
+ *      buffers, and depth/stencil buffers.
+ *
+ *      Surfaces are hierarchial:
+ *
+ *        - Surface may have multiple faces (for cube maps)
+ *
+ *          - Each face has a list of mipmap levels
+ *
+ *             - Each mipmap image may have multiple volume
+ *               slices, if the image is three dimensional.
+ *
+ *                - Each slice is a 2D array of 'blocks'
+ *
+ *                   - Each block may be one or more pixels.
+ *                     (Usually 1, more for DXT or YUV formats.)
+ *
+ *      Surfaces are generic host VRAM objects. The SVGA3D device
+ *      may optimize surfaces according to the format they were
+ *      created with, but this format does not limit the ways in
+ *      which the surface may be used. For example, a depth surface
+ *      can be used as a texture, or a floating point image may
+ *      be used as a vertex buffer. Some surface usages may be
+ *      lower performance, due to software emulation, but any
+ *      usage should work with any surface.
+ *
+ *      If 'sid' is already defined, the old surface is deleted
+ *      and this new surface replaces it.
+ *
+ *      Surface IDs are arbitrary small non-negative integers,
+ *      global to the entire SVGA device.
+ *
+ * Results:
+ *      Returns pointers to arrays allocated in the FIFO for 'faces'
+ *      and 'mipSizes'.
+ *
+ * Side effects:
+ *      Begins a FIFO reservation.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginDefineSurface(struct svga_winsys_context *swc,
+                          struct svga_winsys_surface *sid, // IN
+                          SVGA3dSurfaceFlags flags,    // IN
+                          SVGA3dSurfaceFormat format,  // IN
+                          SVGA3dSurfaceFace **faces,   // OUT
+                          SVGA3dSize **mipSizes,       // OUT
+                          uint32 numMipSizes)          // IN
+{
+   SVGA3dCmdDefineSurface *cmd;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_DEFINE, sizeof *cmd +
+                            sizeof **mipSizes * numMipSizes, 1);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   swc->surface_relocation(swc, &cmd->sid, sid, PIPE_BUFFER_USAGE_GPU_WRITE);
+   cmd->surfaceFlags = flags;
+   cmd->format = format;
+
+   *faces = &cmd->face[0];
+   *mipSizes = (SVGA3dSize*) &cmd[1];
+
+   memset(*faces, 0, sizeof **faces * SVGA3D_MAX_SURFACE_FACES);
+   memset(*mipSizes, 0, sizeof **mipSizes * numMipSizes);
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DefineSurface2D --
+ *
+ *      This is a simplified version of SVGA3D_BeginDefineSurface(),
+ *      which does not support cube maps, mipmaps, or volume textures.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DefineSurface2D(struct svga_winsys_context *swc,    // IN
+                       struct svga_winsys_surface *sid, // IN
+                       uint32 width,                // IN
+                       uint32 height,               // IN
+                       SVGA3dSurfaceFormat format)  // IN
+{
+   SVGA3dSize *mipSizes;
+   SVGA3dSurfaceFace *faces;
+   enum pipe_error ret;
+
+   ret = SVGA3D_BeginDefineSurface(swc,
+                                   sid, 0, format, &faces, &mipSizes, 1);
+   if(ret != PIPE_OK)
+      return ret;
+
+   faces[0].numMipLevels = 1;
+
+   mipSizes[0].width = width;
+   mipSizes[0].height = height;
+   mipSizes[0].depth = 1;
+ 
+   swc->commit(swc);;
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DestroySurface --
+ *
+ *      Release the host VRAM encapsulated by a particular surface ID.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DestroySurface(struct svga_winsys_context *swc,
+                      struct svga_winsys_surface *sid)  // IN
+{
+   SVGA3dCmdDestroySurface *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_DESTROY, sizeof *cmd, 1);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+   
+   swc->surface_relocation(swc, &cmd->sid, sid, PIPE_BUFFER_USAGE_GPU_READ);
+   swc->commit(swc);;
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginSurfaceDMA--
+ *
+ *      Begin a SURFACE_DMA command. This reserves space for it in
+ *      the FIFO, and returns a pointer to the command's box array.
+ *      This function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      When the SVGA3D device asynchronously processes this FIFO
+ *      command, a DMA operation is performed between host VRAM and
+ *      a generic SVGAGuestPtr. The guest pointer may refer to guest
+ *      VRAM (provided by the SVGA PCI device) or to guest system
+ *      memory that has been set up as a Guest Memory Region (GMR)
+ *      by the SVGA device.
+ *
+ *      The guest's DMA buffer must remain valid (not freed, paged out,
+ *      or overwritten) until the host has finished processing this
+ *      command. The guest can determine that the host has finished
+ *      by using the SVGA device's FIFO Fence mechanism.
+ *
+ *      The guest's image buffer can be an arbitrary size and shape.
+ *      Guest image data is interpreted according to the SVGA3D surface
+ *      format specified when the surface was defined.
+ *
+ *      The caller may optionally define the guest image's pitch.
+ *      guestImage->pitch can either be zero (assume image is tightly
+ *      packed) or it must be the number of bytes between vertically
+ *      adjacent image blocks.
+ *
+ *      The provided copybox list specifies which regions of the source
+ *      image are to be copied, and where they appear on the destination.
+ *
+ *      NOTE: srcx/srcy are always on the guest image and x/y are
+ *      always on the host image, regardless of the actual transfer
+ *      direction!
+ *
+ *      For efficiency, the SVGA3D device is free to copy more data
+ *      than specified. For example, it may round copy boxes outwards
+ *      such that they lie on particular alignment boundaries.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SurfaceDMA(struct svga_winsys_context *swc,
+                  struct svga_transfer *st,         // IN
+                  SVGA3dTransferType transfer,      // IN
+                  const SVGA3dCopyBox *boxes,       // IN
+                  uint32 numBoxes)                  // IN
+{
+   struct svga_texture *texture = svga_texture(st->base.texture); 
+   SVGA3dCmdSurfaceDMA *cmd;
+   SVGA3dCmdSurfaceDMASuffix *pSuffix;
+   uint32 boxesSize = sizeof *boxes * numBoxes;
+   unsigned region_flags;
+   unsigned surface_flags;
+   
+   if(transfer == SVGA3D_WRITE_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_READ;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+   }
+   else if(transfer == SVGA3D_READ_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_READ;
+   }
+   else {
+      assert(0);
+      return PIPE_ERROR_BAD_INPUT;
+   }
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_DMA,
+                            sizeof *cmd + boxesSize + sizeof *pSuffix,
+                            2);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   swc->region_relocation(swc, &cmd->guest.ptr, st->hwbuf, 0, region_flags);
+   cmd->guest.pitch = st->base.stride;
+
+   swc->surface_relocation(swc, &cmd->host.sid, texture->handle, surface_flags);
+   cmd->host.face = st->base.face; /* PIPE_TEX_FACE_* and SVGA3D_CUBEFACE_* match */
+   cmd->host.mipmap = st->base.level;
+
+   cmd->transfer = transfer;
+
+   memcpy(&cmd[1], boxes, boxesSize);
+   
+   pSuffix = (SVGA3dCmdSurfaceDMASuffix *)((uint8_t*)cmd + sizeof *cmd + boxesSize);
+   pSuffix->suffixSize = sizeof *pSuffix;
+   pSuffix->maximumOffset = st->hw_nblocksy*st->base.stride;
+   memset(&pSuffix->flags, 0, sizeof pSuffix->flags);
+
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+enum pipe_error
+SVGA3D_BufferDMA(struct svga_winsys_context *swc,
+                 struct svga_winsys_buffer *guest,
+                 struct svga_winsys_surface *host,
+                 SVGA3dTransferType transfer,      // IN
+                 uint32 size,                      // IN
+                 uint32 offset,                    // IN
+                 SVGA3dSurfaceDMAFlags flags)      // IN
+{
+   SVGA3dCmdSurfaceDMA *cmd;
+   SVGA3dCopyBox *box;
+   SVGA3dCmdSurfaceDMASuffix *pSuffix;
+   unsigned region_flags;
+   unsigned surface_flags;
+   
+   if(transfer == SVGA3D_WRITE_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_READ;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+   }
+   else if(transfer == SVGA3D_READ_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_READ;
+   }
+   else {
+      assert(0);
+      return PIPE_ERROR_BAD_INPUT;
+   }
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_DMA,
+                            sizeof *cmd + sizeof *box + sizeof *pSuffix,
+                            2);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   swc->region_relocation(swc, &cmd->guest.ptr, guest, 0, region_flags);
+   cmd->guest.pitch = 0;
+
+   swc->surface_relocation(swc, &cmd->host.sid, host, surface_flags);
+   cmd->host.face = 0;
+   cmd->host.mipmap = 0;
+
+   cmd->transfer = transfer;
+
+   box = (SVGA3dCopyBox *)&cmd[1];
+   box->x = offset;
+   box->y = 0;
+   box->z = 0;
+   box->w = size;
+   box->h = 1;
+   box->d = 1;
+   box->srcx = offset;
+   box->srcy = 0;
+   box->srcz = 0;
+   
+   pSuffix = (SVGA3dCmdSurfaceDMASuffix *)((uint8_t*)cmd + sizeof *cmd + sizeof *box);
+   pSuffix->suffixSize = sizeof *pSuffix;
+   pSuffix->maximumOffset = offset + size;
+   pSuffix->flags = flags;
+
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetRenderTarget --
+ *
+ *      Bind a surface object to a particular render target attachment
+ *      point on the current context. Render target attachment points
+ *      exist for color buffers, a depth buffer, and a stencil buffer.
+ *
+ *      The SVGA3D device is quite lenient about the types of surfaces
+ *      that may be used as render targets. The color buffers must
+ *      all be the same size, but the depth and stencil buffers do not
+ *      have to be the same size as the color buffer. All attachments
+ *      are optional.
+ *
+ *      Some combinations of render target formats may require software
+ *      emulation, depending on the capabilities of the host graphics
+ *      API and graphics hardware.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetRenderTarget(struct svga_winsys_context *swc,
+                       SVGA3dRenderTargetType type,   // IN
+                       struct pipe_surface *surface)  // IN
+{
+   SVGA3dCmdSetRenderTarget *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETRENDERTARGET, sizeof *cmd, 1);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+
+   cmd->cid = swc->cid;
+
+   cmd->type = type;
+
+   surface_to_surfaceid(swc, surface, &cmd->target, PIPE_BUFFER_USAGE_GPU_WRITE);
+
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+
+
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DefineShader --
+ *
+ *      Upload the bytecode for a new shader. The bytecode is "SVGA3D
+ *      format", which is theoretically a binary-compatible superset
+ *      of Microsoft's DirectX shader bytecode. In practice, the
+ *      SVGA3D bytecode doesn't yet have any extensions to DirectX's
+ *      bytecode format.
+ *
+ *      The SVGA3D device supports shader models 1.1 through 2.0.
+ *
+ *      The caller chooses a shader ID (small positive integer) by
+ *      which this shader will be identified in future commands. This
+ *      ID is in a namespace which is per-context and per-shader-type.
+ *
+ *      'bytecodeLen' is specified in bytes. It must be a multiple of 4.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DefineShader(struct svga_winsys_context *swc,
+                    uint32 shid,                  // IN
+                    SVGA3dShaderType type,        // IN
+                    const uint32 *bytecode,       // IN
+                    uint32 bytecodeLen)           // IN
+{
+   SVGA3dCmdDefineShader *cmd;
+
+   assert(bytecodeLen % 4 == 0);
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SHADER_DEFINE, sizeof *cmd + bytecodeLen,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->shid = shid;
+   cmd->type = type;
+   memcpy(&cmd[1], bytecode, bytecodeLen);
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_DestroyShader --
+ *
+ *      Delete a shader that was created by SVGA3D_DefineShader. If
+ *      the shader was the current vertex or pixel shader for its
+ *      context, rendering results are undefined until a new shader is
+ *      bound.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_DestroyShader(struct svga_winsys_context *swc,
+                     uint32 shid,            // IN
+                     SVGA3dShaderType type)  // IN
+{
+   SVGA3dCmdDestroyShader *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SHADER_DESTROY, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->shid = shid;
+   cmd->type = type;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetShaderConst --
+ *
+ *      Set the value of a shader constant.
+ *
+ *      Shader constants are analogous to uniform variables in GLSL,
+ *      except that they belong to the render context rather than to
+ *      an individual shader.
+ *
+ *      Constants may have one of three types: A 4-vector of floats,
+ *      a 4-vector of integers, or a single boolean flag.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetShaderConst(struct svga_winsys_context *swc,
+                      uint32 reg,                   // IN
+                      SVGA3dShaderType type,        // IN
+                      SVGA3dShaderConstType ctype,  // IN
+                      const void *value)            // IN
+{
+   SVGA3dCmdSetShaderConst *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SET_SHADER_CONST, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->reg = reg;
+   cmd->type = type;
+   cmd->ctype = ctype;
+
+   switch (ctype) {
+
+   case SVGA3D_CONST_TYPE_FLOAT:
+   case SVGA3D_CONST_TYPE_INT:
+      memcpy(&cmd->values, value, sizeof cmd->values);
+      break;
+
+   case SVGA3D_CONST_TYPE_BOOL:
+      memset(&cmd->values, 0, sizeof cmd->values);
+      cmd->values[0] = *(uint32*)value;
+      break;
+
+   default:
+      assert(0);
+      break;
+
+   }
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetShader --
+ *
+ *      Switch active shaders. This binds a new vertex or pixel shader
+ *      to the specified context.
+ *
+ *      A shader ID of SVGA3D_INVALID_ID unbinds any shader, switching
+ *      back to the fixed function vertex or pixel pipeline.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetShader(struct svga_winsys_context *swc,
+                 SVGA3dShaderType type,  // IN
+                 uint32 shid)            // IN
+{
+   SVGA3dCmdSetShader *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SET_SHADER, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+   
+   cmd->cid = swc->cid;
+   cmd->type = type;
+   cmd->shid = shid;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginClear --
+ *
+ *      Begin a CLEAR command. This reserves space for it in the FIFO,
+ *      and returns a pointer to the command's rectangle array.  This
+ *      function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      Clear is a rendering operation which fills a list of
+ *      rectangles with constant values on all render target types
+ *      indicated by 'flags'.
+ *
+ *      Clear is not affected by clipping, depth test, or other
+ *      render state which affects the fragment pipeline.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      May write to attached render target surfaces.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginClear(struct svga_winsys_context *swc,
+                  SVGA3dClearFlag flags,  // IN
+                  uint32 color,           // IN
+                  float depth,            // IN
+                  uint32 stencil,         // IN
+                  SVGA3dRect **rects,     // OUT
+                  uint32 numRects)        // IN
+{
+   SVGA3dCmdClear *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_CLEAR, 
+                            sizeof *cmd + sizeof **rects * numRects,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->clearFlag = flags;
+   cmd->color = color;
+   cmd->depth = depth;
+   cmd->stencil = stencil;
+   *rects = (SVGA3dRect*) &cmd[1];
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_ClearRect --
+ *
+ *      This is a simplified version of SVGA3D_BeginClear().
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_ClearRect(struct svga_winsys_context *swc,
+                 SVGA3dClearFlag flags,  // IN
+                 uint32 color,           // IN
+                 float depth,            // IN
+                 uint32 stencil,         // IN
+                 uint32 x,               // IN
+                 uint32 y,               // IN
+                 uint32 w,               // IN
+                 uint32 h)               // IN
+{
+   SVGA3dRect *rect;
+   enum pipe_error ret;
+
+   ret = SVGA3D_BeginClear(swc, flags, color, depth, stencil, &rect, 1);
+   if(ret != PIPE_OK)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   memset(rect, 0, sizeof *rect);
+   rect->x = x;
+   rect->y = y;
+   rect->w = w;
+   rect->h = h;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginDrawPrimitives --
+ *
+ *      Begin a DRAW_PRIMITIVES command. This reserves space for it in
+ *      the FIFO, and returns a pointer to the command's arrays.
+ *      This function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      Drawing commands consist of two variable-length arrays:
+ *      SVGA3dVertexDecl elements declare a set of vertex buffers to
+ *      use while rendering, and SVGA3dPrimitiveRange elements specify
+ *      groups of primitives each with an optional index buffer.
+ *
+ *      The decls and ranges arrays are initialized to zero.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      May write to attached render target surfaces.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginDrawPrimitives(struct svga_winsys_context *swc,
+                           SVGA3dVertexDecl **decls,      // OUT
+                           uint32 numVertexDecls,         // IN
+                           SVGA3dPrimitiveRange **ranges, // OUT
+                           uint32 numRanges)              // IN
+{
+   SVGA3dCmdDrawPrimitives *cmd;
+   SVGA3dVertexDecl *declArray;
+   SVGA3dPrimitiveRange *rangeArray;
+   uint32 declSize = sizeof **decls * numVertexDecls;
+   uint32 rangeSize = sizeof **ranges * numRanges;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_DRAW_PRIMITIVES, 
+                            sizeof *cmd + declSize + rangeSize,
+                            numVertexDecls + numRanges);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->numVertexDecls = numVertexDecls;
+   cmd->numRanges = numRanges;
+
+   declArray = (SVGA3dVertexDecl*) &cmd[1];
+   rangeArray = (SVGA3dPrimitiveRange*) &declArray[numVertexDecls];
+
+   memset(declArray, 0, declSize);
+   memset(rangeArray, 0, rangeSize);
+
+   *decls = declArray;
+   *ranges = rangeArray;
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginSurfaceCopy --
+ *
+ *      Begin a SURFACE_COPY command. This reserves space for it in
+ *      the FIFO, and returns a pointer to the command's arrays.  This
+ *      function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      The box array is initialized with zeroes.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      Asynchronously copies a list of boxes from surface to surface.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginSurfaceCopy(struct svga_winsys_context *swc,
+                        struct pipe_surface *src,    // IN
+                        struct pipe_surface *dest,   // IN
+                        SVGA3dCopyBox **boxes,       // OUT
+                        uint32 numBoxes)             // IN
+{
+   SVGA3dCmdSurfaceCopy *cmd;
+   uint32 boxesSize = sizeof **boxes * numBoxes;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_COPY, sizeof *cmd + boxesSize,
+                            2);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   surface_to_surfaceid(swc, src, &cmd->src, PIPE_BUFFER_USAGE_GPU_READ);
+   surface_to_surfaceid(swc, dest, &cmd->dest, PIPE_BUFFER_USAGE_GPU_WRITE);
+   *boxes = (SVGA3dCopyBox*) &cmd[1];
+
+   memset(*boxes, 0, boxesSize);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SurfaceStretchBlt --
+ *
+ *      Issue a SURFACE_STRETCHBLT command: an asynchronous
+ *      surface-to-surface blit, with scaling.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      Asynchronously copies one box from surface to surface.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SurfaceStretchBlt(struct svga_winsys_context *swc,
+                         struct pipe_surface *src,    // IN
+                         struct pipe_surface *dest,   // IN
+                         SVGA3dBox *boxSrc,           // IN
+                         SVGA3dBox *boxDest,          // IN
+                         SVGA3dStretchBltMode mode)   // IN
+{
+   SVGA3dCmdSurfaceStretchBlt *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_STRETCHBLT, sizeof *cmd,
+                            2);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   surface_to_surfaceid(swc, src, &cmd->src, PIPE_BUFFER_USAGE_GPU_READ);
+   surface_to_surfaceid(swc, dest, &cmd->dest, PIPE_BUFFER_USAGE_GPU_WRITE);
+   cmd->boxSrc = *boxSrc;
+   cmd->boxDest = *boxDest;
+   cmd->mode = mode;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetViewport --
+ *
+ *      Set the current context's viewport rectangle. The viewport
+ *      is clipped to the dimensions of the current render target,
+ *      then all rendering is clipped to the viewport.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetViewport(struct svga_winsys_context *swc,
+                   SVGA3dRect *rect)  // IN
+{
+   SVGA3dCmdSetViewport *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETVIEWPORT, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->rect = *rect;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetScissorRect --
+ *
+ *      Set the current context's scissor rectangle. If scissor
+ *      is enabled then all rendering is clipped to the scissor.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetScissorRect(struct svga_winsys_context *swc,
+                      SVGA3dRect *rect)  // IN
+{
+   SVGA3dCmdSetScissorRect *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETSCISSORRECT, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->rect = *rect;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetClipPlane --
+ *
+ *      Set one of the current context's clip planes. If the clip
+ *      plane is enabled then all 3d rendering is clipped to against
+ *      the plane.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error SVGA3D_SetClipPlane(struct svga_winsys_context *swc,
+                         uint32 index, const float *plane)
+{
+   SVGA3dCmdSetClipPlane *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETCLIPPLANE, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->index = index;
+   cmd->plane[0] = plane[0];
+   cmd->plane[1] = plane[1];
+   cmd->plane[2] = plane[2];
+   cmd->plane[3] = plane[3];
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_SetZRange --
+ *
+ *      Set the range of the depth buffer to use. 'min' and 'max'
+ *      are values between 0.0 and 1.0.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_SetZRange(struct svga_winsys_context *swc,
+                 float zMin,  // IN
+                 float zMax)  // IN
+{
+   SVGA3dCmdSetZRange *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETZRANGE, sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->zRange.min = zMin;
+   cmd->zRange.max = zMax;
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginSetTextureState --
+ *
+ *      Begin a SETTEXTURESTATE command. This reserves space for it in
+ *      the FIFO, and returns a pointer to the command's texture state
+ *      array.  This function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      This command sets rendering state which is per-texture-unit.
+ *
+ *      XXX: Individual texture states need documentation. However,
+ *           they are very similar to the texture states defined by
+ *           Direct3D. The D3D documentation is a good starting point
+ *           for understanding SVGA3D texture states.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginSetTextureState(struct svga_winsys_context *swc,
+                            SVGA3dTextureState **states,  // OUT
+                            uint32 numStates)             // IN
+{
+   SVGA3dCmdSetTextureState *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETTEXTURESTATE, 
+                            sizeof *cmd + sizeof **states * numStates,
+                            numStates);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   *states = (SVGA3dTextureState*) &cmd[1];
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginSetRenderState --
+ *
+ *      Begin a SETRENDERSTATE command. This reserves space for it in
+ *      the FIFO, and returns a pointer to the command's texture state
+ *      array.  This function must be paired with SVGA_FIFOCommitAll().
+ *
+ *      This command sets rendering state which is global to the context.
+ *
+ *      XXX: Individual render states need documentation. However,
+ *           they are very similar to the render states defined by
+ *           Direct3D. The D3D documentation is a good starting point
+ *           for understanding SVGA3D render states.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginSetRenderState(struct svga_winsys_context *swc,
+                           SVGA3dRenderState **states,  // OUT
+                           uint32 numStates)            // IN
+{
+   SVGA3dCmdSetRenderState *cmd;
+   
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SETRENDERSTATE, 
+                            sizeof *cmd + sizeof **states * numStates,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   *states = (SVGA3dRenderState*) &cmd[1];
+
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_BeginQuery--
+ *
+ *      Issues a SVGA_3D_CMD_BEGIN_QUERY command.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      Commits space in the FIFO memory.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_BeginQuery(struct svga_winsys_context *swc,
+                  SVGA3dQueryType type) // IN
+{
+   SVGA3dCmdBeginQuery *cmd;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_BEGIN_QUERY,
+                            sizeof *cmd,
+                            0);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->type = type;
+
+   swc->commit(swc);
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_EndQuery--
+ *
+ *      Issues a SVGA_3D_CMD_END_QUERY command.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      Commits space in the FIFO memory.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_EndQuery(struct svga_winsys_context *swc,
+                SVGA3dQueryType type,              // IN
+                struct svga_winsys_buffer *buffer) // IN/OUT
+{
+   SVGA3dCmdEndQuery *cmd;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_END_QUERY, 
+                            sizeof *cmd,
+                            1);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->type = type;
+
+   swc->region_relocation(swc, &cmd->guestResult, buffer, 0,
+                          PIPE_BUFFER_USAGE_GPU_WRITE);
+
+   swc->commit(swc);
+   
+   return PIPE_OK;
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
+ * SVGA3D_WaitForQuery--
+ *
+ *      Issues a SVGA_3D_CMD_WAIT_FOR_QUERY command.  This reserves space
+ *      for it in the FIFO.  This doesn't actually wait for the query to
+ *      finish but instead tells the host to start a wait at the driver
+ *      level.  The caller can wait on the status variable in the
+ *      guestPtr memory or send an insert fence instruction after this
+ *      command and wait on the fence.
+ *
+ * Results:
+ *      None.
+ *
+ * Side effects:
+ *      Commits space in the FIFO memory.
+ *
+ *----------------------------------------------------------------------
+ */
+
+enum pipe_error
+SVGA3D_WaitForQuery(struct svga_winsys_context *swc,
+                    SVGA3dQueryType type,              // IN
+                    struct svga_winsys_buffer *buffer) // IN/OUT
+{
+   SVGA3dCmdWaitForQuery *cmd;
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_WAIT_FOR_QUERY, 
+                            sizeof *cmd,
+                            1);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   cmd->cid = swc->cid;
+   cmd->type = type;
+   
+   swc->region_relocation(swc, &cmd->guestResult, buffer, 0,
+                          PIPE_BUFFER_USAGE_GPU_WRITE);
+
+   swc->commit(swc);
+   
+   return PIPE_OK;
+}
diff --git a/src/gallium/drivers/svga/svga_cmd.h b/src/gallium/drivers/svga/svga_cmd.h
new file mode 100644
index 0000000000..8041054769
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_cmd.h
@@ -0,0 +1,235 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/*
+ * svga_cmd.h --
+ *
+ *      Command construction utility for the SVGA3D protocol used by
+ *      the VMware SVGA device, based on the svgautil library.
+ */
+
+#ifndef __SVGA3D_H__
+#define __SVGA3D_H__
+
+
+#include "svga_types.h"
+#include "svga_reg.h"
+#include "svga3d_reg.h"
+
+#include "pipe/p_defines.h"
+
+
+struct pipe_buffer;
+struct pipe_surface;
+struct svga_transfer;
+struct svga_winsys_context;
+struct svga_winsys_buffer;
+struct svga_winsys_surface;
+
+
+/*
+ * SVGA Device Interoperability
+ */
+
+void *
+SVGA3D_FIFOReserve(struct svga_winsys_context *swc, uint32 cmd, uint32 cmdSize, uint32 nr_relocs);
+
+void
+SVGA_FIFOCommitAll(struct svga_winsys_context *swc);
+
+
+/*
+ * Context Management
+ */
+
+enum pipe_error
+SVGA3D_DefineContext(struct svga_winsys_context *swc);
+
+enum pipe_error
+SVGA3D_DestroyContext(struct svga_winsys_context *swc);
+
+
+/*
+ * Surface Management
+ */
+
+enum pipe_error
+SVGA3D_BeginDefineSurface(struct svga_winsys_context *swc,
+                          struct svga_winsys_surface *sid,
+                          SVGA3dSurfaceFlags flags,
+                          SVGA3dSurfaceFormat format,
+                          SVGA3dSurfaceFace **faces,
+                          SVGA3dSize **mipSizes,
+                          uint32 numMipSizes);
+enum pipe_error
+SVGA3D_DefineSurface2D(struct svga_winsys_context *swc,
+                       struct svga_winsys_surface *sid,
+                       uint32 width,
+                       uint32 height,
+                       SVGA3dSurfaceFormat format);
+enum pipe_error
+SVGA3D_DestroySurface(struct svga_winsys_context *swc,
+                      struct svga_winsys_surface *sid);
+
+
+/*
+ * Surface Operations
+ */
+
+enum pipe_error
+SVGA3D_SurfaceDMA(struct svga_winsys_context *swc,
+                  struct svga_transfer *st,
+                  SVGA3dTransferType transfer,
+                  const SVGA3dCopyBox *boxes,
+                  uint32 numBoxes);
+
+enum pipe_error
+SVGA3D_BufferDMA(struct svga_winsys_context *swc,
+                 struct svga_winsys_buffer *guest,
+                 struct svga_winsys_surface *host,
+                 SVGA3dTransferType transfer,
+                 uint32 size,
+                 uint32 offset,
+                 SVGA3dSurfaceDMAFlags flags);
+
+/*
+ * Drawing Operations
+ */
+
+
+enum pipe_error
+SVGA3D_BeginClear(struct svga_winsys_context *swc,
+                  SVGA3dClearFlag flags,
+                  uint32 color, float depth, uint32 stencil,
+                  SVGA3dRect **rects, uint32 numRects);
+
+enum pipe_error
+SVGA3D_ClearRect(struct svga_winsys_context *swc,
+                 SVGA3dClearFlag flags, uint32 color, float depth,
+                 uint32 stencil, uint32 x, uint32 y, uint32 w, uint32 h);
+
+enum pipe_error
+SVGA3D_BeginDrawPrimitives(struct svga_winsys_context *swc,
+                           SVGA3dVertexDecl **decls,
+                           uint32 numVertexDecls,
+                           SVGA3dPrimitiveRange **ranges,
+                           uint32 numRanges);
+
+/*
+ * Blits
+ */
+
+enum pipe_error
+SVGA3D_BeginSurfaceCopy(struct svga_winsys_context *swc,
+                        struct pipe_surface *src,
+                        struct pipe_surface *dest,
+                        SVGA3dCopyBox **boxes, uint32 numBoxes);
+
+
+enum pipe_error
+SVGA3D_SurfaceStretchBlt(struct svga_winsys_context *swc,
+                         struct pipe_surface *src,
+                         struct pipe_surface *dest,
+                         SVGA3dBox *boxSrc, SVGA3dBox *boxDest,
+                         SVGA3dStretchBltMode mode);
+
+/*
+ * Shared FFP/Shader Render State
+ */
+
+enum pipe_error
+SVGA3D_SetRenderTarget(struct svga_winsys_context *swc,
+                       SVGA3dRenderTargetType type,
+                       struct pipe_surface *surface);
+
+enum pipe_error
+SVGA3D_SetZRange(struct svga_winsys_context *swc,
+                 float zMin, float zMax);
+
+enum pipe_error
+SVGA3D_SetViewport(struct svga_winsys_context *swc,
+                   SVGA3dRect *rect);
+
+enum pipe_error
+SVGA3D_SetScissorRect(struct svga_winsys_context *swc,
+                      SVGA3dRect *rect);
+
+enum pipe_error
+SVGA3D_SetClipPlane(struct svga_winsys_context *swc,
+                    uint32 index, const float *plane);
+
+enum pipe_error
+SVGA3D_BeginSetTextureState(struct svga_winsys_context *swc,
+                            SVGA3dTextureState **states,
+                            uint32 numStates);
+
+enum pipe_error
+SVGA3D_BeginSetRenderState(struct svga_winsys_context *swc,
+                           SVGA3dRenderState **states,
+                           uint32 numStates);
+
+
+/*
+ * Shaders
+ */
+
+enum pipe_error
+SVGA3D_DefineShader(struct svga_winsys_context *swc,
+                    uint32 shid, SVGA3dShaderType type,
+                    const uint32 *bytecode, uint32 bytecodeLen);
+
+enum pipe_error
+SVGA3D_DestroyShader(struct svga_winsys_context *swc,
+                     uint32 shid, SVGA3dShaderType type);
+
+enum pipe_error
+SVGA3D_SetShaderConst(struct svga_winsys_context *swc,
+                      uint32 reg, SVGA3dShaderType type,
+                      SVGA3dShaderConstType ctype, const void *value);
+
+enum pipe_error
+SVGA3D_SetShader(struct svga_winsys_context *swc,
+                 SVGA3dShaderType type, uint32 shid);
+
+
+/*
+ * Queries
+ */
+
+enum pipe_error
+SVGA3D_BeginQuery(struct svga_winsys_context *swc,
+                  SVGA3dQueryType type);
+
+enum pipe_error
+SVGA3D_EndQuery(struct svga_winsys_context *swc,
+                SVGA3dQueryType type,
+                struct svga_winsys_buffer *buffer);
+
+enum pipe_error
+SVGA3D_WaitForQuery(struct svga_winsys_context *swc,
+                    SVGA3dQueryType type,
+                    struct svga_winsys_buffer *buffer);
+
+#endif /* __SVGA3D_H__ */
diff --git a/src/gallium/drivers/svga/svga_context.c b/src/gallium/drivers/svga/svga_context.c
new file mode 100644
index 0000000000..73233957f3
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_context.c
@@ -0,0 +1,269 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_screen.h"
+#include "util/u_memory.h"
+#include "util/u_upload_mgr.h"
+
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_screen_texture.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_swtnl.h"
+#include "svga_draw.h"
+#include "svga_debug.h"
+#include "svga_state.h"
+
+
+static void svga_destroy( struct pipe_context *pipe )
+{
+   struct svga_context *svga = svga_context( pipe );
+   unsigned shader;
+
+   svga_cleanup_framebuffer( svga );
+   svga_cleanup_tss_binding( svga );
+
+   svga_hwtnl_destroy( svga->hwtnl );
+
+   svga_cleanup_vertex_state(svga);
+   
+   svga->swc->destroy(svga->swc);
+   
+   svga_destroy_swtnl( svga );
+
+   u_upload_destroy( svga->upload_vb );
+   u_upload_destroy( svga->upload_ib );
+
+   for(shader = 0; shader < PIPE_SHADER_TYPES; ++shader)
+      pipe_buffer_reference( &svga->curr.cb[shader], NULL );
+
+   FREE( svga );
+}
+
+static unsigned int
+svga_is_texture_referenced( struct pipe_context *pipe,
+			    struct pipe_texture *texture,
+			    unsigned face, unsigned level)
+{
+   struct svga_texture *tex = svga_texture(texture);
+   struct svga_screen *ss = svga_screen(pipe->screen);
+
+   /**
+    * The screen does not cache texture writes.
+    */
+
+   if (!tex->handle || ss->sws->surface_is_flushed(ss->sws, tex->handle))
+      return PIPE_UNREFERENCED;
+
+   /**
+    * sws->surface_is_flushed() does not distinguish between read references
+    * and write references. So assume a reference is both.
+    */
+
+   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
+}
+
+static unsigned int
+svga_is_buffer_referenced( struct pipe_context *pipe,
+			   struct pipe_buffer *buf)
+
+{
+   struct svga_screen *ss = svga_screen(pipe->screen);
+   struct svga_buffer *sbuf = svga_buffer(buf);
+
+   /**
+    * XXX: Check this.
+    * The screen may cache buffer writes, but when we map, we map out
+    * of those cached writes, so we don't need to set a
+    * PIPE_REFERENCED_FOR_WRITE flag for cached buffers.
+    */
+
+   if (!sbuf->handle || ss->sws->surface_is_flushed(ss->sws, sbuf->handle))
+     return PIPE_UNREFERENCED;
+
+   /**
+    * sws->surface_is_flushed() does not distinguish between read references
+    * and write references. So assume a reference is both,
+    * however, we make an exception for index- and vertex buffers, to avoid
+    * a flush in st_bufferobj_get_subdata, during display list replay.
+    */
+
+   if (sbuf->base.usage & (PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_INDEX))
+      return PIPE_REFERENCED_FOR_READ;
+
+   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
+}
+
+
+struct pipe_context *svga_context_create( struct pipe_screen *screen )
+{
+   struct svga_screen *svgascreen = svga_screen(screen);
+   struct svga_context *svga = NULL;
+   enum pipe_error ret;
+
+   svga = CALLOC_STRUCT(svga_context);
+   if (svga == NULL)
+      goto error1;
+
+   svga->pipe.winsys = screen->winsys;
+   svga->pipe.screen = screen;
+   svga->pipe.destroy = svga_destroy;
+   svga->pipe.clear = svga_clear;
+
+   svga->pipe.is_texture_referenced = svga_is_texture_referenced;
+   svga->pipe.is_buffer_referenced = svga_is_buffer_referenced;
+
+   svga->swc = svgascreen->sws->context_create(svgascreen->sws);
+   if(!svga->swc)
+      goto error2;
+
+   svga_init_blend_functions(svga);
+   svga_init_blit_functions(svga);
+   svga_init_depth_stencil_functions(svga);
+   svga_init_draw_functions(svga);
+   svga_init_flush_functions(svga);
+   svga_init_misc_functions(svga);
+   svga_init_rasterizer_functions(svga);
+   svga_init_sampler_functions(svga);
+   svga_init_fs_functions(svga);
+   svga_init_vs_functions(svga);
+   svga_init_vertex_functions(svga);
+   svga_init_constbuffer_functions(svga);
+   svga_init_query_functions(svga);
+
+   /* debug */
+   svga->debug.no_swtnl = debug_get_bool_option("SVGA_NO_SWTNL", FALSE);
+   svga->debug.force_swtnl = debug_get_bool_option("SVGA_FORCE_SWTNL", FALSE);
+   svga->debug.use_min_mipmap = debug_get_bool_option("SVGA_USE_MIN_MIPMAP", FALSE);
+   svga->debug.disable_shader = debug_get_num_option("SVGA_DISABLE_SHADER", ~0);
+
+   if (!svga_init_swtnl(svga))
+      goto error3;
+
+   svga->upload_ib = u_upload_create( svga->pipe.screen,
+                                      32 * 1024,
+                                      16,
+                                      PIPE_BUFFER_USAGE_INDEX );
+   if (svga->upload_ib == NULL)
+      goto error4;
+
+   svga->upload_vb = u_upload_create( svga->pipe.screen,
+                                      128 * 1024,
+                                      16,
+                                      PIPE_BUFFER_USAGE_VERTEX );
+   if (svga->upload_vb == NULL)
+      goto error5;
+
+   svga->hwtnl = svga_hwtnl_create( svga,
+                                    svga->upload_ib,
+                                    svga->swc );
+   if (svga->hwtnl == NULL)
+      goto error6;
+
+
+   ret = svga_emit_initial_state( svga );
+   if (ret)
+      goto error7;
+   
+   /* Avoid shortcircuiting state with initial value of zero.
+    */
+   memset(&svga->state.hw_clear, 0xcd, sizeof(svga->state.hw_clear));
+   memset(&svga->state.hw_clear.framebuffer, 0x0, 
+          sizeof(svga->state.hw_clear.framebuffer));
+
+   memset(&svga->state.hw_draw, 0xcd, sizeof(svga->state.hw_draw));
+   memset(&svga->state.hw_draw.views, 0x0, sizeof(svga->state.hw_draw.views));
+   svga->state.hw_draw.num_views = 0;
+
+   svga->dirty = ~0;
+   svga->state.white_fs_id = SVGA3D_INVALID_ID;
+
+   LIST_INITHEAD(&svga->dirty_buffers);
+
+   return &svga->pipe;
+
+error7:
+   svga_hwtnl_destroy( svga->hwtnl );
+error6:
+   u_upload_destroy( svga->upload_vb );
+error5:
+   u_upload_destroy( svga->upload_ib );
+error4:
+   svga_destroy_swtnl(svga);
+error3:
+   svga->swc->destroy(svga->swc);
+error2:
+   FREE(svga);
+error1:
+   return NULL;
+}
+
+
+void svga_context_flush( struct svga_context *svga, 
+                         struct pipe_fence_handle **pfence )
+{
+   struct svga_screen *svgascreen = svga_screen(svga->pipe.screen);
+   
+   /* Unmap upload manager buffers: 
+    */
+   u_upload_flush(svga->upload_vb);
+   u_upload_flush(svga->upload_ib);
+
+   /* Flush screen, to ensure that texture dma uploads are processed
+    * before submitting commands.
+    */
+   svga_screen_flush(svgascreen, NULL);
+   
+   svga_context_flush_buffers(svga);
+
+   /* Flush pending commands to hardware:
+    */
+   svga->swc->flush(svga->swc, pfence);
+
+   if (SVGA_DEBUG & DEBUG_SYNC) {
+      if (pfence && *pfence)
+         svga->pipe.screen->fence_finish( svga->pipe.screen, *pfence, 0);
+   }
+}
+
+
+void svga_hwtnl_flush_retry( struct svga_context *svga )
+{
+   enum pipe_error ret = PIPE_OK;
+
+   ret = svga_hwtnl_flush( svga->hwtnl );
+   if (ret == PIPE_ERROR_OUT_OF_MEMORY) {
+      svga_context_flush( svga, NULL );
+      ret = svga_hwtnl_flush( svga->hwtnl );
+   }
+
+   assert(ret == 0);
+}
+
diff --git a/src/gallium/drivers/svga/svga_context.h b/src/gallium/drivers/svga/svga_context.h
new file mode 100644
index 0000000000..9a3e92fd8d
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_context.h
@@ -0,0 +1,443 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_CONTEXT_H
+#define SVGA_CONTEXT_H
+
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "util/u_double_list.h"
+
+#include "tgsi/tgsi_scan.h"
+
+
+#define SVGA_TEX_UNITS 8
+
+struct draw_vertex_shader;
+struct svga_shader_result;
+struct SVGACmdMemory;
+struct u_upload_mgr;
+
+
+struct svga_shader
+{
+   const struct tgsi_token *tokens;
+
+   struct tgsi_shader_info info;
+
+   struct svga_shader_result *results;
+
+   unsigned id;
+
+   boolean use_sm30;
+};
+
+struct svga_fragment_shader
+{
+   struct svga_shader base;
+};
+
+struct svga_vertex_shader
+{
+   struct svga_shader base;
+
+   struct draw_vertex_shader *draw_shader;
+};
+
+
+struct svga_cache_context;
+struct svga_tracked_state;
+
+struct svga_blend_state {
+
+   boolean need_white_fragments;
+
+   /* Should be per-render-target:
+    */
+   struct {
+      uint8_t writemask;
+
+      boolean blend_enable;
+      uint8_t srcblend;
+      uint8_t dstblend;
+      uint8_t blendeq;
+      
+      boolean separate_alpha_blend_enable;
+      uint8_t srcblend_alpha;
+      uint8_t dstblend_alpha;
+      uint8_t blendeq_alpha;
+
+   } rt[1];
+};
+
+struct svga_depth_stencil_state {
+   unsigned zfunc:8;
+   unsigned zenable:1;
+   unsigned zwriteenable:1;
+
+   unsigned alphatestenable:1;
+   unsigned alphafunc:8;
+  
+   struct {
+      unsigned enabled:1;
+      unsigned func:8;
+      unsigned fail:8;
+      unsigned zfail:8;
+      unsigned pass:8;
+   } stencil[2];
+   
+   /* SVGA3D has one ref/mask/writemask triple shared between front &
+    * back face stencil.  We really need two:
+    */
+   unsigned stencil_ref:8;
+   unsigned stencil_mask:8;
+   unsigned stencil_writemask:8;
+
+   float    alpharef;
+};
+
+#define SVGA_UNFILLED_DISABLE 0
+#define SVGA_UNFILLED_LINE    1
+#define SVGA_UNFILLED_POINT   2
+
+#define SVGA_PIPELINE_FLAG_POINTS   (1<<PIPE_PRIM_POINTS)
+#define SVGA_PIPELINE_FLAG_LINES    (1<<PIPE_PRIM_LINES)
+#define SVGA_PIPELINE_FLAG_TRIS     (1<<PIPE_PRIM_TRIANGLES)
+
+struct svga_rasterizer_state {
+   struct pipe_rasterizer_state templ; /* needed for draw module */
+
+   unsigned shademode:8;
+   unsigned cullmode:8;
+   unsigned scissortestenable:1;
+   unsigned multisampleantialias:1;
+   unsigned antialiasedlineenable:1;
+   unsigned lastpixel:1;
+
+   unsigned linepattern;
+
+   float slopescaledepthbias;
+   float depthbias;
+   float pointsize;
+   float pointsize_min;
+   float pointsize_max;
+   
+   unsigned hw_unfilled:16;         /* PIPE_POLYGON_MODE_x */
+   unsigned need_pipeline:16;    /* which prims do we need help for? */
+};
+
+struct svga_sampler_state {
+   unsigned mipfilter;
+   unsigned magfilter;
+   unsigned minfilter;
+   unsigned aniso_level;
+   float lod_bias;
+   unsigned addressu;
+   unsigned addressv;
+   unsigned addressw;
+   unsigned bordercolor;
+   unsigned normalized_coords:1;
+   unsigned compare_mode:1;
+   unsigned compare_func:3;
+
+   unsigned min_lod;
+   unsigned view_min_lod;
+   unsigned view_max_lod;
+};
+
+/* Use to calculate differences between state emitted to hardware and
+ * current driver-calculated state.  
+ */
+struct svga_state 
+{
+   const struct svga_blend_state *blend;
+   const struct svga_depth_stencil_state *depth;
+   const struct svga_rasterizer_state *rast;
+   const struct svga_sampler_state *sampler[PIPE_MAX_SAMPLERS];
+
+   struct pipe_texture *texture[PIPE_MAX_SAMPLERS]; /* or texture ID's? */
+   struct svga_fragment_shader *fs;
+   struct svga_vertex_shader *vs;
+
+   struct pipe_vertex_buffer vb[PIPE_MAX_ATTRIBS];
+   struct pipe_vertex_element ve[PIPE_MAX_ATTRIBS];
+   struct pipe_buffer *cb[PIPE_SHADER_TYPES];
+
+   struct pipe_framebuffer_state framebuffer;
+   float depthscale;
+
+   struct pipe_poly_stipple poly_stipple;
+   struct pipe_scissor_state scissor;
+   struct pipe_blend_color blend_color;
+   struct pipe_clip_state clip;
+   struct pipe_viewport_state viewport;
+
+   const unsigned *edgeflags;
+
+   unsigned num_samplers;
+   unsigned num_textures;
+   unsigned num_vertex_elements;
+   unsigned num_vertex_buffers;
+   unsigned reduced_prim;
+
+   struct {
+      unsigned flag_1d;
+      unsigned flag_srgb;
+   } tex_flags;
+
+   boolean any_user_vertex_buffers;
+
+   unsigned zero_stride_vertex_elements;
+   unsigned num_zero_stride_vertex_elements;
+   /* ### maybe dynamically allocate this */
+   float zero_stride_constants[PIPE_MAX_ATTRIBS*4];
+};
+
+#define RS_MAX 97
+#define TS_MAX 30
+#define CB_MAX 256
+
+struct svga_prescale {
+   float translate[4];
+   float scale[4];
+   boolean enabled;
+};
+
+
+/* Updated by calling svga_update_state( SVGA_STATE_HW_VIEWPORT )
+ */
+struct svga_hw_clear_state
+{
+   struct {
+      unsigned x,y,w,h;
+   } viewport;
+
+   struct {
+      float zmin, zmax;
+   } depthrange;
+   
+   struct pipe_framebuffer_state framebuffer;
+   struct svga_prescale prescale;
+};
+
+struct svga_hw_view_state
+{
+   struct pipe_texture *texture;
+   struct svga_sampler_view *v;
+   unsigned min_lod;
+   unsigned max_lod;
+   int dirty;
+};
+
+/* Updated by calling svga_update_state( SVGA_STATE_HW_DRAW )
+ */
+struct svga_hw_draw_state
+{
+   unsigned rs[RS_MAX];
+   unsigned ts[16][TS_MAX];
+   float cb[PIPE_SHADER_TYPES][CB_MAX][4];
+
+   unsigned shader_id[PIPE_SHADER_TYPES];
+   
+   struct svga_shader_result *fs;
+   struct svga_shader_result *vs;
+   struct svga_hw_view_state views[PIPE_MAX_SAMPLERS];
+
+   unsigned num_views;
+};
+
+
+/* Updated by calling svga_update_state( SVGA_STATE_NEED_SWTNL )
+ */
+struct svga_sw_state
+{
+   unsigned ve_format[PIPE_MAX_ATTRIBS]; /* NEW_VELEMENT */
+
+   /* which parts we need */
+   boolean need_swvfetch;
+   boolean need_pipeline;
+   boolean need_swtnl;
+};
+
+
+/* Queue some state updates (like rss) and submit them to hardware in
+ * a single packet.
+ */
+struct svga_hw_queue;
+
+struct svga_query;
+
+struct svga_context
+{
+   struct pipe_context pipe;
+   struct svga_winsys_context *swc;
+
+   struct {
+      boolean no_swtnl;
+      boolean force_swtnl;
+      boolean use_min_mipmap;
+
+      /* incremented for each shader */
+      unsigned shader_id;
+
+      unsigned disable_shader;
+   } debug;
+
+   struct {
+      struct draw_context *draw;
+      struct vbuf_render *backend;
+      unsigned hw_prim;
+      boolean new_vbuf;
+      boolean new_vdecl;
+   } swtnl;
+
+   struct {
+      unsigned dirty[4];
+
+      unsigned texture_timestamp;
+      unsigned next_fs_id;
+      unsigned next_vs_id;
+
+      /* Internally generated shaders:
+       */
+      unsigned white_fs_id;
+
+      /* 
+       */
+      struct svga_sw_state          sw;
+      struct svga_hw_draw_state     hw_draw;
+      struct svga_hw_clear_state    hw_clear;
+   } state;
+
+   struct svga_state curr;      /* state from the state tracker */
+   unsigned dirty;              /* statechanges since last update_state() */
+
+   struct u_upload_mgr *upload_ib;
+   struct u_upload_mgr *upload_vb;
+   struct svga_hwtnl *hwtnl;
+
+   /** The occlusion query currently in progress */
+   struct svga_query *sq;
+
+   /** List of buffers with queued transfers */
+   struct list_head dirty_buffers;
+};
+
+/* A flag for each state_tracker state object:
+ */
+#define SVGA_NEW_BLEND               0x1
+#define SVGA_NEW_DEPTH_STENCIL       0x2
+#define SVGA_NEW_RAST                0x4
+#define SVGA_NEW_SAMPLER             0x8
+#define SVGA_NEW_TEXTURE             0x10
+#define SVGA_NEW_VBUFFER             0x20
+#define SVGA_NEW_VELEMENT            0x40
+#define SVGA_NEW_FS                  0x80
+#define SVGA_NEW_VS                  0x100
+#define SVGA_NEW_FS_CONST_BUFFER     0x200
+#define SVGA_NEW_VS_CONST_BUFFER     0x400
+#define SVGA_NEW_FRAME_BUFFER        0x800
+#define SVGA_NEW_STIPPLE             0x1000
+#define SVGA_NEW_SCISSOR             0x2000
+#define SVGA_NEW_BLEND_COLOR         0x5000
+#define SVGA_NEW_CLIP                0x8000
+#define SVGA_NEW_VIEWPORT            0x10000
+#define SVGA_NEW_PRESCALE            0x20000
+#define SVGA_NEW_REDUCED_PRIMITIVE   0x40000
+#define SVGA_NEW_TEXTURE_BINDING     0x80000
+#define SVGA_NEW_NEED_PIPELINE       0x100000
+#define SVGA_NEW_NEED_SWVFETCH       0x200000
+#define SVGA_NEW_NEED_SWTNL          0x400000
+#define SVGA_NEW_FS_RESULT           0x800000
+#define SVGA_NEW_VS_RESULT           0x1000000
+#define SVGA_NEW_EDGEFLAGS           0x2000000
+#define SVGA_NEW_ZERO_STRIDE         0x4000000
+#define SVGA_NEW_TEXTURE_FLAGS       0x8000000
+
+
+
+
+
+/***********************************************************************
+ * svga_clear.c: 
+ */
+void svga_clear(struct pipe_context *pipe, 
+                unsigned buffers,
+                const float *rgba,
+                double depth,
+                unsigned stencil);
+
+
+/***********************************************************************
+ * svga_screen_texture.c: 
+ */
+void svga_mark_surfaces_dirty(struct svga_context *svga);
+
+
+
+
+void svga_init_state_functions( struct svga_context *svga );
+void svga_init_flush_functions( struct svga_context *svga );
+void svga_init_string_functions( struct svga_context *svga );
+void svga_init_blit_functions(struct svga_context *svga);
+
+void svga_init_blend_functions( struct svga_context *svga );
+void svga_init_depth_stencil_functions( struct svga_context *svga );
+void svga_init_misc_functions( struct svga_context *svga );
+void svga_init_rasterizer_functions( struct svga_context *svga );
+void svga_init_sampler_functions( struct svga_context *svga );
+void svga_init_fs_functions( struct svga_context *svga );
+void svga_init_vs_functions( struct svga_context *svga );
+void svga_init_vertex_functions( struct svga_context *svga );
+void svga_init_constbuffer_functions( struct svga_context *svga );
+void svga_init_draw_functions( struct svga_context *svga );
+void svga_init_query_functions( struct svga_context *svga );
+
+void svga_cleanup_vertex_state( struct svga_context *svga );
+void svga_cleanup_tss_binding( struct svga_context *svga );
+void svga_cleanup_framebuffer( struct svga_context *svga );
+
+void svga_context_flush( struct svga_context *svga,
+                         struct pipe_fence_handle **pfence );
+
+void svga_hwtnl_flush_retry( struct svga_context *svga );
+
+
+/***********************************************************************
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct svga_context *
+svga_context( struct pipe_context *pipe )
+{
+   return (struct svga_context *)pipe;
+}
+
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_debug.h b/src/gallium/drivers/svga/svga_debug.h
new file mode 100644
index 0000000000..b7bb5686ed
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_debug.h
@@ -0,0 +1,74 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_DEBUG_H
+#define SVGA_DEBUG_H
+
+#include "pipe/p_compiler.h"
+#include "util/u_debug.h"
+
+#define DEBUG_DMA      0x1
+#define DEBUG_TGSI     0x4
+#define DEBUG_PIPE     0x8
+#define DEBUG_STATE    0x10
+#define DEBUG_SCREEN   0x20
+#define DEBUG_TEX      0x40
+#define DEBUG_SWTNL    0x80
+#define DEBUG_CONSTS   0x100
+#define DEBUG_VIEWPORT 0x200
+#define DEBUG_VIEWS    0x400
+#define DEBUG_PERF     0x800    /* print something when we hit any slow path operation */
+#define DEBUG_FLUSH    0x1000   /* flush after every draw */
+#define DEBUG_SYNC     0x2000   /* sync after every flush */
+#define DEBUG_QUERY    0x4000
+
+#ifdef DEBUG
+extern int SVGA_DEBUG;
+#define DBSTR(x) x
+#else
+#define SVGA_DEBUG 0
+#define DBSTR(x) ""
+#endif
+
+static INLINE void
+SVGA_DBG( unsigned flag, const char *fmt, ... )
+{
+#ifdef DEBUG 
+    if (SVGA_DEBUG & flag)
+    {
+        va_list args;
+
+        va_start( args, fmt );
+        debug_vprintf( fmt, args );
+        va_end( args );
+    }
+#else
+    (void)flag;
+    (void)fmt;
+#endif
+}
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_draw.c b/src/gallium/drivers/svga/svga_draw.c
new file mode 100644
index 0000000000..1b371cecc6
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_draw.c
@@ -0,0 +1,370 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "svga_draw.h"
+#include "svga_draw_private.h"
+#include "svga_screen.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_cmd.h"
+
+
+struct svga_hwtnl *svga_hwtnl_create( struct svga_context *svga,
+                                      struct u_upload_mgr *upload_ib,
+                                      struct svga_winsys_context *swc )
+{
+   struct svga_hwtnl *hwtnl = CALLOC_STRUCT(svga_hwtnl);
+   if (hwtnl == NULL)
+      goto fail;
+
+   hwtnl->svga = svga;
+   hwtnl->upload_ib = upload_ib;
+   
+   hwtnl->cmd.swc = swc;
+
+   return hwtnl;
+
+fail:
+   return NULL;
+}
+
+void svga_hwtnl_destroy( struct svga_hwtnl *hwtnl )
+{
+   int i, j;
+
+   for (i = 0; i < PIPE_PRIM_MAX; i++) {
+      for (j = 0; j < IDX_CACHE_MAX; j++) {
+         pipe_buffer_reference( &hwtnl->index_cache[i][j].buffer,
+                                NULL );
+      }
+   }
+
+   for (i = 0; i < hwtnl->cmd.vdecl_count; i++)
+      pipe_buffer_reference(&hwtnl->cmd.vdecl_vb[i], NULL);
+
+   for (i = 0; i < hwtnl->cmd.prim_count; i++)
+      pipe_buffer_reference(&hwtnl->cmd.prim_ib[i], NULL);
+      
+
+   FREE(hwtnl);
+}
+
+
+void svga_hwtnl_set_flatshade( struct svga_hwtnl *hwtnl,
+                               boolean flatshade,
+                               boolean flatshade_first )
+{
+   hwtnl->hw_pv = PV_FIRST;
+   hwtnl->api_pv = (flatshade && !flatshade_first) ? PV_LAST : PV_FIRST;
+}                               
+
+void svga_hwtnl_set_unfilled( struct svga_hwtnl *hwtnl,
+                              unsigned mode )
+{
+   hwtnl->api_fillmode = mode;
+}                               
+
+void svga_hwtnl_reset_vdecl( struct svga_hwtnl *hwtnl,
+                             unsigned count )
+{
+   unsigned i;
+
+   assert(hwtnl->cmd.prim_count == 0);
+
+   for (i = count; i < hwtnl->cmd.vdecl_count; i++) {
+      pipe_buffer_reference(&hwtnl->cmd.vdecl_vb[i],
+                            NULL);
+   }
+
+   hwtnl->cmd.vdecl_count = count;
+}
+
+
+void svga_hwtnl_vdecl( struct svga_hwtnl *hwtnl,
+                          unsigned i,
+                          const SVGA3dVertexDecl *decl,
+                          struct pipe_buffer *vb)
+{
+   assert(hwtnl->cmd.prim_count == 0);
+
+   assert( i < hwtnl->cmd.vdecl_count );
+
+   hwtnl->cmd.vdecl[i] = *decl;
+
+   pipe_buffer_reference(&hwtnl->cmd.vdecl_vb[i],
+                         vb);   
+}
+
+
+
+enum pipe_error
+svga_hwtnl_flush( struct svga_hwtnl *hwtnl )
+{
+   struct svga_winsys_context *swc = hwtnl->cmd.swc;
+   struct svga_context *svga = hwtnl->svga;
+   enum pipe_error ret;
+
+   if (hwtnl->cmd.prim_count) {
+      struct svga_winsys_surface *vb_handle[SVGA3D_INPUTREG_MAX];
+      struct svga_winsys_surface *ib_handle[QSZ];
+      struct svga_winsys_surface *handle;
+      SVGA3dVertexDecl *vdecl;
+      SVGA3dPrimitiveRange *prim;
+      unsigned i;
+
+      for (i = 0; i < hwtnl->cmd.vdecl_count; i++) {
+         handle = svga_buffer_handle(svga, hwtnl->cmd.vdecl_vb[i]);
+         if (handle == NULL)
+            return PIPE_ERROR_OUT_OF_MEMORY;
+
+         vb_handle[i] = handle;
+      }
+
+      for (i = 0; i < hwtnl->cmd.prim_count; i++) {
+         if (hwtnl->cmd.prim_ib[i]) {
+            handle = svga_buffer_handle(svga, hwtnl->cmd.prim_ib[i]);
+            if (handle == NULL)
+               return PIPE_ERROR_OUT_OF_MEMORY;
+         }
+         else
+            handle = NULL;
+
+         ib_handle[i] = handle;
+      }
+
+      ret = SVGA3D_BeginDrawPrimitives(swc, 
+                                       &vdecl, 
+                                       hwtnl->cmd.vdecl_count, 
+                                       &prim, 
+                                       hwtnl->cmd.prim_count);
+      if (ret != PIPE_OK) 
+         return ret;
+
+      
+      memcpy( vdecl,
+              hwtnl->cmd.vdecl,
+              hwtnl->cmd.vdecl_count * sizeof hwtnl->cmd.vdecl[0]);
+
+      for (i = 0; i < hwtnl->cmd.vdecl_count; i++) {
+         /* Given rangeHint is considered to be relative to indexBias, and 
+          * indexBias varies per primitive, we cannot accurately supply an 
+          * rangeHint when emitting more than one primitive per draw command.
+          */
+         if (hwtnl->cmd.prim_count == 1) {
+            vdecl[i].rangeHint.first = hwtnl->cmd.min_index[0];
+            vdecl[i].rangeHint.last = hwtnl->cmd.max_index[0] + 1;
+         }
+         else {
+            vdecl[i].rangeHint.first = 0;
+            vdecl[i].rangeHint.last = 0;
+         }
+
+         swc->surface_relocation(swc,
+                                 &vdecl[i].array.surfaceId,
+                                 vb_handle[i],
+                                 PIPE_BUFFER_USAGE_GPU_READ);
+      }
+
+      memcpy( prim,
+              hwtnl->cmd.prim,
+              hwtnl->cmd.prim_count * sizeof hwtnl->cmd.prim[0]);
+
+      for (i = 0; i < hwtnl->cmd.prim_count; i++) {
+         swc->surface_relocation(swc,
+                                 &prim[i].indexArray.surfaceId,
+                                 ib_handle[i],
+                                 PIPE_BUFFER_USAGE_GPU_READ);
+         pipe_buffer_reference(&hwtnl->cmd.prim_ib[i], NULL);
+      }
+      
+      SVGA_FIFOCommitAll( swc );
+      hwtnl->cmd.prim_count = 0;
+   }
+
+   return PIPE_OK;
+}
+
+
+
+
+
+/***********************************************************************
+ * Internal functions:
+ */
+
+enum pipe_error svga_hwtnl_prim( struct svga_hwtnl *hwtnl,
+                                 const SVGA3dPrimitiveRange *range,
+                                 unsigned min_index,
+                                 unsigned max_index,
+                                 struct pipe_buffer *ib )
+{
+   int ret = PIPE_OK;
+
+#ifdef DEBUG
+   {
+      unsigned i;
+      for (i = 0; i < hwtnl->cmd.vdecl_count; i++) {
+         struct pipe_buffer *vb = hwtnl->cmd.vdecl_vb[i];
+         unsigned size = vb ? vb->size : 0;
+         unsigned offset = hwtnl->cmd.vdecl[i].array.offset;
+         unsigned stride = hwtnl->cmd.vdecl[i].array.stride;
+         unsigned index_bias = range->indexBias;
+         unsigned width;
+
+         assert(vb);
+         assert(size);
+         assert(offset < size);
+         assert(index_bias >= 0);
+         assert(min_index <= max_index);
+         assert(offset + index_bias*stride < size);
+         assert(offset + (index_bias + min_index)*stride < size);
+
+         switch (hwtnl->cmd.vdecl[i].identity.type) {
+         case SVGA3D_DECLTYPE_FLOAT1:
+            width = 4;
+            break;
+         case SVGA3D_DECLTYPE_FLOAT2:
+            width = 4*2;
+            break;
+         case SVGA3D_DECLTYPE_FLOAT3:
+            width = 4*3;
+            break;
+         case SVGA3D_DECLTYPE_FLOAT4:
+            width = 4*4;
+            break;
+         case SVGA3D_DECLTYPE_D3DCOLOR:
+            width = 4;
+            break;
+         case SVGA3D_DECLTYPE_UBYTE4:
+            width = 1*4;
+            break;
+         case SVGA3D_DECLTYPE_SHORT2:
+            width = 2*2;
+            break;
+         case SVGA3D_DECLTYPE_SHORT4:
+            width = 2*4;
+            break;
+         case SVGA3D_DECLTYPE_UBYTE4N:
+            width = 1*4;
+            break;
+         case SVGA3D_DECLTYPE_SHORT2N:
+            width = 2*2;
+            break;
+         case SVGA3D_DECLTYPE_SHORT4N:
+            width = 2*4;
+            break;
+         case SVGA3D_DECLTYPE_USHORT2N:
+            width = 2*2;
+            break;
+         case SVGA3D_DECLTYPE_USHORT4N:
+            width = 2*4;
+            break;
+         case SVGA3D_DECLTYPE_UDEC3:
+            width = 4;
+            break;
+         case SVGA3D_DECLTYPE_DEC3N:
+            width = 4;
+            break;
+         case SVGA3D_DECLTYPE_FLOAT16_2:
+            width = 2*2;
+            break;
+         case SVGA3D_DECLTYPE_FLOAT16_4:
+            width = 2*4;
+            break;
+         default:
+            assert(0);
+            width = 0;
+            break;
+         }
+
+         assert(!stride || width <= stride);
+         assert(offset + (index_bias + max_index)*stride + width <= size);
+      }
+
+      assert(range->indexWidth == range->indexArray.stride);
+
+      if(ib) {
+         unsigned size = ib->size;
+         unsigned offset = range->indexArray.offset;
+         unsigned stride = range->indexArray.stride;
+         unsigned count;
+
+         assert(size);
+         assert(offset < size);
+         assert(stride);
+
+         switch (range->primType) {
+         case SVGA3D_PRIMITIVE_POINTLIST:
+            count = range->primitiveCount;
+            break;
+         case SVGA3D_PRIMITIVE_LINELIST:
+            count = range->primitiveCount * 2;
+            break;
+         case SVGA3D_PRIMITIVE_LINESTRIP:
+            count = range->primitiveCount + 1;
+            break;
+         case SVGA3D_PRIMITIVE_TRIANGLELIST:
+            count = range->primitiveCount * 3;
+            break;
+         case SVGA3D_PRIMITIVE_TRIANGLESTRIP:
+            count = range->primitiveCount + 2;
+            break;
+         case SVGA3D_PRIMITIVE_TRIANGLEFAN:
+            count = range->primitiveCount + 2;
+            break;
+         default:
+            assert(0);
+            count = 0;
+            break;
+         }
+
+         assert(offset + count*stride <= size);
+      }
+   }
+#endif
+
+   if (hwtnl->cmd.prim_count+1 >= QSZ) {
+      ret = svga_hwtnl_flush( hwtnl );
+      if (ret != PIPE_OK)
+         return ret;
+   }
+   
+   /* min/max indices are relative to bias */
+   hwtnl->cmd.min_index[hwtnl->cmd.prim_count] = min_index;
+   hwtnl->cmd.max_index[hwtnl->cmd.prim_count] = max_index;
+
+   hwtnl->cmd.prim[hwtnl->cmd.prim_count] = *range;
+
+   pipe_buffer_reference(&hwtnl->cmd.prim_ib[hwtnl->cmd.prim_count], ib);
+   hwtnl->cmd.prim_count++;
+
+   return ret;
+}
diff --git a/src/gallium/drivers/svga/svga_draw.h b/src/gallium/drivers/svga/svga_draw.h
new file mode 100644
index 0000000000..14553b17b5
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_draw.h
@@ -0,0 +1,83 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_DRAW_H
+#define SVGA_DRAW_H
+
+#include "pipe/p_compiler.h"
+
+#include "svga_hw_reg.h"
+
+struct svga_hwtnl;
+struct svga_winsys_context;
+struct svga_screen;
+struct svga_context;
+struct pipe_buffer;
+struct u_upload_mgr;
+
+struct svga_hwtnl *svga_hwtnl_create( struct svga_context *svga,
+                                      struct u_upload_mgr *upload_ib,
+                                      struct svga_winsys_context *swc );
+
+void svga_hwtnl_destroy( struct svga_hwtnl *hwtnl );
+
+void svga_hwtnl_set_flatshade( struct svga_hwtnl *hwtnl,
+                               boolean flatshade,
+                               boolean flatshade_first );
+
+void svga_hwtnl_set_unfilled( struct svga_hwtnl *hwtnl,
+                              unsigned mode );
+
+void svga_hwtnl_vdecl( struct svga_hwtnl *hwtnl,
+                       unsigned i,
+                       const SVGA3dVertexDecl *decl,
+                       struct pipe_buffer *vb);
+
+void svga_hwtnl_reset_vdecl( struct svga_hwtnl *hwtnl,
+                             unsigned count );
+
+
+enum pipe_error 
+svga_hwtnl_draw_arrays( struct svga_hwtnl *hwtnl,
+                        unsigned prim, 
+                        unsigned start, 
+                        unsigned count);
+
+enum pipe_error
+svga_hwtnl_draw_range_elements( struct svga_hwtnl *hwtnl,
+                                struct pipe_buffer *indexBuffer,
+                                unsigned index_size,
+                                unsigned min_index,
+                                unsigned max_index,
+                                unsigned prim, 
+                                unsigned start, 
+                                unsigned count,
+                                unsigned bias );
+
+enum pipe_error
+svga_hwtnl_flush( struct svga_hwtnl *hwtnl );
+
+
+#endif /* SVGA_DRAW_H_ */
diff --git a/src/gallium/drivers/svga/svga_draw_arrays.c b/src/gallium/drivers/svga/svga_draw_arrays.c
new file mode 100644
index 0000000000..75492dffca
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_draw_arrays.c
@@ -0,0 +1,297 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_inlines.h"
+#include "util/u_prim.h"
+#include "indices/u_indices.h"
+
+#include "svga_hw_reg.h"
+#include "svga_draw.h"
+#include "svga_draw_private.h"
+#include "svga_context.h"
+
+
+#define DBG 0
+
+
+
+
+static enum pipe_error generate_indices( struct svga_hwtnl *hwtnl,
+                                         unsigned nr,
+                                         unsigned index_size,
+                                         u_generate_func generate,
+                                         struct pipe_buffer **out_buf )
+{
+   struct pipe_screen *screen = hwtnl->svga->pipe.screen;
+   unsigned size = index_size * nr;
+   struct pipe_buffer *dst = NULL;
+   void *dst_map = NULL;
+
+   dst = screen->buffer_create( screen, 32, 
+                                PIPE_BUFFER_USAGE_INDEX |
+                                PIPE_BUFFER_USAGE_CPU_WRITE |
+                                PIPE_BUFFER_USAGE_GPU_READ, 
+                                size );
+   if (dst == NULL)
+      goto fail;
+
+   dst_map = pipe_buffer_map( screen, dst, PIPE_BUFFER_USAGE_CPU_WRITE );
+   if (dst_map == NULL)
+      goto fail;
+
+   generate( nr,
+             dst_map );
+
+   pipe_buffer_unmap( screen, dst );
+
+   *out_buf = dst;
+   return PIPE_OK;
+
+fail:
+   if (dst_map)
+      screen->buffer_unmap( screen, dst );
+
+   if (dst)
+      screen->buffer_destroy( dst );
+
+   return PIPE_ERROR_OUT_OF_MEMORY;
+}
+
+static boolean compare( unsigned cached_nr,
+                        unsigned nr,
+                        unsigned type )
+{
+   if (type == U_GENERATE_REUSABLE)
+      return cached_nr >= nr;
+   else
+      return cached_nr == nr;
+}
+
+static enum pipe_error retrieve_or_generate_indices( struct svga_hwtnl *hwtnl,
+                                                     unsigned prim,
+                                                     unsigned gen_type,
+                                                     unsigned gen_nr,
+                                                     unsigned gen_size,
+                                                     u_generate_func generate,
+                                                     struct pipe_buffer **out_buf )
+{
+   enum pipe_error ret = PIPE_OK;
+   int i;
+
+   for (i = 0; i < IDX_CACHE_MAX; i++) {
+      if (hwtnl->index_cache[prim][i].buffer != NULL &&
+          hwtnl->index_cache[prim][i].generate == generate)
+      {
+         if (compare(hwtnl->index_cache[prim][i].gen_nr, gen_nr, gen_type))
+         {
+            pipe_buffer_reference( out_buf,
+                                   hwtnl->index_cache[prim][i].buffer );
+
+            if (DBG) 
+               debug_printf("%s retrieve %d/%d\n", __FUNCTION__, i, gen_nr);
+
+            return PIPE_OK;
+         }
+         else if (gen_type == U_GENERATE_REUSABLE) 
+         {
+            pipe_buffer_reference( &hwtnl->index_cache[prim][i].buffer,
+                                   NULL );
+
+            if (DBG) 
+               debug_printf("%s discard %d/%d\n", __FUNCTION__, 
+                            i, hwtnl->index_cache[prim][i].gen_nr);
+
+            break;
+         }
+      }
+   }
+
+   if (i == IDX_CACHE_MAX)
+   {
+      unsigned smallest = 0;
+      unsigned smallest_size = ~0;
+      
+      for (i = 0; i < IDX_CACHE_MAX && smallest_size; i++) {
+         if (hwtnl->index_cache[prim][i].buffer == NULL)
+         {
+            smallest = i;
+            smallest_size = 0;
+         }
+         else if (hwtnl->index_cache[prim][i].gen_nr < smallest)
+         {
+            smallest = i;
+            smallest_size = hwtnl->index_cache[prim][i].gen_nr;
+         }
+      }
+
+      assert (smallest != IDX_CACHE_MAX);
+
+      pipe_buffer_reference( &hwtnl->index_cache[prim][smallest].buffer,
+                             NULL );
+
+      if (DBG)
+         debug_printf("%s discard smallest %d/%d\n", __FUNCTION__, 
+                      smallest, smallest_size);
+      
+      i = smallest;
+   }
+      
+      
+   ret = generate_indices( hwtnl, 
+                           gen_nr,
+                           gen_size,
+                           generate,
+                           out_buf );
+   if (ret != PIPE_OK)
+      return ret;
+
+
+   hwtnl->index_cache[prim][i].generate = generate;
+   hwtnl->index_cache[prim][i].gen_nr = gen_nr;
+   pipe_buffer_reference( &hwtnl->index_cache[prim][i].buffer,
+                          *out_buf );
+
+   if (DBG)
+      debug_printf("%s cache %d/%d\n", __FUNCTION__, 
+                   i, hwtnl->index_cache[prim][i].gen_nr);
+
+   return PIPE_OK;
+}
+
+
+
+static enum pipe_error
+simple_draw_arrays( struct svga_hwtnl *hwtnl,
+                    unsigned prim, unsigned start, unsigned count )
+{
+   SVGA3dPrimitiveRange range;
+   unsigned hw_prim;
+   unsigned hw_count;
+
+   hw_prim = svga_translate_prim(prim, count, &hw_count);
+   if (hw_count == 0)
+      return PIPE_ERROR_BAD_INPUT;
+      
+   range.primType = hw_prim;
+   range.primitiveCount = hw_count;
+   range.indexArray.surfaceId = SVGA3D_INVALID_ID;
+   range.indexArray.offset = 0;
+   range.indexArray.stride = 0;
+   range.indexWidth = 0;
+   range.indexBias = start;
+
+   /* Min/max index should be calculated prior to applying bias, so we
+    * end up with min_index = 0, max_index = count - 1 and everybody
+    * looking at those numbers knows to adjust them by
+    * range.indexBias.
+    */
+   return svga_hwtnl_prim( hwtnl, &range, 0, count - 1, NULL );
+}
+
+
+
+
+
+
+
+
+
+
+enum pipe_error 
+svga_hwtnl_draw_arrays( struct svga_hwtnl *hwtnl,
+                        unsigned prim, 
+                        unsigned start, 
+                        unsigned count)
+{
+   unsigned gen_prim, gen_size, gen_nr, gen_type;
+   u_generate_func gen_func;
+   enum pipe_error ret = PIPE_OK;
+
+   if (hwtnl->api_fillmode != PIPE_POLYGON_MODE_FILL && 
+       prim >= PIPE_PRIM_TRIANGLES) 
+   {
+      gen_type = u_unfilled_generator( prim,
+                                       start,
+                                       count,
+                                       hwtnl->api_fillmode,
+                                       &gen_prim,
+                                       &gen_size,
+                                       &gen_nr,
+                                       &gen_func );
+   }
+   else {
+      gen_type = u_index_generator( svga_hw_prims,
+                                    prim,
+                                    start,
+                                    count,
+                                    hwtnl->api_pv,
+                                    hwtnl->hw_pv,
+                                    &gen_prim,
+                                    &gen_size,
+                                    &gen_nr,
+                                    &gen_func );
+   }
+
+   if (gen_type == U_GENERATE_LINEAR) {
+      return simple_draw_arrays( hwtnl, gen_prim, start, count );
+   }
+   else {
+      struct pipe_buffer *gen_buf = NULL;
+
+      /* Need to draw as indexed primitive. 
+       * Potentially need to run the gen func to build an index buffer.
+       */
+      ret = retrieve_or_generate_indices( hwtnl,
+                                          prim,
+                                          gen_type,
+                                          gen_nr,
+                                          gen_size,
+                                          gen_func,
+                                          &gen_buf );
+      if (ret)
+         goto done;
+
+      ret = svga_hwtnl_simple_draw_range_elements( hwtnl,
+                                                   gen_buf,
+                                                   gen_size,
+                                                   0,
+                                                   count - 1,
+                                                   gen_prim,
+                                                   0,
+                                                   gen_nr,
+                                                   start );
+      if (ret)
+         goto done;
+
+   done:
+      if (gen_buf)
+         pipe_buffer_reference( &gen_buf, NULL );
+
+      return ret;
+   }
+}
+
diff --git a/src/gallium/drivers/svga/svga_draw_elements.c b/src/gallium/drivers/svga/svga_draw_elements.c
new file mode 100644
index 0000000000..167d817831
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_draw_elements.c
@@ -0,0 +1,255 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "util/u_prim.h"
+#include "util/u_upload_mgr.h"
+#include "indices/u_indices.h"
+
+#include "svga_cmd.h"
+#include "svga_draw.h"
+#include "svga_draw_private.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_context.h"
+
+#include "svga_hw_reg.h"
+
+
+static enum pipe_error
+translate_indices( struct svga_hwtnl *hwtnl,
+                   struct pipe_buffer *src,
+                   unsigned offset,
+                   unsigned nr,
+                   unsigned index_size,
+                   u_translate_func translate,
+                   struct pipe_buffer **out_buf )
+{
+   struct pipe_screen *screen = hwtnl->svga->pipe.screen;
+   unsigned size = index_size * nr;
+   const void *src_map = NULL;
+   struct pipe_buffer *dst = NULL;
+   void *dst_map = NULL;
+
+   dst = screen->buffer_create( screen, 32, 
+                                PIPE_BUFFER_USAGE_INDEX |
+                                PIPE_BUFFER_USAGE_CPU_WRITE |
+                                PIPE_BUFFER_USAGE_GPU_READ, 
+                                size );
+   if (dst == NULL)
+      goto fail;
+
+   src_map = pipe_buffer_map( screen, src, PIPE_BUFFER_USAGE_CPU_READ );
+   if (src_map == NULL)
+      goto fail;
+
+   dst_map = pipe_buffer_map( screen, dst, PIPE_BUFFER_USAGE_CPU_WRITE );
+   if (dst_map == NULL)
+      goto fail;
+
+   translate( (const char *)src_map + offset,
+              nr,
+              dst_map );
+
+   pipe_buffer_unmap( screen, src );
+   pipe_buffer_unmap( screen, dst );
+
+   *out_buf = dst;
+   return PIPE_OK;
+
+fail:
+   if (src_map)
+      screen->buffer_unmap( screen, src );
+
+   if (dst_map)
+      screen->buffer_unmap( screen, dst );
+
+   if (dst)
+      screen->buffer_destroy( dst );
+
+   return PIPE_ERROR_OUT_OF_MEMORY;
+}
+
+
+
+
+
+enum pipe_error
+svga_hwtnl_simple_draw_range_elements( struct svga_hwtnl *hwtnl,
+                                       struct pipe_buffer *index_buffer,
+                                       unsigned index_size,
+                                       unsigned min_index,
+                                       unsigned max_index,
+                                       unsigned prim, 
+                                       unsigned start,
+                                       unsigned count,
+                                       unsigned bias )
+{
+   struct pipe_buffer *upload_buffer = NULL;
+   SVGA3dPrimitiveRange range;
+   unsigned hw_prim;
+   unsigned hw_count;
+   unsigned index_offset = start * index_size;
+   int ret = PIPE_OK;
+
+   hw_prim = svga_translate_prim(prim, count, &hw_count);
+   if (hw_count == 0)
+      goto done;
+
+   if (index_buffer && 
+       svga_buffer_is_user_buffer(index_buffer)) 
+   {
+      assert( index_buffer->size >= index_offset + count * index_size );
+
+      ret = u_upload_buffer( hwtnl->upload_ib,
+                             index_offset,
+                             count * index_size,
+                             index_buffer,
+                             &index_offset,
+                             &upload_buffer );
+      if (ret)
+         goto done;
+
+      /* Don't need to worry about refcounting index_buffer as this is
+       * just a stack variable without a counted reference of its own.
+       * The caller holds the reference.
+       */
+      index_buffer = upload_buffer;
+   }
+
+   range.primType = hw_prim;
+   range.primitiveCount = hw_count;
+   range.indexArray.offset = index_offset;
+   range.indexArray.stride = index_size;
+   range.indexWidth = index_size;
+   range.indexBias = bias;
+      
+   ret = svga_hwtnl_prim( hwtnl, &range, min_index, max_index, index_buffer );
+   if (ret)
+      goto done;
+
+done:
+   if (upload_buffer)
+      pipe_buffer_reference( &upload_buffer, NULL );
+
+   return ret;
+}
+
+
+
+
+enum pipe_error
+svga_hwtnl_draw_range_elements( struct svga_hwtnl *hwtnl,
+                                struct pipe_buffer *index_buffer,
+                                unsigned index_size,
+                                unsigned min_index,
+                                unsigned max_index,
+                                unsigned prim, unsigned start, unsigned count,
+                                unsigned bias)
+{
+   unsigned gen_prim, gen_size, gen_nr, gen_type;
+   u_translate_func gen_func;
+   enum pipe_error ret = PIPE_OK;
+
+   if (hwtnl->api_fillmode != PIPE_POLYGON_MODE_FILL && 
+       prim >= PIPE_PRIM_TRIANGLES) 
+   {
+      gen_type = u_unfilled_translator( prim,
+                                        index_size,
+                                        count,
+                                        hwtnl->api_fillmode,
+                                        &gen_prim,
+                                        &gen_size,
+                                        &gen_nr,
+                                        &gen_func );
+   }
+   else
+   {
+      gen_type = u_index_translator( svga_hw_prims,
+                                     prim,
+                                     index_size,
+                                     count,
+                                     hwtnl->api_pv,
+                                     hwtnl->hw_pv,
+                                     &gen_prim,
+                                     &gen_size,
+                                     &gen_nr,
+                                     &gen_func );
+   }
+
+   
+   if (gen_type == U_TRANSLATE_MEMCPY) {
+      /* No need for translation, just pass through to hardware: 
+       */
+      return svga_hwtnl_simple_draw_range_elements( hwtnl, index_buffer,
+                                                    index_size,
+                                                    min_index,
+                                                    max_index,
+                                                    gen_prim, start, count, bias );
+   }
+   else {
+      struct pipe_buffer *gen_buf = NULL;
+
+      /* Need to allocate a new index buffer and run the translate
+       * func to populate it.  Could potentially cache this translated
+       * index buffer with the original to avoid future
+       * re-translations.  Not much point if we're just accelerating
+       * GL though, as index buffers are typically used only once
+       * there.
+       */
+      ret = translate_indices( hwtnl,
+                               index_buffer,
+                               start * index_size,
+                               gen_nr,
+                               gen_size,
+                               gen_func,
+                               &gen_buf );
+      if (ret)
+         goto done;
+
+      ret = svga_hwtnl_simple_draw_range_elements( hwtnl,
+                                                   gen_buf,
+                                                   gen_size,
+                                                   min_index,
+                                                   max_index,
+                                                   gen_prim,
+                                                   0,
+                                                   gen_nr,
+                                                   bias );
+      if (ret)
+         goto done;
+
+   done:
+      if (gen_buf)
+         pipe_buffer_reference( &gen_buf, NULL );
+
+      return ret;
+   }
+}
+
+
+
+
+
diff --git a/src/gallium/drivers/svga/svga_draw_private.h b/src/gallium/drivers/svga/svga_draw_private.h
new file mode 100644
index 0000000000..9aa40e1664
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_draw_private.h
@@ -0,0 +1,158 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_DRAW_H_
+#define SVGA_DRAW_H_
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_defines.h"
+#include "indices/u_indices.h"
+#include "svga_hw_reg.h"
+#include "svga3d_shaderdefs.h"
+
+struct svga_context;
+struct u_upload_mgr;
+
+/* Should include polygon?
+ */
+static const unsigned svga_hw_prims = 
+   ((1 << PIPE_PRIM_POINTS) |
+    (1 << PIPE_PRIM_LINES) |
+    (1 << PIPE_PRIM_LINE_STRIP) |
+    (1 << PIPE_PRIM_TRIANGLES) |
+    (1 << PIPE_PRIM_TRIANGLE_STRIP) |
+    (1 << PIPE_PRIM_TRIANGLE_FAN));
+
+
+static INLINE unsigned svga_translate_prim(unsigned mode, 
+                                           unsigned count,
+                                           unsigned *out_count)
+{
+   switch (mode) {
+   case PIPE_PRIM_POINTS:
+      *out_count = count;
+      return SVGA3D_PRIMITIVE_POINTLIST;
+
+   case PIPE_PRIM_LINES:
+      *out_count = count / 2;
+      return SVGA3D_PRIMITIVE_LINELIST; 
+
+   case PIPE_PRIM_LINE_STRIP:
+      *out_count = count - 1;
+      return SVGA3D_PRIMITIVE_LINESTRIP; 
+
+   case PIPE_PRIM_TRIANGLES:
+      *out_count = count / 3;
+      return SVGA3D_PRIMITIVE_TRIANGLELIST; 
+
+   case PIPE_PRIM_TRIANGLE_STRIP:
+      *out_count = count - 2;
+      return SVGA3D_PRIMITIVE_TRIANGLESTRIP; 
+
+   case PIPE_PRIM_TRIANGLE_FAN:
+      *out_count = count - 2;
+      return SVGA3D_PRIMITIVE_TRIANGLEFAN; 
+
+   default:
+      assert(0);
+      *out_count = 0;
+      return 0;
+   }
+}
+
+
+struct index_cache {
+   u_generate_func generate;
+   unsigned gen_nr;
+
+   /* If non-null, this buffer is filled by calling 
+    *   generate(nr, map(buffer))
+    */
+   struct pipe_buffer *buffer;
+};
+
+#define QSZ 32
+
+struct draw_cmd {
+   struct svga_winsys_context *swc;
+
+   SVGA3dVertexDecl vdecl[SVGA3D_INPUTREG_MAX];
+   struct pipe_buffer *vdecl_vb[SVGA3D_INPUTREG_MAX];
+   unsigned vdecl_count;
+
+   SVGA3dPrimitiveRange prim[QSZ];
+   struct pipe_buffer *prim_ib[QSZ];
+   unsigned prim_count;
+   unsigned min_index[QSZ];
+   unsigned max_index[QSZ];
+};
+
+#define IDX_CACHE_MAX  8
+
+struct svga_hwtnl {
+   struct svga_context *svga;
+   struct u_upload_mgr *upload_ib;
+   
+   /* Flatshade information:
+    */
+   unsigned api_pv;
+   unsigned hw_pv;
+   unsigned api_fillmode;
+
+   /* Cache the results of running a particular generate func on each
+    * primitive type.
+    */
+   struct index_cache index_cache[PIPE_PRIM_MAX][IDX_CACHE_MAX];
+
+   /* Try to build the maximal draw command packet before emitting:
+    */
+   struct draw_cmd cmd;
+};
+
+
+
+/***********************************************************************
+ * Internal functions
+ */
+enum pipe_error 
+svga_hwtnl_prim( struct svga_hwtnl *hwtnl,
+                 const SVGA3dPrimitiveRange *range,
+                 unsigned min_index,
+                 unsigned max_index,
+                 struct pipe_buffer *ib );
+
+enum pipe_error
+svga_hwtnl_simple_draw_range_elements( struct svga_hwtnl *hwtnl,
+                                       struct pipe_buffer *indexBuffer,
+                                       unsigned index_size,
+                                       unsigned min_index,
+                                       unsigned max_index,
+                                       unsigned prim, 
+                                       unsigned start,
+                                       unsigned count,
+                                       unsigned bias );
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_hw_reg.h b/src/gallium/drivers/svga/svga_hw_reg.h
new file mode 100644
index 0000000000..183f4b918e
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_hw_reg.h
@@ -0,0 +1,42 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_HW_REG_H
+#define SVGA_HW_REG_H
+
+#include "pipe/p_compiler.h"
+
+#if defined(PIPE_CC_GCC)
+#ifndef HAVE_STDINT_H
+#define HAVE_STDINT_H
+#endif
+#endif
+
+#include "svga_types.h"
+
+#include "svga3d_reg.h"
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_pipe_blend.c b/src/gallium/drivers/svga/svga_pipe_blend.c
new file mode 100644
index 0000000000..855d228755
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_blend.c
@@ -0,0 +1,246 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+
+#include "svga_hw_reg.h"
+
+
+static INLINE unsigned
+svga_translate_blend_factor(unsigned factor)
+{
+   switch (factor) {
+   case PIPE_BLENDFACTOR_ZERO:            return SVGA3D_BLENDOP_ZERO;
+   case PIPE_BLENDFACTOR_SRC_ALPHA:       return SVGA3D_BLENDOP_SRCALPHA;
+   case PIPE_BLENDFACTOR_ONE:             return SVGA3D_BLENDOP_ONE;
+   case PIPE_BLENDFACTOR_SRC_COLOR:       return SVGA3D_BLENDOP_SRCCOLOR;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:   return SVGA3D_BLENDOP_INVSRCCOLOR;
+   case PIPE_BLENDFACTOR_DST_COLOR:       return SVGA3D_BLENDOP_DESTCOLOR;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:   return SVGA3D_BLENDOP_INVDESTCOLOR;
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:   return SVGA3D_BLENDOP_INVSRCALPHA;
+   case PIPE_BLENDFACTOR_DST_ALPHA:       return SVGA3D_BLENDOP_DESTALPHA;
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:   return SVGA3D_BLENDOP_INVDESTALPHA;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: return SVGA3D_BLENDOP_SRCALPHASAT;
+   case PIPE_BLENDFACTOR_CONST_COLOR:     return SVGA3D_BLENDOP_BLENDFACTOR;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR: return SVGA3D_BLENDOP_INVBLENDFACTOR;
+   case PIPE_BLENDFACTOR_CONST_ALPHA:     return SVGA3D_BLENDOP_BLENDFACTOR; /* ? */
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA: return SVGA3D_BLENDOP_INVBLENDFACTOR; /* ? */
+   default:
+      assert(0);
+      return SVGA3D_BLENDOP_ZERO;
+   }
+}
+
+static INLINE unsigned
+svga_translate_blend_func(unsigned mode)
+{
+   switch (mode) {
+   case PIPE_BLEND_ADD:              return SVGA3D_BLENDEQ_ADD;
+   case PIPE_BLEND_SUBTRACT:         return SVGA3D_BLENDEQ_SUBTRACT;
+   case PIPE_BLEND_REVERSE_SUBTRACT: return SVGA3D_BLENDEQ_REVSUBTRACT;
+   case PIPE_BLEND_MIN:              return SVGA3D_BLENDEQ_MINIMUM;
+   case PIPE_BLEND_MAX:              return SVGA3D_BLENDEQ_MAXIMUM;
+   default:
+      assert(0);
+      return SVGA3D_BLENDEQ_ADD;
+   }
+}
+
+
+static void *
+svga_create_blend_state(struct pipe_context *pipe,
+                        const struct pipe_blend_state *templ)
+{
+   struct svga_blend_state *blend = CALLOC_STRUCT( svga_blend_state );
+   unsigned i;
+
+ 
+   /* Fill in the per-rendertarget blend state.  We currently only
+    * have one rendertarget.
+    */
+   for (i = 0; i < 1; i++) {
+      /* No way to set this in SVGA3D, and no way to correctly implement it on
+       * top of D3D9 API.  Instead we try to simulate with various blend modes.
+       */
+      if (templ->logicop_enable) {
+         switch (templ->logicop_func) {
+         case PIPE_LOGICOP_XOR:
+            blend->need_white_fragments = TRUE;
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_ONE;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_ONE;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_SUBTRACT;
+            break;
+         case PIPE_LOGICOP_CLEAR:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_ZERO;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_ZERO;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MINIMUM;
+            break;
+         case PIPE_LOGICOP_COPY:
+            blend->rt[i].blend_enable = FALSE;
+            break;
+         case PIPE_LOGICOP_COPY_INVERTED:
+            blend->rt[i].blend_enable   = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_INVSRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_ZERO;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_ADD;
+            break;
+         case PIPE_LOGICOP_NOOP:
+            blend->rt[i].blend_enable   = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_ZERO;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_DESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_ADD;
+            break;
+         case PIPE_LOGICOP_SET:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_ONE;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_ONE;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MAXIMUM;
+            break;
+         case PIPE_LOGICOP_INVERT:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_INVSRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_ZERO;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_ADD;
+            break;
+         case PIPE_LOGICOP_AND:
+            /* Approximate with minimum - works for the 0 & anything case: */
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_SRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_DESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MINIMUM;
+            break;
+         case PIPE_LOGICOP_AND_REVERSE:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_SRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_INVDESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MINIMUM;
+            break;
+         case PIPE_LOGICOP_AND_INVERTED:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_INVSRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_DESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MINIMUM;
+            break;
+         case PIPE_LOGICOP_OR:
+            /* Approximate with maximum - works for the 1 | anything case: */
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_SRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_DESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MAXIMUM;
+            break;
+         case PIPE_LOGICOP_OR_REVERSE:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_SRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_INVDESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MAXIMUM;
+            break;
+         case PIPE_LOGICOP_OR_INVERTED:
+            blend->rt[i].blend_enable = TRUE;
+            blend->rt[i].srcblend       = SVGA3D_BLENDOP_INVSRCCOLOR;
+            blend->rt[i].dstblend       = SVGA3D_BLENDOP_DESTCOLOR;
+            blend->rt[i].blendeq        = SVGA3D_BLENDEQ_MAXIMUM;
+            break;
+         case PIPE_LOGICOP_NAND:
+         case PIPE_LOGICOP_NOR:
+         case PIPE_LOGICOP_EQUIV:
+            /* Fill these in with plausible values */
+            blend->rt[i].blend_enable = FALSE;
+            break;
+         default:
+            assert(0);
+            break;
+         }
+      }
+      else {
+         blend->rt[i].blend_enable   = templ->blend_enable;
+
+         if (templ->blend_enable) {
+            blend->rt[i].srcblend       = svga_translate_blend_factor(templ->rgb_src_factor);
+            blend->rt[i].dstblend       = svga_translate_blend_factor(templ->rgb_dst_factor);
+            blend->rt[i].blendeq        = svga_translate_blend_func(templ->rgb_func);
+            blend->rt[i].srcblend_alpha = svga_translate_blend_factor(templ->alpha_src_factor);
+            blend->rt[i].dstblend_alpha = svga_translate_blend_factor(templ->alpha_dst_factor);
+            blend->rt[i].blendeq_alpha  = svga_translate_blend_func(templ->alpha_func);
+
+            if (blend->rt[i].srcblend_alpha != blend->rt[i].srcblend ||
+                blend->rt[i].dstblend_alpha != blend->rt[i].dstblend ||
+                blend->rt[i].blendeq_alpha  != blend->rt[i].blendeq)
+            {
+               blend->rt[i].separate_alpha_blend_enable = TRUE;
+            }
+         }
+      }
+
+      blend->rt[i].writemask = templ->colormask;
+   }
+
+   return blend;
+}
+
+static void svga_bind_blend_state(struct pipe_context *pipe,
+                                  void *blend)
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.blend = (struct svga_blend_state*)blend;
+   svga->dirty |= SVGA_NEW_BLEND;
+}
+
+
+static void svga_delete_blend_state(struct pipe_context *pipe, void *blend)
+{
+   FREE(blend);
+}
+
+static void svga_set_blend_color( struct pipe_context *pipe,
+                                  const struct pipe_blend_color *blend_color )
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.blend_color = *blend_color;
+
+   svga->dirty |= SVGA_NEW_BLEND;
+}
+
+
+void svga_init_blend_functions( struct svga_context *svga )
+{
+   svga->pipe.create_blend_state = svga_create_blend_state;
+   svga->pipe.bind_blend_state = svga_bind_blend_state;
+   svga->pipe.delete_blend_state = svga_delete_blend_state;
+
+   svga->pipe.set_blend_color = svga_set_blend_color;
+}
+
+
+
diff --git a/src/gallium/drivers/svga/svga_pipe_blit.c b/src/gallium/drivers/svga/svga_pipe_blit.c
new file mode 100644
index 0000000000..5a4a8c0f5f
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_blit.c
@@ -0,0 +1,84 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_screen_texture.h"
+#include "svga_context.h"
+#include "svga_cmd.h"
+
+#define FILE_DEBUG_FLAG DEBUG_BLIT
+
+
+static void svga_surface_copy(struct pipe_context *pipe,
+                              struct pipe_surface *dest,
+                              unsigned destx, unsigned desty,
+                              struct pipe_surface *src,
+                              unsigned srcx, unsigned srcy,
+                              unsigned width, unsigned height)
+{
+   struct svga_context *svga = svga_context(pipe);
+   SVGA3dCopyBox *box;
+   enum pipe_error ret;
+
+   svga_hwtnl_flush_retry( svga );
+
+   ret = SVGA3D_BeginSurfaceCopy(svga->swc,
+                                 src,
+                                 dest,
+                                 &box,
+                                 1);
+   if(ret != PIPE_OK) {
+
+      svga_context_flush(svga, NULL);
+
+      ret = SVGA3D_BeginSurfaceCopy(svga->swc,
+                                    src,
+                                    dest,
+                                    &box,
+                                    1);
+      assert(ret == PIPE_OK);
+   }
+
+   box->x = destx;
+   box->y = desty;
+   box->z = 0;
+   box->w = width;
+   box->h = height;
+   box->d = 1;
+   box->srcx = srcx;
+   box->srcy = srcy;
+   box->srcz = 0;
+
+   SVGA_FIFOCommitAll(svga->swc);
+
+   svga_surface(dest)->dirty = TRUE;
+   svga_propagate_surface(pipe, dest);
+}
+
+
+void
+svga_init_blit_functions(struct svga_context *svga)
+{
+   svga->pipe.surface_copy = svga_surface_copy;
+}
diff --git a/src/gallium/drivers/svga/svga_pipe_clear.c b/src/gallium/drivers/svga/svga_pipe_clear.c
new file mode 100644
index 0000000000..8977d26541
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_clear.c
@@ -0,0 +1,119 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_defines.h"
+#include "util/u_pack_color.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+
+
+static enum pipe_error
+try_clear(struct svga_context *svga, 
+          unsigned buffers,
+          const float *rgba,
+          double depth,
+          unsigned stencil)
+{
+   int ret = PIPE_OK;
+   SVGA3dRect rect = { 0, 0, 0, 0 };
+   boolean restore_viewport = FALSE;
+   SVGA3dClearFlag flags = 0;
+   struct pipe_framebuffer_state *fb = &svga->curr.framebuffer;
+   unsigned color = 0;
+
+   ret = svga_update_state(svga, SVGA_STATE_HW_CLEAR);
+   if (ret)
+      return ret;
+
+   if ((buffers & PIPE_CLEAR_COLOR) && fb->cbufs[0]) {
+      flags |= SVGA3D_CLEAR_COLOR;
+      util_pack_color(rgba, PIPE_FORMAT_A8R8G8B8_UNORM, &color);
+
+      rect.w = fb->cbufs[0]->width;
+      rect.h = fb->cbufs[0]->height;
+   }
+
+   if ((buffers & PIPE_CLEAR_DEPTHSTENCIL) && fb->zsbuf) {
+      flags |= SVGA3D_CLEAR_DEPTH;
+
+      if (svga->curr.framebuffer.zsbuf->format == PIPE_FORMAT_Z24S8_UNORM)
+         flags |= SVGA3D_CLEAR_STENCIL;
+
+      rect.w = MAX2(rect.w, fb->zsbuf->width);
+      rect.h = MAX2(rect.h, fb->zsbuf->height);
+   }
+
+   if (memcmp(&rect, &svga->state.hw_clear.viewport, sizeof(rect)) != 0) {
+      restore_viewport = TRUE;
+      ret = SVGA3D_SetViewport(svga->swc, &rect);
+      if (ret)
+         return ret;
+   }
+
+   ret = SVGA3D_ClearRect(svga->swc, flags, color, depth, stencil,
+                          rect.x, rect.y, rect.w, rect.h);
+   if (ret != PIPE_OK)
+      return ret;
+
+   if (restore_viewport) {
+      memcpy(&rect, &svga->state.hw_clear.viewport, sizeof rect);
+      ret = SVGA3D_SetViewport(svga->swc, &rect);
+   }
+   
+   return ret;
+}
+
+/**
+ * Clear the given surface to the specified value.
+ * No masking, no scissor (clear entire buffer).
+ */
+void
+svga_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba,
+	   double depth, unsigned stencil)
+{
+   struct svga_context *svga = svga_context( pipe );
+   int ret;
+
+   ret = try_clear( svga, buffers, rgba, depth, stencil );
+
+   if (ret == PIPE_ERROR_OUT_OF_MEMORY) {
+      /* Flush command buffer and retry:
+       */
+      svga_context_flush( svga, NULL );
+
+      ret = try_clear( svga, buffers, rgba, depth, stencil );
+   }
+
+   /*
+    * Mark target surfaces as dirty
+    * TODO Mark only cleared surfaces.
+    */
+   svga_mark_surfaces_dirty(svga);
+
+   assert (ret == PIPE_OK);
+}
diff --git a/src/gallium/drivers/svga/svga_pipe_constants.c b/src/gallium/drivers/svga/svga_pipe_constants.c
new file mode 100644
index 0000000000..10e7a12189
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_constants.c
@@ -0,0 +1,74 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "tgsi/tgsi_parse.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_hw_reg.h"
+#include "svga_cmd.h"
+
+/***********************************************************************
+ * Constant buffers 
+ */
+
+struct svga_constbuf 
+{
+   unsigned type;
+   float (*data)[4];
+   unsigned count;
+};
+
+
+
+static void svga_set_constant_buffer(struct pipe_context *pipe,
+                                     uint shader, uint index,
+                                     const struct pipe_constant_buffer *buf)
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   assert(shader < PIPE_SHADER_TYPES);
+   assert(index == 0);
+
+   pipe_buffer_reference( &svga->curr.cb[shader],
+                          buf->buffer );
+
+   if (shader == PIPE_SHADER_FRAGMENT)
+      svga->dirty |= SVGA_NEW_FS_CONST_BUFFER;
+   else
+      svga->dirty |= SVGA_NEW_VS_CONST_BUFFER;
+}
+
+
+
+void svga_init_constbuffer_functions( struct svga_context *svga )
+{
+   svga->pipe.set_constant_buffer = svga_set_constant_buffer;
+}
+
diff --git a/src/gallium/drivers/svga/svga_pipe_depthstencil.c b/src/gallium/drivers/svga/svga_pipe_depthstencil.c
new file mode 100644
index 0000000000..df636c08a0
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_depthstencil.c
@@ -0,0 +1,153 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_hw_reg.h"
+
+
+static INLINE unsigned
+svga_translate_compare_func(unsigned func)
+{
+   switch (func) {
+   case PIPE_FUNC_NEVER:     return SVGA3D_CMP_NEVER;
+   case PIPE_FUNC_LESS:      return SVGA3D_CMP_LESS;
+   case PIPE_FUNC_LEQUAL:    return SVGA3D_CMP_LESSEQUAL;
+   case PIPE_FUNC_GREATER:   return SVGA3D_CMP_GREATER;
+   case PIPE_FUNC_GEQUAL:    return SVGA3D_CMP_GREATEREQUAL;
+   case PIPE_FUNC_NOTEQUAL:  return SVGA3D_CMP_NOTEQUAL;
+   case PIPE_FUNC_EQUAL:     return SVGA3D_CMP_EQUAL;
+   case PIPE_FUNC_ALWAYS:    return SVGA3D_CMP_ALWAYS;
+   default:
+      assert(0);
+      return SVGA3D_CMP_ALWAYS;
+   }
+}
+
+static INLINE unsigned
+svga_translate_stencil_op(unsigned op)
+{
+   switch (op) {
+   case PIPE_STENCIL_OP_KEEP:      return SVGA3D_STENCILOP_KEEP;
+   case PIPE_STENCIL_OP_ZERO:      return SVGA3D_STENCILOP_ZERO;
+   case PIPE_STENCIL_OP_REPLACE:   return SVGA3D_STENCILOP_REPLACE;
+   case PIPE_STENCIL_OP_INCR:      return SVGA3D_STENCILOP_INCR;
+   case PIPE_STENCIL_OP_DECR:      return SVGA3D_STENCILOP_DECR;
+   case PIPE_STENCIL_OP_INCR_WRAP: return SVGA3D_STENCILOP_INCRSAT; /* incorrect? */
+   case PIPE_STENCIL_OP_DECR_WRAP: return SVGA3D_STENCILOP_DECRSAT; /* incorrect? */
+   case PIPE_STENCIL_OP_INVERT:    return SVGA3D_STENCILOP_INVERT;
+   default:
+      assert(0);
+      return SVGA3D_STENCILOP_KEEP;
+   }
+}
+
+
+static void *
+svga_create_depth_stencil_state(struct pipe_context *pipe,
+				const struct pipe_depth_stencil_alpha_state *templ)
+{
+   struct svga_depth_stencil_state *ds = CALLOC_STRUCT( svga_depth_stencil_state );
+
+   /* Don't try to figure out CW/CCW correspondence with
+    * stencil[0]/[1] at this point.  Presumably this can change as
+    * back/front face are modified.
+    */
+   ds->stencil[0].enabled = templ->stencil[0].enabled;
+   if (ds->stencil[0].enabled) {
+      ds->stencil[0].func  = svga_translate_compare_func(templ->stencil[0].func);
+      ds->stencil[0].fail  = svga_translate_stencil_op(templ->stencil[0].fail_op);
+      ds->stencil[0].zfail = svga_translate_stencil_op(templ->stencil[0].zfail_op);
+      ds->stencil[0].pass  = svga_translate_stencil_op(templ->stencil[0].zpass_op);
+      
+      /* SVGA3D has one ref/mask/writemask triple shared between front &
+       * back face stencil.  We really need two:
+       */
+      ds->stencil_ref       = templ->stencil[0].ref_value & 0xff;
+      ds->stencil_mask      = templ->stencil[0].valuemask & 0xff;
+      ds->stencil_writemask = templ->stencil[0].writemask & 0xff;
+   }
+
+
+   ds->stencil[1].enabled = templ->stencil[1].enabled;
+   if (templ->stencil[1].enabled) {
+      ds->stencil[1].func   = svga_translate_compare_func(templ->stencil[1].func);
+      ds->stencil[1].fail   = svga_translate_stencil_op(templ->stencil[1].fail_op);
+      ds->stencil[1].zfail  = svga_translate_stencil_op(templ->stencil[1].zfail_op);
+      ds->stencil[1].pass   = svga_translate_stencil_op(templ->stencil[1].zpass_op);
+
+      ds->stencil_ref       = templ->stencil[1].ref_value & 0xff;
+      ds->stencil_mask      = templ->stencil[1].valuemask & 0xff;
+      ds->stencil_writemask = templ->stencil[1].writemask & 0xff;
+   }
+
+
+   ds->zenable = templ->depth.enabled;
+   if (ds->zenable) {
+      ds->zfunc = svga_translate_compare_func(templ->depth.func);
+      ds->zwriteenable = templ->depth.writemask;
+   }
+
+   ds->alphatestenable = templ->alpha.enabled;
+   if (ds->alphatestenable) {
+      ds->alphafunc = svga_translate_compare_func(templ->alpha.func);
+      ds->alpharef = templ->alpha.ref_value;
+   }
+
+   return ds;
+}
+
+static void svga_bind_depth_stencil_state(struct pipe_context *pipe,
+                                          void *depth_stencil)
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.depth = (const struct svga_depth_stencil_state *)depth_stencil;
+   svga->dirty |= SVGA_NEW_DEPTH_STENCIL;
+}
+
+static void svga_delete_depth_stencil_state(struct pipe_context *pipe,
+                                            void *depth_stencil)
+{
+   FREE(depth_stencil);
+}
+
+
+
+void svga_init_depth_stencil_functions( struct svga_context *svga )
+{
+   svga->pipe.create_depth_stencil_alpha_state = svga_create_depth_stencil_state;
+   svga->pipe.bind_depth_stencil_alpha_state = svga_bind_depth_stencil_state;
+   svga->pipe.delete_depth_stencil_alpha_state = svga_delete_depth_stencil_state;
+}
+
+
+
+
diff --git a/src/gallium/drivers/svga/svga_pipe_draw.c b/src/gallium/drivers/svga/svga_pipe_draw.c
new file mode 100644
index 0000000000..71a552862e
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_draw.c
@@ -0,0 +1,261 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_inlines.h"
+#include "util/u_prim.h"
+#include "util/u_time.h"
+#include "indices/u_indices.h"
+
+#include "svga_hw_reg.h"
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_winsys.h"
+#include "svga_draw.h"
+#include "svga_state.h"
+#include "svga_swtnl.h"
+#include "svga_debug.h"
+
+
+
+static enum pipe_error
+retry_draw_range_elements( struct svga_context *svga,
+                           struct pipe_buffer *index_buffer,
+                           unsigned index_size,
+                           unsigned min_index,
+                           unsigned max_index,
+                           unsigned prim, 
+                           unsigned start, 
+                           unsigned count,
+                           boolean do_retry )
+{
+   enum pipe_error ret = 0;
+
+   svga_hwtnl_set_unfilled( svga->hwtnl,
+                            svga->curr.rast->hw_unfilled );
+
+   svga_hwtnl_set_flatshade( svga->hwtnl,
+                             svga->curr.rast->templ.flatshade,
+                             svga->curr.rast->templ.flatshade_first );
+
+
+   ret = svga_update_state( svga, SVGA_STATE_HW_DRAW );
+   if (ret)
+      goto retry;
+
+   ret = svga_hwtnl_draw_range_elements( svga->hwtnl,
+                                         index_buffer, index_size,
+                                         min_index, max_index,
+                                         prim, start, count, 0 );
+   if (ret)
+      goto retry;
+
+   if (svga->curr.any_user_vertex_buffers) {
+      ret = svga_hwtnl_flush( svga->hwtnl );
+      if (ret)
+         goto retry;
+   }
+
+   return PIPE_OK;
+
+retry:
+   svga_context_flush( svga, NULL );
+
+   if (do_retry)
+   {
+      return retry_draw_range_elements( svga,
+                                        index_buffer, index_size,
+                                        min_index, max_index,
+                                        prim, start, count,
+                                        FALSE );
+   }
+
+   return ret;
+}
+
+
+static enum pipe_error
+retry_draw_arrays( struct svga_context *svga,
+                   unsigned prim, 
+                   unsigned start, 
+                   unsigned count,
+                   boolean do_retry )
+{
+   enum pipe_error ret;
+
+   svga_hwtnl_set_unfilled( svga->hwtnl,
+                            svga->curr.rast->hw_unfilled );
+
+   svga_hwtnl_set_flatshade( svga->hwtnl,
+                             svga->curr.rast->templ.flatshade,
+                             svga->curr.rast->templ.flatshade_first );
+
+   ret = svga_update_state( svga, SVGA_STATE_HW_DRAW );
+   if (ret)
+      goto retry;
+
+   ret = svga_hwtnl_draw_arrays( svga->hwtnl, prim,
+                                 start, count );
+   if (ret)
+      goto retry;
+
+   if (svga->curr.any_user_vertex_buffers) {
+      ret = svga_hwtnl_flush( svga->hwtnl );
+      if (ret)
+         goto retry;
+   }
+
+   return 0;
+
+retry:
+   if (ret == PIPE_ERROR_OUT_OF_MEMORY && do_retry) 
+   {
+      svga_context_flush( svga, NULL );
+
+      return retry_draw_arrays( svga,
+                                prim,
+                                start,
+                                count,
+                                FALSE );
+   }
+
+   return ret;
+}
+
+
+
+
+
+static boolean
+svga_draw_range_elements( struct pipe_context *pipe,
+                          struct pipe_buffer *index_buffer,
+                          unsigned index_size,
+                          unsigned min_index,
+                          unsigned max_index,
+                          unsigned prim, unsigned start, unsigned count)
+{
+   struct svga_context *svga = svga_context( pipe );
+   unsigned reduced_prim = u_reduced_prim(prim);
+   enum pipe_error ret = 0;
+
+   if (!u_trim_pipe_prim( prim, &count ))
+      return TRUE;
+
+   /*
+    * Mark currently bound target surfaces as dirty
+    * doesn't really matter if it is done before drawing.
+    *
+    * TODO If we ever normaly return something other then
+    * true we should not mark it as dirty then.
+    */
+   svga_mark_surfaces_dirty(svga_context(pipe));
+
+   if (svga->curr.reduced_prim != reduced_prim) {
+      svga->curr.reduced_prim = reduced_prim;
+      svga->dirty |= SVGA_NEW_REDUCED_PRIMITIVE;
+   }
+   
+   svga_update_state_retry( svga, SVGA_STATE_NEED_SWTNL );
+
+#ifdef DEBUG
+   if (svga->curr.vs->base.id == svga->debug.disable_shader ||
+       svga->curr.fs->base.id == svga->debug.disable_shader)
+      return 0;
+#endif
+
+   if (svga->state.sw.need_swtnl)
+   {
+      ret = svga_swtnl_draw_range_elements( svga, 
+                                            index_buffer, 
+                                            index_size,
+                                            min_index, max_index,
+                                            prim,
+                                            start, count );
+   }
+   else {
+      if (index_buffer) {
+         ret = retry_draw_range_elements( svga,
+                                          index_buffer,
+                                          index_size,
+                                          min_index,
+                                          max_index,
+                                          prim,
+                                          start,
+                                          count,
+                                          TRUE );
+      }
+      else {
+         ret = retry_draw_arrays( svga, 
+                                  prim, 
+                                  start, 
+                                  count,
+                                  TRUE );
+      }
+   }
+
+   if (SVGA_DEBUG & DEBUG_FLUSH) {
+      static unsigned id;
+      debug_printf("%s %d\n", __FUNCTION__, id++);
+      if (id > 1300)
+         util_time_sleep( 2000 );
+
+      svga_hwtnl_flush_retry( svga );
+      svga_context_flush(svga, NULL);
+   }
+
+   return ret == PIPE_OK;
+}
+
+
+static boolean 
+svga_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *index_buffer,
+                    unsigned index_size,
+                    unsigned prim, unsigned start, unsigned count)
+{
+   return svga_draw_range_elements( pipe, index_buffer,
+                                    index_size,
+                                    0, 0xffffffff,
+                                    prim, start, count );
+}
+
+static boolean 
+svga_draw_arrays( struct pipe_context *pipe,
+                  unsigned prim, unsigned start, unsigned count)
+{
+   return svga_draw_range_elements(pipe, NULL, 0, 
+                                   start, start + count - 1, 
+                                   prim, 
+                                   start, count);
+}
+
+
+void svga_init_draw_functions( struct svga_context *svga )
+{
+   svga->pipe.draw_arrays = svga_draw_arrays;
+   svga->pipe.draw_elements = svga_draw_elements;
+   svga->pipe.draw_range_elements = svga_draw_range_elements;
+}
diff --git a/src/gallium/drivers/svga/svga_pipe_flush.c b/src/gallium/drivers/svga/svga_pipe_flush.c
new file mode 100644
index 0000000000..942366de72
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_flush.c
@@ -0,0 +1,68 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_defines.h"
+#include "svga_screen.h"
+#include "svga_screen_texture.h"
+#include "svga_context.h"
+#include "svga_winsys.h"
+#include "svga_draw.h"
+#include "svga_debug.h"
+
+#include "svga_hw_reg.h"
+
+
+
+
+static void svga_flush( struct pipe_context *pipe,
+                        unsigned flags,
+                        struct pipe_fence_handle **fence )
+{
+   struct svga_context *svga = svga_context(pipe);
+   int i;
+
+   /* Emit buffered drawing commands.
+    */
+   svga_hwtnl_flush_retry( svga );
+
+   /* Emit back-copy from render target view to texture.
+    */
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      if (svga->curr.framebuffer.cbufs[i])
+         svga_propagate_surface(pipe, svga->curr.framebuffer.cbufs[i]);
+   }
+   if (svga->curr.framebuffer.zsbuf)
+      svga_propagate_surface(pipe, svga->curr.framebuffer.zsbuf);
+
+   /* Flush command queue.
+    */
+   svga_context_flush(svga, fence);
+}
+
+
+void svga_init_flush_functions( struct svga_context *svga )
+{
+   svga->pipe.flush = svga_flush;
+}
diff --git a/src/gallium/drivers/svga/svga_pipe_fs.c b/src/gallium/drivers/svga/svga_pipe_fs.c
new file mode 100644
index 0000000000..e3be840d92
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_fs.c
@@ -0,0 +1,124 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_text.h"
+
+#include "svga_screen.h"
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_tgsi.h"
+#include "svga_hw_reg.h"
+#include "svga_cmd.h"
+#include "svga_draw.h"
+#include "svga_debug.h"
+
+
+/***********************************************************************
+ * Fragment shaders 
+ */
+
+static void *
+svga_create_fs_state(struct pipe_context *pipe,
+                     const struct pipe_shader_state *templ)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_screen *svgascreen = svga_screen(pipe->screen);
+   struct svga_fragment_shader *fs;
+
+   fs = CALLOC_STRUCT(svga_fragment_shader);
+   if (!fs)
+      return NULL;
+
+   fs->base.tokens = tgsi_dup_tokens(templ->tokens);
+
+   /* Collect basic info that we'll need later:
+    */
+   tgsi_scan_shader(fs->base.tokens, &fs->base.info);
+
+   fs->base.id = svga->debug.shader_id++;
+   fs->base.use_sm30 = svgascreen->use_ps30;
+   
+   if (SVGA_DEBUG & DEBUG_TGSI || 0) {
+      debug_printf("%s id: %u, inputs: %u, outputs: %u\n",
+                   __FUNCTION__, fs->base.id,
+                   fs->base.info.num_inputs, fs->base.info.num_outputs);
+   }
+
+   return fs;
+}
+
+static void
+svga_bind_fs_state(struct pipe_context *pipe, void *shader)
+{
+   struct svga_fragment_shader *fs = (struct svga_fragment_shader *) shader;
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.fs = fs;
+   svga->dirty |= SVGA_NEW_FS;
+}
+
+static
+void svga_delete_fs_state(struct pipe_context *pipe, void *shader)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_fragment_shader *fs = (struct svga_fragment_shader *) shader;
+   struct svga_shader_result *result, *tmp;
+   enum pipe_error ret;
+
+   svga_hwtnl_flush_retry( svga );
+
+   for (result = fs->base.results; result; result = tmp ) {
+      tmp = result->next;
+
+      ret = SVGA3D_DestroyShader(svga->swc, 
+                                 result->id,
+                                 SVGA3D_SHADERTYPE_PS );
+      if(ret != PIPE_OK) {
+         svga_context_flush(svga, NULL);
+         ret = SVGA3D_DestroyShader(svga->swc, 
+                                    result->id,
+                                    SVGA3D_SHADERTYPE_PS );
+         assert(ret == PIPE_OK);
+      }
+
+      svga_destroy_shader_result( result );
+   }
+
+   FREE((void *)fs->base.tokens);
+   FREE(fs);
+}
+
+
+void svga_init_fs_functions( struct svga_context *svga )
+{
+   svga->pipe.create_fs_state = svga_create_fs_state;
+   svga->pipe.bind_fs_state = svga_bind_fs_state;
+   svga->pipe.delete_fs_state = svga_delete_fs_state;
+}
+
diff --git a/src/gallium/drivers/svga/svga_pipe_misc.c b/src/gallium/drivers/svga/svga_pipe_misc.c
new file mode 100644
index 0000000000..58cb1e6e23
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_misc.c
@@ -0,0 +1,187 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "svga_context.h"
+#include "svga_screen_texture.h"
+#include "svga_state.h"
+#include "svga_winsys.h"
+
+#include "svga_hw_reg.h"
+
+
+
+
+static void svga_set_scissor_state( struct pipe_context *pipe,
+                                 const struct pipe_scissor_state *scissor )
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   memcpy( &svga->curr.scissor, scissor, sizeof(*scissor) );
+   svga->dirty |= SVGA_NEW_SCISSOR;
+}
+
+
+static void svga_set_polygon_stipple( struct pipe_context *pipe,
+                                      const struct pipe_poly_stipple *stipple )
+{
+   /* overridden by the draw module */
+}
+
+
+void svga_cleanup_framebuffer(struct svga_context *svga)
+{
+   struct pipe_framebuffer_state *curr = &svga->curr.framebuffer;
+   struct pipe_framebuffer_state *hw = &svga->state.hw_clear.framebuffer;
+   int i;
+
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      pipe_surface_reference(&curr->cbufs[i], NULL);
+      pipe_surface_reference(&hw->cbufs[i], NULL);
+   }
+
+   pipe_surface_reference(&curr->zsbuf, NULL);
+   pipe_surface_reference(&hw->zsbuf, NULL);
+}
+
+
+#define DEPTH_BIAS_SCALE_FACTOR_D16    ((float)(1<<15))
+#define DEPTH_BIAS_SCALE_FACTOR_D24S8  ((float)(1<<23))
+#define DEPTH_BIAS_SCALE_FACTOR_D32    ((float)(1<<31))
+
+
+static void svga_set_framebuffer_state(struct pipe_context *pipe,
+				       const struct pipe_framebuffer_state *fb)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct pipe_framebuffer_state *dst = &svga->curr.framebuffer;
+   boolean propagate = FALSE;
+   int i;
+
+   dst->width = fb->width;
+   dst->height = fb->height;
+   dst->nr_cbufs = fb->nr_cbufs;
+
+   /* check if we need to propaget any of the target surfaces */
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      if (dst->cbufs[i] && dst->cbufs[i] != fb->cbufs[i])
+         if (svga_surface_needs_propagation(dst->cbufs[i]))
+            propagate = TRUE;
+   }
+
+   if (propagate) {
+      /* make sure that drawing calls comes before propagation calls */
+      svga_hwtnl_flush_retry( svga );
+   
+      for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++)
+         if (dst->cbufs[i] && dst->cbufs[i] != fb->cbufs[i])
+            svga_propagate_surface(pipe, dst->cbufs[i]);
+   }
+
+   /* XXX: Actually the virtual hardware may support rendertargets with
+    * different size, depending on the host API and driver, but since we cannot
+    * know that make no such assumption here. */
+   for(i = 0; i < fb->nr_cbufs; ++i) {
+      if (fb->zsbuf && fb->cbufs[i]) {
+         assert(fb->zsbuf->width == fb->cbufs[i]->width); 
+         assert(fb->zsbuf->height == fb->cbufs[i]->height); 
+      }
+   }
+
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++)
+      pipe_surface_reference(&dst->cbufs[i], fb->cbufs[i]);
+   pipe_surface_reference(&dst->zsbuf, fb->zsbuf);
+
+
+   if (svga->curr.framebuffer.zsbuf)
+   {
+      switch (svga->curr.framebuffer.zsbuf->format) {
+      case PIPE_FORMAT_Z16_UNORM:
+         svga->curr.depthscale = 1.0f / DEPTH_BIAS_SCALE_FACTOR_D16;
+         break;
+      case PIPE_FORMAT_S8Z24_UNORM:
+      case PIPE_FORMAT_X8Z24_UNORM:
+      case PIPE_FORMAT_Z24S8_UNORM:
+      case PIPE_FORMAT_Z24X8_UNORM:
+         svga->curr.depthscale = 1.0f / DEPTH_BIAS_SCALE_FACTOR_D24S8;
+         break;
+      case PIPE_FORMAT_Z32_UNORM:
+         svga->curr.depthscale = 1.0f / DEPTH_BIAS_SCALE_FACTOR_D32;
+         break;
+      case PIPE_FORMAT_Z32_FLOAT:
+         svga->curr.depthscale = 1.0f / ((float)(1<<23));
+         break;
+      default:
+         svga->curr.depthscale = 0.0f;
+         break;
+      }
+   }
+   else {
+      svga->curr.depthscale = 0.0f;
+   }
+
+   svga->dirty |= SVGA_NEW_FRAME_BUFFER;
+}
+
+
+
+static void svga_set_clip_state( struct pipe_context *pipe,
+                                 const struct pipe_clip_state *clip )
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.clip = *clip; /* struct copy */
+
+   svga->dirty |= SVGA_NEW_CLIP;
+}
+
+
+
+/* Called when driver state tracker notices changes to the viewport
+ * matrix:
+ */
+static void svga_set_viewport_state( struct pipe_context *pipe,
+				     const struct pipe_viewport_state *viewport )
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.viewport = *viewport; /* struct copy */
+
+   svga->dirty |= SVGA_NEW_VIEWPORT;
+}
+
+
+
+void svga_init_misc_functions( struct svga_context *svga )
+{
+   svga->pipe.set_scissor_state = svga_set_scissor_state;
+   svga->pipe.set_polygon_stipple = svga_set_polygon_stipple;
+   svga->pipe.set_framebuffer_state = svga_set_framebuffer_state;
+   svga->pipe.set_clip_state = svga_set_clip_state;
+   svga->pipe.set_viewport_state = svga_set_viewport_state;
+}
+
+
diff --git a/src/gallium/drivers/svga/svga_pipe_query.c b/src/gallium/drivers/svga/svga_pipe_query.c
new file mode 100644
index 0000000000..01336b0a2c
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_query.c
@@ -0,0 +1,267 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_state.h"
+#include "pipe/p_context.h"
+#include "util/u_memory.h"
+
+#include "svga_cmd.h"
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_draw.h"
+#include "svga_debug.h"
+
+
+/* Fixme: want a public base class for all pipe structs, even if there
+ * isn't much in them.
+ */
+struct pipe_query {
+   int dummy;
+};
+
+struct svga_query {
+   struct pipe_query base;
+   SVGA3dQueryType type;
+   struct svga_winsys_buffer *hwbuf;
+   volatile SVGA3dQueryResult *queryResult;
+   struct pipe_fence_handle *fence;
+};
+
+/***********************************************************************
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct svga_query *
+svga_query( struct pipe_query *q )
+{
+   return (struct svga_query *)q;
+}
+
+static boolean svga_get_query_result(struct pipe_context *pipe, 
+                                     struct pipe_query *q,
+                                     boolean wait,
+                                     uint64_t *result);
+
+static struct pipe_query *svga_create_query( struct pipe_context *pipe,
+                                             unsigned query_type )
+{
+   struct svga_screen *svgascreen = svga_screen(pipe->screen);
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_query *sq;
+
+   SVGA_DBG(DEBUG_QUERY, "%s\n", __FUNCTION__);
+
+   sq = CALLOC_STRUCT(svga_query);
+   if (!sq)
+      goto no_sq;
+
+   sq->type = SVGA3D_QUERYTYPE_OCCLUSION;
+
+   sq->hwbuf = svga_winsys_buffer_create(svgascreen, 
+                                         1,
+                                         SVGA_BUFFER_USAGE_PINNED,
+                                         sizeof *sq->queryResult);
+   if(!sq->hwbuf)
+      goto no_hwbuf;
+    
+   sq->queryResult = (SVGA3dQueryResult *)sws->buffer_map(sws, 
+                                                          sq->hwbuf, 
+                                                          PIPE_BUFFER_USAGE_CPU_WRITE);
+   if(!sq->queryResult)
+      goto no_query_result;
+
+   sq->queryResult->totalSize = sizeof *sq->queryResult;
+   sq->queryResult->state = SVGA3D_QUERYSTATE_NEW;
+
+   /*
+    * We request the buffer to be pinned and assume it is always mapped.
+    * 
+    * The reason is that we don't want to wait for fences when checking the
+    * query status.
+    */
+   sws->buffer_unmap(sws, sq->hwbuf);
+
+   return &sq->base;
+
+no_query_result:
+   sws->buffer_destroy(sws, sq->hwbuf);
+no_hwbuf:
+   FREE(sq);
+no_sq:
+   return NULL;
+}
+
+static void svga_destroy_query(struct pipe_context *pipe,
+                               struct pipe_query *q)
+{
+   struct svga_screen *svgascreen = svga_screen(pipe->screen);
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_query *sq = svga_query( q );
+
+   SVGA_DBG(DEBUG_QUERY, "%s\n", __FUNCTION__);
+   sws->buffer_destroy(sws, sq->hwbuf);
+   sws->fence_reference(sws, &sq->fence, NULL);
+   FREE(sq);
+}
+
+static void svga_begin_query(struct pipe_context *pipe, 
+                             struct pipe_query *q)
+{
+   struct svga_screen *svgascreen = svga_screen(pipe->screen);
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_context *svga = svga_context( pipe );
+   struct svga_query *sq = svga_query( q );
+   enum pipe_error ret;
+
+   SVGA_DBG(DEBUG_QUERY, "%s\n", __FUNCTION__);
+   
+   assert(!svga->sq);
+
+   /* Need to flush out buffered drawing commands so that they don't
+    * get counted in the query results.
+    */
+   svga_hwtnl_flush_retry(svga);
+   
+   if(sq->queryResult->state == SVGA3D_QUERYSTATE_PENDING) {
+      /* The application doesn't care for the pending query result. We cannot
+       * let go the existing buffer and just get a new one because its storage
+       * may be reused for other purposes and clobbered by the host when it
+       * determines the query result. So the only option here is to wait for
+       * the existing query's result -- not a big deal, given that no sane
+       * application would do this.
+       */
+      uint64_t result;
+
+      svga_get_query_result(pipe, q, TRUE, &result);
+      
+      assert(sq->queryResult->state != SVGA3D_QUERYSTATE_PENDING);
+   }
+   
+   sq->queryResult->state = SVGA3D_QUERYSTATE_NEW;
+   sws->fence_reference(sws, &sq->fence, NULL);
+
+   ret = SVGA3D_BeginQuery(svga->swc, sq->type);
+   if(ret != PIPE_OK) {
+      svga_context_flush(svga, NULL);
+      ret = SVGA3D_BeginQuery(svga->swc, sq->type);
+      assert(ret == PIPE_OK);
+   }
+
+   svga->sq = sq;
+}
+
+static void svga_end_query(struct pipe_context *pipe, 
+                           struct pipe_query *q)
+{
+   struct svga_context *svga = svga_context( pipe );
+   struct svga_query *sq = svga_query( q );
+   enum pipe_error ret;
+
+   SVGA_DBG(DEBUG_QUERY, "%s\n", __FUNCTION__);
+   assert(svga->sq == sq);
+
+   svga_hwtnl_flush_retry(svga);
+   
+   /* Set to PENDING before sending EndQuery. */
+   sq->queryResult->state = SVGA3D_QUERYSTATE_PENDING;
+
+   ret = SVGA3D_EndQuery( svga->swc, sq->type, sq->hwbuf);
+   if(ret != PIPE_OK) {
+      svga_context_flush(svga, NULL);
+      ret = SVGA3D_EndQuery( svga->swc, sq->type, sq->hwbuf);
+      assert(ret == PIPE_OK);
+   }
+   
+   /* TODO: Delay flushing. We don't really need to flush here, just ensure 
+    * that there is one flush before svga_get_query_result attempts to get the
+    * result */
+   svga_context_flush(svga, NULL);
+
+   svga->sq = NULL;
+}
+
+static boolean svga_get_query_result(struct pipe_context *pipe, 
+                                     struct pipe_query *q,
+                                     boolean wait,
+                                     uint64_t *result)
+{
+   struct svga_context *svga = svga_context( pipe );
+   struct svga_screen *svgascreen = svga_screen( pipe->screen );
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_query *sq = svga_query( q );
+   SVGA3dQueryState state;
+   
+   SVGA_DBG(DEBUG_QUERY, "%s wait: %d\n", __FUNCTION__);
+
+   /* The query status won't be updated by the host unless 
+    * SVGA_3D_CMD_WAIT_FOR_QUERY is emitted. Unfortunately this will cause a 
+    * synchronous wait on the host */
+   if(!sq->fence) {
+      enum pipe_error ret;
+
+      ret = SVGA3D_WaitForQuery( svga->swc, sq->type, sq->hwbuf);
+      if(ret != PIPE_OK) {
+         svga_context_flush(svga, NULL);
+         ret = SVGA3D_WaitForQuery( svga->swc, sq->type, sq->hwbuf);
+         assert(ret == PIPE_OK);
+      }
+   
+      svga_context_flush(svga, &sq->fence);
+      
+      assert(sq->fence);
+   }
+
+   state = sq->queryResult->state;
+   if(state == SVGA3D_QUERYSTATE_PENDING) {
+      if(!wait)
+         return FALSE;
+   
+      sws->fence_finish(sws, sq->fence, 0);
+      
+      state = sq->queryResult->state;
+   }
+
+   assert(state == SVGA3D_QUERYSTATE_SUCCEEDED || 
+          state == SVGA3D_QUERYSTATE_FAILED);
+   
+   *result = (uint64_t)sq->queryResult->result32;
+
+   SVGA_DBG(DEBUG_QUERY, "%s result %d\n", __FUNCTION__, (unsigned)*result);
+
+   return TRUE;
+}
+
+
+
+void svga_init_query_functions( struct svga_context *svga )
+{
+   svga->pipe.create_query = svga_create_query;
+   svga->pipe.destroy_query = svga_destroy_query;
+   svga->pipe.begin_query = svga_begin_query;
+   svga->pipe.end_query = svga_end_query;
+   svga->pipe.get_query_result = svga_get_query_result;
+}
diff --git a/src/gallium/drivers/svga/svga_pipe_rasterizer.c b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
new file mode 100644
index 0000000000..b03f8eb9cf
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
@@ -0,0 +1,250 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "draw/draw_context.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+
+#include "svga_hw_reg.h"
+
+/* Hardware frontwinding is always set up as SVGA3D_FRONTWINDING_CW.
+ */
+static SVGA3dFace svga_translate_cullmode( unsigned mode,
+                                           unsigned front_winding )
+{
+   switch (mode) {
+   case PIPE_WINDING_NONE:
+      return SVGA3D_FACE_NONE;
+   case PIPE_WINDING_CCW:
+      return SVGA3D_FACE_BACK;
+   case PIPE_WINDING_CW:
+      return SVGA3D_FACE_FRONT;
+   case PIPE_WINDING_BOTH:
+      return SVGA3D_FACE_FRONT_BACK;
+   default:
+      assert(0);
+      return SVGA3D_FACE_NONE;
+   }
+}
+
+static SVGA3dShadeMode svga_translate_flatshade( unsigned mode )
+{
+   return mode ? SVGA3D_SHADEMODE_FLAT : SVGA3D_SHADEMODE_SMOOTH;
+}
+
+
+static void *
+svga_create_rasterizer_state(struct pipe_context *pipe,
+                             const struct pipe_rasterizer_state *templ)
+{
+   struct svga_rasterizer_state *rast = CALLOC_STRUCT( svga_rasterizer_state );
+   /* need this for draw module. */
+   rast->templ = *templ;
+
+   /* light_twoside          - XXX: need fragment shader varient */
+   /* poly_smooth            - XXX: no fallback available */
+   /* poly_stipple_enable    - draw module */
+   /* point_sprite           - ? */
+   /* point_size_per_vertex  - ? */
+   /* sprite_coord_mode      - ??? */
+   /* bypass_vs_viewport_and_clip        - handled by viewport setup */
+   /* flatshade_first        - handled by index translation */
+   /* gl_rasterization_rules - XXX - viewport code */
+   /* line_width             - draw module */
+   /* fill_cw, fill_ccw      - draw module or index translation */
+
+   rast->shademode = svga_translate_flatshade( templ->flatshade );
+   rast->cullmode = svga_translate_cullmode( templ->cull_mode, 
+                                             templ->front_winding );
+   rast->scissortestenable = templ->scissor;
+   rast->multisampleantialias = templ->multisample;
+   rast->antialiasedlineenable = templ->line_smooth;
+   rast->lastpixel = templ->line_last_pixel;
+   rast->pointsize = templ->point_size;
+   rast->pointsize_min = templ->point_size_min;
+   rast->pointsize_max = templ->point_size_max;
+   rast->hw_unfilled = PIPE_POLYGON_MODE_FILL;
+
+   /* Use swtnl + decomposition implement these:
+    */
+   if (templ->poly_stipple_enable)
+      rast->need_pipeline |= SVGA_PIPELINE_FLAG_TRIS;
+
+   if (templ->line_width != 1.0 &&
+       templ->line_width != 0.0)
+      rast->need_pipeline |= SVGA_PIPELINE_FLAG_LINES;
+
+   if (templ->line_stipple_enable) {
+      /* LinePattern not implemented on all backends. 
+       */
+      if (0) {
+         SVGA3dLinePattern lp;
+         lp.repeat = templ->line_stipple_factor + 1;
+         lp.pattern = templ->line_stipple_pattern;
+         rast->linepattern = lp.uintValue;
+      }
+      else {
+         rast->need_pipeline |= SVGA_PIPELINE_FLAG_LINES;
+      }
+   } 
+
+   if (templ->point_smooth)
+      rast->need_pipeline |= SVGA_PIPELINE_FLAG_POINTS;
+
+   {
+      boolean offset_cw = templ->offset_cw;
+      boolean offset_ccw = templ->offset_ccw;
+      boolean offset  = 0;
+      int fill_cw = templ->fill_cw;
+      int fill_ccw = templ->fill_ccw;
+      int fill = PIPE_POLYGON_MODE_FILL;
+
+      switch (templ->cull_mode) {
+      case PIPE_WINDING_BOTH:
+         offset = 0;
+         fill = PIPE_POLYGON_MODE_FILL;
+         break;
+
+      case PIPE_WINDING_CW:
+         offset = offset_ccw;
+         fill = fill_ccw;
+         break;
+
+      case PIPE_WINDING_CCW:
+         offset = offset_cw;
+         fill = fill_cw;
+         break;
+
+      case PIPE_WINDING_NONE:
+         if (fill_cw != fill_ccw || offset_cw != offset_ccw) 
+         {
+            /* Always need the draw module to work out different
+             * front/back fill modes:
+             */
+            rast->need_pipeline |= SVGA_PIPELINE_FLAG_TRIS;
+         }
+         else {
+            offset = offset_ccw;
+            fill = fill_ccw;
+         }
+         break;
+
+      default:
+         assert(0);
+         break;
+      }
+
+      /* Unfilled primitive modes aren't implemented on all virtual
+       * hardware.  We can do some unfilled processing with index
+       * translation, but otherwise need the draw module:
+       */
+      if (fill != PIPE_POLYGON_MODE_FILL &&
+          (templ->flatshade ||
+           templ->light_twoside ||
+           offset ||
+           templ->cull_mode != PIPE_WINDING_NONE)) 
+      {
+         fill = PIPE_POLYGON_MODE_FILL;
+         rast->need_pipeline |= SVGA_PIPELINE_FLAG_TRIS;
+      }
+
+      /* If we are decomposing to lines, and lines need the pipeline,
+       * then we also need the pipeline for tris.
+       */
+      if (fill == PIPE_POLYGON_MODE_LINE &&
+          (rast->need_pipeline & SVGA_PIPELINE_FLAG_LINES))
+      {
+         fill = PIPE_POLYGON_MODE_FILL;
+         rast->need_pipeline |= SVGA_PIPELINE_FLAG_TRIS;
+      }
+
+      /* Similarly for points:
+       */
+      if (fill == PIPE_POLYGON_MODE_POINT &&
+          (rast->need_pipeline & SVGA_PIPELINE_FLAG_POINTS))
+      {
+         fill = PIPE_POLYGON_MODE_FILL;
+         rast->need_pipeline |= SVGA_PIPELINE_FLAG_TRIS;
+      }
+
+      if (offset) {
+         rast->slopescaledepthbias = templ->offset_scale;
+         rast->depthbias = templ->offset_units;
+      }
+
+      rast->hw_unfilled = fill;
+   }
+
+
+
+
+   if (rast->need_pipeline & SVGA_PIPELINE_FLAG_TRIS) {
+      /* Turn off stuff which will get done in the draw module:
+       */
+      rast->hw_unfilled = PIPE_POLYGON_MODE_FILL;
+      rast->slopescaledepthbias = 0;
+      rast->depthbias = 0;
+   }
+
+   return rast;
+}
+
+static void svga_bind_rasterizer_state( struct pipe_context *pipe,
+                                        void *state )
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_rasterizer_state *raster = (struct svga_rasterizer_state *)state;
+
+   svga->curr.rast = raster;
+
+   draw_set_rasterizer_state(svga->swtnl.draw, raster ? &raster->templ : NULL);
+   
+   svga->dirty |= SVGA_NEW_RAST;
+}
+
+static void svga_delete_rasterizer_state(struct pipe_context *pipe,
+                                         void *raster)
+{
+   FREE(raster);
+}
+
+
+void svga_init_rasterizer_functions( struct svga_context *svga )
+{
+   svga->pipe.create_rasterizer_state = svga_create_rasterizer_state;
+   svga->pipe.bind_rasterizer_state = svga_bind_rasterizer_state;
+   svga->pipe.delete_rasterizer_state = svga_delete_rasterizer_state;
+}
+
+
+/***********************************************************************
+ * Hardware state update
+ */
+
diff --git a/src/gallium/drivers/svga/svga_pipe_sampler.c b/src/gallium/drivers/svga/svga_pipe_sampler.c
new file mode 100644
index 0000000000..3eeca6b784
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_sampler.c
@@ -0,0 +1,243 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "util/u_pack_color.h"
+#include "tgsi/tgsi_parse.h"
+
+#include "svga_context.h"
+#include "svga_screen_texture.h"
+#include "svga_state.h"
+
+#include "svga_hw_reg.h"
+
+#include "svga_debug.h"
+
+static INLINE unsigned
+translate_wrap_mode(unsigned wrap)
+{
+   switch (wrap) {
+   case PIPE_TEX_WRAP_REPEAT: 
+      return SVGA3D_TEX_ADDRESS_WRAP;
+
+   case PIPE_TEX_WRAP_CLAMP: 
+      return SVGA3D_TEX_ADDRESS_CLAMP;
+
+   case PIPE_TEX_WRAP_CLAMP_TO_EDGE: 
+      /* Unfortunately SVGA3D_TEX_ADDRESS_EDGE not respected by
+       * hardware.
+       */
+      return SVGA3D_TEX_ADDRESS_CLAMP;
+
+   case PIPE_TEX_WRAP_CLAMP_TO_BORDER: 
+      return SVGA3D_TEX_ADDRESS_BORDER;
+
+   case PIPE_TEX_WRAP_MIRROR_REPEAT: 
+      return SVGA3D_TEX_ADDRESS_MIRROR;
+
+   case PIPE_TEX_WRAP_MIRROR_CLAMP:  
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:   
+   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: 
+      return SVGA3D_TEX_ADDRESS_MIRRORONCE;
+
+   default:
+      assert(0);
+      return SVGA3D_TEX_ADDRESS_WRAP;
+   }
+}
+
+static INLINE unsigned translate_img_filter( unsigned filter )
+{
+   switch (filter) {
+   case PIPE_TEX_FILTER_NEAREST: return SVGA3D_TEX_FILTER_NEAREST;
+   case PIPE_TEX_FILTER_LINEAR:  return SVGA3D_TEX_FILTER_LINEAR;
+   case PIPE_TEX_FILTER_ANISO:   return SVGA3D_TEX_FILTER_ANISOTROPIC;
+   default:
+      assert(0);
+      return SVGA3D_TEX_FILTER_NEAREST;
+   }
+}
+
+static INLINE unsigned translate_mip_filter( unsigned filter )
+{
+   switch (filter) {
+   case PIPE_TEX_MIPFILTER_NONE:    return SVGA3D_TEX_FILTER_NONE;
+   case PIPE_TEX_MIPFILTER_NEAREST: return SVGA3D_TEX_FILTER_NEAREST;
+   case PIPE_TEX_MIPFILTER_LINEAR:  return SVGA3D_TEX_FILTER_LINEAR;
+   default:
+      assert(0);
+      return SVGA3D_TEX_FILTER_NONE;
+   }
+}
+
+static void *
+svga_create_sampler_state(struct pipe_context *pipe,
+                          const struct pipe_sampler_state *sampler)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_sampler_state *cso = CALLOC_STRUCT( svga_sampler_state );
+   
+   cso->mipfilter = translate_mip_filter(sampler->min_mip_filter);
+   cso->magfilter = translate_img_filter( sampler->mag_img_filter );
+   cso->minfilter = translate_img_filter( sampler->min_img_filter );
+   cso->aniso_level = MAX2( (unsigned) sampler->max_anisotropy, 1 );
+   cso->lod_bias = sampler->lod_bias;
+   cso->addressu = translate_wrap_mode(sampler->wrap_s);
+   cso->addressv = translate_wrap_mode(sampler->wrap_t);
+   cso->addressw = translate_wrap_mode(sampler->wrap_r);
+   cso->normalized_coords = sampler->normalized_coords;
+   cso->compare_mode = sampler->compare_mode;
+   cso->compare_func = sampler->compare_func;
+
+   {
+      ubyte r = float_to_ubyte(sampler->border_color[0]);
+      ubyte g = float_to_ubyte(sampler->border_color[1]);
+      ubyte b = float_to_ubyte(sampler->border_color[2]);
+      ubyte a = float_to_ubyte(sampler->border_color[3]);
+
+      util_pack_color_ub( r, g, b, a,
+                          PIPE_FORMAT_B8G8R8A8_UNORM,
+                          &cso->bordercolor );
+   }
+
+   /* No SVGA3D support for:
+    *    - min/max LOD clamping
+    */
+   cso->min_lod = 0;
+   cso->view_min_lod = MAX2(sampler->min_lod, 0);
+   cso->view_max_lod = MAX2(sampler->max_lod, 0);
+
+   /* Use min_mipmap */
+   if (svga->debug.use_min_mipmap) {
+      if (cso->view_min_lod == cso->view_max_lod) {
+         cso->min_lod = cso->view_min_lod;
+         cso->view_min_lod = 0;
+         cso->view_max_lod = 1000; /* Just a high number */
+         cso->mipfilter = SVGA3D_TEX_FILTER_NONE;
+      }
+   }
+
+   SVGA_DBG(DEBUG_VIEWS, "min %u, view(min %u, max %u) lod, mipfilter %s\n",
+            cso->min_lod, cso->view_min_lod, cso->view_max_lod,
+            cso->mipfilter == SVGA3D_TEX_FILTER_NONE ? "SVGA3D_TEX_FILTER_NONE" : "SOMETHING");
+
+   return cso;
+}
+
+static void svga_bind_sampler_states(struct pipe_context *pipe,
+                                     unsigned num, void **sampler)
+{
+   struct svga_context *svga = svga_context(pipe);
+   unsigned i;
+
+   assert(num <= PIPE_MAX_SAMPLERS);
+
+   /* Check for no-op */
+   if (num == svga->curr.num_samplers &&
+       !memcmp(svga->curr.sampler, sampler, num * sizeof(void *))) {
+      debug_printf("sampler noop\n");
+      return;
+   }
+
+   for (i = 0; i < num; i++)
+      svga->curr.sampler[i] = sampler[i];
+
+   for (i = num; i < svga->curr.num_samplers; i++)
+      svga->curr.sampler[i] = NULL;
+
+   svga->curr.num_samplers = num;
+   svga->dirty |= SVGA_NEW_SAMPLER;
+}
+
+static void svga_delete_sampler_state(struct pipe_context *pipe,
+                                      void *sampler)
+{
+   FREE(sampler);
+}
+
+
+static void svga_set_sampler_textures(struct pipe_context *pipe,
+                                      unsigned num,
+                                      struct pipe_texture **texture)
+{
+   struct svga_context *svga = svga_context(pipe);
+   unsigned flag_1d = 0;
+   unsigned flag_srgb = 0;
+   uint i;
+
+   assert(num <= PIPE_MAX_SAMPLERS);
+
+   /* Check for no-op */
+   if (num == svga->curr.num_textures &&
+       !memcmp(svga->curr.texture, texture, num * sizeof(struct pipe_texture *))) {
+      if (0) debug_printf("texture noop\n");
+      return;
+   }
+
+   for (i = 0; i < num; i++) {
+      pipe_texture_reference(&svga->curr.texture[i],
+                             texture[i]);
+
+      if (!texture[i])
+         continue;
+
+      if (texture[i]->format == PIPE_FORMAT_A8R8G8B8_SRGB)
+         flag_srgb |= 1 << i;
+
+      if (texture[i]->target == PIPE_TEXTURE_1D)
+         flag_1d |= 1 << i;
+   }
+
+   for (i = num; i < svga->curr.num_textures; i++)
+      pipe_texture_reference(&svga->curr.texture[i],
+                             NULL);
+
+   svga->curr.num_textures = num;
+   svga->dirty |= SVGA_NEW_TEXTURE_BINDING;
+
+   if (flag_srgb != svga->curr.tex_flags.flag_srgb ||
+       flag_1d != svga->curr.tex_flags.flag_1d) 
+   {
+      svga->dirty |= SVGA_NEW_TEXTURE_FLAGS;
+      svga->curr.tex_flags.flag_1d = flag_1d;
+      svga->curr.tex_flags.flag_srgb = flag_srgb;
+   }  
+}
+
+
+
+void svga_init_sampler_functions( struct svga_context *svga )
+{
+   svga->pipe.create_sampler_state = svga_create_sampler_state;
+   svga->pipe.bind_sampler_states = svga_bind_sampler_states;
+   svga->pipe.delete_sampler_state = svga_delete_sampler_state;
+   svga->pipe.set_sampler_textures = svga_set_sampler_textures;
+}
+
+
+
diff --git a/src/gallium/drivers/svga/svga_pipe_vertex.c b/src/gallium/drivers/svga/svga_pipe_vertex.c
new file mode 100644
index 0000000000..28e2787e0d
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_vertex.c
@@ -0,0 +1,115 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "tgsi/tgsi_parse.h"
+
+#include "svga_screen.h"
+#include "svga_screen_buffer.h"
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_winsys.h"
+
+#include "svga_hw_reg.h"
+
+
+static void svga_set_vertex_buffers(struct pipe_context *pipe,
+                                    unsigned count,
+                                    const struct pipe_vertex_buffer *buffers)
+{
+   struct svga_context *svga = svga_context(pipe);
+   unsigned i;
+   boolean any_user_buffer = FALSE;
+
+   /* Check for no change */
+   if (count == svga->curr.num_vertex_buffers &&
+       memcmp(svga->curr.vb, buffers, count * sizeof buffers[0]) == 0)
+      return;
+
+   /* Adjust refcounts */
+   for (i = 0; i < count; i++) {
+      pipe_buffer_reference(&svga->curr.vb[i].buffer, buffers[i].buffer);
+      if (svga_buffer(buffers[i].buffer)->user)
+         any_user_buffer = TRUE;
+   }
+
+   for ( ; i < svga->curr.num_vertex_buffers; i++)
+      pipe_buffer_reference(&svga->curr.vb[i].buffer, NULL);
+
+   /* Copy remaining data */
+   memcpy(svga->curr.vb, buffers, count * sizeof buffers[0]);
+   svga->curr.num_vertex_buffers = count;
+   svga->curr.any_user_vertex_buffers = any_user_buffer;
+
+   svga->dirty |= SVGA_NEW_VBUFFER;
+}
+
+static void svga_set_vertex_elements(struct pipe_context *pipe,
+                                     unsigned count,
+                                     const struct pipe_vertex_element *elements)
+{
+   struct svga_context *svga = svga_context(pipe);
+   unsigned i;
+
+   for (i = 0; i < count; i++)
+      svga->curr.ve[i] = elements[i];
+
+   svga->curr.num_vertex_elements = count;
+   svga->dirty |= SVGA_NEW_VELEMENT;
+}
+
+
+static void svga_set_edgeflags(struct pipe_context *pipe,
+                               const unsigned *bitfield)
+{
+   struct svga_context *svga = svga_context(pipe);
+
+   if (bitfield != NULL || svga->curr.edgeflags != NULL) {
+      svga->curr.edgeflags = bitfield;
+      svga->dirty |= SVGA_NEW_EDGEFLAGS;
+   }
+}
+
+
+void svga_cleanup_vertex_state( struct svga_context *svga )
+{
+   unsigned i;
+   
+   for (i = 0 ; i < svga->curr.num_vertex_buffers; i++)
+      pipe_buffer_reference(&svga->curr.vb[i].buffer, NULL);
+}
+
+
+void svga_init_vertex_functions( struct svga_context *svga )
+{
+   svga->pipe.set_vertex_buffers = svga_set_vertex_buffers;
+   svga->pipe.set_vertex_elements = svga_set_vertex_elements;
+   svga->pipe.set_edgeflags = svga_set_edgeflags;
+}
+
+
diff --git a/src/gallium/drivers/svga/svga_pipe_vs.c b/src/gallium/drivers/svga/svga_pipe_vs.c
new file mode 100644
index 0000000000..e5ffe668c3
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_pipe_vs.c
@@ -0,0 +1,189 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "draw/draw_context.h"
+#include "pipe/p_inlines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_text.h"
+
+#include "svga_screen.h"
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_tgsi.h"
+#include "svga_hw_reg.h"
+#include "svga_cmd.h"
+#include "svga_debug.h"
+
+
+static const struct tgsi_token *substitute_vs( 
+   unsigned shader_id,
+   const struct tgsi_token *old_tokens )
+{
+#if 0
+   if (shader_id == 12) {
+   static struct tgsi_token tokens[300];
+
+   const char *text = 
+      "VERT1.1\n"
+      "DCL IN[0]\n"
+      "DCL IN[1]\n"
+      "DCL IN[2]\n"
+      "DCL OUT[0], POSITION\n"
+      "DCL TEMP[0..4]\n"
+      "IMM FLT32 {     1.0000,     1.0000,     1.0000,     1.0000 }\n"
+      "IMM FLT32 {     0.45,     1.0000,     1.0000,     1.0000 }\n"
+      "IMM FLT32 { 1.297863, 0.039245, 0.035993, 0.035976}\n"
+      "IMM FLT32 { -0.019398, 1.696131, -0.202151, -0.202050  }\n"
+      "IMM FLT32 { 0.051711, -0.348713, -0.979204, -0.978714  }\n"
+      "IMM FLT32 { 0.000000, 0.000003, 139.491577, 141.421356 }\n"
+      "DCL CONST[0..7]\n"
+      "DCL CONST[9..16]\n"
+      "  MOV TEMP[2], IMM[0]\n"
+
+      "  MOV TEMP[2].xyz, IN[2]\n"
+      "  MOV TEMP[2].xyz, IN[0]\n"
+      "  MOV TEMP[2].xyz, IN[1]\n"
+
+      "  MUL TEMP[1], IMM[3], TEMP[2].yyyy\n"
+      "  MAD TEMP[3], IMM[2],  TEMP[2].xxxx, TEMP[1]\n"
+      "  MAD TEMP[1], IMM[4], TEMP[2].zzzz, TEMP[3]\n"
+      "  MAD TEMP[4], IMM[5], TEMP[2].wwww, TEMP[1]\n"
+
+      "  MOV OUT[0], TEMP[4]\n"
+      "  END\n";
+
+   if (!tgsi_text_translate( text,
+                             tokens,
+                             Elements(tokens) ))
+   {
+      assert(0);
+      return NULL;
+   }
+
+   return tokens;
+   }
+#endif
+
+   return old_tokens;
+}
+
+
+/***********************************************************************
+ * Vertex shaders 
+ */
+
+static void *
+svga_create_vs_state(struct pipe_context *pipe,
+                     const struct pipe_shader_state *templ)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_screen *svgascreen = svga_screen(pipe->screen);
+   struct svga_vertex_shader *vs = CALLOC_STRUCT(svga_vertex_shader);
+   if (!vs)
+      return NULL;
+
+   /* substitute a debug shader?
+    */
+   vs->base.tokens = tgsi_dup_tokens(substitute_vs(svga->debug.shader_id,
+                                                   templ->tokens));
+
+
+   /* Collect basic info that we'll need later:
+    */
+   tgsi_scan_shader(vs->base.tokens, &vs->base.info);
+
+   {
+      /* Need to do construct a new template in case we substitued a
+       * debug shader.
+       */
+      struct pipe_shader_state tmp2 = *templ;
+      tmp2.tokens = vs->base.tokens;
+      vs->draw_shader = draw_create_vertex_shader(svga->swtnl.draw, &tmp2);
+   }
+
+   vs->base.id = svga->debug.shader_id++;
+   vs->base.use_sm30 = svgascreen->use_vs30;
+
+   if (SVGA_DEBUG & DEBUG_TGSI || 0) {
+      debug_printf("%s id: %u, inputs: %u, outputs: %u\n",
+                   __FUNCTION__, vs->base.id,
+                   vs->base.info.num_inputs, vs->base.info.num_outputs);
+   }
+
+   return vs;
+}
+
+static void svga_bind_vs_state(struct pipe_context *pipe, void *shader)
+{
+   struct svga_vertex_shader *vs = (struct svga_vertex_shader *)shader;
+   struct svga_context *svga = svga_context(pipe);
+
+   svga->curr.vs = vs;
+   svga->dirty |= SVGA_NEW_VS;
+}
+
+
+static void svga_delete_vs_state(struct pipe_context *pipe, void *shader)
+{
+   struct svga_context *svga = svga_context(pipe);
+   struct svga_vertex_shader *vs = (struct svga_vertex_shader *)shader;
+   struct svga_shader_result *result, *tmp;
+   enum pipe_error ret;
+
+   svga_hwtnl_flush_retry( svga );
+
+   draw_delete_vertex_shader(svga->swtnl.draw, vs->draw_shader);
+   
+   for (result = vs->base.results; result; result = tmp ) {
+      tmp = result->next;
+
+      ret = SVGA3D_DestroyShader(svga->swc, 
+                                 result->id,
+                                 SVGA3D_SHADERTYPE_VS );
+      if(ret != PIPE_OK) {
+         svga_context_flush(svga, NULL);
+         ret = SVGA3D_DestroyShader(svga->swc, 
+                                    result->id,
+                                    SVGA3D_SHADERTYPE_VS );
+         assert(ret == PIPE_OK);
+      }
+
+      svga_destroy_shader_result( result );
+   }
+
+   FREE((void *)vs->base.tokens);
+   FREE(vs);
+}
+
+
+void svga_init_vs_functions( struct svga_context *svga )
+{
+   svga->pipe.create_vs_state = svga_create_vs_state;
+   svga->pipe.bind_vs_state = svga_bind_vs_state;
+   svga->pipe.delete_vs_state = svga_delete_vs_state;
+}
+
diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
new file mode 100644
index 0000000000..3afcaffff5
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -0,0 +1,435 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "util/u_memory.h"
+#include "pipe/p_inlines.h"
+#include "util/u_string.h"
+#include "util/u_math.h"
+
+#include "svga_winsys.h"
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_screen_texture.h"
+#include "svga_screen_buffer.h"
+#include "svga_cmd.h"
+#include "svga_debug.h"
+
+#include "svga_hw_reg.h"
+#include "svga3d_shaderdefs.h"
+
+
+#ifdef DEBUG
+int SVGA_DEBUG = 0;
+
+static const struct debug_named_value svga_debug_flags[] = {
+   { "dma",      DEBUG_DMA },
+   { "tgsi",     DEBUG_TGSI },
+   { "pipe",     DEBUG_PIPE },
+   { "state",    DEBUG_STATE },
+   { "screen",   DEBUG_SCREEN },
+   { "tex",      DEBUG_TEX },
+   { "swtnl",    DEBUG_SWTNL },
+   { "const",    DEBUG_CONSTS },
+   { "viewport", DEBUG_VIEWPORT },
+   { "views",    DEBUG_VIEWS },
+   { "perf",     DEBUG_PERF },
+   { "flush",    DEBUG_FLUSH },
+   { "sync",     DEBUG_SYNC },
+   {NULL, 0}
+};
+#endif
+
+static const char *
+svga_get_vendor( struct pipe_screen *pscreen )
+{
+   return "VMware, Inc.";
+}
+
+
+static const char *
+svga_get_name( struct pipe_screen *pscreen )
+{
+#ifdef DEBUG
+   /* Only return internal details in the DEBUG version:
+    */
+   return "SVGA3D; build: DEBUG; mutex: " PIPE_ATOMIC;
+#else
+   return "SVGA3D; build: RELEASE; ";
+#endif
+}
+
+
+
+
+static float
+svga_get_paramf(struct pipe_screen *screen, int param)
+{
+   struct svga_screen *svgascreen = svga_screen(screen);
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   SVGA3dDevCapResult result;
+
+   switch (param) {
+   case PIPE_CAP_MAX_LINE_WIDTH:
+      /* fall-through */
+   case PIPE_CAP_MAX_LINE_WIDTH_AA:
+      return 7.0;
+
+   case PIPE_CAP_MAX_POINT_WIDTH:
+      /* fall-through */
+   case PIPE_CAP_MAX_POINT_WIDTH_AA:
+      /* Keep this to a reasonable size to avoid failures in
+       * conform/pntaa.c:
+       */
+      return 80.0;
+
+   case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+      return 4.0;
+
+   case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+      return 16.0;
+
+   case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+      return 16;
+   case PIPE_CAP_NPOT_TEXTURES:
+      return 1;
+   case PIPE_CAP_TWO_SIDED_STENCIL:
+      return 1;
+   case PIPE_CAP_GLSL:
+      return svgascreen->use_ps30 && svgascreen->use_vs30;
+   case PIPE_CAP_ANISOTROPIC_FILTER:
+      return 1;
+   case PIPE_CAP_POINT_SPRITE:
+      return 1;
+   case PIPE_CAP_MAX_RENDER_TARGETS:
+      if(!sws->get_cap(sws, SVGA3D_DEVCAP_MAX_RENDER_TARGETS, &result))
+         return 1;
+      if(!result.u)
+         return 1;
+      return MIN2(result.u, PIPE_MAX_COLOR_BUFS);
+   case PIPE_CAP_OCCLUSION_QUERY:
+      return 1;
+   case PIPE_CAP_TEXTURE_SHADOW_MAP:
+      return 1;
+   case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+      return SVGA_MAX_TEXTURE_LEVELS;
+   case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+      return 8;  /* max 128x128x128 */
+   case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+      return SVGA_MAX_TEXTURE_LEVELS;
+
+   case PIPE_CAP_TEXTURE_MIRROR_REPEAT: /* req. for GL 1.4 */
+      return 1;
+
+   case PIPE_CAP_BLEND_EQUATION_SEPARATE: /* req. for GL 1.5 */
+      return 1;
+
+   default:
+      return 0;
+   }
+}
+
+
+/* This is a fairly pointless interface
+ */
+static int
+svga_get_param(struct pipe_screen *screen, int param)
+{
+   return (int) svga_get_paramf( screen, param );
+}
+
+
+static INLINE SVGA3dDevCapIndex
+svga_translate_format_cap(enum pipe_format format)
+{
+   switch(format) {
+   
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_A8R8G8B8;
+   case PIPE_FORMAT_X8R8G8B8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_X8R8G8B8;
+
+   case PIPE_FORMAT_R5G6B5_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_R5G6B5;
+   case PIPE_FORMAT_A1R5G5B5_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_A1R5G5B5;
+   case PIPE_FORMAT_A4R4G4B4_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_A4R4G4B4;
+
+   case PIPE_FORMAT_Z16_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_Z_D16;
+   case PIPE_FORMAT_Z24S8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_Z_D24S8;
+   case PIPE_FORMAT_Z24X8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_Z_D24X8;
+
+   case PIPE_FORMAT_A8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_ALPHA8;
+   case PIPE_FORMAT_L8_UNORM:
+      return SVGA3D_DEVCAP_SURFACEFMT_LUMINANCE8;
+
+   case PIPE_FORMAT_DXT1_RGB:
+   case PIPE_FORMAT_DXT1_RGBA:
+      return SVGA3D_DEVCAP_SURFACEFMT_DXT1;
+   case PIPE_FORMAT_DXT3_RGBA:
+      return SVGA3D_DEVCAP_SURFACEFMT_DXT3;
+   case PIPE_FORMAT_DXT5_RGBA:
+      return SVGA3D_DEVCAP_SURFACEFMT_DXT5;
+
+   default:
+      return SVGA3D_DEVCAP_MAX;
+   }
+}
+
+
+static boolean
+svga_is_format_supported( struct pipe_screen *screen,
+                          enum pipe_format format, 
+                          enum pipe_texture_target target,
+                          unsigned tex_usage, 
+                          unsigned geom_flags )
+{
+   struct svga_winsys_screen *sws = svga_screen(screen)->sws;
+   SVGA3dDevCapIndex index;
+   SVGA3dDevCapResult result;
+   
+   assert(tex_usage);
+
+   /* Override host capabilities */
+   if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+      switch(format) { 
+
+      /* Often unsupported/problematic. This means we end up with the same
+       * visuals for all virtual hardware implementations.
+       */
+      case PIPE_FORMAT_A4R4G4B4_UNORM:
+      case PIPE_FORMAT_A1R5G5B5_UNORM:
+         return FALSE;
+         
+      /* Simulate ability to render into compressed textures */
+      case PIPE_FORMAT_DXT1_RGB:
+      case PIPE_FORMAT_DXT1_RGBA:
+      case PIPE_FORMAT_DXT3_RGBA:
+      case PIPE_FORMAT_DXT5_RGBA:
+         return TRUE;
+
+      default:
+         break;
+      }
+   }
+   
+   /* Try to query the host */
+   index = svga_translate_format_cap(format);
+   if( index < SVGA3D_DEVCAP_MAX && 
+       sws->get_cap(sws, index, &result) )
+   {
+      SVGA3dSurfaceFormatCaps mask;
+      
+      mask.value = 0;
+      if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET)
+         mask.offscreenRenderTarget = 1;
+      if (tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL)
+         mask.zStencil = 1;
+      if (tex_usage & PIPE_TEXTURE_USAGE_SAMPLER)
+         mask.texture = 1;
+
+      if ((result.u & mask.value) == mask.value)
+         return TRUE;
+      else
+         return FALSE;
+   }
+
+   /* Use our translate functions directly rather than relying on a
+    * duplicated list of supported formats which is prone to getting
+    * out of sync:
+    */
+   if(tex_usage & (PIPE_TEXTURE_USAGE_RENDER_TARGET | PIPE_TEXTURE_USAGE_DEPTH_STENCIL))
+      return svga_translate_format_render(format) != SVGA3D_FORMAT_INVALID;
+   else
+      return svga_translate_format(format) != SVGA3D_FORMAT_INVALID;
+}
+
+
+static void
+svga_fence_reference(struct pipe_screen *screen,
+                     struct pipe_fence_handle **ptr,
+                     struct pipe_fence_handle *fence)
+{
+   struct svga_winsys_screen *sws = svga_screen(screen)->sws;
+   sws->fence_reference(sws, ptr, fence);
+}
+
+
+static int
+svga_fence_signalled(struct pipe_screen *screen,
+                     struct pipe_fence_handle *fence,
+                     unsigned flag)
+{
+   struct svga_winsys_screen *sws = svga_screen(screen)->sws;
+   return sws->fence_signalled(sws, fence, flag);
+}
+
+
+static int
+svga_fence_finish(struct pipe_screen *screen,
+                  struct pipe_fence_handle *fence,
+                  unsigned flag)
+{
+   struct svga_winsys_screen *sws = svga_screen(screen)->sws;
+   return sws->fence_finish(sws, fence, flag);
+}
+
+
+static void
+svga_destroy_screen( struct pipe_screen *screen )
+{
+   struct svga_screen *svgascreen = svga_screen(screen);
+   
+   svga_screen_cache_cleanup(svgascreen);
+
+   pipe_mutex_destroy(svgascreen->swc_mutex);
+   pipe_mutex_destroy(svgascreen->tex_mutex);
+
+   svgascreen->swc->destroy(svgascreen->swc);
+   
+   svgascreen->sws->destroy(svgascreen->sws);
+   
+   FREE(svgascreen);
+}
+
+
+/**
+ * Create a new svga_screen object
+ */
+struct pipe_screen *
+svga_screen_create(struct svga_winsys_screen *sws)
+{
+   struct svga_screen *svgascreen;
+   struct pipe_screen *screen;
+   SVGA3dDevCapResult result;
+
+#ifdef DEBUG
+   SVGA_DEBUG = debug_get_flags_option("SVGA_DEBUG", svga_debug_flags, 0 );
+#endif
+
+   svgascreen = CALLOC_STRUCT(svga_screen);
+   if (!svgascreen)
+      goto error1;
+
+   svgascreen->debug.force_level_surface_view =
+      debug_get_bool_option("SVGA_FORCE_LEVEL_SURFACE_VIEW", FALSE);
+   svgascreen->debug.force_surface_view =
+      debug_get_bool_option("SVGA_FORCE_SURFACE_VIEW", FALSE);
+   svgascreen->debug.force_sampler_view =
+      debug_get_bool_option("SVGA_FORCE_SAMPLER_VIEW", FALSE);
+   svgascreen->debug.no_surface_view =
+      debug_get_bool_option("SVGA_NO_SURFACE_VIEW", FALSE);
+   svgascreen->debug.no_sampler_view =
+      debug_get_bool_option("SVGA_NO_SAMPLER_VIEW", FALSE);
+
+   screen = &svgascreen->screen;
+
+   screen->destroy = svga_destroy_screen;
+   screen->get_name = svga_get_name;
+   screen->get_vendor = svga_get_vendor;
+   screen->get_param = svga_get_param;
+   screen->get_paramf = svga_get_paramf;
+   screen->is_format_supported = svga_is_format_supported;
+   screen->fence_reference = svga_fence_reference;
+   screen->fence_signalled = svga_fence_signalled;
+   screen->fence_finish = svga_fence_finish;
+   svgascreen->sws = sws;
+
+   svga_screen_init_texture_functions(screen);
+   svga_screen_init_buffer_functions(screen);
+
+   svgascreen->use_ps30 =
+      sws->get_cap(sws, SVGA3D_DEVCAP_FRAGMENT_SHADER_VERSION, &result) &&
+      result.u >= SVGA3DPSVERSION_30 ? TRUE : FALSE;
+
+   svgascreen->use_vs30 =
+      sws->get_cap(sws, SVGA3D_DEVCAP_VERTEX_SHADER_VERSION, &result) &&
+      result.u >= SVGA3DVSVERSION_30 ? TRUE : FALSE;
+
+#if 1
+   /* Shader model 2.0 is unsupported at the moment. */
+   if(!svgascreen->use_ps30 || !svgascreen->use_vs30)
+      goto error2;
+#else
+   if(debug_get_bool_option("SVGA_NO_SM30", FALSE))
+      svgascreen->use_vs30 = svgascreen->use_ps30 = FALSE;
+#endif
+
+   svgascreen->swc = sws->context_create(sws);
+   if(!svgascreen->swc)
+      goto error2;
+
+   pipe_mutex_init(svgascreen->tex_mutex);
+   pipe_mutex_init(svgascreen->swc_mutex);
+
+   LIST_INITHEAD(&svgascreen->cached_buffers);
+   
+   svga_screen_cache_init(svgascreen);
+
+   return screen;
+error2:
+   FREE(svgascreen);
+error1:
+   return NULL;
+}
+
+void svga_screen_flush( struct svga_screen *svgascreen, 
+                        struct pipe_fence_handle **pfence )
+{
+   struct pipe_fence_handle *fence = NULL;
+
+   SVGA_DBG(DEBUG_PERF, "%s\n", __FUNCTION__);
+   
+   pipe_mutex_lock(svgascreen->swc_mutex);
+   svgascreen->swc->flush(svgascreen->swc, &fence);
+   pipe_mutex_unlock(svgascreen->swc_mutex);
+   
+   svga_screen_cache_flush(svgascreen, fence);
+   
+   if(pfence)
+      *pfence = fence;
+   else
+      svgascreen->sws->fence_reference(svgascreen->sws, &fence, NULL);
+}
+
+struct svga_winsys_screen *
+svga_winsys_screen(struct pipe_screen *screen)
+{
+   return svga_screen(screen)->sws;
+}
+
+#ifdef DEBUG
+struct svga_screen *
+svga_screen(struct pipe_screen *screen)
+{
+   assert(screen);
+   assert(screen->destroy == svga_destroy_screen);
+   return (struct svga_screen *)screen;
+}
+#endif
diff --git a/src/gallium/drivers/svga/svga_screen.h b/src/gallium/drivers/svga/svga_screen.h
new file mode 100644
index 0000000000..b94ca7fc1c
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen.h
@@ -0,0 +1,95 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_SCREEN_H
+#define SVGA_SCREEN_H
+
+
+#include "pipe/p_screen.h"
+#include "pipe/p_thread.h"
+
+#include "util/u_double_list.h"
+
+#include "svga_screen_cache.h"
+
+
+struct svga_winsys_screen;
+struct svga_winsys_context;
+struct SVGACmdMemory;
+
+#define SVGA_COMBINE_USERBUFFERS 1
+
+/**
+ * Subclass of pipe_screen
+ */
+struct svga_screen
+{
+   struct pipe_screen screen;
+   struct svga_winsys_screen *sws;
+
+   unsigned use_ps30;
+   unsigned use_vs30;
+   
+   struct {
+      boolean force_level_surface_view;
+      boolean force_surface_view;
+      boolean no_surface_view;
+      boolean force_sampler_view;
+      boolean no_sampler_view;
+   } debug;
+
+   /* The screen needs its own context */
+   struct svga_winsys_context *swc;
+   struct SVGACmdMemory *fifo;
+
+   unsigned texture_timestamp;
+   pipe_mutex tex_mutex; 
+   pipe_mutex swc_mutex; /* Protects the use of swc and dirty_buffers */
+   
+   /** 
+    * List of buffers with cached GMR. Ordered from the most recently used to
+    * the least recently used 
+    */
+   struct list_head cached_buffers;
+   
+   struct svga_host_surface_cache cache;
+};
+
+#ifndef DEBUG
+/** cast wrapper */
+static INLINE struct svga_screen *
+svga_screen(struct pipe_screen *pscreen)
+{
+   return (struct svga_screen *) pscreen;
+}
+#else
+struct svga_screen *
+svga_screen(struct pipe_screen *screen);
+#endif
+
+void svga_screen_flush( struct svga_screen *svga_screen, 
+                        struct pipe_fence_handle **pfence );
+
+#endif /* SVGA_SCREEN_H */
diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
new file mode 100644
index 0000000000..3b7811734e
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -0,0 +1,820 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_thread.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_debug.h"
+
+
+/**
+ * Vertex and index buffers have to be treated slightly differently from 
+ * regular guest memory regions because the SVGA device sees them as 
+ * surfaces, and the state tracker can create/destroy without the pipe 
+ * driver, therefore we must do the uploads from the vws.
+ */
+static INLINE boolean
+svga_buffer_needs_hw_storage(unsigned usage)
+{
+   return usage & (PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_INDEX);
+}
+
+
+static INLINE enum pipe_error
+svga_buffer_create_host_surface(struct svga_screen *ss,
+                                struct svga_buffer *sbuf)
+{
+   if(!sbuf->handle) {
+      sbuf->key.flags = 0;
+      
+      sbuf->key.format = SVGA3D_BUFFER;
+      if(sbuf->base.usage & PIPE_BUFFER_USAGE_VERTEX)
+         sbuf->key.flags |= SVGA3D_SURFACE_HINT_VERTEXBUFFER;
+      if(sbuf->base.usage & PIPE_BUFFER_USAGE_INDEX)
+         sbuf->key.flags |= SVGA3D_SURFACE_HINT_INDEXBUFFER;
+      
+      sbuf->key.size.width = sbuf->base.size;
+      sbuf->key.size.height = 1;
+      sbuf->key.size.depth = 1;
+      
+      sbuf->key.numFaces = 1;
+      sbuf->key.numMipLevels = 1;
+      
+      sbuf->handle = svga_screen_surface_create(ss, &sbuf->key);
+      if(!sbuf->handle)
+         return PIPE_ERROR_OUT_OF_MEMORY;
+   
+      /* Always set the discard flag on the first time the buffer is written
+       * as svga_screen_surface_create might have passed a recycled host
+       * buffer.
+       */
+      sbuf->hw.flags.discard = TRUE;
+
+      SVGA_DBG(DEBUG_DMA, "   grab sid %p sz %d\n", sbuf->handle, sbuf->base.size);
+   }
+   
+   return PIPE_OK;
+}   
+
+
+static INLINE void
+svga_buffer_destroy_host_surface(struct svga_screen *ss,
+                                 struct svga_buffer *sbuf)
+{
+   if(sbuf->handle) {
+      SVGA_DBG(DEBUG_DMA, " ungrab sid %p sz %d\n", sbuf->handle, sbuf->base.size);
+      svga_screen_surface_destroy(ss, &sbuf->key, &sbuf->handle);
+   }
+}   
+
+
+static INLINE void
+svga_buffer_destroy_hw_storage(struct svga_screen *ss, struct svga_buffer *sbuf)
+{
+   struct svga_winsys_screen *sws = ss->sws;
+
+   assert(!sbuf->map.count);
+   assert(sbuf->hw.buf);
+   if(sbuf->hw.buf) {
+      sws->buffer_destroy(sws, sbuf->hw.buf);
+      sbuf->hw.buf = NULL;
+      assert(sbuf->head.prev && sbuf->head.next);
+      LIST_DEL(&sbuf->head);
+#ifdef DEBUG
+      sbuf->head.next = sbuf->head.prev = NULL; 
+#endif
+   }
+}
+
+static INLINE enum pipe_error
+svga_buffer_backup(struct svga_screen *ss, struct svga_buffer *sbuf)
+{
+   if (sbuf->hw.buf && sbuf->hw.num_ranges) {
+      void *src;
+
+      if (!sbuf->swbuf)
+	 sbuf->swbuf = align_malloc(sbuf->base.size, sbuf->base.alignment);
+      if (!sbuf->swbuf)
+	 return PIPE_ERROR_OUT_OF_MEMORY;
+
+      src = ss->sws->buffer_map(ss->sws, sbuf->hw.buf,
+				PIPE_BUFFER_USAGE_CPU_READ);
+      if (!src)
+	 return PIPE_ERROR;
+
+      memcpy(sbuf->swbuf, src, sbuf->base.size);
+      ss->sws->buffer_unmap(ss->sws, sbuf->hw.buf);
+   }
+
+   return PIPE_OK;
+}
+
+/**
+ * Try to make GMR space available by freeing the hardware storage of 
+ * unmapped
+ */
+boolean
+svga_buffer_free_cached_hw_storage(struct svga_screen *ss)
+{
+   struct list_head *curr;
+   struct svga_buffer *sbuf;
+   enum pipe_error ret = PIPE_OK;
+
+   curr = ss->cached_buffers.prev;
+   
+   /* free the least recently used buffer's hw storage which is not mapped */
+   do {
+      if(curr == &ss->cached_buffers)
+         return FALSE;
+
+      sbuf = LIST_ENTRY(struct svga_buffer, curr, head);
+      
+      curr = curr->prev;
+      if (sbuf->map.count == 0)
+	 ret = svga_buffer_backup(ss, sbuf);
+
+   } while(sbuf->map.count != 0 || ret != PIPE_OK);
+   
+   svga_buffer_destroy_hw_storage(ss, sbuf);
+   
+   return TRUE;
+}
+
+struct svga_winsys_buffer *
+svga_winsys_buffer_create( struct svga_screen *ss,
+                           unsigned alignment, 
+                           unsigned usage,
+                           unsigned size )
+{
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_winsys_buffer *buf;
+   
+   /* Just try */
+   buf = sws->buffer_create(sws, alignment, usage, size);
+   if(!buf) {
+
+      SVGA_DBG(DEBUG_DMA|DEBUG_PERF, "flushing screen to find %d bytes GMR\n", 
+               size); 
+      
+      /* Try flushing all pending DMAs */
+      svga_screen_flush(ss, NULL);
+      buf = sws->buffer_create(sws, alignment, usage, size);
+
+      SVGA_DBG(DEBUG_DMA|DEBUG_PERF, "evicting buffers to find %d bytes GMR\n", 
+               size);
+
+      /* Try evicing all buffer storage */
+      while(!buf && svga_buffer_free_cached_hw_storage(ss))
+         buf = sws->buffer_create(sws, alignment, usage, size);
+   }
+   
+   return buf;
+}
+
+
+/**
+ * Allocate DMA'ble storage for the buffer. 
+ * 
+ * Called before mapping a buffer.
+ */
+static INLINE enum pipe_error
+svga_buffer_create_hw_storage(struct svga_screen *ss,
+                              struct svga_buffer *sbuf)
+{
+   if(!sbuf->hw.buf) {
+      unsigned alignment = sbuf->base.alignment;
+      unsigned usage = 0;
+      unsigned size = sbuf->base.size;
+      
+      sbuf->hw.buf = svga_winsys_buffer_create(ss, alignment, usage, size);
+      if(!sbuf->hw.buf)
+         return PIPE_ERROR_OUT_OF_MEMORY;
+      
+      assert(!sbuf->needs_flush);
+      assert(!sbuf->head.prev && !sbuf->head.next);
+      LIST_ADD(&sbuf->head, &ss->cached_buffers);
+   }
+   
+   return PIPE_OK;
+}
+
+
+/**
+ * Variant of SVGA3D_BufferDMA which leaves the copy box temporarily in blank.
+ */
+static enum pipe_error
+svga_buffer_upload_command(struct svga_context *svga,
+                           struct svga_buffer *sbuf)
+{
+   struct svga_winsys_context *swc = svga->swc;
+   struct svga_winsys_buffer *guest = sbuf->hw.buf;
+   struct svga_winsys_surface *host = sbuf->handle;
+   SVGA3dTransferType transfer = SVGA3D_WRITE_HOST_VRAM;
+   SVGA3dSurfaceDMAFlags flags = sbuf->hw.flags;
+   SVGA3dCmdSurfaceDMA *cmd;
+   uint32 numBoxes = sbuf->hw.num_ranges;
+   SVGA3dCopyBox *boxes;
+   SVGA3dCmdSurfaceDMASuffix *pSuffix;
+   unsigned region_flags;
+   unsigned surface_flags;
+   struct pipe_buffer *dummy;
+
+   if(transfer == SVGA3D_WRITE_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_READ;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+   }
+   else if(transfer == SVGA3D_READ_HOST_VRAM) {
+      region_flags = PIPE_BUFFER_USAGE_GPU_WRITE;
+      surface_flags = PIPE_BUFFER_USAGE_GPU_READ;
+   }
+   else {
+      assert(0);
+      return PIPE_ERROR_BAD_INPUT;
+   }
+
+   assert(numBoxes);
+
+   cmd = SVGA3D_FIFOReserve(swc,
+                            SVGA_3D_CMD_SURFACE_DMA,
+                            sizeof *cmd + numBoxes * sizeof *boxes + sizeof *pSuffix,
+                            2);
+   if(!cmd)
+      return PIPE_ERROR_OUT_OF_MEMORY;
+
+   swc->region_relocation(swc, &cmd->guest.ptr, guest, 0, region_flags);
+   cmd->guest.pitch = 0;
+
+   swc->surface_relocation(swc, &cmd->host.sid, host, surface_flags);
+   cmd->host.face = 0;
+   cmd->host.mipmap = 0;
+
+   cmd->transfer = transfer;
+
+   sbuf->hw.boxes = (SVGA3dCopyBox *)&cmd[1];
+   sbuf->hw.svga = svga;
+
+   /* Increment reference count */
+   dummy = NULL;
+   pipe_buffer_reference(&dummy, &sbuf->base);
+
+   pSuffix = (SVGA3dCmdSurfaceDMASuffix *)((uint8_t*)cmd + sizeof *cmd + numBoxes * sizeof *boxes);
+   pSuffix->suffixSize = sizeof *pSuffix;
+   pSuffix->maximumOffset = sbuf->base.size;
+   pSuffix->flags = flags;
+
+   swc->commit(swc);
+
+   return PIPE_OK;
+}
+
+
+/**
+ * Patch up the upload DMA command reserved by svga_buffer_upload_command
+ * with the final ranges.
+ */
+static void
+svga_buffer_upload_flush(struct svga_context *svga,
+                         struct svga_buffer *sbuf)
+{
+   struct svga_screen *ss = svga_screen(svga->pipe.screen);
+   SVGA3dCopyBox *boxes;
+   unsigned i;
+
+   assert(sbuf->handle); 
+   assert(sbuf->hw.buf);
+   assert(sbuf->hw.num_ranges);
+   assert(sbuf->hw.svga == svga);
+   assert(sbuf->hw.boxes);
+   
+   /*
+    * Patch the DMA command with the final copy box.
+    */
+
+   SVGA_DBG(DEBUG_DMA, "dma to sid %p\n", sbuf->handle);
+
+   boxes = sbuf->hw.boxes;
+   for(i = 0; i < sbuf->hw.num_ranges; ++i) {
+      SVGA_DBG(DEBUG_DMA, "  bytes %u - %u\n",
+               sbuf->hw.ranges[i].start, sbuf->hw.ranges[i].end);
+
+      boxes[i].x = sbuf->hw.ranges[i].start;
+      boxes[i].y = 0;
+      boxes[i].z = 0;
+      boxes[i].w = sbuf->hw.ranges[i].end - sbuf->hw.ranges[i].start;
+      boxes[i].h = 1;
+      boxes[i].d = 1;
+      boxes[i].srcx = sbuf->hw.ranges[i].start;
+      boxes[i].srcy = 0;
+      boxes[i].srcz = 0;
+   }
+
+   sbuf->hw.num_ranges = 0;
+   memset(&sbuf->hw.flags, 0, sizeof sbuf->hw.flags);
+
+   assert(sbuf->head.prev && sbuf->head.next);
+   LIST_DEL(&sbuf->head);
+   sbuf->needs_flush = FALSE;
+   /* XXX: do we care about cached_buffers any more ?*/
+   LIST_ADD(&sbuf->head, &ss->cached_buffers);
+
+   sbuf->hw.svga = NULL;
+   sbuf->hw.boxes = NULL;
+
+   /* Decrement reference count */
+   pipe_buffer_reference((struct pipe_buffer **)&sbuf, NULL);
+}
+
+
+/**
+ * Queue a DMA upload of a range of this buffer to the host.
+ *
+ * This function only notes the range down. It doesn't actually emit a DMA
+ * upload command. That only happens when a context tries to refer to this
+ * buffer, and the DMA upload command is added to that context's command buffer.
+ * 
+ * We try to lump as many contiguous DMA transfers together as possible.
+ */
+static void
+svga_buffer_upload_queue(struct svga_buffer *sbuf,
+                         unsigned start,
+                         unsigned end)
+{
+   unsigned i;
+
+   assert(sbuf->hw.buf);
+   assert(end > start);
+   
+   /*
+    * Try to grow one of the ranges.
+    *
+    * Note that it is not this function task to care about overlapping ranges,
+    * as the GMR was already given so it is too late to do anything. Situations
+    * where overlapping ranges may pose a problem should be detected via
+    * pipe_context::is_buffer_referenced and the context that refers to the
+    * buffer should be flushed.
+    */
+
+   for(i = 0; i < sbuf->hw.num_ranges; ++i) {
+      if(start <= sbuf->hw.ranges[i].end && sbuf->hw.ranges[i].start <= end) {
+         sbuf->hw.ranges[i].start = MIN2(sbuf->hw.ranges[i].start, start);
+         sbuf->hw.ranges[i].end   = MAX2(sbuf->hw.ranges[i].end,    end);
+         return;
+      }
+   }
+
+   /*
+    * We cannot add a new range to an existing DMA command, so patch-up the
+    * pending DMA upload and start clean.
+    */
+
+   if(sbuf->needs_flush)
+      svga_buffer_upload_flush(sbuf->hw.svga, sbuf);
+
+   assert(!sbuf->needs_flush);
+   assert(!sbuf->hw.svga);
+   assert(!sbuf->hw.boxes);
+
+   /*
+    * Add a new range.
+    */
+
+   sbuf->hw.ranges[sbuf->hw.num_ranges].start = start;
+   sbuf->hw.ranges[sbuf->hw.num_ranges].end = end;
+   ++sbuf->hw.num_ranges;
+}
+
+
+static void *
+svga_buffer_map_range( struct pipe_screen *screen,
+                       struct pipe_buffer *buf,
+                       unsigned offset, unsigned length,
+                       unsigned usage )
+{
+   struct svga_screen *ss = svga_screen(screen); 
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_buffer *sbuf = svga_buffer( buf );
+   void *map;
+
+   if(sbuf->swbuf) {
+      /* User/malloc buffer */
+      map = sbuf->swbuf;
+   }
+   else {
+      if(!sbuf->hw.buf) {
+         struct svga_winsys_surface *handle = sbuf->handle;
+
+         if(svga_buffer_create_hw_storage(ss, sbuf) != PIPE_OK)
+            return NULL;
+         
+         /* Populate the hardware storage if the host surface pre-existed */
+         if((usage & PIPE_BUFFER_USAGE_CPU_READ) && handle) {
+            SVGA3dSurfaceDMAFlags flags;
+            enum pipe_error ret;
+            struct pipe_fence_handle *fence = NULL;
+            
+            SVGA_DBG(DEBUG_DMA|DEBUG_PERF, "dma from sid %p, bytes %u - %u\n", 
+                     sbuf->handle, 0, sbuf->base.size);
+
+            memset(&flags, 0, sizeof flags);
+            
+            ret = SVGA3D_BufferDMA(ss->swc,
+                                   sbuf->hw.buf,
+                                   sbuf->handle,
+                                   SVGA3D_READ_HOST_VRAM,
+                                   sbuf->base.size,
+                                   0,
+                                   flags);
+            if(ret != PIPE_OK) {
+               ss->swc->flush(ss->swc, NULL);
+               
+               ret = SVGA3D_BufferDMA(ss->swc,
+                                      sbuf->hw.buf,
+                                      sbuf->handle,
+                                      SVGA3D_READ_HOST_VRAM,
+                                      sbuf->base.size,
+                                      0,
+                                      flags);
+               assert(ret == PIPE_OK);
+            }
+            
+            ss->swc->flush(ss->swc, &fence);
+            sws->fence_finish(sws, fence, 0);
+            sws->fence_reference(sws, &fence, NULL);
+         }
+      }
+      else {
+         if((usage & PIPE_BUFFER_USAGE_CPU_READ) && !sbuf->needs_flush) {
+            /* We already had the hardware storage but we would have to issue
+             * a download if we hadn't, so move the buffer to the begginning
+             * of the LRU list.
+             */
+            assert(sbuf->head.prev && sbuf->head.next);
+            LIST_DEL(&sbuf->head);
+            LIST_ADD(&sbuf->head, &ss->cached_buffers);
+         }
+      }
+         
+      map = sws->buffer_map(sws, sbuf->hw.buf, usage);
+   }
+
+   if(map) {
+      pipe_mutex_lock(ss->swc_mutex);
+
+      ++sbuf->map.count;
+
+      if (usage & PIPE_BUFFER_USAGE_CPU_WRITE) {
+         assert(sbuf->map.count <= 1);
+         sbuf->map.writing = TRUE;
+         if (usage & PIPE_BUFFER_USAGE_FLUSH_EXPLICIT)
+            sbuf->map.flush_explicit = TRUE;
+      }
+      
+      pipe_mutex_unlock(ss->swc_mutex);
+   }
+   
+   return map;
+}
+
+static void 
+svga_buffer_flush_mapped_range( struct pipe_screen *screen,
+                                struct pipe_buffer *buf,
+                                unsigned offset, unsigned length)
+{
+   struct svga_buffer *sbuf = svga_buffer( buf );
+   struct svga_screen *ss = svga_screen(screen);
+   
+   pipe_mutex_lock(ss->swc_mutex);
+   assert(sbuf->map.writing);
+   if(sbuf->map.writing) {
+      assert(sbuf->map.flush_explicit);
+      if(sbuf->hw.buf)
+         svga_buffer_upload_queue(sbuf, offset, offset + length);
+   }
+   pipe_mutex_unlock(ss->swc_mutex);
+}
+
+static void 
+svga_buffer_unmap( struct pipe_screen *screen,
+                   struct pipe_buffer *buf)
+{
+   struct svga_screen *ss = svga_screen(screen); 
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_buffer *sbuf = svga_buffer( buf );
+   
+   pipe_mutex_lock(ss->swc_mutex);
+   
+   assert(sbuf->map.count);
+   if(sbuf->map.count)
+      --sbuf->map.count;
+
+   if(sbuf->hw.buf)
+      sws->buffer_unmap(sws, sbuf->hw.buf);
+
+   if(sbuf->map.writing) {
+      if(!sbuf->map.flush_explicit) {
+         /* No mapped range was flushed -- flush the whole buffer */
+         SVGA_DBG(DEBUG_DMA, "flushing the whole buffer\n");
+   
+         if(sbuf->hw.buf)
+            svga_buffer_upload_queue(sbuf, 0, sbuf->base.size);
+      }
+      
+      sbuf->map.writing = FALSE;
+      sbuf->map.flush_explicit = FALSE;
+   }
+
+   pipe_mutex_unlock(ss->swc_mutex);
+}
+
+static void
+svga_buffer_destroy( struct pipe_buffer *buf )
+{
+   struct svga_screen *ss = svga_screen(buf->screen); 
+   struct svga_buffer *sbuf = svga_buffer( buf );
+
+   assert(!p_atomic_read(&buf->reference.count));
+   
+   assert(!sbuf->needs_flush);
+
+   if(sbuf->handle) {
+      SVGA_DBG(DEBUG_DMA, "release sid %p sz %d\n", sbuf->handle, sbuf->base.size);
+      svga_screen_surface_destroy(ss, &sbuf->key, &sbuf->handle);
+   }
+   
+   if(sbuf->hw.buf)
+      svga_buffer_destroy_hw_storage(ss, sbuf);
+   
+   if(sbuf->swbuf && !sbuf->user)
+      align_free(sbuf->swbuf);
+   
+   FREE(sbuf);
+}
+
+static struct pipe_buffer *
+svga_buffer_create(struct pipe_screen *screen,
+                   unsigned alignment,
+                   unsigned usage,
+                   unsigned size)
+{
+   struct svga_screen *ss = svga_screen(screen);
+   struct svga_buffer *sbuf;
+   
+   sbuf = CALLOC_STRUCT(svga_buffer);
+   if(!sbuf)
+      goto error1;
+      
+   sbuf->magic = SVGA_BUFFER_MAGIC;
+   
+   pipe_reference_init(&sbuf->base.reference, 1);
+   sbuf->base.screen = screen;
+   sbuf->base.alignment = alignment;
+   sbuf->base.usage = usage;
+   sbuf->base.size = size;
+
+   if(svga_buffer_needs_hw_storage(usage)) {
+      if(svga_buffer_create_host_surface(ss, sbuf) != PIPE_OK)
+         goto error2;
+   }
+   else {
+      if(alignment < sizeof(void*))
+         alignment = sizeof(void*);
+
+      usage |= PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+      
+      sbuf->swbuf = align_malloc(size, alignment);
+      if(!sbuf->swbuf)
+         goto error2;
+   }
+      
+   return &sbuf->base; 
+
+error2:
+   FREE(sbuf);
+error1:
+   return NULL;
+}
+
+static struct pipe_buffer *
+svga_user_buffer_create(struct pipe_screen *screen,
+                        void *ptr,
+                        unsigned bytes)
+{
+   struct svga_buffer *sbuf;
+   
+   sbuf = CALLOC_STRUCT(svga_buffer);
+   if(!sbuf)
+      goto no_sbuf;
+      
+   sbuf->magic = SVGA_BUFFER_MAGIC;
+   
+   sbuf->swbuf = ptr;
+   sbuf->user = TRUE;
+   
+   pipe_reference_init(&sbuf->base.reference, 1);
+   sbuf->base.screen = screen;
+   sbuf->base.alignment = 1;
+   sbuf->base.usage = 0;
+   sbuf->base.size = bytes;
+   
+   return &sbuf->base; 
+
+no_sbuf:
+   return NULL;
+}
+
+   
+void
+svga_screen_init_buffer_functions(struct pipe_screen *screen)
+{
+   screen->buffer_create = svga_buffer_create;
+   screen->user_buffer_create = svga_user_buffer_create;
+   screen->buffer_map_range = svga_buffer_map_range;
+   screen->buffer_flush_mapped_range = svga_buffer_flush_mapped_range;
+   screen->buffer_unmap = svga_buffer_unmap;
+   screen->buffer_destroy = svga_buffer_destroy;
+}
+
+
+/** 
+ * Copy the contents of the user buffer / malloc buffer to a hardware buffer.
+ */
+static INLINE enum pipe_error
+svga_buffer_update_hw(struct svga_screen *ss, struct svga_buffer *sbuf)
+{
+   if(!sbuf->hw.buf) {
+      enum pipe_error ret;
+      void *map;
+      
+      assert(sbuf->swbuf);
+      if(!sbuf->swbuf)
+         return PIPE_ERROR;
+      
+      ret = svga_buffer_create_hw_storage(ss, sbuf);
+      assert(ret == PIPE_OK);
+      if(ret != PIPE_OK)
+         return ret;
+
+      pipe_mutex_lock(ss->swc_mutex);
+      map = ss->sws->buffer_map(ss->sws, sbuf->hw.buf, PIPE_BUFFER_USAGE_CPU_WRITE);
+      assert(map);
+      if(!map) {
+	 pipe_mutex_unlock(ss->swc_mutex);
+         return PIPE_ERROR_OUT_OF_MEMORY;
+      }
+
+      memcpy(map, sbuf->swbuf, sbuf->base.size);
+      ss->sws->buffer_unmap(ss->sws, sbuf->hw.buf);
+
+      /* This user/malloc buffer is now indistinguishable from a gpu buffer */
+      assert(!sbuf->map.count);
+      if(!sbuf->map.count) {
+         if(sbuf->user)
+            sbuf->user = FALSE;
+         else
+            align_free(sbuf->swbuf);
+         sbuf->swbuf = NULL;
+      }
+      
+      svga_buffer_upload_queue(sbuf, 0, sbuf->base.size);
+   }
+   
+   pipe_mutex_unlock(ss->swc_mutex);
+   return PIPE_OK;
+}
+
+
+struct svga_winsys_surface *
+svga_buffer_handle(struct svga_context *svga,
+                   struct pipe_buffer *buf)
+{
+   struct pipe_screen *screen = svga->pipe.screen;
+   struct svga_screen *ss = svga_screen(screen);
+   struct svga_buffer *sbuf;
+   enum pipe_error ret;
+
+   if(!buf)
+      return NULL;
+
+   sbuf = svga_buffer(buf);
+   
+   assert(!sbuf->map.count);
+   
+   if(!sbuf->handle) {
+      ret = svga_buffer_create_host_surface(ss, sbuf);
+      if(ret != PIPE_OK)
+	 return NULL;
+
+      ret = svga_buffer_update_hw(ss, sbuf);
+      if(ret != PIPE_OK)
+	 return NULL;
+   }
+
+   if(!sbuf->needs_flush && sbuf->hw.num_ranges) {
+      /* Queue the buffer for flushing */
+      ret = svga_buffer_upload_command(svga, sbuf);
+      if(ret != PIPE_OK)
+         /* XXX: Should probably have a richer return value */
+         return NULL;
+
+      assert(sbuf->hw.svga == svga);
+
+      sbuf->needs_flush = TRUE;
+      assert(sbuf->head.prev && sbuf->head.next);
+      LIST_DEL(&sbuf->head);
+      LIST_ADDTAIL(&sbuf->head, &svga->dirty_buffers);
+   }
+
+   return sbuf->handle;
+}
+
+struct pipe_buffer *
+svga_screen_buffer_wrap_surface(struct pipe_screen *screen,
+				enum SVGA3dSurfaceFormat format,
+				struct svga_winsys_surface *srf)
+{
+   struct pipe_buffer *buf;
+   struct svga_buffer *sbuf;
+   struct svga_winsys_screen *sws = svga_winsys_screen(screen);
+
+   buf = svga_buffer_create(screen, 0, SVGA_BUFFER_USAGE_WRAPPED, 0);
+   if (!buf)
+      return NULL;
+
+   sbuf = svga_buffer(buf);
+
+   /*
+    * We are not the creator of this surface and therefore we must not
+    * cache it for reuse. The caching code only caches SVGA3D_BUFFER surfaces
+    * so make sure this isn't one of those.
+    */
+
+   assert(format != SVGA3D_BUFFER);
+   sbuf->key.format = format;
+   sws->surface_reference(sws, &sbuf->handle, srf);
+
+   return buf;
+}
+
+
+struct svga_winsys_surface *
+svga_screen_buffer_get_winsys_surface(struct pipe_buffer *buffer)
+{
+   struct svga_winsys_screen *sws = svga_winsys_screen(buffer->screen);
+   struct svga_winsys_surface *vsurf = NULL;
+
+   sws->surface_reference(sws, &vsurf, svga_buffer(buffer)->handle);
+   return vsurf;
+}
+
+void
+svga_context_flush_buffers(struct svga_context *svga)
+{
+   struct list_head *curr, *next;
+   struct svga_buffer *sbuf;
+
+   curr = svga->dirty_buffers.next;
+   next = curr->next;
+   while(curr != &svga->dirty_buffers) {
+      sbuf = LIST_ENTRY(struct svga_buffer, curr, head);
+
+      assert(p_atomic_read(&sbuf->base.reference.count) != 0);
+      assert(sbuf->needs_flush);
+      
+      svga_buffer_upload_flush(svga, sbuf);
+
+      curr = next; 
+      next = curr->next;
+   }
+}
diff --git a/src/gallium/drivers/svga/svga_screen_buffer.h b/src/gallium/drivers/svga/svga_screen_buffer.h
new file mode 100644
index 0000000000..5d7af5a7c5
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_buffer.h
@@ -0,0 +1,190 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_BUFFER_H
+#define SVGA_BUFFER_H
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_state.h"
+
+#include "util/u_double_list.h"
+
+#include "svga_screen_cache.h"
+
+
+#define SVGA_BUFFER_MAGIC 0x344f9005
+
+/**
+ * Maximum number of discontiguous ranges
+ */
+#define SVGA_BUFFER_MAX_RANGES 32
+
+
+struct svga_screen;
+struct svga_context;
+struct svga_winsys_buffer;
+struct svga_winsys_surface;
+
+
+struct svga_buffer_range
+{
+   unsigned start;
+   unsigned end;
+};
+
+
+/**
+ * Describe a
+ *
+ * This holds the information to emit a SVGA3dCmdSurfaceDMA.
+ */
+struct svga_buffer_upload
+{
+   /**
+    * Guest memory region.
+    */
+   struct svga_winsys_buffer *buf;
+
+   struct svga_buffer_range ranges[SVGA_BUFFER_MAX_RANGES];
+   unsigned num_ranges;
+
+   SVGA3dSurfaceDMAFlags flags;
+
+   /**
+    * Pointer to the DMA copy box *inside* the command buffer.
+    */
+   SVGA3dCopyBox *boxes;
+
+   /**
+    * Context that has the pending DMA to this buffer.
+    */
+   struct svga_context *svga;
+};
+
+
+/**
+ * SVGA pipe buffer.
+ */
+struct svga_buffer 
+{
+   struct pipe_buffer base;
+
+   /** 
+    * Marker to detect bad casts in runtime.
+    */ 
+   uint32_t magic;
+
+   /**
+    * Regular (non DMA'able) memory.
+    * 
+    * Used for user buffers or for buffers which we know before hand that can
+    * never be used by the virtual hardware directly, such as constant buffers.
+    */
+   void *swbuf;
+   
+   /** 
+    * Whether swbuf was created by the user or not.
+    */
+   boolean user;
+   
+   /**
+    * DMA'ble memory.
+    * 
+    * A piece of GMR memory. It is created when mapping the buffer, and will be
+    * used to upload/download vertex data from the host.
+    */
+   struct svga_buffer_upload hw;
+
+   /**
+    * Creation key for the host surface handle.
+    * 
+    * This structure describes all the host surface characteristics so that it 
+    * can be looked up in cache, since creating a host surface is often a slow
+    * operation.
+    */
+   struct svga_host_surface_cache_key key;
+   
+   /**
+    * Host surface handle.
+    * 
+    * This is a platform independent abstraction for host SID. We create when 
+    * trying to bind
+    */
+   struct svga_winsys_surface *handle;
+   
+   struct {
+      unsigned count;
+      boolean writing;
+      boolean flush_explicit;
+   } map;
+   
+   boolean needs_flush;
+   struct list_head head;
+};
+
+
+static INLINE struct svga_buffer *
+svga_buffer(struct pipe_buffer *buffer)
+{
+   if (buffer) {
+      assert(((struct svga_buffer *)buffer)->magic == SVGA_BUFFER_MAGIC);
+      return (struct svga_buffer *)buffer;
+   }
+   return NULL;
+}
+
+
+/**
+ * Returns TRUE for user buffers.  We may
+ * decide to use an alternate upload path for these buffers.
+ */
+static INLINE boolean 
+svga_buffer_is_user_buffer( struct pipe_buffer *buffer )
+{
+   return svga_buffer(buffer)->user;
+}
+
+
+void
+svga_screen_init_buffer_functions(struct pipe_screen *screen);
+
+struct svga_winsys_surface *
+svga_buffer_handle(struct svga_context *svga,
+                   struct pipe_buffer *buf);
+
+void
+svga_context_flush_buffers(struct svga_context *svga);
+
+boolean
+svga_buffer_free_cached_hw_storage(struct svga_screen *ss);
+
+struct svga_winsys_buffer *
+svga_winsys_buffer_create(struct svga_screen *ss,
+                          unsigned alignment, 
+                          unsigned usage,
+                          unsigned size);
+
+#endif /* SVGA_BUFFER_H */
diff --git a/src/gallium/drivers/svga/svga_screen_cache.c b/src/gallium/drivers/svga/svga_screen_cache.c
new file mode 100644
index 0000000000..7360c1688b
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_cache.c
@@ -0,0 +1,307 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "util/u_memory.h"
+
+#include "svga_debug.h"
+#include "svga_winsys.h"
+#include "svga_screen.h"
+#include "svga_screen_cache.h"
+
+
+#define SVGA_SURFACE_CACHE_ENABLED 1
+
+
+/** 
+ * Compute the bucket for this key. 
+ * 
+ * We simply compute log2(width) for now, but
+ */
+static INLINE unsigned
+svga_screen_cache_bucket(const struct svga_host_surface_cache_key *key)
+{
+   unsigned bucket = 0;
+   unsigned size = key->size.width;
+   
+   while ((size >>= 1))
+      ++bucket;
+   
+   if(key->flags & SVGA3D_SURFACE_HINT_INDEXBUFFER)
+      bucket += 32;
+   
+   assert(bucket < SVGA_HOST_SURFACE_CACHE_BUCKETS);
+   
+   return bucket;
+}
+
+
+static INLINE struct svga_winsys_surface *
+svga_screen_cache_lookup(struct svga_screen *svgascreen,
+                         const struct svga_host_surface_cache_key *key)
+{
+   struct svga_host_surface_cache *cache = &svgascreen->cache;
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_host_surface_cache_entry *entry;
+   struct svga_winsys_surface *handle = NULL;
+   struct list_head *curr, *next;
+   unsigned bucket;
+   unsigned tries = 0;
+
+   bucket = svga_screen_cache_bucket(key);
+
+   pipe_mutex_lock(cache->mutex);
+
+   curr = cache->bucket[bucket].next;
+   next = curr->next;
+   while(curr != &cache->bucket[bucket]) {
+      ++tries;
+      
+      entry = LIST_ENTRY(struct svga_host_surface_cache_entry, curr, bucket_head);
+
+      assert(entry->handle);
+      
+      if(memcmp(&entry->key, key, sizeof *key) == 0 &&
+         sws->fence_signalled( sws, entry->fence, 0 ) == 0) {
+         assert(sws->surface_is_flushed(sws, entry->handle));
+         
+         handle = entry->handle; // Reference is transfered here.
+         entry->handle = NULL;
+         
+         LIST_DEL(&entry->bucket_head);
+
+         LIST_DEL(&entry->head);
+         
+         LIST_ADD(&entry->head, &cache->empty);
+
+         break;
+      }
+
+      curr = next; 
+      next = curr->next;
+   }
+
+   pipe_mutex_unlock(cache->mutex);
+   
+#if 0
+   _debug_printf("%s: cache %s after %u tries\n", __FUNCTION__, handle ? "hit" : "miss", tries);
+#else
+   (void)tries;
+#endif
+   
+   return handle;
+}
+
+
+/*
+ * Transfers a handle reference.
+ */
+                           
+static INLINE void
+svga_screen_cache_add(struct svga_screen *svgascreen,
+                      const struct svga_host_surface_cache_key *key, 
+                      struct svga_winsys_surface **p_handle)
+{
+   struct svga_host_surface_cache *cache = &svgascreen->cache;
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_host_surface_cache_entry *entry = NULL;
+   struct svga_winsys_surface *handle = *p_handle;
+   
+
+   assert(handle);
+   if(!handle)
+      return;
+   
+   *p_handle = NULL;
+   pipe_mutex_lock(cache->mutex);
+   
+   if(!LIST_IS_EMPTY(&cache->empty)) {
+        /* use the first empty entry */
+        entry = LIST_ENTRY(struct svga_host_surface_cache_entry, cache->empty.next, head);
+        
+        LIST_DEL(&entry->head);
+     }
+   else if(!LIST_IS_EMPTY(&cache->unused)) {
+      /* free the last used buffer and reuse its entry */
+      entry = LIST_ENTRY(struct svga_host_surface_cache_entry, cache->unused.prev, head);
+      SVGA_DBG(DEBUG_DMA, "unref sid %p\n", entry->handle);
+      sws->surface_reference(sws, &entry->handle, NULL);
+
+      LIST_DEL(&entry->bucket_head);
+
+      LIST_DEL(&entry->head);
+   }
+
+   if(entry) {
+      entry->handle = handle;
+      memcpy(&entry->key, key, sizeof entry->key);
+   
+      LIST_ADD(&entry->head, &cache->validated);
+   }
+   else {
+      /* Couldn't cache the buffer -- this really shouldn't happen */
+      SVGA_DBG(DEBUG_DMA, "unref sid %p\n", handle);
+      sws->surface_reference(sws, &handle, NULL);
+   }
+   
+   pipe_mutex_unlock(cache->mutex);
+}
+
+
+/**
+ * Called during the screen flush to move all buffers not in a validate list
+ * into the unused list.
+ */
+void
+svga_screen_cache_flush(struct svga_screen *svgascreen,
+                        struct pipe_fence_handle *fence)
+{
+   struct svga_host_surface_cache *cache = &svgascreen->cache;
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_host_surface_cache_entry *entry;
+   struct list_head *curr, *next;
+   unsigned bucket;
+
+   pipe_mutex_lock(cache->mutex);
+
+   curr = cache->validated.next;
+   next = curr->next;
+   while(curr != &cache->validated) {
+      entry = LIST_ENTRY(struct svga_host_surface_cache_entry, curr, head);
+
+      assert(entry->handle);
+
+      if(sws->surface_is_flushed(sws, entry->handle)) {
+         LIST_DEL(&entry->head);
+         
+         svgascreen->sws->fence_reference(svgascreen->sws, &entry->fence, fence);
+
+         LIST_ADD(&entry->head, &cache->unused);
+
+         bucket = svga_screen_cache_bucket(&entry->key);
+         LIST_ADD(&entry->bucket_head, &cache->bucket[bucket]);
+      }
+
+      curr = next; 
+      next = curr->next;
+   }
+
+   pipe_mutex_unlock(cache->mutex);
+}
+
+
+void
+svga_screen_cache_cleanup(struct svga_screen *svgascreen)
+{
+   struct svga_host_surface_cache *cache = &svgascreen->cache;
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   unsigned i;
+   
+   for(i = 0; i < SVGA_HOST_SURFACE_CACHE_SIZE; ++i) {
+      if(cache->entries[i].handle) {
+	 SVGA_DBG(DEBUG_DMA, "unref sid %p\n", cache->entries[i].handle);
+	 sws->surface_reference(sws, &cache->entries[i].handle, NULL);
+      }
+
+      if(cache->entries[i].fence)
+         svgascreen->sws->fence_reference(svgascreen->sws, &cache->entries[i].fence, NULL);
+   }
+   
+   pipe_mutex_destroy(cache->mutex);
+}
+
+
+enum pipe_error
+svga_screen_cache_init(struct svga_screen *svgascreen)
+{
+   struct svga_host_surface_cache *cache = &svgascreen->cache;
+   unsigned i;
+
+   pipe_mutex_init(cache->mutex);
+   
+   for(i = 0; i < SVGA_HOST_SURFACE_CACHE_BUCKETS; ++i)
+      LIST_INITHEAD(&cache->bucket[i]);
+
+   LIST_INITHEAD(&cache->unused);
+   
+   LIST_INITHEAD(&cache->validated);
+   
+   LIST_INITHEAD(&cache->empty);
+   for(i = 0; i < SVGA_HOST_SURFACE_CACHE_SIZE; ++i)
+      LIST_ADDTAIL(&cache->entries[i].head, &cache->empty);
+
+   return PIPE_OK;
+}
+
+                           
+struct svga_winsys_surface *
+svga_screen_surface_create(struct svga_screen *svgascreen,
+                           struct svga_host_surface_cache_key *key)
+{
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_winsys_surface *handle = NULL;
+
+   if (SVGA_SURFACE_CACHE_ENABLED && key->format == SVGA3D_BUFFER) {
+      /* round the buffer size up to the nearest power of two to increase the
+       * probability of cache hits */
+      uint32_t size = 1;
+      while(size < key->size.width)
+         size <<= 1;
+      key->size.width = size;
+      
+      handle = svga_screen_cache_lookup(svgascreen, key);
+      if (handle)
+         SVGA_DBG(DEBUG_DMA, "  reuse sid %p sz %d\n", handle, size);
+   }
+
+   if (!handle) {
+      handle = sws->surface_create(sws,
+                                   key->flags,
+                                   key->format,
+                                   key->size, 
+                                   key->numFaces, 
+                                   key->numMipLevels);
+      if (handle)
+         SVGA_DBG(DEBUG_DMA, "create sid %p sz %d\n", handle, key->size);
+   }
+
+   return handle;
+}
+
+
+void
+svga_screen_surface_destroy(struct svga_screen *svgascreen,
+                            const struct svga_host_surface_cache_key *key,
+                            struct svga_winsys_surface **p_handle)
+{
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   
+   if(SVGA_SURFACE_CACHE_ENABLED && key->format == SVGA3D_BUFFER) {
+      svga_screen_cache_add(svgascreen, key, p_handle);
+   }
+   else {
+      SVGA_DBG(DEBUG_DMA, "unref sid %p\n", *p_handle);
+      sws->surface_reference(sws, p_handle, NULL);
+   }
+}
diff --git a/src/gallium/drivers/svga/svga_screen_cache.h b/src/gallium/drivers/svga/svga_screen_cache.h
new file mode 100644
index 0000000000..1bbe987768
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_cache.h
@@ -0,0 +1,135 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_SCREEN_CACHE_H_
+#define SVGA_SCREEN_CACHE_H_
+
+
+#include "svga_types.h"
+#include "svga_reg.h"
+#include "svga3d_reg.h"
+
+#include "pipe/p_thread.h"
+
+#include "util/u_double_list.h"
+
+
+/* TODO: Reduce this once we don't allocate an index buffer per draw call */ 
+#define SVGA_HOST_SURFACE_CACHE_SIZE 1024
+
+#define SVGA_HOST_SURFACE_CACHE_BUCKETS 64
+
+
+struct svga_winsys_surface;
+struct svga_screen;
+
+/**
+ * Same as svga_winsys_screen::surface_create.
+ */
+struct svga_host_surface_cache_key
+{
+   SVGA3dSurfaceFlags flags;
+   SVGA3dSurfaceFormat format;
+   SVGA3dSize size;
+   uint32_t numFaces;
+   uint32_t numMipLevels;
+};
+
+
+struct svga_host_surface_cache_entry 
+{
+   /** 
+    * Head for the LRU list, svga_host_surface_cache::unused, and
+    * svga_host_surface_cache::empty
+    */
+   struct list_head head;
+   
+   /** Head for the bucket lists. */
+   struct list_head bucket_head;
+
+   struct svga_host_surface_cache_key key;
+   struct svga_winsys_surface *handle;
+   
+   struct pipe_fence_handle *fence;
+};
+
+
+/**
+ * Cache of the host surfaces.
+ * 
+ * A cache entry can be in the following stages:
+ * 1. empty
+ * 2. holding a buffer in a validate list
+ * 3. holding a flushed buffer (not in any validate list) with an active fence
+ * 4. holding a flushed buffer with an expired fence
+ * 
+ * An entry progresses from 1 -> 2 -> 3 -> 4. When we need an entry to put a 
+ * buffer into we preferencial take from 1, or from the least recentely used 
+ * buffer from 3/4.
+ */
+struct svga_host_surface_cache 
+{
+   pipe_mutex mutex;
+   
+   /* Unused buffers are put in buckets to speed up lookups */
+   struct list_head bucket[SVGA_HOST_SURFACE_CACHE_BUCKETS];
+   
+   /* Entries with unused buffers, ordered from most to least recently used 
+    * (3 and 4) */
+   struct list_head unused;
+   
+   /* Entries with buffers still in validate lists (2) */
+   struct list_head validated;
+   
+   /** Empty entries (1) */
+   struct list_head empty;
+
+   /** The actual storage for the entries */
+   struct svga_host_surface_cache_entry entries[SVGA_HOST_SURFACE_CACHE_SIZE];
+};
+
+
+void
+svga_screen_cache_cleanup(struct svga_screen *svgascreen);
+
+void
+svga_screen_cache_flush(struct svga_screen *svgascreen,
+                        struct pipe_fence_handle *fence);
+
+enum pipe_error
+svga_screen_cache_init(struct svga_screen *svgascreen);
+
+
+struct svga_winsys_surface *
+svga_screen_surface_create(struct svga_screen *svgascreen,
+                           struct svga_host_surface_cache_key *key);
+
+void
+svga_screen_surface_destroy(struct svga_screen *svgascreen,
+                            const struct svga_host_surface_cache_key *key,
+                            struct svga_winsys_surface **handle);
+
+
+#endif /* SVGA_SCREEN_CACHE_H_ */
diff --git a/src/gallium/drivers/svga/svga_screen_texture.c b/src/gallium/drivers/svga/svga_screen_texture.c
new file mode 100644
index 0000000000..fb11b80dcf
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_texture.c
@@ -0,0 +1,1064 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "svga_cmd.h"
+
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_thread.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#include "svga_screen.h"
+#include "svga_context.h"
+#include "svga_screen_texture.h"
+#include "svga_screen_buffer.h"
+#include "svga_winsys.h"
+#include "svga_debug.h"
+#include "svga_screen_buffer.h"
+
+#include <util/u_string.h>
+
+
+/* XXX: This isn't a real hardware flag, but just a hack for kernel to
+ * know about primary surfaces. Find a better way to accomplish this.
+ */
+#define SVGA3D_SURFACE_HINT_SCANOUT (1 << 9)
+
+
+/*
+ * Helper function and arrays
+ */
+
+SVGA3dSurfaceFormat
+svga_translate_format(enum pipe_format format)
+{
+   switch(format) {
+   
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+      return SVGA3D_A8R8G8B8;
+   case PIPE_FORMAT_X8R8G8B8_UNORM:
+      return SVGA3D_X8R8G8B8;
+
+      /* Required for GL2.1:
+       */
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      return SVGA3D_A8R8G8B8;
+
+   case PIPE_FORMAT_R5G6B5_UNORM:
+      return SVGA3D_R5G6B5;
+   case PIPE_FORMAT_A1R5G5B5_UNORM:
+      return SVGA3D_A1R5G5B5;
+   case PIPE_FORMAT_A4R4G4B4_UNORM:
+      return SVGA3D_A4R4G4B4;
+
+      
+   /* XXX: Doesn't seem to work properly.
+   case PIPE_FORMAT_Z32_UNORM:
+      return SVGA3D_Z_D32;
+    */
+   case PIPE_FORMAT_Z16_UNORM:
+      return SVGA3D_Z_D16;
+   case PIPE_FORMAT_Z24S8_UNORM:
+      return SVGA3D_Z_D24S8;
+   case PIPE_FORMAT_Z24X8_UNORM:
+      return SVGA3D_Z_D24X8;
+
+   case PIPE_FORMAT_A8_UNORM:
+      return SVGA3D_ALPHA8;
+   case PIPE_FORMAT_L8_UNORM:
+      return SVGA3D_LUMINANCE8;
+
+   case PIPE_FORMAT_DXT1_RGB:
+   case PIPE_FORMAT_DXT1_RGBA:
+      return SVGA3D_DXT1;
+   case PIPE_FORMAT_DXT3_RGBA:
+      return SVGA3D_DXT3;
+   case PIPE_FORMAT_DXT5_RGBA:
+      return SVGA3D_DXT5;
+
+   default:
+      return SVGA3D_FORMAT_INVALID;
+   }
+}
+
+
+SVGA3dSurfaceFormat
+svga_translate_format_render(enum pipe_format format)
+{
+   switch(format) { 
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_X8R8G8B8_UNORM:
+   case PIPE_FORMAT_A1R5G5B5_UNORM:
+   case PIPE_FORMAT_A4R4G4B4_UNORM:
+   case PIPE_FORMAT_R5G6B5_UNORM:
+   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24X8_UNORM:
+   case PIPE_FORMAT_Z32_UNORM:
+   case PIPE_FORMAT_Z16_UNORM:
+   case PIPE_FORMAT_L8_UNORM:
+      return svga_translate_format(format);
+
+#if 1
+   /* For on host conversion */
+   case PIPE_FORMAT_DXT1_RGB:
+      return SVGA3D_X8R8G8B8;
+   case PIPE_FORMAT_DXT1_RGBA:
+   case PIPE_FORMAT_DXT3_RGBA:
+   case PIPE_FORMAT_DXT5_RGBA:
+      return SVGA3D_A8R8G8B8;
+#endif
+
+   default:
+      return SVGA3D_FORMAT_INVALID;
+   }
+}
+
+
+static INLINE void
+svga_transfer_dma_band(struct svga_transfer *st,
+                       SVGA3dTransferType transfer,
+                       unsigned y, unsigned h, unsigned srcy)
+{
+   struct svga_texture *texture = svga_texture(st->base.texture); 
+   struct svga_screen *screen = svga_screen(texture->base.screen);
+   SVGA3dCopyBox box;
+   enum pipe_error ret;
+   
+   SVGA_DBG(DEBUG_DMA, "dma %s sid %p, face %u, (%u, %u, %u) - (%u, %u, %u), %ubpp\n",
+                transfer == SVGA3D_WRITE_HOST_VRAM ? "to" : "from", 
+                texture->handle,
+                st->base.face,
+                st->base.x,
+                y,
+                st->base.zslice,
+                st->base.x + st->base.width,
+                y + h,
+                st->base.zslice + 1,
+                texture->base.block.size*8/(texture->base.block.width*texture->base.block.height));
+   
+   box.x = st->base.x;
+   box.y = y;
+   box.z = st->base.zslice;
+   box.w = st->base.width;
+   box.h = h;
+   box.d = 1;
+   box.srcx = 0;
+   box.srcy = srcy;
+   box.srcz = 0;
+
+   pipe_mutex_lock(screen->swc_mutex);
+   ret = SVGA3D_SurfaceDMA(screen->swc, st, transfer, &box, 1);
+   if(ret != PIPE_OK) {
+      screen->swc->flush(screen->swc, NULL);
+      ret = SVGA3D_SurfaceDMA(screen->swc, st, transfer, &box, 1);
+      assert(ret == PIPE_OK);
+   }
+   pipe_mutex_unlock(screen->swc_mutex);
+}
+
+
+static INLINE void
+svga_transfer_dma(struct svga_transfer *st,
+                 SVGA3dTransferType transfer)
+{
+   struct svga_texture *texture = svga_texture(st->base.texture); 
+   struct svga_screen *screen = svga_screen(texture->base.screen);
+   struct svga_winsys_screen *sws = screen->sws;
+   struct pipe_fence_handle *fence = NULL;
+   
+   if (transfer == SVGA3D_READ_HOST_VRAM) {
+      SVGA_DBG(DEBUG_PERF, "%s: readback transfer\n", __FUNCTION__);
+   }
+
+
+   if(!st->swbuf) {
+      /* Do the DMA transfer in a single go */
+      
+      svga_transfer_dma_band(st, transfer, st->base.y, st->base.height, 0);
+
+      if(transfer == SVGA3D_READ_HOST_VRAM) {
+         svga_screen_flush(screen, &fence);
+         sws->fence_finish(sws, fence, 0);
+         //sws->fence_reference(sws, &fence, NULL);
+      }
+   }
+   else {
+      unsigned y, h, srcy;
+      h = st->hw_nblocksy * st->base.block.height;
+      srcy = 0;
+      for(y = 0; y < st->base.height; y += h) {
+         unsigned offset, length;
+         void *hw, *sw;
+
+         if (y + h > st->base.height)
+            h = st->base.height - y;
+
+         /* Transfer band must be aligned to pixel block boundaries */
+         assert(y % st->base.block.height == 0);
+         assert(h % st->base.block.height == 0);
+         
+         offset = y * st->base.stride / st->base.block.height;
+         length = h * st->base.stride / st->base.block.height;
+
+         sw = (uint8_t *)st->swbuf + offset;
+         
+         if(transfer == SVGA3D_WRITE_HOST_VRAM) {
+            /* Wait for the previous DMAs to complete */
+            /* TODO: keep one DMA (at half the size) in the background */
+            if(y) {
+               svga_screen_flush(screen, &fence);
+               sws->fence_finish(sws, fence, 0);
+               //sws->fence_reference(sws, &fence, NULL);
+            }
+
+            hw = sws->buffer_map(sws, st->hwbuf, PIPE_BUFFER_USAGE_CPU_WRITE);
+            assert(hw);
+            if(hw) {
+               memcpy(hw, sw, length);
+               sws->buffer_unmap(sws, st->hwbuf);
+            }
+         }
+         
+         svga_transfer_dma_band(st, transfer, y, h, srcy);
+         
+         if(transfer == SVGA3D_READ_HOST_VRAM) {
+            svga_screen_flush(screen, &fence);
+            sws->fence_finish(sws, fence, 0);
+
+            hw = sws->buffer_map(sws, st->hwbuf, PIPE_BUFFER_USAGE_CPU_READ);
+            assert(hw);
+            if(hw) {
+               memcpy(sw, hw, length);
+               sws->buffer_unmap(sws, st->hwbuf);
+            }
+         }
+      }
+   }
+}
+
+
+static struct pipe_texture *
+svga_texture_create(struct pipe_screen *screen,
+                    const struct pipe_texture *templat)
+{
+   struct svga_screen *svgascreen = svga_screen(screen);
+   struct svga_winsys_screen *sws = svgascreen->sws;
+   struct svga_texture *tex = CALLOC_STRUCT(svga_texture);
+   unsigned width, height, depth;
+   SVGA3dSurfaceFlags flags = 0;
+   SVGA3dSurfaceFormat format;
+   SVGA3dSize size;
+   uint32 numFaces;
+   uint32 numMipLevels;
+   unsigned level;
+   
+   if (!tex)
+      goto error1;
+
+   tex->base = *templat;
+   pipe_reference_init(&tex->base.reference, 1);
+   tex->base.screen = screen;
+
+   assert(templat->last_level < SVGA_MAX_TEXTURE_LEVELS);
+   if(templat->last_level >= SVGA_MAX_TEXTURE_LEVELS)
+      goto error2;
+   
+   width = templat->width0;
+   height = templat->height0;
+   depth = templat->depth0;
+   for(level = 0; level <= templat->last_level; ++level) {
+      tex->base.nblocksx[level] = pf_get_nblocksx(&tex->base.block, width);  
+      tex->base.nblocksy[level] = pf_get_nblocksy(&tex->base.block, height);  
+      width = u_minify(width, 1);
+      height = u_minify(height, 1);
+      depth = u_minify(depth, 1);
+   }
+   
+   size.width = templat->width0;
+   size.height = templat->height0;
+   size.depth = templat->depth0;
+   
+   if(templat->target == PIPE_TEXTURE_CUBE) {
+      flags |= SVGA3D_SURFACE_CUBEMAP;
+      numFaces = 6;
+   }
+   else {
+      numFaces = 1;
+   }
+
+   if(templat->tex_usage & PIPE_TEXTURE_USAGE_SAMPLER)
+      flags |= SVGA3D_SURFACE_HINT_TEXTURE;
+
+   if(templat->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY)
+      flags |= SVGA3D_SURFACE_HINT_SCANOUT;
+   
+   /* 
+    * XXX: Never pass the SVGA3D_SURFACE_HINT_RENDERTARGET hint. Mesa cannot
+    * know beforehand whether a texture will be used as a rendertarget or not
+    * and it always requests PIPE_TEXTURE_USAGE_RENDER_TARGET, therefore
+    * passing the SVGA3D_SURFACE_HINT_RENDERTARGET here defeats its purpose.
+    */
+#if 0
+   if((templat->tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) &&
+      !pf_is_compressed(templat->format))
+      flags |= SVGA3D_SURFACE_HINT_RENDERTARGET;
+#endif
+   
+   if(templat->tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL)
+      flags |= SVGA3D_SURFACE_HINT_DEPTHSTENCIL;
+   
+   numMipLevels = templat->last_level + 1;
+   
+   format = svga_translate_format(templat->format);
+   if(format == SVGA3D_FORMAT_INVALID)
+      goto error2;
+   
+   tex->handle = sws->surface_create(sws, flags, format, size, numFaces, numMipLevels);
+   if (tex->handle)
+      SVGA_DBG(DEBUG_DMA, "create sid %p (texture)\n", tex->handle);
+
+   return &tex->base;
+
+error2:
+   FREE(tex);
+error1:
+   return NULL;
+}
+
+
+static struct pipe_texture *
+svga_texture_blanket(struct pipe_screen * screen,
+                     const struct pipe_texture *base,
+                     const unsigned *stride,
+                     struct pipe_buffer *buffer)
+{
+   struct svga_texture *tex;
+   struct svga_buffer *sbuf = svga_buffer(buffer);
+   struct svga_winsys_screen *sws = svga_winsys_screen(screen);
+   assert(screen);
+
+   /* Only supports one type */
+   if (base->target != PIPE_TEXTURE_2D ||
+       base->last_level != 0 ||
+       base->depth0 != 1) {
+      return NULL;
+   }
+
+   /**
+    * We currently can't do texture blanket on
+    * SVGA3D_BUFFER. Need to blit to a temporary surface?
+    */
+
+   assert(sbuf->handle);
+   if (!sbuf->handle)
+      return NULL;
+
+   if (svga_translate_format(base->format) != sbuf->key.format) {
+      unsigned f1 = svga_translate_format(base->format);
+      unsigned f2 = sbuf->key.format;
+
+      /* It's okay for XRGB and ARGB or depth with/out stencil to get mixed up */
+      if ( !( (f1 == SVGA3D_X8R8G8B8 && f2 == SVGA3D_A8R8G8B8) ||
+              (f1 == SVGA3D_A8R8G8B8 && f2 == SVGA3D_X8R8G8B8) ||
+              (f1 == SVGA3D_Z_D24X8 && f2 == SVGA3D_Z_D24S8) ) ) {
+         debug_printf("%s wrong format %u != %u\n", __FUNCTION__, f1, f2);
+         return NULL;
+      }
+   }
+
+   tex = CALLOC_STRUCT(svga_texture);
+   if (!tex)
+      return NULL;
+
+   tex->base = *base;
+
+   if (sbuf->key.format == 1)
+      tex->base.format = PIPE_FORMAT_X8R8G8B8_UNORM;
+   else if (sbuf->key.format == 2)
+      tex->base.format = PIPE_FORMAT_A8R8G8B8_UNORM;
+
+   pipe_reference_init(&tex->base.reference, 1);
+   tex->base.screen = screen;
+
+   sws->surface_reference(sws, &tex->handle, sbuf->handle);
+
+   return &tex->base;
+}
+
+
+static void
+svga_texture_destroy(struct pipe_texture *pt)
+{
+   struct svga_screen *ss = svga_screen(pt->screen);
+   struct svga_texture *tex = (struct svga_texture *)pt;
+
+   ss->texture_timestamp++;
+
+   svga_sampler_view_reference(&tex->cached_view, NULL);
+
+   /*
+     DBG("%s deleting %p\n", __FUNCTION__, (void *) tex);
+   */
+   SVGA_DBG(DEBUG_DMA, "unref sid %p (texture)\n", tex->handle);
+   ss->sws->surface_reference(ss->sws, &tex->handle, NULL);
+
+   FREE(tex);
+}
+
+
+static void
+svga_texture_copy_handle(struct svga_context *svga,
+                         struct svga_screen *ss,
+                         struct svga_winsys_surface *src_handle,
+                         unsigned src_x, unsigned src_y, unsigned src_z,
+                         unsigned src_level, unsigned src_face,
+                         struct svga_winsys_surface *dst_handle,
+                         unsigned dst_x, unsigned dst_y, unsigned dst_z,
+                         unsigned dst_level, unsigned dst_face,
+                         unsigned width, unsigned height, unsigned depth)
+{
+   struct svga_surface dst, src;
+   enum pipe_error ret;
+   SVGA3dCopyBox box, *boxes;
+
+   assert(svga || ss);
+
+   src.handle = src_handle;
+   src.real_level = src_level;
+   src.real_face = src_face;
+   src.real_zslice = 0;
+
+   dst.handle = dst_handle;
+   dst.real_level = dst_level;
+   dst.real_face = dst_face;
+   dst.real_zslice = 0;
+
+   box.x = dst_x;
+   box.y = dst_y;
+   box.z = dst_z;
+   box.w = width;
+   box.h = height;
+   box.d = depth;
+   box.srcx = src_x;
+   box.srcy = src_y;
+   box.srcz = src_z;
+
+/*
+   SVGA_DBG(DEBUG_VIEWS, "mipcopy src: %p %u (%ux%ux%u), dst: %p %u (%ux%ux%u)\n",
+            src_handle, src_level, src_x, src_y, src_z,
+            dst_handle, dst_level, dst_x, dst_y, dst_z);
+*/
+
+   if (svga) {
+      ret = SVGA3D_BeginSurfaceCopy(svga->swc,
+                                    &src.base,
+                                    &dst.base,
+                                    &boxes, 1);
+      if(ret != PIPE_OK) {
+         svga_context_flush(svga, NULL);
+         ret = SVGA3D_BeginSurfaceCopy(svga->swc,
+                                       &src.base,
+                                       &dst.base,
+                                       &boxes, 1);
+         assert(ret == PIPE_OK);
+      }
+      *boxes = box;
+      SVGA_FIFOCommitAll(svga->swc);
+   } else {
+      pipe_mutex_lock(ss->swc_mutex);
+      ret = SVGA3D_BeginSurfaceCopy(ss->swc,
+                                    &src.base,
+                                    &dst.base,
+                                    &boxes, 1);
+      if(ret != PIPE_OK) {
+         ss->swc->flush(ss->swc, NULL);
+         ret = SVGA3D_BeginSurfaceCopy(ss->swc,
+                                       &src.base,
+                                       &dst.base,
+                                       &boxes, 1);
+         assert(ret == PIPE_OK);
+      }
+      *boxes = box;
+      SVGA_FIFOCommitAll(ss->swc);
+      pipe_mutex_unlock(ss->swc_mutex);
+   }
+}
+
+static struct svga_winsys_surface *
+svga_texture_view_surface(struct pipe_context *pipe,
+                          struct svga_texture *tex,
+                          SVGA3dSurfaceFormat format,
+                          unsigned start_mip,
+                          unsigned num_mip,
+                          int face_pick,
+                          int zslice_pick)
+{
+   struct svga_screen *ss = svga_screen(tex->base.screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_winsys_surface *handle;
+   int i, j;
+   SVGA3dSurfaceFlags flags = 0;
+   SVGA3dSize size;
+   uint32 numFaces;
+   uint32 numMipLevels = num_mip;
+   unsigned z_offset = 0;
+
+   SVGA_DBG(DEBUG_PERF, 
+            "svga: Create surface view: face %d zslice %d mips %d..%d\n",
+            face_pick, zslice_pick, start_mip, start_mip+num_mip-1);
+
+   size.width = u_minify(tex->base.width0, start_mip);
+   size.height = u_minify(tex->base.height0, start_mip);
+   size.depth = zslice_pick < 0 ? u_minify(tex->base.depth0, start_mip) : 1;
+   assert(size.depth == 1);
+   
+   if(tex->base.target == PIPE_TEXTURE_CUBE && face_pick < 0) {
+      flags |= SVGA3D_SURFACE_CUBEMAP;
+      numFaces = 6;
+   } else {
+      numFaces = 1;
+   }
+
+   if(format == SVGA3D_FORMAT_INVALID)
+      return NULL;
+
+   handle = sws->surface_create(sws, flags, format, size, numFaces, numMipLevels);
+
+   if (!handle)
+      return NULL;
+
+   SVGA_DBG(DEBUG_DMA, "create sid %p (texture view)\n", handle);
+
+   if (face_pick < 0)
+      face_pick = 0;
+
+   if (zslice_pick >= 0)
+       z_offset = zslice_pick;
+
+   for (i = 0; i < num_mip; i++) {
+      for (j = 0; j < numFaces; j++) {
+         if(tex->defined[j + face_pick][i + start_mip]) {
+            unsigned depth = zslice_pick < 0 ? u_minify(tex->base.depth0, i + start_mip) : 1;
+            svga_texture_copy_handle(svga_context(pipe), ss,
+                                     tex->handle, 0, 0, z_offset, i + start_mip, j + face_pick,
+                                     handle, 0, 0, 0, i, j,
+                                     u_minify(tex->base.width0, i + start_mip),
+                                     u_minify(tex->base.height0, i + start_mip), depth);
+         }
+      }
+   }
+
+   return handle;
+}
+
+
+static struct pipe_surface *
+svga_get_tex_surface(struct pipe_screen *screen,
+                     struct pipe_texture *pt,
+                     unsigned face, unsigned level, unsigned zslice,
+                     unsigned flags)
+{
+   struct svga_texture *tex = svga_texture(pt);
+   struct svga_surface *s;
+   struct pipe_surface *ps;
+   boolean render = flags & PIPE_BUFFER_USAGE_GPU_WRITE ? TRUE : FALSE;
+   boolean view = FALSE;
+   SVGA3dSurfaceFormat format;
+
+   s = CALLOC_STRUCT(svga_surface);
+   ps = &s->base;
+   if (!ps)
+      return NULL;
+
+   pipe_reference_init(&ps->reference, 1);
+   pipe_texture_reference(&ps->texture, pt);
+   ps->format = pt->format;
+   ps->width = u_minify(pt->width0, level);
+   ps->height = u_minify(pt->height0, level);
+   ps->usage = flags;
+   ps->level = level;
+   ps->face = face;
+   ps->zslice = zslice;
+
+   if (!render)
+      format = svga_translate_format(pt->format);
+   else
+      format = svga_translate_format_render(pt->format);
+
+   assert(format != SVGA3D_FORMAT_INVALID);
+   assert(!(flags & PIPE_BUFFER_USAGE_CPU_READ_WRITE));
+
+
+   if (svga_screen(screen)->debug.force_surface_view)
+      view = TRUE;
+
+   /* Currently only used for compressed textures */
+   if (render && (format != svga_translate_format(pt->format))) {
+      view = TRUE;
+   }
+
+   if (level != 0 && svga_screen(screen)->debug.force_level_surface_view)
+      view = TRUE;
+
+   if (pt->target == PIPE_TEXTURE_3D)
+      view = TRUE;
+
+   if (svga_screen(screen)->debug.no_surface_view)
+      view = FALSE;
+
+   if (view) {
+      SVGA_DBG(DEBUG_VIEWS, "svga: Surface view: yes %p, level %u face %u z %u, %p\n",
+               pt, level, face, zslice, ps);
+
+      s->handle = svga_texture_view_surface(NULL, tex, format, level, 1, face, zslice);
+      s->real_face = 0;
+      s->real_level = 0;
+      s->real_zslice = 0;
+   } else {
+      struct svga_winsys_screen *sws = svga_winsys_screen(screen);
+
+      SVGA_DBG(DEBUG_VIEWS, "svga: Surface view: no %p, level %u, face %u, z %u, %p\n",
+               pt, level, face, zslice, ps);
+
+      sws->surface_reference(sws, &s->handle, tex->handle);
+      s->real_face = face;
+      s->real_level = level;
+      s->real_zslice = zslice;
+   }
+
+   return ps;
+}
+
+
+static void
+svga_tex_surface_destroy(struct pipe_surface *surf)
+{
+   struct svga_surface *s = svga_surface(surf);
+   struct svga_screen *ss = svga_screen(surf->texture->screen);
+
+   SVGA_DBG(DEBUG_DMA, "unref sid %p (tex surface)\n", s->handle);
+   ss->sws->surface_reference(ss->sws, &s->handle, NULL);
+   pipe_texture_reference(&surf->texture, NULL);
+   FREE(surf);
+}
+
+
+static INLINE void 
+svga_mark_surface_dirty(struct pipe_surface *surf)
+{
+   struct svga_surface *s = svga_surface(surf);
+
+   if(!s->dirty) {
+      struct svga_texture *tex = svga_texture(surf->texture);
+
+      s->dirty = TRUE;
+
+      if (s->handle == tex->handle)
+         tex->defined[surf->face][surf->level] = TRUE;
+      else {
+         /* this will happen later in svga_propagate_surface */
+      }
+   }
+}
+
+
+void svga_mark_surfaces_dirty(struct svga_context *svga)
+{
+   unsigned i;
+
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      if (svga->curr.framebuffer.cbufs[i])
+         svga_mark_surface_dirty(svga->curr.framebuffer.cbufs[i]);
+   }
+   if (svga->curr.framebuffer.zsbuf)
+      svga_mark_surface_dirty(svga->curr.framebuffer.zsbuf);
+}
+
+/**
+ * Progagate any changes from surfaces to texture.
+ * pipe is optional context to inline the blit command in.
+ */
+void
+svga_propagate_surface(struct pipe_context *pipe, struct pipe_surface *surf)
+{
+   struct svga_surface *s = svga_surface(surf);
+   struct svga_texture *tex = svga_texture(surf->texture);
+   struct svga_screen *ss = svga_screen(surf->texture->screen);
+
+   if (!s->dirty)
+      return;
+
+   s->dirty = FALSE;
+   ss->texture_timestamp++;
+   tex->view_age[surf->level] = ++(tex->age);
+
+   if (s->handle != tex->handle) {
+      SVGA_DBG(DEBUG_VIEWS, "svga: Surface propagate: tex %p, level %u, from %p\n", tex, surf->level, surf);
+      svga_texture_copy_handle(svga_context(pipe), ss,
+                               s->handle, 0, 0, 0, s->real_level, s->real_face,
+                               tex->handle, 0, 0, surf->zslice, surf->level, surf->face,
+                               u_minify(tex->base.width0, surf->level),
+                               u_minify(tex->base.height0, surf->level), 1);
+      tex->defined[surf->face][surf->level] = TRUE;
+   }
+}
+
+/**
+ * Check if we should call svga_propagate_surface on the surface.
+ */
+extern boolean
+svga_surface_needs_propagation(struct pipe_surface *surf)
+{
+   struct svga_surface *s = svga_surface(surf);
+   struct svga_texture *tex = svga_texture(surf->texture);
+
+   return s->dirty && s->handle != tex->handle;
+}
+
+
+static struct pipe_transfer *
+svga_get_tex_transfer(struct pipe_screen *screen,
+                     struct pipe_texture *texture,
+                     unsigned face, unsigned level, unsigned zslice,
+                     enum pipe_transfer_usage usage, unsigned x, unsigned y,
+                     unsigned w, unsigned h)
+{
+   struct svga_screen *ss = svga_screen(screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_transfer *st;
+
+   /* We can't map texture storage directly */
+   if (usage & PIPE_TRANSFER_MAP_DIRECTLY)
+      return NULL;
+
+   st = CALLOC_STRUCT(svga_transfer);
+   if (!st)
+      return NULL;
+   
+   st->base.format = texture->format;
+   st->base.block = texture->block;
+   st->base.x = x;
+   st->base.y = y;
+   st->base.width = w;
+   st->base.height = h;
+   st->base.nblocksx = pf_get_nblocksx(&texture->block, w);
+   st->base.nblocksy = pf_get_nblocksy(&texture->block, h);
+   st->base.stride = st->base.nblocksx*st->base.block.size;
+   st->base.usage = usage;
+   st->base.face = face;
+   st->base.level = level;
+   st->base.zslice = zslice;
+
+   st->hw_nblocksy = st->base.nblocksy;
+   
+   st->hwbuf = svga_winsys_buffer_create(ss, 
+                                         1, 
+                                         0,
+                                         st->hw_nblocksy*st->base.stride);
+   while(!st->hwbuf && (st->hw_nblocksy /= 2)) {
+      st->hwbuf = svga_winsys_buffer_create(ss, 
+                                            1, 
+                                            0,
+                                            st->hw_nblocksy*st->base.stride);
+   }
+
+   if(!st->hwbuf)
+      goto no_hwbuf;
+
+   if(st->hw_nblocksy < st->base.nblocksy) {
+      /* We couldn't allocate a hardware buffer big enough for the transfer, 
+       * so allocate regular malloc memory instead */
+      debug_printf("%s: failed to allocate %u KB of DMA, splitting into %u x %u KB DMA transfers\n",
+                   __FUNCTION__,
+                   (st->base.nblocksy*st->base.stride + 1023)/1024,
+                   (st->base.nblocksy + st->hw_nblocksy - 1)/st->hw_nblocksy,
+                   (st->hw_nblocksy*st->base.stride + 1023)/1024);
+      st->swbuf = MALLOC(st->base.nblocksy*st->base.stride);
+      if(!st->swbuf)
+         goto no_swbuf;
+   }
+   
+   pipe_texture_reference(&st->base.texture, texture);
+
+   if (usage & PIPE_TRANSFER_READ)
+      svga_transfer_dma(st, SVGA3D_READ_HOST_VRAM);
+
+   return &st->base;
+
+no_swbuf:
+   sws->buffer_destroy(sws, st->hwbuf);
+no_hwbuf:
+   FREE(st);
+   return NULL;
+}
+
+
+static void *
+svga_transfer_map( struct pipe_screen *screen,
+                   struct pipe_transfer *transfer )
+{
+   struct svga_screen *ss = svga_screen(screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_transfer *st = svga_transfer(transfer);
+
+   if(st->swbuf)
+      return st->swbuf;
+   else
+      /* The wait for read transfers already happened when svga_transfer_dma
+       * was called. */
+      return sws->buffer_map(sws, st->hwbuf,
+                             pipe_transfer_buffer_flags(transfer));
+}
+
+
+static void
+svga_transfer_unmap(struct pipe_screen *screen,
+                    struct pipe_transfer *transfer)
+{
+   struct svga_screen *ss = svga_screen(screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_transfer *st = svga_transfer(transfer);
+   
+   if(!st->swbuf)
+      sws->buffer_unmap(sws, st->hwbuf);
+}
+
+
+static void
+svga_tex_transfer_destroy(struct pipe_transfer *transfer)
+{
+   struct svga_texture *tex = svga_texture(transfer->texture);
+   struct svga_screen *ss = svga_screen(transfer->texture->screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_transfer *st = svga_transfer(transfer);
+
+   if (st->base.usage & PIPE_TRANSFER_WRITE) {
+      svga_transfer_dma(st, SVGA3D_WRITE_HOST_VRAM);
+      ss->texture_timestamp++;
+      tex->view_age[transfer->level] = ++(tex->age);
+      tex->defined[transfer->face][transfer->level] = TRUE;
+   }
+
+   pipe_texture_reference(&st->base.texture, NULL);
+   FREE(st->swbuf);
+   sws->buffer_destroy(sws, st->hwbuf);
+   FREE(st);
+}
+
+void
+svga_screen_init_texture_functions(struct pipe_screen *screen)
+{
+   screen->texture_create = svga_texture_create;
+   screen->texture_destroy = svga_texture_destroy;
+   screen->get_tex_surface = svga_get_tex_surface;
+   screen->tex_surface_destroy = svga_tex_surface_destroy;
+   screen->texture_blanket = svga_texture_blanket;
+   screen->get_tex_transfer = svga_get_tex_transfer;
+   screen->transfer_map = svga_transfer_map;
+   screen->transfer_unmap = svga_transfer_unmap;
+   screen->tex_transfer_destroy = svga_tex_transfer_destroy;
+}
+
+/*********************************************************************** 
+ */
+
+struct svga_sampler_view *
+svga_get_tex_sampler_view(struct pipe_context *pipe, struct pipe_texture *pt,
+                          unsigned min_lod, unsigned max_lod)
+{
+   struct svga_screen *ss = svga_screen(pt->screen);
+   struct svga_winsys_screen *sws = ss->sws;
+   struct svga_texture *tex = svga_texture(pt); 
+   struct svga_sampler_view *sv = NULL;
+   SVGA3dSurfaceFormat format = svga_translate_format(pt->format);
+   boolean view = TRUE;
+
+   assert(pt);
+   assert(min_lod >= 0);
+   assert(min_lod <= max_lod);
+   assert(max_lod <= pt->last_level);
+
+
+   /* Is a view needed */
+   {
+      /*
+       * Can't control max lod. For first level views and when we only
+       * look at one level we disable mip filtering to achive the same
+       * results as a view.
+       */
+      if (min_lod == 0 && max_lod >= pt->last_level)
+         view = FALSE;
+
+      if (pf_is_compressed(pt->format) && view) {
+         format = svga_translate_format_render(pt->format);
+      }
+
+      if (ss->debug.no_sampler_view)
+         view = FALSE;
+
+      if (ss->debug.force_sampler_view)
+         view = TRUE;
+   }
+
+   /* First try the cache */
+   if (view) {
+      pipe_mutex_lock(ss->tex_mutex);
+      if (tex->cached_view &&
+          tex->cached_view->min_lod == min_lod &&
+          tex->cached_view->max_lod == max_lod) {
+         svga_sampler_view_reference(&sv, tex->cached_view);
+         pipe_mutex_unlock(ss->tex_mutex);
+         SVGA_DBG(DEBUG_VIEWS, "svga: Sampler view: reuse %p, %u %u, last %u\n",
+                              pt, min_lod, max_lod, pt->last_level);
+         svga_validate_sampler_view(svga_context(pipe), sv);
+         return sv;
+      }
+      pipe_mutex_unlock(ss->tex_mutex);
+   }
+
+   sv = CALLOC_STRUCT(svga_sampler_view);
+   pipe_reference_init(&sv->reference, 1);
+   sv->texture = tex;
+   sv->min_lod = min_lod;
+   sv->max_lod = max_lod;
+
+   /* No view needed just use the whole texture */
+   if (!view) {
+      SVGA_DBG(DEBUG_VIEWS,
+               "svga: Sampler view: no %p, mips %u..%u, nr %u, size (%ux%ux%u), last %u\n",
+               pt, min_lod, max_lod,
+               max_lod - min_lod + 1,
+               pt->width0,
+               pt->height0,
+               pt->depth0,
+               pt->last_level);
+      sws->surface_reference(sws, &sv->handle, tex->handle);
+      return sv;
+   }
+
+   SVGA_DBG(DEBUG_VIEWS,
+            "svga: Sampler view: yes %p, mips %u..%u, nr %u, size (%ux%ux%u), last %u\n",
+            pt, min_lod, max_lod,
+            max_lod - min_lod + 1,
+            pt->width0,
+            pt->height0,
+            pt->depth0,
+            pt->last_level);
+
+   sv->age = tex->age;
+   sv->handle = svga_texture_view_surface(pipe, tex, format,
+                                          min_lod,
+                                          max_lod - min_lod + 1,
+                                          -1, -1);
+
+   if (!sv->handle) {
+      assert(0);
+      sws->surface_reference(sws, &sv->handle, tex->handle);
+      return sv;
+   }
+
+   pipe_mutex_lock(ss->tex_mutex);
+   svga_sampler_view_reference(&tex->cached_view, sv);
+   pipe_mutex_unlock(ss->tex_mutex);
+
+   return sv;
+}
+
+void
+svga_validate_sampler_view(struct svga_context *svga, struct svga_sampler_view *v)
+{
+   struct svga_texture *tex = v->texture;
+   unsigned numFaces;
+   unsigned age = 0;
+   int i, k;
+
+   assert(svga);
+
+   if (v->handle == v->texture->handle)
+      return;
+
+   age = tex->age;
+
+   if(tex->base.target == PIPE_TEXTURE_CUBE)
+      numFaces = 6;
+   else
+      numFaces = 1;
+
+   for (i = v->min_lod; i <= v->max_lod; i++) {
+      for (k = 0; k < numFaces; k++) {
+         if (v->age < tex->view_age[i])
+            svga_texture_copy_handle(svga, NULL,
+                                     tex->handle, 0, 0, 0, i, k,
+                                     v->handle, 0, 0, 0, i - v->min_lod, k,
+                                     u_minify(tex->base.width0, i),
+                                     u_minify(tex->base.height0, i),
+                                     u_minify(tex->base.depth0, i));
+      }
+   }
+
+   v->age = age;
+}
+
+void
+svga_destroy_sampler_view_priv(struct svga_sampler_view *v)
+{
+   struct svga_screen *ss = svga_screen(v->texture->base.screen);
+
+   SVGA_DBG(DEBUG_DMA, "unref sid %p (sampler view)\n", v->handle);
+   ss->sws->surface_reference(ss->sws, &v->handle, NULL);
+
+   FREE(v);
+}
+
+boolean
+svga_screen_buffer_from_texture(struct pipe_texture *texture,
+				struct pipe_buffer **buffer,
+				unsigned *stride)
+{
+   struct svga_texture *stex = svga_texture(texture);
+
+   *buffer = svga_screen_buffer_wrap_surface
+      (texture->screen,
+       svga_translate_format(texture->format),
+       stex->handle);
+
+   *stride = pf_get_nblocksx(&texture->block, texture->width0) *
+      texture->block.size;
+
+   return *buffer != NULL;
+}
+
+
+struct svga_winsys_surface *
+svga_screen_texture_get_winsys_surface(struct pipe_texture *texture)
+{
+   struct svga_winsys_screen *sws = svga_winsys_screen(texture->screen);
+   struct svga_winsys_surface *vsurf = NULL;
+
+   sws->surface_reference(sws, &vsurf, svga_texture(texture)->handle);
+   return vsurf;
+}
diff --git a/src/gallium/drivers/svga/svga_screen_texture.h b/src/gallium/drivers/svga/svga_screen_texture.h
new file mode 100644
index 0000000000..1e6fef59a3
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_screen_texture.h
@@ -0,0 +1,177 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_TEXTURE_H
+#define SVGA_TEXTURE_H
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_state.h"
+
+
+struct pipe_context;
+struct pipe_screen;
+struct svga_context;
+struct svga_winsys_surface;
+enum SVGA3dSurfaceFormat;
+
+
+#define SVGA_MAX_TEXTURE_LEVELS 12 /* 2048x2048 */
+
+
+/**
+ * A sampler's view into a texture
+ *
+ * We currently cache one sampler view on
+ * the texture and in there by holding a reference
+ * from the texture to the sampler view.
+ *
+ * Because of this we can not hold a refernce to the
+ * texture from the sampler view. So the user
+ * of the sampler views must make sure that the
+ * texture has a reference take for as long as
+ * the sampler view is refrenced.
+ *
+ * Just unreferencing the sampler_view before the
+ * texture is enough.
+ */
+struct svga_sampler_view
+{
+   struct pipe_reference reference;
+
+   struct svga_texture *texture;
+
+   int min_lod;
+   int max_lod;
+
+   unsigned age;
+
+   struct svga_winsys_surface *handle;
+};
+
+
+struct svga_texture 
+{
+   struct pipe_texture base;
+
+   struct svga_winsys_surface *handle;
+
+   boolean defined[6][PIPE_MAX_TEXTURE_LEVELS];
+   
+   struct svga_sampler_view *cached_view;
+
+   unsigned view_age[SVGA_MAX_TEXTURE_LEVELS];
+   unsigned age;
+
+   boolean views_modified;
+};
+
+
+struct svga_surface
+{
+   struct pipe_surface base;
+
+   struct svga_winsys_surface *handle;
+
+   unsigned real_face;
+   unsigned real_level;
+   unsigned real_zslice;
+
+   boolean dirty;
+};
+
+
+struct svga_transfer
+{
+   struct pipe_transfer base;
+
+   struct svga_winsys_buffer *hwbuf;
+
+   /* Height of the hardware buffer in pixel blocks */
+   unsigned hw_nblocksy;
+
+   /* Temporary malloc buffer when we can't allocate a hardware buffer
+    * big enough */
+   void *swbuf;
+};
+
+
+static INLINE struct svga_texture *
+svga_texture(struct pipe_texture *texture)
+{
+   return (struct svga_texture *)texture;
+}
+
+static INLINE struct svga_surface *
+svga_surface(struct pipe_surface *surface)
+{
+   assert(surface);
+   return (struct svga_surface *)surface;
+}
+
+static INLINE struct svga_transfer *
+svga_transfer(struct pipe_transfer *transfer)
+{
+   assert(transfer);
+   return (struct svga_transfer *)transfer;
+}
+
+extern struct svga_sampler_view *
+svga_get_tex_sampler_view(struct pipe_context *pipe,
+                          struct pipe_texture *pt,
+                          unsigned min_lod, unsigned max_lod);
+
+void
+svga_validate_sampler_view(struct svga_context *svga, struct svga_sampler_view *v);
+
+void
+svga_destroy_sampler_view_priv(struct svga_sampler_view *v);
+
+static INLINE void
+svga_sampler_view_reference(struct svga_sampler_view **ptr, struct svga_sampler_view *v)
+{
+   struct svga_sampler_view *old = *ptr;
+
+   if (pipe_reference((struct pipe_reference **)ptr, &v->reference))
+      svga_destroy_sampler_view_priv(old);
+}
+
+extern void
+svga_propagate_surface(struct pipe_context *pipe, struct pipe_surface *surf);
+
+extern boolean
+svga_surface_needs_propagation(struct pipe_surface *surf);
+
+extern void
+svga_screen_init_texture_functions(struct pipe_screen *screen);
+
+enum SVGA3dSurfaceFormat
+svga_translate_format(enum pipe_format format);
+
+enum SVGA3dSurfaceFormat
+svga_translate_format_render(enum pipe_format format);
+
+
+#endif /* SVGA_TEXTURE_H */
diff --git a/src/gallium/drivers/svga/svga_state.c b/src/gallium/drivers/svga/svga_state.c
new file mode 100644
index 0000000000..1c21d3acfe
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state.c
@@ -0,0 +1,278 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "util/u_debug.h"
+#include "pipe/p_defines.h"
+#include "util/u_memory.h"
+#include "draw/draw_context.h"
+
+#include "svga_context.h"
+#include "svga_screen.h"
+#include "svga_state.h"
+#include "svga_draw.h"
+#include "svga_cmd.h"
+#include "svga_hw_reg.h"
+
+/* This is just enough to decide whether we need to use the draw
+ * module (swtnl) or not.
+ */
+static const struct svga_tracked_state *need_swtnl_state[] =
+{
+   &svga_update_need_swvfetch,
+   &svga_update_need_pipeline,
+   &svga_update_need_swtnl,
+   NULL
+};
+
+
+/* Atoms to update hardware state prior to emitting a clear or draw
+ * packet.
+ */
+static const struct svga_tracked_state *hw_clear_state[] =
+{
+   &svga_hw_scissor,
+   &svga_hw_viewport,
+   &svga_hw_framebuffer,
+   NULL
+};
+
+
+/* Atoms to update hardware state prior to emitting a draw packet.
+ */
+static const struct svga_tracked_state *hw_draw_state[] =
+{
+   &svga_hw_update_zero_stride,
+   &svga_hw_fs,
+   &svga_hw_vs,
+   &svga_hw_rss,
+   &svga_hw_tss,
+   &svga_hw_tss_binding,
+   &svga_hw_clip_planes,
+   &svga_hw_vdecl,
+   &svga_hw_fs_parameters,
+   &svga_hw_vs_parameters,
+   NULL
+};
+
+
+static const struct svga_tracked_state *swtnl_draw_state[] =
+{
+   &svga_update_swtnl_draw,
+   &svga_update_swtnl_vdecl,
+   NULL
+};
+
+/* Flattens the graph of state dependencies.  Could swap the positions
+ * of hw_clear_state and need_swtnl_state without breaking anything.
+ */
+static const struct svga_tracked_state **state_levels[] = 
+{
+   need_swtnl_state,
+   hw_clear_state,
+   hw_draw_state,
+   swtnl_draw_state
+};
+
+
+
+static unsigned check_state( unsigned a,
+                             unsigned b )
+{
+   return (a & b);
+}
+
+static void accumulate_state( unsigned *a,
+			      unsigned b )
+{
+   *a |= b;
+}
+
+
+static void xor_states( unsigned *result,
+                        unsigned a,
+                        unsigned b )
+{
+   *result = a ^ b;
+}
+
+
+
+static int update_state( struct svga_context *svga,
+                         const struct svga_tracked_state *atoms[],
+                         unsigned *state )
+{
+   boolean debug = TRUE;
+   enum pipe_error ret = 0;
+   unsigned i;
+
+   ret = svga_hwtnl_flush( svga->hwtnl );
+   if (ret != 0)
+      return ret;
+
+   if (debug) {
+      /* Debug version which enforces various sanity checks on the
+       * state flags which are generated and checked to help ensure
+       * state atoms are ordered correctly in the list.
+       */
+      unsigned examined, prev;      
+
+      examined = 0;
+      prev = *state;
+
+      for (i = 0; atoms[i] != NULL; i++) {	 
+	 unsigned generated;
+
+	 assert(atoms[i]->dirty); 
+	 assert(atoms[i]->update);
+
+	 if (check_state(*state, atoms[i]->dirty)) {
+	    if (0)
+               debug_printf("update: %s\n", atoms[i]->name);
+	    ret = atoms[i]->update( svga, *state );
+            if (ret != 0)
+               return ret;
+	 }
+
+	 /* generated = (prev ^ state)
+	  * if (examined & generated)
+	  *     fail;
+	  */
+	 xor_states(&generated, prev, *state);
+	 if (check_state(examined, generated)) {
+	    debug_printf("state atom %s generated state already examined\n", 
+                         atoms[i]->name);
+	    assert(0);
+	 }
+			 
+	 prev = *state;
+	 accumulate_state(&examined, atoms[i]->dirty);
+      }
+   }
+   else {
+      for (i = 0; atoms[i] != NULL; i++) {	 
+	 if (check_state(*state, atoms[i]->dirty)) {
+	    ret = atoms[i]->update( svga, *state );
+            if (ret != 0)
+               return ret;
+         }
+      }
+   }
+
+   return 0;
+}
+
+
+
+int svga_update_state( struct svga_context *svga,
+                       unsigned max_level )
+{
+   struct svga_screen *screen = svga_screen(svga->pipe.screen);
+   int ret = 0;
+   int i;
+
+   /* Check for updates to bound textures.  This can't be done in an
+    * atom as there is no flag which could provoke this test, and we
+    * cannot create one.
+    */
+   if (svga->state.texture_timestamp != screen->texture_timestamp) {
+      svga->state.texture_timestamp = screen->texture_timestamp;
+      svga->dirty |= SVGA_NEW_TEXTURE;
+   }
+
+   for (i = 0; i <= max_level; i++) {
+      svga->dirty |= svga->state.dirty[i];
+
+      if (svga->dirty) {
+         ret = update_state( svga, 
+                             state_levels[i], 
+                             &svga->dirty );
+         if (ret != 0)
+            return ret;
+
+         svga->state.dirty[i] = 0;
+      }
+   }
+   
+   for (; i < SVGA_STATE_MAX; i++) 
+      svga->state.dirty[i] |= svga->dirty;
+
+   svga->dirty = 0;
+   return 0;
+}
+
+
+
+
+void svga_update_state_retry( struct svga_context *svga,
+                              unsigned max_level )
+{
+   int ret;
+
+   ret = svga_update_state( svga, max_level );
+
+   if (ret == PIPE_ERROR_OUT_OF_MEMORY) {
+      svga_context_flush(svga, NULL);
+      ret = svga_update_state( svga, max_level );
+   }
+
+   assert( ret == 0 );
+}
+
+
+
+#define EMIT_RS(_rs, _count, _name, _value)     \
+do {                                            \
+   _rs[_count].state = _name;                   \
+   _rs[_count].uintValue = _value;              \
+   _count++;                                    \
+} while (0)
+
+
+/* Setup any hardware state which will be constant through the life of
+ * a context.
+ */
+enum pipe_error svga_emit_initial_state( struct svga_context *svga )
+{
+   SVGA3dRenderState *rs;
+   unsigned count = 0;
+   const unsigned COUNT = 2;
+   enum pipe_error ret;
+
+   ret = SVGA3D_BeginSetRenderState( svga->swc, &rs, COUNT );
+   if (ret)
+      return ret;
+
+   /* Always use D3D style coordinate space as this is the only one
+    * which is implemented on all backends.
+    */
+   EMIT_RS(rs, count, SVGA3D_RS_COORDINATETYPE, SVGA3D_COORDINATE_LEFTHANDED );
+   EMIT_RS(rs, count, SVGA3D_RS_FRONTWINDING, SVGA3D_FRONTWINDING_CW );
+   
+   assert( COUNT == count );
+   SVGA_FIFOCommitAll( svga->swc );
+
+   return 0;
+
+}
diff --git a/src/gallium/drivers/svga/svga_state.h b/src/gallium/drivers/svga/svga_state.h
new file mode 100644
index 0000000000..22d5a6d552
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state.h
@@ -0,0 +1,95 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_STATE_H
+#define SVGA_STATE_H
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_defines.h"
+
+struct svga_context;
+
+
+void svga_init_state( struct svga_context *svga );
+void svga_destroy_state( struct svga_context *svga );
+
+
+struct svga_tracked_state {
+   const char *name;
+   unsigned dirty;
+   int (*update)( struct svga_context *svga, unsigned dirty );
+};
+
+/* NEED_SWTNL
+ */
+extern struct svga_tracked_state svga_update_need_swvfetch;
+extern struct svga_tracked_state svga_update_need_pipeline;
+extern struct svga_tracked_state svga_update_need_swtnl;
+
+/* HW_CLEAR
+ */
+extern struct svga_tracked_state svga_hw_viewport;
+extern struct svga_tracked_state svga_hw_scissor;
+extern struct svga_tracked_state svga_hw_framebuffer;
+
+/* HW_DRAW
+ */
+extern struct svga_tracked_state svga_hw_vs;
+extern struct svga_tracked_state svga_hw_fs;
+extern struct svga_tracked_state svga_hw_rss;
+extern struct svga_tracked_state svga_hw_tss;
+extern struct svga_tracked_state svga_hw_tss_binding;
+extern struct svga_tracked_state svga_hw_clip_planes;
+extern struct svga_tracked_state svga_hw_vdecl;
+extern struct svga_tracked_state svga_hw_fs_parameters;
+extern struct svga_tracked_state svga_hw_vs_parameters;
+extern struct svga_tracked_state svga_hw_update_zero_stride;
+
+/* SWTNL_DRAW
+ */
+extern struct svga_tracked_state svga_update_swtnl_draw;
+extern struct svga_tracked_state svga_update_swtnl_vdecl;
+
+/* Bring the hardware fully up-to-date so that we can emit draw
+ * commands.
+ */
+#define SVGA_STATE_NEED_SWTNL        0
+#define SVGA_STATE_HW_CLEAR          1
+#define SVGA_STATE_HW_DRAW           2
+#define SVGA_STATE_SWTNL_DRAW        3
+#define SVGA_STATE_MAX               4
+
+
+enum pipe_error svga_update_state( struct svga_context *svga,
+                                   unsigned level );
+
+void svga_update_state_retry( struct svga_context *svga,
+                              unsigned level );
+
+
+enum pipe_error svga_emit_initial_state( struct svga_context *svga );
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_state_constants.c b/src/gallium/drivers/svga/svga_state_constants.c
new file mode 100644
index 0000000000..209ed28245
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_constants.c
@@ -0,0 +1,239 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+#include "svga_tgsi.h"
+#include "svga_debug.h"
+
+#include "svga_hw_reg.h"
+
+/***********************************************************************
+ * Hardware update 
+ */
+
+/* Convert from PIPE_SHADER_* to SVGA3D_SHADERTYPE_*
+ */
+static int svga_shader_type( int unit )
+{
+   return unit + 1;
+}
+
+
+static int emit_const( struct svga_context *svga,
+                       int unit,
+                       int i,
+                       const float *value )
+{
+   int ret = PIPE_OK;
+
+   if (memcmp(svga->state.hw_draw.cb[unit][i], value, 4 * sizeof(float)) != 0) {
+      if (SVGA_DEBUG & DEBUG_CONSTS)
+         debug_printf("%s %s %d: %f %f %f %f\n",
+                      __FUNCTION__,
+                      unit == PIPE_SHADER_VERTEX ? "VERT" : "FRAG",
+                      i,
+                      value[0],
+                      value[1],
+                      value[2],
+                      value[3]);
+
+      ret = SVGA3D_SetShaderConst( svga->swc, 
+                                   i,
+                                   svga_shader_type(unit),
+                                   SVGA3D_CONST_TYPE_FLOAT,
+                                   value );
+      if (ret)
+         return ret;
+
+      memcpy(svga->state.hw_draw.cb[unit][i], value, 4 * sizeof(float));
+   }
+   
+   return ret;
+}
+
+static int emit_consts( struct svga_context *svga,
+                        int offset,
+                        int unit )
+{
+   struct pipe_screen *screen = svga->pipe.screen;
+   unsigned count;
+   const float (*data)[4] = NULL;
+   unsigned i;
+   int ret = PIPE_OK;
+
+   if (svga->curr.cb[unit] == NULL)
+      goto done;
+
+   count = svga->curr.cb[unit]->size / (4 * sizeof(float));
+
+   data = (const float (*)[4])pipe_buffer_map(screen,
+                                              svga->curr.cb[unit],
+                                              PIPE_BUFFER_USAGE_CPU_READ);
+   if (data == NULL) {
+      ret = PIPE_ERROR_OUT_OF_MEMORY;
+      goto done;
+   }
+
+   for (i = 0; i < count; i++) {
+      ret = emit_const( svga, unit, offset + i, data[i] );
+      if (ret)
+         goto done;
+   }
+
+done:
+   if (data)
+      pipe_buffer_unmap(screen, svga->curr.cb[unit]);
+
+   return ret;
+}
+   
+static int emit_fs_consts( struct svga_context *svga,
+                           unsigned dirty )
+{
+   const struct svga_shader_result *result = svga->state.hw_draw.fs;
+   const struct svga_fs_compile_key *key = &result->key.fkey;
+   int ret = 0;
+
+   ret = emit_consts( svga, 0, PIPE_SHADER_FRAGMENT );
+   if (ret)
+      return ret;
+
+   /* The internally generated fragment shader for xor blending
+    * doesn't have a 'result' struct.  It should be fixed to avoid
+    * this special case, but work around it with a NULL check:
+    */
+   if (result != NULL &&
+       key->num_unnormalized_coords)
+   {
+      unsigned offset = result->shader->info.file_max[TGSI_FILE_CONSTANT] + 1;
+      int i;
+
+      for (i = 0; i < key->num_textures; i++) {
+         if (key->tex[i].unnormalized) {
+            struct pipe_texture *tex = svga->curr.texture[i];
+            float data[4];
+
+            data[0] = 1.0 / (float)tex->width0;
+            data[1] = 1.0 / (float)tex->height0;
+            data[2] = 1.0;
+            data[3] = 1.0;
+
+            ret = emit_const( svga,
+                              PIPE_SHADER_FRAGMENT,
+                              key->tex[i].width_height_idx + offset,
+                              data );
+            if (ret)
+               return ret;
+         }
+      }
+
+      offset += key->num_unnormalized_coords;
+   }
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_hw_fs_parameters = 
+{
+   "hw fs params",
+   (SVGA_NEW_FS_CONST_BUFFER |
+    SVGA_NEW_FS_RESULT |
+    SVGA_NEW_TEXTURE_BINDING),
+   emit_fs_consts
+};
+
+/***********************************************************************
+ */
+
+static int emit_vs_consts( struct svga_context *svga,
+                           unsigned dirty )
+{
+   const struct svga_shader_result *result = svga->state.hw_draw.vs;
+   const struct svga_vs_compile_key *key = &result->key.vkey;
+   int ret = 0;
+   unsigned offset;
+
+   /* SVGA_NEW_VS_RESULT
+    */
+   if (result == NULL) 
+      return 0;
+
+   /* SVGA_NEW_VS_CONST_BUFFER 
+    */
+   ret = emit_consts( svga, 0, PIPE_SHADER_VERTEX );
+   if (ret)
+      return ret;
+
+   offset = result->shader->info.file_max[TGSI_FILE_CONSTANT] + 1;
+
+   /* SVGA_NEW_VS_RESULT
+    */
+   if (key->need_prescale) {
+      ret = emit_const( svga, PIPE_SHADER_VERTEX, offset++,
+                        svga->state.hw_clear.prescale.scale );
+      if (ret)
+         return ret;
+
+      ret = emit_const( svga, PIPE_SHADER_VERTEX, offset++,
+                        svga->state.hw_clear.prescale.translate );
+      if (ret)
+         return ret;
+   }
+
+   /* SVGA_NEW_ZERO_STRIDE
+    */
+   if (key->zero_stride_vertex_elements) {
+      unsigned i, curr_zero_stride = 0;
+      for (i = 0; i < PIPE_MAX_ATTRIBS; ++i) {
+         if (key->zero_stride_vertex_elements & (1 << i)) {
+            ret = emit_const( svga, PIPE_SHADER_VERTEX, offset++,
+                              svga->curr.zero_stride_constants +
+                              4 * curr_zero_stride );
+            if (ret)
+               return ret;
+            ++curr_zero_stride;
+         }
+      }
+   }
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_hw_vs_parameters = 
+{
+   "hw vs params",
+   (SVGA_NEW_VS_CONST_BUFFER |
+    SVGA_NEW_ZERO_STRIDE |
+    SVGA_NEW_VS_RESULT),
+   emit_vs_consts
+};
+
diff --git a/src/gallium/drivers/svga/svga_state_framebuffer.c b/src/gallium/drivers/svga/svga_state_framebuffer.c
new file mode 100644
index 0000000000..7d7f93d8e3
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_framebuffer.c
@@ -0,0 +1,455 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+#include "svga_debug.h"
+
+#include "svga_hw_reg.h"
+
+
+/***********************************************************************
+ * Hardware state update
+ */
+
+
+static int emit_framebuffer( struct svga_context *svga,
+                             unsigned dirty )
+{
+   const struct pipe_framebuffer_state *curr = &svga->curr.framebuffer;
+   struct pipe_framebuffer_state *hw = &svga->state.hw_clear.framebuffer;
+   unsigned i;
+   enum pipe_error ret;
+
+   /* XXX: Need shadow state in svga->hw to eliminate redundant
+    * uploads, especially of NULL buffers.
+    */
+   
+   for(i = 0; i < PIPE_MAX_COLOR_BUFS; ++i) {
+      if (curr->cbufs[i] != hw->cbufs[i]) {
+         ret = SVGA3D_SetRenderTarget(svga->swc, SVGA3D_RT_COLOR0 + i, curr->cbufs[i]);
+         if (ret != PIPE_OK)
+            return ret;
+         
+         pipe_surface_reference(&hw->cbufs[i], curr->cbufs[i]);
+      }
+   }
+
+   
+   if (curr->zsbuf != hw->zsbuf) {
+      ret = SVGA3D_SetRenderTarget(svga->swc, SVGA3D_RT_DEPTH, curr->zsbuf);
+      if (ret != PIPE_OK)
+         return ret;
+
+      if (curr->zsbuf &&
+          curr->zsbuf->format == PIPE_FORMAT_Z24S8_UNORM) {
+         ret = SVGA3D_SetRenderTarget(svga->swc, SVGA3D_RT_STENCIL, curr->zsbuf);
+         if (ret != PIPE_OK)
+            return ret;
+      }
+      else {
+         ret = SVGA3D_SetRenderTarget(svga->swc, SVGA3D_RT_STENCIL, NULL);
+         if (ret != PIPE_OK)
+            return ret;
+      }
+      
+      pipe_surface_reference(&hw->zsbuf, curr->zsbuf);
+   }
+
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_hw_framebuffer = 
+{
+   "hw framebuffer state",
+   SVGA_NEW_FRAME_BUFFER,
+   emit_framebuffer
+};
+
+
+
+
+/*********************************************************************** 
+ */
+
+static int emit_viewport( struct svga_context *svga,
+                          unsigned dirty )
+{
+   const struct pipe_viewport_state *viewport = &svga->curr.viewport;
+   struct svga_prescale prescale;
+   SVGA3dRect rect;
+   /* Not sure if this state is relevant with POSITIONT.  Probably
+    * not, but setting to 0,1 avoids some state pingponging.
+    */
+   float range_min = 0.0;
+   float range_max = 1.0;
+   float flip = -1.0;
+   boolean degenerate = FALSE;
+   enum pipe_error ret;
+
+   float fb_width = svga->curr.framebuffer.width;
+   float fb_height = svga->curr.framebuffer.height;
+
+   memset( &prescale, 0, sizeof(prescale) );
+
+   if (svga->curr.rast->templ.bypass_vs_clip_and_viewport) {
+
+      /* Avoid POSITIONT as it has a non trivial implementation outside the D3D
+       * API. Always generate a vertex shader.
+       */
+      rect.x = 0;
+      rect.y = 0;
+      rect.w = svga->curr.framebuffer.width;
+      rect.h = svga->curr.framebuffer.height;
+
+      prescale.scale[0] = 2.0 / (float)rect.w;
+      prescale.scale[1] = - 2.0 / (float)rect.h;
+      prescale.scale[2] = 1.0;
+      prescale.scale[3] = 1.0;
+      prescale.translate[0] = -1.0f;
+      prescale.translate[1] = 1.0f;
+      prescale.translate[2] = 0;
+      prescale.translate[3] = 0;
+      prescale.enabled = TRUE;
+   } else {
+
+      /* Examine gallium viewport transformation and produce a screen
+       * rectangle and possibly vertex shader pre-transformation to
+       * get the same results.
+       */
+      float fx =        viewport->scale[0] * -1.0 + viewport->translate[0];
+      float fy = flip * viewport->scale[1] * -1.0 + viewport->translate[1];
+      float fw =        viewport->scale[0] * 2; 
+      float fh = flip * viewport->scale[1] * 2; 
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "\ninitial %f,%f %fx%f\n",
+               fx,
+               fy,
+               fw,
+               fh);
+
+      prescale.scale[0] = 1.0;
+      prescale.scale[1] = 1.0;
+      prescale.scale[2] = 1.0;
+      prescale.scale[3] = 1.0;
+      prescale.translate[0] = 0;
+      prescale.translate[1] = 0;
+      prescale.translate[2] = 0;
+      prescale.translate[3] = 0;
+      prescale.enabled = TRUE;
+
+
+
+      if (fw < 0) {
+         prescale.scale[0] *= -1.0;
+         prescale.translate[0] += -fw;
+         fw = -fw;
+         fx =        viewport->scale[0] * 1.0 + viewport->translate[0];
+      }
+
+      if (fh < 0) {
+         prescale.scale[1] *= -1.0;
+         prescale.translate[1] += -fh;
+         fh = -fh;
+         fy = flip * viewport->scale[1] * 1.0 + viewport->translate[1];
+      }
+
+      if (fx < 0) {
+         prescale.translate[0] += fx;
+         prescale.scale[0] *= fw / (fw + fx); 
+         fw += fx;
+         fx = 0;
+      }
+
+      if (fy < 0) {
+         prescale.translate[1] += fy;
+         prescale.scale[1] *= fh / (fh + fy); 
+         fh += fy;
+         fy = 0;
+      }
+
+      if (fx + fw > fb_width) {
+         prescale.scale[0] *= fw / (fb_width - fx); 
+         prescale.translate[0] -= fx * (fw / (fb_width - fx));
+         prescale.translate[0] += fx;
+         fw = fb_width - fx;
+         
+      }
+
+      if (fy + fh > fb_height) {
+         prescale.scale[1] *= fh / (fb_height - fy);
+         prescale.translate[1] -= fy * (fh / (fb_height - fy));
+         prescale.translate[1] += fy;
+         fh = fb_height - fy;
+      }
+
+      if (fw < 0 || fh < 0) {
+         fw = fh = fx = fy = 0;
+         degenerate = TRUE;
+         goto out;
+      }
+
+
+      /* D3D viewport is integer space.  Convert fx,fy,etc. to
+       * integers.
+       *
+       * TODO: adjust pretranslate correct for any subpixel error
+       * introduced converting to integers.
+       */
+      rect.x = fx;
+      rect.y = fy;
+      rect.w = fw;
+      rect.h = fh;
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "viewport error %f,%f %fx%f\n",
+               fabs((float)rect.x - fx),
+               fabs((float)rect.y - fy),
+               fabs((float)rect.w - fw),
+               fabs((float)rect.h - fh));
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "viewport %d,%d %dx%d\n",
+               rect.x,
+               rect.y,
+               rect.w,
+               rect.h);
+      
+
+      /* Finally, to get GL rasterization rules, need to tweak the
+       * screen-space coordinates slightly relative to D3D which is
+       * what hardware implements natively.
+       */
+      if (svga->curr.rast->templ.gl_rasterization_rules) {
+         float adjust_x = 0.0;
+         float adjust_y = 0.0;
+
+         switch (svga->curr.reduced_prim) {
+         case PIPE_PRIM_LINES:
+            adjust_x = -0.5;
+            adjust_y = 0;
+            break;
+         case PIPE_PRIM_POINTS:
+         case PIPE_PRIM_TRIANGLES:
+            adjust_x = -0.375;
+            adjust_y = -0.5;
+            break;
+         }
+
+         prescale.translate[0] += adjust_x;
+         prescale.translate[1] += adjust_y;
+         prescale.translate[2] = 0.5; /* D3D clip space */
+         prescale.scale[2]     = 0.5; /* D3D clip space */
+      }
+
+
+      range_min = viewport->scale[2] * -1.0 + viewport->translate[2];
+      range_max = viewport->scale[2] *  1.0 + viewport->translate[2];
+
+      /* D3D (and by implication SVGA) doesn't like dealing with zmax
+       * less than zmin.  Detect that case, flip the depth range and
+       * invert our z-scale factor to achieve the same effect.
+       */
+      if (range_min > range_max) {
+         float range_tmp;
+         range_tmp = range_min; 
+         range_min = range_max; 
+         range_max = range_tmp;
+         prescale.scale[2]     = -prescale.scale[2];
+      }
+   }
+
+   if (prescale.enabled) {
+      float H[2];
+      float J[2];
+      int i;
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "prescale %f,%f %fx%f\n",
+               prescale.translate[0],
+               prescale.translate[1],
+               prescale.scale[0],
+               prescale.scale[1]);
+
+      H[0] = (float)rect.w / 2.0;
+      H[1] = -(float)rect.h / 2.0;
+      J[0] = (float)rect.x + (float)rect.w / 2.0;
+      J[1] = (float)rect.y + (float)rect.h / 2.0;
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "H %f,%f\n"
+               "J %fx%f\n",
+               H[0],
+               H[1],
+               J[0],
+               J[1]);
+
+      /* Adjust prescale to take into account the fact that it is
+       * going to be applied prior to the perspective divide and
+       * viewport transformation.
+       * 
+       * Vwin = H(Vc/Vc.w) + J
+       *
+       * We want to tweak Vwin with scale and translation from above,
+       * as in:
+       *
+       * Vwin' = S Vwin + T
+       *
+       * But we can only modify the values at Vc.  Plugging all the
+       * above together, and rearranging, eventually we get:
+       *
+       *   Vwin' = H(Vc'/Vc'.w) + J
+       * where:
+       *   Vc' = SVc + KVc.w
+       *   K = (T + (S-1)J) / H
+       *
+       * Overwrite prescale.translate with values for K:
+       */
+      for (i = 0; i < 2; i++) {
+         prescale.translate[i] = ((prescale.translate[i] +
+                                   (prescale.scale[i] - 1.0) * J[i]) / H[i]);
+      }
+
+      SVGA_DBG(DEBUG_VIEWPORT,
+               "clipspace %f,%f %fx%f\n",
+               prescale.translate[0],
+               prescale.translate[1],
+               prescale.scale[0],
+               prescale.scale[1]);
+   }
+
+out:
+   if (degenerate) {
+      rect.x = 0;
+      rect.y = 0;
+      rect.w = 1;
+      rect.h = 1;
+      prescale.enabled = FALSE;
+   }
+
+   if (memcmp(&rect, &svga->state.hw_clear.viewport, sizeof(rect)) != 0) {
+      ret = SVGA3D_SetViewport(svga->swc, &rect);
+      if(ret != PIPE_OK)
+         return ret;
+
+      memcpy(&svga->state.hw_clear.viewport, &rect, sizeof(rect));
+      assert(sizeof(rect) == sizeof(svga->state.hw_clear.viewport));
+   }
+
+   if (svga->state.hw_clear.depthrange.zmin != range_min ||
+       svga->state.hw_clear.depthrange.zmax != range_max) 
+   {
+      ret = SVGA3D_SetZRange(svga->swc, range_min, range_max );
+      if(ret != PIPE_OK)
+         return ret;
+
+      svga->state.hw_clear.depthrange.zmin = range_min;
+      svga->state.hw_clear.depthrange.zmax = range_max;
+   }
+
+   if (memcmp(&prescale, &svga->state.hw_clear.prescale, sizeof prescale) != 0) {
+      svga->dirty |= SVGA_NEW_PRESCALE;
+      svga->state.hw_clear.prescale = prescale;
+   }
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_hw_viewport = 
+{
+   "hw viewport state",
+   ( SVGA_NEW_FRAME_BUFFER |
+     SVGA_NEW_VIEWPORT |
+     SVGA_NEW_RAST |
+     SVGA_NEW_REDUCED_PRIMITIVE ),
+   emit_viewport
+};
+
+
+/***********************************************************************
+ * Scissor state
+ */
+static int emit_scissor_rect( struct svga_context *svga,
+                              unsigned dirty )
+{
+   const struct pipe_scissor_state *scissor = &svga->curr.scissor;
+   SVGA3dRect rect;
+
+   rect.x = scissor->minx;
+   rect.y = scissor->miny;
+   rect.w = scissor->maxx - scissor->minx; /* + 1 ?? */
+   rect.h = scissor->maxy - scissor->miny; /* + 1 ?? */
+   
+   return SVGA3D_SetScissorRect(svga->swc, &rect);
+}
+
+
+struct svga_tracked_state svga_hw_scissor = 
+{
+   "hw scissor state",
+   SVGA_NEW_SCISSOR,
+   emit_scissor_rect
+};
+
+
+/***********************************************************************
+ * Userclip state
+ */
+
+static int emit_clip_planes( struct svga_context *svga,
+                             unsigned dirty )
+{
+   unsigned i;
+   enum pipe_error ret;
+
+   /* TODO: just emit directly from svga_set_clip_state()?
+    */
+   for (i = 0; i < svga->curr.clip.nr; i++) {
+      ret = SVGA3D_SetClipPlane( svga->swc,
+                                 i,
+                                 svga->curr.clip.ucp[i] );
+      if(ret != PIPE_OK)
+         return ret;
+   }
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_hw_clip_planes = 
+{
+   "hw viewport state",
+   SVGA_NEW_CLIP,
+   emit_clip_planes
+};
diff --git a/src/gallium/drivers/svga/svga_state_fs.c b/src/gallium/drivers/svga/svga_state_fs.c
new file mode 100644
index 0000000000..6ec38ed3e4
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_fs.c
@@ -0,0 +1,282 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+#include "svga_tgsi.h"
+
+#include "svga_hw_reg.h"
+
+
+
+static INLINE int compare_fs_keys( const struct svga_fs_compile_key *a,
+                                   const struct svga_fs_compile_key *b )
+{
+   unsigned keysize = svga_fs_key_size( a );
+   return memcmp( a, b, keysize );
+}
+
+
+static struct svga_shader_result *search_fs_key( struct svga_fragment_shader *fs,
+                                                 const struct svga_fs_compile_key *key )
+{
+   struct svga_shader_result *result = fs->base.results;
+
+   assert(key);
+
+   for ( ; result; result = result->next) {
+      if (compare_fs_keys( key, &result->key.fkey ) == 0)
+         return result;
+   }
+   
+   return NULL;
+}
+
+
+static enum pipe_error compile_fs( struct svga_context *svga,
+                                   struct svga_fragment_shader *fs,
+                                   const struct svga_fs_compile_key *key,
+                                   struct svga_shader_result **out_result )
+{
+   struct svga_shader_result *result;
+   enum pipe_error ret;
+
+   result = svga_translate_fragment_program( fs, key );
+   if (result == NULL) {
+      ret = PIPE_ERROR_OUT_OF_MEMORY;
+      goto fail;
+   }
+
+
+   ret = SVGA3D_DefineShader(svga->swc, 
+                             svga->state.next_fs_id,
+                             SVGA3D_SHADERTYPE_PS,
+                             result->tokens, 
+                             result->nr_tokens * sizeof result->tokens[0]);
+   if (ret)
+      goto fail;
+
+   *out_result = result;
+   result->id = svga->state.next_fs_id++;
+   result->next = fs->base.results;
+   fs->base.results = result;
+   return PIPE_OK;
+
+fail:
+   if (result)
+      svga_destroy_shader_result( result );
+   return ret;
+}
+
+/* The blend workaround for simulating logicop xor behaviour requires
+ * that the incoming fragment color be white.  This change achieves
+ * that by hooking up a hard-wired fragment shader that just emits
+ * color 1,1,1,1
+ *   
+ * This is a slightly incomplete solution as it assumes that the
+ * actual bound shader has no other effects beyond generating a
+ * fragment color.  In particular shaders containing TEXKIL and/or
+ * depth-write will not have the correct behaviour, nor will those
+ * expecting to use alphatest.
+ *   
+ * These are avoidable issues, but they are not much worse than the
+ * unavoidable ones associated with this technique, so it's not clear
+ * how much effort should be expended trying to resolve them - the
+ * ultimate result will still not be correct in most cases.
+ *
+ * Shader below was generated with:
+ *   SVGA_DEBUG=tgsi ./mesa/progs/fp/fp-tri white.txt
+ */
+static int emit_white_fs( struct svga_context *svga )
+{
+   int ret;
+
+   /* ps_3_0
+    * def c0, 1.000000, 0.000000, 0.000000, 1.000000
+    * mov oC0, c0.x
+    * end
+    */
+   static const unsigned white_tokens[] = {
+      0xffff0300,
+      0x05000051,
+      0xa00f0000,
+      0x3f800000,
+      0x00000000,
+      0x00000000,
+      0x3f800000,
+      0x02000001,
+      0x800f0800,
+      0xa0000000,
+      0x0000ffff,
+   };
+
+   ret = SVGA3D_DefineShader(svga->swc, 
+                             svga->state.next_fs_id,
+                             SVGA3D_SHADERTYPE_PS,
+                             white_tokens, 
+                             sizeof(white_tokens));
+   if (ret)
+      return ret;
+
+   svga->state.white_fs_id = svga->state.next_fs_id++;
+   return 0;
+}
+
+
+/* SVGA_NEW_TEXTURE_BINDING
+ * SVGA_NEW_RAST
+ * SVGA_NEW_NEED_SWTNL
+ * SVGA_NEW_SAMPLER
+ */
+static int make_fs_key( const struct svga_context *svga,
+                        struct svga_fs_compile_key *key )
+{
+   int i;
+   int idx = 0;
+
+   memset(key, 0, sizeof *key);
+
+   /* Only need fragment shader fixup for twoside lighting if doing
+    * hwtnl.  Otherwise the draw module does the whole job for us.
+    *
+    * SVGA_NEW_SWTNL
+    */
+   if (!svga->state.sw.need_swtnl) {
+      /* SVGA_NEW_RAST
+       */
+      key->light_twoside = svga->curr.rast->templ.light_twoside;
+      key->front_cw = (svga->curr.rast->templ.front_winding == 
+                       PIPE_WINDING_CW);
+   }
+
+   
+   /* XXX: want to limit this to the textures that the shader actually
+    * refers to.
+    *
+    * SVGA_NEW_TEXTURE_BINDING | SVGA_NEW_SAMPLER
+    */
+   for (i = 0; i < svga->curr.num_textures; i++) {
+      if (svga->curr.texture[i]) {
+         assert(svga->curr.sampler[i]);
+         key->tex[i].texture_target = svga->curr.texture[i]->target;
+         if (!svga->curr.sampler[i]->normalized_coords) {
+            key->tex[i].width_height_idx = idx++;
+            key->tex[i].unnormalized = TRUE;
+            ++key->num_unnormalized_coords;
+         }
+      }
+   }
+   key->num_textures = svga->curr.num_textures;
+
+   idx = 0;
+   for (i = 0; i < svga->curr.num_samplers; ++i) {
+      if (svga->curr.sampler[i]) {
+         key->tex[i].compare_mode = svga->curr.sampler[i]->compare_mode;
+         key->tex[i].compare_func = svga->curr.sampler[i]->compare_func;
+      }
+   }
+
+   return 0;
+}
+
+
+
+static int emit_hw_fs( struct svga_context *svga,
+                       unsigned dirty )
+{
+   struct svga_shader_result *result = NULL;
+   unsigned id = SVGA3D_INVALID_ID;
+   int ret = 0;
+
+   /* SVGA_NEW_BLEND
+    */
+   if (svga->curr.blend->need_white_fragments) {
+      if (svga->state.white_fs_id == SVGA3D_INVALID_ID) {
+         ret = emit_white_fs( svga );
+         if (ret)
+            return ret;
+      }
+      id = svga->state.white_fs_id;
+   }
+   else {
+      struct svga_fragment_shader *fs = svga->curr.fs;
+      struct svga_fs_compile_key key;
+
+      /* SVGA_NEW_TEXTURE_BINDING
+       * SVGA_NEW_RAST
+       * SVGA_NEW_NEED_SWTNL
+       * SVGA_NEW_SAMPLER
+       */
+      ret = make_fs_key( svga, &key );
+      if (ret)
+         return ret;
+
+      result = search_fs_key( fs, &key );
+      if (!result) {
+         ret = compile_fs( svga, fs, &key, &result );
+         if (ret)
+            return ret;
+      }
+
+      assert (result);
+      id = result->id;
+   }
+
+   assert(id != SVGA3D_INVALID_ID);
+
+   if (id != svga->state.hw_draw.shader_id[PIPE_SHADER_FRAGMENT]) {
+      ret = SVGA3D_SetShader(svga->swc, 
+                             SVGA3D_SHADERTYPE_PS, 
+                             id );
+      if (ret)
+         return ret;
+
+      svga->dirty |= SVGA_NEW_FS_RESULT;
+      svga->state.hw_draw.shader_id[PIPE_SHADER_FRAGMENT] = id;
+      svga->state.hw_draw.fs = result;      
+   }
+
+   return 0;
+}
+
+struct svga_tracked_state svga_hw_fs = 
+{
+   "fragment shader (hwtnl)",
+   (SVGA_NEW_FS |
+    SVGA_NEW_TEXTURE_BINDING |
+    SVGA_NEW_NEED_SWTNL |
+    SVGA_NEW_RAST |
+    SVGA_NEW_SAMPLER |
+    SVGA_NEW_BLEND),
+   emit_hw_fs
+};
+
+
+
diff --git a/src/gallium/drivers/svga/svga_state_need_swtnl.c b/src/gallium/drivers/svga/svga_state_need_swtnl.c
new file mode 100644
index 0000000000..00201b8091
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_need_swtnl.c
@@ -0,0 +1,200 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_state.h"
+
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_debug.h"
+#include "svga_hw_reg.h"
+
+/***********************************************************************
+ */
+
+static INLINE SVGA3dDeclType 
+svga_translate_vertex_format(enum pipe_format format)
+{
+   switch (format) {
+   case PIPE_FORMAT_R32_FLOAT:            return SVGA3D_DECLTYPE_FLOAT1;
+   case PIPE_FORMAT_R32G32_FLOAT:         return SVGA3D_DECLTYPE_FLOAT2;
+   case PIPE_FORMAT_R32G32B32_FLOAT:      return SVGA3D_DECLTYPE_FLOAT3;
+   case PIPE_FORMAT_R32G32B32A32_FLOAT:   return SVGA3D_DECLTYPE_FLOAT4;
+   case PIPE_FORMAT_B8G8R8A8_UNORM:       return SVGA3D_DECLTYPE_D3DCOLOR;
+   case PIPE_FORMAT_R8G8B8A8_USCALED:     return SVGA3D_DECLTYPE_UBYTE4;
+   case PIPE_FORMAT_R16G16_SSCALED:       return SVGA3D_DECLTYPE_SHORT2;
+   case PIPE_FORMAT_R16G16B16A16_SSCALED: return SVGA3D_DECLTYPE_SHORT4;
+   case PIPE_FORMAT_R8G8B8A8_UNORM:       return SVGA3D_DECLTYPE_UBYTE4N;
+   case PIPE_FORMAT_R16G16_SNORM:         return SVGA3D_DECLTYPE_SHORT2N;
+   case PIPE_FORMAT_R16G16B16A16_SNORM:   return SVGA3D_DECLTYPE_SHORT4N;
+   case PIPE_FORMAT_R16G16_UNORM:         return SVGA3D_DECLTYPE_USHORT2N;
+   case PIPE_FORMAT_R16G16B16A16_UNORM:   return SVGA3D_DECLTYPE_USHORT4N;
+
+   /* These formats don't exist yet:
+    * 
+   case PIPE_FORMAT_R10G10B10_USCALED:    return SVGA3D_DECLTYPE_UDEC3;
+   case PIPE_FORMAT_R10G10B10_SNORM:      return SVGA3D_DECLTYPE_DEC3N;
+   case PIPE_FORMAT_R16G16_FLOAT:         return SVGA3D_DECLTYPE_FLOAT16_2;
+   case PIPE_FORMAT_R16G16B16A16_FLOAT:   return SVGA3D_DECLTYPE_FLOAT16_4;
+   */
+
+   default:
+      /* There are many formats without hardware support.  This case
+       * will be hit regularly, meaning we'll need swvfetch.
+       */
+      return SVGA3D_DECLTYPE_MAX;
+   }
+}
+
+
+static int update_need_swvfetch( struct svga_context *svga,
+                                 unsigned dirty )
+{
+   unsigned i;
+   boolean need_swvfetch = FALSE;
+
+   for (i = 0; i < svga->curr.num_vertex_elements; i++) {
+      svga->state.sw.ve_format[i] = svga_translate_vertex_format(svga->curr.ve[i].src_format);
+      if (svga->state.sw.ve_format[i] == SVGA3D_DECLTYPE_MAX) {
+         need_swvfetch = TRUE;
+         break;
+      }
+   }
+
+   if (need_swvfetch != svga->state.sw.need_swvfetch) {
+      svga->state.sw.need_swvfetch = need_swvfetch;
+      svga->dirty |= SVGA_NEW_NEED_SWVFETCH;
+   }
+   
+   return 0;
+}
+
+struct svga_tracked_state svga_update_need_swvfetch = 
+{
+   "update need_swvfetch",
+   ( SVGA_NEW_VELEMENT ),
+   update_need_swvfetch
+};
+
+
+/*********************************************************************** 
+ */
+
+static int update_need_pipeline( struct svga_context *svga,
+                                 unsigned dirty )
+{
+   
+   boolean need_pipeline = FALSE;
+
+   /* SVGA_NEW_RAST, SVGA_NEW_REDUCED_PRIMITIVE
+    */
+   if (svga->curr.rast->need_pipeline & (1 << svga->curr.reduced_prim)) {
+      SVGA_DBG(DEBUG_SWTNL, "%s: rast need_pipeline (%d) & prim (%x)\n", 
+                 __FUNCTION__,
+                 svga->curr.rast->need_pipeline,
+                 (1 << svga->curr.reduced_prim) );
+      need_pipeline = TRUE;
+   }
+
+   /* SVGA_NEW_EDGEFLAGS
+    */
+   if (svga->curr.rast->hw_unfilled != PIPE_POLYGON_MODE_FILL &&
+       svga->curr.reduced_prim == PIPE_PRIM_TRIANGLES && 
+       svga->curr.edgeflags != NULL) {
+      SVGA_DBG(DEBUG_SWTNL, "%s: edgeflags\n", __FUNCTION__);
+      need_pipeline = TRUE;
+   }
+
+   /* SVGA_NEW_CLIP 
+    */
+   if (!svga->curr.rast->templ.bypass_vs_clip_and_viewport &&
+       svga->curr.clip.nr) {
+      SVGA_DBG(DEBUG_SWTNL, "%s: userclip\n", __FUNCTION__);
+      need_pipeline = TRUE;
+   }
+
+   if (need_pipeline != svga->state.sw.need_pipeline) {
+      svga->state.sw.need_pipeline = need_pipeline;
+      svga->dirty |= SVGA_NEW_NEED_PIPELINE;
+   }
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_update_need_pipeline = 
+{
+   "need pipeline",
+   (SVGA_NEW_RAST |
+    SVGA_NEW_CLIP |
+    SVGA_NEW_REDUCED_PRIMITIVE),
+   update_need_pipeline
+};
+
+
+/*********************************************************************** 
+ */
+
+static int update_need_swtnl( struct svga_context *svga,
+                              unsigned dirty )
+{
+   boolean need_swtnl;
+
+   if (svga->debug.no_swtnl) {
+      svga->state.sw.need_swvfetch = 0;
+      svga->state.sw.need_pipeline = 0;
+   }
+
+   need_swtnl = (svga->state.sw.need_swvfetch ||
+                 svga->state.sw.need_pipeline);
+
+   if (svga->debug.force_swtnl) {
+      need_swtnl = 1;
+   }
+
+   if (need_swtnl != svga->state.sw.need_swtnl) {
+      SVGA_DBG(DEBUG_SWTNL|DEBUG_PERF,
+               "%s need_swvfetch: %s, need_pipeline %s\n",
+               __FUNCTION__,
+               svga->state.sw.need_swvfetch ? "true" : "false",
+               svga->state.sw.need_pipeline ? "true" : "false");
+
+      svga->state.sw.need_swtnl = need_swtnl;
+      svga->dirty |= SVGA_NEW_NEED_SWTNL;
+      svga->swtnl.new_vdecl = TRUE;
+   }
+  
+   return 0;
+}
+
+
+struct svga_tracked_state svga_update_need_swtnl =
+{
+   "need swtnl",
+   (SVGA_NEW_NEED_PIPELINE |
+    SVGA_NEW_NEED_SWVFETCH),
+   update_need_swtnl
+};
diff --git a/src/gallium/drivers/svga/svga_state_rss.c b/src/gallium/drivers/svga/svga_state_rss.c
new file mode 100644
index 0000000000..8b6803a285
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_rss.c
@@ -0,0 +1,268 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+
+#include "svga_hw_reg.h"
+
+
+
+struct rs_queue {
+   unsigned rs_count;
+   SVGA3dRenderState rs[SVGA3D_RS_MAX];
+};
+
+
+#define EMIT_RS(svga, value, token, fail)                       \
+do {                                                            \
+   if (svga->state.hw_draw.rs[SVGA3D_RS_##token] != value) {    \
+      svga_queue_rs( &queue, SVGA3D_RS_##token, value );        \
+      svga->state.hw_draw.rs[SVGA3D_RS_##token] = value;        \
+   }                                                            \
+} while (0)
+
+#define EMIT_RS_FLOAT(svga, fvalue, token, fail)                \
+do {                                                            \
+   unsigned value = fui(fvalue);                                \
+   if (svga->state.hw_draw.rs[SVGA3D_RS_##token] != value) {    \
+      svga_queue_rs( &queue, SVGA3D_RS_##token, value );        \
+      svga->state.hw_draw.rs[SVGA3D_RS_##token] = value;        \
+   }                                                            \
+} while (0)
+
+
+static INLINE void
+svga_queue_rs( struct rs_queue *q,
+               unsigned rss,
+               unsigned value )
+{
+   q->rs[q->rs_count].state = rss;
+   q->rs[q->rs_count].uintValue = value;
+   q->rs_count++;
+}
+
+
+/* Compare old and new render states and emit differences between them
+ * to hardware.  Simplest implementation would be to emit the whole of
+ * the "to" state.
+ */
+static int emit_rss( struct svga_context *svga,
+                     unsigned dirty )
+{
+   struct rs_queue queue;
+
+   queue.rs_count = 0;
+
+   if (dirty & SVGA_NEW_BLEND) {
+      const struct svga_blend_state *curr = svga->curr.blend;
+
+      EMIT_RS( svga, curr->rt[0].writemask, COLORWRITEENABLE, fail );
+      EMIT_RS( svga, curr->rt[0].blend_enable, BLENDENABLE, fail );
+
+      if (curr->rt[0].blend_enable) {
+         EMIT_RS( svga, curr->rt[0].srcblend, SRCBLEND, fail );
+         EMIT_RS( svga, curr->rt[0].dstblend, DSTBLEND, fail );
+         EMIT_RS( svga, curr->rt[0].blendeq, BLENDEQUATION, fail );
+
+         EMIT_RS( svga, curr->rt[0].separate_alpha_blend_enable, 
+                  SEPARATEALPHABLENDENABLE, fail );
+
+         if (curr->rt[0].separate_alpha_blend_enable) {
+            EMIT_RS( svga, curr->rt[0].srcblend_alpha, SRCBLENDALPHA, fail );
+            EMIT_RS( svga, curr->rt[0].dstblend_alpha, DSTBLENDALPHA, fail );
+            EMIT_RS( svga, curr->rt[0].blendeq_alpha, BLENDEQUATIONALPHA, fail );
+         }
+      }
+   }
+
+
+   if (dirty & (SVGA_NEW_DEPTH_STENCIL | SVGA_NEW_RAST)) {
+      const struct svga_depth_stencil_state *curr = svga->curr.depth; 
+      const struct svga_rasterizer_state *rast = svga->curr.rast; 
+
+      if (!curr->stencil[0].enabled) 
+      {
+         /* Stencil disabled
+          */
+         EMIT_RS( svga, FALSE, STENCILENABLE, fail );
+         EMIT_RS( svga, FALSE, STENCILENABLE2SIDED, fail );
+      }
+      else if (curr->stencil[0].enabled && !curr->stencil[1].enabled)
+      {
+         /* Regular stencil
+          */
+         EMIT_RS( svga, TRUE, STENCILENABLE, fail );
+         EMIT_RS( svga, FALSE, STENCILENABLE2SIDED, fail );
+
+         EMIT_RS( svga, curr->stencil[0].func,  STENCILFUNC, fail );
+         EMIT_RS( svga, curr->stencil[0].fail,  STENCILFAIL, fail );
+         EMIT_RS( svga, curr->stencil[0].zfail, STENCILZFAIL, fail );
+         EMIT_RS( svga, curr->stencil[0].pass,  STENCILPASS, fail );
+         
+         EMIT_RS( svga, curr->stencil_ref, STENCILREF, fail );
+         EMIT_RS( svga, curr->stencil_mask, STENCILMASK, fail );
+         EMIT_RS( svga, curr->stencil_writemask, STENCILWRITEMASK, fail );
+      }
+      else 
+      {
+         int cw, ccw;
+
+         /* Hardware frontwinding is always CW, so if ours is also CW,
+          * then our definition of front face agrees with hardware.
+          * Otherwise need to flip.
+          */
+         if (rast->templ.front_winding == PIPE_WINDING_CW) {
+            cw = 0;
+            ccw = 1;
+         }
+         else {
+            cw = 1;
+            ccw = 0;
+         }
+
+         /* Twoside stencil
+          */
+         EMIT_RS( svga, TRUE, STENCILENABLE, fail );
+         EMIT_RS( svga, TRUE, STENCILENABLE2SIDED, fail );
+
+         EMIT_RS( svga, curr->stencil[cw].func,  STENCILFUNC, fail );
+         EMIT_RS( svga, curr->stencil[cw].fail,  STENCILFAIL, fail );
+         EMIT_RS( svga, curr->stencil[cw].zfail, STENCILZFAIL, fail );
+         EMIT_RS( svga, curr->stencil[cw].pass,  STENCILPASS, fail );
+
+         EMIT_RS( svga, curr->stencil[ccw].func,  CCWSTENCILFUNC, fail );
+         EMIT_RS( svga, curr->stencil[ccw].fail,  CCWSTENCILFAIL, fail );
+         EMIT_RS( svga, curr->stencil[ccw].zfail, CCWSTENCILZFAIL, fail );
+         EMIT_RS( svga, curr->stencil[ccw].pass,  CCWSTENCILPASS, fail );
+
+         EMIT_RS( svga, curr->stencil_ref, STENCILREF, fail );
+         EMIT_RS( svga, curr->stencil_mask, STENCILMASK, fail );
+         EMIT_RS( svga, curr->stencil_writemask, STENCILWRITEMASK, fail );
+      }
+
+      EMIT_RS( svga, curr->zenable, ZENABLE, fail );
+      if (curr->zenable) {
+         EMIT_RS( svga, curr->zfunc, ZFUNC, fail );
+         EMIT_RS( svga, curr->zwriteenable, ZWRITEENABLE, fail );
+      }
+
+      EMIT_RS( svga, curr->alphatestenable, ALPHATESTENABLE, fail );
+      if (curr->alphatestenable) {
+         EMIT_RS( svga, curr->alphafunc, ALPHAFUNC, fail );
+         EMIT_RS_FLOAT( svga, curr->alpharef, ALPHAREF, fail );
+      }
+   }
+
+
+   if (dirty & SVGA_NEW_RAST)
+   {
+      const struct svga_rasterizer_state *curr = svga->curr.rast; 
+
+      /* Shademode: still need to rearrange index list to move
+       * flat-shading PV first vertex.
+       */
+      EMIT_RS( svga, curr->shademode, SHADEMODE, fail );
+      EMIT_RS( svga, curr->cullmode, CULLMODE, fail );
+      EMIT_RS( svga, curr->scissortestenable, SCISSORTESTENABLE, fail );
+      EMIT_RS( svga, curr->multisampleantialias, MULTISAMPLEANTIALIAS, fail );
+      EMIT_RS( svga, curr->lastpixel, LASTPIXEL, fail );
+      EMIT_RS( svga, curr->linepattern, LINEPATTERN, fail );
+      EMIT_RS_FLOAT( svga, curr->pointsize, POINTSIZE, fail );
+      EMIT_RS_FLOAT( svga, curr->pointsize_min, POINTSIZEMIN, fail );
+      EMIT_RS_FLOAT( svga, curr->pointsize_max, POINTSIZEMAX, fail );
+   }
+
+   if (dirty & (SVGA_NEW_RAST | SVGA_NEW_FRAME_BUFFER | SVGA_NEW_NEED_PIPELINE))
+   {
+      const struct svga_rasterizer_state *curr = svga->curr.rast; 
+      float slope = 0.0;
+      float bias  = 0.0;
+
+      /* Need to modify depth bias according to bound depthbuffer
+       * format.  Don't do hardware depthbias while the software
+       * pipeline is active.
+       */
+      if (!svga->state.sw.need_pipeline &&
+          svga->curr.framebuffer.zsbuf)
+      {
+         slope = curr->slopescaledepthbias;
+         bias  = svga->curr.depthscale * curr->depthbias;
+      }
+
+      EMIT_RS_FLOAT( svga, slope, SLOPESCALEDEPTHBIAS, fail );
+      EMIT_RS_FLOAT( svga, bias, DEPTHBIAS, fail );
+   }
+
+
+   if (queue.rs_count) {
+      SVGA3dRenderState *rs;
+
+      if (SVGA3D_BeginSetRenderState( svga->swc,
+                                      &rs,
+                                      queue.rs_count ) != PIPE_OK)
+         goto fail;
+
+      memcpy( rs,
+              queue.rs,
+              queue.rs_count * sizeof queue.rs[0]);
+      
+      SVGA_FIFOCommitAll( svga->swc );
+   }
+
+   /* Also blend color:
+    */
+
+   return 0;
+
+fail:
+   /* XXX: need to poison cached hardware state on failure to ensure
+    * dirty state gets re-emitted.  Fix this by re-instating partial
+    * FIFOCommit command and only updating cached hw state once the
+    * initial allocation has succeeded.
+    */
+   memset(svga->state.hw_draw.rs, 0xcd, sizeof(svga->state.hw_draw.rs));
+
+   return PIPE_ERROR_OUT_OF_MEMORY;
+}
+
+
+struct svga_tracked_state svga_hw_rss = 
+{
+   "hw rss state",
+
+   (SVGA_NEW_BLEND |
+    SVGA_NEW_DEPTH_STENCIL |
+    SVGA_NEW_RAST |
+    SVGA_NEW_FRAME_BUFFER |
+    SVGA_NEW_NEED_PIPELINE),
+
+   emit_rss
+};
diff --git a/src/gallium/drivers/svga/svga_state_tss.c b/src/gallium/drivers/svga/svga_state_tss.c
new file mode 100644
index 0000000000..b313794520
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_tss.c
@@ -0,0 +1,279 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+
+#include "svga_screen_texture.h"
+#include "svga_winsys.h"
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+
+#include "svga_hw_reg.h"
+
+
+void svga_cleanup_tss_binding(struct svga_context *svga)
+{
+   int i;
+   unsigned count = MAX2( svga->curr.num_textures,
+                          svga->state.hw_draw.num_views );
+
+   for (i = 0; i < count; i++) {
+      struct svga_hw_view_state *view = &svga->state.hw_draw.views[i];
+
+      svga_sampler_view_reference(&view->v, NULL);
+      pipe_texture_reference( &svga->curr.texture[i], NULL );
+      pipe_texture_reference( &view->texture, NULL );
+
+      view->dirty = 1;
+   }
+}
+
+
+static int
+update_tss_binding(struct svga_context *svga, 
+                   unsigned dirty )
+{
+   unsigned i;
+   unsigned count = MAX2( svga->curr.num_textures,
+                          svga->state.hw_draw.num_views );
+   unsigned min_lod;
+   unsigned max_lod;
+
+
+   struct {
+      struct {
+         unsigned unit;
+         struct svga_hw_view_state *view;
+      } bind[PIPE_MAX_SAMPLERS];
+
+      unsigned bind_count;
+   } queue;
+
+   queue.bind_count = 0;
+   
+   for (i = 0; i < count; i++) {
+      const struct svga_sampler_state *s = svga->curr.sampler[i];
+      struct svga_hw_view_state *view = &svga->state.hw_draw.views[i];
+
+      /* get min max lod */
+      if (svga->curr.texture[i]) {
+         min_lod = MAX2(s->view_min_lod, 0);
+         max_lod = MIN2(s->view_max_lod, svga->curr.texture[i]->last_level);
+      } else {
+         min_lod = 0;
+         max_lod = 0;
+      }
+
+      if (view->texture != svga->curr.texture[i] ||
+          view->min_lod != min_lod ||
+          view->max_lod != max_lod) {
+
+         svga_sampler_view_reference(&view->v, NULL);
+         pipe_texture_reference( &view->texture, svga->curr.texture[i] );
+
+         view->dirty = TRUE;
+         view->min_lod = min_lod;
+         view->max_lod = max_lod;
+
+         if (svga->curr.texture[i])
+            view->v = svga_get_tex_sampler_view(&svga->pipe, 
+                                                svga->curr.texture[i], 
+                                                min_lod,
+                                                max_lod);
+      }
+
+      if (view->dirty) {
+         queue.bind[queue.bind_count].unit = i;
+         queue.bind[queue.bind_count].view = view;
+         queue.bind_count++;
+      } 
+      else if (view->v) {
+         svga_validate_sampler_view(svga, view->v);
+      }
+   }
+
+   svga->state.hw_draw.num_views = svga->curr.num_textures;
+
+   if (queue.bind_count) {
+      SVGA3dTextureState *ts;
+
+      if (SVGA3D_BeginSetTextureState( svga->swc,
+                                       &ts,
+                                       queue.bind_count ) != PIPE_OK)
+         goto fail;
+
+      for (i = 0; i < queue.bind_count; i++) {
+         ts[i].stage = queue.bind[i].unit;
+         ts[i].name = SVGA3D_TS_BIND_TEXTURE;
+
+         if (queue.bind[i].view->v) {
+            svga->swc->surface_relocation(svga->swc,
+                                          &ts[i].value,
+                                          queue.bind[i].view->v->handle,
+                                          PIPE_BUFFER_USAGE_GPU_READ);
+         }
+         else {
+            ts[i].value = SVGA3D_INVALID_ID;
+         }
+         
+         queue.bind[i].view->dirty = FALSE;
+      }
+
+      SVGA_FIFOCommitAll( svga->swc );
+   }
+
+   return 0;
+
+fail:
+   return PIPE_ERROR_OUT_OF_MEMORY;
+}
+
+
+struct svga_tracked_state svga_hw_tss_binding = {
+   "texture binding emit",
+   SVGA_NEW_TEXTURE_BINDING |
+   SVGA_NEW_SAMPLER,
+   update_tss_binding
+};
+
+
+/***********************************************************************
+ */
+
+struct ts_queue {
+   unsigned ts_count;
+   SVGA3dTextureState ts[PIPE_MAX_SAMPLERS*SVGA3D_TS_MAX];
+};
+
+
+#define EMIT_TS(svga, unit, val, token, fail)                           \
+do {                                                                    \
+   if (svga->state.hw_draw.ts[unit][SVGA3D_TS_##token] != val) {        \
+      svga_queue_tss( &queue, unit, SVGA3D_TS_##token, val );           \
+      svga->state.hw_draw.ts[unit][SVGA3D_TS_##token] = val;            \
+   }                                                                    \
+} while (0)
+
+#define EMIT_TS_FLOAT(svga, unit, fvalue, token, fail)                  \
+do {                                                                    \
+   unsigned val = fui(fvalue);                                          \
+   if (svga->state.hw_draw.ts[unit][SVGA3D_TS_##token] != val) {        \
+      svga_queue_tss( &queue, unit, SVGA3D_TS_##token, val );           \
+      svga->state.hw_draw.ts[unit][SVGA3D_TS_##token] = val;            \
+   }                                                                    \
+} while (0)
+
+
+static INLINE void 
+svga_queue_tss( struct ts_queue *q,
+                unsigned unit,
+                unsigned tss,
+                unsigned value )
+{
+   assert(q->ts_count < sizeof(q->ts)/sizeof(q->ts[0]));
+   q->ts[q->ts_count].stage = unit;
+   q->ts[q->ts_count].name = tss;
+   q->ts[q->ts_count].value = value;
+   q->ts_count++;
+}
+
+
+static int
+update_tss(struct svga_context *svga, 
+           unsigned dirty )
+{
+   unsigned i;
+   struct ts_queue queue;
+
+   queue.ts_count = 0;
+   for (i = 0; i < svga->curr.num_samplers; i++) {
+      if (svga->curr.sampler[i]) {
+         const struct svga_sampler_state *curr = svga->curr.sampler[i];
+
+         EMIT_TS(svga, i, curr->mipfilter, MIPFILTER, fail);
+         EMIT_TS(svga, i, curr->min_lod, TEXTURE_MIPMAP_LEVEL, fail);
+         EMIT_TS(svga, i, curr->magfilter, MAGFILTER, fail);
+         EMIT_TS(svga, i, curr->minfilter, MINFILTER, fail);
+         EMIT_TS(svga, i, curr->aniso_level, TEXTURE_ANISOTROPIC_LEVEL, fail);
+         EMIT_TS_FLOAT(svga, i, curr->lod_bias, TEXTURE_LOD_BIAS, fail);
+         EMIT_TS(svga, i, curr->addressu, ADDRESSU, fail);
+         EMIT_TS(svga, i, curr->addressw, ADDRESSW, fail);
+         EMIT_TS(svga, i, curr->bordercolor, BORDERCOLOR, fail);
+         // TEXCOORDINDEX -- hopefully not needed
+
+         if (svga->curr.tex_flags.flag_1d & (1 << i)) {
+            debug_printf("wrap 1d tex %d\n", i);
+            EMIT_TS(svga, i, SVGA3D_TEX_ADDRESS_WRAP, ADDRESSV, fail);
+         }
+         else
+            EMIT_TS(svga, i, curr->addressv, ADDRESSV, fail);
+
+         if (svga->curr.tex_flags.flag_srgb & (1 << i))
+            EMIT_TS_FLOAT(svga, i, 2.2f, GAMMA, fail);
+         else
+            EMIT_TS_FLOAT(svga, i, 1.0f, GAMMA, fail);
+
+      }
+   }
+ 
+   if (queue.ts_count) {
+      SVGA3dTextureState *ts;
+
+      if (SVGA3D_BeginSetTextureState( svga->swc,
+                                       &ts,
+                                       queue.ts_count ) != PIPE_OK)
+         goto fail;
+
+      memcpy( ts,
+              queue.ts,
+              queue.ts_count * sizeof queue.ts[0]);
+      
+      SVGA_FIFOCommitAll( svga->swc );
+   }
+
+   return 0;
+
+fail:
+   /* XXX: need to poison cached hardware state on failure to ensure
+    * dirty state gets re-emitted.  Fix this by re-instating partial
+    * FIFOCommit command and only updating cached hw state once the
+    * initial allocation has succeeded.
+    */
+   memset(svga->state.hw_draw.ts, 0xcd, sizeof(svga->state.hw_draw.ts));
+
+   return PIPE_ERROR_OUT_OF_MEMORY;
+}
+
+
+struct svga_tracked_state svga_hw_tss = {
+   "texture state emit",
+   (SVGA_NEW_SAMPLER |
+    SVGA_NEW_TEXTURE_FLAGS),
+   update_tss
+};
+
diff --git a/src/gallium/drivers/svga/svga_state_vdecl.c b/src/gallium/drivers/svga/svga_state_vdecl.c
new file mode 100644
index 0000000000..c534308f50
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_vdecl.c
@@ -0,0 +1,182 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "util/u_upload_mgr.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_draw.h"
+#include "svga_tgsi.h"
+#include "svga_screen.h"
+#include "svga_screen_buffer.h"
+
+#include "svga_hw_reg.h"
+
+
+static int
+upload_user_buffers( struct svga_context *svga )
+{
+   enum pipe_error ret = PIPE_OK;
+   int i;
+   int nr;
+
+   if (0) 
+      debug_printf("%s: %d\n", __FUNCTION__, svga->curr.num_vertex_buffers);
+
+   nr = svga->curr.num_vertex_buffers;
+
+   for (i = 0; i < nr; i++) 
+   {
+      if (svga_buffer_is_user_buffer(svga->curr.vb[i].buffer))
+      {
+         struct pipe_buffer *upload_buffer = NULL;
+         unsigned offset = /*svga->curr.vb[i].buffer_offset*/ 0;
+         unsigned size = svga->curr.vb[i].buffer->size /*- offset*/;
+         unsigned upload_offset;
+
+         ret = u_upload_buffer( svga->upload_vb,
+                                offset,
+                                size,
+                                svga->curr.vb[i].buffer,
+                                &upload_offset,
+                                &upload_buffer );
+         if (ret)
+            return ret;
+
+         if (0)
+            debug_printf("%s: %d: orig buf %p upl buf %p ofs %d sz %d\n", 
+                         __FUNCTION__, 
+                         i,
+                         svga->curr.vb[i].buffer,
+                         upload_buffer, upload_offset, size);
+
+         /* Make sure we release the old buffer and end up with the
+          * correct refcount on the uploaded buffer.
+          */
+         pipe_buffer_reference( &svga->curr.vb[i].buffer, NULL );
+         svga->curr.vb[i].buffer = upload_buffer;
+         svga->curr.vb[i].buffer_offset = upload_offset;
+      }
+   }
+
+   if (0)
+      debug_printf("%s: DONE\n", __FUNCTION__);
+
+   return ret;
+}
+
+
+/***********************************************************************
+ */
+
+
+static int emit_hw_vs_vdecl( struct svga_context *svga,
+                             unsigned dirty )
+{
+   const struct pipe_vertex_element *ve = svga->curr.ve;
+   SVGA3dVertexDecl decl;
+   unsigned i;
+
+   assert(svga->curr.num_vertex_elements >=
+          svga->curr.vs->base.info.file_count[TGSI_FILE_INPUT]);
+
+   svga_hwtnl_reset_vdecl( svga->hwtnl, 
+                           svga->curr.num_vertex_elements );
+
+   for (i = 0; i < svga->curr.num_vertex_elements; i++) {
+      const struct pipe_vertex_buffer *vb = &svga->curr.vb[ve[i].vertex_buffer_index];
+      unsigned usage, index;
+
+
+      svga_generate_vdecl_semantics( i, &usage, &index );
+
+      /* SVGA_NEW_VELEMENT
+       */
+      decl.identity.type = svga->state.sw.ve_format[i];
+      decl.identity.method = SVGA3D_DECLMETHOD_DEFAULT;
+      decl.identity.usage = usage;
+      decl.identity.usageIndex = index;
+      decl.array.stride = vb->stride;
+      decl.array.offset = (vb->buffer_offset +
+                           ve[i].src_offset);
+
+      svga_hwtnl_vdecl( svga->hwtnl,
+                        i,
+                        &decl,
+                        vb->buffer );
+   }
+
+   return 0;
+}
+
+
+static int emit_hw_vdecl( struct svga_context *svga,
+                          unsigned dirty )
+{
+   int ret = 0;
+
+   /* SVGA_NEW_NEED_SWTNL
+    */
+   if (svga->state.sw.need_swtnl)
+      return 0; /* Do not emit during swtnl */
+
+   /* If we get to here, we know that we're going to draw.  Upload
+    * userbuffers now and try to combine multiple userbuffers from
+    * multiple draw calls into a single host buffer for performance.
+    */
+   if (svga->curr.any_user_vertex_buffers &&
+       SVGA_COMBINE_USERBUFFERS)
+   {
+      ret = upload_user_buffers( svga );
+      if (ret)
+         return ret;
+
+      svga->curr.any_user_vertex_buffers = FALSE;
+   }
+
+   return emit_hw_vs_vdecl( svga, dirty );
+}
+
+
+struct svga_tracked_state svga_hw_vdecl = 
+{
+   "hw vertex decl state (hwtnl version)",
+   ( SVGA_NEW_NEED_SWTNL |
+     SVGA_NEW_VELEMENT |
+     SVGA_NEW_VBUFFER |
+     SVGA_NEW_RAST |
+     SVGA_NEW_FS |
+     SVGA_NEW_VS ),
+   emit_hw_vdecl
+};
+
+
+
+
+
+
diff --git a/src/gallium/drivers/svga/svga_state_vs.c b/src/gallium/drivers/svga/svga_state_vs.c
new file mode 100644
index 0000000000..a947745732
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_state_vs.c
@@ -0,0 +1,239 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "pipe/p_inlines.h"
+#include "pipe/p_defines.h"
+#include "util/u_math.h"
+#include "translate/translate.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_cmd.h"
+#include "svga_tgsi.h"
+
+#include "svga_hw_reg.h"
+
+/***********************************************************************
+ */
+
+
+static INLINE int compare_vs_keys( const struct svga_vs_compile_key *a,
+                                   const struct svga_vs_compile_key *b )
+{
+   unsigned keysize = svga_vs_key_size( a );
+   return memcmp( a, b, keysize );
+}
+
+
+static struct svga_shader_result *search_vs_key( struct svga_vertex_shader *vs,
+                                                 const struct svga_vs_compile_key *key )
+{
+   struct svga_shader_result *result = vs->base.results;
+
+   assert(key);
+
+   for ( ; result; result = result->next) {
+      if (compare_vs_keys( key, &result->key.vkey ) == 0)
+         return result;
+   }
+   
+   return NULL;
+}
+
+
+static enum pipe_error compile_vs( struct svga_context *svga,
+                                   struct svga_vertex_shader *vs,
+                                   const struct svga_vs_compile_key *key,
+                                   struct svga_shader_result **out_result )
+{
+   struct svga_shader_result *result;
+   enum pipe_error ret = PIPE_OK;
+
+   result = svga_translate_vertex_program( vs, key );
+   if (result == NULL) {
+      ret = PIPE_ERROR_OUT_OF_MEMORY;
+      goto fail;
+   }
+
+   ret = SVGA3D_DefineShader(svga->swc, 
+                             svga->state.next_vs_id,
+                             SVGA3D_SHADERTYPE_VS,
+                             result->tokens, 
+                             result->nr_tokens * sizeof result->tokens[0]);
+   if (ret)
+      goto fail;
+
+   *out_result = result;
+   result->id = svga->state.next_vs_id++;
+   result->next = vs->base.results;
+   vs->base.results = result;
+   return PIPE_OK;
+
+fail:
+   if (result)
+      svga_destroy_shader_result( result );
+   return ret;
+}
+
+/* SVGA_NEW_PRESCALE, SVGA_NEW_RAST, SVGA_NEW_ZERO_STRIDE
+ */
+static int make_vs_key( struct svga_context *svga,
+                        struct svga_vs_compile_key *key )
+{
+   memset(key, 0, sizeof *key);
+   key->need_prescale = svga->state.hw_clear.prescale.enabled;
+   key->allow_psiz = svga->curr.rast->templ.point_size_per_vertex;
+   key->zero_stride_vertex_elements =
+      svga->curr.zero_stride_vertex_elements;
+   key->num_zero_stride_vertex_elements =
+      svga->curr.num_zero_stride_vertex_elements;
+   return 0;
+}
+
+
+
+static int emit_hw_vs( struct svga_context *svga,
+                       unsigned dirty )
+{
+   struct svga_shader_result *result = NULL;
+   unsigned id = SVGA3D_INVALID_ID;
+   int ret = 0;
+
+   /* SVGA_NEW_NEED_SWTNL */
+   if (!svga->state.sw.need_swtnl) {
+      struct svga_vertex_shader *vs = svga->curr.vs;
+      struct svga_vs_compile_key key;
+
+      ret = make_vs_key( svga, &key );
+      if (ret)
+         return ret;
+
+      result = search_vs_key( vs, &key );
+      if (!result) {
+         ret = compile_vs( svga, vs, &key, &result );
+         if (ret)
+            return ret;
+      }
+
+      assert (result);
+      id = result->id;
+   }
+
+   if (id != svga->state.hw_draw.shader_id[PIPE_SHADER_VERTEX]) {
+      ret = SVGA3D_SetShader(svga->swc, 
+                             SVGA3D_SHADERTYPE_VS, 
+                             id );
+      if (ret)
+         return ret;
+
+      svga->dirty |= SVGA_NEW_VS_RESULT;
+      svga->state.hw_draw.shader_id[PIPE_SHADER_VERTEX] = id;
+      svga->state.hw_draw.vs = result;      
+   }
+
+   return 0;
+}
+
+struct svga_tracked_state svga_hw_vs = 
+{
+   "vertex shader (hwtnl)",
+   (SVGA_NEW_VS |
+    SVGA_NEW_PRESCALE |
+    SVGA_NEW_NEED_SWTNL |
+    SVGA_NEW_ZERO_STRIDE),
+   emit_hw_vs
+};
+
+
+/***********************************************************************
+ */
+static int update_zero_stride( struct svga_context *svga,
+                               unsigned dirty )
+{
+   unsigned i;
+
+   svga->curr.zero_stride_vertex_elements = 0;
+   svga->curr.num_zero_stride_vertex_elements = 0;
+
+   for (i = 0; i < svga->curr.num_vertex_elements; i++) {
+      const struct pipe_vertex_element *vel = &svga->curr.ve[i];
+      const struct pipe_vertex_buffer *vbuffer = &svga->curr.vb[
+         vel->vertex_buffer_index];
+      if (vbuffer->stride == 0) {
+         unsigned const_idx =
+            svga->curr.num_zero_stride_vertex_elements;
+         struct translate *translate;
+         struct translate_key key;
+         void *mapped_buffer;
+
+         svga->curr.zero_stride_vertex_elements |= (1 << i);
+         ++svga->curr.num_zero_stride_vertex_elements;
+
+         key.output_stride = 4 * sizeof(float);
+         key.nr_elements = 1;
+         key.element[0].input_format = vel->src_format;
+         key.element[0].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+         key.element[0].input_buffer = vel->vertex_buffer_index;
+         key.element[0].input_offset = vel->src_offset;
+         key.element[0].output_offset = const_idx * 4 * sizeof(float);
+
+         translate_key_sanitize(&key);
+         /* translate_generic_create is technically private but
+          * we don't want to code-generate, just want generic
+          * translation */
+         translate = translate_generic_create(&key);
+
+         assert(vel->src_offset == 0);
+         
+         mapped_buffer = pipe_buffer_map_range(svga->pipe.screen, 
+                                               vbuffer->buffer,
+                                               vel->src_offset,
+                                               pf_get_size(vel->src_format),
+                                               PIPE_BUFFER_USAGE_CPU_READ);
+         translate->set_buffer(translate, vel->vertex_buffer_index,
+                               mapped_buffer,
+                               vbuffer->stride);
+         translate->run(translate, 0, 1,
+                        svga->curr.zero_stride_constants);
+
+         pipe_buffer_unmap(svga->pipe.screen,
+                           vbuffer->buffer);
+         translate->release(translate);
+      }
+   }
+
+   if (svga->curr.num_zero_stride_vertex_elements)
+      svga->dirty |= SVGA_NEW_ZERO_STRIDE;
+
+   return 0;
+}
+
+struct svga_tracked_state svga_hw_update_zero_stride =
+{
+   "update zero_stride",
+   ( SVGA_NEW_VELEMENT |
+     SVGA_NEW_VBUFFER ),
+   update_zero_stride
+};
diff --git a/src/gallium/drivers/svga/svga_swtnl.h b/src/gallium/drivers/svga/svga_swtnl.h
new file mode 100644
index 0000000000..4882f26b17
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_swtnl.h
@@ -0,0 +1,52 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_SWTNL_H
+#define SVGA_SWTNL_H
+
+#include "pipe/p_compiler.h"
+
+struct svga_context;
+struct pipe_context;
+struct pipe_buffer;
+struct vbuf_render;
+
+
+boolean svga_init_swtnl( struct svga_context *svga );
+void svga_destroy_swtnl( struct svga_context *svga );
+
+
+enum pipe_error
+svga_swtnl_draw_range_elements(struct svga_context *svga,
+                               struct pipe_buffer *indexBuffer,
+                               unsigned indexSize,
+                               unsigned min_index,
+                               unsigned max_index,
+                               unsigned prim, 
+                               unsigned start, 
+                               unsigned count);
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_swtnl_backend.c b/src/gallium/drivers/svga/svga_swtnl_backend.c
new file mode 100644
index 0000000000..b4f757a47a
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_swtnl_backend.c
@@ -0,0 +1,349 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "draw/draw_vbuf.h"
+#include "draw/draw_context.h"
+#include "draw/draw_vertex.h"
+
+#include "util/u_debug.h"
+#include "pipe/p_inlines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "util/u_simple_shaders.h"
+
+#include "svga_context.h"
+#include "svga_state.h"
+#include "svga_swtnl.h"
+
+#include "svga_types.h"
+#include "svga_reg.h"
+#include "svga3d_reg.h"
+#include "svga_draw.h"
+#include "svga_swtnl_private.h"
+
+
+static const struct vertex_info *
+svga_vbuf_render_get_vertex_info( struct vbuf_render *render )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+
+   svga_swtnl_update_vdecl(svga);
+
+   return &svga_render->vertex_info;
+}
+
+
+static boolean
+svga_vbuf_render_allocate_vertices( struct vbuf_render *render,
+                                    ushort vertex_size,
+                                    ushort nr_vertices )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+   struct pipe_screen *screen = svga->pipe.screen;
+   size_t size = (size_t)nr_vertices * (size_t)vertex_size;
+   boolean new_vbuf = FALSE;
+   boolean new_ibuf = FALSE;
+
+   if (svga_render->vertex_size != vertex_size)
+      svga->swtnl.new_vdecl = TRUE;
+   svga_render->vertex_size = (size_t)vertex_size;
+
+   if (svga->swtnl.new_vbuf)
+      new_ibuf = new_vbuf = TRUE;
+   svga->swtnl.new_vbuf = FALSE;
+
+   if (svga_render->vbuf_size < svga_render->vbuf_offset + svga_render->vbuf_used + size)
+      new_vbuf = TRUE;
+
+   if (new_vbuf)
+      pipe_buffer_reference(&svga_render->vbuf, NULL);
+   if (new_ibuf)
+      pipe_buffer_reference(&svga_render->ibuf, NULL);
+
+   if (!svga_render->vbuf) {
+      svga_render->vbuf_size = MAX2(size, svga_render->vbuf_alloc_size);
+      svga_render->vbuf = pipe_buffer_create(screen,
+                                             0,
+                                             PIPE_BUFFER_USAGE_VERTEX,
+                                             svga_render->vbuf_size);
+      if(!svga_render->vbuf) {
+         svga_context_flush(svga, NULL);
+         svga_render->vbuf = pipe_buffer_create(screen,
+                                                0,
+                                                PIPE_BUFFER_USAGE_VERTEX,
+                                                svga_render->vbuf_size);
+         assert(svga_render->vbuf);
+      }
+
+      svga->swtnl.new_vdecl = TRUE;
+      svga_render->vbuf_offset = 0;
+   } else {
+      svga_render->vbuf_offset += svga_render->vbuf_used;
+   }
+
+   svga_render->vbuf_used = 0;
+
+   if (svga->swtnl.new_vdecl)
+      svga_render->vdecl_offset = svga_render->vbuf_offset;
+
+   return TRUE;
+}
+
+static void *
+svga_vbuf_render_map_vertices( struct vbuf_render *render )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+   struct pipe_screen *screen = svga->pipe.screen;
+
+   char *ptr = (char*)pipe_buffer_map(screen,
+                                      svga_render->vbuf,
+                                      PIPE_BUFFER_USAGE_CPU_WRITE | 
+                                      PIPE_BUFFER_USAGE_FLUSH_EXPLICIT);
+   return ptr + svga_render->vbuf_offset;
+}
+
+static void
+svga_vbuf_render_unmap_vertices( struct vbuf_render *render,
+                                 ushort min_index,
+                                 ushort max_index )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+   struct pipe_screen *screen = svga->pipe.screen;
+   unsigned offset, length;
+   size_t used = svga_render->vertex_size * ((size_t)max_index + 1);
+
+   offset = svga_render->vbuf_offset + svga_render->vertex_size * min_index;
+   length = svga_render->vertex_size * (max_index + 1 - min_index);
+   pipe_buffer_flush_mapped_range(screen, svga_render->vbuf, offset, length);
+   pipe_buffer_unmap(screen, svga_render->vbuf);
+   svga_render->min_index = min_index;
+   svga_render->max_index = max_index;
+   svga_render->vbuf_used = MAX2(svga_render->vbuf_used, used);
+}
+
+static boolean
+svga_vbuf_render_set_primitive( struct vbuf_render *render,
+                                unsigned prim )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   svga_render->prim = prim;
+
+   return TRUE;
+}
+
+static void
+svga_vbuf_sumbit_state( struct svga_vbuf_render *svga_render )
+{
+   struct svga_context *svga = svga_render->svga;
+   SVGA3dVertexDecl vdecl[PIPE_MAX_ATTRIBS];
+   enum pipe_error ret;
+   int i;
+
+   /* if the vdecl or vbuf hasn't changed do nothing */
+   if (!svga->swtnl.new_vdecl)
+      return;
+
+   memcpy(vdecl, svga_render->vdecl, sizeof(vdecl));
+
+   /* flush the hw state */
+   ret = svga_hwtnl_flush(svga->hwtnl);
+   if (ret) {
+      svga_context_flush(svga, NULL);
+      ret = svga_hwtnl_flush(svga->hwtnl);
+      /* if we hit this path we might become synced with hw */
+      svga->swtnl.new_vbuf = TRUE;
+      assert(ret == 0);
+   }
+
+   svga_hwtnl_reset_vdecl(svga->hwtnl, svga_render->vdecl_count);
+
+   for (i = 0; i < svga_render->vdecl_count; i++) {
+      vdecl[i].array.offset += svga_render->vdecl_offset;
+
+      svga_hwtnl_vdecl( svga->hwtnl,
+                        i,
+                        &vdecl[i],
+                        svga_render->vbuf );
+   }
+
+   /* We have already taken care of flatshading, so let the hwtnl
+    * module use whatever is most convenient:
+    */
+   if (svga->state.sw.need_pipeline) {
+      svga_hwtnl_set_flatshade(svga->hwtnl, FALSE, FALSE);
+      svga_hwtnl_set_unfilled(svga->hwtnl, PIPE_POLYGON_MODE_FILL);
+   }
+   else {
+      svga_hwtnl_set_flatshade( svga->hwtnl,
+                                svga->curr.rast->templ.flatshade,
+                                svga->curr.rast->templ.flatshade_first );
+
+      svga_hwtnl_set_unfilled( svga->hwtnl,
+                               svga->curr.rast->hw_unfilled );
+   }
+
+   svga->swtnl.new_vdecl = FALSE;
+}
+
+static void
+svga_vbuf_render_draw_arrays( struct vbuf_render *render,
+                              unsigned start,
+                              uint nr )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+   unsigned bias = (svga_render->vbuf_offset - svga_render->vdecl_offset) / svga_render->vertex_size;
+   enum pipe_error ret = 0;
+
+   svga_vbuf_sumbit_state(svga_render);
+
+   /* Need to call update_state() again as the draw module may have
+    * altered some of our state behind our backs.  Testcase:
+    * redbook/polys.c
+    */
+   svga_update_state_retry( svga, SVGA_STATE_HW_DRAW );
+
+   ret = svga_hwtnl_draw_arrays(svga->hwtnl, svga_render->prim, start + bias, nr);
+   if (ret != PIPE_OK) {
+      svga_context_flush(svga, NULL);
+      ret = svga_hwtnl_draw_arrays(svga->hwtnl, svga_render->prim, start + bias, nr);
+      svga->swtnl.new_vbuf = TRUE;
+      assert(ret == PIPE_OK);
+   }
+}
+
+
+static void
+svga_vbuf_render_draw( struct vbuf_render *render,
+                       const ushort *indices,
+                       uint nr_indices)
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+   struct svga_context *svga = svga_render->svga;
+   struct pipe_screen *screen = svga->pipe.screen;
+   unsigned bias = (svga_render->vbuf_offset - svga_render->vdecl_offset) / svga_render->vertex_size;
+   boolean ret;
+   size_t size = 2 * nr_indices;
+
+   assert(( svga_render->vbuf_offset - svga_render->vdecl_offset) % svga_render->vertex_size == 0);
+   
+   if (svga_render->ibuf_size < svga_render->ibuf_offset + size)
+      pipe_buffer_reference(&svga_render->ibuf, NULL);
+
+   if (!svga_render->ibuf) {
+      svga_render->ibuf_size = MAX2(size, svga_render->ibuf_alloc_size);
+      svga_render->ibuf = pipe_buffer_create(screen,
+                                             0,
+                                             PIPE_BUFFER_USAGE_VERTEX,
+                                             svga_render->ibuf_size);
+      svga_render->ibuf_offset = 0;
+   }
+
+   pipe_buffer_write(screen, svga_render->ibuf,
+                     svga_render->ibuf_offset, 2 * nr_indices, indices);
+
+
+   /* off to hardware */
+   svga_vbuf_sumbit_state(svga_render);
+
+   /* Need to call update_state() again as the draw module may have
+    * altered some of our state behind our backs.  Testcase:
+    * redbook/polys.c
+    */
+   svga_update_state_retry( svga, SVGA_STATE_HW_DRAW );
+
+   ret = svga_hwtnl_draw_range_elements(svga->hwtnl,
+                                        svga_render->ibuf,
+                                        2,
+                                        svga_render->min_index,
+                                        svga_render->max_index,
+                                        svga_render->prim,
+                                        svga_render->ibuf_offset / 2, nr_indices, bias);
+   if(ret != PIPE_OK) {
+      svga_context_flush(svga, NULL);
+      ret = svga_hwtnl_draw_range_elements(svga->hwtnl,
+                                           svga_render->ibuf,
+                                           2,
+                                           svga_render->min_index,
+                                           svga_render->max_index,
+                                           svga_render->prim,
+                                           svga_render->ibuf_offset / 2, nr_indices, bias);
+      svga->swtnl.new_vbuf = TRUE;
+      assert(ret == PIPE_OK);
+   }
+
+   svga_render->ibuf_offset += size;
+}
+
+
+static void
+svga_vbuf_render_release_vertices( struct vbuf_render *render )
+{
+
+}
+
+
+static void
+svga_vbuf_render_destroy( struct vbuf_render *render )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(render);
+
+   pipe_buffer_reference(&svga_render->vbuf, NULL);
+   pipe_buffer_reference(&svga_render->ibuf, NULL);
+   FREE(svga_render);
+}
+
+
+/**
+ * Create a new primitive render.
+ */
+struct vbuf_render *
+svga_vbuf_render_create( struct svga_context *svga )
+{
+   struct svga_vbuf_render *svga_render = CALLOC_STRUCT(svga_vbuf_render);
+
+   svga_render->svga = svga;
+   svga_render->ibuf_size = 0;
+   svga_render->vbuf_size = 0;
+   svga_render->ibuf_alloc_size = 4*1024;
+   svga_render->vbuf_alloc_size = 64*1024;
+   svga_render->base.max_vertex_buffer_bytes = 64*1024/10;
+   svga_render->base.max_indices = 65536;
+   svga_render->base.get_vertex_info = svga_vbuf_render_get_vertex_info;
+   svga_render->base.allocate_vertices = svga_vbuf_render_allocate_vertices;
+   svga_render->base.map_vertices = svga_vbuf_render_map_vertices;
+   svga_render->base.unmap_vertices = svga_vbuf_render_unmap_vertices;
+   svga_render->base.set_primitive = svga_vbuf_render_set_primitive;
+   svga_render->base.draw = svga_vbuf_render_draw;
+   svga_render->base.draw_arrays = svga_vbuf_render_draw_arrays;
+   svga_render->base.release_vertices = svga_vbuf_render_release_vertices;
+   svga_render->base.destroy = svga_vbuf_render_destroy;
+
+   return &svga_render->base;
+}
diff --git a/src/gallium/drivers/svga/svga_swtnl_draw.c b/src/gallium/drivers/svga/svga_swtnl_draw.c
new file mode 100644
index 0000000000..8b14c913f7
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_swtnl_draw.c
@@ -0,0 +1,170 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "draw/draw_context.h"
+#include "draw/draw_vbuf.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_state.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_swtnl.h"
+#include "svga_state.h"
+#include "svga_swtnl_private.h"
+
+
+
+enum pipe_error
+svga_swtnl_draw_range_elements(struct svga_context *svga,
+                               struct pipe_buffer *indexBuffer,
+                               unsigned indexSize,
+                               unsigned min_index,
+                               unsigned max_index,
+                               unsigned prim, unsigned start, unsigned count)
+{
+   struct draw_context *draw = svga->swtnl.draw;
+   unsigned i;
+   const void *map;
+   enum pipe_error ret;
+
+   assert(!svga->dirty);
+   assert(svga->state.sw.need_swtnl);
+   assert(draw);
+
+   ret = svga_update_state(svga, SVGA_STATE_SWTNL_DRAW);
+   if (ret) {
+      svga_context_flush(svga, NULL);
+      ret = svga_update_state(svga, SVGA_STATE_SWTNL_DRAW);
+      svga->swtnl.new_vbuf = TRUE;
+      assert(ret == PIPE_OK);
+   }
+
+   /*
+    * Map vertex buffers
+    */
+   for (i = 0; i < svga->curr.num_vertex_buffers; i++) {
+      map = pipe_buffer_map(svga->pipe.screen,
+                            svga->curr.vb[i].buffer,
+                            PIPE_BUFFER_USAGE_CPU_READ);
+
+      draw_set_mapped_vertex_buffer(draw, i, map);
+   }
+
+   /* Map index buffer, if present */
+   if (indexBuffer) {
+      map = pipe_buffer_map(svga->pipe.screen, indexBuffer,
+                            PIPE_BUFFER_USAGE_CPU_READ);
+
+      draw_set_mapped_element_buffer_range(draw, 
+                                           indexSize, 
+                                           min_index,
+                                           max_index,
+                                           map);
+   }
+   
+   if (svga->curr.cb[PIPE_SHADER_VERTEX]) {
+      map = pipe_buffer_map(svga->pipe.screen,
+                            svga->curr.cb[PIPE_SHADER_VERTEX],
+                            PIPE_BUFFER_USAGE_CPU_READ);
+      assert(map);
+      draw_set_mapped_constant_buffer(
+         draw, 
+         map,
+         svga->curr.cb[PIPE_SHADER_VERTEX]->size);
+   }
+
+   draw_arrays(svga->swtnl.draw, prim, start, count);
+
+   draw_flush(svga->swtnl.draw);
+
+   /* Ensure the draw module didn't touch this */
+   assert(i == svga->curr.num_vertex_buffers);
+   
+   /*
+    * unmap vertex/index buffers
+    */
+   for (i = 0; i < svga->curr.num_vertex_buffers; i++) {
+      pipe_buffer_unmap(svga->pipe.screen, svga->curr.vb[i].buffer);
+      draw_set_mapped_vertex_buffer(draw, i, NULL);
+   }
+
+   if (indexBuffer) {
+      pipe_buffer_unmap(svga->pipe.screen, indexBuffer);
+      draw_set_mapped_element_buffer(draw, 0, NULL);
+   }
+
+   if (svga->curr.cb[PIPE_SHADER_VERTEX]) {
+      pipe_buffer_unmap(svga->pipe.screen,
+                        svga->curr.cb[PIPE_SHADER_VERTEX]);
+   }
+
+   return ret;
+}
+
+
+
+
+boolean svga_init_swtnl( struct svga_context *svga )
+{
+   svga->swtnl.backend = svga_vbuf_render_create(svga);
+   if(!svga->swtnl.backend)
+      goto fail;
+
+   /*
+    * Create drawing context and plug our rendering stage into it.
+    */
+   svga->swtnl.draw = draw_create();
+   if (svga->swtnl.draw == NULL)
+      goto fail;
+
+
+   draw_set_rasterize_stage(svga->swtnl.draw, 
+                            draw_vbuf_stage( svga->swtnl.draw, svga->swtnl.backend ));
+
+   draw_set_render(svga->swtnl.draw, svga->swtnl.backend);
+
+   draw_install_aaline_stage(svga->swtnl.draw, &svga->pipe);
+   draw_install_aapoint_stage(svga->swtnl.draw, &svga->pipe);
+   draw_install_pstipple_stage(svga->swtnl.draw, &svga->pipe);
+
+   draw_set_driver_clipping(svga->swtnl.draw, debug_get_bool_option("SVGA_SWTNL_FSE", FALSE));
+
+   return TRUE;
+
+fail:
+   if (svga->swtnl.backend)
+      svga->swtnl.backend->destroy( svga->swtnl.backend );
+
+   if (svga->swtnl.draw)
+      draw_destroy( svga->swtnl.draw );
+
+   return FALSE;
+}
+
+
+void svga_destroy_swtnl( struct svga_context *svga )
+{
+   draw_destroy( svga->swtnl.draw );
+}
diff --git a/src/gallium/drivers/svga/svga_swtnl_private.h b/src/gallium/drivers/svga/svga_swtnl_private.h
new file mode 100644
index 0000000000..9bbb42910f
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_swtnl_private.h
@@ -0,0 +1,93 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_SWTNL_PRIVATE_H
+#define SVGA_SWTNL_PRIVATE_H
+
+#include "svga_swtnl.h"
+#include "draw/draw_vertex.h"
+
+#include "svga_types.h"
+#include "svga3d_reg.h"
+
+/**
+ * Primitive renderer for svga.
+ */
+struct svga_vbuf_render {
+   struct vbuf_render base;
+
+   struct svga_context *svga;
+   struct vertex_info vertex_info;
+
+   unsigned vertex_size;
+
+   unsigned prim;
+
+   struct pipe_buffer *vbuf;
+   struct pipe_buffer *ibuf;
+
+   /* current size of buffer */
+   size_t vbuf_size;
+   size_t ibuf_size;
+
+   /* size of that the buffer should be */
+   size_t vbuf_alloc_size;
+   size_t ibuf_alloc_size;
+
+   /* current write place */
+   size_t vbuf_offset;
+   size_t ibuf_offset;
+
+   /* currently used */
+   size_t vbuf_used;
+
+   SVGA3dVertexDecl vdecl[PIPE_MAX_ATTRIBS];
+   unsigned vdecl_offset;
+   unsigned vdecl_count;
+
+   ushort min_index;
+   ushort max_index;
+};
+
+/**
+ * Basically a cast wrapper.
+ */
+static INLINE struct svga_vbuf_render *
+svga_vbuf_render( struct vbuf_render *render )
+{
+   assert(render);
+   return (struct svga_vbuf_render *)render;
+}
+
+
+struct vbuf_render *
+svga_vbuf_render_create( struct svga_context *svga );
+
+
+int
+svga_swtnl_update_vdecl( struct svga_context *svga );
+
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_swtnl_state.c b/src/gallium/drivers/svga/svga_swtnl_state.c
new file mode 100644
index 0000000000..1616312113
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_swtnl_state.c
@@ -0,0 +1,242 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#include "draw/draw_context.h"
+#include "draw/draw_vbuf.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_state.h"
+#include "util/u_memory.h"
+
+#include "svga_context.h"
+#include "svga_swtnl.h"
+#include "svga_state.h"
+
+#include "svga_swtnl_private.h"
+
+
+#define SVGA_POINT_ADJ_X -0.375
+#define SVGA_POINT_ADJ_Y -0.5
+
+#define SVGA_LINE_ADJ_X -0.5
+#define SVGA_LINE_ADJ_Y -0.5
+
+#define SVGA_TRIANGLE_ADJ_X -0.375
+#define SVGA_TRIANGLE_ADJ_Y -0.5
+
+
+static void set_draw_viewport( struct svga_context *svga )
+{
+   struct pipe_viewport_state vp = svga->curr.viewport;
+   float adjx = 0;
+   float adjy = 0;
+
+   switch (svga->curr.reduced_prim) {
+   case PIPE_PRIM_POINTS:
+      adjx = SVGA_POINT_ADJ_X;
+      adjy = SVGA_POINT_ADJ_Y;
+      break;
+   case PIPE_PRIM_LINES:
+      /* XXX: This is to compensate for the fact that wide lines are
+       * going to be drawn with triangles, but we're not catching all
+       * cases where that will happen.
+       */
+      if (svga->curr.rast->templ.line_width > 1.0) 
+      {
+         adjx = SVGA_LINE_ADJ_X + 0.175;
+         adjy = SVGA_LINE_ADJ_Y - 0.175;
+      }
+      else {
+         adjx = SVGA_LINE_ADJ_X;
+         adjy = SVGA_LINE_ADJ_Y;
+      }
+      break;
+   case PIPE_PRIM_TRIANGLES:
+      adjx += SVGA_TRIANGLE_ADJ_X;
+      adjy += SVGA_TRIANGLE_ADJ_Y;
+      break;
+   }
+
+   vp.translate[0] += adjx;
+   vp.translate[1] += adjy;
+
+   draw_set_viewport_state(svga->swtnl.draw, &vp);
+}
+
+static int update_swtnl_draw( struct svga_context *svga,
+                              unsigned dirty )
+{
+   draw_flush( svga->swtnl.draw );
+
+   if (dirty & SVGA_NEW_VS) 
+      draw_bind_vertex_shader(svga->swtnl.draw,
+                              svga->curr.vs->draw_shader);
+
+   if (dirty & SVGA_NEW_VBUFFER)
+      draw_set_vertex_buffers(svga->swtnl.draw, 
+                              svga->curr.num_vertex_buffers, 
+                              svga->curr.vb);
+
+   if (dirty & SVGA_NEW_VELEMENT)
+      draw_set_vertex_elements(svga->swtnl.draw, 
+                               svga->curr.num_vertex_elements, 
+                               svga->curr.ve );
+
+   if (dirty & SVGA_NEW_CLIP)
+      draw_set_clip_state(svga->swtnl.draw, 
+                          &svga->curr.clip);
+
+   if (dirty & (SVGA_NEW_VIEWPORT |
+                SVGA_NEW_REDUCED_PRIMITIVE | 
+                SVGA_NEW_RAST))
+      set_draw_viewport( svga );
+
+   if (dirty & SVGA_NEW_RAST)
+      draw_set_rasterizer_state(svga->swtnl.draw,
+                                &svga->curr.rast->templ);
+
+   if (dirty & SVGA_NEW_FRAME_BUFFER)
+      draw_set_mrd(svga->swtnl.draw, 
+                   svga->curr.depthscale);
+
+   if (dirty & SVGA_NEW_EDGEFLAGS)
+      draw_set_edgeflags( svga->swtnl.draw, 
+                          svga->curr.edgeflags );
+
+   return 0;
+}
+
+
+struct svga_tracked_state svga_update_swtnl_draw =
+{
+   "update draw module state",
+   (SVGA_NEW_VS |
+    SVGA_NEW_VBUFFER |
+    SVGA_NEW_VELEMENT |
+    SVGA_NEW_CLIP |
+    SVGA_NEW_VIEWPORT |
+    SVGA_NEW_RAST |
+    SVGA_NEW_FRAME_BUFFER |
+    SVGA_NEW_REDUCED_PRIMITIVE |
+    SVGA_NEW_EDGEFLAGS),
+   update_swtnl_draw
+};
+
+
+int svga_swtnl_update_vdecl( struct svga_context *svga )
+{
+   struct svga_vbuf_render *svga_render = svga_vbuf_render(svga->swtnl.backend);
+   struct draw_context *draw = svga->swtnl.draw;
+   struct vertex_info *vinfo = &svga_render->vertex_info;
+   SVGA3dVertexDecl vdecl[PIPE_MAX_ATTRIBS];
+   const enum interp_mode colorInterp =
+      svga->curr.rast->templ.flatshade ? INTERP_CONSTANT : INTERP_LINEAR;
+   const struct svga_fragment_shader *fs = svga->curr.fs;
+   int offset = 0;
+   int nr_decls = 0;
+   int src, i;
+
+   memset(vinfo, 0, sizeof(*vinfo));
+   memset(vdecl, 0, sizeof(vdecl));
+
+   /* always add position */
+   src = draw_find_vs_output(draw, TGSI_SEMANTIC_POSITION, 0);
+   draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_LINEAR, src);
+   vinfo->attrib[0].emit = EMIT_4F;
+   vdecl[0].array.offset = offset;
+   vdecl[0].identity.type = SVGA3D_DECLTYPE_FLOAT4;
+   vdecl[0].identity.usage = SVGA3D_DECLUSAGE_POSITIONT;
+   vdecl[0].identity.usageIndex = 0;
+   offset += 16;
+   nr_decls++;
+
+   for (i = 0; i < fs->base.info.num_inputs; i++) {
+      unsigned name = fs->base.info.input_semantic_name[i];
+      unsigned index = fs->base.info.input_semantic_index[i];
+      src = draw_find_vs_output(draw, name, index);
+      vdecl[nr_decls].array.offset = offset;
+      vdecl[nr_decls].identity.usageIndex = fs->base.info.input_semantic_index[i];
+
+      switch (name) {
+      case TGSI_SEMANTIC_COLOR:
+         draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
+         vdecl[nr_decls].identity.usage = SVGA3D_DECLUSAGE_COLOR;
+         vdecl[nr_decls].identity.type = SVGA3D_DECLTYPE_FLOAT4;
+         offset += 16;
+         nr_decls++;
+         break;
+      case TGSI_SEMANTIC_GENERIC:
+         draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+         vdecl[nr_decls].identity.usage = SVGA3D_DECLUSAGE_TEXCOORD;
+         vdecl[nr_decls].identity.type = SVGA3D_DECLTYPE_FLOAT4;
+         vdecl[nr_decls].identity.usageIndex += 1;
+         offset += 16;
+         nr_decls++;
+         break;
+      case TGSI_SEMANTIC_FOG:
+         draw_emit_vertex_attr(vinfo, EMIT_1F, INTERP_PERSPECTIVE, src);
+         vdecl[nr_decls].identity.usage = SVGA3D_DECLUSAGE_TEXCOORD;
+         vdecl[nr_decls].identity.type = SVGA3D_DECLTYPE_FLOAT1;
+         assert(vdecl[nr_decls].identity.usageIndex == 0);
+         offset += 4;
+         nr_decls++;
+         break;
+      case TGSI_SEMANTIC_POSITION:
+         /* generated internally, not a vertex shader output */
+         break;
+      default:
+         assert(0);
+      }
+   }
+
+   draw_compute_vertex_size(vinfo);
+
+   svga_render->vdecl_count = nr_decls;
+   for (i = 0; i < svga_render->vdecl_count; i++)
+      vdecl[i].array.stride = offset;
+
+   if (memcmp(svga_render->vdecl, vdecl, sizeof(vdecl)) == 0)
+      return 0;
+
+   memcpy(svga_render->vdecl, vdecl, sizeof(vdecl));
+   svga->swtnl.new_vdecl = TRUE;
+
+   return 0;
+}
+
+
+static int update_swtnl_vdecl( struct svga_context *svga,
+                               unsigned dirty )
+{
+   return svga_swtnl_update_vdecl( svga );
+}
+
+
+struct svga_tracked_state svga_update_swtnl_vdecl =
+{
+   "update draw module vdecl",
+   (SVGA_NEW_VS |
+    SVGA_NEW_FS),
+   update_swtnl_vdecl
+};
diff --git a/src/gallium/drivers/svga/svga_tgsi.c b/src/gallium/drivers/svga/svga_tgsi.c
new file mode 100644
index 0000000000..81eea1a145
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi.c
@@ -0,0 +1,266 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_shader_tokens.h"
+#include "pipe/p_defines.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_dump.h"
+#include "tgsi/tgsi_scan.h"
+#include "util/u_memory.h"
+
+#include "svgadump/svga_shader_dump.h"
+
+#include "svga_context.h"
+#include "svga_tgsi.h"
+#include "svga_tgsi_emit.h"
+#include "svga_debug.h"
+
+#include "svga_hw_reg.h"
+#include "svga3d_shaderdefs.h"
+
+
+/* Sinkhole used only in error conditions.
+ */
+static char err_buf[128];
+
+#if 0
+static void svga_destroy_shader_emitter( struct svga_shader_emitter *emit )
+{
+   if (emit->buf != err_buf)
+      FREE(emit->buf);
+}
+#endif
+
+
+static boolean svga_shader_expand( struct svga_shader_emitter *emit )
+{
+   char *new_buf;
+   unsigned newsize = emit->size * 2;
+
+   if(emit->buf != err_buf)
+      new_buf = REALLOC(emit->buf, emit->size, newsize);
+   else
+      new_buf = NULL;
+
+   if (new_buf == NULL) {
+      emit->ptr = err_buf;
+      emit->buf = err_buf;
+      emit->size = sizeof(err_buf);
+      return FALSE;
+   }
+
+   emit->size = newsize;
+   emit->ptr = new_buf + (emit->ptr - emit->buf);
+   emit->buf = new_buf;
+   return TRUE;
+}   
+
+static INLINE boolean reserve(  struct svga_shader_emitter *emit,
+                                unsigned nr_dwords )
+{
+   if (emit->ptr - emit->buf + nr_dwords * sizeof(unsigned) >= emit->size) {
+      if (!svga_shader_expand( emit ))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+boolean svga_shader_emit_dword( struct svga_shader_emitter *emit,
+                                unsigned dword )
+{
+   if (!reserve(emit, 1))
+      return FALSE;
+
+   *(unsigned *)emit->ptr = dword;
+   emit->ptr += sizeof dword;
+   return TRUE;
+}
+
+boolean svga_shader_emit_dwords( struct svga_shader_emitter *emit,
+                                 const unsigned *dwords,
+                                 unsigned nr )
+{
+   if (!reserve(emit, nr))
+      return FALSE;
+
+   memcpy( emit->ptr, dwords, nr * sizeof *dwords );
+   emit->ptr += nr * sizeof *dwords;
+   return TRUE;
+}
+
+boolean svga_shader_emit_opcode( struct svga_shader_emitter *emit,
+                                 unsigned opcode )
+{
+   SVGA3dShaderInstToken *here;
+
+   if (!reserve(emit, 1))
+      return FALSE;
+
+   here = (SVGA3dShaderInstToken *)emit->ptr;
+   here->value = opcode;
+
+   if (emit->insn_offset) {
+      SVGA3dShaderInstToken *prev = (SVGA3dShaderInstToken *)(emit->buf + 
+                                                              emit->insn_offset);
+      prev->size = (here - prev) - 1;
+   }
+   
+   emit->insn_offset = emit->ptr - emit->buf;
+   emit->ptr += sizeof(unsigned);
+   return TRUE;
+}
+
+#define SVGA3D_PS_2X (SVGA3D_PS_20 | 1)
+#define SVGA3D_VS_2X (SVGA3D_VS_20 | 1)
+
+static boolean svga_shader_emit_header( struct svga_shader_emitter *emit )
+{
+   SVGA3dShaderVersion header;
+
+   memset( &header, 0, sizeof header );
+
+   switch (emit->unit) {
+   case PIPE_SHADER_FRAGMENT:
+      header.value = emit->use_sm30 ? SVGA3D_PS_30 : SVGA3D_PS_2X;
+      break;
+   case PIPE_SHADER_VERTEX:
+      header.value = emit->use_sm30 ? SVGA3D_VS_30 : SVGA3D_VS_2X;
+      break;
+   }
+ 
+   return svga_shader_emit_dword( emit, header.value );
+}
+
+
+
+
+
+/* Parse TGSI shader and translate to SVGA/DX9 serialized
+ * representation.  
+ *
+ * In this function SVGA shader is emitted to an in-memory buffer that
+ * can be dynamically grown.  Once we've finished and know how large
+ * it is, it will be copied to a hardware buffer for upload.
+ */
+static struct svga_shader_result *
+svga_tgsi_translate( const struct svga_shader *shader,
+                     union svga_compile_key key,
+                     unsigned unit )
+{
+   struct svga_shader_result *result = NULL;
+   struct svga_shader_emitter emit;
+   int ret = 0;
+
+   memset(&emit, 0, sizeof(emit));
+
+   emit.use_sm30 = shader->use_sm30;
+   emit.size = 1024;
+   emit.buf = MALLOC(emit.size);
+   if (emit.buf == NULL) {
+      ret = PIPE_ERROR_OUT_OF_MEMORY;
+      goto fail;
+   }
+
+   emit.ptr = emit.buf;
+   emit.unit = unit;
+   emit.key = key;
+
+   tgsi_scan_shader( shader->tokens, &emit.info);
+
+   emit.imm_start = emit.info.file_max[TGSI_FILE_CONSTANT] + 1;
+   
+   if (unit == PIPE_SHADER_FRAGMENT)
+      emit.imm_start += key.fkey.num_unnormalized_coords;
+
+   if (unit == PIPE_SHADER_VERTEX) {
+      emit.imm_start += key.vkey.need_prescale ? 2 : 0;
+      emit.imm_start += key.vkey.num_zero_stride_vertex_elements;
+   }
+
+   emit.nr_hw_const = (emit.imm_start + emit.info.file_max[TGSI_FILE_IMMEDIATE] + 1);
+
+   emit.nr_hw_temp = emit.info.file_max[TGSI_FILE_TEMPORARY] + 1;
+   emit.in_main_func = TRUE;
+
+   if (!svga_shader_emit_header( &emit ))
+      goto fail;
+
+   if (!svga_shader_emit_instructions( &emit, shader->tokens ))
+      goto fail;
+   
+   result = CALLOC_STRUCT(svga_shader_result);
+   if (result == NULL)
+      goto fail;
+
+   result->shader = shader;
+   result->tokens = (const unsigned *)emit.buf;
+   result->nr_tokens = (emit.ptr - emit.buf) / sizeof(unsigned);
+   memcpy(&result->key, &key, sizeof key);
+
+   return result;
+
+fail:
+   FREE(result);
+   FREE(emit.buf);
+   return NULL;
+}
+
+
+
+
+struct svga_shader_result *
+svga_translate_fragment_program( const struct svga_fragment_shader *fs,
+                                 const struct svga_fs_compile_key *fkey )
+{
+   union svga_compile_key key;
+   memcpy(&key.fkey, fkey, sizeof *fkey);
+
+   return svga_tgsi_translate( &fs->base, 
+                               key,
+                               PIPE_SHADER_FRAGMENT );
+}
+
+struct svga_shader_result *
+svga_translate_vertex_program( const struct svga_vertex_shader *vs,
+                               const struct svga_vs_compile_key *vkey )
+{
+   union svga_compile_key key;
+   memcpy(&key.vkey, vkey, sizeof *vkey);
+
+   return svga_tgsi_translate( &vs->base, 
+                               key,
+                               PIPE_SHADER_VERTEX );
+}
+
+
+void svga_destroy_shader_result( struct svga_shader_result *result )
+{
+   FREE((unsigned *)result->tokens);
+   FREE(result);
+}
+
diff --git a/src/gallium/drivers/svga/svga_tgsi.h b/src/gallium/drivers/svga/svga_tgsi.h
new file mode 100644
index 0000000000..896c90a89a
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi.h
@@ -0,0 +1,139 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_TGSI_H
+#define SVGA_TGSI_H
+
+#include "pipe/p_state.h"
+
+#include "svga_hw_reg.h"
+
+struct svga_fragment_shader;
+struct svga_vertex_shader;
+struct svga_shader;
+struct tgsi_shader_info;
+struct tgsi_token;
+
+
+struct svga_vs_compile_key
+{
+   ubyte need_prescale:1;
+   ubyte allow_psiz:1;
+   unsigned zero_stride_vertex_elements;
+   ubyte num_zero_stride_vertex_elements:6;
+};
+
+struct svga_fs_compile_key
+{
+   boolean light_twoside:1;
+   boolean front_cw:1;
+   ubyte num_textures;
+   ubyte num_unnormalized_coords;
+   struct {
+      ubyte compare_mode       : 1;
+      ubyte compare_func       : 3;
+      ubyte unnormalized       : 1;
+
+      ubyte width_height_idx   : 7;
+
+      ubyte texture_target;
+   } tex[PIPE_MAX_SAMPLERS];
+};
+
+union svga_compile_key {
+   struct svga_vs_compile_key vkey;
+   struct svga_fs_compile_key fkey;
+};
+
+struct svga_shader_result
+{
+   const struct svga_shader *shader;
+
+   /* Parameters used to generate this compilation result:
+    */
+   union svga_compile_key key;
+
+   /* Compiled shader tokens:
+    */
+   const unsigned *tokens;
+   unsigned nr_tokens;
+
+   /* SVGA Shader ID:
+    */
+   unsigned id;
+   
+   /* Next compilation result:
+    */
+   struct svga_shader_result *next;
+};
+
+
+/* TGSI doesn't provide use with VS input semantics (they're actually
+ * pretty meaningless), so we just generate some plausible ones here.
+ * This is called both from within the TGSI translator and when
+ * building vdecls to ensure they match up.
+ *
+ * The real use of this information is matching vertex elements to
+ * fragment shader inputs in the case where vertex shader is disabled.
+ */
+static INLINE void svga_generate_vdecl_semantics( unsigned idx,
+                                                  unsigned *usage,
+                                                  unsigned *usage_index )
+{
+   if (idx == 0) {
+      *usage = SVGA3D_DECLUSAGE_POSITION;
+      *usage_index = 0;
+   }
+   else {
+      *usage = SVGA3D_DECLUSAGE_TEXCOORD;
+      *usage_index = idx - 1;
+   }
+}
+
+
+
+static INLINE unsigned svga_vs_key_size( const struct svga_vs_compile_key *key )
+{
+   return sizeof *key;
+}
+
+static INLINE unsigned svga_fs_key_size( const struct svga_fs_compile_key *key )
+{
+   return (const char *)&key->tex[key->num_textures].texture_target -
+      (const char *)key;
+}
+
+struct svga_shader_result *
+svga_translate_fragment_program( const struct svga_fragment_shader *fs,
+                                 const struct svga_fs_compile_key *fkey );
+
+struct svga_shader_result *
+svga_translate_vertex_program( const struct svga_vertex_shader *fs,
+                               const struct svga_vs_compile_key *vkey );
+
+
+void svga_destroy_shader_result( struct svga_shader_result *result );
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c b/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c
new file mode 100644
index 0000000000..54457082a0
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c
@@ -0,0 +1,280 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "util/u_memory.h"
+
+#include "svga_tgsi_emit.h"
+#include "svga_context.h"
+
+
+
+
+static boolean ps20_input( struct svga_shader_emitter *emit,
+                           struct tgsi_declaration_semantic semantic,
+                           unsigned idx )
+{
+   struct src_register reg;
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   switch (semantic.SemanticName) {
+   case TGSI_SEMANTIC_POSITION:
+      /* Special case:
+       */
+      reg = src_register( SVGA3DREG_MISCTYPE, 
+                          SVGA3DMISCREG_POSITION );
+      break;
+   case TGSI_SEMANTIC_COLOR:
+      reg = src_register( SVGA3DREG_INPUT, 
+                          semantic.SemanticIndex );
+      break;
+   case TGSI_SEMANTIC_FOG:
+      assert(semantic.SemanticIndex == 0);
+      reg = src_register( SVGA3DREG_TEXTURE, 0 );
+      break;
+   case TGSI_SEMANTIC_GENERIC:
+      reg = src_register( SVGA3DREG_TEXTURE,
+                          semantic.SemanticIndex + 1 );
+      break;
+   default:
+      assert(0);
+      return TRUE;
+   }
+
+   emit->input_map[idx] = reg;
+
+   dcl.dst = dst( reg );
+
+   dcl.usage = 0;
+   dcl.index = 0;
+
+   dcl.values[0] |= 1<<31;
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+
+static boolean ps20_output( struct svga_shader_emitter *emit,
+                            struct tgsi_declaration_semantic semantic,
+                            unsigned idx )
+{
+   SVGA3dShaderDestToken reg;
+
+   switch (semantic.SemanticName) {
+   case TGSI_SEMANTIC_COLOR:
+      if (semantic.SemanticIndex < PIPE_MAX_COLOR_BUFS) {
+         unsigned cbuf = semantic.SemanticIndex;
+
+         emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                               emit->nr_hw_temp++ );
+         emit->temp_col[cbuf] = emit->output_map[idx];
+         emit->true_col[cbuf] = dst_register( SVGA3DREG_COLOROUT, 
+                                              semantic.SemanticIndex );
+      }
+      else {
+         assert(0);
+         reg = dst_register( SVGA3DREG_COLOROUT, 0 );
+      }
+      break;
+   case TGSI_SEMANTIC_POSITION:
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_pos = emit->output_map[idx];
+      emit->true_pos = dst_register( SVGA3DREG_DEPTHOUT, 
+                                     semantic.SemanticIndex );
+      break;
+   default:
+      assert(0);
+      reg = dst_register( SVGA3DREG_COLOROUT, 0 );
+      break;
+   }
+
+   return TRUE;
+}
+
+
+static boolean vs20_input( struct svga_shader_emitter *emit,
+                           struct tgsi_declaration_semantic semantic,
+                           unsigned idx )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   emit->input_map[idx] = src_register( SVGA3DREG_INPUT, idx );
+   dcl.dst = dst_register( SVGA3DREG_INPUT, idx );
+
+   assert(dcl.dst.reserved0);
+
+   /* Mesa doesn't provide use with VS input semantics (they're
+    * actually pretty meaningless), so we just generate some plausible
+    * ones here.  This has to match what we declare in the vdecl code
+    * in svga_pipe_vertex.c.
+    */
+   if (idx == 0) {
+      dcl.usage = SVGA3D_DECLUSAGE_POSITION;
+      dcl.index = 0;
+   }
+   else {
+      dcl.usage = SVGA3D_DECLUSAGE_TEXCOORD;
+      dcl.index = idx - 1;
+   }
+
+   dcl.values[0] |= 1<<31;
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+
+static boolean vs20_output( struct svga_shader_emitter *emit,
+                         struct tgsi_declaration_semantic semantic,
+                         unsigned idx )
+{
+   /* Don't emit dcl instruction for vs20 inputs
+    */
+
+   /* Just build the register map table: 
+    */
+   switch (semantic.SemanticName) {
+   case TGSI_SEMANTIC_POSITION:
+      assert(semantic.SemanticIndex == 0);
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_pos = emit->output_map[idx];
+      emit->true_pos = dst_register( SVGA3DREG_RASTOUT, 
+                                     SVGA3DRASTOUT_POSITION);
+      break;
+   case TGSI_SEMANTIC_PSIZE:
+      assert(semantic.SemanticIndex == 0);
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_psiz = emit->output_map[idx];
+      emit->true_psiz = dst_register( SVGA3DREG_RASTOUT, 
+                                      SVGA3DRASTOUT_PSIZE );
+      break;
+   case TGSI_SEMANTIC_FOG:
+      assert(semantic.SemanticIndex == 0);
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEXCRDOUT, 0 );
+      break;
+   case TGSI_SEMANTIC_COLOR:
+      /* oD0 */
+      emit->output_map[idx] = dst_register( SVGA3DREG_ATTROUT,
+                                            semantic.SemanticIndex );
+      break;
+   case TGSI_SEMANTIC_GENERIC:
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEXCRDOUT,
+                                            semantic.SemanticIndex + 1 );
+      break;
+   default:
+      assert(0);
+      emit->output_map[idx] = dst_register(  SVGA3DREG_TEMP, 0 );
+      return FALSE;
+   }
+
+   return TRUE;
+}
+
+static boolean ps20_sampler( struct svga_shader_emitter *emit,
+                          struct tgsi_declaration_semantic semantic,
+                          unsigned idx )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   dcl.dst = dst_register( SVGA3DREG_SAMPLER, idx );
+   dcl.type = svga_tgsi_sampler_type( emit, idx );
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+
+boolean svga_translate_decl_sm20( struct svga_shader_emitter *emit,
+                             const struct tgsi_full_declaration *decl )
+{
+   unsigned first = decl->DeclarationRange.First;
+   unsigned last = decl->DeclarationRange.Last;
+   unsigned semantic = 0;
+   unsigned semantic_idx = 0;
+   unsigned idx;
+   
+   if (decl->Declaration.Semantic) {
+      semantic = decl->Semantic.SemanticName;
+      semantic_idx = decl->Semantic.SemanticIndex;
+   }
+
+   for( idx = first; idx <= last; idx++ ) {
+      boolean ok;
+
+      switch (decl->Declaration.File) {
+      case TGSI_FILE_SAMPLER:
+         assert (emit->unit == PIPE_SHADER_FRAGMENT);
+         ok = ps20_sampler( emit, decl->Semantic, idx );
+         break;
+
+      case TGSI_FILE_INPUT:
+         if (emit->unit == PIPE_SHADER_VERTEX)
+            ok = vs20_input( emit, decl->Semantic, idx );
+         else
+            ok = ps20_input( emit, decl->Semantic, idx );
+         break;
+
+      case TGSI_FILE_OUTPUT:
+         if (emit->unit == PIPE_SHADER_VERTEX)
+            ok = vs20_output( emit, decl->Semantic, idx );
+         else
+            ok = ps20_output( emit, decl->Semantic, idx );
+         break;
+
+      default:
+         /* don't need to declare other vars */
+         ok = TRUE;
+      }
+
+      if (!ok)
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+
+
diff --git a/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c b/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c
new file mode 100644
index 0000000000..08e7dfb117
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c
@@ -0,0 +1,385 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "util/u_memory.h"
+
+#include "svga_tgsi_emit.h"
+#include "svga_context.h"
+
+static boolean translate_vs_ps_semantic( struct tgsi_declaration_semantic semantic,
+                                         unsigned *usage,
+                                         unsigned *idx )
+{
+   switch (semantic.SemanticName) {
+   case TGSI_SEMANTIC_POSITION:  
+      *idx = semantic.SemanticIndex;
+      *usage = SVGA3D_DECLUSAGE_POSITION;
+      break;
+   case TGSI_SEMANTIC_COLOR:     
+
+      *idx = semantic.SemanticIndex;
+      *usage = SVGA3D_DECLUSAGE_COLOR;
+      break;
+   case TGSI_SEMANTIC_BCOLOR:
+      *idx = semantic.SemanticIndex + 2; /* sharing with COLOR */
+      *usage = SVGA3D_DECLUSAGE_COLOR;
+      break;
+   case TGSI_SEMANTIC_FOG:       
+      *idx = 0;
+      assert(semantic.SemanticIndex == 0);
+      *usage = SVGA3D_DECLUSAGE_TEXCOORD;
+      break;
+   case TGSI_SEMANTIC_PSIZE:     
+      *idx = semantic.SemanticIndex;
+      *usage = SVGA3D_DECLUSAGE_PSIZE;
+      break;
+   case TGSI_SEMANTIC_GENERIC:   
+      *idx = semantic.SemanticIndex + 1; /* texcoord[0] is reserved for fog */
+      *usage = SVGA3D_DECLUSAGE_TEXCOORD;
+      break;
+   case TGSI_SEMANTIC_NORMAL:    
+      *idx = semantic.SemanticIndex;
+      *usage = SVGA3D_DECLUSAGE_NORMAL;
+      break;
+   default:
+      assert(0);
+      *usage = SVGA3D_DECLUSAGE_TEXCOORD;
+      *idx = 0;
+      return FALSE;
+   }
+
+   return TRUE;
+}
+
+
+static boolean emit_decl( struct svga_shader_emitter *emit,
+                          SVGA3dShaderDestToken reg,
+                          unsigned usage, 
+                          unsigned index )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   dcl.dst = reg;
+   dcl.usage = usage;
+   dcl.index = index;
+   dcl.values[0] |= 1<<31;
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+static boolean emit_vface_decl( struct svga_shader_emitter *emit )
+{
+   if (!emit->emitted_vface) {
+      SVGA3dShaderDestToken reg =
+         dst_register( SVGA3DREG_MISCTYPE,
+                       SVGA3DMISCREG_FACE );
+
+      if (!emit_decl( emit, reg, 0, 0 ))
+         return FALSE;
+
+      emit->emitted_vface = TRUE;
+   }
+   return TRUE;
+}
+
+static boolean ps30_input( struct svga_shader_emitter *emit,
+                           struct tgsi_declaration_semantic semantic,
+                           unsigned idx )
+{
+   unsigned usage, index;
+   SVGA3dShaderDestToken reg;
+
+   if (semantic.SemanticName == TGSI_SEMANTIC_POSITION) {
+      emit->input_map[idx] = src_register( SVGA3DREG_MISCTYPE,
+                                           SVGA3DMISCREG_POSITION );
+
+      emit->input_map[idx].base.swizzle = TRANSLATE_SWIZZLE( TGSI_SWIZZLE_X,
+                                                             TGSI_SWIZZLE_Y,
+                                                             TGSI_SWIZZLE_Y,
+                                                             TGSI_SWIZZLE_Y );
+
+      reg = writemask( dst(emit->input_map[idx]),
+                       TGSI_WRITEMASK_XY );
+
+      return emit_decl( emit, reg, 0, 0 );
+   }
+   else if (emit->key.fkey.light_twoside &&
+            (semantic.SemanticName == TGSI_SEMANTIC_COLOR)) {
+
+      if (!translate_vs_ps_semantic( semantic, &usage, &index ))
+         return FALSE;
+
+      emit->internal_color_idx[emit->internal_color_count] = idx;
+      emit->input_map[idx] = src_register( SVGA3DREG_INPUT, emit->ps30_input_count );
+      emit->ps30_input_count++;
+      emit->internal_color_count++;
+
+      reg = dst( emit->input_map[idx] );
+
+      if (!emit_decl( emit, reg, usage, index ))
+         return FALSE;
+
+      semantic.SemanticName = TGSI_SEMANTIC_BCOLOR;
+      if (!translate_vs_ps_semantic( semantic, &usage, &index ))
+         return FALSE;
+
+      reg = dst_register( SVGA3DREG_INPUT, emit->ps30_input_count++ );
+
+      if (!emit_decl( emit, reg, usage, index ))
+         return FALSE;
+
+      if (!emit_vface_decl( emit ))
+         return FALSE;
+
+      return TRUE;
+   }
+   else if (semantic.SemanticName == TGSI_SEMANTIC_FACE) {
+      if (!emit_vface_decl( emit ))
+         return FALSE;
+      emit->emit_frontface = TRUE;
+      emit->internal_frontface_idx = idx;
+      return TRUE;
+   }
+   else {
+
+      if (!translate_vs_ps_semantic( semantic, &usage, &index ))
+         return FALSE;
+
+      emit->input_map[idx] = src_register( SVGA3DREG_INPUT, emit->ps30_input_count++ );
+      reg = dst( emit->input_map[idx] );
+
+      return emit_decl( emit, reg, usage, index );
+   }
+
+}
+
+
+/* PS output registers are the same as 2.0
+ */
+static boolean ps30_output( struct svga_shader_emitter *emit,
+                            struct tgsi_declaration_semantic semantic,
+                            unsigned idx )
+{
+   SVGA3dShaderDestToken reg;
+
+   switch (semantic.SemanticName) {
+   case TGSI_SEMANTIC_COLOR:
+      emit->output_map[idx] = dst_register( SVGA3DREG_COLOROUT, 
+                                            semantic.SemanticIndex );
+      break;
+   case TGSI_SEMANTIC_POSITION:
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_pos = emit->output_map[idx];
+      emit->true_pos = dst_register( SVGA3DREG_DEPTHOUT, 
+                                     semantic.SemanticIndex );
+      break;
+   default:
+      assert(0);
+      reg = dst_register( SVGA3DREG_COLOROUT, 0 );
+      break;
+   }
+
+   return TRUE;
+}
+
+
+/* We still make up the input semantics the same as in 2.0
+ */
+static boolean vs30_input( struct svga_shader_emitter *emit,
+                           struct tgsi_declaration_semantic semantic,
+                           unsigned idx )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+   unsigned usage, index;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   if (emit->key.vkey.zero_stride_vertex_elements & (1 << idx)) {
+      unsigned i;
+      unsigned offset = 0;
+      unsigned start_idx = emit->info.file_max[TGSI_FILE_CONSTANT] + 1;
+      /* adjust for prescale constants */
+      start_idx += emit->key.vkey.need_prescale ? 2 : 0;
+      /* compute the offset from the start of zero stride constants */
+      for (i = 0; i < PIPE_MAX_ATTRIBS && i < idx; ++i) {
+         if (emit->key.vkey.zero_stride_vertex_elements & (1<<i))
+            ++offset;
+      }
+      emit->input_map[idx] = src_register( SVGA3DREG_CONST,
+                                           start_idx + offset );
+   } else {
+      emit->input_map[idx] = src_register( SVGA3DREG_INPUT, idx );
+      dcl.dst = dst_register( SVGA3DREG_INPUT, idx );
+
+      assert(dcl.dst.reserved0);
+
+      svga_generate_vdecl_semantics( idx, &usage, &index );
+
+      dcl.usage = usage;
+      dcl.index = index;
+      dcl.values[0] |= 1<<31;
+
+      return  (emit_instruction(emit, opcode) &&
+               svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+   }
+   return TRUE;
+}
+
+/* VS3.0 outputs have proper declarations and semantic info for
+ * matching against PS inputs.
+ */
+static boolean vs30_output( struct svga_shader_emitter *emit,
+                         struct tgsi_declaration_semantic semantic,
+                         unsigned idx )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+   unsigned usage, index;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   if (!translate_vs_ps_semantic( semantic, &usage, &index ))
+      return FALSE;
+
+   dcl.dst = dst_register( SVGA3DREG_OUTPUT, idx );
+   dcl.usage = usage;
+   dcl.index = index;
+   dcl.values[0] |= 1<<31;
+
+   if (semantic.SemanticName == TGSI_SEMANTIC_POSITION) {
+      assert(idx == 0);
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_pos = emit->output_map[idx];
+      emit->true_pos = dcl.dst;
+   }
+   else if (semantic.SemanticName == TGSI_SEMANTIC_PSIZE) {
+      emit->output_map[idx] = dst_register( SVGA3DREG_TEMP,
+                                            emit->nr_hw_temp++ );
+      emit->temp_psiz = emit->output_map[idx];
+
+      /* This has the effect of not declaring psiz (below) and not 
+       * emitting the final MOV to true_psiz in the postamble.
+       */
+      if (!emit->key.vkey.allow_psiz)
+         return TRUE;
+
+      emit->true_psiz = dcl.dst;
+   }
+   else {
+      emit->output_map[idx] = dcl.dst;
+   }
+
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+static boolean ps30_sampler( struct svga_shader_emitter *emit,
+                          struct tgsi_declaration_semantic semantic,
+                          unsigned idx )
+{
+   SVGA3DOpDclArgs dcl;
+   SVGA3dShaderInstToken opcode;
+
+   opcode = inst_token( SVGA3DOP_DCL );
+   dcl.values[0] = 0;
+   dcl.values[1] = 0;
+
+   dcl.dst = dst_register( SVGA3DREG_SAMPLER, idx );
+   dcl.type = svga_tgsi_sampler_type( emit, idx );
+   dcl.values[0] |= 1<<31;
+
+   return  (emit_instruction(emit, opcode) &&
+            svga_shader_emit_dwords( emit, dcl.values, Elements(dcl.values)));
+}
+
+
+boolean svga_translate_decl_sm30( struct svga_shader_emitter *emit,
+                             const struct tgsi_full_declaration *decl )
+{
+   unsigned first = decl->DeclarationRange.First;
+   unsigned last = decl->DeclarationRange.Last;
+   unsigned semantic = 0;
+   unsigned semantic_idx = 0;
+   unsigned idx;
+
+   if (decl->Declaration.Semantic) {
+      semantic = decl->Semantic.SemanticName;
+      semantic_idx = decl->Semantic.SemanticIndex;
+   }
+
+   for( idx = first; idx <= last; idx++ ) {
+      boolean ok;
+
+      switch (decl->Declaration.File) {
+      case TGSI_FILE_SAMPLER:
+         assert (emit->unit == PIPE_SHADER_FRAGMENT);
+         ok = ps30_sampler( emit, decl->Semantic, idx );
+         break;
+
+      case TGSI_FILE_INPUT:
+         if (emit->unit == PIPE_SHADER_VERTEX)
+            ok = vs30_input( emit, decl->Semantic, idx );
+         else
+            ok = ps30_input( emit, decl->Semantic, idx );
+         break;
+
+      case TGSI_FILE_OUTPUT:
+         if (emit->unit == PIPE_SHADER_VERTEX)
+            ok = vs30_output( emit, decl->Semantic, idx );
+         else
+            ok = ps30_output( emit, decl->Semantic, idx );
+         break;
+
+      default:
+         /* don't need to declare other vars */
+         ok = TRUE;
+      }
+
+      if (!ok)
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+
+
diff --git a/src/gallium/drivers/svga/svga_tgsi_emit.h b/src/gallium/drivers/svga/svga_tgsi_emit.h
new file mode 100644
index 0000000000..2557824293
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi_emit.h
@@ -0,0 +1,345 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_TGSI_EMIT_H
+#define SVGA_TGSI_EMIT_H
+
+#include "tgsi/tgsi_scan.h"
+#include "svga_hw_reg.h"
+#include "svga_tgsi.h"
+#include "svga3d_shaderdefs.h"
+
+struct src_register
+{
+   SVGA3dShaderSrcToken base;
+   SVGA3dShaderSrcToken indirect;
+};
+
+
+struct svga_arl_consts {
+   int number;
+   int idx;
+   int swizzle;
+   int arl_num;
+};
+
+/* Internal functions:
+ */
+
+struct svga_shader_emitter
+{
+   boolean use_sm30;
+   
+   unsigned size;
+   char *buf;
+   char *ptr;
+
+   union svga_compile_key key;
+   struct tgsi_shader_info info;
+   int unit;
+
+   int imm_start;
+
+   int nr_hw_const;
+   int nr_hw_temp;
+   
+   int insn_offset;
+
+   int internal_temp_count;
+   int internal_imm_count;
+
+   int internal_color_idx[2]; /* diffuse, specular */
+   int internal_color_count;
+
+   boolean emitted_vface;
+   boolean emit_frontface;
+   int internal_frontface_idx;
+
+   int ps30_input_count;
+
+   boolean in_main_func;
+
+   boolean created_zero_immediate;
+   int zero_immediate_idx;
+
+   boolean created_loop_const;
+   int loop_const_idx;
+
+   boolean created_sincos_consts;
+   int sincos_consts_idx;
+
+   unsigned label[32];
+   unsigned nr_labels;
+
+   struct src_register input_map[PIPE_MAX_ATTRIBS];
+   SVGA3dShaderDestToken output_map[PIPE_MAX_ATTRIBS];
+
+   struct src_register imm_0055;
+   SVGA3dShaderDestToken temp_pos;
+   SVGA3dShaderDestToken true_pos;
+
+   SVGA3dShaderDestToken temp_col[PIPE_MAX_COLOR_BUFS];
+   SVGA3dShaderDestToken true_col[PIPE_MAX_COLOR_BUFS];
+
+   SVGA3dShaderDestToken temp_psiz;
+   SVGA3dShaderDestToken true_psiz;
+
+   struct svga_arl_consts arl_consts[12];
+   int num_arl_consts;
+   int current_arl;
+};
+
+
+boolean svga_shader_emit_dword( struct svga_shader_emitter *emit,
+                                unsigned dword );
+
+boolean svga_shader_emit_dwords( struct svga_shader_emitter *emit,
+                                 const unsigned *dwords,
+                                 unsigned nr );
+
+boolean svga_shader_emit_opcode( struct svga_shader_emitter *emit,
+                                 unsigned opcode );
+
+boolean svga_shader_emit_instructions( struct svga_shader_emitter *emit,
+                                       const struct tgsi_token *tokens );
+
+boolean svga_translate_decl_sm20( struct svga_shader_emitter *emit,
+                               const struct tgsi_full_declaration *decl );
+
+boolean svga_translate_decl_sm30( struct svga_shader_emitter *emit,
+                               const struct tgsi_full_declaration *decl );
+
+
+static INLINE boolean emit_dst( struct svga_shader_emitter *emit,
+                         SVGA3dShaderDestToken dest )
+{
+   assert(dest.reserved0);
+   return svga_shader_emit_dword( emit, dest.value );
+}
+
+static INLINE boolean emit_src( struct svga_shader_emitter *emit,
+                         const struct src_register src )
+{
+   if (src.base.relAddr) {
+      assert(src.base.reserved0);
+      assert(src.indirect.reserved0);
+      return (svga_shader_emit_dword( emit, src.base.value ) &&
+              svga_shader_emit_dword( emit, src.indirect.value ));
+   }
+   else {
+      assert(src.base.reserved0);
+      return svga_shader_emit_dword( emit, src.base.value );
+   }
+}
+
+
+static INLINE boolean emit_instruction( struct svga_shader_emitter *emit,
+                                 SVGA3dShaderInstToken opcode )
+{
+   return svga_shader_emit_opcode( emit, opcode.value );
+}
+
+
+static INLINE boolean emit_op1( struct svga_shader_emitter *emit,
+                         SVGA3dShaderInstToken inst,
+                         SVGA3dShaderDestToken dest,
+                         struct src_register src0 )
+{
+   return (emit_instruction( emit, inst ) &&
+           emit_dst( emit, dest ) &&
+           emit_src( emit, src0 ));
+}
+
+static INLINE boolean emit_op2( struct svga_shader_emitter *emit,
+                     SVGA3dShaderInstToken inst,
+                     SVGA3dShaderDestToken dest,
+                     struct src_register src0,
+                     struct src_register src1 )
+{
+   return (emit_instruction( emit, inst ) &&
+           emit_dst( emit, dest ) &&
+           emit_src( emit, src0 ) &&
+           emit_src( emit, src1 ));
+}
+
+static INLINE boolean emit_op3( struct svga_shader_emitter *emit,
+                         SVGA3dShaderInstToken inst,
+                         SVGA3dShaderDestToken dest,
+                         struct src_register src0,
+                         struct src_register src1,
+                         struct src_register src2 )
+{
+   return (emit_instruction( emit, inst ) &&
+           emit_dst( emit, dest ) &&
+           emit_src( emit, src0 ) &&
+           emit_src( emit, src1 ) &&
+           emit_src( emit, src2 ));
+}
+
+
+#define TRANSLATE_SWIZZLE(x,y,z,w)  ((x) | ((y) << 2) | ((z) << 4) | ((w) << 6))
+#define SWIZZLE_XYZW  \
+ TRANSLATE_SWIZZLE(TGSI_SWIZZLE_X,TGSI_SWIZZLE_Y,TGSI_SWIZZLE_Z,TGSI_SWIZZLE_W)
+#define SWIZZLE_XXXX  \
+ TRANSLATE_SWIZZLE(TGSI_SWIZZLE_X,TGSI_SWIZZLE_X,TGSI_SWIZZLE_X,TGSI_SWIZZLE_X)
+#define SWIZZLE_YYYY  \
+ TRANSLATE_SWIZZLE(TGSI_SWIZZLE_Y,TGSI_SWIZZLE_Y,TGSI_SWIZZLE_Y,TGSI_SWIZZLE_Y)
+#define SWIZZLE_ZZZZ  \
+ TRANSLATE_SWIZZLE(TGSI_SWIZZLE_Z,TGSI_SWIZZLE_Z,TGSI_SWIZZLE_Z,TGSI_SWIZZLE_Z)
+#define SWIZZLE_WWWW  \
+ TRANSLATE_SWIZZLE(TGSI_SWIZZLE_W,TGSI_SWIZZLE_W,TGSI_SWIZZLE_W,TGSI_SWIZZLE_W)
+
+
+
+static INLINE SVGA3dShaderInstToken
+inst_token( unsigned opcode )
+{
+   SVGA3dShaderInstToken inst;
+
+   inst.value = 0;
+   inst.op = opcode;
+
+   return inst;
+}
+
+static INLINE SVGA3dShaderDestToken 
+dst_register( unsigned file,
+              int number )
+{
+   SVGA3dShaderDestToken dest;
+
+   dest.value = 0;
+   dest.num = number;
+   dest.type_upper = file >> 3;
+   dest.relAddr = 0;
+   dest.reserved1 = 0;
+   dest.mask = 0xf;
+   dest.dstMod = 0;
+   dest.shfScale = 0;
+   dest.type_lower = file & 0x7;
+   dest.reserved0 = 1;          /* is_reg */
+   
+   return dest;
+}
+
+static INLINE SVGA3dShaderDestToken
+writemask( SVGA3dShaderDestToken dest,
+           unsigned mask )
+{
+   dest.mask &= mask;
+   return dest;
+}
+
+
+static INLINE SVGA3dShaderSrcToken 
+src_token( unsigned file, int number )
+{
+   SVGA3dShaderSrcToken src;
+
+   src.value = 0;
+   src.num = number;
+   src.type_upper = file >> 3;
+   src.relAddr = 0;
+   src.reserved1 = 0;
+   src.swizzle = SWIZZLE_XYZW;
+   src.srcMod = 0;
+   src.type_lower = file & 0x7;
+   src.reserved0 = 1;           /* is_reg */
+
+   return src;
+}
+
+
+static INLINE struct src_register 
+absolute( struct src_register src )
+{
+   src.base.srcMod = SVGA3DSRCMOD_ABS;
+
+   return src;
+}
+
+
+static INLINE struct src_register 
+negate( struct src_register src )
+{
+   switch (src.base.srcMod) {
+   case SVGA3DSRCMOD_ABS:
+      src.base.srcMod = SVGA3DSRCMOD_ABSNEG;
+      break;
+   case SVGA3DSRCMOD_ABSNEG:
+      src.base.srcMod = SVGA3DSRCMOD_ABS;
+      break;
+   case SVGA3DSRCMOD_NEG:
+      src.base.srcMod = SVGA3DSRCMOD_NONE;
+      break;
+   case SVGA3DSRCMOD_NONE:
+      src.base.srcMod = SVGA3DSRCMOD_NEG;
+      break;
+   }
+   return src;
+}
+
+
+static INLINE struct src_register 
+src_register( unsigned file, int number )
+{
+   struct src_register src;
+   
+   src.base = src_token( file, number );
+   src.indirect.value = 0;
+
+   return src;
+}
+
+static INLINE SVGA3dShaderDestToken dst( struct src_register src )
+{
+   return dst_register( SVGA3dShaderGetRegType( src.base.value ),
+                        src.base.num );
+}
+
+static INLINE struct src_register src( SVGA3dShaderDestToken dst )
+{
+   return src_register( SVGA3dShaderGetRegType( dst.value ),
+                        dst.num );
+}
+
+static INLINE ubyte svga_tgsi_sampler_type( struct svga_shader_emitter *emit,
+                                            int idx )
+{
+   switch (emit->key.fkey.tex[idx].texture_target) {
+   case PIPE_TEXTURE_1D:
+      return SVGA3DSAMP_2D;
+   case PIPE_TEXTURE_2D:
+      return SVGA3DSAMP_2D;
+   case PIPE_TEXTURE_3D:
+      return SVGA3DSAMP_VOLUME;
+   case PIPE_TEXTURE_CUBE:
+      return SVGA3DSAMP_CUBE;
+   }
+
+   return SVGA3DSAMP_UNKNOWN;
+}
+
+#endif
diff --git a/src/gallium/drivers/svga/svga_tgsi_insn.c b/src/gallium/drivers/svga/svga_tgsi_insn.c
new file mode 100644
index 0000000000..ea409b7e16
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_tgsi_insn.c
@@ -0,0 +1,2716 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "util/u_memory.h"
+
+#include "svga_tgsi_emit.h"
+#include "svga_context.h"
+
+
+static boolean emit_vs_postamble( struct svga_shader_emitter *emit );
+static boolean emit_ps_postamble( struct svga_shader_emitter *emit );
+
+
+
+ 
+static unsigned
+translate_opcode(
+   uint opcode )
+{
+   switch (opcode) {
+   case TGSI_OPCODE_ABS:        return SVGA3DOP_ABS;
+   case TGSI_OPCODE_ADD:        return SVGA3DOP_ADD;
+   case TGSI_OPCODE_BREAKC:     return SVGA3DOP_BREAKC;
+   case TGSI_OPCODE_DDX:        return SVGA3DOP_DSX;
+   case TGSI_OPCODE_DDY:        return SVGA3DOP_DSY;
+   case TGSI_OPCODE_DP2A:       return SVGA3DOP_DP2ADD;
+   case TGSI_OPCODE_DP3:        return SVGA3DOP_DP3;
+   case TGSI_OPCODE_DP4:        return SVGA3DOP_DP4;
+   case TGSI_OPCODE_ENDFOR:     return SVGA3DOP_ENDLOOP;
+   case TGSI_OPCODE_FRC:        return SVGA3DOP_FRC;
+   case TGSI_OPCODE_BGNFOR:     return SVGA3DOP_LOOP;
+   case TGSI_OPCODE_MAD:        return SVGA3DOP_MAD;
+   case TGSI_OPCODE_MAX:        return SVGA3DOP_MAX;
+   case TGSI_OPCODE_MIN:        return SVGA3DOP_MIN;
+   case TGSI_OPCODE_MOV:        return SVGA3DOP_MOV;
+   case TGSI_OPCODE_MUL:        return SVGA3DOP_MUL;
+   case TGSI_OPCODE_NOP:        return SVGA3DOP_NOP;
+   case TGSI_OPCODE_NRM4:       return SVGA3DOP_NRM;
+   case TGSI_OPCODE_SSG:        return SVGA3DOP_SGN;
+   default:
+      debug_printf("Unkown opcode %u\n", opcode);
+      assert( 0 );
+      return SVGA3DOP_LAST_INST;
+   }
+}
+
+
+static unsigned translate_file( unsigned file )
+{
+   switch (file) {
+   case TGSI_FILE_TEMPORARY: return SVGA3DREG_TEMP;
+   case TGSI_FILE_INPUT:     return SVGA3DREG_INPUT;
+   case TGSI_FILE_OUTPUT:    return SVGA3DREG_OUTPUT; /* VS3.0+ only */
+   case TGSI_FILE_IMMEDIATE: return SVGA3DREG_CONST;
+   case TGSI_FILE_CONSTANT:  return SVGA3DREG_CONST;
+   case TGSI_FILE_SAMPLER:   return SVGA3DREG_SAMPLER;
+   case TGSI_FILE_ADDRESS:   return SVGA3DREG_ADDR;
+   default:
+      assert( 0 );
+      return SVGA3DREG_TEMP;
+   }
+}
+
+
+
+
+
+
+static SVGA3dShaderDestToken 
+translate_dst_register( struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn,
+                        unsigned idx )
+{
+   const struct tgsi_full_dst_register *reg = &insn->FullDstRegisters[idx];
+   SVGA3dShaderDestToken dest;
+
+   switch (reg->DstRegister.File) {
+   case TGSI_FILE_OUTPUT:
+      /* Output registers encode semantic information in their name.
+       * Need to lookup a table built at decl time:
+       */
+      dest = emit->output_map[reg->DstRegister.Index];
+      break;
+
+   default:
+      dest = dst_register( translate_file( reg->DstRegister.File ),
+                           reg->DstRegister.Index );
+      break;
+   }
+
+   dest.mask = reg->DstRegister.WriteMask;
+
+   if (insn->Instruction.Saturate) 
+      dest.dstMod = SVGA3DDSTMOD_SATURATE;
+
+   return dest;
+}
+
+
+static struct src_register 
+swizzle( struct src_register src,
+         int x,
+         int y,
+         int z,
+         int w )
+{
+   x = (src.base.swizzle >> (x * 2)) & 0x3;
+   y = (src.base.swizzle >> (y * 2)) & 0x3;
+   z = (src.base.swizzle >> (z * 2)) & 0x3;
+   w = (src.base.swizzle >> (w * 2)) & 0x3;
+
+   src.base.swizzle = TRANSLATE_SWIZZLE(x,y,z,w);
+
+   return src;
+}
+
+static struct src_register
+scalar( struct src_register src,
+        int comp )
+{
+   return swizzle( src, comp, comp, comp, comp );
+}
+
+static INLINE boolean
+svga_arl_needs_adjustment( const struct svga_shader_emitter *emit )
+{
+   int i;
+
+   for (i = 0; i < emit->num_arl_consts; ++i) {
+      if (emit->arl_consts[i].arl_num == emit->current_arl)
+         return TRUE;
+   }
+   return FALSE;
+}
+
+static INLINE int
+svga_arl_adjustment( const struct svga_shader_emitter *emit )
+{
+   int i;
+
+   for (i = 0; i < emit->num_arl_consts; ++i) {
+      if (emit->arl_consts[i].arl_num == emit->current_arl)
+         return emit->arl_consts[i].number;
+   }
+   return 0;
+}
+
+static struct src_register 
+translate_src_register( const struct svga_shader_emitter *emit,
+                        const struct tgsi_full_src_register *reg )
+{
+   struct src_register src;
+
+   switch (reg->SrcRegister.File) {
+   case TGSI_FILE_INPUT:
+      /* Input registers are referred to by their semantic name rather
+       * than by index.  Use the mapping build up from the decls:
+       */
+      src = emit->input_map[reg->SrcRegister.Index];
+      break;
+       
+   case TGSI_FILE_IMMEDIATE:
+      /* Immediates are appended after TGSI constants in the D3D
+       * constant buffer.
+       */
+      src = src_register( translate_file( reg->SrcRegister.File ),
+                          reg->SrcRegister.Index + 
+                          emit->imm_start );
+      break;
+
+   default:
+      src = src_register( translate_file( reg->SrcRegister.File ),
+                          reg->SrcRegister.Index );
+
+      break;
+   }
+
+   /* Indirect addressing (for coninstant buffer lookups only)
+    */
+   if (reg->SrcRegister.Indirect)
+   {
+      /* we shift the offset towards the minimum */
+      if (svga_arl_needs_adjustment( emit )) {
+         src.base.num -= svga_arl_adjustment( emit );
+      }
+      src.base.relAddr = 1;
+
+      /* Not really sure what should go in the second token:
+       */
+      src.indirect = src_token( SVGA3DREG_ADDR,
+                                reg->SrcRegisterInd.Index );
+
+      src.indirect.swizzle = SWIZZLE_XXXX;
+   }
+
+   src = swizzle( src,
+                  reg->SrcRegister.SwizzleX,
+                  reg->SrcRegister.SwizzleY,
+                  reg->SrcRegister.SwizzleZ,
+                  reg->SrcRegister.SwizzleW );
+
+   /* src.mod isn't a bitfield, unfortunately:
+    * See tgsi_util_get_full_src_register_sign_mode for implementation details.
+    */
+   if (reg->SrcRegisterExtMod.Absolute) {
+      if (reg->SrcRegisterExtMod.Negate)
+         src.base.srcMod = SVGA3DSRCMOD_ABSNEG;
+      else
+         src.base.srcMod = SVGA3DSRCMOD_ABS;
+   }
+   else {
+      if (reg->SrcRegister.Negate != reg->SrcRegisterExtMod.Negate)
+         src.base.srcMod = SVGA3DSRCMOD_NEG;
+      else
+         src.base.srcMod = SVGA3DSRCMOD_NONE;
+   }
+
+   return src;
+}
+
+
+/*
+ * Get a temporary register, return -1 if none available
+ */
+static INLINE SVGA3dShaderDestToken 
+get_temp( struct svga_shader_emitter *emit )
+{
+   int i = emit->nr_hw_temp + emit->internal_temp_count++;
+
+   return dst_register( SVGA3DREG_TEMP, i );
+}
+
+/* Release a single temp.  Currently only effective if it was the last
+ * allocated temp, otherwise release will be delayed until the next
+ * call to reset_temp_regs().
+ */
+static INLINE void 
+release_temp( struct svga_shader_emitter *emit,
+              SVGA3dShaderDestToken temp )
+{
+   if (temp.num == emit->internal_temp_count - 1)
+      emit->internal_temp_count--;
+}
+
+static void reset_temp_regs( struct svga_shader_emitter *emit )
+{
+   emit->internal_temp_count = 0;
+}
+   
+
+static boolean submit_op0( struct svga_shader_emitter *emit,
+                           SVGA3dShaderInstToken inst,
+                           SVGA3dShaderDestToken dest )
+{
+   return (emit_instruction( emit, inst ) && 
+           emit_dst( emit, dest ));
+}
+
+static boolean submit_op1( struct svga_shader_emitter *emit,
+                           SVGA3dShaderInstToken inst,
+                           SVGA3dShaderDestToken dest,
+                           struct src_register src0 )
+{
+   return emit_op1( emit, inst, dest, src0 );
+}
+
+
+/* SVGA shaders may not refer to >1 constant register in a single
+ * instruction.  This function checks for that usage and inserts a
+ * move to temporary if detected.
+ *
+ * The same applies to input registers -- at most a single input
+ * register may be read by any instruction.
+ */
+static boolean submit_op2( struct svga_shader_emitter *emit,
+                           SVGA3dShaderInstToken inst,
+                           SVGA3dShaderDestToken dest,
+                           struct src_register src0,
+                           struct src_register src1 )
+{
+   SVGA3dShaderDestToken temp;
+   SVGA3dShaderRegType type0, type1;
+   boolean need_temp = FALSE;
+
+   temp.value = 0;
+   type0 = SVGA3dShaderGetRegType( src0.base.value );
+   type1 = SVGA3dShaderGetRegType( src1.base.value );
+
+   if (type0 == SVGA3DREG_CONST &&
+       type1 == SVGA3DREG_CONST &&
+       src0.base.num != src1.base.num)
+      need_temp = TRUE;
+
+   if (type0 == SVGA3DREG_INPUT &&
+       type1 == SVGA3DREG_INPUT &&
+       src0.base.num != src1.base.num)
+      need_temp = TRUE;
+
+   if (need_temp)
+   {
+      temp = get_temp( emit );
+
+      if (!emit_op1( emit, inst_token( SVGA3DOP_MOV ), temp, src0 ))
+         return FALSE;
+
+      src0 = src( temp );
+   }
+
+   if (!emit_op2( emit, inst, dest, src0, src1 ))
+      return FALSE;
+
+   if (need_temp)
+      release_temp( emit, temp );
+
+   return TRUE;
+}
+
+
+/* SVGA shaders may not refer to >1 constant register in a single
+ * instruction.  This function checks for that usage and inserts a
+ * move to temporary if detected.
+ */
+static boolean submit_op3( struct svga_shader_emitter *emit,
+                           SVGA3dShaderInstToken inst,
+                           SVGA3dShaderDestToken dest,
+                           struct src_register src0,
+                           struct src_register src1,
+                           struct src_register src2 )
+{
+   SVGA3dShaderDestToken temp0;
+   SVGA3dShaderDestToken temp1;
+   boolean need_temp0 = FALSE;
+   boolean need_temp1 = FALSE;
+   SVGA3dShaderRegType type0, type1, type2;
+
+   temp0.value = 0;
+   temp1.value = 0;
+   type0 = SVGA3dShaderGetRegType( src0.base.value );
+   type1 = SVGA3dShaderGetRegType( src1.base.value );
+   type2 = SVGA3dShaderGetRegType( src2.base.value );
+
+   if (inst.op != SVGA3DOP_SINCOS) {
+      if (type0 == SVGA3DREG_CONST &&
+          ((type1 == SVGA3DREG_CONST && src0.base.num != src1.base.num) ||
+           (type2 == SVGA3DREG_CONST && src0.base.num != src2.base.num)))
+         need_temp0 = TRUE;
+
+      if (type1 == SVGA3DREG_CONST &&
+          (type2 == SVGA3DREG_CONST && src1.base.num != src2.base.num))
+         need_temp1 = TRUE;
+   }
+
+   if (type0 == SVGA3DREG_INPUT &&
+       ((type1 == SVGA3DREG_INPUT && src0.base.num != src1.base.num) ||
+        (type2 == SVGA3DREG_INPUT && src0.base.num != src2.base.num)))
+      need_temp0 = TRUE;
+
+   if (type1 == SVGA3DREG_INPUT &&
+       (type2 == SVGA3DREG_INPUT && src1.base.num != src2.base.num))
+      need_temp1 = TRUE;
+
+   if (need_temp0)
+   {
+      temp0 = get_temp( emit );
+ 
+      if (!emit_op1( emit, inst_token( SVGA3DOP_MOV ), temp0, src0 ))
+         return FALSE;
+         
+      src0 = src( temp0 );
+   }
+
+   if (need_temp1)
+   {
+      temp1 = get_temp( emit );
+
+      if (!emit_op1( emit, inst_token( SVGA3DOP_MOV ), temp1, src1 ))
+         return FALSE;
+
+      src1 = src( temp1 );
+   }
+
+   if (!emit_op3( emit, inst, dest, src0, src1, src2 ))
+      return FALSE;
+
+   if (need_temp1)
+      release_temp( emit, temp1 );
+   if (need_temp0)
+      release_temp( emit, temp0 );
+   return TRUE;
+}
+
+
+static boolean emit_def_const( struct svga_shader_emitter *emit,
+                               SVGA3dShaderConstType type,
+                               unsigned idx,
+                               float a,
+                               float b,
+                               float c,
+                               float d )
+{
+   SVGA3DOpDefArgs def;
+   SVGA3dShaderInstToken opcode;
+
+   switch (type) {
+   case SVGA3D_CONST_TYPE_FLOAT:
+      opcode = inst_token( SVGA3DOP_DEF );
+      def.dst = dst_register( SVGA3DREG_CONST, idx );
+      def.constValues[0] = a;
+      def.constValues[1] = b;
+      def.constValues[2] = c;
+      def.constValues[3] = d;
+      break;
+   case SVGA3D_CONST_TYPE_INT:
+      opcode = inst_token( SVGA3DOP_DEFI );
+      def.dst = dst_register( SVGA3DREG_CONSTINT, idx );
+      def.constIValues[0] = (int)a;
+      def.constIValues[1] = (int)b;
+      def.constIValues[2] = (int)c;
+      def.constIValues[3] = (int)d;
+      break;
+   default:
+      assert(0);
+      break;
+   }
+
+   if (!emit_instruction(emit, opcode) ||
+       !svga_shader_emit_dwords( emit, def.values, Elements(def.values)))
+      return FALSE;
+
+   return TRUE;
+}
+
+static INLINE boolean
+create_zero_immediate( struct svga_shader_emitter *emit )
+{
+   unsigned idx = emit->nr_hw_const++;
+
+   if (!emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT,
+                        idx, 0, 0, 0, 1 ))
+      return FALSE;
+
+   emit->zero_immediate_idx = idx;
+   emit->created_zero_immediate = TRUE;
+
+   return TRUE;
+}
+
+static INLINE boolean
+create_loop_const( struct svga_shader_emitter *emit )
+{
+   unsigned idx = emit->nr_hw_const++;
+
+   if (!emit_def_const( emit, SVGA3D_CONST_TYPE_INT, idx,
+                        255, /* iteration count */
+                        0, /* initial value */
+                        1, /* step size */
+                        0 /* not used, must be 0 */))
+      return FALSE;
+
+   emit->loop_const_idx = idx;
+   emit->created_loop_const = TRUE;
+
+   return TRUE;
+}
+
+static INLINE boolean
+create_sincos_consts( struct svga_shader_emitter *emit )
+{
+   unsigned idx = emit->nr_hw_const++;
+
+   if (!emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT, idx,
+                        -1.5500992e-006f,
+                        -2.1701389e-005f,
+                        0.0026041667f,
+                        0.00026041668f ))
+      return FALSE;
+
+   emit->sincos_consts_idx = idx;
+   idx = emit->nr_hw_const++;
+
+   if (!emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT, idx,
+                        -0.020833334f,
+                        -0.12500000f,
+                        1.0f,
+                        0.50000000f ))
+      return FALSE;
+
+   emit->created_sincos_consts = TRUE;
+
+   return TRUE;
+}
+
+static INLINE boolean
+create_arl_consts( struct svga_shader_emitter *emit )
+{
+   int i;
+
+   for (i = 0; i < emit->num_arl_consts; i += 4) {
+      int j;
+      unsigned idx = emit->nr_hw_const++;
+      float vals[4];
+      for (j = 0; j < 4 && (j + i) < emit->num_arl_consts; ++j) {
+         vals[j] = emit->arl_consts[i + j].number;
+         emit->arl_consts[i + j].idx = idx;
+         switch (j) {
+         case 0:
+            emit->arl_consts[i + 0].swizzle = TGSI_SWIZZLE_X;
+            break;
+         case 1:
+            emit->arl_consts[i + 0].swizzle = TGSI_SWIZZLE_Y;
+            break;
+         case 2:
+            emit->arl_consts[i + 0].swizzle = TGSI_SWIZZLE_Z;
+            break;
+         case 3:
+            emit->arl_consts[i + 0].swizzle = TGSI_SWIZZLE_W;
+            break;
+         }
+      }
+      while (j < 4)
+         vals[j++] = 0;
+
+      if (!emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT, idx,
+                           vals[0], vals[1],
+                           vals[2], vals[3]))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+static INLINE struct src_register
+get_vface( struct svga_shader_emitter *emit )
+{
+   assert(emit->emitted_vface);
+   return src_register(SVGA3DREG_MISCTYPE, 
+                       SVGA3DMISCREG_FACE);
+}
+
+/* returns {0, 0, 0, 1} immediate */
+static INLINE struct src_register
+get_zero_immediate( struct svga_shader_emitter *emit )
+{
+   assert(emit->created_zero_immediate);
+   assert(emit->zero_immediate_idx >= 0);
+   return src_register( SVGA3DREG_CONST,
+                        emit->zero_immediate_idx );
+}
+
+/* returns the loop const */
+static INLINE struct src_register
+get_loop_const( struct svga_shader_emitter *emit )
+{
+   assert(emit->created_loop_const);
+   assert(emit->loop_const_idx >= 0);
+   return src_register( SVGA3DREG_CONSTINT,
+                        emit->loop_const_idx );
+}
+
+/* returns a sincos const */
+static INLINE struct src_register
+get_sincos_const( struct svga_shader_emitter *emit,
+                  unsigned index )
+{
+   assert(emit->created_sincos_consts);
+   assert(emit->sincos_consts_idx >= 0);
+   assert(index == 0 || index == 1);
+   return src_register( SVGA3DREG_CONST,
+                        emit->sincos_consts_idx + index );
+}
+
+static INLINE struct src_register
+get_fake_arl_const( struct svga_shader_emitter *emit )
+{
+   struct src_register reg;
+   int idx = 0, swizzle = 0, i;
+
+   for (i = 0; i < emit->num_arl_consts; ++ i) {
+      if (emit->arl_consts[i].arl_num == emit->current_arl) {
+         idx = emit->arl_consts[i].idx;
+         swizzle = emit->arl_consts[i].swizzle;
+      }
+   }
+
+   reg = src_register( SVGA3DREG_CONST, idx );
+   return scalar(reg, swizzle);
+}
+
+static INLINE struct src_register
+get_tex_dimensions( struct svga_shader_emitter *emit, int sampler_num )
+{
+   int idx;
+   struct src_register reg;
+
+   /* the width/height indexes start right after constants */
+   idx = emit->key.fkey.tex[sampler_num].width_height_idx +
+         emit->info.file_max[TGSI_FILE_CONSTANT] + 1;
+
+   reg = src_register( SVGA3DREG_CONST, idx );
+   return reg;
+}
+
+static boolean emit_fake_arl(struct svga_shader_emitter *emit,
+                             const struct tgsi_full_instruction *insn)
+{
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 = get_fake_arl_const( emit );
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   SVGA3dShaderDestToken tmp = get_temp( emit );
+
+   if (!submit_op1(emit, inst_token( SVGA3DOP_MOV ), tmp, src0))
+      return FALSE;
+
+   if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ), tmp, src( tmp ),
+                    src1))
+      return FALSE;
+
+   /* replicate the original swizzle */
+   src1 = src(tmp);
+   src1.base.swizzle = src0.base.swizzle;
+
+   return submit_op1( emit, inst_token( SVGA3DOP_MOVA ),
+                      dst, src1 );
+}
+
+static boolean emit_if(struct svga_shader_emitter *emit,
+                       const struct tgsi_full_instruction *insn)
+{
+   const struct src_register src = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register zero = get_zero_immediate( emit );
+   SVGA3dShaderInstToken if_token = inst_token( SVGA3DOP_IFC );
+
+   if_token.control = SVGA3DOPCOMPC_NE;
+   zero = scalar(zero, TGSI_SWIZZLE_X);
+
+   return (emit_instruction( emit, if_token ) &&
+           emit_src( emit, src ) &&
+           emit_src( emit, zero ) );
+}
+
+static boolean emit_endif(struct svga_shader_emitter *emit,
+                       const struct tgsi_full_instruction *insn)
+{
+   return (emit_instruction( emit,
+                             inst_token( SVGA3DOP_ENDIF )));
+}
+
+static boolean emit_else(struct svga_shader_emitter *emit,
+                         const struct tgsi_full_instruction *insn)
+{
+   return (emit_instruction( emit,
+                             inst_token( SVGA3DOP_ELSE )));
+}
+
+/* Translate the following TGSI FLR instruction.
+ *    FLR  DST, SRC
+ * To the following SVGA3D instruction sequence.
+ *    FRC  TMP, SRC
+ *    SUB  DST, SRC, TMP
+ */
+static boolean emit_floor(struct svga_shader_emitter *emit,
+                          const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* FRC  TMP, SRC */
+   if (!submit_op1( emit, inst_token( SVGA3DOP_FRC ), temp, src0 ))
+      return FALSE;
+
+   /* SUB  DST, SRC, TMP */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ), dst, src0,
+                    negate( src( temp ) ) ))
+      return FALSE;
+
+   return TRUE;
+}
+
+
+/* Translate the following TGSI CMP instruction.
+ *    CMP  DST, SRC0, SRC1, SRC2
+ * To the following SVGA3D instruction sequence.
+ *    CMP  DST, SRC0, SRC2, SRC1
+ */
+static boolean emit_cmp(struct svga_shader_emitter *emit,
+                          const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   const struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   const struct src_register src2 = translate_src_register(
+      emit, &insn->FullSrcRegisters[2] );
+
+   /* CMP  DST, SRC0, SRC2, SRC1 */
+   return submit_op3( emit, inst_token( SVGA3DOP_CMP ), dst, src0, src2, src1);
+}
+
+
+
+/* Translate the following TGSI DIV instruction.
+ *    DIV  DST.xy, SRC0, SRC1
+ * To the following SVGA3D instruction sequence.
+ *    RCP  TMP.x, SRC1.xxxx
+ *    RCP  TMP.y, SRC1.yyyy
+ *    MUL  DST.xy, SRC0, TMP
+ */
+static boolean emit_div(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   const struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+   int i;
+
+   /* For each enabled element, perform a RCP instruction.  Note that
+    * RCP is scalar in SVGA3D:
+    */
+   for (i = 0; i < 4; i++) {
+      unsigned channel = 1 << i;
+      if (dst.mask & channel) {
+         /* RCP  TMP.?, SRC1.???? */
+         if (!submit_op1( emit, inst_token( SVGA3DOP_RCP ), 
+                          writemask(temp, channel), 
+                          scalar(src1, i) ))
+            return FALSE;
+      }
+   }
+
+   /* Then multiply them out with a single mul:
+    *
+    * MUL  DST, SRC0, TMP
+    */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ), dst, src0,
+                    src( temp ) ))
+      return FALSE;
+
+   return TRUE;
+}
+
+/* Translate the following TGSI DP2 instruction.
+ *    DP2  DST, SRC1, SRC2
+ * To the following SVGA3D instruction sequence.
+ *    MUL  TMP, SRC1, SRC2
+ *    ADD  DST, TMP.xxxx, TMP.yyyy
+ */
+static boolean emit_dp2(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   const struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+   struct src_register temp_src0, temp_src1;
+
+   /* MUL  TMP, SRC1, SRC2 */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ), temp, src0, src1 ))
+      return FALSE;
+
+   temp_src0 = scalar(src( temp ), TGSI_SWIZZLE_X);
+   temp_src1 = scalar(src( temp ), TGSI_SWIZZLE_Y);
+
+   /* ADD  DST, TMP.xxxx, TMP.yyyy */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ), dst,
+                    temp_src0, temp_src1 ))
+      return FALSE;
+
+   return TRUE;
+}
+
+
+/* Translate the following TGSI DPH instruction.
+ *    DPH  DST, SRC1, SRC2
+ * To the following SVGA3D instruction sequence.
+ *    DP3  TMP, SRC1, SRC2
+ *    ADD  DST, TMP, SRC2.wwww
+ */
+static boolean emit_dph(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* DP3  TMP, SRC1, SRC2 */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_DP3 ), temp, src0, src1 ))
+      return FALSE;
+
+   src1 = scalar(src1, TGSI_SWIZZLE_W);
+
+   /* ADD  DST, TMP, SRC2.wwww */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ), dst,
+                    src( temp ), src1 ))
+      return FALSE;
+
+   return TRUE;
+}
+
+/* Translate the following TGSI DST instruction.
+ *    NRM  DST, SRC
+ * To the following SVGA3D instruction sequence.
+ *    DP3  TMP, SRC, SRC
+ *    RSQ  TMP, TMP
+ *    MUL  DST, SRC, TMP
+ */
+static boolean emit_nrm(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* DP3  TMP, SRC, SRC */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_DP3 ), temp, src0, src0 ))
+      return FALSE;
+
+   /* RSQ  TMP, TMP */
+   if (!submit_op1( emit, inst_token( SVGA3DOP_RSQ ), temp, src( temp )))
+      return FALSE;
+
+   /* MUL  DST, SRC, TMP */
+   if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ), dst,
+                    src0, src( temp )))
+      return FALSE;
+
+   return TRUE;
+
+}
+
+static boolean do_emit_sincos(struct svga_shader_emitter *emit,
+                              SVGA3dShaderDestToken dst,
+                              struct src_register src0)
+{
+   src0 = scalar(src0, TGSI_SWIZZLE_X);
+
+   if (emit->use_sm30) {
+      return submit_op1( emit, inst_token( SVGA3DOP_SINCOS ),
+                         dst, src0 );
+   } else {
+      struct src_register const1 = get_sincos_const( emit, 0 );
+      struct src_register const2 = get_sincos_const( emit, 1 );
+
+      return submit_op3( emit, inst_token( SVGA3DOP_SINCOS ),
+                         dst, src0, const1, const2 );
+   }
+}
+
+static boolean emit_sincos(struct svga_shader_emitter *emit,
+                           const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* SCS TMP SRC */
+   if (!do_emit_sincos(emit, writemask(temp, TGSI_WRITEMASK_XY), src0 ))
+      return FALSE;
+
+   /* MOV DST TMP */
+   if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), dst, src( temp ) ))
+      return FALSE;
+
+   return TRUE;
+}
+
+/*
+ * SCS TMP SRC
+ * MOV DST TMP.yyyy
+ */
+static boolean emit_sin(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* SCS TMP SRC */
+   if (!do_emit_sincos(emit, writemask(temp, TGSI_WRITEMASK_Y), src0))
+      return FALSE;
+
+   src0 = scalar(src( temp ), TGSI_SWIZZLE_Y);
+
+   /* MOV DST TMP.yyyy */
+   if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), dst, src0 ))
+      return FALSE;
+
+   return TRUE;
+}
+
+/*
+ * SCS TMP SRC
+ * MOV DST TMP.xxxx
+ */
+static boolean emit_cos(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   SVGA3dShaderDestToken temp = get_temp( emit );
+
+   /* SCS TMP SRC */
+   if (!do_emit_sincos( emit, writemask(temp, TGSI_WRITEMASK_X), src0 ))
+      return FALSE;
+
+   src0 = scalar(src( temp ), TGSI_SWIZZLE_X);
+
+   /* MOV DST TMP.xxxx */
+   if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), dst, src0 ))
+      return FALSE;
+
+   return TRUE;
+}
+
+
+/*
+ * ADD DST SRC0, negate(SRC0)
+ */
+static boolean emit_sub(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+
+   src1 = negate(src1);
+
+   if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ), dst,
+                    src0, src1 ))
+      return FALSE;
+
+   return TRUE;
+}
+
+
+static boolean emit_kil(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst;
+   const struct tgsi_full_src_register *reg = &insn->FullSrcRegisters[0];
+   struct src_register src0;
+
+   inst = inst_token( SVGA3DOP_TEXKILL );
+   src0 = translate_src_register( emit, reg );
+
+   if (reg->SrcRegisterExtMod.Absolute ||
+       reg->SrcRegister.Negate != reg->SrcRegisterExtMod.Negate ||
+       reg->SrcRegister.Indirect ||
+       reg->SrcRegister.SwizzleX != 0 ||
+       reg->SrcRegister.SwizzleY != 1 ||
+       reg->SrcRegister.SwizzleZ != 2 ||
+       reg->SrcRegister.File != TGSI_FILE_TEMPORARY)
+   {
+      SVGA3dShaderDestToken temp = get_temp( emit );
+
+      submit_op1( emit, inst_token( SVGA3DOP_MOV ), temp, src0 );
+      src0 = src( temp );
+   }
+
+   return submit_op0( emit, inst, dst(src0) );
+}
+
+
+/* mesa state tracker always emits kilp as an unconditional
+ * kil */
+static boolean emit_kilp(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst;
+   SVGA3dShaderDestToken temp;
+   struct src_register one = get_zero_immediate( emit );
+
+   inst = inst_token( SVGA3DOP_TEXKILL );
+   one = scalar( one, TGSI_SWIZZLE_W );
+
+   /* texkill doesn't allow negation on the operand so lets move
+    * negation of {1} to a temp register */
+   temp = get_temp( emit );
+   if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), temp,
+                    negate( one ) ))
+      return FALSE;
+
+   return submit_op0( emit, inst, temp );
+}
+
+/* Implement conditionals by initializing destination reg to 'fail',
+ * then set predicate reg with UFOP_SETP, then move 'pass' to dest
+ * based on predicate reg.
+ *
+ * SETP src0, cmp, src1  -- do this first to avoid aliasing problems.
+ * MOV dst, fail
+ * MOV dst, pass, p0 
+ */
+static boolean
+emit_conditional(struct svga_shader_emitter *emit,
+                 unsigned compare_func,
+                 SVGA3dShaderDestToken dst,
+                 struct src_register src0,
+                 struct src_register src1,
+                 struct src_register pass,
+                 struct src_register fail)
+{
+   SVGA3dShaderDestToken pred_reg = dst_register( SVGA3DREG_PREDICATE, 0 );
+   SVGA3dShaderInstToken setp_token, mov_token;
+   setp_token = inst_token( SVGA3DOP_SETP );
+
+   switch (compare_func) {
+   case PIPE_FUNC_NEVER:
+      return submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                         dst, fail );
+      break;
+   case PIPE_FUNC_LESS:
+      setp_token.control = SVGA3DOPCOMP_LT;
+      break;
+   case PIPE_FUNC_EQUAL:
+      setp_token.control = SVGA3DOPCOMP_EQ;
+      break;
+   case PIPE_FUNC_LEQUAL:
+      setp_token.control = SVGA3DOPCOMP_LE;
+      break;
+   case PIPE_FUNC_GREATER:
+      setp_token.control = SVGA3DOPCOMP_GT;
+      break;
+   case PIPE_FUNC_NOTEQUAL:
+      setp_token.control = SVGA3DOPCOMPC_NE;
+      break;
+   case PIPE_FUNC_GEQUAL:
+      setp_token.control = SVGA3DOPCOMP_GE;
+      break;
+   case PIPE_FUNC_ALWAYS:
+      return submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                         dst, pass );
+      break;
+   }
+
+   /* SETP src0, COMPOP, src1 */
+   if (!submit_op2( emit, setp_token, pred_reg,
+                    src0, src1 ))
+      return FALSE;
+
+   mov_token = inst_token( SVGA3DOP_MOV );
+
+   /* MOV dst, fail */
+   if (!submit_op1( emit, mov_token, dst,
+                    fail ))
+      return FALSE;
+
+   /* MOV dst, pass (predicated)
+    *
+    * Note that the predicate reg (and possible modifiers) is passed
+    * as the first source argument.
+    */
+   mov_token.predicated = 1;
+   if (!submit_op2( emit, mov_token, dst,
+                    src( pred_reg ), pass ))
+      return FALSE;
+
+   return TRUE;
+}
+
+
+static boolean
+emit_select(struct svga_shader_emitter *emit,
+            unsigned compare_func,
+            SVGA3dShaderDestToken dst,
+            struct src_register src0,
+            struct src_register src1 )
+{
+   /* There are some SVGA instructions which implement some selects
+    * directly, but they are only available in the vertex shader.
+    */
+   if (emit->unit == PIPE_SHADER_VERTEX) {
+      switch (compare_func) {
+      case PIPE_FUNC_GEQUAL:
+         return submit_op2( emit, inst_token( SVGA3DOP_SGE ), dst, src0, src1 );
+      case PIPE_FUNC_LEQUAL:
+         return submit_op2( emit, inst_token( SVGA3DOP_SGE ), dst, src1, src0 );
+      case PIPE_FUNC_GREATER:
+         return submit_op2( emit, inst_token( SVGA3DOP_SLT ), dst, src1, src0 );
+      case PIPE_FUNC_LESS:
+         return submit_op2( emit, inst_token( SVGA3DOP_SLT ), dst, src0, src1 );
+      default:
+         break;
+      }
+   }
+
+
+   /* Otherwise, need to use the setp approach:
+    */
+   {
+      struct src_register one, zero;
+      /* zero immediate is 0,0,0,1 */
+      zero = get_zero_immediate( emit );
+      one  = scalar( zero, TGSI_SWIZZLE_W );
+      zero = scalar( zero, TGSI_SWIZZLE_X );
+
+      return emit_conditional(
+         emit,
+         compare_func,
+         dst,
+         src0,
+         src1,
+         one, zero);
+   }
+}
+
+
+static boolean emit_select_op(struct svga_shader_emitter *emit,
+                              unsigned compare,
+                              const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+      
+   return emit_select( emit, compare, dst, src0, src1 );
+}
+
+
+/* Translate texture instructions to SVGA3D representation.
+ */
+static boolean emit_tex2(struct svga_shader_emitter *emit,
+                         const struct tgsi_full_instruction *insn,
+                         SVGA3dShaderDestToken dst )
+{
+   SVGA3dShaderInstToken inst;
+   struct src_register src0;
+   struct src_register src1;
+
+   inst.value = 0;
+   inst.op = SVGA3DOP_TEX;
+
+   switch (insn->Instruction.Opcode) {
+   case TGSI_OPCODE_TEX:
+      break;
+   case TGSI_OPCODE_TXP:
+      inst.control = SVGA3DOPCONT_PROJECT;
+      break;
+   case TGSI_OPCODE_TXB:
+      inst.control = SVGA3DOPCONT_BIAS;
+      break;
+   default:
+      assert(0);
+      return FALSE;
+   }
+
+   src0 = translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   src1 = translate_src_register( emit, &insn->FullSrcRegisters[1] );
+
+   if (emit->key.fkey.tex[src1.base.num].unnormalized) {
+      struct src_register wh = get_tex_dimensions( emit, src1.base.num );
+      SVGA3dShaderDestToken tmp = get_temp( emit );
+
+      /* MUL  tmp, SRC0, WH */
+      if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ),
+                       tmp, src0, wh ))
+         return FALSE;
+      src0 = src( tmp );
+   }
+
+   return submit_op2( emit, inst, dst, src0, src1 );
+}
+
+
+
+
+/* Translate texture instructions to SVGA3D representation.
+ */
+static boolean emit_tex3(struct svga_shader_emitter *emit,
+                         const struct tgsi_full_instruction *insn,
+                         SVGA3dShaderDestToken dst )
+{
+   SVGA3dShaderInstToken inst;
+   struct src_register src0;
+   struct src_register src1;
+   struct src_register src2;
+
+   inst.value = 0;
+
+   switch (insn->Instruction.Opcode) {
+   case TGSI_OPCODE_TXD: 
+      inst.op = SVGA3DOP_TEXLDD;
+      break;
+   case TGSI_OPCODE_TXL:
+      inst.op = SVGA3DOP_TEXLDL;
+      break;
+   }
+
+   src0 = translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   src1 = translate_src_register( emit, &insn->FullSrcRegisters[1] );
+   src2 = translate_src_register( emit, &insn->FullSrcRegisters[2] );
+
+   return submit_op3( emit, inst, dst, src0, src1, src2 );
+}
+
+
+static boolean emit_tex(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderDestToken dst = 
+      translate_dst_register( emit, insn, 0 );
+   struct src_register src0 =
+      translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 =
+      translate_src_register( emit, &insn->FullSrcRegisters[1] );
+
+   SVGA3dShaderDestToken tex_result;
+
+   /* check for shadow samplers */
+   boolean compare = (emit->key.fkey.tex[src1.base.num].compare_mode ==
+                      PIPE_TEX_COMPARE_R_TO_TEXTURE);
+
+
+   /* If doing compare processing, need to put this value into a
+    * temporary so it can be used as a source later on.
+    */
+   if (compare ||
+       (!emit->use_sm30 && dst.mask != TGSI_WRITEMASK_XYZW) ) {
+      tex_result = get_temp( emit );
+   }
+   else {
+      tex_result = dst;
+   }
+
+   switch(insn->Instruction.Opcode) {
+   case TGSI_OPCODE_TEX:
+   case TGSI_OPCODE_TXB:
+   case TGSI_OPCODE_TXP:
+      if (!emit_tex2( emit, insn, tex_result ))
+         return FALSE;
+      break;
+   case TGSI_OPCODE_TXL:
+   case TGSI_OPCODE_TXD:
+      if (!emit_tex3( emit, insn, tex_result ))
+         return FALSE;
+      break;
+   default:
+      assert(0);
+   }
+
+
+   if (compare) {
+      SVGA3dShaderDestToken src0_zdivw = get_temp( emit );
+      struct src_register tex_src_x = scalar(src(tex_result), TGSI_SWIZZLE_Y);
+      struct src_register one =
+         scalar( get_zero_immediate( emit ), TGSI_SWIZZLE_W );
+
+      /* Divide texcoord R by Q */
+      if (!submit_op1( emit, inst_token( SVGA3DOP_RCP ),
+                       src0_zdivw,
+                       scalar(src0, TGSI_SWIZZLE_W) ))
+         return FALSE;
+
+      if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ),
+                       src0_zdivw,
+                       scalar(src0, TGSI_SWIZZLE_Z),
+                       src(src0_zdivw) ))
+         return FALSE;
+
+      if (!emit_select(
+             emit,
+             emit->key.fkey.tex[src1.base.num].compare_func,
+             dst,
+             src(src0_zdivw),
+             tex_src_x))
+         return FALSE;
+
+      return submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                         writemask( dst, TGSI_WRITEMASK_W),
+                         one );
+   }
+   else if (!emit->use_sm30 && dst.mask != TGSI_WRITEMASK_XYZW) 
+   {
+      if (!emit_op1( emit, inst_token( SVGA3DOP_MOV ), dst, src(tex_result) ))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+static boolean emit_bgnloop2( struct svga_shader_emitter *emit,
+                              const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst = inst_token( SVGA3DOP_LOOP );
+   struct src_register loop_reg = src_register( SVGA3DREG_LOOP, 0 );
+   struct src_register const_int = get_loop_const( emit );
+
+   return (emit_instruction( emit, inst ) &&
+           emit_src( emit, loop_reg ) &&
+           emit_src( emit, const_int ) );
+}
+
+static boolean emit_endloop2( struct svga_shader_emitter *emit,
+                              const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst = inst_token( SVGA3DOP_ENDLOOP );
+   return emit_instruction( emit, inst );
+}
+
+static boolean emit_brk( struct svga_shader_emitter *emit,
+                         const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst = inst_token( SVGA3DOP_BREAK );
+   return emit_instruction( emit, inst );
+}
+
+static boolean emit_scalar_op1( struct svga_shader_emitter *emit,
+                                unsigned opcode,
+                                const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst;
+   SVGA3dShaderDestToken dst;
+   struct src_register src;
+
+   inst = inst_token( opcode );
+   dst = translate_dst_register( emit, insn, 0 );
+   src = translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   src = scalar( src, TGSI_SWIZZLE_X );
+
+   return submit_op1( emit, inst, dst, src );
+}
+
+
+static boolean emit_simple_instruction(struct svga_shader_emitter *emit,
+                                       unsigned opcode,
+                                       const struct tgsi_full_instruction *insn )
+{
+   const struct tgsi_full_src_register *src = insn->FullSrcRegisters;
+   SVGA3dShaderInstToken inst;
+   SVGA3dShaderDestToken dst;
+
+   inst = inst_token( opcode );
+   dst = translate_dst_register( emit, insn, 0 );
+
+   switch (insn->Instruction.NumSrcRegs) {
+   case 0:
+      return submit_op0( emit, inst, dst );
+   case 1:
+      return submit_op1( emit, inst, dst,
+                         translate_src_register( emit, &src[0] ));
+   case 2:
+      return submit_op2( emit, inst, dst,
+                         translate_src_register( emit, &src[0] ),
+                         translate_src_register( emit, &src[1] ) );
+   case 3:
+      return submit_op3( emit, inst, dst,
+                         translate_src_register( emit, &src[0] ),
+                         translate_src_register( emit, &src[1] ),
+                         translate_src_register( emit, &src[2] ) );
+   default:
+      assert(0);
+      return FALSE;
+   }
+}
+
+static boolean emit_arl(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   ++emit->current_arl;
+   if (svga_arl_needs_adjustment( emit )) {
+      return emit_fake_arl( emit, insn );
+   } else {
+      /* no need to adjust, just emit straight arl */
+      return emit_simple_instruction(emit, SVGA3DOP_MOVA, insn);
+   }
+}
+
+static boolean alias_src_dst( struct src_register src,
+                              SVGA3dShaderDestToken dst )
+{
+   if (src.base.num != dst.num)
+      return FALSE;
+
+   if (SVGA3dShaderGetRegType(dst.value) != 
+       SVGA3dShaderGetRegType(src.base.value))
+      return FALSE;
+
+   return TRUE;
+}
+
+static boolean emit_pow(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   boolean need_tmp = FALSE;
+   
+   /* POW can only output to a temporary */
+   if (insn->FullDstRegisters[0].DstRegister.File != TGSI_FILE_TEMPORARY)
+      need_tmp = TRUE;
+   
+   /* POW src1 must not be the same register as dst */
+   if (alias_src_dst( src1, dst ))
+      need_tmp = TRUE;
+
+   /* it's a scalar op */
+   src0 = scalar( src0, TGSI_SWIZZLE_X );
+   src1 = scalar( src1, TGSI_SWIZZLE_X );
+
+   if (need_tmp) {
+      SVGA3dShaderDestToken tmp = writemask(get_temp( emit ), TGSI_WRITEMASK_X );
+
+      if (!submit_op2(emit, inst_token( SVGA3DOP_POW ), tmp, src0, src1))
+         return FALSE;
+
+      return submit_op1(emit, inst_token( SVGA3DOP_MOV ), dst, scalar(src(tmp), 0) );
+   } 
+   else {
+      return submit_op2(emit, inst_token( SVGA3DOP_POW ), dst, src0, src1);
+   }
+}
+
+static boolean emit_xpd(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   const struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   boolean need_dst_tmp = FALSE;
+
+   /* XPD can only output to a temporary */
+   if (SVGA3dShaderGetRegType(dst.value) != SVGA3DREG_TEMP) 
+      need_dst_tmp = TRUE;
+
+   /* The dst reg must not be the same as src0 or src1*/
+   if (alias_src_dst(src0, dst) ||
+       alias_src_dst(src1, dst))
+      need_dst_tmp = TRUE;
+
+   if (need_dst_tmp) {
+      SVGA3dShaderDestToken tmp = get_temp( emit );
+
+      /* Obey DX9 restrictions on mask:
+       */
+      tmp.mask = dst.mask & TGSI_WRITEMASK_XYZ;
+
+      if (!submit_op2(emit, inst_token( SVGA3DOP_CRS ), tmp, src0, src1))
+         return FALSE;
+
+      if (!submit_op1(emit, inst_token( SVGA3DOP_MOV ), dst, src( tmp )))
+         return FALSE;
+   } 
+   else {
+      if (!submit_op2(emit, inst_token( SVGA3DOP_CRS ), dst, src0, src1))
+         return FALSE;
+   }
+
+   /* Need to emit 1.0 to dst.w?
+    */
+   if (dst.mask & TGSI_WRITEMASK_W) {
+      struct src_register zero = get_zero_immediate( emit );
+
+      if (!submit_op1(emit, 
+                      inst_token( SVGA3DOP_MOV ), 
+                      writemask(dst, TGSI_WRITEMASK_W),
+                      zero))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+
+static boolean emit_lrp(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   SVGA3dShaderDestToken tmp;
+   const struct src_register src0 = translate_src_register(
+      emit, &insn->FullSrcRegisters[0] );
+   const struct src_register src1 = translate_src_register(
+      emit, &insn->FullSrcRegisters[1] );
+   const struct src_register src2 = translate_src_register(
+      emit, &insn->FullSrcRegisters[2] );
+   boolean need_dst_tmp = FALSE;
+
+   /* The dst reg must not be the same as src0 or src2 */
+   if (alias_src_dst(src0, dst) ||
+       alias_src_dst(src2, dst))
+      need_dst_tmp = TRUE;
+
+   if (need_dst_tmp) {
+      tmp = get_temp( emit );
+      tmp.mask = dst.mask;
+   }
+   else {
+      tmp = dst;
+   }
+
+   if (!submit_op3(emit, inst_token( SVGA3DOP_LRP ), tmp, src0, src1, src2))
+      return FALSE;
+
+   if (need_dst_tmp) {
+      if (!submit_op1(emit, inst_token( SVGA3DOP_MOV ), dst, src( tmp )))
+         return FALSE;      
+   } 
+
+   return TRUE;
+}
+
+
+static boolean emit_dst_insn(struct svga_shader_emitter *emit,
+                             const struct tgsi_full_instruction *insn )
+{
+   if (emit->unit == PIPE_SHADER_VERTEX) {
+      /* SVGA/DX9 has a DST instruction, but only for vertex shaders:
+       */
+      return emit_simple_instruction(emit, SVGA3DOP_DST, insn);
+   }
+   else {
+
+      /* result[0] = 1    * 1;
+       * result[1] = a[1] * b[1];
+       * result[2] = a[2] * 1;
+       * result[3] = 1    * b[3];
+       */
+
+      SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+      SVGA3dShaderDestToken tmp;
+      const struct src_register src0 = translate_src_register(
+         emit, &insn->FullSrcRegisters[0] );
+      const struct src_register src1 = translate_src_register(
+         emit, &insn->FullSrcRegisters[1] );
+      struct src_register zero = get_zero_immediate( emit );
+      boolean need_tmp = FALSE;
+
+      if (SVGA3dShaderGetRegType(dst.value) != SVGA3DREG_TEMP ||
+          alias_src_dst(src0, dst) ||
+          alias_src_dst(src1, dst))
+         need_tmp = TRUE;
+
+      if (need_tmp) {
+         tmp = get_temp( emit );
+      }
+      else {
+         tmp = dst;
+      }
+
+      /* tmp.xw = 1.0
+       */
+      if (tmp.mask & TGSI_WRITEMASK_XW) {
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), 
+                          writemask(tmp, TGSI_WRITEMASK_XW ),
+                          scalar( zero, 3 )))
+            return FALSE;
+      }
+      
+      /* tmp.yz = src0
+       */
+      if (tmp.mask & TGSI_WRITEMASK_YZ) {
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), 
+                          writemask(tmp, TGSI_WRITEMASK_YZ ),
+                          src0))
+            return FALSE;
+      }
+
+      /* tmp.yw = tmp * src1
+       */
+      if (tmp.mask & TGSI_WRITEMASK_YW) {
+         if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ), 
+                          writemask(tmp, TGSI_WRITEMASK_YW ),
+                          src(tmp),
+                          src1))
+            return FALSE;
+      }
+
+      /* dst = tmp
+       */
+      if (need_tmp) {
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), 
+                          dst,
+                          src(tmp)))
+            return FALSE;
+      }      
+   }
+   
+   return TRUE;
+}
+
+
+static boolean emit_exp(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 =
+      translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   struct src_register zero = get_zero_immediate( emit );
+   SVGA3dShaderDestToken fraction;
+
+   if (dst.mask & TGSI_WRITEMASK_Y)
+      fraction = dst;
+   else if (dst.mask & TGSI_WRITEMASK_X)
+      fraction = get_temp( emit );
+
+   /* If y is being written, fill it with src0 - floor(src0).
+    */
+   if (dst.mask & TGSI_WRITEMASK_XY) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_FRC ),
+                       writemask( fraction, TGSI_WRITEMASK_Y ),
+                       src0 ))
+         return FALSE;
+   }
+
+   /* If x is being written, fill it with 2 ^ floor(src0).
+    */
+   if (dst.mask & TGSI_WRITEMASK_X) {
+      if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ),
+                       writemask( dst, dst.mask & TGSI_WRITEMASK_X ),
+                       src0,
+                       scalar( negate( src( fraction ) ), TGSI_SWIZZLE_Y ) ) )
+         return FALSE;
+
+      if (!submit_op1( emit, inst_token( SVGA3DOP_EXP ),
+                       writemask( dst, dst.mask & TGSI_WRITEMASK_X ),
+                       scalar( src( dst ), TGSI_SWIZZLE_X ) ) )
+         return FALSE;
+
+      if (!(dst.mask & TGSI_WRITEMASK_Y))
+         release_temp( emit, fraction );
+   }
+
+   /* If z is being written, fill it with 2 ^ src0 (partial precision).
+    */
+   if (dst.mask & TGSI_WRITEMASK_Z) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_EXPP ),
+                       writemask( dst, dst.mask & TGSI_WRITEMASK_Z ),
+                       src0 ) )
+         return FALSE;
+   }
+
+   /* If w is being written, fill it with one.
+    */
+   if (dst.mask & TGSI_WRITEMASK_W) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                       writemask(dst, TGSI_WRITEMASK_W),
+                       scalar( zero, TGSI_SWIZZLE_W ) ))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+static boolean emit_lit(struct svga_shader_emitter *emit,
+                             const struct tgsi_full_instruction *insn )
+{
+   if (emit->unit == PIPE_SHADER_VERTEX) {
+      /* SVGA/DX9 has a LIT instruction, but only for vertex shaders:
+       */
+      return emit_simple_instruction(emit, SVGA3DOP_LIT, insn);
+   }
+   else {
+
+      /* D3D vs. GL semantics can be fairly easily accomodated by
+       * variations on this sequence.
+       *
+       * GL:
+       *   tmp.y = src.x
+       *   tmp.z = pow(src.y,src.w)
+       *   p0 = src0.xxxx > 0
+       *   result = zero.wxxw
+       *   (p0) result.yz = tmp
+       *
+       * D3D:
+       *   tmp.y = src.x
+       *   tmp.z = pow(src.y,src.w)
+       *   p0 = src0.xxyy > 0
+       *   result = zero.wxxw
+       *   (p0) result.yz = tmp
+       *
+       * Will implement the GL version for now.
+       */
+
+      SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+      SVGA3dShaderDestToken tmp = get_temp( emit );
+      const struct src_register src0 = translate_src_register(
+         emit, &insn->FullSrcRegisters[0] );
+      struct src_register zero = get_zero_immediate( emit );
+
+      /* tmp = pow(src.y, src.w)
+       */
+      if (dst.mask & TGSI_WRITEMASK_Z) {
+         if (!submit_op2(emit, inst_token( SVGA3DOP_POW ), 
+                         tmp, 
+                         scalar(src0, 1), 
+                         scalar(src0, 3)))
+            return FALSE;
+      }
+
+      /* tmp.y = src.x
+       */
+      if (dst.mask & TGSI_WRITEMASK_Y) {
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), 
+                          writemask(tmp, TGSI_WRITEMASK_Y ),
+                          scalar(src0, 0)))
+            return FALSE;
+      }
+      
+      /* Can't quite do this with emit conditional due to the extra
+       * writemask on the predicated mov:
+       */
+      {
+         SVGA3dShaderDestToken pred_reg = dst_register( SVGA3DREG_PREDICATE, 0 );
+         SVGA3dShaderInstToken setp_token, mov_token;
+         struct src_register predsrc;
+
+         setp_token = inst_token( SVGA3DOP_SETP );
+         mov_token = inst_token( SVGA3DOP_MOV );
+
+         setp_token.control = SVGA3DOPCOMP_GT;
+
+         /* D3D vs GL semantics:
+          */
+         if (0)
+            predsrc = swizzle(src0, 0, 0, 1, 1); /* D3D */
+         else
+            predsrc = swizzle(src0, 0, 0, 0, 0); /* GL */
+
+         /* SETP src0.xxyy, GT, {0}.x */
+         if (!submit_op2( emit, setp_token, pred_reg,
+                          predsrc, 
+                          swizzle(zero, 0, 0, 0, 0) ))
+            return FALSE;
+         
+         /* MOV dst, fail */
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), dst,
+                          swizzle(zero, 3, 0, 0, 3 )))
+             return FALSE;
+
+         /* MOV dst.yz, tmp (predicated)
+          *
+          * Note that the predicate reg (and possible modifiers) is passed
+          * as the first source argument.
+          */
+         if (dst.mask & TGSI_WRITEMASK_YZ) {
+            mov_token.predicated = 1;
+            if (!submit_op2( emit, mov_token,
+                             writemask(dst, TGSI_WRITEMASK_YZ),
+                             src( pred_reg ), src( tmp ) ))
+               return FALSE;
+         }
+      }
+   }
+
+   return TRUE;
+}
+
+
+
+
+static boolean emit_ex2( struct svga_shader_emitter *emit,
+                         const struct tgsi_full_instruction *insn )
+{
+   SVGA3dShaderInstToken inst;
+   SVGA3dShaderDestToken dst;
+   struct src_register src0;
+
+   inst = inst_token( SVGA3DOP_EXP );
+   dst = translate_dst_register( emit, insn, 0 );
+   src0 = translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   src0 = scalar( src0, TGSI_SWIZZLE_X );
+
+   if (dst.mask != TGSI_WRITEMASK_XYZW) {
+      SVGA3dShaderDestToken tmp = get_temp( emit );
+
+      if (!submit_op1( emit, inst, tmp, src0 ))
+         return FALSE;
+
+      return submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                         dst,
+                         scalar( src( tmp ), TGSI_SWIZZLE_X ) );
+   }
+
+   return submit_op1( emit, inst, dst, src0 );
+}
+
+
+static boolean emit_log(struct svga_shader_emitter *emit,
+                        const struct tgsi_full_instruction *insn)
+{
+   SVGA3dShaderDestToken dst = translate_dst_register( emit, insn, 0 );
+   struct src_register src0 =
+      translate_src_register( emit, &insn->FullSrcRegisters[0] );
+   struct src_register zero = get_zero_immediate( emit );
+   SVGA3dShaderDestToken abs_tmp;
+   struct src_register abs_src0;
+   SVGA3dShaderDestToken log2_abs;
+
+   if (dst.mask & TGSI_WRITEMASK_Z)
+      log2_abs = dst;
+   else if (dst.mask & TGSI_WRITEMASK_XY)
+      log2_abs = get_temp( emit );
+
+   /* If z is being written, fill it with log2( abs( src0 ) ).
+    */
+   if (dst.mask & TGSI_WRITEMASK_XYZ) {
+      if (!src0.base.srcMod || src0.base.srcMod == SVGA3DSRCMOD_ABS)
+         abs_src0 = src0;
+      else {
+         abs_tmp = get_temp( emit );
+
+         if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                          abs_tmp,
+                          src0 ) )
+            return FALSE;
+
+         abs_src0 = src( abs_tmp );
+      }
+
+      abs_src0 = absolute( scalar( abs_src0, TGSI_SWIZZLE_X ) );
+
+      if (!submit_op1( emit, inst_token( SVGA3DOP_LOG ),
+                       writemask( log2_abs, TGSI_WRITEMASK_Z ),
+                       abs_src0 ) )
+         return FALSE;
+   }
+
+   if (dst.mask & TGSI_WRITEMASK_XY) {
+      SVGA3dShaderDestToken floor_log2;
+
+      if (dst.mask & TGSI_WRITEMASK_X)
+         floor_log2 = dst;
+      else
+         floor_log2 = get_temp( emit );
+
+      /* If x is being written, fill it with floor( log2( abs( src0 ) ) ).
+       */
+      if (!submit_op1( emit, inst_token( SVGA3DOP_FRC ),
+                       writemask( floor_log2, TGSI_WRITEMASK_X ),
+                       scalar( src( log2_abs ), TGSI_SWIZZLE_Z ) ) )
+         return FALSE;
+
+      if (!submit_op2( emit, inst_token( SVGA3DOP_ADD ),
+                       writemask( floor_log2, TGSI_WRITEMASK_X ),
+                       scalar( src( log2_abs ), TGSI_SWIZZLE_Z ),
+                       negate( src( floor_log2 ) ) ) )
+         return FALSE;
+
+      /* If y is being written, fill it with
+       * abs ( src0 ) / ( 2 ^ floor( log2( abs( src0 ) ) ) ).
+       */
+      if (dst.mask & TGSI_WRITEMASK_Y) {
+         if (!submit_op1( emit, inst_token( SVGA3DOP_EXP ),
+                          writemask( dst, TGSI_WRITEMASK_Y ),
+                          negate( scalar( src( floor_log2 ),
+                                          TGSI_SWIZZLE_X ) ) ) )
+            return FALSE;
+
+         if (!submit_op2( emit, inst_token( SVGA3DOP_MUL ),
+                          writemask( dst, TGSI_WRITEMASK_Y ),
+                          src( dst ),
+                          abs_src0 ) )
+            return FALSE;
+      }
+
+      if (!(dst.mask & TGSI_WRITEMASK_X))
+         release_temp( emit, floor_log2 );
+
+      if (!(dst.mask & TGSI_WRITEMASK_Z))
+         release_temp( emit, log2_abs );
+   }
+
+   if (dst.mask & TGSI_WRITEMASK_XYZ && src0.base.srcMod &&
+       src0.base.srcMod != SVGA3DSRCMOD_ABS)
+      release_temp( emit, abs_tmp );
+
+   /* If w is being written, fill it with one.
+    */
+   if (dst.mask & TGSI_WRITEMASK_W) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ),
+                       writemask(dst, TGSI_WRITEMASK_W),
+                       scalar( zero, TGSI_SWIZZLE_W ) ))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+
+static boolean emit_bgnsub( struct svga_shader_emitter *emit,
+                           unsigned position,
+                           const struct tgsi_full_instruction *insn )
+{
+   unsigned i;
+
+   /* Note that we've finished the main function and are now emitting
+    * subroutines.  This affects how we terminate the generated
+    * shader.
+    */
+   emit->in_main_func = FALSE;
+   
+   for (i = 0; i < emit->nr_labels; i++) {
+      if (emit->label[i] == position) {
+         return (emit_instruction( emit, inst_token( SVGA3DOP_RET ) ) &&
+                 emit_instruction( emit, inst_token( SVGA3DOP_LABEL ) ) &&
+                 emit_src( emit, src_register( SVGA3DREG_LABEL, i )));
+      }
+   }
+
+   assert(0);
+   return TRUE;
+}
+
+static boolean emit_call( struct svga_shader_emitter *emit,
+                           const struct tgsi_full_instruction *insn )
+{
+   unsigned position = insn->InstructionExtLabel.Label;
+   unsigned i;
+   
+   for (i = 0; i < emit->nr_labels; i++) {
+      if (emit->label[i] == position) 
+         break;
+   }
+
+   if (emit->nr_labels == Elements(emit->label))
+      return FALSE;
+
+   if (i == emit->nr_labels) {
+      emit->label[i] = position;
+      emit->nr_labels++;
+   }
+
+   return (emit_instruction( emit, inst_token( SVGA3DOP_CALL ) ) &&
+           emit_src( emit, src_register( SVGA3DREG_LABEL, i )));
+}
+
+
+static boolean emit_end( struct svga_shader_emitter *emit )
+{
+   if (emit->unit == PIPE_SHADER_VERTEX) {
+      return emit_vs_postamble( emit );
+   }
+   else {
+      return emit_ps_postamble( emit );
+   }
+}
+
+
+
+static boolean svga_emit_instruction( struct svga_shader_emitter *emit,
+                                      unsigned position,
+                                      const struct tgsi_full_instruction *insn )
+{
+   switch (insn->Instruction.Opcode) {
+
+   case TGSI_OPCODE_ARL:
+      return emit_arl( emit, insn );
+
+   case TGSI_OPCODE_TEX:
+   case TGSI_OPCODE_TXB:
+   case TGSI_OPCODE_TXP:
+   case TGSI_OPCODE_TXL:
+   case TGSI_OPCODE_TXD:
+      return emit_tex( emit, insn );
+
+   case TGSI_OPCODE_BGNSUB:
+      return emit_bgnsub( emit, position, insn );
+
+   case TGSI_OPCODE_ENDSUB:
+      return TRUE;
+
+   case TGSI_OPCODE_CAL:
+      return emit_call( emit, insn );
+
+   case TGSI_OPCODE_FLR:
+   case TGSI_OPCODE_TRUNC:        /* should be TRUNC, not FLR */
+      return emit_floor( emit, insn );
+
+   case TGSI_OPCODE_CMP:
+      return emit_cmp( emit, insn );
+
+   case TGSI_OPCODE_DIV:
+      return emit_div( emit, insn );
+
+   case TGSI_OPCODE_DP2:
+      return emit_dp2( emit, insn );
+
+   case TGSI_OPCODE_DPH:
+      return emit_dph( emit, insn );
+
+   case TGSI_OPCODE_NRM:
+      return emit_nrm( emit, insn );
+
+   case TGSI_OPCODE_COS:
+      return emit_cos( emit, insn );
+
+   case TGSI_OPCODE_SIN:
+      return emit_sin( emit, insn );
+
+   case TGSI_OPCODE_SCS:
+      return emit_sincos( emit, insn );
+
+   case TGSI_OPCODE_END:
+      /* TGSI always finishes the main func with an END */
+      return emit_end( emit );
+
+   case TGSI_OPCODE_KIL:
+      return emit_kil( emit, insn );
+
+      /* Selection opcodes.  The underlying language is fairly
+       * non-orthogonal about these.
+       */
+   case TGSI_OPCODE_SEQ:
+      return emit_select_op( emit, PIPE_FUNC_EQUAL, insn );
+
+   case TGSI_OPCODE_SNE:
+      return emit_select_op( emit, PIPE_FUNC_NOTEQUAL, insn );
+
+   case TGSI_OPCODE_SGT:
+      return emit_select_op( emit, PIPE_FUNC_GREATER, insn );
+
+   case TGSI_OPCODE_SGE:
+      return emit_select_op( emit, PIPE_FUNC_GEQUAL, insn );
+
+   case TGSI_OPCODE_SLT:
+      return emit_select_op( emit, PIPE_FUNC_LESS, insn );
+
+   case TGSI_OPCODE_SLE:
+      return emit_select_op( emit, PIPE_FUNC_LEQUAL, insn );
+
+   case TGSI_OPCODE_SUB:
+      return emit_sub( emit, insn );
+
+   case TGSI_OPCODE_POW:
+      return emit_pow( emit, insn );
+
+   case TGSI_OPCODE_EX2:
+      return emit_ex2( emit, insn );
+
+   case TGSI_OPCODE_EXP:
+      return emit_exp( emit, insn );
+
+   case TGSI_OPCODE_LOG:
+      return emit_log( emit, insn );
+
+   case TGSI_OPCODE_LG2:
+      return emit_scalar_op1( emit, SVGA3DOP_LOG, insn );
+
+   case TGSI_OPCODE_RSQ:
+      return emit_scalar_op1( emit, SVGA3DOP_RSQ, insn );
+
+   case TGSI_OPCODE_RCP:
+      return emit_scalar_op1( emit, SVGA3DOP_RCP, insn );
+
+   case TGSI_OPCODE_CONT:
+   case TGSI_OPCODE_RET:
+      /* This is a noop -- we tell mesa that we can't support RET
+       * within a function (early return), so this will always be
+       * followed by an ENDSUB.
+       */
+      return TRUE;
+
+      /* These aren't actually used by any of the frontends we care
+       * about:
+       */
+   case TGSI_OPCODE_CLAMP:
+   case TGSI_OPCODE_ROUND:
+   case TGSI_OPCODE_AND:
+   case TGSI_OPCODE_OR:
+   case TGSI_OPCODE_I2F:
+   case TGSI_OPCODE_NOT:
+   case TGSI_OPCODE_SHL:
+   case TGSI_OPCODE_SHR:
+   case TGSI_OPCODE_XOR:
+      return FALSE;
+
+   case TGSI_OPCODE_IF:
+      return emit_if( emit, insn );
+   case TGSI_OPCODE_ELSE:
+      return emit_else( emit, insn );
+   case TGSI_OPCODE_ENDIF:
+      return emit_endif( emit, insn );
+
+   case TGSI_OPCODE_BGNLOOP:
+      return emit_bgnloop2( emit, insn );
+   case TGSI_OPCODE_ENDLOOP:
+      return emit_endloop2( emit, insn );
+   case TGSI_OPCODE_BRK:
+      return emit_brk( emit, insn );
+
+   case TGSI_OPCODE_XPD:
+      return emit_xpd( emit, insn );
+
+   case TGSI_OPCODE_KILP:
+      return emit_kilp( emit, insn );
+
+   case TGSI_OPCODE_DST:
+      return emit_dst_insn( emit, insn );
+
+   case TGSI_OPCODE_LIT:
+      return emit_lit( emit, insn );
+
+   case TGSI_OPCODE_LRP:
+      return emit_lrp( emit, insn );
+
+   default: {
+      unsigned opcode = translate_opcode(insn->Instruction.Opcode);
+
+      if (opcode == SVGA3DOP_LAST_INST)
+         return FALSE;
+
+      if (!emit_simple_instruction( emit, opcode, insn ))
+         return FALSE;
+   }
+   }
+
+   return TRUE;
+}
+
+
+static boolean svga_emit_immediate( struct svga_shader_emitter *emit,
+                                    struct tgsi_full_immediate *imm)
+{
+   static const float id[4] = {0,0,0,1};
+   float value[4];
+   unsigned i;
+
+   assert(1 <= imm->Immediate.NrTokens && imm->Immediate.NrTokens <= 5);
+   for (i = 0; i < imm->Immediate.NrTokens - 1; i++)
+      value[i] = imm->u[i].Float;
+
+   for ( ; i < 4; i++ )
+      value[i] = id[i];
+
+   return emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT,
+                          emit->imm_start + emit->internal_imm_count++,
+                          value[0], value[1], value[2], value[3]);
+}
+
+static boolean make_immediate( struct svga_shader_emitter *emit,
+                               float a,
+                               float b,
+                               float c,
+                               float d,
+                               struct src_register *out )
+{
+   unsigned idx = emit->nr_hw_const++;
+
+   if (!emit_def_const( emit, SVGA3D_CONST_TYPE_FLOAT,
+                        idx, a, b, c, d ))
+      return FALSE;
+
+   *out = src_register( SVGA3DREG_CONST, idx );
+
+   return TRUE;
+}
+
+static boolean emit_vs_preamble( struct svga_shader_emitter *emit )
+{
+   if (!emit->key.vkey.need_prescale) {
+      if (!make_immediate( emit, 0, 0, .5, .5,
+                           &emit->imm_0055))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+static boolean emit_ps_preamble( struct svga_shader_emitter *emit )
+{
+   unsigned i;
+
+   /* For SM20, need to initialize the temporaries we're using to hold
+    * color outputs to some value.  Shaders which don't set all of
+    * these values are likely to be rejected by the DX9 runtime.
+    */
+   if (!emit->use_sm30) {
+      struct src_register zero = get_zero_immediate( emit );
+      for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+         if (SVGA3dShaderGetRegType(emit->true_col[i].value) != 0) {
+            
+            if (!submit_op1( emit,
+                             inst_token(SVGA3DOP_MOV),
+                             emit->temp_col[i],
+                             zero ))
+               return FALSE;
+         }
+      }
+   }
+   
+   return TRUE;
+}
+
+static boolean emit_ps_postamble( struct svga_shader_emitter *emit )
+{
+   unsigned i;
+
+   /* PS oDepth is incredibly fragile and it's very hard to catch the
+    * types of usage that break it during shader emit.  Easier just to
+    * redirect the main program to a temporary and then only touch
+    * oDepth with a hand-crafted MOV below.
+    */
+   if (SVGA3dShaderGetRegType(emit->true_pos.value) != 0) {
+
+      if (!submit_op1( emit,
+                       inst_token(SVGA3DOP_MOV),
+                       emit->true_pos,
+                       scalar(src(emit->temp_pos), TGSI_SWIZZLE_Z) ))
+         return FALSE;
+   }
+
+   /* Similarly for SM20 color outputs...  Luckily SM30 isn't so
+    * fragile.
+    */
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      if (SVGA3dShaderGetRegType(emit->true_col[i].value) != 0) {
+
+         if (!submit_op1( emit,
+                          inst_token(SVGA3DOP_MOV),
+                          emit->true_col[i],
+                          src(emit->temp_col[i]) ))
+            return FALSE;
+      }
+   }
+
+   return TRUE;
+}
+
+static boolean emit_vs_postamble( struct svga_shader_emitter *emit )
+{
+   /* PSIZ output is incredibly fragile and it's very hard to catch
+    * the types of usage that break it during shader emit.  Easier
+    * just to redirect the main program to a temporary and then only
+    * touch PSIZ with a hand-crafted MOV below.
+    */
+   if (SVGA3dShaderGetRegType(emit->true_psiz.value) != 0) {
+      
+      if (!submit_op1( emit,
+                       inst_token(SVGA3DOP_MOV),
+                       emit->true_psiz,
+                       scalar(src(emit->temp_psiz), TGSI_SWIZZLE_X) ))
+         return FALSE;
+   }
+
+   /* Need to perform various manipulations on vertex position to cope
+    * with the different GL and D3D clip spaces.
+    */
+   if (emit->key.vkey.need_prescale) {
+      SVGA3dShaderDestToken temp_pos = emit->temp_pos;
+      SVGA3dShaderDestToken pos = emit->true_pos;
+      unsigned offset = emit->info.file_max[TGSI_FILE_CONSTANT] + 1;
+      struct src_register prescale_scale = src_register( SVGA3DREG_CONST, 
+                                                         offset + 0 ); 
+      struct src_register prescale_trans = src_register( SVGA3DREG_CONST, 
+                                                         offset + 1 ); 
+
+      /* MUL temp_pos.xyz,    temp_pos,      prescale.scale
+       * MAD result.position, temp_pos.wwww, prescale.trans, temp_pos
+       *   --> Note that prescale.trans.w == 0
+       */
+      if (!submit_op2( emit, 
+                       inst_token(SVGA3DOP_MUL), 
+                       writemask(temp_pos, TGSI_WRITEMASK_XYZ), 
+                       src(temp_pos),
+                       prescale_scale ))
+         return FALSE;
+
+      if (!submit_op3( emit, 
+                       inst_token(SVGA3DOP_MAD), 
+                       pos, 
+                       swizzle(src(temp_pos), 3, 3, 3, 3),
+                       prescale_trans,
+                       src(temp_pos)))
+         return FALSE;
+   }
+   else {
+      SVGA3dShaderDestToken temp_pos = emit->temp_pos;
+      SVGA3dShaderDestToken pos = emit->true_pos;
+      struct src_register imm_0055 = emit->imm_0055;
+
+      /* Adjust GL clipping coordinate space to hardware (D3D-style):
+       *
+       * DP4 temp_pos.z, {0,0,.5,.5}, temp_pos
+       * MOV result.position, temp_pos 
+       */
+      if (!submit_op2( emit, 
+                       inst_token(SVGA3DOP_DP4), 
+                       writemask(temp_pos, TGSI_WRITEMASK_Z), 
+                       imm_0055, 
+                       src(temp_pos) ))
+         return FALSE;
+
+      if (!submit_op1( emit,
+                       inst_token(SVGA3DOP_MOV),
+                       pos,
+                       src(temp_pos) ))
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
+/*
+  0: IF VFACE :4
+  1:   COLOR = FrontColor;
+  2: ELSE
+  3:   COLOR = BackColor;
+  4: ENDIF
+ */
+static boolean emit_light_twoside( struct svga_shader_emitter *emit )
+{
+   struct src_register vface, zero;
+   struct src_register front[2];
+   struct src_register back[2];
+   SVGA3dShaderDestToken color[2];
+   int count =  emit->internal_color_count;
+   int i;
+   SVGA3dShaderInstToken if_token;
+
+   if (count == 0)
+      return TRUE;
+
+   vface = get_vface( emit );
+   zero = get_zero_immediate( emit );
+
+   /* Can't use get_temp() to allocate the color reg as such
+    * temporaries will be reclaimed after each instruction by the call
+    * to reset_temp_regs().
+    */
+   for (i = 0; i < count; i++) {
+      color[i] = dst_register( SVGA3DREG_TEMP, 
+                               emit->nr_hw_temp++ );
+
+      front[i] = emit->input_map[emit->internal_color_idx[i]];
+
+      /* Back is always the next input:
+       */
+      back[i] = front[i];
+      back[i].base.num = front[i].base.num + 1;
+
+      /* Reassign the input_map to the actual front-face color:
+       */
+      emit->input_map[emit->internal_color_idx[i]] = src(color[i]);
+   }
+   
+   if_token = inst_token( SVGA3DOP_IFC );
+
+   if (emit->key.fkey.front_cw)
+      if_token.control = SVGA3DOPCOMP_GT;
+   else
+      if_token.control = SVGA3DOPCOMP_LT;
+
+   zero = scalar(zero, TGSI_SWIZZLE_X);
+
+   if (!(emit_instruction( emit, if_token ) &&
+         emit_src( emit, vface ) &&
+         emit_src( emit, zero ) ))
+      return FALSE;
+
+   for (i = 0; i < count; i++) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), color[i], front[i] ))
+         return FALSE;
+   }
+
+   if (!(emit_instruction( emit, inst_token( SVGA3DOP_ELSE))))
+      return FALSE;
+   
+   for (i = 0; i < count; i++) {
+      if (!submit_op1( emit, inst_token( SVGA3DOP_MOV ), color[i], back[i] ))
+         return FALSE;
+   }
+
+   if (!emit_instruction( emit, inst_token( SVGA3DOP_ENDIF ) ))
+      return FALSE;
+
+   return TRUE;
+}
+
+/*
+  0: SETP_GT TEMP, VFACE, 0
+  where TEMP is a fake frontface register
+ */
+static boolean emit_frontface( struct svga_shader_emitter *emit )
+{
+   struct src_register vface, zero;
+   SVGA3dShaderDestToken temp;
+   struct src_register pass, fail;
+
+   vface = get_vface( emit );
+   zero = get_zero_immediate( emit );
+
+   /* Can't use get_temp() to allocate the fake frontface reg as such
+    * temporaries will be reclaimed after each instruction by the call
+    * to reset_temp_regs().
+    */
+   temp = dst_register( SVGA3DREG_TEMP,
+                        emit->nr_hw_temp++ );
+
+   if (emit->key.fkey.front_cw) {
+      pass = scalar( zero, TGSI_SWIZZLE_W );
+      fail = scalar( zero, TGSI_SWIZZLE_X );
+   } else {
+      pass = scalar( zero, TGSI_SWIZZLE_X );
+      fail = scalar( zero, TGSI_SWIZZLE_W );
+   }
+
+   if (!emit_conditional(emit, PIPE_FUNC_GREATER,
+                         temp, vface, scalar( zero, TGSI_SWIZZLE_X ),
+                         pass, fail))
+      return FALSE;
+
+   /* Reassign the input_map to the actual front-face color:
+    */
+   emit->input_map[emit->internal_frontface_idx] = src(temp);
+
+   return TRUE;
+}
+
+static INLINE boolean
+needs_to_create_zero( struct svga_shader_emitter *emit )
+{
+   int i;
+
+   if (emit->unit == PIPE_SHADER_FRAGMENT) {
+      if (!emit->use_sm30)
+         return TRUE;
+
+      if (emit->key.fkey.light_twoside)
+         return TRUE;
+
+      if (emit->emit_frontface)
+         return TRUE;
+
+      if (emit->info.opcode_count[TGSI_OPCODE_DST] >= 1 ||
+          emit->info.opcode_count[TGSI_OPCODE_LIT] >= 1)
+         return TRUE;
+   }
+
+   if (emit->info.opcode_count[TGSI_OPCODE_IF] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SGE] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SGT] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SLE] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SLT] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SNE] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_SEQ] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_EXP] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_LOG] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_XPD] >= 1 ||
+       emit->info.opcode_count[TGSI_OPCODE_KILP] >= 1)
+      return TRUE;
+
+   for (i = 0; i < emit->key.fkey.num_textures; i++) {
+      if (emit->key.fkey.tex[i].compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE)
+         return TRUE;
+   }
+
+   return FALSE;
+}
+
+static INLINE boolean
+needs_to_create_loop_const( struct svga_shader_emitter *emit )
+{
+   return (emit->info.opcode_count[TGSI_OPCODE_BGNLOOP] >= 1);
+}
+
+static INLINE boolean
+needs_to_create_sincos_consts( struct svga_shader_emitter *emit )
+{
+   return !emit->use_sm30 && (emit->info.opcode_count[TGSI_OPCODE_SIN] >= 1 ||
+                              emit->info.opcode_count[TGSI_OPCODE_COS] >= 1 ||
+                              emit->info.opcode_count[TGSI_OPCODE_SCS] >= 1);
+}
+
+static INLINE boolean
+needs_to_create_arl_consts( struct svga_shader_emitter *emit )
+{
+   return (emit->num_arl_consts > 0);
+}
+
+static INLINE boolean
+pre_parse_add_indirect( struct svga_shader_emitter *emit,
+                        int num, int current_arl)
+{
+   int i;
+   assert(num < 0);
+
+   for (i = 0; i < emit->num_arl_consts; ++i) {
+      if (emit->arl_consts[i].arl_num == current_arl)
+         break;
+   }
+   /* new entry */
+   if (emit->num_arl_consts == i) {
+      ++emit->num_arl_consts;
+   }
+   emit->arl_consts[i].number = (emit->arl_consts[i].number > num) ?
+                                num :
+                                emit->arl_consts[i].number;
+   emit->arl_consts[i].arl_num = current_arl;
+   return TRUE;
+}
+
+static boolean
+pre_parse_instruction( struct svga_shader_emitter *emit,
+                       const struct tgsi_full_instruction *insn,
+                       int current_arl)
+{
+   if (insn->FullSrcRegisters[0].SrcRegister.Indirect &&
+       insn->FullSrcRegisters[0].SrcRegisterInd.File == TGSI_FILE_ADDRESS) {
+      const struct tgsi_full_src_register *reg = &insn->FullSrcRegisters[0];
+      if (reg->SrcRegister.Index < 0) {
+         pre_parse_add_indirect(emit, reg->SrcRegister.Index, current_arl);
+      }
+   }
+
+   if (insn->FullSrcRegisters[1].SrcRegister.Indirect &&
+       insn->FullSrcRegisters[1].SrcRegisterInd.File == TGSI_FILE_ADDRESS) {
+      const struct tgsi_full_src_register *reg = &insn->FullSrcRegisters[1];
+      if (reg->SrcRegister.Index < 0) {
+         pre_parse_add_indirect(emit, reg->SrcRegister.Index, current_arl);
+      }
+   }
+
+   if (insn->FullSrcRegisters[2].SrcRegister.Indirect &&
+       insn->FullSrcRegisters[2].SrcRegisterInd.File == TGSI_FILE_ADDRESS) {
+      const struct tgsi_full_src_register *reg = &insn->FullSrcRegisters[2];
+      if (reg->SrcRegister.Index < 0) {
+         pre_parse_add_indirect(emit, reg->SrcRegister.Index, current_arl);
+      }
+   }
+
+   return TRUE;
+}
+
+static boolean
+pre_parse_tokens( struct svga_shader_emitter *emit,
+                  const struct tgsi_token *tokens )
+{
+   struct tgsi_parse_context parse;
+   int current_arl = 0;
+
+   tgsi_parse_init( &parse, tokens );
+
+   while (!tgsi_parse_end_of_tokens( &parse )) {
+      tgsi_parse_token( &parse );
+      switch (parse.FullToken.Token.Type) {
+      case TGSI_TOKEN_TYPE_IMMEDIATE:
+      case TGSI_TOKEN_TYPE_DECLARATION:
+         break;
+      case TGSI_TOKEN_TYPE_INSTRUCTION:
+         if (parse.FullToken.FullInstruction.Instruction.Opcode ==
+             TGSI_OPCODE_ARL) {
+            ++current_arl;
+         }
+         if (!pre_parse_instruction( emit, &parse.FullToken.FullInstruction,
+                                     current_arl ))
+            return FALSE;
+         break;
+      default:
+         break;
+      }
+
+   }
+   return TRUE;
+}
+
+static boolean svga_shader_emit_helpers( struct svga_shader_emitter *emit )
+
+{
+   if (needs_to_create_zero( emit )) {
+      create_zero_immediate( emit );
+   }
+   if (needs_to_create_loop_const( emit )) {
+      create_loop_const( emit );
+   }
+   if (needs_to_create_sincos_consts( emit )) {
+      create_sincos_consts( emit );
+   }
+   if (needs_to_create_arl_consts( emit )) {
+      create_arl_consts( emit );
+   }
+
+   if (emit->unit == PIPE_SHADER_FRAGMENT) {
+      if (!emit_ps_preamble( emit ))
+         return FALSE;
+
+      if (emit->key.fkey.light_twoside) {
+         if (!emit_light_twoside( emit ))
+            return FALSE;
+      }
+      if (emit->emit_frontface) {
+         if (!emit_frontface( emit ))
+            return FALSE;
+      }
+   }
+
+   return TRUE;
+}
+
+boolean svga_shader_emit_instructions( struct svga_shader_emitter *emit,
+                                       const struct tgsi_token *tokens )
+{
+   struct tgsi_parse_context parse;
+   boolean ret = TRUE;
+   boolean helpers_emitted = FALSE;
+   unsigned line_nr = 0;
+
+   tgsi_parse_init( &parse, tokens );
+   emit->internal_imm_count = 0;
+
+   if (emit->unit == PIPE_SHADER_VERTEX) {
+      ret = emit_vs_preamble( emit );
+      if (!ret)
+         goto done;
+   }
+
+   pre_parse_tokens(emit, tokens);
+
+   while (!tgsi_parse_end_of_tokens( &parse )) {
+      tgsi_parse_token( &parse );
+
+      switch (parse.FullToken.Token.Type) {
+      case TGSI_TOKEN_TYPE_IMMEDIATE:
+         ret = svga_emit_immediate( emit, &parse.FullToken.FullImmediate );
+         if (!ret)
+            goto done;
+         break;
+
+      case TGSI_TOKEN_TYPE_DECLARATION:
+         if (emit->use_sm30)
+            ret = svga_translate_decl_sm30( emit, &parse.FullToken.FullDeclaration );
+         else
+            ret = svga_translate_decl_sm20( emit, &parse.FullToken.FullDeclaration );
+         if (!ret)
+            goto done;
+         break;
+         
+      case TGSI_TOKEN_TYPE_INSTRUCTION:
+         if (!helpers_emitted) {
+            if (!svga_shader_emit_helpers( emit ))
+               goto done;
+            helpers_emitted = TRUE;
+         }
+         ret = svga_emit_instruction( emit, 
+                                      line_nr++,
+                                      &parse.FullToken.FullInstruction );
+         if (!ret)
+            goto done;
+         break;
+      default:
+         break;
+      }
+      
+      reset_temp_regs( emit );
+   }
+
+   /* Need to terminate the current subroutine.  Note that the
+    * hardware doesn't tolerate shaders without sub-routines
+    * terminating with RET+END.
+    */
+   if (!emit->in_main_func) {
+      ret = emit_instruction( emit, inst_token( SVGA3DOP_RET ) );
+      if (!ret)
+         goto done;
+   }
+
+   /* Need to terminate the whole shader:
+    */
+   ret = emit_instruction( emit, inst_token( SVGA3DOP_END ) );
+   if (!ret)
+      goto done;
+
+done:
+   assert(ret);
+   tgsi_parse_free( &parse );
+   return ret;
+}
+
diff --git a/src/gallium/drivers/svga/svga_winsys.h b/src/gallium/drivers/svga/svga_winsys.h
new file mode 100644
index 0000000000..59f299c185
--- /dev/null
+++ b/src/gallium/drivers/svga/svga_winsys.h
@@ -0,0 +1,299 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * VMware SVGA specific winsys interface.
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ * 
+ * Documentation taken from the VMware SVGA DDK.
+ */
+
+#ifndef SVGA_WINSYS_H_
+#define SVGA_WINSYS_H_
+
+
+#include "svga_types.h"
+#include "svga_reg.h"
+#include "svga3d_reg.h"
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_defines.h"
+
+
+struct svga_winsys_screen;
+struct svga_winsys_buffer;
+struct pipe_screen;
+struct pipe_context;
+struct pipe_fence_handle;
+struct pipe_texture;
+struct svga_region;
+
+
+#define SVGA_BUFFER_USAGE_PINNED  (PIPE_BUFFER_USAGE_CUSTOM << 0)
+#define SVGA_BUFFER_USAGE_WRAPPED (PIPE_BUFFER_USAGE_CUSTOM << 1)
+
+
+/** Opaque surface handle */
+struct svga_winsys_surface;
+
+/** Opaque buffer handle */
+struct svga_winsys_handle;
+
+
+/**
+ * SVGA per-context winsys interface.
+ */
+struct svga_winsys_context
+{
+   void
+   (*destroy)(struct svga_winsys_context *swc);
+
+   void *       
+   (*reserve)(struct svga_winsys_context *swc, 
+	      uint32_t nr_bytes, uint32_t nr_relocs );
+   
+   /**
+    * Emit a relocation for a host surface.
+    * 
+    * @param flags PIPE_BUFFER_USAGE_GPU_READ/WRITE
+    * 
+    * NOTE: Order of this call does matter. It should be the same order
+    * as relocations appear in the command buffer.
+    */
+   void
+   (*surface_relocation)(struct svga_winsys_context *swc, 
+	                 uint32 *sid, 
+	                 struct svga_winsys_surface *surface,
+	                 unsigned flags);
+   
+   /**
+    * Emit a relocation for a guest memory region.
+    * 
+    * @param flags PIPE_BUFFER_USAGE_GPU_READ/WRITE
+    * 
+    * NOTE: Order of this call does matter. It should be the same order
+    * as relocations appear in the command buffer.
+    */
+   void
+   (*region_relocation)(struct svga_winsys_context *swc, 
+	                struct SVGAGuestPtr *ptr, 
+	                struct svga_winsys_buffer *buffer,
+	                uint32 offset,
+                        unsigned flags);
+
+   void
+   (*commit)(struct svga_winsys_context *swc);
+   
+   enum pipe_error
+   (*flush)(struct svga_winsys_context *swc, 
+	    struct pipe_fence_handle **pfence);
+
+   /** 
+    * Context ID used to fill in the commands
+    * 
+    * Context IDs are arbitrary small non-negative integers,
+    * global to the entire SVGA device.
+    */
+   uint32 cid;
+};
+
+
+/**
+ * SVGA per-screen winsys interface.
+ */
+struct svga_winsys_screen
+{
+   void
+   (*destroy)(struct svga_winsys_screen *sws);
+   
+   boolean
+   (*get_cap)(struct svga_winsys_screen *sws,
+              SVGA3dDevCapIndex index,
+              SVGA3dDevCapResult *result);
+   
+   /**
+    * Create a new context.
+    *
+    * Context objects encapsulate all render state, and shader
+    * objects are per-context.
+    *
+    * Surfaces are not per-context. The same surface can be shared
+    * between multiple contexts, and surface operations can occur
+    * without a context.
+    */
+   struct svga_winsys_context *
+   (*context_create)(struct svga_winsys_screen *sws);
+   
+   
+   /**
+    * This creates a "surface" object in the SVGA3D device,
+    * and returns the surface ID (sid). Surfaces are generic
+    * containers for host VRAM objects like textures, vertex
+    * buffers, and depth/stencil buffers.
+    *
+    * Surfaces are hierarchial:
+    *
+    * - Surface may have multiple faces (for cube maps)
+    *
+    * - Each face has a list of mipmap levels
+    *
+    * - Each mipmap image may have multiple volume
+    *   slices, if the image is three dimensional.
+    *
+    * - Each slice is a 2D array of 'blocks'
+    *
+    * - Each block may be one or more pixels.
+    *   (Usually 1, more for DXT or YUV formats.)
+    *
+    * Surfaces are generic host VRAM objects. The SVGA3D device
+    * may optimize surfaces according to the format they were
+    * created with, but this format does not limit the ways in
+    * which the surface may be used. For example, a depth surface
+    * can be used as a texture, or a floating point image may
+    * be used as a vertex buffer. Some surface usages may be
+    * lower performance, due to software emulation, but any
+    * usage should work with any surface.
+    */
+   struct svga_winsys_surface *
+   (*surface_create)(struct svga_winsys_screen *sws,
+                     SVGA3dSurfaceFlags flags,
+                     SVGA3dSurfaceFormat format,
+                     SVGA3dSize size,
+                     uint32 numFaces,
+                     uint32 numMipLevels);
+
+   /**
+    * Whether this surface is sitting in a validate list
+    */
+   boolean
+   (*surface_is_flushed)(struct svga_winsys_screen *sws,
+                         struct svga_winsys_surface *surface);
+
+   /**
+    * Reference a SVGA3D surface object. This allows sharing of a
+    * surface between different objects.
+    */
+   void 
+   (*surface_reference)(struct svga_winsys_screen *sws,
+			struct svga_winsys_surface **pdst,
+			struct svga_winsys_surface *src);
+
+   /**
+    * Buffer management. Buffer attributes are mostly fixed over its lifetime.
+    *
+    * Remember that gallium gets to choose the interface it needs, and the
+    * window systems must then implement that interface (rather than the
+    * other way around...).
+    *
+    * usage is a bitmask of PIPE_BUFFER_USAGE_PIXEL/VERTEX/INDEX/CONSTANT. This
+    * usage argument is only an optimization hint, not a guarantee, therefore 
+    * proper behavior must be observed in all circumstances.
+    *
+    * alignment indicates the client's alignment requirements, eg for
+    * SSE instructions.
+    */
+   struct svga_winsys_buffer *
+   (*buffer_create)( struct svga_winsys_screen *sws, 
+	             unsigned alignment, 
+	             unsigned usage,
+	             unsigned size );
+
+   /** 
+    * Map the entire data store of a buffer object into the client's address.
+    * flags is a bitmask of:
+    * - PIPE_BUFFER_USAGE_CPU_READ/WRITE
+    * - PIPE_BUFFER_USAGE_DONTBLOCK
+    * - PIPE_BUFFER_USAGE_UNSYNCHRONIZED
+    */
+   void *
+   (*buffer_map)( struct svga_winsys_screen *sws, 
+	          struct svga_winsys_buffer *buf,
+		  unsigned usage );
+   
+   void 
+   (*buffer_unmap)( struct svga_winsys_screen *sws, 
+                    struct svga_winsys_buffer *buf );
+
+   void 
+   (*buffer_destroy)( struct svga_winsys_screen *sws,
+	              struct svga_winsys_buffer *buf );
+
+
+   /**
+    * Reference a fence object.
+    */
+   void
+   (*fence_reference)( struct svga_winsys_screen *sws,
+                       struct pipe_fence_handle **pdst,
+                       struct pipe_fence_handle *src );
+
+   /**
+    * Checks whether the fence has been signalled.
+    * \param flags  driver-specific meaning
+    * \return zero on success.
+    */
+   int (*fence_signalled)( struct svga_winsys_screen *sws,
+                           struct pipe_fence_handle *fence,
+                           unsigned flag );
+
+   /**
+    * Wait for the fence to finish.
+    * \param flags  driver-specific meaning
+    * \return zero on success.
+    */
+   int (*fence_finish)( struct svga_winsys_screen *sws,
+                        struct pipe_fence_handle *fence,
+                        unsigned flag );
+
+};
+
+
+struct pipe_context *
+svga_context_create(struct pipe_screen *screen);
+
+struct pipe_screen *
+svga_screen_create(struct svga_winsys_screen *sws);
+
+struct svga_winsys_screen *
+svga_winsys_screen(struct pipe_screen *screen);
+
+struct pipe_buffer *
+svga_screen_buffer_wrap_surface(struct pipe_screen *screen,
+				enum SVGA3dSurfaceFormat format,
+				struct svga_winsys_surface *srf);
+
+struct svga_winsys_surface *
+svga_screen_texture_get_winsys_surface(struct pipe_texture *texture);
+struct svga_winsys_surface *
+svga_screen_buffer_get_winsys_surface(struct pipe_buffer *buffer);
+
+boolean
+svga_screen_buffer_from_texture(struct pipe_texture *texture,
+				struct pipe_buffer **buffer,
+				unsigned *stride);
+
+#endif /* SVGA_WINSYS_H_ */
diff --git a/src/gallium/drivers/svga/svgadump/svga_dump.c b/src/gallium/drivers/svga/svgadump/svga_dump.c
new file mode 100644
index 0000000000..910afa2528
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_dump.c
@@ -0,0 +1,1736 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * Dump SVGA commands.
+ *
+ * Generated automatically from svga3d_reg.h by svga_dump.py.
+ */
+
+#include "svga_types.h"
+#include "svga_shader_dump.h"
+#include "svga3d_reg.h"
+
+#include "util/u_debug.h"
+#include "svga_dump.h"
+
+static void
+dump_SVGA3dVertexDecl(const SVGA3dVertexDecl *cmd)
+{
+   switch((*cmd).identity.type) {
+   case SVGA3D_DECLTYPE_FLOAT1:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT1\n");
+      break;
+   case SVGA3D_DECLTYPE_FLOAT2:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT2\n");
+      break;
+   case SVGA3D_DECLTYPE_FLOAT3:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT3\n");
+      break;
+   case SVGA3D_DECLTYPE_FLOAT4:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT4\n");
+      break;
+   case SVGA3D_DECLTYPE_D3DCOLOR:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_D3DCOLOR\n");
+      break;
+   case SVGA3D_DECLTYPE_UBYTE4:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4\n");
+      break;
+   case SVGA3D_DECLTYPE_SHORT2:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2\n");
+      break;
+   case SVGA3D_DECLTYPE_SHORT4:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4\n");
+      break;
+   case SVGA3D_DECLTYPE_UBYTE4N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UBYTE4N\n");
+      break;
+   case SVGA3D_DECLTYPE_SHORT2N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT2N\n");
+      break;
+   case SVGA3D_DECLTYPE_SHORT4N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_SHORT4N\n");
+      break;
+   case SVGA3D_DECLTYPE_USHORT2N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT2N\n");
+      break;
+   case SVGA3D_DECLTYPE_USHORT4N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_USHORT4N\n");
+      break;
+   case SVGA3D_DECLTYPE_UDEC3:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_UDEC3\n");
+      break;
+   case SVGA3D_DECLTYPE_DEC3N:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_DEC3N\n");
+      break;
+   case SVGA3D_DECLTYPE_FLOAT16_2:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_2\n");
+      break;
+   case SVGA3D_DECLTYPE_FLOAT16_4:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_FLOAT16_4\n");
+      break;
+   case SVGA3D_DECLTYPE_MAX:
+      debug_printf("\t\t.identity.type = SVGA3D_DECLTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.identity.type = %i\n", (*cmd).identity.type);
+      break;
+   }
+   switch((*cmd).identity.method) {
+   case SVGA3D_DECLMETHOD_DEFAULT:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_DEFAULT\n");
+      break;
+   case SVGA3D_DECLMETHOD_PARTIALU:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALU\n");
+      break;
+   case SVGA3D_DECLMETHOD_PARTIALV:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_PARTIALV\n");
+      break;
+   case SVGA3D_DECLMETHOD_CROSSUV:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_CROSSUV\n");
+      break;
+   case SVGA3D_DECLMETHOD_UV:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_UV\n");
+      break;
+   case SVGA3D_DECLMETHOD_LOOKUP:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUP\n");
+      break;
+   case SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED:
+      debug_printf("\t\t.identity.method = SVGA3D_DECLMETHOD_LOOKUPPRESAMPLED\n");
+      break;
+   default:
+      debug_printf("\t\t.identity.method = %i\n", (*cmd).identity.method);
+      break;
+   }
+   switch((*cmd).identity.usage) {
+   case SVGA3D_DECLUSAGE_POSITION:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITION\n");
+      break;
+   case SVGA3D_DECLUSAGE_BLENDWEIGHT:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDWEIGHT\n");
+      break;
+   case SVGA3D_DECLUSAGE_BLENDINDICES:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BLENDINDICES\n");
+      break;
+   case SVGA3D_DECLUSAGE_NORMAL:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_NORMAL\n");
+      break;
+   case SVGA3D_DECLUSAGE_PSIZE:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_PSIZE\n");
+      break;
+   case SVGA3D_DECLUSAGE_TEXCOORD:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TEXCOORD\n");
+      break;
+   case SVGA3D_DECLUSAGE_TANGENT:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TANGENT\n");
+      break;
+   case SVGA3D_DECLUSAGE_BINORMAL:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_BINORMAL\n");
+      break;
+   case SVGA3D_DECLUSAGE_TESSFACTOR:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_TESSFACTOR\n");
+      break;
+   case SVGA3D_DECLUSAGE_POSITIONT:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_POSITIONT\n");
+      break;
+   case SVGA3D_DECLUSAGE_COLOR:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_COLOR\n");
+      break;
+   case SVGA3D_DECLUSAGE_FOG:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_FOG\n");
+      break;
+   case SVGA3D_DECLUSAGE_DEPTH:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_DEPTH\n");
+      break;
+   case SVGA3D_DECLUSAGE_SAMPLE:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_SAMPLE\n");
+      break;
+   case SVGA3D_DECLUSAGE_MAX:
+      debug_printf("\t\t.identity.usage = SVGA3D_DECLUSAGE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.identity.usage = %i\n", (*cmd).identity.usage);
+      break;
+   }
+   debug_printf("\t\t.identity.usageIndex = %u\n", (*cmd).identity.usageIndex);
+   debug_printf("\t\t.array.surfaceId = %u\n", (*cmd).array.surfaceId);
+   debug_printf("\t\t.array.offset = %u\n", (*cmd).array.offset);
+   debug_printf("\t\t.array.stride = %u\n", (*cmd).array.stride);
+   debug_printf("\t\t.rangeHint.first = %u\n", (*cmd).rangeHint.first);
+   debug_printf("\t\t.rangeHint.last = %u\n", (*cmd).rangeHint.last);
+}
+
+static void
+dump_SVGA3dTextureState(const SVGA3dTextureState *cmd)
+{
+   debug_printf("\t\t.stage = %u\n", (*cmd).stage);
+   switch((*cmd).name) {
+   case SVGA3D_TS_INVALID:
+      debug_printf("\t\t.name = SVGA3D_TS_INVALID\n");
+      break;
+   case SVGA3D_TS_BIND_TEXTURE:
+      debug_printf("\t\t.name = SVGA3D_TS_BIND_TEXTURE\n");
+      break;
+   case SVGA3D_TS_COLOROP:
+      debug_printf("\t\t.name = SVGA3D_TS_COLOROP\n");
+      break;
+   case SVGA3D_TS_COLORARG1:
+      debug_printf("\t\t.name = SVGA3D_TS_COLORARG1\n");
+      break;
+   case SVGA3D_TS_COLORARG2:
+      debug_printf("\t\t.name = SVGA3D_TS_COLORARG2\n");
+      break;
+   case SVGA3D_TS_ALPHAOP:
+      debug_printf("\t\t.name = SVGA3D_TS_ALPHAOP\n");
+      break;
+   case SVGA3D_TS_ALPHAARG1:
+      debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG1\n");
+      break;
+   case SVGA3D_TS_ALPHAARG2:
+      debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG2\n");
+      break;
+   case SVGA3D_TS_ADDRESSU:
+      debug_printf("\t\t.name = SVGA3D_TS_ADDRESSU\n");
+      break;
+   case SVGA3D_TS_ADDRESSV:
+      debug_printf("\t\t.name = SVGA3D_TS_ADDRESSV\n");
+      break;
+   case SVGA3D_TS_MIPFILTER:
+      debug_printf("\t\t.name = SVGA3D_TS_MIPFILTER\n");
+      break;
+   case SVGA3D_TS_MAGFILTER:
+      debug_printf("\t\t.name = SVGA3D_TS_MAGFILTER\n");
+      break;
+   case SVGA3D_TS_MINFILTER:
+      debug_printf("\t\t.name = SVGA3D_TS_MINFILTER\n");
+      break;
+   case SVGA3D_TS_BORDERCOLOR:
+      debug_printf("\t\t.name = SVGA3D_TS_BORDERCOLOR\n");
+      break;
+   case SVGA3D_TS_TEXCOORDINDEX:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDINDEX\n");
+      break;
+   case SVGA3D_TS_TEXTURETRANSFORMFLAGS:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXTURETRANSFORMFLAGS\n");
+      break;
+   case SVGA3D_TS_TEXCOORDGEN:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXCOORDGEN\n");
+      break;
+   case SVGA3D_TS_BUMPENVMAT00:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT00\n");
+      break;
+   case SVGA3D_TS_BUMPENVMAT01:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT01\n");
+      break;
+   case SVGA3D_TS_BUMPENVMAT10:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT10\n");
+      break;
+   case SVGA3D_TS_BUMPENVMAT11:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVMAT11\n");
+      break;
+   case SVGA3D_TS_TEXTURE_MIPMAP_LEVEL:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_MIPMAP_LEVEL\n");
+      break;
+   case SVGA3D_TS_TEXTURE_LOD_BIAS:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_LOD_BIAS\n");
+      break;
+   case SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL:
+      debug_printf("\t\t.name = SVGA3D_TS_TEXTURE_ANISOTROPIC_LEVEL\n");
+      break;
+   case SVGA3D_TS_ADDRESSW:
+      debug_printf("\t\t.name = SVGA3D_TS_ADDRESSW\n");
+      break;
+   case SVGA3D_TS_GAMMA:
+      debug_printf("\t\t.name = SVGA3D_TS_GAMMA\n");
+      break;
+   case SVGA3D_TS_BUMPENVLSCALE:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLSCALE\n");
+      break;
+   case SVGA3D_TS_BUMPENVLOFFSET:
+      debug_printf("\t\t.name = SVGA3D_TS_BUMPENVLOFFSET\n");
+      break;
+   case SVGA3D_TS_COLORARG0:
+      debug_printf("\t\t.name = SVGA3D_TS_COLORARG0\n");
+      break;
+   case SVGA3D_TS_ALPHAARG0:
+      debug_printf("\t\t.name = SVGA3D_TS_ALPHAARG0\n");
+      break;
+   case SVGA3D_TS_MAX:
+      debug_printf("\t\t.name = SVGA3D_TS_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.name = %i\n", (*cmd).name);
+      break;
+   }
+   debug_printf("\t\t.value = %u\n", (*cmd).value);
+   debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue);
+}
+
+static void
+dump_SVGA3dCopyBox(const SVGA3dCopyBox *cmd)
+{
+   debug_printf("\t\t.x = %u\n", (*cmd).x);
+   debug_printf("\t\t.y = %u\n", (*cmd).y);
+   debug_printf("\t\t.z = %u\n", (*cmd).z);
+   debug_printf("\t\t.w = %u\n", (*cmd).w);
+   debug_printf("\t\t.h = %u\n", (*cmd).h);
+   debug_printf("\t\t.d = %u\n", (*cmd).d);
+   debug_printf("\t\t.srcx = %u\n", (*cmd).srcx);
+   debug_printf("\t\t.srcy = %u\n", (*cmd).srcy);
+   debug_printf("\t\t.srcz = %u\n", (*cmd).srcz);
+}
+
+static void
+dump_SVGA3dCmdSetClipPlane(const SVGA3dCmdSetClipPlane *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.index = %u\n", (*cmd).index);
+   debug_printf("\t\t.plane[0] = %f\n", (*cmd).plane[0]);
+   debug_printf("\t\t.plane[1] = %f\n", (*cmd).plane[1]);
+   debug_printf("\t\t.plane[2] = %f\n", (*cmd).plane[2]);
+   debug_printf("\t\t.plane[3] = %f\n", (*cmd).plane[3]);
+}
+
+static void
+dump_SVGA3dCmdWaitForQuery(const SVGA3dCmdWaitForQuery *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_QUERYTYPE_OCCLUSION:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n");
+      break;
+   case SVGA3D_QUERYTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+   debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId);
+   debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset);
+}
+
+static void
+dump_SVGA3dCmdSetRenderTarget(const SVGA3dCmdSetRenderTarget *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_RT_DEPTH:
+      debug_printf("\t\t.type = SVGA3D_RT_DEPTH\n");
+      break;
+   case SVGA3D_RT_STENCIL:
+      debug_printf("\t\t.type = SVGA3D_RT_STENCIL\n");
+      break;
+   default:
+      debug_printf("\t\t.type = SVGA3D_RT_COLOR%u\n", (*cmd).type - SVGA3D_RT_COLOR0);
+      break;
+   }
+   debug_printf("\t\t.target.sid = %u\n", (*cmd).target.sid);
+   debug_printf("\t\t.target.face = %u\n", (*cmd).target.face);
+   debug_printf("\t\t.target.mipmap = %u\n", (*cmd).target.mipmap);
+}
+
+static void
+dump_SVGA3dCmdSetTextureState(const SVGA3dCmdSetTextureState *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+}
+
+static void
+dump_SVGA3dCmdSurfaceCopy(const SVGA3dCmdSurfaceCopy *cmd)
+{
+   debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid);
+   debug_printf("\t\t.src.face = %u\n", (*cmd).src.face);
+   debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap);
+   debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid);
+   debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face);
+   debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap);
+}
+
+static void
+dump_SVGA3dCmdSetMaterial(const SVGA3dCmdSetMaterial *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).face) {
+   case SVGA3D_FACE_INVALID:
+      debug_printf("\t\t.face = SVGA3D_FACE_INVALID\n");
+      break;
+   case SVGA3D_FACE_NONE:
+      debug_printf("\t\t.face = SVGA3D_FACE_NONE\n");
+      break;
+   case SVGA3D_FACE_FRONT:
+      debug_printf("\t\t.face = SVGA3D_FACE_FRONT\n");
+      break;
+   case SVGA3D_FACE_BACK:
+      debug_printf("\t\t.face = SVGA3D_FACE_BACK\n");
+      break;
+   case SVGA3D_FACE_FRONT_BACK:
+      debug_printf("\t\t.face = SVGA3D_FACE_FRONT_BACK\n");
+      break;
+   case SVGA3D_FACE_MAX:
+      debug_printf("\t\t.face = SVGA3D_FACE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.face = %i\n", (*cmd).face);
+      break;
+   }
+   debug_printf("\t\t.material.diffuse[0] = %f\n", (*cmd).material.diffuse[0]);
+   debug_printf("\t\t.material.diffuse[1] = %f\n", (*cmd).material.diffuse[1]);
+   debug_printf("\t\t.material.diffuse[2] = %f\n", (*cmd).material.diffuse[2]);
+   debug_printf("\t\t.material.diffuse[3] = %f\n", (*cmd).material.diffuse[3]);
+   debug_printf("\t\t.material.ambient[0] = %f\n", (*cmd).material.ambient[0]);
+   debug_printf("\t\t.material.ambient[1] = %f\n", (*cmd).material.ambient[1]);
+   debug_printf("\t\t.material.ambient[2] = %f\n", (*cmd).material.ambient[2]);
+   debug_printf("\t\t.material.ambient[3] = %f\n", (*cmd).material.ambient[3]);
+   debug_printf("\t\t.material.specular[0] = %f\n", (*cmd).material.specular[0]);
+   debug_printf("\t\t.material.specular[1] = %f\n", (*cmd).material.specular[1]);
+   debug_printf("\t\t.material.specular[2] = %f\n", (*cmd).material.specular[2]);
+   debug_printf("\t\t.material.specular[3] = %f\n", (*cmd).material.specular[3]);
+   debug_printf("\t\t.material.emissive[0] = %f\n", (*cmd).material.emissive[0]);
+   debug_printf("\t\t.material.emissive[1] = %f\n", (*cmd).material.emissive[1]);
+   debug_printf("\t\t.material.emissive[2] = %f\n", (*cmd).material.emissive[2]);
+   debug_printf("\t\t.material.emissive[3] = %f\n", (*cmd).material.emissive[3]);
+   debug_printf("\t\t.material.shininess = %f\n", (*cmd).material.shininess);
+}
+
+static void
+dump_SVGA3dCmdSetLightData(const SVGA3dCmdSetLightData *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.index = %u\n", (*cmd).index);
+   switch((*cmd).data.type) {
+   case SVGA3D_LIGHTTYPE_INVALID:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_INVALID\n");
+      break;
+   case SVGA3D_LIGHTTYPE_POINT:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_POINT\n");
+      break;
+   case SVGA3D_LIGHTTYPE_SPOT1:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT1\n");
+      break;
+   case SVGA3D_LIGHTTYPE_SPOT2:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_SPOT2\n");
+      break;
+   case SVGA3D_LIGHTTYPE_DIRECTIONAL:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_DIRECTIONAL\n");
+      break;
+   case SVGA3D_LIGHTTYPE_MAX:
+      debug_printf("\t\t.data.type = SVGA3D_LIGHTTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.data.type = %i\n", (*cmd).data.type);
+      break;
+   }
+   debug_printf("\t\t.data.inWorldSpace = %u\n", (*cmd).data.inWorldSpace);
+   debug_printf("\t\t.data.diffuse[0] = %f\n", (*cmd).data.diffuse[0]);
+   debug_printf("\t\t.data.diffuse[1] = %f\n", (*cmd).data.diffuse[1]);
+   debug_printf("\t\t.data.diffuse[2] = %f\n", (*cmd).data.diffuse[2]);
+   debug_printf("\t\t.data.diffuse[3] = %f\n", (*cmd).data.diffuse[3]);
+   debug_printf("\t\t.data.specular[0] = %f\n", (*cmd).data.specular[0]);
+   debug_printf("\t\t.data.specular[1] = %f\n", (*cmd).data.specular[1]);
+   debug_printf("\t\t.data.specular[2] = %f\n", (*cmd).data.specular[2]);
+   debug_printf("\t\t.data.specular[3] = %f\n", (*cmd).data.specular[3]);
+   debug_printf("\t\t.data.ambient[0] = %f\n", (*cmd).data.ambient[0]);
+   debug_printf("\t\t.data.ambient[1] = %f\n", (*cmd).data.ambient[1]);
+   debug_printf("\t\t.data.ambient[2] = %f\n", (*cmd).data.ambient[2]);
+   debug_printf("\t\t.data.ambient[3] = %f\n", (*cmd).data.ambient[3]);
+   debug_printf("\t\t.data.position[0] = %f\n", (*cmd).data.position[0]);
+   debug_printf("\t\t.data.position[1] = %f\n", (*cmd).data.position[1]);
+   debug_printf("\t\t.data.position[2] = %f\n", (*cmd).data.position[2]);
+   debug_printf("\t\t.data.position[3] = %f\n", (*cmd).data.position[3]);
+   debug_printf("\t\t.data.direction[0] = %f\n", (*cmd).data.direction[0]);
+   debug_printf("\t\t.data.direction[1] = %f\n", (*cmd).data.direction[1]);
+   debug_printf("\t\t.data.direction[2] = %f\n", (*cmd).data.direction[2]);
+   debug_printf("\t\t.data.direction[3] = %f\n", (*cmd).data.direction[3]);
+   debug_printf("\t\t.data.range = %f\n", (*cmd).data.range);
+   debug_printf("\t\t.data.falloff = %f\n", (*cmd).data.falloff);
+   debug_printf("\t\t.data.attenuation0 = %f\n", (*cmd).data.attenuation0);
+   debug_printf("\t\t.data.attenuation1 = %f\n", (*cmd).data.attenuation1);
+   debug_printf("\t\t.data.attenuation2 = %f\n", (*cmd).data.attenuation2);
+   debug_printf("\t\t.data.theta = %f\n", (*cmd).data.theta);
+   debug_printf("\t\t.data.phi = %f\n", (*cmd).data.phi);
+}
+
+static void
+dump_SVGA3dCmdSetViewport(const SVGA3dCmdSetViewport *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x);
+   debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y);
+   debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w);
+   debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h);
+}
+
+static void
+dump_SVGA3dCmdSetScissorRect(const SVGA3dCmdSetScissorRect *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.rect.x = %u\n", (*cmd).rect.x);
+   debug_printf("\t\t.rect.y = %u\n", (*cmd).rect.y);
+   debug_printf("\t\t.rect.w = %u\n", (*cmd).rect.w);
+   debug_printf("\t\t.rect.h = %u\n", (*cmd).rect.h);
+}
+
+static void
+dump_SVGA3dCopyRect(const SVGA3dCopyRect *cmd)
+{
+   debug_printf("\t\t.x = %u\n", (*cmd).x);
+   debug_printf("\t\t.y = %u\n", (*cmd).y);
+   debug_printf("\t\t.w = %u\n", (*cmd).w);
+   debug_printf("\t\t.h = %u\n", (*cmd).h);
+   debug_printf("\t\t.srcx = %u\n", (*cmd).srcx);
+   debug_printf("\t\t.srcy = %u\n", (*cmd).srcy);
+}
+
+static void
+dump_SVGA3dCmdSetShader(const SVGA3dCmdSetShader *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_SHADERTYPE_COMPILED_DX8:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n");
+      break;
+   case SVGA3D_SHADERTYPE_VS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n");
+      break;
+   case SVGA3D_SHADERTYPE_PS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n");
+      break;
+   case SVGA3D_SHADERTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+   debug_printf("\t\t.shid = %u\n", (*cmd).shid);
+}
+
+static void
+dump_SVGA3dCmdEndQuery(const SVGA3dCmdEndQuery *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_QUERYTYPE_OCCLUSION:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n");
+      break;
+   case SVGA3D_QUERYTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+   debug_printf("\t\t.guestResult.gmrId = %u\n", (*cmd).guestResult.gmrId);
+   debug_printf("\t\t.guestResult.offset = %u\n", (*cmd).guestResult.offset);
+}
+
+static void
+dump_SVGA3dSize(const SVGA3dSize *cmd)
+{
+   debug_printf("\t\t.width = %u\n", (*cmd).width);
+   debug_printf("\t\t.height = %u\n", (*cmd).height);
+   debug_printf("\t\t.depth = %u\n", (*cmd).depth);
+}
+
+static void
+dump_SVGA3dCmdDestroySurface(const SVGA3dCmdDestroySurface *cmd)
+{
+   debug_printf("\t\t.sid = %u\n", (*cmd).sid);
+}
+
+static void
+dump_SVGA3dCmdDefineContext(const SVGA3dCmdDefineContext *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+}
+
+static void
+dump_SVGA3dRect(const SVGA3dRect *cmd)
+{
+   debug_printf("\t\t.x = %u\n", (*cmd).x);
+   debug_printf("\t\t.y = %u\n", (*cmd).y);
+   debug_printf("\t\t.w = %u\n", (*cmd).w);
+   debug_printf("\t\t.h = %u\n", (*cmd).h);
+}
+
+static void
+dump_SVGA3dCmdBeginQuery(const SVGA3dCmdBeginQuery *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_QUERYTYPE_OCCLUSION:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_OCCLUSION\n");
+      break;
+   case SVGA3D_QUERYTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_QUERYTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dRenderState(const SVGA3dRenderState *cmd)
+{
+   switch((*cmd).state) {
+   case SVGA3D_RS_INVALID:
+      debug_printf("\t\t.state = SVGA3D_RS_INVALID\n");
+      break;
+   case SVGA3D_RS_ZENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_ZENABLE\n");
+      break;
+   case SVGA3D_RS_ZWRITEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_ZWRITEENABLE\n");
+      break;
+   case SVGA3D_RS_ALPHATESTENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_ALPHATESTENABLE\n");
+      break;
+   case SVGA3D_RS_DITHERENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_DITHERENABLE\n");
+      break;
+   case SVGA3D_RS_BLENDENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_BLENDENABLE\n");
+      break;
+   case SVGA3D_RS_FOGENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGENABLE\n");
+      break;
+   case SVGA3D_RS_SPECULARENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_SPECULARENABLE\n");
+      break;
+   case SVGA3D_RS_STENCILENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE\n");
+      break;
+   case SVGA3D_RS_LIGHTINGENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_LIGHTINGENABLE\n");
+      break;
+   case SVGA3D_RS_NORMALIZENORMALS:
+      debug_printf("\t\t.state = SVGA3D_RS_NORMALIZENORMALS\n");
+      break;
+   case SVGA3D_RS_POINTSPRITEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSPRITEENABLE\n");
+      break;
+   case SVGA3D_RS_POINTSCALEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSCALEENABLE\n");
+      break;
+   case SVGA3D_RS_STENCILREF:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILREF\n");
+      break;
+   case SVGA3D_RS_STENCILMASK:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILMASK\n");
+      break;
+   case SVGA3D_RS_STENCILWRITEMASK:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILWRITEMASK\n");
+      break;
+   case SVGA3D_RS_FOGSTART:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGSTART\n");
+      break;
+   case SVGA3D_RS_FOGEND:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGEND\n");
+      break;
+   case SVGA3D_RS_FOGDENSITY:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGDENSITY\n");
+      break;
+   case SVGA3D_RS_POINTSIZE:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSIZE\n");
+      break;
+   case SVGA3D_RS_POINTSIZEMIN:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMIN\n");
+      break;
+   case SVGA3D_RS_POINTSIZEMAX:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSIZEMAX\n");
+      break;
+   case SVGA3D_RS_POINTSCALE_A:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_A\n");
+      break;
+   case SVGA3D_RS_POINTSCALE_B:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_B\n");
+      break;
+   case SVGA3D_RS_POINTSCALE_C:
+      debug_printf("\t\t.state = SVGA3D_RS_POINTSCALE_C\n");
+      break;
+   case SVGA3D_RS_FOGCOLOR:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGCOLOR\n");
+      break;
+   case SVGA3D_RS_AMBIENT:
+      debug_printf("\t\t.state = SVGA3D_RS_AMBIENT\n");
+      break;
+   case SVGA3D_RS_CLIPPLANEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_CLIPPLANEENABLE\n");
+      break;
+   case SVGA3D_RS_FOGMODE:
+      debug_printf("\t\t.state = SVGA3D_RS_FOGMODE\n");
+      break;
+   case SVGA3D_RS_FILLMODE:
+      debug_printf("\t\t.state = SVGA3D_RS_FILLMODE\n");
+      break;
+   case SVGA3D_RS_SHADEMODE:
+      debug_printf("\t\t.state = SVGA3D_RS_SHADEMODE\n");
+      break;
+   case SVGA3D_RS_LINEPATTERN:
+      debug_printf("\t\t.state = SVGA3D_RS_LINEPATTERN\n");
+      break;
+   case SVGA3D_RS_SRCBLEND:
+      debug_printf("\t\t.state = SVGA3D_RS_SRCBLEND\n");
+      break;
+   case SVGA3D_RS_DSTBLEND:
+      debug_printf("\t\t.state = SVGA3D_RS_DSTBLEND\n");
+      break;
+   case SVGA3D_RS_BLENDEQUATION:
+      debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATION\n");
+      break;
+   case SVGA3D_RS_CULLMODE:
+      debug_printf("\t\t.state = SVGA3D_RS_CULLMODE\n");
+      break;
+   case SVGA3D_RS_ZFUNC:
+      debug_printf("\t\t.state = SVGA3D_RS_ZFUNC\n");
+      break;
+   case SVGA3D_RS_ALPHAFUNC:
+      debug_printf("\t\t.state = SVGA3D_RS_ALPHAFUNC\n");
+      break;
+   case SVGA3D_RS_STENCILFUNC:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILFUNC\n");
+      break;
+   case SVGA3D_RS_STENCILFAIL:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILFAIL\n");
+      break;
+   case SVGA3D_RS_STENCILZFAIL:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILZFAIL\n");
+      break;
+   case SVGA3D_RS_STENCILPASS:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILPASS\n");
+      break;
+   case SVGA3D_RS_ALPHAREF:
+      debug_printf("\t\t.state = SVGA3D_RS_ALPHAREF\n");
+      break;
+   case SVGA3D_RS_FRONTWINDING:
+      debug_printf("\t\t.state = SVGA3D_RS_FRONTWINDING\n");
+      break;
+   case SVGA3D_RS_COORDINATETYPE:
+      debug_printf("\t\t.state = SVGA3D_RS_COORDINATETYPE\n");
+      break;
+   case SVGA3D_RS_ZBIAS:
+      debug_printf("\t\t.state = SVGA3D_RS_ZBIAS\n");
+      break;
+   case SVGA3D_RS_RANGEFOGENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_RANGEFOGENABLE\n");
+      break;
+   case SVGA3D_RS_COLORWRITEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE\n");
+      break;
+   case SVGA3D_RS_VERTEXMATERIALENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_VERTEXMATERIALENABLE\n");
+      break;
+   case SVGA3D_RS_DIFFUSEMATERIALSOURCE:
+      debug_printf("\t\t.state = SVGA3D_RS_DIFFUSEMATERIALSOURCE\n");
+      break;
+   case SVGA3D_RS_SPECULARMATERIALSOURCE:
+      debug_printf("\t\t.state = SVGA3D_RS_SPECULARMATERIALSOURCE\n");
+      break;
+   case SVGA3D_RS_AMBIENTMATERIALSOURCE:
+      debug_printf("\t\t.state = SVGA3D_RS_AMBIENTMATERIALSOURCE\n");
+      break;
+   case SVGA3D_RS_EMISSIVEMATERIALSOURCE:
+      debug_printf("\t\t.state = SVGA3D_RS_EMISSIVEMATERIALSOURCE\n");
+      break;
+   case SVGA3D_RS_TEXTUREFACTOR:
+      debug_printf("\t\t.state = SVGA3D_RS_TEXTUREFACTOR\n");
+      break;
+   case SVGA3D_RS_LOCALVIEWER:
+      debug_printf("\t\t.state = SVGA3D_RS_LOCALVIEWER\n");
+      break;
+   case SVGA3D_RS_SCISSORTESTENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_SCISSORTESTENABLE\n");
+      break;
+   case SVGA3D_RS_BLENDCOLOR:
+      debug_printf("\t\t.state = SVGA3D_RS_BLENDCOLOR\n");
+      break;
+   case SVGA3D_RS_STENCILENABLE2SIDED:
+      debug_printf("\t\t.state = SVGA3D_RS_STENCILENABLE2SIDED\n");
+      break;
+   case SVGA3D_RS_CCWSTENCILFUNC:
+      debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFUNC\n");
+      break;
+   case SVGA3D_RS_CCWSTENCILFAIL:
+      debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILFAIL\n");
+      break;
+   case SVGA3D_RS_CCWSTENCILZFAIL:
+      debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILZFAIL\n");
+      break;
+   case SVGA3D_RS_CCWSTENCILPASS:
+      debug_printf("\t\t.state = SVGA3D_RS_CCWSTENCILPASS\n");
+      break;
+   case SVGA3D_RS_VERTEXBLEND:
+      debug_printf("\t\t.state = SVGA3D_RS_VERTEXBLEND\n");
+      break;
+   case SVGA3D_RS_SLOPESCALEDEPTHBIAS:
+      debug_printf("\t\t.state = SVGA3D_RS_SLOPESCALEDEPTHBIAS\n");
+      break;
+   case SVGA3D_RS_DEPTHBIAS:
+      debug_printf("\t\t.state = SVGA3D_RS_DEPTHBIAS\n");
+      break;
+   case SVGA3D_RS_OUTPUTGAMMA:
+      debug_printf("\t\t.state = SVGA3D_RS_OUTPUTGAMMA\n");
+      break;
+   case SVGA3D_RS_ZVISIBLE:
+      debug_printf("\t\t.state = SVGA3D_RS_ZVISIBLE\n");
+      break;
+   case SVGA3D_RS_LASTPIXEL:
+      debug_printf("\t\t.state = SVGA3D_RS_LASTPIXEL\n");
+      break;
+   case SVGA3D_RS_CLIPPING:
+      debug_printf("\t\t.state = SVGA3D_RS_CLIPPING\n");
+      break;
+   case SVGA3D_RS_WRAP0:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP0\n");
+      break;
+   case SVGA3D_RS_WRAP1:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP1\n");
+      break;
+   case SVGA3D_RS_WRAP2:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP2\n");
+      break;
+   case SVGA3D_RS_WRAP3:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP3\n");
+      break;
+   case SVGA3D_RS_WRAP4:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP4\n");
+      break;
+   case SVGA3D_RS_WRAP5:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP5\n");
+      break;
+   case SVGA3D_RS_WRAP6:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP6\n");
+      break;
+   case SVGA3D_RS_WRAP7:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP7\n");
+      break;
+   case SVGA3D_RS_WRAP8:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP8\n");
+      break;
+   case SVGA3D_RS_WRAP9:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP9\n");
+      break;
+   case SVGA3D_RS_WRAP10:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP10\n");
+      break;
+   case SVGA3D_RS_WRAP11:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP11\n");
+      break;
+   case SVGA3D_RS_WRAP12:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP12\n");
+      break;
+   case SVGA3D_RS_WRAP13:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP13\n");
+      break;
+   case SVGA3D_RS_WRAP14:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP14\n");
+      break;
+   case SVGA3D_RS_WRAP15:
+      debug_printf("\t\t.state = SVGA3D_RS_WRAP15\n");
+      break;
+   case SVGA3D_RS_MULTISAMPLEANTIALIAS:
+      debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEANTIALIAS\n");
+      break;
+   case SVGA3D_RS_MULTISAMPLEMASK:
+      debug_printf("\t\t.state = SVGA3D_RS_MULTISAMPLEMASK\n");
+      break;
+   case SVGA3D_RS_INDEXEDVERTEXBLENDENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_INDEXEDVERTEXBLENDENABLE\n");
+      break;
+   case SVGA3D_RS_TWEENFACTOR:
+      debug_printf("\t\t.state = SVGA3D_RS_TWEENFACTOR\n");
+      break;
+   case SVGA3D_RS_ANTIALIASEDLINEENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_ANTIALIASEDLINEENABLE\n");
+      break;
+   case SVGA3D_RS_COLORWRITEENABLE1:
+      debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE1\n");
+      break;
+   case SVGA3D_RS_COLORWRITEENABLE2:
+      debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE2\n");
+      break;
+   case SVGA3D_RS_COLORWRITEENABLE3:
+      debug_printf("\t\t.state = SVGA3D_RS_COLORWRITEENABLE3\n");
+      break;
+   case SVGA3D_RS_SEPARATEALPHABLENDENABLE:
+      debug_printf("\t\t.state = SVGA3D_RS_SEPARATEALPHABLENDENABLE\n");
+      break;
+   case SVGA3D_RS_SRCBLENDALPHA:
+      debug_printf("\t\t.state = SVGA3D_RS_SRCBLENDALPHA\n");
+      break;
+   case SVGA3D_RS_DSTBLENDALPHA:
+      debug_printf("\t\t.state = SVGA3D_RS_DSTBLENDALPHA\n");
+      break;
+   case SVGA3D_RS_BLENDEQUATIONALPHA:
+      debug_printf("\t\t.state = SVGA3D_RS_BLENDEQUATIONALPHA\n");
+      break;
+   case SVGA3D_RS_MAX:
+      debug_printf("\t\t.state = SVGA3D_RS_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.state = %i\n", (*cmd).state);
+      break;
+   }
+   debug_printf("\t\t.uintValue = %u\n", (*cmd).uintValue);
+   debug_printf("\t\t.floatValue = %f\n", (*cmd).floatValue);
+}
+
+static void
+dump_SVGA3dVertexDivisor(const SVGA3dVertexDivisor *cmd)
+{
+   debug_printf("\t\t.value = %u\n", (*cmd).value);
+   debug_printf("\t\t.count = %u\n", (*cmd).count);
+   debug_printf("\t\t.indexedData = %u\n", (*cmd).indexedData);
+   debug_printf("\t\t.instanceData = %u\n", (*cmd).instanceData);
+}
+
+static void
+dump_SVGA3dCmdDefineShader(const SVGA3dCmdDefineShader *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.shid = %u\n", (*cmd).shid);
+   switch((*cmd).type) {
+   case SVGA3D_SHADERTYPE_COMPILED_DX8:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n");
+      break;
+   case SVGA3D_SHADERTYPE_VS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n");
+      break;
+   case SVGA3D_SHADERTYPE_PS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n");
+      break;
+   case SVGA3D_SHADERTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dCmdSetShaderConst(const SVGA3dCmdSetShaderConst *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.reg = %u\n", (*cmd).reg);
+   switch((*cmd).type) {
+   case SVGA3D_SHADERTYPE_COMPILED_DX8:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n");
+      break;
+   case SVGA3D_SHADERTYPE_VS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n");
+      break;
+   case SVGA3D_SHADERTYPE_PS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n");
+      break;
+   case SVGA3D_SHADERTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+   switch((*cmd).ctype) {
+   case SVGA3D_CONST_TYPE_FLOAT:
+      debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_FLOAT\n");
+      debug_printf("\t\t.values[0] = %f\n", *(const float *)&(*cmd).values[0]);
+      debug_printf("\t\t.values[1] = %f\n", *(const float *)&(*cmd).values[1]);
+      debug_printf("\t\t.values[2] = %f\n", *(const float *)&(*cmd).values[2]);
+      debug_printf("\t\t.values[3] = %f\n", *(const float *)&(*cmd).values[3]);
+      break;
+   case SVGA3D_CONST_TYPE_INT:
+      debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_INT\n");
+      debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]);
+      debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]);
+      debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]);
+      debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]);
+      break;
+   case SVGA3D_CONST_TYPE_BOOL:
+      debug_printf("\t\t.ctype = SVGA3D_CONST_TYPE_BOOL\n");
+      debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]);
+      debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]);
+      debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]);
+      debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]);
+      break;
+   default:
+      debug_printf("\t\t.ctype = %i\n", (*cmd).ctype);
+      debug_printf("\t\t.values[0] = %u\n", (*cmd).values[0]);
+      debug_printf("\t\t.values[1] = %u\n", (*cmd).values[1]);
+      debug_printf("\t\t.values[2] = %u\n", (*cmd).values[2]);
+      debug_printf("\t\t.values[3] = %u\n", (*cmd).values[3]);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dCmdSetZRange(const SVGA3dCmdSetZRange *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.zRange.min = %f\n", (*cmd).zRange.min);
+   debug_printf("\t\t.zRange.max = %f\n", (*cmd).zRange.max);
+}
+
+static void
+dump_SVGA3dCmdDrawPrimitives(const SVGA3dCmdDrawPrimitives *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.numVertexDecls = %u\n", (*cmd).numVertexDecls);
+   debug_printf("\t\t.numRanges = %u\n", (*cmd).numRanges);
+}
+
+static void
+dump_SVGA3dCmdSetLightEnabled(const SVGA3dCmdSetLightEnabled *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.index = %u\n", (*cmd).index);
+   debug_printf("\t\t.enabled = %u\n", (*cmd).enabled);
+}
+
+static void
+dump_SVGA3dPrimitiveRange(const SVGA3dPrimitiveRange *cmd)
+{
+   switch((*cmd).primType) {
+   case SVGA3D_PRIMITIVE_INVALID:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_INVALID\n");
+      break;
+   case SVGA3D_PRIMITIVE_TRIANGLELIST:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLELIST\n");
+      break;
+   case SVGA3D_PRIMITIVE_POINTLIST:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_POINTLIST\n");
+      break;
+   case SVGA3D_PRIMITIVE_LINELIST:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINELIST\n");
+      break;
+   case SVGA3D_PRIMITIVE_LINESTRIP:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_LINESTRIP\n");
+      break;
+   case SVGA3D_PRIMITIVE_TRIANGLESTRIP:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLESTRIP\n");
+      break;
+   case SVGA3D_PRIMITIVE_TRIANGLEFAN:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_TRIANGLEFAN\n");
+      break;
+   case SVGA3D_PRIMITIVE_MAX:
+      debug_printf("\t\t.primType = SVGA3D_PRIMITIVE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.primType = %i\n", (*cmd).primType);
+      break;
+   }
+   debug_printf("\t\t.primitiveCount = %u\n", (*cmd).primitiveCount);
+   debug_printf("\t\t.indexArray.surfaceId = %u\n", (*cmd).indexArray.surfaceId);
+   debug_printf("\t\t.indexArray.offset = %u\n", (*cmd).indexArray.offset);
+   debug_printf("\t\t.indexArray.stride = %u\n", (*cmd).indexArray.stride);
+   debug_printf("\t\t.indexWidth = %u\n", (*cmd).indexWidth);
+   debug_printf("\t\t.indexBias = %i\n", (*cmd).indexBias);
+}
+
+static void
+dump_SVGA3dCmdPresent(const SVGA3dCmdPresent *cmd)
+{
+   debug_printf("\t\t.sid = %u\n", (*cmd).sid);
+}
+
+static void
+dump_SVGA3dCmdSetRenderState(const SVGA3dCmdSetRenderState *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+}
+
+static void
+dump_SVGA3dCmdSurfaceStretchBlt(const SVGA3dCmdSurfaceStretchBlt *cmd)
+{
+   debug_printf("\t\t.src.sid = %u\n", (*cmd).src.sid);
+   debug_printf("\t\t.src.face = %u\n", (*cmd).src.face);
+   debug_printf("\t\t.src.mipmap = %u\n", (*cmd).src.mipmap);
+   debug_printf("\t\t.dest.sid = %u\n", (*cmd).dest.sid);
+   debug_printf("\t\t.dest.face = %u\n", (*cmd).dest.face);
+   debug_printf("\t\t.dest.mipmap = %u\n", (*cmd).dest.mipmap);
+   debug_printf("\t\t.boxSrc.x = %u\n", (*cmd).boxSrc.x);
+   debug_printf("\t\t.boxSrc.y = %u\n", (*cmd).boxSrc.y);
+   debug_printf("\t\t.boxSrc.z = %u\n", (*cmd).boxSrc.z);
+   debug_printf("\t\t.boxSrc.w = %u\n", (*cmd).boxSrc.w);
+   debug_printf("\t\t.boxSrc.h = %u\n", (*cmd).boxSrc.h);
+   debug_printf("\t\t.boxSrc.d = %u\n", (*cmd).boxSrc.d);
+   debug_printf("\t\t.boxDest.x = %u\n", (*cmd).boxDest.x);
+   debug_printf("\t\t.boxDest.y = %u\n", (*cmd).boxDest.y);
+   debug_printf("\t\t.boxDest.z = %u\n", (*cmd).boxDest.z);
+   debug_printf("\t\t.boxDest.w = %u\n", (*cmd).boxDest.w);
+   debug_printf("\t\t.boxDest.h = %u\n", (*cmd).boxDest.h);
+   debug_printf("\t\t.boxDest.d = %u\n", (*cmd).boxDest.d);
+   switch((*cmd).mode) {
+   case SVGA3D_STRETCH_BLT_POINT:
+      debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_POINT\n");
+      break;
+   case SVGA3D_STRETCH_BLT_LINEAR:
+      debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_LINEAR\n");
+      break;
+   case SVGA3D_STRETCH_BLT_MAX:
+      debug_printf("\t\t.mode = SVGA3D_STRETCH_BLT_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.mode = %i\n", (*cmd).mode);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dCmdSurfaceDMA(const SVGA3dCmdSurfaceDMA *cmd)
+{
+   debug_printf("\t\t.guest.ptr.gmrId = %u\n", (*cmd).guest.ptr.gmrId);
+   debug_printf("\t\t.guest.ptr.offset = %u\n", (*cmd).guest.ptr.offset);
+   debug_printf("\t\t.guest.pitch = %u\n", (*cmd).guest.pitch);
+   debug_printf("\t\t.host.sid = %u\n", (*cmd).host.sid);
+   debug_printf("\t\t.host.face = %u\n", (*cmd).host.face);
+   debug_printf("\t\t.host.mipmap = %u\n", (*cmd).host.mipmap);
+   switch((*cmd).transfer) {
+   case SVGA3D_WRITE_HOST_VRAM:
+      debug_printf("\t\t.transfer = SVGA3D_WRITE_HOST_VRAM\n");
+      break;
+   case SVGA3D_READ_HOST_VRAM:
+      debug_printf("\t\t.transfer = SVGA3D_READ_HOST_VRAM\n");
+      break;
+   default:
+      debug_printf("\t\t.transfer = %i\n", (*cmd).transfer);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dCmdSurfaceDMASuffix(const SVGA3dCmdSurfaceDMASuffix *cmd)
+{
+   debug_printf("\t\t.suffixSize = %u\n", (*cmd).suffixSize);
+   debug_printf("\t\t.maximumOffset = %u\n", (*cmd).maximumOffset);
+   debug_printf("\t\t.flags.discard = %u\n", (*cmd).flags.discard);
+   debug_printf("\t\t.flags.unsynchronized = %u\n", (*cmd).flags.unsynchronized);
+}
+
+static void
+dump_SVGA3dCmdSetTransform(const SVGA3dCmdSetTransform *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).type) {
+   case SVGA3D_TRANSFORM_INVALID:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_INVALID\n");
+      break;
+   case SVGA3D_TRANSFORM_WORLD:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD\n");
+      break;
+   case SVGA3D_TRANSFORM_VIEW:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_VIEW\n");
+      break;
+   case SVGA3D_TRANSFORM_PROJECTION:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_PROJECTION\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE0:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE0\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE1:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE1\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE2:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE2\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE3:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE3\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE4:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE4\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE5:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE5\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE6:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE6\n");
+      break;
+   case SVGA3D_TRANSFORM_TEXTURE7:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_TEXTURE7\n");
+      break;
+   case SVGA3D_TRANSFORM_WORLD1:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD1\n");
+      break;
+   case SVGA3D_TRANSFORM_WORLD2:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD2\n");
+      break;
+   case SVGA3D_TRANSFORM_WORLD3:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_WORLD3\n");
+      break;
+   case SVGA3D_TRANSFORM_MAX:
+      debug_printf("\t\t.type = SVGA3D_TRANSFORM_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+   debug_printf("\t\t.matrix[0] = %f\n", (*cmd).matrix[0]);
+   debug_printf("\t\t.matrix[1] = %f\n", (*cmd).matrix[1]);
+   debug_printf("\t\t.matrix[2] = %f\n", (*cmd).matrix[2]);
+   debug_printf("\t\t.matrix[3] = %f\n", (*cmd).matrix[3]);
+   debug_printf("\t\t.matrix[4] = %f\n", (*cmd).matrix[4]);
+   debug_printf("\t\t.matrix[5] = %f\n", (*cmd).matrix[5]);
+   debug_printf("\t\t.matrix[6] = %f\n", (*cmd).matrix[6]);
+   debug_printf("\t\t.matrix[7] = %f\n", (*cmd).matrix[7]);
+   debug_printf("\t\t.matrix[8] = %f\n", (*cmd).matrix[8]);
+   debug_printf("\t\t.matrix[9] = %f\n", (*cmd).matrix[9]);
+   debug_printf("\t\t.matrix[10] = %f\n", (*cmd).matrix[10]);
+   debug_printf("\t\t.matrix[11] = %f\n", (*cmd).matrix[11]);
+   debug_printf("\t\t.matrix[12] = %f\n", (*cmd).matrix[12]);
+   debug_printf("\t\t.matrix[13] = %f\n", (*cmd).matrix[13]);
+   debug_printf("\t\t.matrix[14] = %f\n", (*cmd).matrix[14]);
+   debug_printf("\t\t.matrix[15] = %f\n", (*cmd).matrix[15]);
+}
+
+static void
+dump_SVGA3dCmdDestroyShader(const SVGA3dCmdDestroyShader *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   debug_printf("\t\t.shid = %u\n", (*cmd).shid);
+   switch((*cmd).type) {
+   case SVGA3D_SHADERTYPE_COMPILED_DX8:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_COMPILED_DX8\n");
+      break;
+   case SVGA3D_SHADERTYPE_VS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_VS\n");
+      break;
+   case SVGA3D_SHADERTYPE_PS:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_PS\n");
+      break;
+   case SVGA3D_SHADERTYPE_MAX:
+      debug_printf("\t\t.type = SVGA3D_SHADERTYPE_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.type = %i\n", (*cmd).type);
+      break;
+   }
+}
+
+static void
+dump_SVGA3dCmdDestroyContext(const SVGA3dCmdDestroyContext *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+}
+
+static void
+dump_SVGA3dCmdClear(const SVGA3dCmdClear *cmd)
+{
+   debug_printf("\t\t.cid = %u\n", (*cmd).cid);
+   switch((*cmd).clearFlag) {
+   case SVGA3D_CLEAR_COLOR:
+      debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_COLOR\n");
+      break;
+   case SVGA3D_CLEAR_DEPTH:
+      debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_DEPTH\n");
+      break;
+   case SVGA3D_CLEAR_STENCIL:
+      debug_printf("\t\t.clearFlag = SVGA3D_CLEAR_STENCIL\n");
+      break;
+   default:
+      debug_printf("\t\t.clearFlag = %i\n", (*cmd).clearFlag);
+      break;
+   }
+   debug_printf("\t\t.color = %u\n", (*cmd).color);
+   debug_printf("\t\t.depth = %f\n", (*cmd).depth);
+   debug_printf("\t\t.stencil = %u\n", (*cmd).stencil);
+}
+
+static void
+dump_SVGA3dCmdDefineSurface(const SVGA3dCmdDefineSurface *cmd)
+{
+   debug_printf("\t\t.sid = %u\n", (*cmd).sid);
+   switch((*cmd).surfaceFlags) {
+   case SVGA3D_SURFACE_CUBEMAP:
+      debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_CUBEMAP\n");
+      break;
+   case SVGA3D_SURFACE_HINT_STATIC:
+      debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_STATIC\n");
+      break;
+   case SVGA3D_SURFACE_HINT_DYNAMIC:
+      debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_DYNAMIC\n");
+      break;
+   case SVGA3D_SURFACE_HINT_INDEXBUFFER:
+      debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_INDEXBUFFER\n");
+      break;
+   case SVGA3D_SURFACE_HINT_VERTEXBUFFER:
+      debug_printf("\t\t.surfaceFlags = SVGA3D_SURFACE_HINT_VERTEXBUFFER\n");
+      break;
+   default:
+      debug_printf("\t\t.surfaceFlags = %i\n", (*cmd).surfaceFlags);
+      break;
+   }
+   switch((*cmd).format) {
+   case SVGA3D_FORMAT_INVALID:
+      debug_printf("\t\t.format = SVGA3D_FORMAT_INVALID\n");
+      break;
+   case SVGA3D_X8R8G8B8:
+      debug_printf("\t\t.format = SVGA3D_X8R8G8B8\n");
+      break;
+   case SVGA3D_A8R8G8B8:
+      debug_printf("\t\t.format = SVGA3D_A8R8G8B8\n");
+      break;
+   case SVGA3D_R5G6B5:
+      debug_printf("\t\t.format = SVGA3D_R5G6B5\n");
+      break;
+   case SVGA3D_X1R5G5B5:
+      debug_printf("\t\t.format = SVGA3D_X1R5G5B5\n");
+      break;
+   case SVGA3D_A1R5G5B5:
+      debug_printf("\t\t.format = SVGA3D_A1R5G5B5\n");
+      break;
+   case SVGA3D_A4R4G4B4:
+      debug_printf("\t\t.format = SVGA3D_A4R4G4B4\n");
+      break;
+   case SVGA3D_Z_D32:
+      debug_printf("\t\t.format = SVGA3D_Z_D32\n");
+      break;
+   case SVGA3D_Z_D16:
+      debug_printf("\t\t.format = SVGA3D_Z_D16\n");
+      break;
+   case SVGA3D_Z_D24S8:
+      debug_printf("\t\t.format = SVGA3D_Z_D24S8\n");
+      break;
+   case SVGA3D_Z_D15S1:
+      debug_printf("\t\t.format = SVGA3D_Z_D15S1\n");
+      break;
+   case SVGA3D_LUMINANCE8:
+      debug_printf("\t\t.format = SVGA3D_LUMINANCE8\n");
+      break;
+   case SVGA3D_LUMINANCE4_ALPHA4:
+      debug_printf("\t\t.format = SVGA3D_LUMINANCE4_ALPHA4\n");
+      break;
+   case SVGA3D_LUMINANCE16:
+      debug_printf("\t\t.format = SVGA3D_LUMINANCE16\n");
+      break;
+   case SVGA3D_LUMINANCE8_ALPHA8:
+      debug_printf("\t\t.format = SVGA3D_LUMINANCE8_ALPHA8\n");
+      break;
+   case SVGA3D_DXT1:
+      debug_printf("\t\t.format = SVGA3D_DXT1\n");
+      break;
+   case SVGA3D_DXT2:
+      debug_printf("\t\t.format = SVGA3D_DXT2\n");
+      break;
+   case SVGA3D_DXT3:
+      debug_printf("\t\t.format = SVGA3D_DXT3\n");
+      break;
+   case SVGA3D_DXT4:
+      debug_printf("\t\t.format = SVGA3D_DXT4\n");
+      break;
+   case SVGA3D_DXT5:
+      debug_printf("\t\t.format = SVGA3D_DXT5\n");
+      break;
+   case SVGA3D_BUMPU8V8:
+      debug_printf("\t\t.format = SVGA3D_BUMPU8V8\n");
+      break;
+   case SVGA3D_BUMPL6V5U5:
+      debug_printf("\t\t.format = SVGA3D_BUMPL6V5U5\n");
+      break;
+   case SVGA3D_BUMPX8L8V8U8:
+      debug_printf("\t\t.format = SVGA3D_BUMPX8L8V8U8\n");
+      break;
+   case SVGA3D_BUMPL8V8U8:
+      debug_printf("\t\t.format = SVGA3D_BUMPL8V8U8\n");
+      break;
+   case SVGA3D_ARGB_S10E5:
+      debug_printf("\t\t.format = SVGA3D_ARGB_S10E5\n");
+      break;
+   case SVGA3D_ARGB_S23E8:
+      debug_printf("\t\t.format = SVGA3D_ARGB_S23E8\n");
+      break;
+   case SVGA3D_A2R10G10B10:
+      debug_printf("\t\t.format = SVGA3D_A2R10G10B10\n");
+      break;
+   case SVGA3D_V8U8:
+      debug_printf("\t\t.format = SVGA3D_V8U8\n");
+      break;
+   case SVGA3D_Q8W8V8U8:
+      debug_printf("\t\t.format = SVGA3D_Q8W8V8U8\n");
+      break;
+   case SVGA3D_CxV8U8:
+      debug_printf("\t\t.format = SVGA3D_CxV8U8\n");
+      break;
+   case SVGA3D_X8L8V8U8:
+      debug_printf("\t\t.format = SVGA3D_X8L8V8U8\n");
+      break;
+   case SVGA3D_A2W10V10U10:
+      debug_printf("\t\t.format = SVGA3D_A2W10V10U10\n");
+      break;
+   case SVGA3D_ALPHA8:
+      debug_printf("\t\t.format = SVGA3D_ALPHA8\n");
+      break;
+   case SVGA3D_R_S10E5:
+      debug_printf("\t\t.format = SVGA3D_R_S10E5\n");
+      break;
+   case SVGA3D_R_S23E8:
+      debug_printf("\t\t.format = SVGA3D_R_S23E8\n");
+      break;
+   case SVGA3D_RG_S10E5:
+      debug_printf("\t\t.format = SVGA3D_RG_S10E5\n");
+      break;
+   case SVGA3D_RG_S23E8:
+      debug_printf("\t\t.format = SVGA3D_RG_S23E8\n");
+      break;
+   case SVGA3D_BUFFER:
+      debug_printf("\t\t.format = SVGA3D_BUFFER\n");
+      break;
+   case SVGA3D_Z_D24X8:
+      debug_printf("\t\t.format = SVGA3D_Z_D24X8\n");
+      break;
+   case SVGA3D_FORMAT_MAX:
+      debug_printf("\t\t.format = SVGA3D_FORMAT_MAX\n");
+      break;
+   default:
+      debug_printf("\t\t.format = %i\n", (*cmd).format);
+      break;
+   }
+   debug_printf("\t\t.face[0].numMipLevels = %u\n", (*cmd).face[0].numMipLevels);
+   debug_printf("\t\t.face[1].numMipLevels = %u\n", (*cmd).face[1].numMipLevels);
+   debug_printf("\t\t.face[2].numMipLevels = %u\n", (*cmd).face[2].numMipLevels);
+   debug_printf("\t\t.face[3].numMipLevels = %u\n", (*cmd).face[3].numMipLevels);
+   debug_printf("\t\t.face[4].numMipLevels = %u\n", (*cmd).face[4].numMipLevels);
+   debug_printf("\t\t.face[5].numMipLevels = %u\n", (*cmd).face[5].numMipLevels);
+}
+
+
+void            
+svga_dump_commands(const void *commands, uint32_t size)
+{
+   const uint8_t *next = commands;
+   const uint8_t *last = next + size;
+   
+   assert(size % sizeof(uint32_t) == 0);
+   
+   while(next < last) {
+      const uint32_t cmd_id = *(const uint32_t *)next;
+
+      if(SVGA_3D_CMD_BASE <= cmd_id && cmd_id < SVGA_3D_CMD_MAX) {
+         const SVGA3dCmdHeader *header = (const SVGA3dCmdHeader *)next;
+         const uint8_t *body = (const uint8_t *)&header[1];
+
+         next = (const uint8_t *)body + header->size;
+         if(next > last)
+            break;
+
+         switch(cmd_id) {
+         case SVGA_3D_CMD_SURFACE_DEFINE:
+            debug_printf("\tSVGA_3D_CMD_SURFACE_DEFINE\n");
+            {
+               const SVGA3dCmdDefineSurface *cmd = (const SVGA3dCmdDefineSurface *)body;
+               dump_SVGA3dCmdDefineSurface(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dSize) <= next) {
+                  dump_SVGA3dSize((const SVGA3dSize *)body);
+                  body += sizeof(SVGA3dSize);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SURFACE_DESTROY:
+            debug_printf("\tSVGA_3D_CMD_SURFACE_DESTROY\n");
+            {
+               const SVGA3dCmdDestroySurface *cmd = (const SVGA3dCmdDestroySurface *)body;
+               dump_SVGA3dCmdDestroySurface(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SURFACE_COPY:
+            debug_printf("\tSVGA_3D_CMD_SURFACE_COPY\n");
+            {
+               const SVGA3dCmdSurfaceCopy *cmd = (const SVGA3dCmdSurfaceCopy *)body;
+               dump_SVGA3dCmdSurfaceCopy(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dCopyBox) <= next) {
+                  dump_SVGA3dCopyBox((const SVGA3dCopyBox *)body);
+                  body += sizeof(SVGA3dCopyBox);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SURFACE_STRETCHBLT:
+            debug_printf("\tSVGA_3D_CMD_SURFACE_STRETCHBLT\n");
+            {
+               const SVGA3dCmdSurfaceStretchBlt *cmd = (const SVGA3dCmdSurfaceStretchBlt *)body;
+               dump_SVGA3dCmdSurfaceStretchBlt(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SURFACE_DMA:
+            debug_printf("\tSVGA_3D_CMD_SURFACE_DMA\n");
+            {
+               const SVGA3dCmdSurfaceDMA *cmd = (const SVGA3dCmdSurfaceDMA *)body;
+               dump_SVGA3dCmdSurfaceDMA(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dCopyBox) <= next) {
+                  dump_SVGA3dCopyBox((const SVGA3dCopyBox *)body);
+                  body += sizeof(SVGA3dCopyBox);
+               }
+               while(body + sizeof(SVGA3dCmdSurfaceDMASuffix) <= next) {
+                  dump_SVGA3dCmdSurfaceDMASuffix((const SVGA3dCmdSurfaceDMASuffix *)body);
+                  body += sizeof(SVGA3dCmdSurfaceDMASuffix);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_CONTEXT_DEFINE:
+            debug_printf("\tSVGA_3D_CMD_CONTEXT_DEFINE\n");
+            {
+               const SVGA3dCmdDefineContext *cmd = (const SVGA3dCmdDefineContext *)body;
+               dump_SVGA3dCmdDefineContext(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_CONTEXT_DESTROY:
+            debug_printf("\tSVGA_3D_CMD_CONTEXT_DESTROY\n");
+            {
+               const SVGA3dCmdDestroyContext *cmd = (const SVGA3dCmdDestroyContext *)body;
+               dump_SVGA3dCmdDestroyContext(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETTRANSFORM:
+            debug_printf("\tSVGA_3D_CMD_SETTRANSFORM\n");
+            {
+               const SVGA3dCmdSetTransform *cmd = (const SVGA3dCmdSetTransform *)body;
+               dump_SVGA3dCmdSetTransform(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETZRANGE:
+            debug_printf("\tSVGA_3D_CMD_SETZRANGE\n");
+            {
+               const SVGA3dCmdSetZRange *cmd = (const SVGA3dCmdSetZRange *)body;
+               dump_SVGA3dCmdSetZRange(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETRENDERSTATE:
+            debug_printf("\tSVGA_3D_CMD_SETRENDERSTATE\n");
+            {
+               const SVGA3dCmdSetRenderState *cmd = (const SVGA3dCmdSetRenderState *)body;
+               dump_SVGA3dCmdSetRenderState(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dRenderState) <= next) {
+                  dump_SVGA3dRenderState((const SVGA3dRenderState *)body);
+                  body += sizeof(SVGA3dRenderState);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SETRENDERTARGET:
+            debug_printf("\tSVGA_3D_CMD_SETRENDERTARGET\n");
+            {
+               const SVGA3dCmdSetRenderTarget *cmd = (const SVGA3dCmdSetRenderTarget *)body;
+               dump_SVGA3dCmdSetRenderTarget(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETTEXTURESTATE:
+            debug_printf("\tSVGA_3D_CMD_SETTEXTURESTATE\n");
+            {
+               const SVGA3dCmdSetTextureState *cmd = (const SVGA3dCmdSetTextureState *)body;
+               dump_SVGA3dCmdSetTextureState(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dTextureState) <= next) {
+                  dump_SVGA3dTextureState((const SVGA3dTextureState *)body);
+                  body += sizeof(SVGA3dTextureState);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SETMATERIAL:
+            debug_printf("\tSVGA_3D_CMD_SETMATERIAL\n");
+            {
+               const SVGA3dCmdSetMaterial *cmd = (const SVGA3dCmdSetMaterial *)body;
+               dump_SVGA3dCmdSetMaterial(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETLIGHTDATA:
+            debug_printf("\tSVGA_3D_CMD_SETLIGHTDATA\n");
+            {
+               const SVGA3dCmdSetLightData *cmd = (const SVGA3dCmdSetLightData *)body;
+               dump_SVGA3dCmdSetLightData(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETLIGHTENABLED:
+            debug_printf("\tSVGA_3D_CMD_SETLIGHTENABLED\n");
+            {
+               const SVGA3dCmdSetLightEnabled *cmd = (const SVGA3dCmdSetLightEnabled *)body;
+               dump_SVGA3dCmdSetLightEnabled(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETVIEWPORT:
+            debug_printf("\tSVGA_3D_CMD_SETVIEWPORT\n");
+            {
+               const SVGA3dCmdSetViewport *cmd = (const SVGA3dCmdSetViewport *)body;
+               dump_SVGA3dCmdSetViewport(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SETCLIPPLANE:
+            debug_printf("\tSVGA_3D_CMD_SETCLIPPLANE\n");
+            {
+               const SVGA3dCmdSetClipPlane *cmd = (const SVGA3dCmdSetClipPlane *)body;
+               dump_SVGA3dCmdSetClipPlane(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_CLEAR:
+            debug_printf("\tSVGA_3D_CMD_CLEAR\n");
+            {
+               const SVGA3dCmdClear *cmd = (const SVGA3dCmdClear *)body;
+               dump_SVGA3dCmdClear(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dRect) <= next) {
+                  dump_SVGA3dRect((const SVGA3dRect *)body);
+                  body += sizeof(SVGA3dRect);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_PRESENT:
+            debug_printf("\tSVGA_3D_CMD_PRESENT\n");
+            {
+               const SVGA3dCmdPresent *cmd = (const SVGA3dCmdPresent *)body;
+               dump_SVGA3dCmdPresent(cmd);
+               body = (const uint8_t *)&cmd[1];
+               while(body + sizeof(SVGA3dCopyRect) <= next) {
+                  dump_SVGA3dCopyRect((const SVGA3dCopyRect *)body);
+                  body += sizeof(SVGA3dCopyRect);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SHADER_DEFINE:
+            debug_printf("\tSVGA_3D_CMD_SHADER_DEFINE\n");
+            {
+               const SVGA3dCmdDefineShader *cmd = (const SVGA3dCmdDefineShader *)body;
+               dump_SVGA3dCmdDefineShader(cmd);
+               body = (const uint8_t *)&cmd[1];
+               svga_shader_dump((const uint32_t *)body, 
+                            (unsigned)(next - body)/sizeof(uint32_t),
+                            FALSE );
+               body = next;
+            }
+            break;
+         case SVGA_3D_CMD_SHADER_DESTROY:
+            debug_printf("\tSVGA_3D_CMD_SHADER_DESTROY\n");
+            {
+               const SVGA3dCmdDestroyShader *cmd = (const SVGA3dCmdDestroyShader *)body;
+               dump_SVGA3dCmdDestroyShader(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SET_SHADER:
+            debug_printf("\tSVGA_3D_CMD_SET_SHADER\n");
+            {
+               const SVGA3dCmdSetShader *cmd = (const SVGA3dCmdSetShader *)body;
+               dump_SVGA3dCmdSetShader(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_SET_SHADER_CONST:
+            debug_printf("\tSVGA_3D_CMD_SET_SHADER_CONST\n");
+            {
+               const SVGA3dCmdSetShaderConst *cmd = (const SVGA3dCmdSetShaderConst *)body;
+               dump_SVGA3dCmdSetShaderConst(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_DRAW_PRIMITIVES:
+            debug_printf("\tSVGA_3D_CMD_DRAW_PRIMITIVES\n");
+            {
+               const SVGA3dCmdDrawPrimitives *cmd = (const SVGA3dCmdDrawPrimitives *)body;
+               unsigned i, j;
+               dump_SVGA3dCmdDrawPrimitives(cmd);
+               body = (const uint8_t *)&cmd[1];
+               for(i = 0; i < cmd->numVertexDecls; ++i) {
+                  dump_SVGA3dVertexDecl((const SVGA3dVertexDecl *)body);
+                  body += sizeof(SVGA3dVertexDecl);
+               }
+               for(j = 0; j < cmd->numRanges; ++j) {
+                  dump_SVGA3dPrimitiveRange((const SVGA3dPrimitiveRange *)body);
+                  body += sizeof(SVGA3dPrimitiveRange);
+               }
+               while(body + sizeof(SVGA3dVertexDivisor) <= next) {
+                  dump_SVGA3dVertexDivisor((const SVGA3dVertexDivisor *)body);
+                  body += sizeof(SVGA3dVertexDivisor);
+               }
+            }
+            break;
+         case SVGA_3D_CMD_SETSCISSORRECT:
+            debug_printf("\tSVGA_3D_CMD_SETSCISSORRECT\n");
+            {
+               const SVGA3dCmdSetScissorRect *cmd = (const SVGA3dCmdSetScissorRect *)body;
+               dump_SVGA3dCmdSetScissorRect(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_BEGIN_QUERY:
+            debug_printf("\tSVGA_3D_CMD_BEGIN_QUERY\n");
+            {
+               const SVGA3dCmdBeginQuery *cmd = (const SVGA3dCmdBeginQuery *)body;
+               dump_SVGA3dCmdBeginQuery(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_END_QUERY:
+            debug_printf("\tSVGA_3D_CMD_END_QUERY\n");
+            {
+               const SVGA3dCmdEndQuery *cmd = (const SVGA3dCmdEndQuery *)body;
+               dump_SVGA3dCmdEndQuery(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         case SVGA_3D_CMD_WAIT_FOR_QUERY:
+            debug_printf("\tSVGA_3D_CMD_WAIT_FOR_QUERY\n");
+            {
+               const SVGA3dCmdWaitForQuery *cmd = (const SVGA3dCmdWaitForQuery *)body;
+               dump_SVGA3dCmdWaitForQuery(cmd);
+               body = (const uint8_t *)&cmd[1];
+            }
+            break;
+         default:
+            debug_printf("\t0x%08x\n", cmd_id);
+            break;
+         }
+
+         while(body + sizeof(uint32_t) <= next) {
+            debug_printf("\t\t0x%08x\n", *(const uint32_t *)body);
+            body += sizeof(uint32_t);
+         }
+         while(body + sizeof(uint32_t) <= next)
+            debug_printf("\t\t0x%02x\n", *body++);
+      }
+      else if(cmd_id == SVGA_CMD_FENCE) {
+         debug_printf("\tSVGA_CMD_FENCE\n");
+         debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]);
+         next += 2*sizeof(uint32_t);
+      }
+      else {
+         debug_printf("\t0x%08x\n", cmd_id);
+         next += sizeof(uint32_t);
+      }
+   }
+}
+
diff --git a/src/gallium/drivers/svga/svgadump/svga_dump.h b/src/gallium/drivers/svga/svgadump/svga_dump.h
new file mode 100644
index 0000000000..69a8702087
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_dump.h
@@ -0,0 +1,34 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+#ifndef SVGA_DUMP_H_
+#define SVGA_DUMP_H_
+
+#include "pipe/p_compiler.h"
+
+void
+svga_dump_commands(const void *commands, uint32_t size);
+
+#endif /* SVGA_DUMP_H_ */
diff --git a/src/gallium/drivers/svga/svgadump/svga_dump.py b/src/gallium/drivers/svga/svgadump/svga_dump.py
new file mode 100755
index 0000000000..288e753296
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_dump.py
@@ -0,0 +1,329 @@
+#!/usr/bin/env python
+'''
+Generates dumper for the SVGA 3D command stream using pygccxml.
+
+Jose Fonseca <jfonseca@vmware.com>
+'''
+
+copyright = '''
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+ '''
+
+import os
+import sys
+
+from pygccxml import parser
+from pygccxml import declarations
+
+from pygccxml.declarations import algorithm
+from pygccxml.declarations import decl_visitor
+from pygccxml.declarations import type_traits
+from pygccxml.declarations import type_visitor
+
+
+enums = True
+
+
+class decl_dumper_t(decl_visitor.decl_visitor_t):
+
+    def __init__(self, instance = '', decl = None):
+        decl_visitor.decl_visitor_t.__init__(self)
+        self._instance = instance
+        self.decl = decl
+
+    def clone(self):
+        return decl_dumper_t(self._instance, self.decl)
+
+    def visit_class(self):
+        class_ = self.decl
+        assert self.decl.class_type in ('struct', 'union')
+
+        for variable in class_.variables():
+            if variable.name != '':
+                #print 'variable = %r' % variable.name
+                dump_type(self._instance + '.' + variable.name, variable.type)
+
+    def visit_enumeration(self):
+        if enums:
+            print '   switch(%s) {' % ("(*cmd)" + self._instance,)
+            for name, value in self.decl.values:
+                print '   case %s:' % (name,)
+                print '      debug_printf("\\t\\t%s = %s\\n");' % (self._instance, name)
+                print '      break;'
+            print '   default:'
+            print '      debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance)
+            print '      break;'
+            print '   }'
+        else:
+            print '   debug_printf("\\t\\t%s = %%i\\n", %s);' % (self._instance, "(*cmd)" + self._instance)
+
+
+def dump_decl(instance, decl):
+    dumper = decl_dumper_t(instance, decl)
+    algorithm.apply_visitor(dumper, decl)
+
+
+class type_dumper_t(type_visitor.type_visitor_t):
+
+    def __init__(self, instance, type_):
+        type_visitor.type_visitor_t.__init__(self)
+        self.instance = instance
+        self.type = type_
+
+    def clone(self):
+        return type_dumper_t(self.instance, self.type)
+
+    def visit_char(self):
+        self.print_instance('%i')
+        
+    def visit_unsigned_char(self):
+        self.print_instance('%u')
+
+    def visit_signed_char(self):
+        self.print_instance('%i')
+    
+    def visit_wchar(self):
+        self.print_instance('%i')
+        
+    def visit_short_int(self):
+        self.print_instance('%i')
+        
+    def visit_short_unsigned_int(self):
+        self.print_instance('%u')
+        
+    def visit_bool(self):
+        self.print_instance('%i')
+        
+    def visit_int(self):
+        self.print_instance('%i')
+        
+    def visit_unsigned_int(self):
+        self.print_instance('%u')
+        
+    def visit_long_int(self):
+        self.print_instance('%li')
+        
+    def visit_long_unsigned_int(self):
+        self.print_instance('%lu')
+        
+    def visit_long_long_int(self):
+        self.print_instance('%lli')
+        
+    def visit_long_long_unsigned_int(self):
+        self.print_instance('%llu')
+        
+    def visit_float(self):
+        self.print_instance('%f')
+        
+    def visit_double(self):
+        self.print_instance('%f')
+        
+    def visit_array(self):
+        for i in range(type_traits.array_size(self.type)):
+            dump_type(self.instance + '[%i]' % i, type_traits.base_type(self.type))
+
+    def visit_pointer(self):
+        self.print_instance('%p')
+
+    def visit_declarated(self):
+        #print 'decl = %r' % self.type.decl_string
+        decl = type_traits.remove_declarated(self.type)
+        dump_decl(self.instance, decl)
+
+    def print_instance(self, format):
+        print '   debug_printf("\\t\\t%s = %s\\n", %s);' % (self.instance, format, "(*cmd)" + self.instance)
+
+
+def dump_type(instance, type_):
+    type_ = type_traits.remove_alias(type_)
+    visitor = type_dumper_t(instance, type_)
+    algorithm.apply_visitor(visitor, type_)
+
+
+def dump_struct(decls, class_):
+    print 'static void'
+    print 'dump_%s(const %s *cmd)' % (class_.name, class_.name)
+    print '{'
+    dump_decl('', class_)
+    print '}'
+    print ''
+
+
+cmds = [
+    ('SVGA_3D_CMD_SURFACE_DEFINE', 'SVGA3dCmdDefineSurface', (), 'SVGA3dSize'),
+    ('SVGA_3D_CMD_SURFACE_DESTROY', 'SVGA3dCmdDestroySurface', (), None),
+    ('SVGA_3D_CMD_SURFACE_COPY', 'SVGA3dCmdSurfaceCopy', (), 'SVGA3dCopyBox'),
+    ('SVGA_3D_CMD_SURFACE_STRETCHBLT', 'SVGA3dCmdSurfaceStretchBlt', (), None),
+    ('SVGA_3D_CMD_SURFACE_DMA', 'SVGA3dCmdSurfaceDMA', (), 'SVGA3dCopyBox'),
+    ('SVGA_3D_CMD_CONTEXT_DEFINE', 'SVGA3dCmdDefineContext', (), None),
+    ('SVGA_3D_CMD_CONTEXT_DESTROY', 'SVGA3dCmdDestroyContext', (), None),
+    ('SVGA_3D_CMD_SETTRANSFORM', 'SVGA3dCmdSetTransform', (), None),
+    ('SVGA_3D_CMD_SETZRANGE', 'SVGA3dCmdSetZRange', (), None),
+    ('SVGA_3D_CMD_SETRENDERSTATE', 'SVGA3dCmdSetRenderState', (), 'SVGA3dRenderState'),
+    ('SVGA_3D_CMD_SETRENDERTARGET', 'SVGA3dCmdSetRenderTarget', (), None),
+    ('SVGA_3D_CMD_SETTEXTURESTATE', 'SVGA3dCmdSetTextureState', (), 'SVGA3dTextureState'),
+    ('SVGA_3D_CMD_SETMATERIAL', 'SVGA3dCmdSetMaterial', (), None),
+    ('SVGA_3D_CMD_SETLIGHTDATA', 'SVGA3dCmdSetLightData', (), None),
+    ('SVGA_3D_CMD_SETLIGHTENABLED', 'SVGA3dCmdSetLightEnabled', (), None),
+    ('SVGA_3D_CMD_SETVIEWPORT', 'SVGA3dCmdSetViewport', (), None),
+    ('SVGA_3D_CMD_SETCLIPPLANE', 'SVGA3dCmdSetClipPlane', (), None),
+    ('SVGA_3D_CMD_CLEAR', 'SVGA3dCmdClear', (), 'SVGA3dRect'),
+    ('SVGA_3D_CMD_PRESENT', 'SVGA3dCmdPresent', (), 'SVGA3dCopyRect'),
+    ('SVGA_3D_CMD_SHADER_DEFINE', 'SVGA3dCmdDefineShader', (), None),
+    ('SVGA_3D_CMD_SHADER_DESTROY', 'SVGA3dCmdDestroyShader', (), None),
+    ('SVGA_3D_CMD_SET_SHADER', 'SVGA3dCmdSetShader', (), None),
+    ('SVGA_3D_CMD_SET_SHADER_CONST', 'SVGA3dCmdSetShaderConst', (), None),
+    ('SVGA_3D_CMD_DRAW_PRIMITIVES', 'SVGA3dCmdDrawPrimitives', (('SVGA3dVertexDecl', 'numVertexDecls'), ('SVGA3dPrimitiveRange', 'numRanges')), 'SVGA3dVertexDivisor'),
+    ('SVGA_3D_CMD_SETSCISSORRECT', 'SVGA3dCmdSetScissorRect', (), None),
+    ('SVGA_3D_CMD_BEGIN_QUERY', 'SVGA3dCmdBeginQuery', (), None),
+    ('SVGA_3D_CMD_END_QUERY', 'SVGA3dCmdEndQuery', (), None),
+    ('SVGA_3D_CMD_WAIT_FOR_QUERY', 'SVGA3dCmdWaitForQuery', (), None),
+    #('SVGA_3D_CMD_PRESENT_READBACK', None, (), None),
+]
+
+def dump_cmds():
+    print r'''
+void            
+svga_dump_commands(const void *commands, uint32_t size)
+{
+   const uint8_t *next = commands;
+   const uint8_t *last = next + size;
+   
+   assert(size % sizeof(uint32_t) == 0);
+   
+   while(next < last) {
+      const uint32_t cmd_id = *(const uint32_t *)next;
+
+      if(SVGA_3D_CMD_BASE <= cmd_id && cmd_id < SVGA_3D_CMD_MAX) {
+         const SVGA3dCmdHeader *header = (const SVGA3dCmdHeader *)next;
+         const uint8_t *body = (const uint8_t *)&header[1];
+
+         next = (const uint8_t *)body + header->size;
+         if(next > last)
+            break;
+'''
+
+    print '         switch(cmd_id) {'
+    indexes = 'ijklmn'
+    for id, header, body, footer in cmds:
+        print '         case %s:' % id
+        print '            debug_printf("\\t%s\\n");' % id
+        print '            {'
+        print '               const %s *cmd = (const %s *)body;' % (header, header)
+        if len(body):
+            print '               unsigned ' + ', '.join(indexes[:len(body)]) + ';'
+        print '               dump_%s(cmd);' % header
+        print '               body = (const uint8_t *)&cmd[1];'
+        for i in range(len(body)):
+            struct, count = body[i]
+            idx = indexes[i]
+            print '               for(%s = 0; %s < cmd->%s; ++%s) {' % (idx, idx, count, idx)
+            print '                  dump_%s((const %s *)body);' % (struct, struct)
+            print '                  body += sizeof(%s);' % struct
+            print '               }'
+        if footer is not None:
+            print '               while(body + sizeof(%s) <= next) {' % footer
+            print '                  dump_%s((const %s *)body);' % (footer, footer)
+            print '                  body += sizeof(%s);' % footer
+            print '               }'
+        if id == 'SVGA_3D_CMD_SHADER_DEFINE':
+            print '               sh_svga_dump((const uint32_t *)body, (unsigned)(next - body)/sizeof(uint32_t));'
+            print '               body = next;'
+        print '            }'
+        print '            break;'
+    print '         default:'
+    print '            debug_printf("\\t0x%08x\\n", cmd_id);'
+    print '            break;'
+    print '         }'
+            
+    print r'''
+         while(body + sizeof(uint32_t) <= next) {
+            debug_printf("\t\t0x%08x\n", *(const uint32_t *)body);
+            body += sizeof(uint32_t);
+         }
+         while(body + sizeof(uint32_t) <= next)
+            debug_printf("\t\t0x%02x\n", *body++);
+      }
+      else if(cmd_id == SVGA_CMD_FENCE) {
+         debug_printf("\tSVGA_CMD_FENCE\n");
+         debug_printf("\t\t0x%08x\n", ((const uint32_t *)next)[1]);
+         next += 2*sizeof(uint32_t);
+      }
+      else {
+         debug_printf("\t0x%08x\n", cmd_id);
+         next += sizeof(uint32_t);
+      }
+   }
+}
+'''
+
+def main():
+    print copyright.strip()
+    print
+    print '/**'
+    print ' * @file'
+    print ' * Dump SVGA commands.'
+    print ' *'
+    print ' * Generated automatically from svga3d_reg.h by svga_dump.py.'
+    print ' */'
+    print
+    print '#include "svga_types.h"'
+    print '#include "svga_shader_dump.h"'
+    print '#include "svga3d_reg.h"'
+    print
+    print '#include "pipe/p_debug.h"'
+    print '#include "svga_dump.h"'
+    print
+
+    config = parser.config_t(
+        include_paths = ['include'],
+        compiler = 'gcc',
+    )
+
+    headers = [
+        'include/svga_types.h', 
+        'include/svga3d_reg.h', 
+    ]
+
+    decls = parser.parse(headers, config, parser.COMPILATION_MODE.ALL_AT_ONCE)
+    global_ns = declarations.get_global_namespace(decls)
+
+    names = set()
+    for id, header, body, footer in cmds:
+        names.add(header)
+        for struct, count in body:
+            names.add(struct)
+        if footer is not None:
+            names.add(footer)
+
+    for class_ in global_ns.classes(lambda decl: decl.name in names):
+        dump_struct(decls, class_)
+
+    dump_cmds()
+
+
+if __name__ == '__main__':
+    main()
diff --git a/src/gallium/drivers/svga/svgadump/svga_shader.h b/src/gallium/drivers/svga/svgadump/svga_shader.h
new file mode 100644
index 0000000000..9217af2dd9
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_shader.h
@@ -0,0 +1,220 @@
+/**********************************************************
+ * Copyright 2007-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA Shader Token Definitions
+ * 
+ * @author Michal Krol <michal@vmware.com>
+ */
+
+#ifndef ST_SHADER_SVGA_H
+#define ST_SHADER_SVGA_H
+
+#include "pipe/p_compiler.h"
+
+struct sh_op
+{
+   unsigned opcode:16;
+   unsigned control:8;
+   unsigned length:4;
+   unsigned predicated:1;
+   unsigned unused:1;
+   unsigned coissue:1;
+   unsigned is_reg:1;
+};
+
+struct sh_reg
+{
+   unsigned number:11;
+   unsigned type_hi:2;
+   unsigned relative:1;
+   unsigned unused:14;
+   unsigned type_lo:3;
+   unsigned is_reg:1;
+};
+
+static INLINE unsigned
+sh_reg_type( struct sh_reg reg )
+{
+   return reg.type_lo | (reg.type_hi << 3);
+}
+
+struct sh_cdata
+{
+   float xyzw[4];
+};
+
+struct sh_def
+{
+   struct sh_op op;
+   struct sh_reg reg;
+   struct sh_cdata cdata;
+};
+
+struct sh_defb
+{
+   struct sh_op op;
+   struct sh_reg reg;
+   uint data;
+};
+
+struct sh_idata
+{
+   int xyzw[4];
+};
+
+struct sh_defi
+{
+   struct sh_op op;
+   struct sh_reg reg;
+   struct sh_idata idata;
+};
+
+#define PS_TEXTURETYPE_UNKNOWN   SVGA3DSAMP_UNKNOWN
+#define PS_TEXTURETYPE_2D        SVGA3DSAMP_2D
+#define PS_TEXTURETYPE_CUBE      SVGA3DSAMP_CUBE
+#define PS_TEXTURETYPE_VOLUME    SVGA3DSAMP_VOLUME
+
+struct ps_sampleinfo
+{
+   unsigned unused:27;
+   unsigned texture_type:4;
+   unsigned is_reg:1;
+};
+
+struct vs_semantic
+{
+   unsigned usage:5;
+   unsigned unused1:11;
+   unsigned usage_index:4;
+   unsigned unused2:12;
+};
+
+struct sh_dstreg
+{
+   unsigned number:11;
+   unsigned type_hi:2;
+   unsigned relative:1;
+   unsigned unused:2;
+   unsigned write_mask:4;
+   unsigned modifier:4;
+   unsigned shift_scale:4;
+   unsigned type_lo:3;
+   unsigned is_reg:1;
+};
+
+static INLINE unsigned
+sh_dstreg_type( struct sh_dstreg reg )
+{
+   return reg.type_lo | (reg.type_hi << 3);
+}
+
+struct sh_dcl
+{
+   struct sh_op op;
+   union {
+      struct {
+         struct ps_sampleinfo sampleinfo;
+      } ps;
+      struct {
+         struct vs_semantic semantic;
+      } vs;
+   } u;
+   struct sh_dstreg reg;
+};
+
+
+struct sh_srcreg
+{
+   unsigned number:11;
+   unsigned type_hi:2;
+   unsigned relative:1;
+   unsigned unused:2;
+   unsigned swizzle_x:2;
+   unsigned swizzle_y:2;
+   unsigned swizzle_z:2;
+   unsigned swizzle_w:2;
+   unsigned modifier:4;
+   unsigned type_lo:3;
+   unsigned is_reg:1;
+};
+
+static INLINE unsigned
+sh_srcreg_type( struct sh_srcreg reg )
+{
+   return reg.type_lo | (reg.type_hi << 3);
+}
+
+struct sh_dstop
+{
+   struct sh_op op;
+   struct sh_dstreg dst;
+};
+
+struct sh_srcop
+{
+   struct sh_op op;
+   struct sh_srcreg src;
+};
+
+struct sh_src2op
+{
+   struct sh_op op;
+   struct sh_srcreg src0;
+   struct sh_srcreg src1;
+};
+
+struct sh_unaryop
+{
+   struct sh_op op;
+   struct sh_dstreg dst;
+   struct sh_srcreg src;
+};
+
+struct sh_binaryop
+{
+   struct sh_op op;
+   struct sh_dstreg dst;
+   struct sh_srcreg src0;
+   struct sh_srcreg src1;
+};
+
+struct sh_trinaryop
+{
+   struct sh_op op;
+   struct sh_dstreg dst;
+   struct sh_srcreg src0;
+   struct sh_srcreg src1;
+   struct sh_srcreg src2;
+};
+
+struct sh_comment
+{
+   unsigned opcode:16;
+   unsigned size:16;
+};
+
+#endif /* ST_SHADER_SVGA_H */
diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_dump.c b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c
new file mode 100644
index 0000000000..b0e7fdf378
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c
@@ -0,0 +1,654 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA Shader Dump Facilities
+ * 
+ * @author Michal Krol <michal@vmware.com>
+ */
+
+#include "svga_shader.h"
+#include "svga_shader_dump.h"
+#include "svga_shader_op.h"
+#include "util/u_debug.h"
+
+#include "../svga_hw_reg.h"
+#include "svga3d_shaderdefs.h"
+
+struct dump_info
+{
+   SVGA3dShaderVersion version;
+   boolean is_ps;
+};
+
+static void dump_op( struct sh_op op, const char *mnemonic )
+{
+   assert( op.predicated == 0 );
+   assert( op.is_reg == 0 );
+
+   if (op.coissue)
+      debug_printf( "+" );
+   debug_printf( "%s", mnemonic );
+   switch (op.control) {
+   case 0:
+      break;
+   case SVGA3DOPCONT_PROJECT:
+      debug_printf( "p" );
+      break;
+   case SVGA3DOPCONT_BIAS:
+      debug_printf( "b" );
+      break;
+   default:
+      assert( 0 );
+   }
+}
+
+
+static void dump_comp_op( struct sh_op op, const char *mnemonic )
+{
+   assert( op.is_reg == 0 );
+
+   if (op.coissue)
+      debug_printf( "+" );
+   debug_printf( "%s", mnemonic );
+   switch (op.control) {
+   case SVGA3DOPCOMP_RESERVED0:
+      break;
+   case SVGA3DOPCOMP_GT:
+      debug_printf("_gt");
+      break;
+   case SVGA3DOPCOMP_EQ:
+      debug_printf("_eq");
+      break;
+   case SVGA3DOPCOMP_GE:
+      debug_printf("_ge");
+      break;
+   case SVGA3DOPCOMP_LT:
+      debug_printf("_lt");
+      break;
+   case SVGA3DOPCOMPC_NE:
+      debug_printf("_ne");
+      break;
+   case SVGA3DOPCOMP_LE:
+      debug_printf("_le");
+      break;
+   case SVGA3DOPCOMP_RESERVED1:
+   default:
+      assert( 0 );
+   }
+}
+
+
+static void dump_reg( struct sh_reg reg, struct sh_srcreg *indreg, const struct dump_info *di )
+{
+   assert( sh_reg_type( reg ) == SVGA3DREG_CONST || reg.relative == 0 );
+   assert( reg.is_reg == 1 );
+
+   switch (sh_reg_type( reg )) {
+   case SVGA3DREG_TEMP:
+      debug_printf( "r%u", reg.number );
+      break;
+
+   case SVGA3DREG_INPUT:
+      debug_printf( "v%u", reg.number );
+      break;
+
+   case SVGA3DREG_CONST:
+      if (reg.relative) {
+         if (sh_srcreg_type( *indreg ) == SVGA3DREG_LOOP)
+            debug_printf( "c[aL+%u]", reg.number );
+         else
+            debug_printf( "c[a%u.x+%u]", indreg->number, reg.number );
+      }
+      else
+         debug_printf( "c%u", reg.number );
+      break;
+
+   case SVGA3DREG_ADDR:    /* VS */
+   /* SVGA3DREG_TEXTURE */ /* PS */
+      if (di->is_ps)
+         debug_printf( "t%u", reg.number );
+      else
+         debug_printf( "a%u", reg.number );
+      break;
+
+   case SVGA3DREG_RASTOUT:
+      switch (reg.number) {
+      case 0 /*POSITION*/:
+         debug_printf( "oPos" );
+         break;
+      case 1 /*FOG*/:
+         debug_printf( "oFog" );
+         break;
+      case 2 /*POINT_SIZE*/:
+         debug_printf( "oPts" );
+         break;
+      default:
+         assert( 0 );
+         debug_printf( "???" );
+      }
+      break;
+
+   case SVGA3DREG_ATTROUT:
+      assert( reg.number < 2 );
+      debug_printf( "oD%u", reg.number );
+      break;
+
+   case SVGA3DREG_TEXCRDOUT:
+   /* SVGA3DREG_OUTPUT */
+      debug_printf( "oT%u", reg.number );
+      break;
+
+   case SVGA3DREG_COLOROUT:
+      debug_printf( "oC%u", reg.number );
+      break;
+
+   case SVGA3DREG_DEPTHOUT:
+      debug_printf( "oD%u", reg.number );
+      break;
+
+   case SVGA3DREG_SAMPLER:
+      debug_printf( "s%u", reg.number );
+      break;
+
+   case SVGA3DREG_CONSTBOOL:
+      assert( !reg.relative );
+      debug_printf( "b%u", reg.number );
+      break;
+
+   case SVGA3DREG_CONSTINT:
+      assert( !reg.relative );
+      debug_printf( "i%u", reg.number );
+      break;
+
+   case SVGA3DREG_LOOP:
+      assert( reg.number == 0 );
+      debug_printf( "aL" );
+      break;
+
+   case SVGA3DREG_MISCTYPE:
+      switch (reg.number) {
+      case SVGA3DMISCREG_POSITION:
+         debug_printf( "vPos" );
+         break;
+      case SVGA3DMISCREG_FACE:
+         debug_printf( "vFace" );
+         break;
+      default:
+         assert(0);
+         break;
+      }
+      break;
+
+   case SVGA3DREG_LABEL:
+      debug_printf( "l%u", reg.number );
+      break;
+
+   case SVGA3DREG_PREDICATE:
+      debug_printf( "p%u", reg.number );
+      break;
+
+
+   default:
+      assert( 0 );
+      debug_printf( "???" );
+   }
+}
+
+static void dump_cdata( struct sh_cdata cdata )
+{
+   debug_printf( "%f, %f, %f, %f", cdata.xyzw[0], cdata.xyzw[1], cdata.xyzw[2], cdata.xyzw[3] );
+}
+
+static void dump_idata( struct sh_idata idata )
+{
+   debug_printf( "%d, %d, %d, %d", idata.xyzw[0], idata.xyzw[1], idata.xyzw[2], idata.xyzw[3] );
+}
+
+static void dump_bdata( boolean bdata )
+{
+   debug_printf( bdata ? "TRUE" : "FALSE" );
+}
+
+static void dump_sampleinfo( struct ps_sampleinfo sampleinfo )
+{
+   switch (sampleinfo.texture_type) {
+   case SVGA3DSAMP_2D:
+      debug_printf( "_2d" );
+      break;
+   case SVGA3DSAMP_CUBE:
+      debug_printf( "_cube" );
+      break;
+   case SVGA3DSAMP_VOLUME:
+      debug_printf( "_volume" );
+      break;
+   default:
+      assert( 0 );
+   }
+}
+
+
+static void dump_usageinfo( struct vs_semantic semantic )
+{
+   switch (semantic.usage) {
+   case SVGA3D_DECLUSAGE_POSITION:
+      debug_printf("_position" );
+      break;
+   case SVGA3D_DECLUSAGE_BLENDWEIGHT:
+      debug_printf("_blendweight" );
+      break;
+   case SVGA3D_DECLUSAGE_BLENDINDICES:
+      debug_printf("_blendindices" );
+      break;
+   case SVGA3D_DECLUSAGE_NORMAL:
+      debug_printf("_normal" );
+      break;
+   case SVGA3D_DECLUSAGE_PSIZE:
+      debug_printf("_psize" );
+      break;
+   case SVGA3D_DECLUSAGE_TEXCOORD:
+      debug_printf("_texcoord");
+      break;
+   case SVGA3D_DECLUSAGE_TANGENT:
+      debug_printf("_tangent" );
+      break;
+   case SVGA3D_DECLUSAGE_BINORMAL:
+      debug_printf("_binormal" );
+      break;
+   case SVGA3D_DECLUSAGE_TESSFACTOR:
+      debug_printf("_tessfactor" );
+      break;
+   case SVGA3D_DECLUSAGE_POSITIONT:
+      debug_printf("_positiont" );
+      break;
+   case SVGA3D_DECLUSAGE_COLOR:
+      debug_printf("_color" );
+      break;
+   case SVGA3D_DECLUSAGE_FOG:
+      debug_printf("_fog" );
+      break;
+   case SVGA3D_DECLUSAGE_DEPTH:
+      debug_printf("_depth" );
+      break;
+   case SVGA3D_DECLUSAGE_SAMPLE:
+      debug_printf("_sample");
+      break;
+   default:
+      assert( 0 );
+      return;
+   }
+
+   if (semantic.usage_index != 0) {
+      debug_printf("%d", semantic.usage_index );
+   }
+}
+
+static void dump_dstreg( struct sh_dstreg dstreg, const struct dump_info *di )
+{
+   union {
+      struct sh_reg reg;
+      struct sh_dstreg dstreg;
+   } u;
+
+   assert( (dstreg.modifier & (SVGA3DDSTMOD_SATURATE | SVGA3DDSTMOD_PARTIALPRECISION)) == dstreg.modifier );
+
+   if (dstreg.modifier & SVGA3DDSTMOD_SATURATE)
+      debug_printf( "_sat" );
+   if (dstreg.modifier & SVGA3DDSTMOD_PARTIALPRECISION)
+      debug_printf( "_pp" );
+   switch (dstreg.shift_scale) {
+   case 0:
+      break;
+   case 1:
+      debug_printf( "_x2" );
+      break;
+   case 2:
+      debug_printf( "_x4" );
+      break;
+   case 3:
+      debug_printf( "_x8" );
+      break;
+   case 13:
+      debug_printf( "_d8" );
+      break;
+   case 14:
+      debug_printf( "_d4" );
+      break;
+   case 15:
+      debug_printf( "_d2" );
+      break;
+   default:
+      assert( 0 );
+   }
+   debug_printf( " " );
+
+   u.dstreg = dstreg;
+   dump_reg( u.reg, NULL, di );
+   if (dstreg.write_mask != SVGA3DWRITEMASK_ALL) {
+      debug_printf( "." );
+      if (dstreg.write_mask & SVGA3DWRITEMASK_0)
+         debug_printf( "x" );
+      if (dstreg.write_mask & SVGA3DWRITEMASK_1)
+         debug_printf( "y" );
+      if (dstreg.write_mask & SVGA3DWRITEMASK_2)
+         debug_printf( "z" );
+      if (dstreg.write_mask & SVGA3DWRITEMASK_3)
+         debug_printf( "w" );
+   }
+}
+
+static void dump_srcreg( struct sh_srcreg srcreg, struct sh_srcreg *indreg, const struct dump_info *di )
+{
+   union {
+      struct sh_reg reg;
+      struct sh_srcreg srcreg;
+   } u;
+
+   switch (srcreg.modifier) {
+   case SVGA3DSRCMOD_NEG:
+   case SVGA3DSRCMOD_BIASNEG:
+   case SVGA3DSRCMOD_SIGNNEG:
+   case SVGA3DSRCMOD_X2NEG:
+      debug_printf( "-" );
+      break;
+   case SVGA3DSRCMOD_ABS:
+      debug_printf( "|" );
+      break;
+   case SVGA3DSRCMOD_ABSNEG:
+      debug_printf( "-|" );
+      break;
+   case SVGA3DSRCMOD_COMP:
+      debug_printf( "1-" );
+      break;
+   case SVGA3DSRCMOD_NOT:
+      debug_printf( "!" );
+   }
+
+   u.srcreg = srcreg;
+   dump_reg( u.reg, indreg, di );
+   switch (srcreg.modifier) {
+   case SVGA3DSRCMOD_NONE:
+   case SVGA3DSRCMOD_NEG:
+   case SVGA3DSRCMOD_COMP:
+   case SVGA3DSRCMOD_NOT:
+      break;
+   case SVGA3DSRCMOD_ABS:
+   case SVGA3DSRCMOD_ABSNEG:
+      debug_printf( "|" );
+      break;
+   case SVGA3DSRCMOD_BIAS:
+   case SVGA3DSRCMOD_BIASNEG:
+      debug_printf( "_bias" );
+      break;
+   case SVGA3DSRCMOD_SIGN:
+   case SVGA3DSRCMOD_SIGNNEG:
+      debug_printf( "_bx2" );
+      break;
+   case SVGA3DSRCMOD_X2:
+   case SVGA3DSRCMOD_X2NEG:
+      debug_printf( "_x2" );
+      break;
+   case SVGA3DSRCMOD_DZ:
+      debug_printf( "_dz" );
+      break;
+   case SVGA3DSRCMOD_DW:
+      debug_printf( "_dw" );
+      break;
+   default:
+      assert( 0 );
+   }
+   if (srcreg.swizzle_x != 0 || srcreg.swizzle_y != 1 || srcreg.swizzle_z != 2 || srcreg.swizzle_w != 3) {
+      debug_printf( "." );
+      if (srcreg.swizzle_x == srcreg.swizzle_y && srcreg.swizzle_y == srcreg.swizzle_z && srcreg.swizzle_z == srcreg.swizzle_w) {
+         debug_printf( "%c", "xyzw"[srcreg.swizzle_x] );
+      }
+      else {
+         debug_printf( "%c", "xyzw"[srcreg.swizzle_x] );
+         debug_printf( "%c", "xyzw"[srcreg.swizzle_y] );
+         debug_printf( "%c", "xyzw"[srcreg.swizzle_z] );
+         debug_printf( "%c", "xyzw"[srcreg.swizzle_w] );
+      }
+   }
+}
+
+void
+svga_shader_dump(
+   const unsigned *assem,
+   unsigned dwords,
+   unsigned do_binary )
+{
+   const unsigned *start = assem;
+   boolean finished = FALSE;
+   struct dump_info di;
+   unsigned i;
+
+   if (do_binary) {
+      for (i = 0; i < dwords; i++) 
+         debug_printf("  0x%08x,\n", assem[i]);
+      
+      debug_printf("\n\n");
+   }
+
+   di.version.value = *assem++;
+   di.is_ps = (di.version.type == SVGA3D_PS_TYPE);
+
+   debug_printf(
+      "%s_%u_%u\n",
+      di.is_ps ? "ps" : "vs",
+      di.version.major,
+      di.version.minor );
+
+   while (!finished) {
+      struct sh_op op = *(struct sh_op *) assem;
+
+      if (assem - start >= dwords) {
+         debug_printf("... ran off end of buffer\n");
+         assert(0);
+         return;
+      }
+
+      switch (op.opcode) {
+      case SVGA3DOP_DCL:
+         {
+            struct sh_dcl dcl = *(struct sh_dcl *) assem;
+
+            debug_printf( "dcl" );
+            if (sh_dstreg_type( dcl.reg ) == SVGA3DREG_SAMPLER)
+               dump_sampleinfo( dcl.u.ps.sampleinfo );
+            else if (di.is_ps) {
+               if (di.version.major == 3 && 
+                   sh_dstreg_type( dcl.reg ) != SVGA3DREG_MISCTYPE)
+                  dump_usageinfo( dcl.u.vs.semantic );
+            }
+            else
+               dump_usageinfo( dcl.u.vs.semantic );
+            dump_dstreg( dcl.reg, &di );
+            debug_printf( "\n" );
+            assem += sizeof( struct sh_dcl ) / sizeof( unsigned );
+         }
+         break;
+
+      case SVGA3DOP_DEFB:
+         {
+            struct sh_defb defb = *(struct sh_defb *) assem;
+
+            debug_printf( "defb " );
+            dump_reg( defb.reg, NULL, &di );
+            debug_printf( ", " );
+            dump_bdata( defb.data );
+            debug_printf( "\n" );
+            assem += sizeof( struct sh_defb ) / sizeof( unsigned );
+         }
+         break;
+
+      case SVGA3DOP_DEFI:
+         {
+            struct sh_defi defi = *(struct sh_defi *) assem;
+
+            debug_printf( "defi " );
+            dump_reg( defi.reg, NULL, &di );
+            debug_printf( ", " );
+            dump_idata( defi.idata );
+            debug_printf( "\n" );
+            assem += sizeof( struct sh_defi ) / sizeof( unsigned );
+         }
+         break;
+
+      case SVGA3DOP_TEXCOORD:
+         assert( di.is_ps );
+         dump_op( op, "texcoord" );
+         if (0) {
+            struct sh_dstop dstop = *(struct sh_dstop *) assem;
+            dump_dstreg( dstop.dst, &di );
+            assem += sizeof( struct sh_dstop ) / sizeof( unsigned );
+         }
+         else {
+            struct sh_unaryop unaryop = *(struct sh_unaryop *) assem;
+            dump_dstreg( unaryop.dst, &di );
+            debug_printf( ", " );
+            dump_srcreg( unaryop.src, NULL, &di );
+            assem += sizeof( struct sh_unaryop ) / sizeof( unsigned );
+         }
+         debug_printf( "\n" );
+         break;
+
+      case SVGA3DOP_TEX:
+         assert( di.is_ps );
+         if (0) {
+            dump_op( op, "tex" );
+            if (0) {
+               struct sh_dstop dstop = *(struct sh_dstop *) assem;
+
+               dump_dstreg( dstop.dst, &di );
+               assem += sizeof( struct sh_dstop ) / sizeof( unsigned );
+            }
+            else {
+               struct sh_unaryop unaryop = *(struct sh_unaryop *) assem;
+
+               dump_dstreg( unaryop.dst, &di );
+               debug_printf( ", " );
+               dump_srcreg( unaryop.src, NULL, &di );
+               assem += sizeof( struct sh_unaryop ) / sizeof( unsigned );
+            }
+         }
+         else {
+            struct sh_binaryop binaryop = *(struct sh_binaryop *) assem;
+
+            dump_op( op, "texld" );
+            dump_dstreg( binaryop.dst, &di );
+            debug_printf( ", " );
+            dump_srcreg( binaryop.src0, NULL, &di );
+            debug_printf( ", " );
+            dump_srcreg( binaryop.src1, NULL, &di );
+            assem += sizeof( struct sh_binaryop ) / sizeof( unsigned );
+         }
+         debug_printf( "\n" );
+         break;
+
+      case SVGA3DOP_DEF:
+         {
+            struct sh_def def = *(struct sh_def *) assem;
+
+            debug_printf( "def " );
+            dump_reg( def.reg, NULL, &di );
+            debug_printf( ", " );
+            dump_cdata( def.cdata );
+            debug_printf( "\n" );
+            assem += sizeof( struct sh_def ) / sizeof( unsigned );
+         }
+         break;
+
+      case SVGA3DOP_PHASE:
+         debug_printf( "phase\n" );
+         assem += sizeof( struct sh_op ) / sizeof( unsigned );
+         break;
+
+      case SVGA3DOP_COMMENT:
+         {
+            struct sh_comment comment = *(struct sh_comment *)assem;
+
+            /* Ignore comment contents. */
+            assem += sizeof(struct sh_comment) / sizeof(unsigned) + comment.size;
+         }
+         break;
+
+      case SVGA3DOP_RET:
+         debug_printf( "ret\n" );
+         assem += sizeof( struct sh_op ) / sizeof( unsigned );
+         break;
+
+      case SVGA3DOP_END:
+         debug_printf( "end\n" );
+         finished = TRUE;
+         break;
+
+      default:
+         {
+            const struct sh_opcode_info *info = svga_opcode_info( op.opcode );
+            uint i;
+            uint num_src = info->num_src + op.predicated;
+            boolean not_first_arg = FALSE;
+
+            assert( info->num_dst <= 1 );
+
+            if (op.opcode == SVGA3DOP_SINCOS && di.version.major < 3)
+               num_src += 2;
+
+            dump_comp_op( op, info->mnemonic );
+            assem += sizeof( struct sh_op ) / sizeof( unsigned );
+
+            if (info->num_dst > 0) {
+               struct sh_dstreg dstreg = *(struct sh_dstreg *) assem;
+
+               dump_dstreg( dstreg, &di );
+               assem += sizeof( struct sh_dstreg ) / sizeof( unsigned );
+               not_first_arg = TRUE;
+            }
+
+            for (i = 0; i < num_src; i++) {
+               struct sh_srcreg srcreg;
+               struct sh_srcreg indreg;
+
+               srcreg = *(struct sh_srcreg *) assem;
+               assem += sizeof( struct sh_srcreg ) / sizeof( unsigned );
+               if (srcreg.relative && !di.is_ps && di.version.major >= 2) {
+                  indreg = *(struct sh_srcreg *) assem;
+                  assem += sizeof( struct sh_srcreg ) / sizeof( unsigned );
+               }
+
+               if (not_first_arg)
+                  debug_printf( ", " );
+               else
+                  debug_printf( " " );
+               dump_srcreg( srcreg, &indreg, &di );
+               not_first_arg = TRUE;
+            }
+
+            debug_printf( "\n" );
+         }
+      }
+   }
+}
diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_dump.h b/src/gallium/drivers/svga/svgadump/svga_shader_dump.h
new file mode 100644
index 0000000000..a2657acb2f
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_shader_dump.h
@@ -0,0 +1,42 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA Shader Dump Facilities
+ * 
+ * @author Michal Krol <michal@vmware.com>
+ */
+
+#ifndef SVGA_SHADER_DUMP_H
+#define SVGA_SHADER_DUMP_H
+
+void
+svga_shader_dump(
+   const unsigned *assem,
+   unsigned dwords,
+   unsigned do_binary );
+
+#endif /* SVGA_SHADER_DUMP_H */
diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_op.c b/src/gallium/drivers/svga/svgadump/svga_shader_op.c
new file mode 100644
index 0000000000..8343bfdaab
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_shader_op.c
@@ -0,0 +1,168 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA Shader Token Opcode Info
+ * 
+ * @author Michal Krol <michal@vmware.com>
+ */
+
+#include "util/u_debug.h"
+#include "svga_shader_op.h"
+
+#include "../svga_hw_reg.h"
+#include "svga3d_shaderdefs.h"
+
+#define SVGA3DOP_INVALID SVGA3DOP_END
+#define TGSI_OPCODE_INVALID TGSI_OPCODE_LAST
+
+static struct sh_opcode_info opcode_info[] =
+{
+   { "nop",          0, 0, SVGA3DOP_NOP          },
+   { "mov",          1, 1, SVGA3DOP_MOV,         },
+   { "add",          1, 2, SVGA3DOP_ADD,         },
+   { "sub",          1, 2, SVGA3DOP_SUB,         },
+   { "mad",          1, 3, SVGA3DOP_MAD,         },
+   { "mul",          1, 2, SVGA3DOP_MUL,         },
+   { "rcp",          1, 1, SVGA3DOP_RCP,         },
+   { "rsq",          1, 1, SVGA3DOP_RSQ,         },
+   { "dp3",          1, 2, SVGA3DOP_DP3,         },
+   { "dp4",          1, 2, SVGA3DOP_DP4,         },
+   { "min",          1, 2, SVGA3DOP_MIN,         },
+   { "max",          1, 2, SVGA3DOP_MAX,         },
+   { "slt",          1, 2, SVGA3DOP_SLT,         },
+   { "sge",          1, 2, SVGA3DOP_SGE,         },
+   { "exp",          1, 1, SVGA3DOP_EXP,         },
+   { "log",          1, 1, SVGA3DOP_LOG,         },
+   { "lit",          1, 1, SVGA3DOP_LIT,         },
+   { "dst",          1, 2, SVGA3DOP_DST,         },
+   { "lrp",          1, 3, SVGA3DOP_LRP,         },
+   { "frc",          1, 1, SVGA3DOP_FRC,         },
+   { "m4x4",         1, 2, SVGA3DOP_M4x4,        },
+   { "m4x3",         1, 2, SVGA3DOP_M4x3,        },
+   { "m3x4",         1, 2, SVGA3DOP_M3x4,        },
+   { "m3x3",         1, 2, SVGA3DOP_M3x3,        },
+   { "m3x2",         1, 2, SVGA3DOP_M3x2,        },
+   { "call",         0, 1, SVGA3DOP_CALL,        },
+   { "callnz",       0, 2, SVGA3DOP_CALLNZ,      },
+   { "loop",         0, 2, SVGA3DOP_LOOP,        },
+   { "ret",          0, 0, SVGA3DOP_RET,         },
+   { "endloop",      0, 0, SVGA3DOP_ENDLOOP,     },
+   { "label",        0, 1, SVGA3DOP_LABEL,       },
+   { "dcl",          0, 0, SVGA3DOP_DCL,         },
+   { "pow",          1, 2, SVGA3DOP_POW,         },
+   { "crs",          1, 2, SVGA3DOP_CRS,         },
+   { "sgn",          1, 3, SVGA3DOP_SGN,         },
+   { "abs",          1, 1, SVGA3DOP_ABS,         },
+   { "nrm",          1, 1, SVGA3DOP_NRM,         }, /* 3-componenet normalization */
+   { "sincos",       1, 1, SVGA3DOP_SINCOS,      },
+   { "rep",          0, 1, SVGA3DOP_REP,         },
+   { "endrep",       0, 0, SVGA3DOP_ENDREP,      },
+   { "if",           0, 1, SVGA3DOP_IF,          },
+   { "ifc",          0, 2, SVGA3DOP_IFC,         },
+   { "else",         0, 0, SVGA3DOP_ELSE,        },
+   { "endif",        0, 0, SVGA3DOP_ENDIF,       },
+   { "break",        0, 0, SVGA3DOP_BREAK,       },
+   { "breakc",       0, 0, SVGA3DOP_BREAKC,      },
+   { "mova",         1, 1, SVGA3DOP_MOVA,        },
+   { "defb",         0, 0, SVGA3DOP_DEFB,        },
+   { "defi",         0, 0, SVGA3DOP_DEFI,        },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "???",          0, 0, SVGA3DOP_INVALID,     },
+   { "texcoord",     0, 0, SVGA3DOP_TEXCOORD,    },
+   { "texkill",      1, 0, SVGA3DOP_TEXKILL,     },
+   { "tex",          0, 0, SVGA3DOP_TEX,         },
+   { "texbem",       1, 1, SVGA3DOP_TEXBEM,      },
+   { "texbeml",      1, 1, SVGA3DOP_TEXBEML,     },
+   { "texreg2ar",    1, 1, SVGA3DOP_TEXREG2AR,   },
+   { "texreg2gb",    1, 1, SVGA3DOP_TEXREG2GB,   },
+   { "texm3x2pad",   1, 1, SVGA3DOP_TEXM3x2PAD,  },
+   { "texm3x2tex",   1, 1, SVGA3DOP_TEXM3x2TEX,  },
+   { "texm3x3pad",   1, 1, SVGA3DOP_TEXM3x3PAD,  },
+   { "texm3x3tex",   1, 1, SVGA3DOP_TEXM3x3TEX,  },
+   { "reserved0",    0, 0, SVGA3DOP_RESERVED0,   },
+   { "texm3x3spec",  1, 2, SVGA3DOP_TEXM3x3SPEC, },
+   { "texm3x3vspec", 1, 1, SVGA3DOP_TEXM3x3VSPEC,},
+   { "expp",         1, 1, SVGA3DOP_EXPP,        },
+   { "logp",         1, 1, SVGA3DOP_LOGP,        },
+   { "cnd",          1, 3, SVGA3DOP_CND,         },
+   { "def",          0, 0, SVGA3DOP_DEF,         },
+   { "texreg2rgb",   1, 1, SVGA3DOP_TEXREG2RGB,  },
+   { "texdp3tex",    1, 1, SVGA3DOP_TEXDP3TEX,   },
+   { "texm3x2depth", 1, 1, SVGA3DOP_TEXM3x2DEPTH,},
+   { "texdp3",       1, 1, SVGA3DOP_TEXDP3,      },
+   { "texm3x3",      1, 1, SVGA3DOP_TEXM3x3,     },
+   { "texdepth",     1, 0, SVGA3DOP_TEXDEPTH,    },
+   { "cmp",          1, 3, SVGA3DOP_CMP,         },
+   { "bem",          1, 2, SVGA3DOP_BEM,         },
+   { "dp2add",       1, 3, SVGA3DOP_DP2ADD,      },
+   { "dsx",          1, 1, SVGA3DOP_INVALID,     },
+   { "dsy",          1, 1, SVGA3DOP_INVALID,     },
+   { "texldd",       1, 1, SVGA3DOP_INVALID,     },
+   { "setp",         1, 2, SVGA3DOP_SETP,        },
+   { "texldl",       1, 1, SVGA3DOP_INVALID,     },
+   { "breakp",       1, 1, SVGA3DOP_INVALID,     },
+};
+
+const struct sh_opcode_info *svga_opcode_info( uint op )
+{
+   struct sh_opcode_info *info;
+
+   if (op >= sizeof( opcode_info ) / sizeof( opcode_info[0] )) {
+      /* The opcode is either PHASE, COMMENT, END or out of range.
+       */
+      assert( 0 );
+      return NULL;
+   }
+
+   info = &opcode_info[op];
+
+   if (info->svga_opcode == SVGA3DOP_INVALID) {
+      /* No valid information. Please provide number of dst/src registers.
+       */
+      assert( 0 );
+      return NULL;
+   }
+
+   /* Sanity check.
+    */
+   assert( op == info->svga_opcode );
+
+   return info;
+}
diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_op.h b/src/gallium/drivers/svga/svgadump/svga_shader_op.h
new file mode 100644
index 0000000000..e558de02c5
--- /dev/null
+++ b/src/gallium/drivers/svga/svgadump/svga_shader_op.h
@@ -0,0 +1,46 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA Shader Token Opcode Info
+ * 
+ * @author Michal Krol <michal@vmware.com>
+ */
+
+#ifndef SVGA_SHADER_OP_H
+#define SVGA_SHADER_OP_H
+
+struct sh_opcode_info
+{
+   const char *mnemonic;
+   unsigned num_dst:8;
+   unsigned num_src:8;
+   unsigned svga_opcode:16;
+};
+
+const struct sh_opcode_info *svga_opcode_info( unsigned op );
+
+#endif /* SVGA_SHADER_OP_H */
diff --git a/src/gallium/drivers/trace/tr_dump_state.c b/src/gallium/drivers/trace/tr_dump_state.c
index bcf6751af4..6d58209294 100644
--- a/src/gallium/drivers/trace/tr_dump_state.c
+++ b/src/gallium/drivers/trace/tr_dump_state.c
@@ -83,15 +83,15 @@ void trace_dump_template(const struct pipe_texture *templat)
    trace_dump_member(format, templat, format);
 
    trace_dump_member_begin("width");
-   trace_dump_array(uint, templat->width, 1);
+   trace_dump_uint(templat->width0);
    trace_dump_member_end();
 
    trace_dump_member_begin("height");
-   trace_dump_array(uint, templat->height, 1);
+   trace_dump_uint(templat->height0);
    trace_dump_member_end();
 
    trace_dump_member_begin("depth");
-   trace_dump_array(uint, templat->depth, 1);
+   trace_dump_uint(templat->depth0);
    trace_dump_member_end();
 
    trace_dump_member_begin("block");
diff --git a/src/gallium/drivers/trace/tr_rbug.c b/src/gallium/drivers/trace/tr_rbug.c
index 81e0a6f3b0..b59458c0e3 100644
--- a/src/gallium/drivers/trace/tr_rbug.c
+++ b/src/gallium/drivers/trace/tr_rbug.c
@@ -200,9 +200,9 @@ trace_rbug_texture_info(struct trace_rbug *tr_rbug, struct rbug_header *header,
    t = tr_tex->texture;
    rbug_send_texture_info_reply(tr_rbug->con, serial,
                                t->target, t->format,
-                               t->width, t->last_level + 1,
-                               t->height, t->last_level + 1,
-                               t->depth, t->last_level + 1,
+                               &t->width0, 1,
+                               &t->height0, 1,
+                               &t->depth0, 1,
                                t->block.width, t->block.height, t->block.size,
                                t->last_level,
                                t->nr_samples,
diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h
index c4e05a5526..8cf3942cc7 100644
--- a/src/gallium/include/pipe/p_shader_tokens.h
+++ b/src/gallium/include/pipe/p_shader_tokens.h
@@ -295,6 +295,8 @@ union tgsi_immediate_data
  * respectively. For a given operation code, those numbers are fixed and are
  * present here only for convenience.
  *
+ * If Predicate is TRUE, tgsi_instruction_predicate token immediately follows.
+ *
  * If Extended is TRUE, it is now executed.
  *
  * Saturate controls how are final results in destination registers modified.
@@ -308,7 +310,8 @@ struct tgsi_instruction
    unsigned Saturate   : 2;  /* TGSI_SAT_ */
    unsigned NumDstRegs : 2;  /* UINT */
    unsigned NumSrcRegs : 4;  /* UINT */
-   unsigned Padding    : 3;
+   unsigned Predicate  : 1;  /* BOOL */
+   unsigned Padding    : 2;
    unsigned Extended   : 1;  /* BOOL */
 };
 
@@ -325,7 +328,6 @@ struct tgsi_instruction
 
 #define TGSI_INSTRUCTION_EXT_TYPE_LABEL     1
 #define TGSI_INSTRUCTION_EXT_TYPE_TEXTURE   2
-#define TGSI_INSTRUCTION_EXT_TYPE_PREDICATE 3
 
 struct tgsi_instruction_ext
 {
@@ -341,9 +343,6 @@ struct tgsi_instruction_ext
  * If tgsi_instruction_ext::Type is TGSI_INSTRUCTION_EXT_TYPE_TEXTURE, it
  * should be cast to tgsi_instruction_ext_texture.
  * 
- * If tgsi_instruction_ext::Type is TGSI_INSTRUCTION_EXT_TYPE_PREDICATE, it
- * should be cast to tgsi_instruction_ext_predicate.
- * 
  * If tgsi_instruction_ext::Extended is TRUE, another tgsi_instruction_ext
  * follows.
  */
@@ -384,17 +383,15 @@ struct tgsi_instruction_ext_texture
  * For SM3, the following constraint applies.
  *   - Swizzle is either set to identity or replicate.
  */
-struct tgsi_instruction_ext_predicate
+struct tgsi_instruction_predicate
 {
-   unsigned Type     : 4;  /* TGSI_INSTRUCTION_EXT_TYPE_PREDICATE */
+   int      Index    : 16; /* SINT */
    unsigned SwizzleX : 2;  /* TGSI_SWIZZLE_x */
    unsigned SwizzleY : 2;  /* TGSI_SWIZZLE_x */
    unsigned SwizzleZ : 2;  /* TGSI_SWIZZLE_x */
    unsigned SwizzleW : 2;  /* TGSI_SWIZZLE_x */
    unsigned Negate   : 1;  /* BOOL */
-   unsigned SrcIndex : 8;  /* UINT */
-   unsigned Padding  : 10;
-   unsigned Extended : 1;  /* BOOL */
+   unsigned Padding  : 7;
 };
 
 /**
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index 3c9f0f076f..12433ad4c2 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -350,9 +350,9 @@ struct pipe_texture
    enum pipe_texture_target target; /**< PIPE_TEXTURE_x */
    enum pipe_format format;         /**< PIPE_FORMAT_x */
 
-   unsigned width[PIPE_MAX_TEXTURE_LEVELS];
-   unsigned height[PIPE_MAX_TEXTURE_LEVELS];
-   unsigned depth[PIPE_MAX_TEXTURE_LEVELS];
+   unsigned width0;
+   unsigned height0;
+   unsigned depth0;
 
    struct pipe_format_block block;
    unsigned nblocksx[PIPE_MAX_TEXTURE_LEVELS]; /**< allocated width in blocks */
diff --git a/src/gallium/state_trackers/dri/dri_drawable.c b/src/gallium/state_trackers/dri/dri_drawable.c
index 5625ff53cf..45a6059ea8 100644
--- a/src/gallium/state_trackers/dri/dri_drawable.c
+++ b/src/gallium/state_trackers/dri/dri_drawable.c
@@ -46,7 +46,7 @@
 
 #include "util/u_memory.h"
 #include "util/u_rect.h"
-
+ 
 static struct pipe_surface *
 dri_surface_from_handle(struct drm_api *api,
 			struct pipe_screen *screen,
@@ -62,10 +62,10 @@ dri_surface_from_handle(struct drm_api *api,
    templat.tex_usage |= PIPE_TEXTURE_USAGE_RENDER_TARGET;
    templat.target = PIPE_TEXTURE_2D;
    templat.last_level = 0;
-   templat.depth[0] = 1;
+   templat.depth0 = 1;
    templat.format = format;
-   templat.width[0] = width;
-   templat.height[0] = height;
+   templat.width0 = width;
+   templat.height0 = height;
    pf_get_block(templat.format, &templat.block);
 
    texture = api->texture_from_shared_handle(api, screen, &templat,
diff --git a/src/gallium/state_trackers/egl/egl_surface.c b/src/gallium/state_trackers/egl/egl_surface.c
index 91615abebe..ddd9b04cd4 100644
--- a/src/gallium/state_trackers/egl/egl_surface.c
+++ b/src/gallium/state_trackers/egl/egl_surface.c
@@ -114,10 +114,10 @@ drm_create_texture(_EGLDisplay *dpy,
 	templat.tex_usage |= PIPE_TEXTURE_USAGE_PRIMARY;
 	templat.target = PIPE_TEXTURE_2D;
 	templat.last_level = 0;
-	templat.depth[0] = 1;
+	templat.depth0 = 1;
 	templat.format = PIPE_FORMAT_A8R8G8B8_UNORM;
-	templat.width[0] = w;
-	templat.height[0] = h;
+	templat.width0 = w;
+	templat.height0 = h;
 	pf_get_block(templat.format, &templat.block);
 
 	texture = screen->texture_create(dev->screen,
diff --git a/src/gallium/state_trackers/egl/egl_tracker.c b/src/gallium/state_trackers/egl/egl_tracker.c
index 5811c20f03..745803c7eb 100644
--- a/src/gallium/state_trackers/egl/egl_tracker.c
+++ b/src/gallium/state_trackers/egl/egl_tracker.c
@@ -88,11 +88,11 @@ drm_get_device_id(struct drm_device *device)
 	}
 
 	ret = fgets(path, sizeof( path ), file);
+	fclose(file);
 	if (!ret)
 		return;
 
 	sscanf(path, "%x", &device->deviceID);
-	fclose(file);
 }
 
 static void
diff --git a/src/gallium/state_trackers/python/p_device.i b/src/gallium/state_trackers/python/p_device.i
index f16fe5b0ff..a83bcc71a1 100644
--- a/src/gallium/state_trackers/python/p_device.i
+++ b/src/gallium/state_trackers/python/p_device.i
@@ -113,9 +113,9 @@ struct st_device {
       memset(&templat, 0, sizeof(templat));
       templat.format = format;
       pf_get_block(templat.format, &templat.block);
-      templat.width[0] = width;
-      templat.height[0] = height;
-      templat.depth[0] = depth;
+      templat.width0 = width;
+      templat.height0 = height;
+      templat.depth0 = depth;
       templat.last_level = last_level;
       templat.target = target;
       templat.tex_usage = tex_usage;
diff --git a/src/gallium/state_trackers/python/p_texture.i b/src/gallium/state_trackers/python/p_texture.i
index 1d513abf3c..5416b872f5 100644
--- a/src/gallium/state_trackers/python/p_texture.i
+++ b/src/gallium/state_trackers/python/p_texture.i
@@ -59,15 +59,15 @@
    }
    
    unsigned get_width(unsigned level=0) {
-      return $self->width[level];
+      return u_minify($self->width0, level);
    }
    
    unsigned get_height(unsigned level=0) {
-      return $self->height[level];
+      return u_minify($self->height0, level);
    }
    
    unsigned get_depth(unsigned level=0) {
-      return $self->depth[level];
+      return u_minify($self->depth0, level);
    }
    
    unsigned get_nblocksx(unsigned level=0) {
@@ -88,7 +88,7 @@
          SWIG_exception(SWIG_ValueError, "face out of bounds");
       if(level > $self->last_level)
          SWIG_exception(SWIG_ValueError, "level out of bounds");
-      if(zslice >= $self->depth[level])
+      if(zslice >= u_minify($self->depth0, level))
          SWIG_exception(SWIG_ValueError, "zslice out of bounds");
       
       surface = CALLOC_STRUCT(st_surface);
@@ -375,13 +375,13 @@ struct st_surface
    static unsigned
    st_surface_width_get(struct st_surface *surface)
    {
-      return surface->texture->width[surface->level];
+      return u_minify(surface->texture->width0, surface->level);
    }
    
    static unsigned
    st_surface_height_get(struct st_surface *surface)
    {
-      return surface->texture->height[surface->level];
+      return u_minify(surface->texture->height0, surface->level);
    }
 
    static unsigned
diff --git a/src/gallium/state_trackers/python/retrace/interpreter.py b/src/gallium/state_trackers/python/retrace/interpreter.py
index 348f2e4368..d0bcb690a9 100755
--- a/src/gallium/state_trackers/python/retrace/interpreter.py
+++ b/src/gallium/state_trackers/python/retrace/interpreter.py
@@ -279,9 +279,9 @@ class Screen(Object):
     def texture_create(self, templat):
         return self.real.texture_create(
             format = templat.format,
-            width = templat.width[0],
-            height = templat.height[0],
-            depth = templat.depth[0],
+            width = templat.width0,
+            height = templat.height0,
+            depth = templat.depth0,
             last_level = templat.last_level,
             target = templat.target,
             tex_usage = templat.tex_usage,
diff --git a/src/gallium/state_trackers/python/st_device.c b/src/gallium/state_trackers/python/st_device.c
index ea7d18738f..a791113aba 100644
--- a/src/gallium/state_trackers/python/st_device.c
+++ b/src/gallium/state_trackers/python/st_device.c
@@ -252,9 +252,9 @@ st_context_create(struct st_device *st_dev)
       templat.block.size = 4;
       templat.block.width = 1;
       templat.block.height = 1;
-      templat.width[0] = 1;
-      templat.height[0] = 1;
-      templat.depth[0] = 1;
+      templat.width0 = 1;
+      templat.height0 = 1;
+      templat.depth0 = 1;
       templat.last_level = 0;
    
       st_ctx->default_texture = screen->texture_create( screen, &templat );
@@ -264,8 +264,8 @@ st_context_create(struct st_device *st_dev)
                                              0, 0, 0,
                                              PIPE_TRANSFER_WRITE,
                                              0, 0,
-                                             st_ctx->default_texture->width[0],
-                                             st_ctx->default_texture->height[0]);
+                                             st_ctx->default_texture->width0,
+                                             st_ctx->default_texture->height0);
          if (transfer) {
             uint32_t *map;
             map = (uint32_t *) screen->transfer_map(screen, transfer);
diff --git a/src/gallium/state_trackers/python/st_sample.c b/src/gallium/state_trackers/python/st_sample.c
index 53a01891e1..6fee90afda 100644
--- a/src/gallium/state_trackers/python/st_sample.c
+++ b/src/gallium/state_trackers/python/st_sample.c
@@ -528,8 +528,8 @@ st_sample_surface(struct st_surface *surface, float *rgba)
 {
    struct pipe_texture *texture = surface->texture;
    struct pipe_screen *screen = texture->screen;
-   unsigned width = texture->width[surface->level];
-   unsigned height = texture->height[surface->level];
+   unsigned width = u_minify(texture->width0, surface->level);
+   unsigned height = u_minify(texture->height0, surface->level);
    uint rgba_stride = width * 4;
    struct pipe_transfer *transfer;
    void *raw;
diff --git a/src/gallium/state_trackers/vega/api_filters.c b/src/gallium/state_trackers/vega/api_filters.c
index 862cbb03c4..faf396d087 100644
--- a/src/gallium/state_trackers/vega/api_filters.c
+++ b/src/gallium/state_trackers/vega/api_filters.c
@@ -68,9 +68,9 @@ static INLINE struct pipe_texture *create_texture_1d(struct vg_context *ctx,
    templ.target = PIPE_TEXTURE_1D;
    templ.format = PIPE_FORMAT_A8R8G8B8_UNORM;
    templ.last_level = 0;
-   templ.width[0] = color_data_len;
-   templ.height[0] = 1;
-   templ.depth[0] = 1;
+   templ.width0 = color_data_len;
+   templ.height0 = 1;
+   templ.depth0 = 1;
    pf_get_block(PIPE_FORMAT_A8R8G8B8_UNORM, &templ.block);
    templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
 
@@ -81,7 +81,7 @@ static INLINE struct pipe_texture *create_texture_1d(struct vg_context *ctx,
          screen->get_tex_transfer(screen, tex,
                                   0, 0, 0,
                                   PIPE_TRANSFER_READ_WRITE ,
-                                  0, 0, tex->width[0], tex->height[0]);
+                                  0, 0, tex->width0, tex->height0);
       void *map = screen->transfer_map(screen, transfer);
       memcpy(map, color_data, sizeof(VGint)*color_data_len);
       screen->transfer_unmap(screen, transfer);
diff --git a/src/gallium/state_trackers/vega/arc.c b/src/gallium/state_trackers/vega/arc.c
index e74c7f0334..8b04d21ea7 100644
--- a/src/gallium/state_trackers/vega/arc.c
+++ b/src/gallium/state_trackers/vega/arc.c
@@ -528,6 +528,7 @@ static INLINE int num_beziers_needed(struct arc *arc)
    double threshold = 0.05;
    VGboolean found = VG_FALSE;
    int n = 1;
+   int i;
    double min_eta, max_eta;
 
    min_eta = MIN2(arc->eta1, arc->eta2);
@@ -538,7 +539,7 @@ static INLINE int num_beziers_needed(struct arc *arc)
       if (d_eta <= 0.5 * M_PI) {
          double eta_b = min_eta;
          found = VG_TRUE;
-         for (int i = 0; found && (i < n); ++i) {
+         for (i = 0; found && (i < n); ++i) {
             double etaA = eta_b;
             eta_b += d_eta;
             found = (estimate_error(arc, etaA, eta_b) <= threshold);
@@ -554,6 +555,7 @@ static void arc_to_beziers(struct arc *arc,
                            struct arc_cb cb,
                            struct matrix *matrix)
 {
+   int i;
    int n = 1;
    double d_eta, eta_b, cos_eta_b,
       sin_eta_b, a_cos_eta_b, b_sin_eta_b, a_sin_eta_b,
@@ -607,7 +609,7 @@ static void arc_to_beziers(struct arc *arc,
    t     = tan(0.5 * d_eta);
    alpha = sin(d_eta) * (sqrt(4 + 3 * t * t) - 1) / 3;
 
-   for (int i = 0; i < n; ++i) {
+   for (i = 0; i < n; ++i) {
       struct bezier bezier;
       double xA    = x_b;
       double yA    = y_b;
diff --git a/src/gallium/state_trackers/vega/bezier.c b/src/gallium/state_trackers/vega/bezier.c
index 39a7ade016..0d5504004c 100644
--- a/src/gallium/state_trackers/vega/bezier.c
+++ b/src/gallium/state_trackers/vega/bezier.c
@@ -255,7 +255,9 @@ static enum shift_result good_offset(const struct bezier *b1,
    const float max_dist_line = threshold*offset*offset;
    const float max_dist_normal = threshold*offset;
    const float spacing = 0.25;
-   for (float i = spacing; i < 0.99; i += spacing) {
+   float i;
+
+   for (i = spacing; i < 0.99; i += spacing) {
       float p1[2],p2[2], d, l;
       float normal[2];
       bezier_point_at(b1, i, p1);
@@ -330,6 +332,7 @@ static enum shift_result shift(const struct bezier *orig,
                                struct bezier *shifted,
                                float offset, float threshold)
 {
+   int i;
    int map[4];
    VGboolean p1_p2_equal = (orig->x1 == orig->x2 && orig->y1 == orig->y2);
    VGboolean p2_p3_equal = (orig->x2 == orig->x3 && orig->y2 == orig->y3);
@@ -404,7 +407,7 @@ static enum shift_result shift(const struct bezier *orig,
    points_shifted[0][0] = points[0][0] + offset * prev_normal[0];
    points_shifted[0][1] = points[0][1] + offset * prev_normal[1];
 
-   for (int i = 1; i < np - 1; ++i) {
+   for (i = 1; i < np - 1; ++i) {
       float normal_sum[2], r;
       float next_normal[2];
       compute_pt_normal(points[i], points[i + 1], next_normal);
diff --git a/src/gallium/state_trackers/vega/image.c b/src/gallium/state_trackers/vega/image.c
index 9a722980d5..4684a5727d 100644
--- a/src/gallium/state_trackers/vega/image.c
+++ b/src/gallium/state_trackers/vega/image.c
@@ -93,8 +93,8 @@ static void vg_copy_texture(struct vg_context *ctx,
    dst_loc[3] = height;
    dst_bounds[0] = 0.f;
    dst_bounds[1] = 0.f;
-   dst_bounds[2] = dst->width[0];
-   dst_bounds[3] = dst->height[0];
+   dst_bounds[2] = dst->width0;
+   dst_bounds[3] = dst->height0;
 
    src_loc[0] = sx;
    src_loc[1] = sy;
@@ -102,8 +102,8 @@ static void vg_copy_texture(struct vg_context *ctx,
    src_loc[3] = height;
    src_bounds[0] = 0.f;
    src_bounds[1] = 0.f;
-   src_bounds[2] = src->width[0];
-   src_bounds[3] = src->height[0];
+   src_bounds[2] = src->width0;
+   src_bounds[3] = src->height0;
 
    vg_bound_rect(src_loc, src_bounds, src_shift);
    vg_bound_rect(dst_loc, dst_bounds, dst_shift);
@@ -272,9 +272,9 @@ struct vg_image * image_create(VGImageFormat format,
    pt.format = pformat;
    pf_get_block(pformat, &pt.block);
    pt.last_level = 0;
-   pt.width[0] = width;
-   pt.height[0] = height;
-   pt.depth[0] = 1;
+   pt.width0 = width;
+   pt.height0 = height;
+   pt.depth0 = 1;
    pt.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
 
    newtex = screen->texture_create(screen, &pt);
@@ -414,7 +414,7 @@ void image_sub_data(struct vg_image *image,
    { /* upload color_data */
       struct pipe_transfer *transfer = screen->get_tex_transfer(
          screen, texture, 0, 0, 0,
-         PIPE_TRANSFER_WRITE, 0, 0, texture->width[0], texture->height[0]);
+         PIPE_TRANSFER_WRITE, 0, 0, texture->width0, texture->height0);
       src += (dataStride * yoffset);
       for (i = 0; i < height; i++) {
          _vega_unpack_float_span_rgba(ctx, width, xoffset, src, dataFormat, temp);
diff --git a/src/gallium/state_trackers/vega/mask.c b/src/gallium/state_trackers/vega/mask.c
index 24650a37d5..b84103fdba 100644
--- a/src/gallium/state_trackers/vega/mask.c
+++ b/src/gallium/state_trackers/vega/mask.c
@@ -426,7 +426,7 @@ static void mask_using_texture(struct pipe_texture *texture,
    if (!surface)
       return;
    if (!intersect_rectangles(surface->width, surface->height,
-                             texture->width[0], texture->height[0],
+                             texture->width0, texture->height0,
                              x, y, width, height,
                              offsets, loc))
       return;
@@ -493,9 +493,9 @@ struct vg_mask_layer * mask_layer_create(VGint width, VGint height)
       pt.format = PIPE_FORMAT_A8R8G8B8_UNORM;
       pf_get_block(PIPE_FORMAT_A8R8G8B8_UNORM, &pt.block);
       pt.last_level = 0;
-      pt.width[0] = width;
-      pt.height[0] = height;
-      pt.depth[0] = 1;
+      pt.width0 = width;
+      pt.height0 = height;
+      pt.depth0 = 1;
       pt.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
       pt.compressed = 0;
 
@@ -607,8 +607,8 @@ void mask_render_to(struct path *path,
    struct vg_mask_layer *temp_layer;
    VGint width, height;
 
-   width = fb_buffers->alpha_mask->width[0];
-   height = fb_buffers->alpha_mask->width[0];
+   width = fb_buffers->alpha_mask->width0;
+   height = fb_buffers->alpha_mask->width0;
 
    temp_layer = mask_layer_create(width, height);
 
diff --git a/src/gallium/state_trackers/vega/paint.c b/src/gallium/state_trackers/vega/paint.c
index 04a6ba9cdc..e8ca7d9e89 100644
--- a/src/gallium/state_trackers/vega/paint.c
+++ b/src/gallium/state_trackers/vega/paint.c
@@ -151,9 +151,9 @@ static INLINE struct pipe_texture *create_gradient_texture(struct vg_paint *p)
    templ.target = PIPE_TEXTURE_1D;
    templ.format = PIPE_FORMAT_A8R8G8B8_UNORM;
    templ.last_level = 0;
-   templ.width[0] = 1024;
-   templ.height[0] = 1;
-   templ.depth[0] = 1;
+   templ.width0 = 1024;
+   templ.height0 = 1;
+   templ.depth0 = 1;
    pf_get_block(PIPE_FORMAT_A8R8G8B8_UNORM, &templ.block);
    templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
 
@@ -328,8 +328,8 @@ static INLINE void  paint_pattern_buffer(struct vg_paint *paint, void *buffer)
 
    map[4] = 0.f;
    map[5] = 1.f;
-   map[6] = paint->pattern.texture->width[0];
-   map[7] = paint->pattern.texture->height[0];
+   map[6] = paint->pattern.texture->width0;
+   map[7] = paint->pattern.texture->height0;
    {
       struct matrix mat;
       memcpy(&mat, &ctx->state.vg.fill_paint_to_user_matrix,
diff --git a/src/gallium/state_trackers/vega/renderer.c b/src/gallium/state_trackers/vega/renderer.c
index 396c88aa3d..9085ed1bfe 100644
--- a/src/gallium/state_trackers/vega/renderer.c
+++ b/src/gallium/state_trackers/vega/renderer.c
@@ -230,13 +230,13 @@ void renderer_draw_texture(struct renderer *r,
    struct pipe_buffer *buf;
    VGfloat s0, t0, s1, t1;
 
-   assert(tex->width[0] != 0);
-   assert(tex->height[0] != 0);
+   assert(tex->width0 != 0);
+   assert(tex->height0 != 0);
 
-   s0 = x1offset / tex->width[0];
-   s1 = x2offset / tex->width[0];
-   t0 = y1offset / tex->height[0];
-   t1 = y2offset / tex->height[0];
+   s0 = x1offset / tex->width0;
+   s1 = x2offset / tex->width0;
+   t0 = y1offset / tex->height0;
+   t1 = y2offset / tex->height0;
 
    cso_save_vertex_shader(r->cso);
    /* shaders */
@@ -276,10 +276,10 @@ void renderer_copy_texture(struct renderer *ctx,
    struct pipe_framebuffer_state fb;
    float s0, t0, s1, t1;
 
-   assert(src->width[0] != 0);
-   assert(src->height[0] != 0);
-   assert(dst->width[0] != 0);
-   assert(dst->height[0] != 0);
+   assert(src->width0 != 0);
+   assert(src->height0 != 0);
+   assert(dst->width0 != 0);
+   assert(dst->height0 != 0);
 
 #if 0
    debug_printf("copy texture [%f, %f, %f, %f], [%f, %f, %f, %f]\n",
@@ -287,10 +287,10 @@ void renderer_copy_texture(struct renderer *ctx,
 #endif
 
 #if 1
-   s0 = sx1 / src->width[0];
-   s1 = sx2 / src->width[0];
-   t0 = sy1 / src->height[0];
-   t1 = sy2 / src->height[0];
+   s0 = sx1 / src->width0;
+   s1 = sx2 / src->width0;
+   t0 = sy1 / src->height0;
+   t1 = sy2 / src->height0;
 #else
    s0 = 0;
    s1 = 1;
@@ -445,9 +445,9 @@ void renderer_copy_surface(struct renderer *ctx,
    texTemp.target = PIPE_TEXTURE_2D;
    texTemp.format = src->format;
    texTemp.last_level = 0;
-   texTemp.width[0] = srcW;
-   texTemp.height[0] = srcH;
-   texTemp.depth[0] = 1;
+   texTemp.width0 = srcW;
+   texTemp.height0 = srcH;
+   texTemp.depth0 = 1;
    pf_get_block(src->format, &texTemp.block);
 
    tex = screen->texture_create(screen, &texTemp);
@@ -570,13 +570,13 @@ void renderer_texture_quad(struct renderer *r,
    struct pipe_buffer *buf;
    VGfloat s0, t0, s1, t1;
 
-   assert(tex->width[0] != 0);
-   assert(tex->height[0] != 0);
+   assert(tex->width0 != 0);
+   assert(tex->height0 != 0);
 
-   s0 = x1offset / tex->width[0];
-   s1 = x2offset / tex->width[0];
-   t0 = y1offset / tex->height[0];
-   t1 = y2offset / tex->height[0];
+   s0 = x1offset / tex->width0;
+   s1 = x2offset / tex->width0;
+   t0 = y1offset / tex->height0;
+   t1 = y2offset / tex->height0;
 
    cso_save_vertex_shader(r->cso);
    /* shaders */
diff --git a/src/gallium/state_trackers/vega/vg_context.c b/src/gallium/state_trackers/vega/vg_context.c
index e0ff02f3a9..00d23f5c22 100644
--- a/src/gallium/state_trackers/vega/vg_context.c
+++ b/src/gallium/state_trackers/vega/vg_context.c
@@ -231,6 +231,8 @@ static void update_clip_state(struct vg_context *ctx)
    if (state->scissoring) {
       struct pipe_blend_state *blend = &ctx->state.g3d.blend;
       struct pipe_framebuffer_state *fb = &ctx->state.g3d.fb;
+      int i;
+
       dsa->depth.writemask = 1;/*glDepthMask(TRUE);*/
       dsa->depth.func = PIPE_FUNC_ALWAYS;
       dsa->depth.enabled = 1;
@@ -254,7 +256,7 @@ static void update_clip_state(struct vg_context *ctx)
       cso_set_blend(ctx->cso_context, blend);
 
       /* enable scissoring */
-      for (int i = 0; i < state->scissor_rects_num; ++i) {
+      for (i = 0; i < state->scissor_rects_num; ++i) {
          const float x      = state->scissor_rects[i * 4 + 0].f;
          const float y      = state->scissor_rects[i * 4 + 1].f;
          const float width  = state->scissor_rects[i * 4 + 2].f;
diff --git a/src/gallium/state_trackers/vega/vg_tracker.c b/src/gallium/state_trackers/vega/vg_tracker.c
index c4da01e52c..d28463dd1b 100644
--- a/src/gallium/state_trackers/vega/vg_tracker.c
+++ b/src/gallium/state_trackers/vega/vg_tracker.c
@@ -51,9 +51,9 @@ create_texture(struct pipe_context *pipe, enum pipe_format format,
 
    templ.target = PIPE_TEXTURE_2D;
    pf_get_block(templ.format, &templ.block);
-   templ.width[0] = width;
-   templ.height[0] = height;
-   templ.depth[0] = 1;
+   templ.width0 = width;
+   templ.height0 = height;
+   templ.depth0 = 1;
    templ.last_level = 0;
 
    if (pf_get_component_bits(format, PIPE_FORMAT_COMP_S)) {
diff --git a/src/gallium/state_trackers/wgl/stw_winsys.h b/src/gallium/state_trackers/wgl/stw_winsys.h
index 1ead47d6e6..1de6e906d0 100644
--- a/src/gallium/state_trackers/wgl/stw_winsys.h
+++ b/src/gallium/state_trackers/wgl/stw_winsys.h
@@ -73,7 +73,7 @@ struct stw_winsys
                           HANDLE hSharedSurface);
 
    /**
-    * Open a shared surface (optional).
+    * Close a shared surface (optional).
     */
    void
    (*shared_surface_close)(struct pipe_screen *screen,
diff --git a/src/gallium/state_trackers/xorg/xorg_composite.c b/src/gallium/state_trackers/xorg/xorg_composite.c
index 93a3e1b8cf..69a83b1523 100644
--- a/src/gallium/state_trackers/xorg/xorg_composite.c
+++ b/src/gallium/state_trackers/xorg/xorg_composite.c
@@ -151,8 +151,11 @@ render_filter_to_gallium(int xrender_filter, int *out_filter)
    case PictFilterBest:
       *out_filter = PIPE_TEX_FILTER_LINEAR;
       break;
+   case PictFilterConvolution:
+      *out_filter = PIPE_TEX_FILTER_NEAREST;
+      return FALSE;
    default:
-      debug_printf("Unkown xrender filter");
+      debug_printf("Unknown xrender filter\n");
       *out_filter = PIPE_TEX_FILTER_NEAREST;
       return FALSE;
    }
@@ -237,7 +240,7 @@ picture_format_fixups(struct exa_pixmap_priv *pSrc, PicturePtr pSrcPicture, bool
 
    if (pSrc->picture_format == pSrcPicture->format) {
       if (pSrc->picture_format == PICT_a8)
-         return mask ? FS_MASK_LUMINANCE : FS_MASK_LUMINANCE;
+         return mask ? FS_MASK_LUMINANCE : FS_SRC_LUMINANCE;
       return 0;
    }
 
@@ -436,8 +439,8 @@ setup_fs_constant_buffer(struct exa_context *exa)
 static void
 setup_constant_buffers(struct exa_context *exa, struct exa_pixmap_priv *pDst)
 {
-   int width = pDst->tex->width[0];
-   int height = pDst->tex->height[0];
+   int width = pDst->tex->width0;
+   int height = pDst->tex->height0;
 
    setup_vs_constant_buffer(exa, width, height);
    setup_fs_constant_buffer(exa);
@@ -564,6 +567,8 @@ boolean xorg_solid_bind_state(struct exa_context *exa,
    renderer_bind_viewport(exa->renderer, pixmap);
    renderer_bind_rasterizer(exa->renderer);
    bind_blend_state(exa, PictOpSrc, NULL, NULL, NULL);
+   cso_set_samplers(exa->renderer->cso, 0, NULL);
+   cso_set_sampler_textures(exa->renderer->cso, 0, NULL);
    setup_constant_buffers(exa, pixmap);
 
    shader = xorg_shaders_get(exa->renderer->shaders, vs_traits, fs_traits);
diff --git a/src/gallium/state_trackers/xorg/xorg_crtc.c b/src/gallium/state_trackers/xorg/xorg_crtc.c
index 85b9162d4c..c4751724c9 100644
--- a/src/gallium/state_trackers/xorg/xorg_crtc.c
+++ b/src/gallium/state_trackers/xorg/xorg_crtc.c
@@ -187,10 +187,10 @@ crtc_load_cursor_argb(xf86CrtcPtr crtc, CARD32 * image)
 	templat.tex_usage |= PIPE_TEXTURE_USAGE_PRIMARY;
 	templat.target = PIPE_TEXTURE_2D;
 	templat.last_level = 0;
-	templat.depth[0] = 1;
+	templat.depth0 = 1;
 	templat.format = PIPE_FORMAT_A8R8G8B8_UNORM;
-	templat.width[0] = 64;
-	templat.height[0] = 64;
+	templat.width0 = 64;
+	templat.height0 = 64;
 	pf_get_block(templat.format, &templat.block);
 
 	crtcp->cursor_tex = ms->screen->texture_create(ms->screen,
diff --git a/src/gallium/state_trackers/xorg/xorg_dri2.c b/src/gallium/state_trackers/xorg/xorg_dri2.c
index c41a7cd639..406e0afff4 100644
--- a/src/gallium/state_trackers/xorg/xorg_dri2.c
+++ b/src/gallium/state_trackers/xorg/xorg_dri2.c
@@ -42,6 +42,12 @@
 
 #include "util/u_rect.h"
 
+/* Make all the #if cases in the code esier to read */
+/* XXX can it be set to 1? */
+#ifndef DRI2INFOREC_VERSION
+#define DRI2INFOREC_VERSION 0
+#endif
+
 typedef struct {
     PixmapPtr pPixmap;
     struct pipe_texture *tex;
@@ -79,7 +85,7 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
     case DRI2BufferFrontLeft:
 	break;
     case DRI2BufferStencil:
-#if defined(DRI2INFOREC_VERSION) && DRI2INFOREC_VERSION > 2
+#if DRI2INFOREC_VERSION >= 3
     case DRI2BufferDepthStencil:
 #else
     /* Works on old X servers because sanity checking is for the weak */
@@ -103,9 +109,9 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
 		template.format = ms->ds_depth_bits_last ?
 		    PIPE_FORMAT_S8Z24_UNORM : PIPE_FORMAT_Z24S8_UNORM;
 	    pf_get_block(template.format, &template.block);
-	    template.width[0] = pDraw->width;
-	    template.height[0] = pDraw->height;
-	    template.depth[0] = 1;
+	    template.width0 = pDraw->width;
+	    template.height0 = pDraw->height;
+	    template.depth0 = 1;
 	    template.last_level = 0;
 	    template.tex_usage = PIPE_TEXTURE_USAGE_DEPTH_STENCIL |
 		PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
@@ -121,9 +127,12 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
     }
 
     if (!tex) {
+	/* First call to make sure we have a pixmap private */
 	exaMoveInPixmap(private->pPixmap);
 	xorg_exa_set_shared_usage(private->pPixmap);
 	pScreen->ModifyPixmapHeader(private->pPixmap, 0, 0, 0, 0, 0, NULL);
+	/* Second call to make sure texture has valid contents */
+	exaMoveInPixmap(private->pPixmap);
 	tex = xorg_exa_get_texture(private->pPixmap);
     }
 
@@ -137,6 +146,11 @@ driDoCreateBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int format)
     buffer->cpp = 4;
     buffer->driverPrivate = private;
     buffer->flags = 0; /* not tiled */
+#if DRI2INFOREC_VERSION == 2
+    ((DRI2Buffer2Ptr)buffer)->format = 0;
+#elif DRI2INFOREC_VERSION >= 3
+    buffer->format = 0;
+#endif
     private->tex = tex;
 
     return TRUE;
@@ -157,12 +171,12 @@ driDoDestroyBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer)
     (*pScreen->DestroyPixmap)(private->pPixmap);
 }
 
-#if defined(DRI2INFOREC_VERSION) && DRI2INFOREC_VERSION > 2
+#if DRI2INFOREC_VERSION >= 2
 
-static DRI2BufferPtr
+static DRI2Buffer2Ptr
 driCreateBuffer(DrawablePtr pDraw, unsigned int attachment, unsigned int format)
 {
-    DRI2BufferPtr buffer;
+    DRI2Buffer2Ptr buffer;
     BufferPrivatePtr private;
 
     buffer = xcalloc(1, sizeof *buffer);
@@ -177,7 +191,8 @@ driCreateBuffer(DrawablePtr pDraw, unsigned int attachment, unsigned int format)
     buffer->attachment = attachment;
     buffer->driverPrivate = private;
 
-    if (driDoCreateBuffer(pDraw, buffer, format))
+    /* So far it is safe to downcast a DRI2Buffer2Ptr to DRI2BufferPtr */
+    if (driDoCreateBuffer(pDraw, (DRI2BufferPtr)buffer, format))
 	return buffer;
 
     xfree(private);
@@ -187,15 +202,16 @@ fail:
 }
 
 static void
-driDestroyBuffer(DrawablePtr pDraw, DRI2BufferPtr buffer)
+driDestroyBuffer(DrawablePtr pDraw, DRI2Buffer2Ptr buffer)
 {
-    driDoDestroyBuffer(pDraw, buffer);
+    /* So far it is safe to downcast a DRI2Buffer2Ptr to DRI2BufferPtr */
+    driDoDestroyBuffer(pDraw, (DRI2BufferPtr)buffer);
 
     xfree(buffer->driverPrivate);
     xfree(buffer);
 }
 
-#else /* DRI2INFOREC_VERSION <= 2 */
+#else /* DRI2INFOREC_VERSION < 2 */
 
 static DRI2BufferPtr
 driCreateBuffers(DrawablePtr pDraw, unsigned int *attachments, int count)
@@ -245,7 +261,7 @@ driDestroyBuffers(DrawablePtr pDraw, DRI2BufferPtr buffers, int count)
     }
 }
 
-#endif /* DRI2INFOREC_VERSION */
+#endif /* DRI2INFOREC_VERSION >= 2 */
 
 static void
 driCopyRegion(DrawablePtr pDraw, RegionPtr pRegion,
@@ -260,6 +276,7 @@ driCopyRegion(DrawablePtr pDraw, RegionPtr pRegion,
     PixmapPtr dst_pixmap;
     GCPtr gc;
     RegionPtr copy_clip;
+    Bool save_accel;
 
     /*
      * In driCreateBuffers we dewrap windows into the
@@ -325,8 +342,11 @@ driCopyRegion(DrawablePtr pDraw, RegionPtr pRegion,
 	}
     }
 
+    save_accel = ms->exa->accel;
+    ms->exa->accel = TRUE;
     (*gc->ops->CopyArea)(&src_pixmap->drawable, &dst_pixmap->drawable, gc,
 			 0, 0, pDraw->width, pDraw->height, 0, 0);
+    ms->exa->accel = save_accel;
 
     FreeScratchGC(gc);
 
@@ -342,7 +362,7 @@ driScreenInit(ScreenPtr pScreen)
     modesettingPtr ms = modesettingPTR(pScrn);
     DRI2InfoRec dri2info;
 
-#if defined(DRI2INFOREC_VERSION)
+#if DRI2INFOREC_VERSION >= 2
     dri2info.version = DRI2INFOREC_VERSION;
 #else
     dri2info.version = 1;
@@ -352,7 +372,7 @@ driScreenInit(ScreenPtr pScreen)
     dri2info.driverName = pScrn->driverName;
     dri2info.deviceName = "/dev/dri/card0"; /* FIXME */
 
-#if defined(DRI2INFOREC_VERSION) && DRI2INFOREC_VERSION > 2
+#if DRI2INFOREC_VERSION >= 2
     dri2info.CreateBuffer = driCreateBuffer;
     dri2info.DestroyBuffer = driDestroyBuffer;
 #else
diff --git a/src/gallium/state_trackers/xorg/xorg_driver.c b/src/gallium/state_trackers/xorg/xorg_driver.c
index 26cf2dd772..d949167adc 100644
--- a/src/gallium/state_trackers/xorg/xorg_driver.c
+++ b/src/gallium/state_trackers/xorg/xorg_driver.c
@@ -75,10 +75,12 @@ static Bool PreInit(ScrnInfoPtr pScrn, int flags);
 typedef enum
 {
     OPTION_SW_CURSOR,
+    OPTION_2D_ACCEL,
 } modesettingOpts;
 
 static const OptionInfoRec Options[] = {
     {OPTION_SW_CURSOR, "SWcursor", OPTV_BOOLEAN, {0}, FALSE},
+    {OPTION_2D_ACCEL, "2DAccel", OPTV_BOOLEAN, {0}, FALSE},
     {-1, NULL, OPTV_NONE, {0}, FALSE}
 };
 
@@ -609,7 +611,8 @@ ScreenInit(int scrnIndex, ScreenPtr pScreen, int argc, char **argv)
 
     xf86SetBlackWhitePixels(pScreen);
 
-    ms->exa = xorg_exa_init(pScrn);
+    ms->exa = xorg_exa_init(pScrn, xf86ReturnOptValBool(ms->Options,
+                                                        OPTION_2D_ACCEL, TRUE));
     ms->debug_fallback = debug_get_bool_option("XORG_DEBUG_FALLBACK", TRUE);
 
     xorg_init_video(pScreen);
diff --git a/src/gallium/state_trackers/xorg/xorg_exa.c b/src/gallium/state_trackers/xorg/xorg_exa.c
index 6fa274eb0a..a68a626fa4 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa.c
+++ b/src/gallium/state_trackers/xorg/xorg_exa.c
@@ -46,7 +46,6 @@
 #include "util/u_rect.h"
 
 #define DEBUG_PRINT 0
-#define ACCEL_ENABLED TRUE
 
 /*
  * Helper functions
@@ -170,15 +169,7 @@ xorg_exa_common_done(struct exa_context *exa)
 static void
 ExaWaitMarker(ScreenPtr pScreen, int marker)
 {
-   ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];
-   modesettingPtr ms = modesettingPTR(pScrn);
-   struct exa_context *exa = ms->exa;
-
-#if 0
-   xorg_exa_flush(exa, PIPE_FLUSH_RENDER_CACHE, NULL);
-#else
-   xorg_exa_finish(exa);
-#endif
+   /* Nothing to do, handled in the PrepareAccess hook */
 }
 
 static int
@@ -239,6 +230,11 @@ ExaUploadToScreen(PixmapPtr pPix, int x, int y, int w, int h, char *src,
     if (!priv || !priv->tex)
 	return FALSE;
 
+    /* make sure that any pending operations are flushed to hardware */
+    if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) &
+	(PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE))
+	xorg_exa_flush(exa, 0, NULL);
+
     transfer = exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0,
 					   PIPE_TRANSFER_WRITE, x, y, w, h);
     if (!transfer)
@@ -288,7 +284,7 @@ ExaPrepareAccess(PixmapPtr pPix, int index)
 					PIPE_TRANSFER_MAP_DIRECTLY |
 #endif
 					PIPE_TRANSFER_READ_WRITE,
-					0, 0, priv->tex->width[0], priv->tex->height[0]);
+					0, 0, priv->tex->width0, priv->tex->height0);
 	if (!priv->map_transfer)
 #ifdef EXA_MIXED_PIXMAPS
 	    return FALSE;
@@ -384,7 +380,7 @@ ExaPrepareSolid(PixmapPtr pPixmap, int alu, Pixel planeMask, Pixel fg)
 	XORG_FALLBACK("format %s", pf_name(priv->tex->format));
     }
 
-    return ACCEL_ENABLED && xorg_solid_bind_state(exa, priv, fg);
+    return exa->accel && xorg_solid_bind_state(exa, priv, fg);
 }
 
 static void
@@ -443,7 +439,7 @@ ExaPrepareCopy(PixmapPtr pSrcPixmap, PixmapPtr pDstPixmap, int xdir,
     exa->copy.src = src_priv;
     exa->copy.dst = priv;
 
-    return ACCEL_ENABLED;
+    return exa->accel;
 }
 
 static void
@@ -572,7 +568,7 @@ ExaPrepareComposite(int op, PicturePtr pSrcPicture,
                        render_format_name(pMaskPicture->format));
    }
 
-   return ACCEL_ENABLED &&
+   return exa->accel &&
           xorg_composite_bind_state(exa, op, pSrcPicture, pMaskPicture,
                                     pDstPicture,
                                     pSrc ? exaGetPixmapDriverPrivate(pSrc) : NULL,
@@ -605,6 +601,9 @@ ExaCheckComposite(int op,
 		  PicturePtr pSrcPicture, PicturePtr pMaskPicture,
 		  PicturePtr pDstPicture)
 {
+   ScrnInfoPtr pScrn = xf86Screens[pDstPicture->pDrawable->pScreen->myNum];
+   modesettingPtr ms = modesettingPTR(pScrn);
+   struct exa_context *exa = ms->exa;
    boolean accelerated = xorg_composite_accelerated(op,
                                                     pSrcPicture,
                                                     pMaskPicture,
@@ -613,7 +612,7 @@ ExaCheckComposite(int op,
    debug_printf("ExaCheckComposite(%d, %p, %p, %p) = %d\n",
                 op, pSrcPicture, pMaskPicture, pDstPicture, accelerated);
 #endif
-   return ACCEL_ENABLED && accelerated;
+   return exa->accel && accelerated;
 }
 
 static void *
@@ -751,10 +750,11 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height,
 			     bitsPerPixel, devKind, NULL);
 
     /* Deal with screen resize */
-    if (!priv->tex ||
-        (priv->tex->width[0] != width ||
-         priv->tex->height[0] != height ||
-         priv->tex_flags != priv->flags)) {
+    if ((exa->accel || priv->flags) &&
+        (!priv->tex ||
+         (priv->tex->width0 != width ||
+          priv->tex->height0 != height ||
+          priv->tex_flags != priv->flags))) {
 	struct pipe_texture *texture = NULL;
 	struct pipe_texture template;
 
@@ -762,9 +762,9 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height,
 	template.target = PIPE_TEXTURE_2D;
 	exa_get_pipe_format(depth, &template.format, &bitsPerPixel, &priv->picture_format);
 	pf_get_block(template.format, &template.block);
-	template.width[0] = width;
-	template.height[0] = height;
-	template.depth[0] = 1;
+	template.width0 = width;
+	template.height0 = height;
+	template.depth0 = 1;
 	template.last_level = 0;
 	template.tex_usage = PIPE_TEXTURE_USAGE_RENDER_TARGET | priv->flags;
 	priv->tex_flags = priv->flags;
@@ -779,12 +779,12 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height,
 	    src_surf = xorg_gpu_surface(exa->pipe->screen, priv);
         if (exa->pipe->surface_copy) {
             exa->pipe->surface_copy(exa->pipe, dst_surf, 0, 0, src_surf,
-                        0, 0, min(width, texture->width[0]),
-                        min(height, texture->height[0]));
+                        0, 0, min(width, texture->width0),
+                        min(height, texture->height0));
         } else {
             util_surface_copy(exa->pipe, FALSE, dst_surf, 0, 0, src_surf,
-                        0, 0, min(width, texture->width[0]),
-                        min(height, texture->height[0]));
+                        0, 0, min(width, texture->width0),
+                        min(height, texture->height0));
         }
 	    exa->scrn->tex_surface_destroy(dst_surf);
 	    exa->scrn->tex_surface_destroy(src_surf);
@@ -817,8 +817,8 @@ xorg_exa_set_texture(PixmapPtr pPixmap, struct  pipe_texture *tex)
     if (!priv)
 	return FALSE;
 
-    if (pPixmap->drawable.width != tex->width[0] ||
-	pPixmap->drawable.height != tex->height[0])
+    if (pPixmap->drawable.width != tex->width0 ||
+	pPixmap->drawable.height != tex->height0)
 	return FALSE;
 
     pipe_texture_reference(&priv->tex, tex);
@@ -841,9 +841,9 @@ xorg_exa_create_root_texture(ScrnInfoPtr pScrn,
     template.target = PIPE_TEXTURE_2D;
     exa_get_pipe_format(depth, &template.format, &bitsPerPixel, &dummy);
     pf_get_block(template.format, &template.block);
-    template.width[0] = width;
-    template.height[0] = height;
-    template.depth[0] = 1;
+    template.width0 = width;
+    template.height0 = height;
+    template.depth0 = 1;
     template.last_level = 0;
     template.tex_usage |= PIPE_TEXTURE_USAGE_RENDER_TARGET;
     template.tex_usage |= PIPE_TEXTURE_USAGE_PRIMARY;
@@ -869,7 +869,7 @@ xorg_exa_close(ScrnInfoPtr pScrn)
 }
 
 void *
-xorg_exa_init(ScrnInfoPtr pScrn)
+xorg_exa_init(ScrnInfoPtr pScrn, Bool accel)
 {
    modesettingPtr ms = modesettingPTR(pScrn);
    struct exa_context *exa;
@@ -934,6 +934,7 @@ xorg_exa_init(ScrnInfoPtr pScrn)
    ms->ctx = exa->pipe;
 
    exa->renderer = renderer_create(exa->pipe);
+   exa->accel = accel;
 
    return (void *)exa;
 
diff --git a/src/gallium/state_trackers/xorg/xorg_exa.h b/src/gallium/state_trackers/xorg/xorg_exa.h
index 7f4aebb9c3..15cc29d662 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa.h
+++ b/src/gallium/state_trackers/xorg/xorg_exa.h
@@ -24,6 +24,8 @@ struct exa_context
    float solid_color[4];
    boolean has_solid_color;
 
+   boolean accel;
+
    /* float[9] projective matrix bound to pictures */
    struct {
       float    src[9];
diff --git a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
index 3bf64b6331..52b97af163 100644
--- a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
+++ b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c
@@ -47,22 +47,22 @@ static void
 print_fs_traits(int fs_traits)
 {
    const char *strings[] = {
-      "FS_COMPOSITE",       // = 1 << 0,
-      "FS_MASK",            // = 1 << 1,
-      "FS_SOLID_FILL",      // = 1 << 2,
-      "FS_LINGRAD_FILL",    // = 1 << 3,
-      "FS_RADGRAD_FILL",    // = 1 << 4,
-      "FS_CA_FULL",         // = 1 << 5, /* src.rgba * mask.rgba */
-      "FS_CA_SRCALPHA",     // = 1 << 6, /* src.aaaa * mask.rgba */
-      "FS_YUV",             // = 1 << 7,
-      "FS_SRC_REPEAT_NONE", // = 1 << 8,
-      "FS_MASK_REPEAT_NONE",// = 1 << 9,
-      "FS_SRC_SWIZZLE_RGB", // = 1 << 10,
-      "FS_MASK_SWIZZLE_RGB",// = 1 << 11,
-      "FS_SRC_SET_ALPHA",   // = 1 << 12,
-      "FS_MASK_SET_ALPHA",  // = 1 << 13,
-      "FS_SRC_LUMINANCE",   // = 1 << 14,
-      "FS_MASK_LUMINANCE",  // = 1 << 15,
+      "FS_COMPOSITE",       /* = 1 << 0 */
+      "FS_MASK",            /* = 1 << 1 */
+      "FS_SOLID_FILL",      /* = 1 << 2 */
+      "FS_LINGRAD_FILL",    /* = 1 << 3 */
+      "FS_RADGRAD_FILL",    /* = 1 << 4 */
+      "FS_CA_FULL",         /* = 1 << 5 - src.rgba * mask.rgba */
+      "FS_CA_SRCALPHA",     /* = 1 << 6 - src.aaaa * mask.rgba */
+      "FS_YUV",             /* = 1 << 7 */
+      "FS_SRC_REPEAT_NONE", /* = 1 << 8 */
+      "FS_MASK_REPEAT_NONE",/* = 1 << 9 */
+      "FS_SRC_SWIZZLE_RGB", /* = 1 << 10 */
+      "FS_MASK_SWIZZLE_RGB",/* = 1 << 11 */
+      "FS_SRC_SET_ALPHA",   /* = 1 << 12 */
+      "FS_MASK_SET_ALPHA",  /* = 1 << 13 */
+      "FS_SRC_LUMINANCE",   /* = 1 << 14 */
+      "FS_MASK_LUMINANCE",  /* = 1 << 15 */
    };
    int i, k;
    debug_printf("%s: ", __func__);
diff --git a/src/gallium/state_trackers/xorg/xorg_renderer.c b/src/gallium/state_trackers/xorg/xorg_renderer.c
index 723605312c..418a8dd88b 100644
--- a/src/gallium/state_trackers/xorg/xorg_renderer.c
+++ b/src/gallium/state_trackers/xorg/xorg_renderer.c
@@ -167,14 +167,14 @@ add_vertex_data1(struct xorg_renderer *r,
       map_point(src_matrix, pt3[0], pt3[1], &pt3[0], &pt3[1]);
    }
 
-   s0 =  pt0[0] / src->width[0];
-   s1 =  pt1[0] / src->width[0];
-   s2 =  pt2[0] / src->width[0];
-   s3 =  pt3[0] / src->width[0];
-   t0 =  pt0[1] / src->height[0];
-   t1 =  pt1[1] / src->height[0];
-   t2 =  pt2[1] / src->height[0];
-   t3 =  pt3[1] / src->height[0];
+   s0 =  pt0[0] / src->width0;
+   s1 =  pt1[0] / src->width0;
+   s2 =  pt2[0] / src->width0;
+   s3 =  pt3[0] / src->width0;
+   t0 =  pt0[1] / src->height0;
+   t1 =  pt1[1] / src->height0;
+   t2 =  pt2[1] / src->height0;
+   t3 =  pt3[1] / src->height0;
 
    /* 1st vertex */
    add_vertex_1tex(r, dstX, dstY, s0, t0);
@@ -262,15 +262,15 @@ add_vertex_data2(struct xorg_renderer *r,
       map_point(mask_matrix, mpt1[0], mpt1[1], &mpt1[0], &mpt1[1]);
    }
 
-   src_s0 = spt0[0] / src->width[0];
-   src_t0 = spt0[1] / src->height[0];
-   src_s1 = spt1[0] / src->width[0];
-   src_t1 = spt1[1] / src->height[0];
+   src_s0 = spt0[0] / src->width0;
+   src_t0 = spt0[1] / src->height0;
+   src_s1 = spt1[0] / src->width0;
+   src_t1 = spt1[1] / src->height0;
 
-   mask_s0 = mpt0[0] / mask->width[0];
-   mask_t0 = mpt0[1] / mask->height[0];
-   mask_s1 = mpt1[0] / mask->width[0];
-   mask_t1 = mpt1[1] / mask->height[0];
+   mask_s0 = mpt0[0] / mask->width0;
+   mask_t0 = mpt0[1] / mask->height0;
+   mask_s1 = mpt1[0] / mask->width0;
+   mask_t1 = mpt1[1] / mask->height0;
 
    /* 1st vertex */
    add_vertex_2tex(r, dstX, dstY,
@@ -300,10 +300,10 @@ setup_vertex_data_yuv(struct xorg_renderer *r,
    spt1[0] = srcX + srcW;
    spt1[1] = srcY + srcH;
 
-   s0 = spt0[0] / tex[0]->width[0];
-   t0 = spt0[1] / tex[0]->height[0];
-   s1 = spt1[0] / tex[0]->width[0];
-   t1 = spt1[1] / tex[0]->height[0];
+   s0 = spt0[0] / tex[0]->width0;
+   t0 = spt0[1] / tex[0]->height0;
+   s1 = spt1[0] / tex[0]->width0;
+   t1 = spt1[1] / tex[0]->height0;
 
    /* 1st vertex */
    add_vertex_1tex(r, dstX, dstY, s0, t0);
@@ -387,8 +387,8 @@ void renderer_bind_framebuffer(struct xorg_renderer *r,
    struct pipe_surface *surface = xorg_gpu_surface(r->pipe->screen, priv);
    memset(&state, 0, sizeof(struct pipe_framebuffer_state));
 
-   state.width  = priv->tex->width[0];
-   state.height = priv->tex->height[0];
+   state.width  = priv->tex->width0;
+   state.height = priv->tex->height0;
 
    state.nr_cbufs = 1;
    state.cbufs[0] = surface;
@@ -407,8 +407,8 @@ void renderer_bind_framebuffer(struct xorg_renderer *r,
 void renderer_bind_viewport(struct xorg_renderer *r,
                             struct exa_pixmap_priv *dst)
 {
-   int width = dst->tex->width[0];
-   int height = dst->tex->height[0];
+   int width = dst->tex->width0;
+   int height = dst->tex->height0;
 
    /*debug_printf("Bind viewport (%d, %d)\n", width, height);*/
 
@@ -584,16 +584,16 @@ static void renderer_copy_texture(struct xorg_renderer *r,
    float s0, t0, s1, t1;
    struct xorg_shader shader;
 
-   assert(src->width[0] != 0);
-   assert(src->height[0] != 0);
-   assert(dst->width[0] != 0);
-   assert(dst->height[0] != 0);
+   assert(src->width0 != 0);
+   assert(src->height0 != 0);
+   assert(dst->width0 != 0);
+   assert(dst->height0 != 0);
 
 #if 1
-   s0 = sx1 / src->width[0];
-   s1 = sx2 / src->width[0];
-   t0 = sy1 / src->height[0];
-   t1 = sy2 / src->height[0];
+   s0 = sx1 / src->width0;
+   s1 = sx2 / src->width0;
+   t0 = sy1 / src->height0;
+   t1 = sy2 / src->height0;
 #else
    s0 = 0;
    s1 = 1;
@@ -730,9 +730,9 @@ create_sampler_texture(struct xorg_renderer *r,
    templ.target = PIPE_TEXTURE_2D;
    templ.format = format;
    templ.last_level = 0;
-   templ.width[0] = src->width[0];
-   templ.height[0] = src->height[0];
-   templ.depth[0] = 1;
+   templ.width0 = src->width0;
+   templ.height0 = src->height0;
+   templ.depth0 = 1;
    pf_get_block(format, &templ.block);
    templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
 
@@ -754,13 +754,13 @@ create_sampler_texture(struct xorg_renderer *r,
                 ps_tex, /* dest */
                 0, 0, /* destx/y */
                 ps_read,
-                0, 0, src->width[0], src->height[0]);
+                0, 0, src->width0, src->height0);
       } else {
           util_surface_copy(pipe, FALSE,
                 ps_tex, /* dest */
                 0, 0, /* destx/y */
                 ps_read,
-                0, 0, src->width[0], src->height[0]);
+                0, 0, src->width0, src->height0);
       }
       pipe_surface_reference(&ps_read, NULL);
       pipe_surface_reference(&ps_tex, NULL);
@@ -791,8 +791,8 @@ void renderer_copy_pixmap(struct xorg_renderer *r,
    dst_loc[3] = height;
    dst_bounds[0] = 0.f;
    dst_bounds[1] = 0.f;
-   dst_bounds[2] = dst->width[0];
-   dst_bounds[3] = dst->height[0];
+   dst_bounds[2] = dst->width0;
+   dst_bounds[3] = dst->height0;
 
    src_loc[0] = sx;
    src_loc[1] = sy;
@@ -800,8 +800,8 @@ void renderer_copy_pixmap(struct xorg_renderer *r,
    src_loc[3] = height;
    src_bounds[0] = 0.f;
    src_bounds[1] = 0.f;
-   src_bounds[2] = src->width[0];
-   src_bounds[3] = src->height[0];
+   src_bounds[2] = src->width0;
+   src_bounds[3] = src->height0;
 
    bound_rect(src_loc, src_bounds, src_shift);
    bound_rect(dst_loc, dst_bounds, dst_shift);
diff --git a/src/gallium/state_trackers/xorg/xorg_tracker.h b/src/gallium/state_trackers/xorg/xorg_tracker.h
index 6130cf6621..20c9259c7b 100644
--- a/src/gallium/state_trackers/xorg/xorg_tracker.h
+++ b/src/gallium/state_trackers/xorg/xorg_tracker.h
@@ -131,7 +131,7 @@ xorg_exa_create_root_texture(ScrnInfoPtr pScrn,
 			     int depth, int bpp);
 
 void *
-xorg_exa_init(ScrnInfoPtr pScrn);
+xorg_exa_init(ScrnInfoPtr pScrn, Bool accel);
 
 void
 xorg_exa_close(ScrnInfoPtr pScrn);
diff --git a/src/gallium/state_trackers/xorg/xorg_xv.c b/src/gallium/state_trackers/xorg/xorg_xv.c
index 2b935c0f73..bb515a0f49 100644
--- a/src/gallium/state_trackers/xorg/xorg_xv.c
+++ b/src/gallium/state_trackers/xorg/xorg_xv.c
@@ -73,10 +73,11 @@ static XF86VideoEncodingRec DummyEncoding[1] = {
    }
 };
 
-#define NUM_IMAGES 2
+#define NUM_IMAGES 3
 static XF86ImageRec Images[NUM_IMAGES] = {
    XVIMAGE_UYVY,
    XVIMAGE_YUY2,
+   XVIMAGE_YV12,
 };
 
 struct xorg_xv_port_priv {
@@ -166,9 +167,9 @@ create_component_texture(struct pipe_context *pipe,
    templ.target = PIPE_TEXTURE_2D;
    templ.format = PIPE_FORMAT_L8_UNORM;
    templ.last_level = 0;
-   templ.width[0] = width;
-   templ.height[0] = height;
-   templ.depth[0] = 1;
+   templ.width0 = width;
+   templ.height0 = height;
+   templ.depth0 = 1;
    pf_get_block(PIPE_FORMAT_L8_UNORM, &templ.block);
    templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
 
@@ -182,18 +183,18 @@ check_yuv_textures(struct xorg_xv_port_priv *priv,  int width, int height)
 {
    struct pipe_texture **dst = priv->yuv[priv->current_set];
    if (!dst[0] ||
-       dst[0]->width[0] != width ||
-       dst[0]->height[0] != height) {
+       dst[0]->width0 != width ||
+       dst[0]->height0 != height) {
       pipe_texture_reference(&dst[0], NULL);
    }
    if (!dst[1] ||
-       dst[1]->width[0] != width ||
-       dst[1]->height[0] != height) {
+       dst[1]->width0 != width ||
+       dst[1]->height0 != height) {
       pipe_texture_reference(&dst[1], NULL);
    }
    if (!dst[2] ||
-       dst[2]->width[0] != width ||
-       dst[2]->height[0] != height) {
+       dst[2]->width0 != width ||
+       dst[2]->height0 != height) {
       pipe_texture_reference(&dst[2], NULL);
    }
 
@@ -256,7 +257,7 @@ copy_packed_data(ScrnInfoPtr pScrn,
    switch (id) {
    case FOURCC_YV12: {
       for (i = 0; i < w; ++i) {
-         for (j = 0; i < h; ++j) {
+         for (j = 0; j < h; ++j) {
             /*XXX use src? */
             y1  = buf[j*w + i];
             u   = buf[(j/2) * (w/2) + i/2 + y_array_size];
@@ -320,8 +321,8 @@ copy_packed_data(ScrnInfoPtr pScrn,
 static void
 setup_vs_video_constants(struct xorg_renderer *r, struct exa_pixmap_priv *dst)
 {
-   int width = dst->tex->width[0];
-   int height = dst->tex->height[0];
+   int width = dst->tex->width0;
+   int height = dst->tex->height0;
    const int param_bytes = 8 * sizeof(float);
    float vs_consts[8] = {
       2.f/width, 2.f/height, 1, 1,
@@ -446,6 +447,11 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
    int x, y, w, h;
    struct exa_pixmap_priv *dst = exaGetPixmapDriverPrivate(pPixmap);
 
+   if (dst && !dst->tex) {
+	xorg_exa_set_shared_usage(pPixmap);
+	pScrn->pScreen->ModifyPixmapHeader(pPixmap, 0, 0, 0, 0, 0, NULL);
+   }
+
    if (!dst || !dst->tex)
       XORG_FALLBACK("Xv destination %s", !dst ? "!dst" : "!dst->tex");
 
@@ -469,6 +475,9 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
    setup_vs_video_constants(pPriv->r, dst);
    setup_fs_video_constants(pPriv->r, hdtv);
 
+   exaMoveInPixmap(pPixmap);
+   DamageDamageRegion(&pPixmap->drawable, dstRegion);
+
    while (nbox--) {
       int box_x1 = pbox->x1;
       int box_y1 = pbox->y1;
@@ -476,8 +485,8 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
       int box_y2 = pbox->y2;
       float diff_x = (float)src_w / (float)dst_w;
       float diff_y = (float)src_h / (float)dst_h;
-      int offset_x = box_x1 - dstX;
-      int offset_y = box_y1 - dstY;
+      int offset_x = box_x1 - dstX + pPixmap->screen_x;
+      int offset_y = box_y1 - dstY + pPixmap->screen_y;
       int offset_w;
       int offset_h;
 
@@ -495,7 +504,7 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
 
       pbox++;
    }
-   DamageDamageRegion(&pPixmap->drawable, dstRegion);
+   DamageRegionProcessPending(&pPixmap->drawable);
 
    return TRUE;
 }
@@ -537,6 +546,7 @@ put_image(ScrnInfoPtr pScrn,
    switch (id) {
    case FOURCC_UYVY:
    case FOURCC_YUY2:
+   case FOURCC_YV12:
    default:
       srcPitch = width << 1;
       break;
@@ -585,6 +595,7 @@ query_image_attributes(ScrnInfoPtr pScrn,
    switch (id) {
    case FOURCC_UYVY:
    case FOURCC_YUY2:
+   case FOURCC_YV12:
    default:
       size = *w << 1;
       if (pitches)
diff --git a/src/gallium/state_trackers/xorg/xvmc/surface.c b/src/gallium/state_trackers/xorg/xvmc/surface.c
index bf9038f356..8cb73f4897 100644
--- a/src/gallium/state_trackers/xorg/xvmc/surface.c
+++ b/src/gallium/state_trackers/xorg/xvmc/surface.c
@@ -103,9 +103,9 @@ CreateOrResizeBackBuffer(struct pipe_video_context *vpipe, unsigned int width, u
    /* XXX: Needs to match the drawable's format? */
    template.format = PIPE_FORMAT_X8R8G8B8_UNORM;
    template.last_level = 0;
-   template.width[0] = width;
-   template.height[0] = height;
-   template.depth[0] = 1;
+   template.width0 = width;
+   template.height0 = height;
+   template.depth0 = 1;
    pf_get_block(template.format, &template.block);
    template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
 
diff --git a/src/gallium/winsys/drm/SConscript b/src/gallium/winsys/drm/SConscript
index a9e9f2682a..9f7b383d2d 100644
--- a/src/gallium/winsys/drm/SConscript
+++ b/src/gallium/winsys/drm/SConscript
@@ -48,6 +48,11 @@ if env['dri']:
 	#	$(INSTALL) -d $(DRI_DRIVER_INSTALL_DIR)
 	#	$(INSTALL) -m 755 $(LIBNAME) $(DRI_DRIVER_INSTALL_DIR)
 
+	if 'vmware' in env['winsys']:
+		SConscript([
+			'vmware/SConscript',
+		])
+
 	if 'intel' in env['winsys']:
 		SConscript([
 			'intel/SConscript',
diff --git a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
index 317dc44d22..d497861324 100644
--- a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
+++ b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
@@ -24,10 +24,10 @@ dri_surface_from_handle(struct drm_api *api, struct pipe_screen *pscreen,
 	tmpl.tex_usage = PIPE_TEXTURE_USAGE_PRIMARY;
 	tmpl.target = PIPE_TEXTURE_2D;
 	tmpl.last_level = 0;
-	tmpl.depth[0] = 1;
+	tmpl.depth0 = 1;
 	tmpl.format = format;
-	tmpl.width[0] = width;
-	tmpl.height[0] = height;
+	tmpl.width0 = width;
+	tmpl.height0 = height;
 	pf_get_block(tmpl.format, &tmpl.block);
 
 	pt = api->texture_from_shared_handle(api, pscreen, &tmpl,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
index 81cd9dc4fb..74afffc9cf 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
@@ -317,9 +317,9 @@ struct pipe_surface *radeon_surface_from_handle(struct radeon_context *radeon_co
     memset(&tmpl, 0, sizeof(tmpl));
     tmpl.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
     tmpl.target = PIPE_TEXTURE_2D;
-    tmpl.width[0] = w;
-    tmpl.height[0] = h;
-    tmpl.depth[0] = 1;
+    tmpl.width0 = w;
+    tmpl.height0 = h;
+    tmpl.depth0 = 1;
     tmpl.format = format;
     pf_get_block(tmpl.format, &tmpl.block);
     tmpl.nblocksx[0] = pf_get_nblocksx(&tmpl.block, w);
diff --git a/src/gallium/winsys/drm/vmware/Makefile b/src/gallium/winsys/drm/vmware/Makefile
new file mode 100644
index 0000000000..2ae6dead5c
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/Makefile
@@ -0,0 +1,12 @@
+# src/gallium/winsys/drm/vmware/Makefile
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+SUBDIRS = core $(GALLIUM_STATE_TRACKERS_DIRS)
+
+default install clean:
+	@for dir in $(SUBDIRS) ; do \
+		if [ -d $$dir ] ; then \
+			(cd $$dir && $(MAKE) $@) || exit 1; \
+		fi \
+	done
diff --git a/src/gallium/winsys/drm/vmware/SConscript b/src/gallium/winsys/drm/vmware/SConscript
new file mode 100644
index 0000000000..06e6d5be9c
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/SConscript
@@ -0,0 +1,11 @@
+Import('*')
+
+SConscript(['core/SConscript',])
+
+if 'mesa' in env['statetrackers']:
+
+    SConscript(['dri/SConscript'])
+
+if 'xorg' in env['statetrackers']:
+
+    SConscript(['xorg/SConscript'])
diff --git a/src/gallium/winsys/drm/vmware/core/Makefile b/src/gallium/winsys/drm/vmware/core/Makefile
new file mode 100644
index 0000000000..a52957c1a5
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/Makefile
@@ -0,0 +1,35 @@
+TOP = ../../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = svgadrm
+
+C_SOURCES = \
+        vmw_buffer.c \
+        vmw_context.c  \
+        vmw_fence.c  \
+        vmw_screen.c  \
+        vmw_screen_dri.c  \
+        vmw_screen_ioctl.c  \
+        vmw_screen_pools.c  \
+        vmw_screen_svga.c  \
+        vmw_surface.c
+
+LIBRARY_INCLUDES = \
+       -I$(TOP)/src/gallium/drivers/svga \
+       -I$(TOP)/src/gallium/drivers/svga/include \
+       -I$(GALLIUM)/src/mesa/drivers/dri/common \
+       -I$(GALLIUM)/include \
+       -I$(GALLIUM)/include/GL/internal \
+       -I$(GALLIUM)/src/mesa \
+       -I$(GALLIUM)/src/mesa/main \
+       -I$(GALLIUM)/src/mesa/glapi \
+       -I$(GALLIUM)/src/egl/main \
+       -I$(GALLIUM)/src/egl/drivers/dri \
+       $(shell pkg-config libdrm --cflags-only-I)
+
+LIBRARY_DEFINES = \
+       -std=gnu99 -fvisibility=hidden \
+       -DHAVE_STDINT_H -D_FILE_OFFSET_BITS=64 \
+       $(shell pkg-config libdrm --cflags-only-other)
+
+include ../../../../Makefile.template
diff --git a/src/gallium/winsys/drm/vmware/core/SConscript b/src/gallium/winsys/drm/vmware/core/SConscript
new file mode 100644
index 0000000000..edaf9458be
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/SConscript
@@ -0,0 +1,39 @@
+Import('*')
+
+env = env.Clone()
+
+if env['gcc']:
+	env.Append(CCFLAGS = ['-fvisibility=hidden'])
+	env.Append(CPPDEFINES = [
+		'HAVE_STDINT_H', 
+		'HAVE_SYS_TYPES_H',
+                '-D_FILE_OFFSET_BITS=64',
+	])
+	
+env.Prepend(CPPPATH = [
+	'include',
+        '#/src/gallium/drivers/svga',
+        '#/src/gallium/drivers/svga/include',
+])
+
+env.Append(CPPDEFINES = [
+])
+
+sources = [
+        'vmw_buffer.c',
+        'vmw_context.c',
+        'vmw_fence.c',
+        'vmw_screen.c',
+        'vmw_screen_dri.c',
+        'vmw_screen_ioctl.c',
+        'vmw_screen_pools.c',
+        'vmw_screen_svga.c',
+        'vmw_surface.c',
+]
+
+svgadrm = env.ConvenienceLibrary(
+	target = 'svgadrm',
+	source = sources,
+)
+
+Export('svgadrm')
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_buffer.c b/src/gallium/winsys/drm/vmware/core/vmw_buffer.c
new file mode 100644
index 0000000000..b812fb59d3
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_buffer.c
@@ -0,0 +1,274 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * SVGA buffer manager for Guest Memory Regions (GMRs).
+ * 
+ * GMRs are used for pixel and vertex data upload/download to/from the virtual
+ * SVGA hardware. There is a limited number of GMRs available, and 
+ * creating/destroying them is also a slow operation so we must suballocate 
+ * them.
+ * 
+ * This file implements a pipebuffer library's buffer manager, so that we can
+ * use pipepbuffer's suballocation, fencing, and debugging facilities with GMRs. 
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#include "svga_cmd.h"
+
+#include "pipe/p_inlines.h"
+#include "util/u_memory.h"
+#include "pipebuffer/pb_buffer.h"
+#include "pipebuffer/pb_bufmgr.h"
+
+#include "svga_winsys.h"
+
+#include "vmw_screen.h"
+#include "vmw_buffer.h"
+
+
+struct vmw_gmr_bufmgr;
+
+
+struct vmw_gmr_buffer
+{
+   struct pb_buffer base;
+   
+   struct vmw_gmr_bufmgr *mgr;
+   
+   struct vmw_region *region;
+   void *map;
+   
+#ifdef DEBUG
+   struct pipe_fence_handle *last_fence;
+#endif
+};
+
+
+extern const struct pb_vtbl vmw_gmr_buffer_vtbl;
+
+
+static INLINE struct vmw_gmr_buffer *
+vmw_gmr_buffer(struct pb_buffer *buf)
+{
+   assert(buf);
+   assert(buf->vtbl == &vmw_gmr_buffer_vtbl);
+   return (struct vmw_gmr_buffer *)buf;
+}
+
+
+struct vmw_gmr_bufmgr
+{
+   struct pb_manager base;
+   
+   struct vmw_winsys_screen *vws;
+};
+
+
+static INLINE struct vmw_gmr_bufmgr *
+vmw_gmr_bufmgr(struct pb_manager *mgr)
+{
+   assert(mgr);
+   return (struct vmw_gmr_bufmgr *)mgr;
+}
+
+
+static void
+vmw_gmr_buffer_destroy(struct pb_buffer *_buf)
+{
+   struct vmw_gmr_buffer *buf = vmw_gmr_buffer(_buf);
+
+#ifdef DEBUG
+   if(buf->last_fence) {
+      struct svga_winsys_screen *sws = &buf->mgr->vws->base;
+      assert(sws->fence_signalled(sws, buf->last_fence, 0) == 0);
+   }
+#endif
+
+   vmw_ioctl_region_unmap(buf->region);
+   
+   vmw_ioctl_region_destroy(buf->region);
+
+   FREE(buf);
+}
+
+
+static void *
+vmw_gmr_buffer_map(struct pb_buffer *_buf,
+               unsigned flags)
+{
+   struct vmw_gmr_buffer *buf = vmw_gmr_buffer(_buf);
+   return buf->map;
+}
+
+
+static void
+vmw_gmr_buffer_unmap(struct pb_buffer *_buf)
+{
+   /* Do nothing */
+   (void)_buf;
+}
+
+
+static void
+vmw_gmr_buffer_get_base_buffer(struct pb_buffer *buf,
+                           struct pb_buffer **base_buf,
+                           unsigned *offset)
+{
+   *base_buf = buf;
+   *offset = 0;
+}
+
+
+static enum pipe_error
+vmw_gmr_buffer_validate( struct pb_buffer *_buf, 
+                         struct pb_validate *vl,
+                         unsigned flags )
+{
+   /* Always pinned */
+   return PIPE_OK;
+}
+
+
+static void
+vmw_gmr_buffer_fence( struct pb_buffer *_buf, 
+                      struct pipe_fence_handle *fence )
+{
+   /* We don't need to do anything, as the pipebuffer library
+    * will take care of delaying the destruction of fenced buffers */  
+#ifdef DEBUG
+   struct vmw_gmr_buffer *buf = vmw_gmr_buffer(_buf);
+   if(fence)
+      buf->last_fence = fence;
+#endif
+}
+
+
+const struct pb_vtbl vmw_gmr_buffer_vtbl = {
+   vmw_gmr_buffer_destroy,
+   vmw_gmr_buffer_map,
+   vmw_gmr_buffer_unmap,
+   vmw_gmr_buffer_validate,
+   vmw_gmr_buffer_fence,
+   vmw_gmr_buffer_get_base_buffer
+};
+
+
+static struct pb_buffer *
+vmw_gmr_bufmgr_create_buffer(struct pb_manager *_mgr,
+                         pb_size size,
+                         const struct pb_desc *desc) 
+{
+   struct vmw_gmr_bufmgr *mgr = vmw_gmr_bufmgr(_mgr);
+   struct vmw_winsys_screen *vws = mgr->vws;
+   struct vmw_gmr_buffer *buf;
+   
+   buf = CALLOC_STRUCT(vmw_gmr_buffer);
+   if(!buf)
+      goto error1;
+
+   pipe_reference_init(&buf->base.base.reference, 1);
+   buf->base.base.alignment = desc->alignment;
+   buf->base.base.usage = desc->usage;
+   buf->base.base.size = size;
+   buf->base.vtbl = &vmw_gmr_buffer_vtbl;
+   buf->mgr = mgr;
+
+   buf->region = vmw_ioctl_region_create(vws, size);
+   if(!buf->region)
+      goto error2;
+	 
+   buf->map = vmw_ioctl_region_map(buf->region);
+   if(!buf->map)
+      goto error3;
+
+   return &buf->base;
+
+error3:
+   vmw_ioctl_region_destroy(buf->region);
+error2:
+   FREE(buf);
+error1:
+   return NULL;
+}
+
+
+static void
+vmw_gmr_bufmgr_flush(struct pb_manager *mgr) 
+{
+   /* No-op */
+}
+
+
+static void
+vmw_gmr_bufmgr_destroy(struct pb_manager *_mgr) 
+{
+   struct vmw_gmr_bufmgr *mgr = vmw_gmr_bufmgr(_mgr);
+   FREE(mgr);
+}
+
+
+struct pb_manager *
+vmw_gmr_bufmgr_create(struct vmw_winsys_screen *vws) 
+{
+   struct vmw_gmr_bufmgr *mgr;
+   
+   mgr = CALLOC_STRUCT(vmw_gmr_bufmgr);
+   if(!mgr)
+      return NULL;
+
+   mgr->base.destroy = vmw_gmr_bufmgr_destroy;
+   mgr->base.create_buffer = vmw_gmr_bufmgr_create_buffer;
+   mgr->base.flush = vmw_gmr_bufmgr_flush;
+   
+   mgr->vws = vws;
+   
+   return &mgr->base;
+}
+
+
+boolean
+vmw_gmr_bufmgr_region_ptr(struct pb_buffer *buf, 
+                          struct SVGAGuestPtr *ptr)
+{
+   struct pb_buffer *base_buf;
+   unsigned offset = 0;
+   struct vmw_gmr_buffer *gmr_buf;
+   
+   pb_get_base_buffer( buf, &base_buf, &offset );
+   
+   gmr_buf = vmw_gmr_buffer(base_buf);
+   if(!gmr_buf)
+      return FALSE;
+   
+   *ptr = vmw_ioctl_region_ptr(gmr_buf->region);
+   
+   ptr->offset += offset;
+   
+   return TRUE;
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_buffer.h b/src/gallium/winsys/drm/vmware/core/vmw_buffer.h
new file mode 100644
index 0000000000..634bdcabd2
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_buffer.h
@@ -0,0 +1,65 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#ifndef VMW_BUFFER_H_
+#define VMW_BUFFER_H_
+
+
+#include "pipe/p_compiler.h"
+
+struct SVGAGuestPtr;
+struct pb_buffer;
+struct pb_manager;
+struct svga_winsys_buffer;
+struct svga_winsys_surface;
+struct vmw_winsys_screen;
+
+
+static INLINE struct pb_buffer *
+vmw_pb_buffer(struct svga_winsys_buffer *buffer)
+{
+   assert(buffer);
+   return (struct pb_buffer *)buffer;
+}
+
+
+static INLINE struct svga_winsys_buffer *
+vmw_svga_winsys_buffer(struct pb_buffer *buffer)
+{
+   assert(buffer);
+   return (struct svga_winsys_buffer *)buffer;
+}
+
+
+struct pb_manager *
+vmw_gmr_bufmgr_create(struct vmw_winsys_screen *vws);
+
+boolean
+vmw_gmr_bufmgr_region_ptr(struct pb_buffer *buf, 
+                          struct SVGAGuestPtr *ptr);
+
+
+#endif /* VMW_BUFFER_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_context.c b/src/gallium/winsys/drm/vmware/core/vmw_context.c
new file mode 100644
index 0000000000..b6997588de
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_context.c
@@ -0,0 +1,297 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "svga_cmd.h"
+
+#include "util/u_debug.h"
+#include "util/u_memory.h"
+#include "util/u_debug_stack.h"
+#include "pipebuffer/pb_buffer.h"
+#include "pipebuffer/pb_validate.h"
+
+#include "svga_winsys.h"
+#include "vmw_context.h"
+#include "vmw_screen.h"
+#include "vmw_buffer.h"
+#include "vmw_surface.h"
+#include "vmw_fence.h"
+
+#define VMW_COMMAND_SIZE (64*1024)
+#define VMW_SURFACE_RELOCS (1024)
+
+#define VMW_MUST_FLUSH_STACK 8
+
+struct vmw_svga_winsys_context
+{
+   struct svga_winsys_context base;
+
+   struct vmw_winsys_screen *vws;
+
+#ifdef DEBUG
+   boolean must_flush;
+   struct debug_stack_frame must_flush_stack[VMW_MUST_FLUSH_STACK];
+#endif
+
+   struct {
+      uint8_t buffer[VMW_COMMAND_SIZE];
+      uint32_t size;
+      uint32_t used;
+      uint32_t reserved;
+   } command;
+
+   struct {
+      struct vmw_svga_winsys_surface *handles[VMW_SURFACE_RELOCS];
+      uint32_t size;
+      uint32_t used;
+      uint32_t staged;
+      uint32_t reserved;
+   } surface;
+
+   struct pb_validate *validate;
+
+   uint32_t last_fence;
+};
+
+
+static INLINE struct vmw_svga_winsys_context *
+vmw_svga_winsys_context(struct svga_winsys_context *swc)
+{
+   assert(swc);
+   return (struct vmw_svga_winsys_context *)swc;
+}
+
+
+static enum pipe_error
+vmw_swc_flush(struct svga_winsys_context *swc,
+              struct pipe_fence_handle **pfence)
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+   struct pipe_fence_handle *fence = NULL;
+   unsigned i;
+   enum pipe_error ret;
+
+   ret = pb_validate_validate(vswc->validate);
+   assert(ret == PIPE_OK);
+   if(ret == PIPE_OK) {
+
+      if (vswc->command.used)
+         vmw_ioctl_command(vswc->vws,
+                           vswc->command.buffer,
+                           vswc->command.used,
+                           &vswc->last_fence);
+
+      fence = vmw_pipe_fence(vswc->last_fence);
+
+      pb_validate_fence(vswc->validate, fence);
+   }
+
+   vswc->command.used = 0;
+   vswc->command.reserved = 0;
+
+   for(i = 0; i < vswc->surface.used + vswc->surface.staged; ++i) {
+      struct vmw_svga_winsys_surface *vsurf =
+	 vswc->surface.handles[i];
+      p_atomic_dec(&vsurf->validated);
+      vmw_svga_winsys_surface_reference(&vswc->surface.handles[i], NULL);
+   }
+
+   vswc->surface.used = 0;
+   vswc->surface.reserved = 0;
+
+#ifdef DEBUG
+   vswc->must_flush = FALSE;
+#endif
+
+   if(pfence)
+      *pfence = fence;
+
+   return ret;
+}
+
+
+static void *
+vmw_swc_reserve(struct svga_winsys_context *swc,
+                uint32_t nr_bytes, uint32_t nr_relocs )
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+
+#ifdef DEBUG
+   /* Check if somebody forgot to check the previous failure */
+   if(vswc->must_flush) {
+      debug_printf("Forgot to flush:\n");
+      debug_backtrace_dump(vswc->must_flush_stack, VMW_MUST_FLUSH_STACK);
+      assert(!vswc->must_flush);
+   }
+#endif
+
+   assert(nr_bytes <= vswc->command.size);
+   if(nr_bytes > vswc->command.size)
+      return NULL;
+
+   if(vswc->command.used + nr_bytes > vswc->command.size ||
+      vswc->surface.used + nr_relocs > vswc->surface.size) {
+#ifdef DEBUG
+      vswc->must_flush = TRUE;
+      debug_backtrace_capture(vswc->must_flush_stack, 1,
+                              VMW_MUST_FLUSH_STACK);
+#endif
+      return NULL;
+   }
+
+   assert(vswc->command.used + nr_bytes <= vswc->command.size);
+   assert(vswc->surface.used + nr_relocs <= vswc->surface.size);
+
+   vswc->command.reserved = nr_bytes;
+   vswc->surface.reserved = nr_relocs;
+   vswc->surface.staged = 0;
+
+   return vswc->command.buffer + vswc->command.used;
+}
+
+
+static void
+vmw_swc_surface_relocation(struct svga_winsys_context *swc,
+                           uint32 *where,
+                           struct svga_winsys_surface *surface,
+                           unsigned flags)
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+   struct vmw_svga_winsys_surface *vsurf;
+
+   if(!surface) {
+      *where = SVGA3D_INVALID_ID;
+      return;
+   }
+
+   assert(vswc->surface.staged < vswc->surface.reserved);
+
+   vsurf = vmw_svga_winsys_surface(surface);
+
+   *where = vsurf->sid;
+
+   vmw_svga_winsys_surface_reference(&vswc->surface.handles[vswc->surface.used + vswc->surface.staged], vsurf);
+   p_atomic_inc(&vsurf->validated);
+   ++vswc->surface.staged;
+}
+
+
+static void
+vmw_swc_region_relocation(struct svga_winsys_context *swc,
+                          struct SVGAGuestPtr *where,
+                          struct svga_winsys_buffer *buffer,
+                          uint32 offset,
+                          unsigned flags)
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+   struct SVGAGuestPtr ptr;
+   struct pb_buffer *buf = vmw_pb_buffer(buffer);
+   enum pipe_error ret;
+
+   if(!vmw_gmr_bufmgr_region_ptr(buf, &ptr))
+      assert(0);
+
+   ptr.offset += offset;
+
+   *where = ptr;
+
+   ret = pb_validate_add_buffer(vswc->validate, buf, flags);
+   /* TODO: Update pipebuffer to reserve buffers and not fail here */
+   assert(ret == PIPE_OK);
+}
+
+
+static void
+vmw_swc_commit(struct svga_winsys_context *swc)
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+
+   assert(vswc->command.reserved);
+   assert(vswc->command.used + vswc->command.reserved <= vswc->command.size);
+   vswc->command.used += vswc->command.reserved;
+   vswc->command.reserved = 0;
+
+   assert(vswc->surface.staged <= vswc->surface.reserved);
+   assert(vswc->surface.used + vswc->surface.staged <= vswc->surface.size);
+   vswc->surface.used += vswc->surface.staged;
+   vswc->surface.staged = 0;
+   vswc->surface.reserved = 0;
+}
+
+
+static void
+vmw_swc_destroy(struct svga_winsys_context *swc)
+{
+   struct vmw_svga_winsys_context *vswc = vmw_svga_winsys_context(swc);
+   unsigned i;
+   for(i = 0; i < vswc->surface.used; ++i) {
+      p_atomic_dec(&vswc->surface.handles[i]->validated);
+      vmw_svga_winsys_surface_reference(&vswc->surface.handles[i], NULL);
+   }
+   pb_validate_destroy(vswc->validate);
+   vmw_ioctl_context_destroy(vswc->vws, swc->cid);
+   FREE(vswc);
+}
+
+
+struct svga_winsys_context *
+vmw_svga_winsys_context_create(struct svga_winsys_screen *sws)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   struct vmw_svga_winsys_context *vswc;
+
+   vswc = CALLOC_STRUCT(vmw_svga_winsys_context);
+   if(!vswc)
+      return NULL;
+
+   vswc->base.destroy = vmw_swc_destroy;
+   vswc->base.reserve = vmw_swc_reserve;
+   vswc->base.surface_relocation = vmw_swc_surface_relocation;
+   vswc->base.region_relocation = vmw_swc_region_relocation;
+   vswc->base.commit = vmw_swc_commit;
+   vswc->base.flush = vmw_swc_flush;
+
+   vswc->base.cid = vmw_ioctl_context_create(vws);
+
+   vswc->vws = vws;
+
+   vswc->command.size = VMW_COMMAND_SIZE;
+   vswc->surface.size = VMW_SURFACE_RELOCS;
+
+   vswc->validate = pb_validate_create();
+   if(!vswc->validate) {
+      FREE(vswc);
+      return NULL;
+   }
+
+   return &vswc->base;
+}
+
+
+struct pipe_context *
+vmw_svga_context_create(struct pipe_screen *screen)
+{
+   return svga_context_create(screen);
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_context.h b/src/gallium/winsys/drm/vmware/core/vmw_context.h
new file mode 100644
index 0000000000..305ce9b5be
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_context.h
@@ -0,0 +1,59 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#ifndef VMW_CONTEXT_H_
+#define VMW_CONTEXT_H_
+
+#include "pipe/p_compiler.h"
+
+struct svga_winsys_screen;
+struct svga_winsys_context;
+struct pipe_context;
+struct pipe_screen;
+
+#define VMW_DEBUG 0
+
+#if VMW_DEBUG
+#define vmw_printf debug_printf
+#define VMW_FUNC  debug_printf("%s\n", __FUNCTION__)
+#else
+#define VMW_FUNC
+#define vmw_printf(...)
+#endif
+
+
+struct svga_winsys_context *
+vmw_svga_winsys_context_create(struct svga_winsys_screen *sws);
+
+struct pipe_context *
+vmw_svga_context_create(struct pipe_screen *screen);
+
+
+#endif /* VMW_CONTEXT_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_fence.c b/src/gallium/winsys/drm/vmware/core/vmw_fence.c
new file mode 100644
index 0000000000..873dd51166
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_fence.c
@@ -0,0 +1,108 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "util/u_memory.h"
+#include "pipebuffer/pb_buffer_fenced.h"
+
+#include "vmw_screen.h"
+#include "vmw_fence.h"
+
+
+
+struct vmw_fence_ops 
+{
+   struct pb_fence_ops base;
+
+   struct vmw_winsys_screen *vws;
+};
+
+
+static INLINE struct vmw_fence_ops *
+vmw_fence_ops(struct pb_fence_ops *ops)
+{
+   assert(ops);
+   return (struct vmw_fence_ops *)ops;
+}
+
+
+static void
+vmw_fence_ops_fence_reference(struct pb_fence_ops *ops,
+                              struct pipe_fence_handle **ptr,
+                              struct pipe_fence_handle *fence)
+{
+   *ptr = fence;
+}
+
+
+static int
+vmw_fence_ops_fence_signalled(struct pb_fence_ops *ops,
+                              struct pipe_fence_handle *fence,
+                              unsigned flag)
+{
+   struct vmw_winsys_screen *vws = vmw_fence_ops(ops)->vws;
+   (void)flag;
+   return vmw_ioctl_fence_signalled(vws, vmw_fence(fence));
+}
+
+
+static int
+vmw_fence_ops_fence_finish(struct pb_fence_ops *ops,
+                           struct pipe_fence_handle *fence,
+                           unsigned flag)
+{
+   struct vmw_winsys_screen *vws = vmw_fence_ops(ops)->vws;
+   (void)flag;
+   return vmw_ioctl_fence_finish(vws, vmw_fence(fence));
+}
+
+
+static void
+vmw_fence_ops_destroy(struct pb_fence_ops *ops)
+{
+   FREE(ops);
+}
+
+
+struct pb_fence_ops *
+vmw_fence_ops_create(struct vmw_winsys_screen *vws) 
+{
+   struct vmw_fence_ops *ops;
+
+   ops = CALLOC_STRUCT(vmw_fence_ops);
+   if(!ops)
+      return NULL;
+
+   ops->base.destroy = &vmw_fence_ops_destroy;
+   ops->base.fence_reference = &vmw_fence_ops_fence_reference;
+   ops->base.fence_signalled = &vmw_fence_ops_fence_signalled;
+   ops->base.fence_finish = &vmw_fence_ops_fence_finish;
+
+   ops->vws = vws;
+
+   return &ops->base;
+}
+
+
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_fence.h b/src/gallium/winsys/drm/vmware/core/vmw_fence.h
new file mode 100644
index 0000000000..5357b4f61d
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_fence.h
@@ -0,0 +1,59 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#ifndef VMW_FENCE_H_
+#define VMW_FENCE_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+struct pipe_fence_handle;
+struct pb_fence_ops;
+struct vmw_winsys_screen;
+
+
+/** Cast from a pipe_fence_handle pointer into a SVGA fence */
+static INLINE uint32_t
+vmw_fence( struct pipe_fence_handle *fence )
+{
+   return (uint32_t)(uintptr_t)fence;
+}
+
+
+/** Cast from a SVGA fence number to pipe_fence_handle pointer */
+static INLINE struct pipe_fence_handle *
+vmw_pipe_fence( uint32_t fence )
+{
+   return (struct pipe_fence_handle *)(uintptr_t)fence;
+}
+
+
+struct pb_fence_ops *
+vmw_fence_ops_create(struct vmw_winsys_screen *vws); 
+
+
+#endif /* VMW_FENCE_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen.c b/src/gallium/winsys/drm/vmware/core/vmw_screen.c
new file mode 100644
index 0000000000..911eec5e25
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen.c
@@ -0,0 +1,74 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "vmw_screen.h"
+
+#include "vmw_context.h"
+
+#include "util/u_memory.h"
+#include "pipe/p_compiler.h"
+
+
+/* Called from vmw_drm_create_screen(), creates and initializes the
+ * vmw_winsys_screen structure, which is the main entity in this
+ * module.
+ */
+struct vmw_winsys_screen *
+vmw_winsys_create( int fd )
+{
+   struct vmw_winsys_screen *vws = CALLOC_STRUCT(vmw_winsys_screen);
+   if (!vws)
+      goto out_no_vws;
+
+   vws->ioctl.drm_fd = fd;
+
+   if (!vmw_ioctl_init(vws))
+      goto out_no_ioctl;
+
+   if(!vmw_pools_init(vws))
+      goto out_no_pools;
+
+   if (!vmw_winsys_screen_init_svga(vws))
+      goto out_no_svga;
+
+   return vws;
+out_no_svga:
+   vmw_pools_cleanup(vws);
+out_no_pools:
+   vmw_ioctl_cleanup(vws);
+out_no_ioctl:
+   FREE(vws);
+out_no_vws:
+   return NULL;
+}
+
+void
+vmw_winsys_destroy(struct vmw_winsys_screen *vws)
+{
+   vmw_pools_cleanup(vws);
+   vmw_ioctl_cleanup(vws);
+   FREE(vws);
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen.h b/src/gallium/winsys/drm/vmware/core/vmw_screen.h
new file mode 100644
index 0000000000..a875107370
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen.h
@@ -0,0 +1,134 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * Common definitions for the VMware SVGA winsys.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#ifndef VMW_SCREEN_H_
+#define VMW_SCREEN_H_
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_state.h"
+
+#include "svga_winsys.h"
+
+struct pb_manager;
+struct vmw_region;
+
+
+struct vmw_winsys_screen
+{
+   struct svga_winsys_screen base;
+
+   struct {
+      volatile uint32_t *fifo_map;
+      uint64_t last_fence;
+      int drm_fd;
+   } ioctl;
+
+   struct {
+      struct pb_manager *gmr;
+      struct pb_manager *gmr_mm;
+      struct pb_manager *gmr_fenced;
+   } pools;
+};
+
+
+static INLINE struct vmw_winsys_screen *
+vmw_winsys_screen(struct svga_winsys_screen *base)
+{
+   return (struct vmw_winsys_screen *)base;
+}
+
+/*  */
+uint32
+vmw_ioctl_context_create(struct vmw_winsys_screen *vws);
+
+void
+vmw_ioctl_context_destroy(struct vmw_winsys_screen *vws,
+                          uint32 cid);
+
+uint32
+vmw_ioctl_surface_create(struct vmw_winsys_screen *vws,
+                              SVGA3dSurfaceFlags flags,
+                              SVGA3dSurfaceFormat format,
+                              SVGA3dSize size,
+                              uint32 numFaces,
+                              uint32 numMipLevels);
+
+void
+vmw_ioctl_surface_destroy(struct vmw_winsys_screen *vws,
+                          uint32 sid);
+
+void
+vmw_ioctl_command(struct vmw_winsys_screen *vws,
+                       void *commands,
+                       uint32_t size,
+                       uint32_t *fence);
+
+struct vmw_region *
+vmw_ioctl_region_create(struct vmw_winsys_screen *vws, uint32_t size);
+
+void
+vmw_ioctl_region_destroy(struct vmw_region *region);
+
+struct SVGAGuestPtr
+vmw_ioctl_region_ptr(struct vmw_region *region);
+
+void *
+vmw_ioctl_region_map(struct vmw_region *region);
+void
+vmw_ioctl_region_unmap(struct vmw_region *region);
+
+
+int
+vmw_ioctl_fence_finish(struct vmw_winsys_screen *vws,
+                       uint32_t fence);
+
+int
+vmw_ioctl_fence_signalled(struct vmw_winsys_screen *vws,
+                          uint32_t fence);
+
+
+/* Initialize parts of vmw_winsys_screen at startup:
+ */
+boolean vmw_ioctl_init(struct vmw_winsys_screen *vws);
+boolean vmw_pools_init(struct vmw_winsys_screen *vws);
+boolean vmw_winsys_screen_init_svga(struct vmw_winsys_screen *vws);
+
+void vmw_ioctl_cleanup(struct vmw_winsys_screen *vws);
+void vmw_pools_cleanup(struct vmw_winsys_screen *vws);
+
+struct vmw_winsys_screen *vmw_winsys_create(int fd);
+void vmw_winsys_destroy(struct vmw_winsys_screen *sws);
+
+
+#endif /* VMW_SCREEN_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
new file mode 100644
index 0000000000..5995eee34b
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
@@ -0,0 +1,371 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_inlines.h"
+#include "util/u_memory.h"
+#include "vmw_screen.h"
+
+#include "trace/tr_drm.h"
+
+#include "vmw_screen.h"
+#include "vmw_surface.h"
+#include "vmw_fence.h"
+#include "vmw_context.h"
+
+#include <state_tracker/dri1_api.h>
+#include <state_tracker/drm_api.h>
+#include <vmwgfx_drm.h>
+#include <xf86drm.h>
+
+#include <stdio.h>
+
+static struct dri1_api dri1_api_hooks;
+static struct dri1_api_version ddx_required = { 0, 1, 0 };
+static struct dri1_api_version ddx_compat = { 0, 0, 0 };
+static struct dri1_api_version dri_required = { 4, 0, 0 };
+static struct dri1_api_version dri_compat = { 4, 0, 0 };
+static struct dri1_api_version drm_required = { 0, 1, 0 };
+static struct dri1_api_version drm_compat = { 0, 0, 0 };
+
+static boolean
+vmw_dri1_check_version(const struct dri1_api_version *cur,
+		       const struct dri1_api_version *required,
+		       const struct dri1_api_version *compat,
+		       const char component[])
+{
+   if (cur->major > required->major && cur->major <= compat->major)
+      return TRUE;
+   if (cur->major == required->major && cur->minor >= required->minor)
+      return TRUE;
+
+   fprintf(stderr, "%s version failure.\n", component);
+   fprintf(stderr, "%s version is %d.%d.%d and this driver can only work\n"
+	   "with versions %d.%d.x through %d.x.x.\n",
+	   component,
+	   cur->major,
+	   cur->minor,
+	   cur->patch_level, required->major, required->minor, compat->major);
+   return FALSE;
+}
+
+/* This is actually the entrypoint to the entire driver, called by the
+ * libGL (or EGL, or ...) code via the drm_api_hooks table at the
+ * bottom of the file.
+ */
+static struct pipe_screen *
+vmw_drm_create_screen(struct drm_api *drm_api,
+                      int fd,
+                      struct drm_create_screen_arg *arg)
+{
+   struct vmw_winsys_screen *vws;
+   struct pipe_screen *screen;
+   struct dri1_create_screen_arg *dri1;
+
+   if (arg != NULL) {
+      switch (arg->mode) {
+      case DRM_CREATE_NORMAL:
+	 break;
+      case DRM_CREATE_DRI1:
+	 dri1 = (struct dri1_create_screen_arg *)arg;
+	 if (!vmw_dri1_check_version(&dri1->ddx_version, &ddx_required,
+				     &ddx_compat, "ddx - driver api"))
+	    return NULL;
+	 if (!vmw_dri1_check_version(&dri1->dri_version, &dri_required,
+				     &dri_compat, "dri info"))
+	    return NULL;
+	 if (!vmw_dri1_check_version(&dri1->drm_version, &drm_required,
+				     &drm_compat, "vmwgfx drm driver"))
+	    return NULL;
+	 dri1->api = &dri1_api_hooks;
+	 break;
+      default:
+	 return NULL;
+      }
+   }
+
+   vws = vmw_winsys_create( fd );
+   if (!vws)
+      goto out_no_vws;
+
+   screen = svga_screen_create( &vws->base );
+   if (!screen)
+      goto out_no_screen;
+
+   return screen;
+
+   /* Failure cases:
+    */
+out_no_screen:
+   vmw_winsys_destroy( vws );
+
+out_no_vws:
+   return NULL;
+}
+
+static INLINE boolean
+vmw_dri1_intersect_src_bbox(struct drm_clip_rect *dst,
+			    int dst_x,
+			    int dst_y,
+			    const struct drm_clip_rect *src,
+			    const struct drm_clip_rect *bbox)
+{
+   int xy1;
+   int xy2;
+
+   xy1 = ((int)src->x1 > (int)bbox->x1 + dst_x) ? src->x1 :
+      (int)bbox->x1 + dst_x;
+   xy2 = ((int)src->x2 < (int)bbox->x2 + dst_x) ? src->x2 :
+      (int)bbox->x2 + dst_x;
+   if (xy1 >= xy2 || xy1 < 0)
+      return FALSE;
+
+   dst->x1 = xy1;
+   dst->x2 = xy2;
+
+   xy1 = ((int)src->y1 > (int)bbox->y1 + dst_y) ? src->y1 :
+      (int)bbox->y1 + dst_y;
+   xy2 = ((int)src->y2 < (int)bbox->y2 + dst_y) ? src->y2 :
+      (int)bbox->y2 + dst_y;
+   if (xy1 >= xy2 || xy1 < 0)
+      return FALSE;
+
+   dst->y1 = xy1;
+   dst->y2 = xy2;
+   return TRUE;
+}
+
+/**
+ * No fancy get-surface-from-sarea stuff here.
+ * Just use the present blit.
+ */
+
+static void
+vmw_dri1_present_locked(struct pipe_context *locked_pipe,
+			struct pipe_surface *surf,
+			const struct drm_clip_rect *rect,
+			unsigned int num_clip,
+			int x_draw, int y_draw,
+			const struct drm_clip_rect *bbox,
+			struct pipe_fence_handle **p_fence)
+{
+   struct svga_winsys_surface *srf =
+      svga_screen_texture_get_winsys_surface(surf->texture);
+   struct vmw_svga_winsys_surface *vsrf = vmw_svga_winsys_surface(srf);
+   struct vmw_winsys_screen *vws =
+      vmw_winsys_screen(svga_winsys_screen(locked_pipe->screen));
+   struct drm_clip_rect clip;
+   int i;
+   struct
+   {
+      SVGA3dCmdHeader header;
+      SVGA3dCmdPresent body;
+      SVGA3dCopyRect rect;
+   } cmd;
+   boolean visible = FALSE;
+   uint32_t fence_seq = 0;
+
+   VMW_FUNC;
+   cmd.header.id = SVGA_3D_CMD_PRESENT;
+   cmd.header.size = sizeof cmd.body + sizeof cmd.rect;
+   cmd.body.sid = vsrf->sid;
+
+   for (i = 0; i < num_clip; ++i) {
+      if (!vmw_dri1_intersect_src_bbox(&clip, x_draw, y_draw, rect++, bbox))
+	 continue;
+
+      cmd.rect.x = clip.x1;
+      cmd.rect.y = clip.y1;
+      cmd.rect.w = clip.x2 - clip.x1;
+      cmd.rect.h = clip.y2 - clip.y1;
+      cmd.rect.srcx = (int)clip.x1 - x_draw;
+      cmd.rect.srcy = (int)clip.y1 - y_draw;
+
+      vmw_printf("%s: Clip %d x %d y %d w %d h %d srcx %d srcy %d\n",
+		   __FUNCTION__,
+		   i,
+		   cmd.rect.x,
+		   cmd.rect.y,
+		   cmd.rect.w, cmd.rect.h, cmd.rect.srcx, cmd.rect.srcy);
+
+      vmw_ioctl_command(vws, &cmd, sizeof cmd.header + cmd.header.size,
+                        &fence_seq);
+      visible = TRUE;
+   }
+
+   *p_fence = (visible) ? vmw_pipe_fence(fence_seq) : NULL;
+   vmw_svga_winsys_surface_reference(&vsrf, NULL);
+}
+
+/**
+ * FIXME: We'd probably want to cache these buffers in the
+ * screen, based on handle.
+ */
+
+static struct pipe_buffer *
+vmw_drm_buffer_from_handle(struct drm_api *drm_api,
+                           struct pipe_screen *screen,
+			   const char *name,
+			   unsigned handle)
+{
+    struct vmw_svga_winsys_surface *vsrf;
+    struct svga_winsys_surface *ssrf;
+    struct vmw_winsys_screen *vws =
+	vmw_winsys_screen(svga_winsys_screen(screen));
+    struct pipe_buffer *buf;
+    union drm_vmw_surface_reference_arg arg;
+    struct drm_vmw_surface_arg *req = &arg.req;
+    struct drm_vmw_surface_create_req *rep = &arg.rep;
+    int ret;
+    int i;
+
+    /**
+     * The vmware device specific handle is the hardware SID.
+     * FIXME: We probably want to move this to the ioctl implementations.
+     */
+
+    memset(&arg, 0, sizeof(arg));
+    req->sid = handle;
+
+    ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_REF_SURFACE,
+			      &arg, sizeof(arg));
+
+    if (ret) {
+	fprintf(stderr, "Failed referencing shared surface. SID %d.\n"
+		"Error %d (%s).\n",
+		handle, ret, strerror(-ret));
+	return NULL;
+    }
+
+    if (rep->mip_levels[0] != 1) {
+	fprintf(stderr, "Incorrect number of mipmap levels on shared surface."
+		" SID %d, levels %d\n",
+		handle, rep->mip_levels[0]);
+	goto out_mip;
+    }
+
+    for (i=1; i < DRM_VMW_MAX_SURFACE_FACES; ++i) {
+	if (rep->mip_levels[i] != 0) {
+	    fprintf(stderr, "Incorrect number of faces levels on shared surface."
+		    " SID %d, face %d present.\n",
+		    handle, i);
+	    goto out_mip;
+	}
+    }
+
+    vsrf = CALLOC_STRUCT(vmw_svga_winsys_surface);
+    if (!vsrf)
+	goto out_mip;
+
+    pipe_reference_init(&vsrf->refcnt, 1);
+    p_atomic_set(&vsrf->validated, 0);
+    vsrf->sid = handle;
+    ssrf = svga_winsys_surface(vsrf);
+    buf = svga_screen_buffer_wrap_surface(screen, rep->format, ssrf);
+    if (!buf)
+	vmw_svga_winsys_surface_reference(&vsrf, NULL);
+
+    return buf;
+  out_mip:
+    vmw_ioctl_surface_destroy(vws, handle);
+    return NULL;
+}
+
+static struct pipe_texture *
+vmw_drm_texture_from_handle(struct drm_api *drm_api,
+			    struct pipe_screen *screen,
+			    struct pipe_texture *templat,
+			    const char *name,
+			    unsigned stride,
+			    unsigned handle)
+{
+    struct pipe_buffer *buffer;
+    buffer = vmw_drm_buffer_from_handle(drm_api, screen, name, handle);
+
+    if (!buffer)
+	return NULL;
+
+    return screen->texture_blanket(screen, templat, &stride, buffer);
+}
+
+static boolean
+vmw_drm_handle_from_buffer(struct drm_api *drm_api,
+                           struct pipe_screen *screen,
+			   struct pipe_buffer *buffer,
+			   unsigned *handle)
+{
+    struct svga_winsys_surface *surface =
+	svga_screen_buffer_get_winsys_surface(buffer);
+    struct vmw_svga_winsys_surface *vsrf;
+
+    if (!surface)
+	return FALSE;
+
+    vsrf = vmw_svga_winsys_surface(surface);
+    *handle = vsrf->sid;
+    vmw_svga_winsys_surface_reference(&vsrf, NULL);
+    return TRUE;
+}
+
+static boolean
+vmw_drm_handle_from_texture(struct drm_api *drm_api,
+			    struct pipe_screen *screen,
+			    struct pipe_texture *texture,
+			    unsigned *stride,
+			    unsigned *handle)
+{
+    struct pipe_buffer *buffer;
+
+    if (!svga_screen_buffer_from_texture(texture, &buffer, stride))
+	return FALSE;
+
+    return vmw_drm_handle_from_buffer(drm_api, screen, buffer, handle);
+}
+
+static struct pipe_context*
+vmw_drm_create_context(struct drm_api *drm_api,
+                       struct pipe_screen *screen)
+{
+   return vmw_svga_context_create(screen);
+}
+
+static struct dri1_api dri1_api_hooks = {
+   .front_srf_locked = NULL,
+   .present_locked = vmw_dri1_present_locked
+};
+
+static struct drm_api vmw_drm_api_hooks = {
+   .create_screen = vmw_drm_create_screen,
+   .create_context = vmw_drm_create_context,
+   .texture_from_shared_handle = vmw_drm_texture_from_handle,
+   .shared_handle_from_texture = vmw_drm_handle_from_texture,
+   .local_handle_from_texture = vmw_drm_handle_from_texture,
+};
+
+struct drm_api* drm_api_create()
+{
+   return trace_drm_create(&vmw_drm_api_hooks);
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_ioctl.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_ioctl.c
new file mode 100644
index 0000000000..51e455f925
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_ioctl.c
@@ -0,0 +1,504 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ *
+ * Wrappers for DRM ioctl functionlaity used by the rest of the vmw
+ * drm winsys.
+ *
+ * Based on svgaicd_escape.c
+ */
+
+
+#include "svga_cmd.h"
+#include "util/u_memory.h"
+#include "util/u_math.h"
+#include "svgadump/svga_dump.h"
+#include "vmw_screen.h"
+#include "vmw_context.h"
+#include "xf86drm.h"
+#include "vmwgfx_drm.h"
+
+#include <sys/mman.h>
+#include <errno.h>
+#include <unistd.h>
+
+struct vmw_region
+{
+   SVGAGuestPtr ptr;
+   uint32_t handle;
+   uint64_t map_handle;
+   void *data;
+   uint32_t map_count;
+   int drm_fd;
+   uint32_t size;
+};
+
+static void
+vmw_check_last_cmd(struct vmw_winsys_screen *vws)
+{
+   static uint32_t buffer[16384];
+   struct drm_vmw_fifo_debug_arg arg;
+   int ret;
+
+   return;
+   memset(&arg, 0, sizeof(arg));
+   arg.debug_buffer = (unsigned long)buffer;
+   arg.debug_buffer_size = 65536;
+
+   ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_FIFO_DEBUG,
+			     &arg, sizeof(arg));
+
+   if (ret) {
+      debug_printf("%s Ioctl error: \"%s\".\n", __FUNCTION__, strerror(-ret));
+      return;
+   }
+
+   if (arg.did_not_fit) {
+      debug_printf("%s Command did not fit completely.\n", __FUNCTION__);
+   }
+
+   svga_dump_commands(buffer, arg.used_size);
+}
+
+static void
+vmw_ioctl_fifo_unmap(struct vmw_winsys_screen *vws, void *mapping)
+{
+   VMW_FUNC;
+   (void)munmap(mapping, getpagesize());
+}
+
+
+static void *
+vmw_ioctl_fifo_map(struct vmw_winsys_screen *vws,
+                   uint32_t fifo_offset )
+{
+   void *map;
+
+   VMW_FUNC;
+
+   map = mmap(NULL, getpagesize(), PROT_READ, MAP_SHARED,
+	      vws->ioctl.drm_fd, fifo_offset);
+
+   if (map == MAP_FAILED) {
+      debug_printf("Map failed %s\n", strerror(errno));
+      return NULL;
+   }
+
+   vmw_printf("Fifo (min) is 0x%08x\n", ((uint32_t *) map)[SVGA_FIFO_MIN]);
+
+   return map;
+}
+
+uint32
+vmw_ioctl_context_create(struct vmw_winsys_screen *vws)
+{
+   struct drm_vmw_context_arg c_arg;
+   int ret;
+
+   VMW_FUNC;
+
+   ret = drmCommandRead(vws->ioctl.drm_fd, DRM_VMW_CREATE_CONTEXT,
+			&c_arg, sizeof(c_arg));
+
+   if (ret)
+      return -1;
+
+   vmw_check_last_cmd(vws);
+   vmw_printf("Context id is %d\n", c_arg.cid);
+
+   return c_arg.cid;
+}
+
+void
+vmw_ioctl_context_destroy(struct vmw_winsys_screen *vws, uint32 cid)
+{
+   struct drm_vmw_context_arg c_arg;
+
+   VMW_FUNC;
+
+   memset(&c_arg, 0, sizeof(c_arg));
+   c_arg.cid = cid;
+
+   (void)drmCommandWrite(vws->ioctl.drm_fd, DRM_VMW_UNREF_CONTEXT,
+			 &c_arg, sizeof(c_arg));
+
+   vmw_check_last_cmd(vws);
+}
+
+uint32
+vmw_ioctl_surface_create(struct vmw_winsys_screen *vws,
+			      SVGA3dSurfaceFlags flags,
+			      SVGA3dSurfaceFormat format,
+			      SVGA3dSize size,
+			      uint32_t numFaces, uint32_t numMipLevels)
+{
+   union drm_vmw_surface_create_arg s_arg;
+   struct drm_vmw_surface_create_req *req = &s_arg.req;
+   struct drm_vmw_surface_arg *rep = &s_arg.rep;
+   struct drm_vmw_size sizes[DRM_VMW_MAX_SURFACE_FACES*
+			     DRM_VMW_MAX_MIP_LEVELS];
+   struct drm_vmw_size *cur_size;
+   uint32_t iFace;
+   uint32_t iMipLevel;
+   int ret;
+
+   vmw_printf("%s flags %d format %d\n", __FUNCTION__, flags, format);
+
+   memset(&s_arg, 0, sizeof(s_arg));
+   req->flags = (uint32_t) flags;
+   req->format = (uint32_t) format;
+   req->shareable = 1;
+
+   assert(numFaces * numMipLevels < DRM_VMW_MAX_SURFACE_FACES*
+	  DRM_VMW_MAX_MIP_LEVELS);
+   cur_size = sizes;
+   for (iFace = 0; iFace < numFaces; ++iFace) {
+      SVGA3dSize mipSize = size;
+
+      req->mip_levels[iFace] = numMipLevels;
+      for (iMipLevel = 0; iMipLevel < numMipLevels; ++iMipLevel) {
+	 cur_size->width = mipSize.width;
+	 cur_size->height = mipSize.height;
+	 cur_size->depth = mipSize.depth;
+	 mipSize.width = MAX2(mipSize.width >> 1, 1);
+	 mipSize.height = MAX2(mipSize.height >> 1, 1);
+	 mipSize.depth = MAX2(mipSize.depth >> 1, 1);
+	 cur_size++;
+      }
+   }
+   for (iFace = numFaces; iFace < SVGA3D_MAX_SURFACE_FACES; ++iFace) {
+      req->mip_levels[iFace] = 0;
+   }
+
+   req->size_addr = (unsigned long)&sizes;
+
+   ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_CREATE_SURFACE,
+			     &s_arg, sizeof(s_arg));
+
+   if (ret)
+      return -1;
+
+   vmw_printf("Surface id is %d\n", rep->sid);
+   vmw_check_last_cmd(vws);
+
+   return rep->sid;
+}
+
+void
+vmw_ioctl_surface_destroy(struct vmw_winsys_screen *vws, uint32 sid)
+{
+   struct drm_vmw_surface_arg s_arg;
+
+   VMW_FUNC;
+
+   memset(&s_arg, 0, sizeof(s_arg));
+   s_arg.sid = sid;
+
+   (void)drmCommandWrite(vws->ioctl.drm_fd, DRM_VMW_UNREF_SURFACE,
+			 &s_arg, sizeof(s_arg));
+   vmw_check_last_cmd(vws);
+
+}
+
+void
+vmw_ioctl_command(struct vmw_winsys_screen *vws, void *commands, uint32_t size,
+		       uint32_t * pfence)
+{
+   struct drm_vmw_execbuf_arg arg;
+   struct drm_vmw_fence_rep rep;
+   int ret;
+
+#ifdef DEBUG
+   {
+      static boolean firsttime = TRUE;
+      static boolean debug = FALSE;
+      static boolean skip = FALSE;
+      if (firsttime) {
+         debug = debug_get_bool_option("SVGA_DUMP_CMD", FALSE);
+         skip = debug_get_bool_option("SVGA_SKIP_CMD", FALSE);
+      }
+      if (debug) {
+         VMW_FUNC;
+         svga_dump_commands(commands, size);
+      }
+      firsttime = FALSE;
+      if (skip) {
+         size = 0;
+      }
+   }
+#endif
+
+   memset(&arg, 0, sizeof(arg));
+   memset(&rep, 0, sizeof(rep));
+
+   rep.error = -EFAULT;
+   arg.fence_rep = (unsigned long)&rep;
+   arg.commands = (unsigned long)commands;
+   arg.command_size = size;
+
+   do {
+       ret = drmCommandWrite(vws->ioctl.drm_fd, DRM_VMW_EXECBUF, &arg, sizeof(arg));
+   } while(ret == -ERESTART);
+   if (ret) {
+      debug_printf("%s error %s.\n", __FUNCTION__, strerror(-ret));
+   }
+   if (rep.error) {
+
+      /*
+       * Kernel has synced and put the last fence sequence in the FIFO
+       * register.
+       */
+
+      if (rep.error == -EFAULT)
+	 rep.fence_seq = vws->ioctl.fifo_map[SVGA_FIFO_FENCE];
+
+      debug_printf("%s Fence error %s.\n", __FUNCTION__,
+		   strerror(-rep.error));
+   }
+
+   vws->ioctl.last_fence = rep.fence_seq;
+
+   if (pfence)
+      *pfence = rep.fence_seq;
+   vmw_check_last_cmd(vws);
+
+}
+
+
+struct vmw_region *
+vmw_ioctl_region_create(struct vmw_winsys_screen *vws, uint32_t size)
+{
+   struct vmw_region *region;
+   union drm_vmw_alloc_dmabuf_arg arg;
+   struct drm_vmw_alloc_dmabuf_req *req = &arg.req;
+   struct drm_vmw_dmabuf_rep *rep = &arg.rep;
+   int ret;
+
+   vmw_printf("%s: size = %u\n", __FUNCTION__, size);
+
+   region = CALLOC_STRUCT(vmw_region);
+   if (!region)
+      goto out_err1;
+
+   memset(&arg, 0, sizeof(arg));
+   req->size = size;
+   do {
+      ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_ALLOC_DMABUF, &arg,
+				sizeof(arg));
+   } while (ret == -ERESTART);
+
+   if (ret) {
+      debug_printf("IOCTL failed %d: %s\n", ret, strerror(-ret));
+      goto out_err1;
+   }
+
+   region->ptr.gmrId = rep->cur_gmr_id;
+   region->ptr.offset = rep->cur_gmr_offset;
+   region->data = NULL;
+   region->handle = rep->handle;
+   region->map_handle = rep->map_handle;
+   region->map_count = 0;
+   region->size = size;
+   region->drm_fd = vws->ioctl.drm_fd;
+
+   vmw_printf("   gmrId = %u, offset = %u\n",
+              region->ptr.gmrId, region->ptr.offset);
+
+   return region;
+
+ out_err1:
+   FREE(region);
+   return NULL;
+}
+
+void
+vmw_ioctl_region_destroy(struct vmw_region *region)
+{
+   struct drm_vmw_unref_dmabuf_arg arg;
+
+   vmw_printf("%s: gmrId = %u, offset = %u\n", __FUNCTION__,
+              region->ptr.gmrId, region->ptr.offset);
+
+   if (region->data) {
+      munmap(region->data, region->size);
+      region->data = NULL;
+   }
+
+   memset(&arg, 0, sizeof(arg));
+   arg.handle = region->handle;
+   drmCommandWrite(region->drm_fd, DRM_VMW_UNREF_DMABUF, &arg, sizeof(arg));
+
+   FREE(region);
+}
+
+SVGAGuestPtr
+vmw_ioctl_region_ptr(struct vmw_region *region)
+{
+   return region->ptr;
+}
+
+void *
+vmw_ioctl_region_map(struct vmw_region *region)
+{
+   void *map;
+
+   vmw_printf("%s: gmrId = %u, offset = %u\n", __FUNCTION__,
+              region->ptr.gmrId, region->ptr.offset);
+
+   if (region->data == NULL) {
+      map = mmap(NULL, region->size, PROT_READ | PROT_WRITE, MAP_SHARED,
+		 region->drm_fd, region->map_handle);
+      if (map == MAP_FAILED) {
+	 debug_printf("%s: Map failed.\n", __FUNCTION__);
+	 return NULL;
+      }
+
+      region->data = map;
+   }
+
+   ++region->map_count;
+
+   return region->data;
+}
+
+void
+vmw_ioctl_region_unmap(struct vmw_region *region)
+{
+   vmw_printf("%s: gmrId = %u, offset = %u\n", __FUNCTION__,
+              region->ptr.gmrId, region->ptr.offset);
+   --region->map_count;
+}
+
+
+int
+vmw_ioctl_fence_signalled(struct vmw_winsys_screen *vws,
+                          uint32_t fence)
+{
+   uint32_t expected;
+   uint32_t current;
+   
+   assert(fence);
+   if(!fence)
+      return 0;
+   
+   expected = fence;
+   current = vws->ioctl.fifo_map[SVGA_FIFO_FENCE];
+   
+   if ((int32)(current - expected) >= 0)
+      return 0; /* fence passed */
+   else
+      return -1;
+}
+
+
+static void
+vmw_ioctl_sync(struct vmw_winsys_screen *vws, 
+		    uint32_t fence)
+{
+   uint32_t cur_fence;
+   struct drm_vmw_fence_wait_arg arg;
+   int ret;
+
+   vmw_printf("%s: fence = %lu\n", __FUNCTION__,
+              (unsigned long)fence);
+
+   cur_fence = vws->ioctl.fifo_map[SVGA_FIFO_FENCE];
+   vmw_printf("%s: Fence id read is 0x%08x\n", __FUNCTION__,
+              (unsigned int)cur_fence);
+
+   if ((cur_fence - fence) < (1 << 24))
+      return;
+
+   memset(&arg, 0, sizeof(arg));
+   arg.sequence = fence;
+
+   do {
+       ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_FENCE_WAIT, &arg,
+				 sizeof(arg));
+   } while (ret == -ERESTART);
+}
+
+
+int
+vmw_ioctl_fence_finish(struct vmw_winsys_screen *vws,
+                       uint32_t fence)
+{
+   assert(fence);
+   
+   if(fence) {
+      if(vmw_ioctl_fence_signalled(vws, fence) != 0) {
+         vmw_ioctl_sync(vws, fence);
+      }
+   }
+   
+   return 0;
+}
+
+
+boolean
+vmw_ioctl_init(struct vmw_winsys_screen *vws)
+{
+   struct drm_vmw_getparam_arg gp_arg;
+   int ret;
+
+   VMW_FUNC;
+
+   memset(&gp_arg, 0, sizeof(gp_arg));
+   gp_arg.param = DRM_VMW_PARAM_FIFO_OFFSET;
+   ret = drmCommandWriteRead(vws->ioctl.drm_fd, DRM_VMW_GET_PARAM,
+			     &gp_arg, sizeof(gp_arg));
+
+   if (ret) {
+      debug_printf("GET_PARAM on %d returned %d: %s\n",
+		   vws->ioctl.drm_fd, ret, strerror(-ret));
+      goto out_err1;
+   }
+
+   vmw_printf("Offset to map is 0x%08llx\n",
+              (unsigned long long)gp_arg.value);
+
+   vws->ioctl.fifo_map = vmw_ioctl_fifo_map(vws, gp_arg.value);
+   if (vws->ioctl.fifo_map == NULL)
+      goto out_err1;
+
+   vmw_printf("%s OK\n", __FUNCTION__);
+   return TRUE;
+
+ out_err1:
+   debug_printf("%s Failed\n", __FUNCTION__);
+   return FALSE;
+}
+
+
+
+void
+vmw_ioctl_cleanup(struct vmw_winsys_screen *vws)
+{
+   VMW_FUNC;
+
+   vmw_ioctl_fifo_unmap(vws, (void *)vws->ioctl.fifo_map);
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_pools.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_pools.c
new file mode 100644
index 0000000000..b1c24b0cb6
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_pools.c
@@ -0,0 +1,79 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "vmw_screen.h"
+
+#include "vmw_buffer.h"
+#include "vmw_fence.h"
+
+#include "pipebuffer/pb_buffer.h"
+#include "pipebuffer/pb_bufmgr.h"
+
+void
+vmw_pools_cleanup(struct vmw_winsys_screen *vws)
+{
+   if(vws->pools.gmr_fenced)
+      vws->pools.gmr_fenced->destroy(vws->pools.gmr_fenced);
+
+   /* gmr_mm pool is already destroyed above */
+
+   if(vws->pools.gmr)
+      vws->pools.gmr->destroy(vws->pools.gmr);
+}
+
+
+boolean
+vmw_pools_init(struct vmw_winsys_screen *vws)
+{
+   vws->pools.gmr = vmw_gmr_bufmgr_create(vws);
+   if(!vws->pools.gmr)
+      goto error;
+
+   vws->pools.gmr_mm = mm_bufmgr_create(vws->pools.gmr,
+                                        16*1024*1024,
+                                        12 /* 4096 alignment */);
+   if(!vws->pools.gmr_mm)
+      goto error;
+
+   vws->pools.gmr_fenced = fenced_bufmgr_create(
+      vws->pools.gmr_mm,
+      vmw_fence_ops_create(vws));
+
+#ifdef DEBUG
+   vws->pools.gmr_fenced = pb_debug_manager_create(vws->pools.gmr_fenced,
+						   4096,
+						   4096);
+#endif
+   if(!vws->pools.gmr_fenced)
+      goto error;
+
+   return TRUE;
+
+error:
+   vmw_pools_cleanup(vws);
+   return FALSE;
+}
+
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_svga.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_svga.c
new file mode 100644
index 0000000000..d7d008859b
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_svga.c
@@ -0,0 +1,295 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * This file implements the SVGA interface into this winsys, defined
+ * in drivers/svga/svga_winsys.h.
+ *
+ * @author Keith Whitwell
+ * @author Jose Fonseca
+ */
+
+
+#include "svga_cmd.h"
+#include "svga3d_caps.h"
+
+#include "pipe/p_inlines.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "pipebuffer/pb_buffer.h"
+#include "pipebuffer/pb_bufmgr.h"
+#include "svga_winsys.h"
+#include "vmw_context.h"
+#include "vmw_screen.h"
+#include "vmw_surface.h"
+#include "vmw_buffer.h"
+#include "vmw_fence.h"
+
+
+static struct svga_winsys_buffer *
+vmw_svga_winsys_buffer_create(struct svga_winsys_screen *sws,
+                              unsigned alignment,
+                              unsigned usage,
+                              unsigned size)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   struct pb_desc desc;
+   struct pb_manager *provider;
+   struct pb_buffer *buffer;
+
+   memset(&desc, 0, sizeof desc);
+   desc.alignment = alignment;
+   desc.usage = usage;
+
+   provider = vws->pools.gmr_fenced;
+
+   assert(provider);
+   buffer = provider->create_buffer(provider, size, &desc);
+   if(!buffer)
+      return NULL;
+
+   return vmw_svga_winsys_buffer(buffer);
+}
+
+
+static void *
+vmw_svga_winsys_buffer_map(struct svga_winsys_screen *sws,
+                           struct svga_winsys_buffer *buf,
+                           unsigned flags)
+{
+   (void)sws;
+   return pb_map(vmw_pb_buffer(buf), flags);
+}
+
+
+static void
+vmw_svga_winsys_buffer_unmap(struct svga_winsys_screen *sws,
+                             struct svga_winsys_buffer *buf)
+{
+   (void)sws;
+   pb_unmap(vmw_pb_buffer(buf));
+}
+
+
+static void
+vmw_svga_winsys_buffer_destroy(struct svga_winsys_screen *sws,
+                               struct svga_winsys_buffer *buf)
+{
+   struct pb_buffer *pbuf = vmw_pb_buffer(buf);
+   (void)sws;
+   pb_reference(&pbuf, NULL);
+}
+
+
+static void
+vmw_svga_winsys_fence_reference(struct svga_winsys_screen *sws,
+                                struct pipe_fence_handle **pdst,
+                                struct pipe_fence_handle *src)
+{
+   (void)sws;
+   *pdst = src;
+}
+
+
+static int
+vmw_svga_winsys_fence_signalled(struct svga_winsys_screen *sws,
+                                struct pipe_fence_handle *fence,
+                                unsigned flag)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   (void)flag;
+   return vmw_ioctl_fence_signalled(vws, vmw_fence(fence));
+}
+
+
+static int
+vmw_svga_winsys_fence_finish(struct svga_winsys_screen *sws,
+                             struct pipe_fence_handle *fence,
+                             unsigned flag)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   (void)flag;
+   return vmw_ioctl_fence_finish(vws, vmw_fence(fence));
+}
+
+
+
+static struct svga_winsys_surface *
+vmw_svga_winsys_surface_create(struct svga_winsys_screen *sws,
+                               SVGA3dSurfaceFlags flags,
+                               SVGA3dSurfaceFormat format,
+                               SVGA3dSize size,
+                               uint32 numFaces,
+                               uint32 numMipLevels)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   struct vmw_svga_winsys_surface *surface;
+
+   surface = CALLOC_STRUCT(vmw_svga_winsys_surface);
+   if(!surface)
+      goto no_surface;
+
+   pipe_reference_init(&surface->refcnt, 1);
+   p_atomic_set(&surface->validated, 0);
+   surface->screen = vws;
+   surface->sid = vmw_ioctl_surface_create(vws,
+                                           flags, format, size,
+                                           numFaces, numMipLevels);
+   if(surface->sid == SVGA3D_INVALID_ID)
+      goto no_sid;
+
+   return svga_winsys_surface(surface);
+
+no_sid:
+   FREE(surface);
+no_surface:
+   return NULL;
+}
+
+
+static boolean
+vmw_svga_winsys_surface_is_flushed(struct svga_winsys_screen *sws,
+                                   struct svga_winsys_surface *surface)
+{
+   struct vmw_svga_winsys_surface *vsurf = vmw_svga_winsys_surface(surface);
+   return (p_atomic_read(&vsurf->validated) == 0);
+}
+
+
+static void
+vmw_svga_winsys_surface_ref(struct svga_winsys_screen *sws,
+			    struct svga_winsys_surface **pDst,
+			    struct svga_winsys_surface *src)
+{
+   struct vmw_svga_winsys_surface *d_vsurf = vmw_svga_winsys_surface(*pDst);
+   struct vmw_svga_winsys_surface *s_vsurf = vmw_svga_winsys_surface(src);
+
+   vmw_svga_winsys_surface_reference(&d_vsurf, s_vsurf);
+   *pDst = svga_winsys_surface(d_vsurf);
+}
+
+
+static void
+vmw_svga_winsys_destroy(struct svga_winsys_screen *sws)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+
+   vmw_winsys_destroy(vws);
+}
+
+
+static boolean
+vmw_svga_winsys_get_cap(struct svga_winsys_screen *sws,
+                        SVGA3dDevCapIndex index,
+                        SVGA3dDevCapResult *result)
+{
+   struct vmw_winsys_screen *vws = vmw_winsys_screen(sws);
+   const uint32 *capsBlock;
+   const SVGA3dCapsRecord *capsRecord = NULL;
+   uint32 offset;
+   const SVGA3dCapPair *capArray;
+   int numCaps, first, last;
+
+   if(!vws->ioctl.fifo_map)
+      return FALSE;
+
+   if(vws->ioctl.fifo_map[SVGA_FIFO_3D_HWVERSION] < SVGA3D_HWVERSION_WS6_B1)
+      return FALSE;
+
+   /*
+    * Search linearly through the caps block records for the specified type.
+    */
+   capsBlock = (const uint32 *)&vws->ioctl.fifo_map[SVGA_FIFO_3D_CAPS];
+   for (offset = 0; capsBlock[offset] != 0; offset += capsBlock[offset]) {
+      const SVGA3dCapsRecord *record;
+      assert(offset < SVGA_FIFO_3D_CAPS_SIZE);
+      record = (const SVGA3dCapsRecord *) (capsBlock + offset);
+      if ((record->header.type >= SVGA3DCAPS_RECORD_DEVCAPS_MIN) &&
+          (record->header.type <= SVGA3DCAPS_RECORD_DEVCAPS_MAX) &&
+          (!capsRecord || (record->header.type > capsRecord->header.type))) {
+         capsRecord = record;
+      }
+   }
+
+   if(!capsRecord)
+      return FALSE;
+
+   /*
+    * Calculate the number of caps from the size of the record.
+    */
+   capArray = (const SVGA3dCapPair *) capsRecord->data;
+   numCaps = (int) ((capsRecord->header.length * sizeof(uint32) -
+                     sizeof capsRecord->header) / (2 * sizeof(uint32)));
+
+   /*
+    * Binary-search for the cap with the specified index.
+    */
+   for (first = 0, last = numCaps - 1; first <= last; ) {
+      int mid = (first + last) / 2;
+
+      if ((SVGA3dDevCapIndex) capArray[mid][0] == index) {
+         /*
+          * Found it.
+          */
+         result->u = capArray[mid][1];
+         return TRUE;
+      }
+
+      /*
+       * Divide and conquer.
+       */
+      if ((SVGA3dDevCapIndex) capArray[mid][0] > index) {
+         last = mid - 1;
+      } else {
+         first = mid + 1;
+      }
+   }
+
+   return FALSE;
+}
+
+
+boolean
+vmw_winsys_screen_init_svga(struct vmw_winsys_screen *vws)
+{
+   vws->base.destroy = vmw_svga_winsys_destroy;
+   vws->base.get_cap = vmw_svga_winsys_get_cap;
+   vws->base.context_create = vmw_svga_winsys_context_create;
+   vws->base.surface_create = vmw_svga_winsys_surface_create;
+   vws->base.surface_is_flushed = vmw_svga_winsys_surface_is_flushed;
+   vws->base.surface_reference = vmw_svga_winsys_surface_ref;
+   vws->base.buffer_create = vmw_svga_winsys_buffer_create;
+   vws->base.buffer_map = vmw_svga_winsys_buffer_map;
+   vws->base.buffer_unmap = vmw_svga_winsys_buffer_unmap;
+   vws->base.buffer_destroy = vmw_svga_winsys_buffer_destroy;
+   vws->base.fence_reference = vmw_svga_winsys_fence_reference;
+   vws->base.fence_signalled = vmw_svga_winsys_fence_signalled;
+   vws->base.fence_finish = vmw_svga_winsys_fence_finish;
+
+   return TRUE;
+}
+
+
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_surface.c b/src/gallium/winsys/drm/vmware/core/vmw_surface.c
new file mode 100644
index 0000000000..c19e556df9
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_surface.c
@@ -0,0 +1,59 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+
+#include "svga_cmd.h"
+#include "util/u_debug.h"
+#include "util/u_memory.h"
+
+#include "vmw_surface.h"
+#include "vmw_screen.h"
+
+void
+vmw_svga_winsys_surface_reference(struct vmw_svga_winsys_surface **pdst,
+                                  struct vmw_svga_winsys_surface *src)
+{
+   struct pipe_reference *src_ref;
+   struct pipe_reference *dst_ref;
+   struct vmw_svga_winsys_surface *dst = *pdst;
+   
+   if(pdst == NULL || *pdst == src)
+      return;
+   
+   src_ref = src ? &src->refcnt : NULL;
+   dst_ref = dst ? &dst->refcnt : NULL;
+
+   if (pipe_reference(&dst_ref, src_ref)) {
+      vmw_ioctl_surface_destroy(dst->screen, dst->sid);
+#ifdef DEBUG
+      /* to detect dangling pointers */
+      assert(p_atomic_read(&dst->validated) == 0);
+      dst->sid = SVGA3D_INVALID_ID;
+#endif
+      FREE(dst);
+   }
+
+   *pdst = src;
+}
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_surface.h b/src/gallium/winsys/drm/vmware/core/vmw_surface.h
new file mode 100644
index 0000000000..340cc1532e
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmw_surface.h
@@ -0,0 +1,79 @@
+/**********************************************************
+ * Copyright 2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * Surfaces for VMware SVGA winsys.
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+
+#ifndef VMW_SURFACE_H_
+#define VMW_SURFACE_H_
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_atomic.h"
+#include "pipe/p_refcnt.h"
+
+#define VMW_MAX_PRESENTS 3
+
+
+
+struct vmw_svga_winsys_surface
+{
+   struct pipe_atomic validated;
+   struct pipe_reference refcnt;
+
+   struct vmw_winsys_screen *screen;
+   uint32_t sid;
+
+   /* FIXME: make this thread safe */
+   unsigned next_present_no;
+   uint32_t present_fences[VMW_MAX_PRESENTS];
+};
+
+
+static INLINE struct svga_winsys_surface *
+svga_winsys_surface(struct vmw_svga_winsys_surface *surf)
+{
+   assert(!surf || surf->sid != SVGA3D_INVALID_ID);
+   return (struct svga_winsys_surface *)surf;
+}
+
+
+static INLINE struct vmw_svga_winsys_surface *
+vmw_svga_winsys_surface(struct svga_winsys_surface *surf)
+{
+   return (struct vmw_svga_winsys_surface *)surf;
+}
+
+
+void
+vmw_svga_winsys_surface_reference(struct vmw_svga_winsys_surface **pdst,
+                                  struct vmw_svga_winsys_surface *src);
+
+#endif /* VMW_SURFACE_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmwgfx_drm.h b/src/gallium/winsys/drm/vmware/core/vmwgfx_drm.h
new file mode 100644
index 0000000000..6705dd4289
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/core/vmwgfx_drm.h
@@ -0,0 +1,442 @@
+/**************************************************************************
+ *
+ * Copyright © 2009 VMware, Inc., Palo Alto, CA., USA
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef _VMWGFX_DRM_H_
+#define _VMWGFX_DRM_H_
+
+#define DRM_VMW_MAX_SURFACE_FACES 6
+#define DRM_VMW_MAX_MIP_LEVELS 24
+
+#define DRM_VMW_EXT_NAME_LEN 128
+
+#define DRM_VMW_GET_PARAM            1
+#define DRM_VMW_EXTENSION            2
+#define DRM_VMW_CREATE_CONTEXT       3
+#define DRM_VMW_UNREF_CONTEXT        4
+#define DRM_VMW_CREATE_SURFACE       5
+#define DRM_VMW_UNREF_SURFACE        6
+#define DRM_VMW_REF_SURFACE          7
+#define DRM_VMW_EXECBUF              8
+#define DRM_VMW_ALLOC_DMABUF         9
+#define DRM_VMW_UNREF_DMABUF         10
+#define DRM_VMW_FIFO_DEBUG           11
+#define DRM_VMW_FENCE_WAIT           12
+
+
+/*************************************************************************/
+/**
+ * DRM_VMW_GET_PARAM - get device information.
+ *
+ * Currently we support only one parameter:
+ *
+ * DRM_VMW_PARAM_FIFO_OFFSET:
+ * Offset to use to map the first page of the FIFO read-only.
+ * The fifo is mapped using the mmap() system call on the drm device.
+ */
+
+#define DRM_VMW_PARAM_FIFO_OFFSET    0
+
+/**
+ * struct drm_vmw_getparam_arg
+ *
+ * @value: Returned value. //Out
+ * @param: Parameter to query. //In.
+ *
+ * Argument to the DRM_VMW_GET_PARAM Ioctl.
+ */
+
+struct drm_vmw_getparam_arg {
+	uint64_t value;
+	uint32_t param;
+	uint32_t pad64;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_EXTENSION - Query device extensions.
+ */
+
+/**
+ * struct drm_vmw_extension_rep
+ *
+ * @exists: The queried extension exists.
+ * @driver_ioctl_offset: Ioctl number of the first ioctl in the extension.
+ * @driver_sarea_offset: Offset to any space in the DRI SAREA
+ * used by the extension.
+ * @major: Major version number of the extension.
+ * @minor: Minor version number of the extension.
+ * @pl: Patch level version number of the extension.
+ *
+ * Output argument to the DRM_VMW_EXTENSION Ioctl.
+ */
+
+struct drm_vmw_extension_rep {
+	int32_t exists;
+	uint32_t driver_ioctl_offset;
+	uint32_t driver_sarea_offset;
+	uint32_t major;
+	uint32_t minor;
+	uint32_t pl;
+	uint32_t pad64;
+};
+
+/**
+ * union drm_vmw_extension_arg
+ *
+ * @extension - Ascii name of the extension to be queried. //In
+ * @rep - Reply as defined above. //Out
+ *
+ * Argument to the DRM_VMW_EXTENSION Ioctl.
+ */
+
+union drm_vmw_extension_arg {
+	char extension[DRM_VMW_EXT_NAME_LEN];
+	struct drm_vmw_extension_rep rep;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_CREATE_CONTEXT - Create a host context.
+ *
+ * Allocates a device unique context id, and queues a create context command
+ * for the host. Does not wait for host completion.
+ */
+
+/**
+ * struct drm_vmw_context_arg
+ *
+ * @cid: Device unique context ID.
+ *
+ * Output argument to the DRM_VMW_CREATE_CONTEXT Ioctl.
+ * Input argument to the DRM_VMW_UNREF_CONTEXT Ioctl.
+ */
+
+struct drm_vmw_context_arg {
+	int32_t cid;
+	uint32_t pad64;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_UNREF_CONTEXT - Create a host context.
+ *
+ * Frees a global context id, and queues a destroy host command for the host.
+ * Does not wait for host completion. The context ID can be used directly
+ * in the command stream and shows up as the same context ID on the host.
+ */
+
+/*************************************************************************/
+/**
+ * DRM_VMW_CREATE_SURFACE - Create a host suface.
+ *
+ * Allocates a device unique surface id, and queues a create surface command
+ * for the host. Does not wait for host completion. The surface ID can be
+ * used directly in the command stream and shows up as the same surface
+ * ID on the host.
+ */
+
+/**
+ * struct drm_wmv_surface_create_req
+ *
+ * @flags: Surface flags as understood by the host.
+ * @format: Surface format as understood by the host.
+ * @mip_levels: Number of mip levels for each face.
+ * An unused face should have 0 encoded.
+ * @size_addr: Address of a user-space array of sruct drm_vmw_size
+ * cast to an uint64_t for 32-64 bit compatibility.
+ * The size of the array should equal the total number of mipmap levels.
+ * @shareable: Boolean whether other clients (as identified by file descriptors)
+ * may reference this surface.
+ *
+ * Input data to the DRM_VMW_CREATE_SURFACE Ioctl.
+ * Output data from the DRM_VMW_REF_SURFACE Ioctl.
+ */
+
+struct drm_vmw_surface_create_req {
+	uint32_t flags;
+	uint32_t format;
+	uint32_t mip_levels[DRM_VMW_MAX_SURFACE_FACES];
+	uint64_t size_addr;
+	int32_t shareable;
+	uint32_t pad64;
+};
+
+/**
+ * struct drm_wmv_surface_arg
+ *
+ * @sid: Surface id of created surface or surface to destroy or reference.
+ *
+ * Output data from the DRM_VMW_CREATE_SURFACE Ioctl.
+ * Input argument to the DRM_VMW_UNREF_SURFACE Ioctl.
+ * Input argument to the DRM_VMW_REF_SURFACE Ioctl.
+ */
+
+struct drm_vmw_surface_arg {
+	int32_t sid;
+	uint32_t pad64;
+};
+
+/**
+ * struct drm_vmw_size ioctl.
+ *
+ * @width - mip level width
+ * @height - mip level height
+ * @depth - mip level depth
+ *
+ * Description of a mip level.
+ * Input data to the DRM_WMW_CREATE_SURFACE Ioctl.
+ */
+
+struct drm_vmw_size {
+	uint32_t width;
+	uint32_t height;
+	uint32_t depth;
+	uint32_t pad64;
+};
+
+/**
+ * union drm_vmw_surface_create_arg
+ *
+ * @rep: Output data as described above.
+ * @req: Input data as described above.
+ *
+ * Argument to the DRM_VMW_CREATE_SURFACE Ioctl.
+ */
+
+union drm_vmw_surface_create_arg {
+	struct drm_vmw_surface_arg rep;
+	struct drm_vmw_surface_create_req req;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_REF_SURFACE - Reference a host surface.
+ *
+ * Puts a reference on a host surface with a give sid, as previously
+ * returned by the DRM_VMW_CREATE_SURFACE ioctl.
+ * A reference will make sure the surface isn't destroyed while we hold
+ * it and will allow the calling client to use the surface ID in the command
+ * stream.
+ *
+ * On successful return, the Ioctl returns the surface information given
+ * in the DRM_VMW_CREATE_SURFACE ioctl.
+ */
+
+/**
+ * union drm_vmw_surface_reference_arg
+ *
+ * @rep: Output data as described above.
+ * @req: Input data as described above.
+ *
+ * Argument to the DRM_VMW_REF_SURFACE Ioctl.
+ */
+
+union drm_vmw_surface_reference_arg {
+	struct drm_vmw_surface_create_req rep;
+	struct drm_vmw_surface_arg req;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_UNREF_SURFACE - Unreference a host surface.
+ *
+ * Clear a reference previously put on a host surface.
+ * When all references are gone, including the one implicitly placed
+ * on creation,
+ * a destroy surface command will be queued for the host.
+ * Does not wait for completion.
+ */
+
+/*************************************************************************/
+/**
+ * DRM_VMW_EXECBUF
+ *
+ * Submit a command buffer for execution on the host, and return a
+ * fence sequence that when signaled, indicates that the command buffer has
+ * executed.
+ */
+
+/**
+ * struct drm_vmw_execbuf_arg
+ *
+ * @commands: User-space address of a command buffer cast to an uint64_t.
+ * @command-size: Size in bytes of the command buffer.
+ * @fence_rep: User-space address of a struct drm_vmw_fence_rep cast to an
+ * uint64_t.
+ *
+ * Argument to the DRM_VMW_EXECBUF Ioctl.
+ */
+
+struct drm_vmw_execbuf_arg {
+	uint64_t commands;
+	uint32_t command_size;
+	uint32_t pad64;
+	uint64_t fence_rep;
+};
+
+/**
+ * struct drm_vmw_fence_rep
+ *
+ * @fence_seq: Fence sequence associated with a command submission.
+ * @error: This member should've been set to -EFAULT on submission.
+ * The following actions should be take on completion:
+ * error == -EFAULT: Fence communication failed. The host is synchronized.
+ * Use the last fence id read from the FIFO fence register.
+ * error != 0 && error != -EFAULT:
+ * Fence submission failed. The host is synchronized. Use the fence_seq member.
+ * error == 0: All is OK, The host may not be synchronized.
+ * Use the fence_seq member.
+ *
+ * Input / Output data to the DRM_VMW_EXECBUF Ioctl.
+ */
+
+struct drm_vmw_fence_rep {
+	uint64_t fence_seq;
+	int32_t error;
+	uint32_t pad64;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_ALLOC_DMABUF
+ *
+ * Allocate a DMA buffer that is visible also to the host.
+ * NOTE: The buffer is
+ * identified by a handle and an offset, which are private to the guest, but
+ * useable in the command stream. The guest kernel may translate these
+ * and patch up the command stream accordingly. In the future, the offset may
+ * be zero at all times, or it may disappear from the interface before it is
+ * fixed.
+ *
+ * The DMA buffer may stay user-space mapped in the guest at all times,
+ * and is thus suitable for sub-allocation.
+ *
+ * DMA buffers are mapped using the mmap() syscall on the drm device.
+ */
+
+/**
+ * struct drm_vmw_alloc_dmabuf_req
+ *
+ * @size: Required minimum size of the buffer.
+ *
+ * Input data to the DRM_VMW_ALLOC_DMABUF Ioctl.
+ */
+
+struct drm_vmw_alloc_dmabuf_req {
+	uint32_t size;
+	uint32_t pad64;
+};
+
+/**
+ * struct drm_vmw_dmabuf_rep
+ *
+ * @map_handle: Offset to use in the mmap() call used to map the buffer.
+ * @handle: Handle unique to this buffer. Used for unreferencing.
+ * @cur_gmr_id: GMR id to use in the command stream when this buffer is
+ * referenced. See not above.
+ * @cur_gmr_offset: Offset to use in the command stream when this buffer is
+ * referenced. See note above.
+ *
+ * Output data from the DRM_VMW_ALLOC_DMABUF Ioctl.
+ */
+
+struct drm_vmw_dmabuf_rep {
+	uint64_t map_handle;
+	uint32_t handle;
+	uint32_t cur_gmr_id;
+	uint32_t cur_gmr_offset;
+	uint32_t pad64;
+};
+
+/**
+ * union drm_vmw_dmabuf_arg
+ *
+ * @req: Input data as described above.
+ * @rep: Output data as described above.
+ *
+ * Argument to the DRM_VMW_ALLOC_DMABUF Ioctl.
+ */
+
+union drm_vmw_alloc_dmabuf_arg {
+	struct drm_vmw_alloc_dmabuf_req req;
+	struct drm_vmw_dmabuf_rep rep;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_UNREF_DMABUF - Free a DMA buffer.
+ *
+ */
+
+/**
+ * struct drm_vmw_unref_dmabuf_arg
+ *
+ * @handle: Handle indicating what buffer to free. Obtained from the
+ * DRM_VMW_ALLOC_DMABUF Ioctl.
+ *
+ * Argument to the DRM_VMW_UNREF_DMABUF Ioctl.
+ */
+
+struct drm_vmw_unref_dmabuf_arg {
+	uint32_t handle;
+	uint32_t pad64;
+};
+
+/*************************************************************************/
+/**
+ * DRM_VMW_FIFO_DEBUG - Get last FIFO submission.
+ *
+ * This IOCTL copies the last FIFO submission directly out of the FIFO buffer.
+ */
+
+/**
+ * struct drm_vmw_fifo_debug_arg
+ *
+ * @debug_buffer: User space address of a debug_buffer cast to an uint64_t //In
+ * @debug_buffer_size: Size in bytes of debug buffer //In
+ * @used_size: Number of bytes copied to the buffer // Out
+ * @did_not_fit: Boolean indicating that the fifo contents did not fit. //Out
+ *
+ * Argument to the DRM_VMW_FIFO_DEBUG Ioctl.
+ */
+
+struct drm_vmw_fifo_debug_arg {
+	uint64_t debug_buffer;
+	uint32_t debug_buffer_size;
+	uint32_t used_size;
+	int32_t did_not_fit;
+	uint32_t pad64;
+};
+
+struct drm_vmw_fence_wait_arg {
+	uint64_t sequence;
+	uint64_t kernel_cookie;
+	int32_t cookie_valid;
+	int32_t pad64;
+};
+
+#endif
diff --git a/src/gallium/winsys/drm/vmware/dri/Makefile b/src/gallium/winsys/drm/vmware/dri/Makefile
new file mode 100644
index 0000000000..8a39e23da6
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/dri/Makefile
@@ -0,0 +1,18 @@
+
+TOP = ../../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = vmwgfx_dri.so
+
+PIPE_DRIVERS = \
+	$(TOP)/src/gallium/state_trackers/dri/libdridrm.a \
+	$(TOP)/src/gallium/winsys/drm/vmware/core/libsvgadrm.a \
+	$(TOP)/src/gallium/drivers/trace/libtrace.a \
+	$(TOP)/src/gallium/drivers/svga/libsvga.a
+
+C_SOURCES = \
+	$(COMMON_GALLIUM_SOURCES)
+
+include ../../Makefile.template
+
+symlinks:
diff --git a/src/gallium/winsys/drm/vmware/dri/SConscript b/src/gallium/winsys/drm/vmware/dri/SConscript
new file mode 100644
index 0000000000..1019f577a5
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/dri/SConscript
@@ -0,0 +1,62 @@
+import os
+import os.path
+
+Import('*')
+
+if env['platform'] == 'linux':
+
+   if env['dri']:
+      env = env.Clone()
+
+      sources = [
+        '#/src/mesa/drivers/dri/common/utils.c',
+        '#/src/mesa/drivers/dri/common/vblank.c',
+        '#/src/mesa/drivers/dri/common/dri_util.c',
+        '#/src/mesa/drivers/dri/common/xmlconfig.c',
+         ]
+   
+      
+      env.ParseConfig('pkg-config --cflags --libs libdrm')
+      
+      env.Prepend(CPPPATH = [
+            '#/src/mesa/state_tracker',
+            '#/src/mesa/drivers/dri/common',
+            '#/src/mesa/main',
+            '#/src/mesa/glapi',
+            '#/src/mesa',
+            '#/include',
+            '#/src/gallium/drivers/svga',
+            '#/src/gallium/drivers/svga/include',
+            ])
+      
+      env.Append(CPPDEFINES = [
+            'HAVE_STDINT_H', 
+            'HAVE_SYS_TYPES_H',
+            ])
+
+      env.Append(CFLAGS = [
+            '-std=gnu99',
+            '-D_FILE_OFFSET_BITS=64',
+            ])
+      
+      env.Prepend(LIBPATH = [
+            ])
+      
+      env.Prepend(LIBS = [
+            trace,
+            st_dri,
+            svgadrm,
+            svga,
+            mesa,
+            auxiliaries,
+            ])
+      
+      # TODO: write a wrapper function http://www.scons.org/wiki/WrapperFunctions
+      env.LoadableModule(
+         target ='vmwgfx_dri.so',
+         source = sources,
+         LIBS = env['LIBS'],
+         SHLIBPREFIX = '',
+         )
+      
+
diff --git a/src/gallium/winsys/drm/vmware/egl/Makefile b/src/gallium/winsys/drm/vmware/egl/Makefile
new file mode 100644
index 0000000000..8e2980c318
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/egl/Makefile
@@ -0,0 +1,18 @@
+
+TOP = ../../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = EGL_svga.so
+
+PIPE_DRIVERS = \
+	$(TOP)/src/gallium/state_trackers/egl/libegldrm.a \
+	$(TOP)/src/gallium/winsys/drm/vmware/core/libsvgadrm.a \
+	$(TOP)/src/gallium/drivers/trace/libtrace.a \
+	$(TOP)/src/gallium/drivers/svga/libsvga.a
+
+C_SOURCES = \
+	$(COMMON_GALLIUM_SOURCES)
+
+include ../../Makefile.template
+
+symlinks:
diff --git a/src/gallium/winsys/drm/vmware/xorg/Makefile b/src/gallium/winsys/drm/vmware/xorg/Makefile
new file mode 100644
index 0000000000..e152263256
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/xorg/Makefile
@@ -0,0 +1,54 @@
+TARGET     = vmwgfx_drv.so
+CFILES     = $(wildcard ./*.c)
+OBJECTS    = $(patsubst ./%.c,./%.o,$(CFILES))
+TOP        = ../../../../../..
+
+include $(TOP)/configs/current
+
+INCLUDES = \
+	$(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto) \
+	-I../gem \
+	-I$(TOP)/src/gallium/include \
+	-I$(TOP)/src/gallium/drivers \
+	-I$(TOP)/src/gallium/auxiliary \
+	-I$(TOP)/src/gallium
+
+LIBS = \
+	$(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a \
+	$(TOP)/src/gallium/winsys/drm/vmware/core/libsvgadrm.a \
+	$(TOP)/src/gallium/drivers/trace/libtrace.a \
+	$(TOP)/src/gallium/drivers/svga/libsvga.a \
+	$(GALLIUM_AUXILIARIES)
+
+DRIVER_DEFINES = \
+	-DHAVE_CONFIG_H
+
+
+#############################################
+
+
+
+all default: $(TARGET)
+
+$(TARGET): $(OBJECTS) Makefile $(TOP)/src/gallium/state_trackers/xorg/libxorgtracker.a $(LIBS)
+	$(TOP)/bin/mklib -noprefix -o $@ \
+	$(OBJECTS) $(LIBS) $(shell pkg-config --libs libdrm) -ldrm_intel
+
+clean:
+	rm -rf $(OBJECTS) $(TARGET)
+
+install:
+	$(INSTALL) -d $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR)
+	$(MINSTALL) -m 755 $(TARGET) $(DESTDIR)/$(XORG_DRIVER_INSTALL_DIR)
+
+
+##############################################
+
+
+.c.o:
+	$(CC) -c $(CFLAGS) $(INCLUDES) $(DRIVER_DEFINES) $< -o $@
+
+
+##############################################
+
+.PHONY	= all clean install
diff --git a/src/gallium/winsys/drm/vmware/xorg/SConscript b/src/gallium/winsys/drm/vmware/xorg/SConscript
new file mode 100644
index 0000000000..ff7b2ed34e
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/xorg/SConscript
@@ -0,0 +1,54 @@
+import os.path
+
+Import('*')
+
+if env['platform'] == 'linux':
+
+	env = env.Clone()
+
+	env.ParseConfig('pkg-config --cflags --libs libdrm xorg-server')
+
+	env.Prepend(CPPPATH = [
+		'#/include',
+		'#/src/gallium',
+		'#/src/mesa',
+		'#/src/gallium/drivers/svga',
+		'#/src/gallium/drivers/svga/include',
+	])
+
+	env.Append(CPPDEFINES = [
+	])
+
+	if env['gcc']:
+		env.Append(CPPDEFINES = [
+			'HAVE_STDINT_H',
+			'HAVE_SYS_TYPES_H',
+		])
+
+	env.Append(CFLAGS = [
+		'-std=gnu99',
+		'-D_FILE_OFFSET_BITS=64',
+	])
+
+	env.Prepend(LIBPATH = [
+	])
+
+	env.Prepend(LIBS = [
+		trace,
+		st_xorg,
+		svgadrm,
+		svga,
+                auxiliaries,
+	])
+
+	sources = [
+		'vmw_xorg.c',
+	]
+
+	# TODO: write a wrapper function http://www.scons.org/wiki/WrapperFunctions
+	env.LoadableModule(
+		target ='vmwgfx_drv.so',
+		source = sources,
+		LIBS = env['LIBS'],
+		SHLIBPREFIX = '',
+	)
diff --git a/src/gallium/winsys/drm/vmware/xorg/vmw_xorg.c b/src/gallium/winsys/drm/vmware/xorg/vmw_xorg.c
new file mode 100644
index 0000000000..3acc110ae7
--- /dev/null
+++ b/src/gallium/winsys/drm/vmware/xorg/vmw_xorg.c
@@ -0,0 +1,150 @@
+/**********************************************************
+ * Copyright 2008-2009 VMware, Inc.  All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ **********************************************************/
+
+/**
+ * @file
+ * Glue file for Xorg State Tracker.
+ *
+ * @author Alan Hourihane <alanh@tungstengraphics.com>
+ * @author Jakob Bornecrantz <wallbraker@gmail.com>
+ */
+
+#include "state_trackers/xorg/xorg_winsys.h"
+
+static void vmw_xorg_identify(int flags);
+static Bool vmw_xorg_pci_probe(DriverPtr driver,
+			       int entity_num,
+			       struct pci_device *device,
+			       intptr_t match_data);
+
+static const struct pci_id_match vmw_xorg_device_match[] = {
+    {0x15ad, PCI_MATCH_ANY, PCI_MATCH_ANY, PCI_MATCH_ANY, 0, 0, 0},
+    {0, 0, 0, 0, 0, 0, 0},
+};
+
+static SymTabRec vmw_xorg_chipsets[] = {
+    {PCI_MATCH_ANY, "VMware SVGA Device"},
+    {-1, NULL}
+};
+
+static PciChipsets vmw_xorg_pci_devices[] = {
+    {PCI_MATCH_ANY, PCI_MATCH_ANY, NULL},
+    {-1, -1, NULL}
+};
+
+static XF86ModuleVersionInfo vmw_xorg_version = {
+    "vmwgfx",
+    MODULEVENDORSTRING,
+    MODINFOSTRING1,
+    MODINFOSTRING2,
+    XORG_VERSION_CURRENT,
+    0, 1, 0, /* major, minor, patch */
+    ABI_CLASS_VIDEODRV,
+    ABI_VIDEODRV_VERSION,
+    MOD_CLASS_VIDEODRV,
+    {0, 0, 0, 0}
+};
+
+/*
+ * Xorg driver exported structures
+ */
+
+_X_EXPORT DriverRec vmwgfx = {
+    1,
+    "vmwgfx",
+    vmw_xorg_identify,
+    NULL,
+    xorg_tracker_available_options,
+    NULL,
+    0,
+    NULL,
+    vmw_xorg_device_match,
+    vmw_xorg_pci_probe
+};
+
+static MODULESETUPPROTO(vmw_xorg_setup);
+
+_X_EXPORT XF86ModuleData vmwgfxModuleData = {
+    &vmw_xorg_version,
+    vmw_xorg_setup,
+    NULL
+};
+
+/*
+ * Xorg driver functions
+ */
+
+static pointer
+vmw_xorg_setup(pointer module, pointer opts, int *errmaj, int *errmin)
+{
+    static Bool setupDone = 0;
+
+    /* This module should be loaded only once, but check to be sure.
+     */
+    if (!setupDone) {
+	setupDone = 1;
+	xf86AddDriver(&vmwgfx, module, HaveDriverFuncs);
+
+	/*
+	 * The return value must be non-NULL on success even though there
+	 * is no TearDownProc.
+	 */
+	return (pointer) 1;
+    } else {
+	if (errmaj)
+	    *errmaj = LDR_ONCEONLY;
+	return NULL;
+    }
+}
+
+static void
+vmw_xorg_identify(int flags)
+{
+    xf86PrintChipsets("vmwgfx", "Driver for VMware SVGA device",
+		      vmw_xorg_chipsets);
+}
+
+static Bool
+vmw_xorg_pci_probe(DriverPtr driver,
+	  int entity_num, struct pci_device *device, intptr_t match_data)
+{
+    ScrnInfoPtr scrn = NULL;
+    EntityInfoPtr entity;
+
+    scrn = xf86ConfigPciEntity(scrn, 0, entity_num, vmw_xorg_pci_devices,
+			       NULL, NULL, NULL, NULL, NULL);
+    if (scrn != NULL) {
+	scrn->driverVersion = 1;
+	scrn->driverName = "vmwgfx";
+	scrn->name = "vmwgfx";
+	scrn->Probe = NULL;
+
+	entity = xf86GetEntityInfo(entity_num);
+
+	/* Use all the functions from the xorg tracker */
+	xorg_tracker_set_functions(scrn);
+    }
+    return scrn != NULL;
+}
diff --git a/src/gallium/winsys/egl_xlib/egl_xlib.c b/src/gallium/winsys/egl_xlib/egl_xlib.c
index d02f825047..599973ce12 100644
--- a/src/gallium/winsys/egl_xlib/egl_xlib.c
+++ b/src/gallium/winsys/egl_xlib/egl_xlib.c
@@ -41,6 +41,7 @@
 #include "pipe/p_state.h"
 #include "pipe/internal/p_winsys_screen.h"
 #include "util/u_memory.h"
+#include "util/u_math.h"
 #include "softpipe/sp_winsys.h"
 #include "softpipe/sp_texture.h"
 
@@ -138,17 +139,6 @@ lookup_context(_EGLContext *ctx)
 }
 
 
-static unsigned int
-bitcount(unsigned int n)
-{
-   unsigned int bits;
-   for (bits = 0; n > 0; n = n >> 1) {
-      bits += (n & 1);
-   }
-   return bits;
-}
-
-
 /**
  * Create the EGLConfigs.  (one per X visual)
  */
@@ -174,9 +164,9 @@ create_configs(struct xlib_egl_display *xdpy, _EGLDisplay *disp)
    for (i = 0; i < num_visuals; i++) {
       _EGLConfig *config = calloc(1, sizeof(_EGLConfig));
       int id = i + 1;
-      int rbits = bitcount(visInfo[i].red_mask);
-      int gbits = bitcount(visInfo[i].green_mask);
-      int bbits = bitcount(visInfo[i].blue_mask);
+      int rbits = util_bitcount(visInfo[i].red_mask);
+      int gbits = util_bitcount(visInfo[i].green_mask);
+      int bbits = util_bitcount(visInfo[i].blue_mask);
       int abits = bbits == 8 ? 8 : 0;
       int zbits = 24;
       int sbits = 8;
diff --git a/src/glu/sgi/libtess/mesh.c b/src/glu/sgi/libtess/mesh.c
index ae861f8642..95f87cdc94 100644
--- a/src/glu/sgi/libtess/mesh.c
+++ b/src/glu/sgi/libtess/mesh.c
@@ -284,7 +284,12 @@ GLUhalfEdge *__gl_meshMakeEdge( GLUmesh *mesh )
   } 
 
   e = MakeEdge( &mesh->eHead );
-  if (e == NULL) return NULL;
+  if (e == NULL) {
+     memFree(newVertex1);
+     memFree(newVertex2);
+     memFree(newFace);
+     return NULL;
+  }
 
   MakeVertex( newVertex1, e, &mesh->vHead );
   MakeVertex( newVertex2, e->Sym, &mesh->vHead );
diff --git a/src/glu/sgi/libutil/mipmap.c b/src/glu/sgi/libutil/mipmap.c
index af647af73c..c5faebd6a3 100644
--- a/src/glu/sgi/libutil/mipmap.c
+++ b/src/glu/sgi/libutil/mipmap.c
@@ -3526,6 +3526,8 @@ gluScaleImage(GLenum format, GLsizei widthin, GLsizei heightin,
     afterImage =
 	malloc(image_size(widthout, heightout, format, GL_UNSIGNED_SHORT));
     if (beforeImage == NULL || afterImage == NULL) {
+	free(beforeImage);
+	free(afterImage);
 	return GLU_OUT_OF_MEMORY;
     }
 
@@ -3760,6 +3762,7 @@ static int bitmapBuild2DMipmaps(GLenum target, GLint internalFormat,
 		    glPixelStorei(GL_UNPACK_SKIP_PIXELS,psm.unpack_skip_pixels);
 		    glPixelStorei(GL_UNPACK_ROW_LENGTH, psm.unpack_row_length);
 		    glPixelStorei(GL_UNPACK_SWAP_BYTES, psm.unpack_swap_bytes);
+		    free(newImage);
 		    return GLU_OUT_OF_MEMORY;
 		}
 	    }
@@ -7382,6 +7385,8 @@ int gluScaleImage3D(GLenum format,
    afterImage = malloc(imageSize3D(widthOut, heightOut, depthOut, format,
 				   GL_UNSIGNED_SHORT));
    if (beforeImage == NULL || afterImage == NULL) {
+       free(beforeImage);
+       free(afterImage);
        return GLU_OUT_OF_MEMORY;
    }
    retrieveStoreModes3D(&psm);
diff --git a/src/glx/x11/drisw_glx.c b/src/glx/x11/drisw_glx.c
index 15e1586658..1866b2cc87 100644
--- a/src/glx/x11/drisw_glx.c
+++ b/src/glx/x11/drisw_glx.c
@@ -250,12 +250,14 @@ driCreateContext(__GLXscreenConfigs * psc,
 {
    __GLXDRIcontextPrivate *pcp, *pcp_shared;
    __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) mode;
-   const __DRIcoreExtension *core = psc->core;
+   const __DRIcoreExtension *core;
    __DRIcontext *shared = NULL;
 
    if (!psc || !psc->driScreen)
       return NULL;
 
+   core = psc->core;
+
    if (shareList) {
       pcp_shared = (__GLXDRIcontextPrivate *) shareList->driContext;
       shared = pcp_shared->driContext;
diff --git a/src/glx/x11/glxextensions.c b/src/glx/x11/glxextensions.c
index 473f46d1e5..6852128e2a 100644
--- a/src/glx/x11/glxextensions.c
+++ b/src/glx/x11/glxextensions.c
@@ -112,6 +112,7 @@ static const struct extension_info known_gl_extensions[] = {
    { GL(ARB_draw_buffers),               VER(0,0), Y, N, N, N },
    { GL(ARB_fragment_program),           VER(0,0), Y, N, N, N },
    { GL(ARB_fragment_program_shadow),    VER(0,0), Y, N, N, N },
+   { GL(ARB_framebuffer_object),         VER(0,0), Y, N, N, N },
    { GL(ARB_imaging),                    VER(0,0), Y, N, N, N },
    { GL(ARB_multisample),                VER(1,3), Y, N, N, N },
    { GL(ARB_multitexture),               VER(1,3), Y, N, N, N },
@@ -150,8 +151,11 @@ static const struct extension_info known_gl_extensions[] = {
    { GL(EXT_depth_bounds_test),          VER(0,0), N, N, N, N },
    { GL(EXT_draw_range_elements),        VER(1,2), Y, N, Y, N },
    { GL(EXT_fog_coord),                  VER(1,4), Y, N, N, N },
+   { GL(EXT_framebuffer_blit),           VER(0,0), Y, N, N, N },
+   { GL(EXT_framebuffer_multisample),    VER(0,0), Y, N, N, N },
    { GL(EXT_framebuffer_object),         VER(0,0), Y, N, N, N },
    { GL(EXT_multi_draw_arrays),          VER(1,4), Y, N, Y, N },
+   { GL(EXT_packed_depth_stencil),       VER(0,0), Y, N, N, N },
    { GL(EXT_packed_pixels),              VER(1,2), Y, N, N, N },
    { GL(EXT_paletted_texture),           VER(0,0), Y, N, N, N },
    { GL(EXT_pixel_buffer_object),        VER(0,0), N, N, N, N },
@@ -209,6 +213,7 @@ static const struct extension_info known_gl_extensions[] = {
    { GL(NV_fragment_program2),           VER(0,0), Y, N, N, N },
    { GL(NV_light_max_exponent),          VER(0,0), Y, N, N, N },
    { GL(NV_multisample_filter_hint),     VER(0,0), Y, N, N, N },
+   { GL(NV_packed_depth_stencil),        VER(0,0), Y, N, N, N },
    { GL(NV_point_sprite),                VER(0,0), Y, N, N, N },
    { GL(NV_texgen_reflection),           VER(0,0), Y, N, N, N },
    { GL(NV_texture_compression_vtc),     VER(0,0), Y, N, N, N },
diff --git a/src/glx/x11/glxextensions.h b/src/glx/x11/glxextensions.h
index 9f1c697487..652c5db1c8 100644
--- a/src/glx/x11/glxextensions.h
+++ b/src/glx/x11/glxextensions.h
@@ -74,6 +74,7 @@ enum
    GL_ARB_draw_buffers_bit,
    GL_ARB_fragment_program_bit,
    GL_ARB_fragment_program_shadow_bit,
+   GL_ARB_framebuffer_object_bit,
    GL_ARB_imaging_bit,
    GL_ARB_multisample_bit,
    GL_ARB_multitexture_bit,
@@ -112,8 +113,11 @@ enum
    GL_EXT_depth_bounds_test_bit,
    GL_EXT_draw_range_elements_bit,
    GL_EXT_fog_coord_bit,
+   GL_EXT_framebuffer_blit_bit,
+   GL_EXT_framebuffer_multisample_bit,
    GL_EXT_framebuffer_object_bit,
    GL_EXT_multi_draw_arrays_bit,
+   GL_EXT_packed_depth_stencil_bit,
    GL_EXT_packed_pixels_bit,
    GL_EXT_paletted_texture_bit,
    GL_EXT_pixel_buffer_object_bit,
@@ -164,6 +168,7 @@ enum
    GL_NV_fragment_program2_bit,
    GL_NV_light_max_exponent_bit,
    GL_NV_multisample_filter_hint_bit,
+   GL_NV_packed_depth_stencil_bit,
    GL_NV_point_sprite_bit,
    GL_NV_texgen_reflection_bit,
    GL_NV_texture_compression_vtc_bit,
diff --git a/src/glx/x11/indirect.c b/src/glx/x11/indirect.c
index e0cafd43bc..ea90ce4463 100644
--- a/src/glx/x11/indirect.c
+++ b/src/glx/x11/indirect.c
@@ -8604,6 +8604,26 @@ __indirect_glDrawBuffersARB(GLsizei n, const GLenum * bufs)
     }
 }
 
+#define X_GLrop_RenderbufferStorageMultisample 4331
+void
+__indirect_glRenderbufferStorageMultisample(GLenum target, GLsizei samples,
+                                            GLenum internalformat,
+                                            GLsizei width, GLsizei height)
+{
+    __GLXcontext *const gc = __glXGetCurrentContext();
+    const GLuint cmdlen = 24;
+    emit_header(gc->pc, X_GLrop_RenderbufferStorageMultisample, cmdlen);
+    (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4);
+    (void) memcpy((void *) (gc->pc + 8), (void *) (&samples), 4);
+    (void) memcpy((void *) (gc->pc + 12), (void *) (&internalformat), 4);
+    (void) memcpy((void *) (gc->pc + 16), (void *) (&width), 4);
+    (void) memcpy((void *) (gc->pc + 20), (void *) (&height), 4);
+    gc->pc += cmdlen;
+    if (__builtin_expect(gc->pc > gc->limit, 0)) {
+        (void) __glXFlushRenderBuffer(gc, gc->pc);
+    }
+}
+
 #define X_GLrop_SampleMaskSGIS 2048
 void
 __indirect_glSampleMaskSGIS(GLclampf value, GLboolean invert)
@@ -10597,6 +10617,52 @@ __indirect_glRenderbufferStorageEXT(GLenum target, GLenum internalformat,
     }
 }
 
+#define X_GLrop_BlitFramebufferEXT 4330
+void
+__indirect_glBlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1,
+                                GLint srcY1, GLint dstX0, GLint dstY0,
+                                GLint dstX1, GLint dstY1, GLbitfield mask,
+                                GLenum filter)
+{
+    __GLXcontext *const gc = __glXGetCurrentContext();
+    const GLuint cmdlen = 44;
+    emit_header(gc->pc, X_GLrop_BlitFramebufferEXT, cmdlen);
+    (void) memcpy((void *) (gc->pc + 4), (void *) (&srcX0), 4);
+    (void) memcpy((void *) (gc->pc + 8), (void *) (&srcY0), 4);
+    (void) memcpy((void *) (gc->pc + 12), (void *) (&srcX1), 4);
+    (void) memcpy((void *) (gc->pc + 16), (void *) (&srcY1), 4);
+    (void) memcpy((void *) (gc->pc + 20), (void *) (&dstX0), 4);
+    (void) memcpy((void *) (gc->pc + 24), (void *) (&dstY0), 4);
+    (void) memcpy((void *) (gc->pc + 28), (void *) (&dstX1), 4);
+    (void) memcpy((void *) (gc->pc + 32), (void *) (&dstY1), 4);
+    (void) memcpy((void *) (gc->pc + 36), (void *) (&mask), 4);
+    (void) memcpy((void *) (gc->pc + 40), (void *) (&filter), 4);
+    gc->pc += cmdlen;
+    if (__builtin_expect(gc->pc > gc->limit, 0)) {
+        (void) __glXFlushRenderBuffer(gc, gc->pc);
+    }
+}
+
+#define X_GLrop_FramebufferTextureLayerEXT 237
+void
+__indirect_glFramebufferTextureLayerEXT(GLenum target, GLenum attachment,
+                                        GLuint texture, GLint level,
+                                        GLint layer)
+{
+    __GLXcontext *const gc = __glXGetCurrentContext();
+    const GLuint cmdlen = 24;
+    emit_header(gc->pc, X_GLrop_FramebufferTextureLayerEXT, cmdlen);
+    (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4);
+    (void) memcpy((void *) (gc->pc + 8), (void *) (&attachment), 4);
+    (void) memcpy((void *) (gc->pc + 12), (void *) (&texture), 4);
+    (void) memcpy((void *) (gc->pc + 16), (void *) (&level), 4);
+    (void) memcpy((void *) (gc->pc + 20), (void *) (&layer), 4);
+    gc->pc += cmdlen;
+    if (__builtin_expect(gc->pc > gc->limit, 0)) {
+        (void) __glXFlushRenderBuffer(gc, gc->pc);
+    }
+}
+
 
 #  undef FASTCALL
 #  undef NOINLINE
diff --git a/src/glx/x11/indirect.h b/src/glx/x11/indirect.h
index 0719a1b302..19a8c0d134 100644
--- a/src/glx/x11/indirect.h
+++ b/src/glx/x11/indirect.h
@@ -572,6 +572,7 @@ extern HIDDEN void __indirect_glGetQueryObjectuivARB(GLuint id, GLenum pname, GL
 extern HIDDEN void __indirect_glGetQueryivARB(GLenum target, GLenum pname, GLint * params);
 extern HIDDEN GLboolean __indirect_glIsQueryARB(GLuint id);
 extern HIDDEN void __indirect_glDrawBuffersARB(GLsizei n, const GLenum * bufs);
+extern HIDDEN void __indirect_glRenderbufferStorageMultisample(GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height);
 extern HIDDEN void __indirect_glSampleMaskSGIS(GLclampf value, GLboolean invert);
 extern HIDDEN void __indirect_glSamplePatternSGIS(GLenum pattern);
 extern HIDDEN void __indirect_glColorPointerEXT(GLint size, GLenum type, GLsizei stride, GLsizei count, const GLvoid * pointer);
@@ -710,6 +711,8 @@ extern HIDDEN void __indirect_glGetRenderbufferParameterivEXT(GLenum target, GLe
 extern HIDDEN GLboolean __indirect_glIsFramebufferEXT(GLuint framebuffer);
 extern HIDDEN GLboolean __indirect_glIsRenderbufferEXT(GLuint renderbuffer);
 extern HIDDEN void __indirect_glRenderbufferStorageEXT(GLenum target, GLenum internalformat, GLsizei width, GLsizei height);
+extern HIDDEN void __indirect_glBlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter);
+extern HIDDEN void __indirect_glFramebufferTextureLayerEXT(GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer);
 
 #  undef HIDDEN
 #  undef FASTCALL
diff --git a/src/glx/x11/indirect_init.c b/src/glx/x11/indirect_init.c
index 852fe712c6..73ca993027 100644
--- a/src/glx/x11/indirect_init.c
+++ b/src/glx/x11/indirect_init.c
@@ -588,6 +588,10 @@ __GLapi * __glXNewIndirectAPI( void )
 
     glAPI->DrawBuffersARB = __indirect_glDrawBuffersARB;
 
+    /*  45. GL_ARB_framebuffer_object */
+
+    glAPI->RenderbufferStorageMultisample = __indirect_glRenderbufferStorageMultisample;
+
     /*  25. GL_SGIS_multisample */
 
     glAPI->SampleMaskSGIS = __indirect_glSampleMaskSGIS;
@@ -768,6 +772,14 @@ __GLapi * __glXNewIndirectAPI( void )
     glAPI->IsRenderbufferEXT = __indirect_glIsRenderbufferEXT;
     glAPI->RenderbufferStorageEXT = __indirect_glRenderbufferStorageEXT;
 
+    /* 316. GL_EXT_framebuffer_blit */
+
+    glAPI->BlitFramebufferEXT = __indirect_glBlitFramebufferEXT;
+
+    /* 329. GL_EXT_texture_array */
+
+    glAPI->FramebufferTextureLayerEXT = __indirect_glFramebufferTextureLayerEXT;
+
     return glAPI;
 }
 
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 1ce9315530..39d25ce3f4 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -60,9 +60,13 @@ SHARED_INCLUDES = \
 
 ##### TARGETS #####
 
-default: symlinks subdirs depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+default: subdirs lib
 
 
+.PHONY: lib
+lib: symlinks subdirs depend
+	@$(MAKE) $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+
 $(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(EXTRA_MODULES) $(WINOBJ) Makefile \
 		$(TOP)/src/mesa/drivers/dri/Makefile.template
 	$(MKLIB) -o $@ -noprefix -linker '$(CC)' -ldflags '$(LDFLAGS)' \
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h
index 89c815722f..95f97414a9 100644
--- a/src/mesa/drivers/dri/common/spantmp2.h
+++ b/src/mesa/drivers/dri/common/spantmp2.h
@@ -82,6 +82,72 @@
       rgba[3] = 0xff;							\
    } while (0)
 
+#elif (SPANTMP_PIXEL_FMT == GL_RGB)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_5_6_5_REV)
+
+/**
+ ** GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+  p = PACK_COLOR_565_REV( color[0], color[1], color[2] )
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_565_REV( r, g, b ))
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      p = p << 8 | p >> 8;						\
+      rgba[0] = ((p >> 8) & 0xf8) * 255 / 0xf8;				\
+      rgba[1] = ((p >> 3) & 0xfc) * 255 / 0xfc;				\
+      rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8;				\
+      rgba[3] = 0xff;							\
+   } while (0)
+
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_4_4_4_4)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_SHORT_4_4_4_4
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+   p = PACK_COLOR_4444_REV(color[3], color[0], color[1], color[2])
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_4444_REV(a, r, g, b))			\
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      rgba[0] = ((p >> 0) & 0xf) * 0x11;				\
+      rgba[1] = ((p >> 12) & 0xf) * 0x11;				\
+      rgba[2] = ((p >> 4) & 0xf) * 0x11;				\
+      rgba[3] = ((p >> 8) & 0xf) * 0x11;				\
+   } while (0)
+
+
 #elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_4_4_4_4_REV)
 
 /**
@@ -147,6 +213,39 @@
       rgba[3] = ((p >> 15) & 0x1) * 0xff;				\
    } while (0)
 
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_1_5_5_5)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+   p = PACK_COLOR_1555_REV(color[3], color[0], color[1], color[2])
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_1555_REV(a, r, g, b))			\
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      p = p << 8 | p >> 8;						\
+      rgba[0] = ((p >> 7) & 0xf8) * 255 / 0xf8;				\
+      rgba[1] = ((p >> 2) & 0xf8) * 255 / 0xf8;				\
+      rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8;				\
+      rgba[3] = ((p >> 15) & 0x1) * 0xff;				\
+   } while (0)
+
 #elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
 
 /**
@@ -202,6 +301,61 @@
      } while (0)
 # endif
 
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_INT_8_8_8_8
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (     buf + (_x) * 4 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLuint *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLuint *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+# define INIT_MONO_PIXEL(p, color)                       \
+     p = PACK_COLOR_8888(color[2], color[1], color[0], color[3]) 
+
+# define WRITE_RGBA(_x, _y, r, g, b, a)                                 \
+   PUT_VALUE(_x, _y, ((r << 8) |					\
+		      (g << 16) |					\
+		      (b << 24) |					\
+		      (a << 0)))
+
+#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p)
+
+# if defined( USE_X86_ASM )
+#  define READ_RGBA(rgba, _x, _y)                                       \
+    do {                                                                \
+       GLuint p = GET_VALUE(_x, _y);					\
+       __asm__ __volatile__( "rorl $8, %0"				\
+				: "=r" (p) : "0" (p) );                 \
+       ((GLuint *)rgba)[0] = p;                                         \
+    } while (0)
+# elif defined( MESA_BIG_ENDIAN )
+    /* On PowerPC with GCC 3.4.2 the shift madness below becomes a single
+     * rotlwi instruction.  It also produces good code on SPARC.
+     */
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = CPU_TO_LE32(GET_VALUE(_x, _y));                      \
+        GLuint t = p;                                                   \
+        *((uint32_t *) rgba) = (t >> 24) | (p << 8);                    \
+     } while (0)
+# else
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = GET_VALUE(_x, _y);					\
+	rgba[0] = (p >>  8) & 0xff;					\
+	rgba[1] = (p >> 16) & 0xff;					\
+	rgba[2] = (p >> 24) & 0xff;					\
+	rgba[3] = (p >>  0) & 0xff;					\
+     } while (0)
+# endif
+
 #else
 #error SPANTMP_PIXEL_FMT must be set to a valid value!
 #endif
diff --git a/src/mesa/drivers/dri/ffb/ffb_vbtmp.h b/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
index 0495d0e276..c548ef3ad5 100644
--- a/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
+++ b/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
@@ -38,11 +38,11 @@ static void TAG(emit)(GLcontext *ctx, GLuint start, GLuint end)
 #endif
 
 #if (IND & (FFB_VB_RGBA_BIT))
-	col0 = VB->ColorPtr[0]->data;
-	col0_stride = VB->ColorPtr[0]->stride;
+	col0 = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+	col0_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 #if (IND & (FFB_VB_TWOSIDE_BIT))
-	col1 = VB->ColorPtr[1]->data;
-	col1_stride = VB->ColorPtr[1]->stride;
+	col1 = VB->BackfaceColorPtr->data;
+	col1_stride = VB->BackfaceColorPtr->stride;
 #endif
 #endif
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_render.c b/src/mesa/drivers/dri/gamma/gamma_render.c
index 1b9fd169f4..a03a93d132 100644
--- a/src/mesa/drivers/dri/gamma/gamma_render.c
+++ b/src/mesa/drivers/dri/gamma/gamma_render.c
@@ -53,13 +53,13 @@ static void gamma_emit( GLcontext *ctx, GLuint start, GLuint end)
    GLfloat (*tc0)[4] = 0;
    GLuint tc0_size = 0;
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    if (ctx->Texture.Unit[0]._ReallyEnabled) {
-      tc0_stride = VB->TexCoordPtr[0]->stride;
-      tc0 = VB->TexCoordPtr[0]->data;
-      tc0_size = VB->TexCoordPtr[0]->size;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0]->data;
+      tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0]->size;
       coord = VB->ClipPtr->data;
       coord_stride = VB->ClipPtr->stride;
    } else {
diff --git a/src/mesa/drivers/dri/i915/i830_texstate.c b/src/mesa/drivers/dri/i915/i830_texstate.c
index f4bbb53b86..c62281d341 100644
--- a/src/mesa/drivers/dri/i915/i830_texstate.c
+++ b/src/mesa/drivers/dri/i915/i830_texstate.c
@@ -27,6 +27,7 @@
 
 #include "main/mtypes.h"
 #include "main/enums.h"
+#include "main/colormac.h"
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
@@ -311,11 +312,10 @@ i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
    CLAMPED_FLOAT_TO_UBYTE(border[2], tObj->BorderColor[2]);
    CLAMPED_FLOAT_TO_UBYTE(border[3], tObj->BorderColor[3]);
 
-   state[I830_TEXREG_TM0S4] = INTEL_PACKCOLOR8888(border[0],
-                                                  border[1],
-                                                  border[2],
-                                                  border[3]);
-
+   state[I830_TEXREG_TM0S4] = PACK_COLOR_8888(border[3],
+					      border[0],
+					      border[1],
+					      border[2]);
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
    /* memcmp was already disabled, but definitely won't work as the
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index a6f554701e..c05c7759ac 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -126,7 +126,7 @@ i830_render_start(struct intel_context *intel)
 
       for (i = 0; i < I830_TEX_UNITS; i++) {
          if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_TEX(i))) {
-            GLuint sz = VB->TexCoordPtr[i]->size;
+            GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
             GLuint emit;
             GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] &
                           ~TEXCOORDTYPE_MASK);
@@ -714,9 +714,6 @@ i830_new_batch(struct intel_context *intel)
 {
    struct i830_context *i830 = i830_context(&intel->ctx);
    i830->state.emitted = 0;
-
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!intel->no_batch_wrap);
 }
 
 static void 
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index d9c61446f5..9e4d318036 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -1301,7 +1301,7 @@ i915ValidateFragmentProgram(struct i915_context *i915)
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
       if (inputsRead & FRAG_BIT_TEX(i)) {
-         int sz = VB->TexCoordPtr[i]->size;
+         int sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
          s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
          s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index d6689af53f..1bacd51aec 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -28,6 +28,7 @@
 #include "main/mtypes.h"
 #include "main/enums.h"
 #include "main/macros.h"
+#include "main/colormac.h"
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
@@ -363,15 +364,15 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
        * R channel, while the hardware uses A.  Spam R into all the channels
        * for safety.
        */
-      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(border[0],
-						   border[0],
-						   border[0],
-						   border[0]);
+      state[I915_TEXREG_SS4] = PACK_COLOR_8888(border[0],
+					       border[0],
+					       border[0],
+					       border[0]);
    } else {
-      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(border[0],
-						   border[1],
-						   border[2],
-						   border[3]);
+      state[I915_TEXREG_SS4] = PACK_COLOR_8888(border[3],
+					       border[0],
+					       border[1],
+					       border[2]);
    }
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index 77ba8d5581..3e7b5101cc 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -667,9 +667,6 @@ i915_new_batch(struct intel_context *intel)
     * difficulties associated with them (physical address requirements).
     */
    i915->state.emitted = 0;
-
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!intel->no_batch_wrap);
 }
 
 static void 
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index bc527aae47..8a3ab39bc2 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -1250,81 +1250,6 @@ union fi
    GLint i;
 };
 
-
-/**********************************************************************/
-/*             Used only with the metaops callbacks.                  */
-/**********************************************************************/
-static void
-intel_meta_draw_poly(struct intel_context *intel,
-                     GLuint n,
-                     GLfloat xy[][2],
-                     GLfloat z, GLuint color, GLfloat tex[][2])
-{
-   union fi *vb;
-   GLint i;
-   unsigned int saved_vertex_size = intel->vertex_size;
-
-   LOCK_HARDWARE(intel);
-
-   intel->vertex_size = 6;
-
-   /* All 3d primitives should be emitted with LOOP_CLIPRECTS,
-    * otherwise the drawing origin (DR4) might not be set correctly.
-    */
-   intel_set_prim(intel, PRIM3D_TRIFAN);
-   vb = (union fi *) intel_get_prim_space(intel, n);
-
-   for (i = 0; i < n; i++) {
-      vb[0].f = xy[i][0];
-      vb[1].f = xy[i][1];
-      vb[2].f = z;
-      vb[3].i = color;
-      vb[4].f = tex[i][0];
-      vb[5].f = tex[i][1];
-      vb += 6;
-   }
-
-   INTEL_FIREVERTICES(intel);
-
-   intel->vertex_size = saved_vertex_size;
-
-   UNLOCK_HARDWARE(intel);
-}
-
-static void
-intel_meta_draw_quad(struct intel_context *intel,
-                     GLfloat x0, GLfloat x1,
-                     GLfloat y0, GLfloat y1,
-                     GLfloat z,
-                     GLuint color,
-                     GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1)
-{
-   GLfloat xy[4][2];
-   GLfloat tex[4][2];
-
-   xy[0][0] = x0;
-   xy[0][1] = y0;
-   xy[1][0] = x1;
-   xy[1][1] = y0;
-   xy[2][0] = x1;
-   xy[2][1] = y1;
-   xy[3][0] = x0;
-   xy[3][1] = y1;
-
-   tex[0][0] = s0;
-   tex[0][1] = t0;
-   tex[1][0] = s1;
-   tex[1][1] = t0;
-   tex[2][0] = s1;
-   tex[2][1] = t1;
-   tex[3][0] = s0;
-   tex[3][1] = t1;
-
-   intel_meta_draw_poly(intel, 4, xy, z, color, tex);
-}
-
-
-
 /**********************************************************************/
 /*                            Initialization.                         */
 /**********************************************************************/
@@ -1333,7 +1258,6 @@ intel_meta_draw_quad(struct intel_context *intel,
 void
 intelInitTriFuncs(GLcontext * ctx)
 {
-   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    static int firsttime = 1;
 
@@ -1350,6 +1274,4 @@ intelInitTriFuncs(GLcontext * ctx)
    tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
    tnl->Driver.Render.CopyPV = _tnl_copy_pv;
    tnl->Driver.Render.Interp = _tnl_interp;
-
-   intel->vtbl.meta_draw_quad = intel_meta_draw_quad;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_clip.c b/src/mesa/drivers/dri/i965/brw_clip.c
index f45dcf8282..dbd10a5297 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.c
+++ b/src/mesa/drivers/dri/i965/brw_clip.c
@@ -78,7 +78,7 @@ static void compile_clip_prog( struct brw_context *brw,
        delta = REG_SIZE;
 
    for (i = 0; i < VERT_RESULT_MAX; i++)
-      if (c.key.attrs & (1<<i)) {
+      if (c.key.attrs & BITFIELD64_BIT(i)) {
 	 c.offset[i] = delta;
 	 delta += ATTR_SIZE;
       }
diff --git a/src/mesa/drivers/dri/i965/brw_clip.h b/src/mesa/drivers/dri/i965/brw_clip.h
index dc550ac793..1c6825510a 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.h
+++ b/src/mesa/drivers/dri/i965/brw_clip.h
@@ -42,7 +42,7 @@
  * up polygon offset and flatshading at this point:
  */
 struct brw_clip_prog_key {
-   GLuint attrs:32;		
+   GLbitfield64 attrs;
    GLuint primitive:4;
    GLuint nr_userclip:3;
    GLuint do_flat_shading:1;
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 87e8a6aad4..e73e21433c 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -172,8 +172,8 @@ struct brw_fragment_program {
    GLuint id;  /**< serial no. to identify frag progs, never re-used */
    GLboolean isGLSL;  /**< really, any IF/LOOP/CONT/BREAK instructions */
 
-   dri_bo *const_buffer;    /** Program constant buffer/surface */
    GLboolean use_const_buffer;
+   dri_bo *const_buffer;    /** Program constant buffer/surface */
 
    /** for debugging, which texture units are referenced */
    GLbitfield tex_units_used;
@@ -231,7 +231,7 @@ struct brw_vs_prog_data {
    GLuint curb_read_length;
    GLuint urb_read_length;
    GLuint total_grf;
-   GLuint outputs_written;
+   GLbitfield64 outputs_written;
    GLuint nr_params;       /**< number of float params/constants */
 
    GLuint inputs_read;
@@ -438,7 +438,6 @@ struct brw_context
    GLuint primitive;
 
    GLboolean emit_state_always;
-   GLboolean no_batch_wrap;
 
    struct {
       struct brw_state_flags dirty;
@@ -742,9 +741,5 @@ brw_fragment_program_const(const struct gl_fragment_program *p)
    return (const struct brw_fragment_program *) p;
 }
 
-
-
-#define DO_SETUP_BITS ((1<<(FRAG_ATTRIB_MAX)) - 1)
-
 #endif
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 8bcb6083f7..7ad860898f 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -145,7 +145,7 @@ static void brw_emit_prim(struct brw_context *brw,
    prim_packet.base_vert_location = prim->basevertex;
 
    /* Can't wrap here, since we rely on the validated state. */
-   brw->no_batch_wrap = GL_TRUE;
+   intel->no_batch_wrap = GL_TRUE;
 
    /* If we're set to always flush, do it before and after the primitive emit.
     * We want to catch both missed flushes that hurt instruction/state cache
@@ -163,7 +163,7 @@ static void brw_emit_prim(struct brw_context *brw,
       intel_batchbuffer_emit_mi_flush(intel->batch);
    }
 
-   brw->no_batch_wrap = GL_FALSE;
+   intel->no_batch_wrap = GL_FALSE;
 }
 
 static void brw_merge_inputs( struct brw_context *brw,
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 271a88dae0..7c796dae93 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -243,14 +243,6 @@ static void wrap_buffers( struct brw_context *brw,
       dri_bo_unreference(brw->vb.upload.bo);
    brw->vb.upload.bo = dri_bo_alloc(brw->intel.bufmgr, "temporary VBO",
 				    size, 1);
-
-   /* Set the internal VBO\ to no-backing-store.  We only use them as a
-    * temporary within a brw_try_draw_prims while the lock is held.
-    */
-   /* DON'T DO THIS AS IF WE HAVE TO RE-ORG MEMORY WE NEED SOMEWHERE WITH
-      FAKE TO PUSH THIS STUFF */
-//   if (!brw->intel.ttm)
-//      dri_bo_fake_disable_backing_store(brw->vb.upload.bo, NULL, NULL);
 }
 
 static void get_space( struct brw_context *brw,
diff --git a/src/mesa/drivers/dri/i965/brw_gs.h b/src/mesa/drivers/dri/i965/brw_gs.h
index e0cf07256b..010c1c2352 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.h
+++ b/src/mesa/drivers/dri/i965/brw_gs.h
@@ -40,7 +40,7 @@
 #define MAX_GS_VERTS (4)	     
 
 struct brw_gs_prog_key {
-   GLuint attrs:32;
+   GLbitfield64 attrs;
    GLuint primitive:4;
    GLuint hint_gs_always:1;
    GLuint pv_first:1;
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index f350cbd74e..968890f7fb 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -61,7 +61,7 @@ static void compile_sf_prog( struct brw_context *brw,
    c.key = *key;
    c.nr_attrs = brw_count_bits(c.key.attrs);
    c.nr_attr_regs = (c.nr_attrs+1)/2;
-   c.nr_setup_attrs = brw_count_bits(c.key.attrs & DO_SETUP_BITS);
+   c.nr_setup_attrs = brw_count_bits(c.key.attrs);
    c.nr_setup_regs = (c.nr_setup_attrs+1)/2;
 
    c.prog_data.urb_read_length = c.nr_attr_regs;
@@ -70,7 +70,7 @@ static void compile_sf_prog( struct brw_context *brw,
    /* Construct map from attribute number to position in the vertex.
     */
    for (i = idx = 0; i < VERT_RESULT_MAX; i++) 
-      if (c.key.attrs & (1<<i)) {
+      if (c.key.attrs & BITFIELD64_BIT(i)) {
 	 c.attr_to_idx[i] = idx;
 	 c.idx_to_attr[idx] = i;
 	 if (i >= VERT_RESULT_TEX0 && i <= VERT_RESULT_TEX7) {
@@ -147,7 +147,7 @@ static void upload_sf_prog(struct brw_context *brw)
        * edgeflag testing here, it is already done in the clip
        * program.
        */
-      if (key.attrs & (1<<VERT_RESULT_EDGE))
+      if (key.attrs & BITFIELD64_BIT(VERT_RESULT_EDGE))
 	 key.primitive = SF_UNFILLED_TRIS;
       else
 	 key.primitive = SF_TRIANGLES;
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index e835229a1d..0ba731fac9 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -45,7 +45,7 @@
 #define SF_UNFILLED_TRIS   3
 
 struct brw_sf_prog_key {
-   GLuint attrs:32;
+   GLbitfield64 attrs;
    GLuint primitive:2;
    GLuint do_twoside_color:1;
    GLuint do_flat_shading:1;
diff --git a/src/mesa/drivers/dri/i965/brw_sf_emit.c b/src/mesa/drivers/dri/i965/brw_sf_emit.c
index 561fcd501b..3eae41ee74 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_emit.c
@@ -56,7 +56,7 @@ static struct brw_reg get_vert_attr(struct brw_sf_compile *c,
 static GLboolean have_attr(struct brw_sf_compile *c,
 			   GLuint attr)
 {
-   return (c->key.attrs & (1<<attr)) ? 1 : 0;
+   return (c->key.attrs & BITFIELD64_BIT(attr)) ? 1 : 0;
 }
 
 /*********************************************************************** 
@@ -122,8 +122,8 @@ static void do_twoside_color( struct brw_sf_compile *c )
  * Flat shading
  */
 
-#define VERT_RESULT_COLOR_BITS ((1<<VERT_RESULT_COL0) | \
-                                 (1<<VERT_RESULT_COL1))
+#define VERT_RESULT_COLOR_BITS (BITFIELD64_BIT(VERT_RESULT_COL0) | \
+				BITFIELD64_BIT(VERT_RESULT_COL1))
 
 static void copy_colors( struct brw_sf_compile *c,
 		     struct brw_reg dst,
@@ -312,8 +312,8 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
 				  GLushort *pc_linear)
 {
    GLboolean is_last_attr = (reg == c->nr_setup_regs - 1);
-   GLuint persp_mask;
-   GLuint linear_mask;
+   GLbitfield64 persp_mask;
+   GLbitfield64 linear_mask;
 
    if (c->key.do_flat_shading || c->key.linear_color)
       persp_mask = c->key.attrs & ~(FRAG_BIT_WPOS |
@@ -331,10 +331,10 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
    *pc_linear = 0;
    *pc = 0xf;
       
-   if (persp_mask & (1 << c->idx_to_attr[reg*2])) 
+   if (persp_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2]))
       *pc_persp = 0xf;
 
-   if (linear_mask & (1 << c->idx_to_attr[reg*2])) 
+   if (linear_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2]))
       *pc_linear = 0xf;
 
    /* Maybe only processs one attribute on the final round:
@@ -342,10 +342,10 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
    if (reg*2+1 < c->nr_setup_attrs) {
       *pc |= 0xf0;
 
-      if (persp_mask & (1 << c->idx_to_attr[reg*2+1])) 
+      if (persp_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2+1]))
 	 *pc_persp |= 0xf0;
 
-      if (linear_mask & (1 << c->idx_to_attr[reg*2+1])) 
+      if (linear_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2+1]))
 	 *pc_linear |= 0xf0;
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index ab6f158080..b129b1f1c3 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -112,6 +112,7 @@ void brw_validate_state(struct brw_context *brw);
 void brw_upload_state(struct brw_context *brw);
 void brw_init_state(struct brw_context *brw);
 void brw_destroy_state(struct brw_context *brw);
+void brw_clear_validated_bos(struct brw_context *brw);
 
 /***********************************************************************
  * brw_state_cache.c
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index f4283bda1b..af8dfb4c15 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -143,7 +143,7 @@ static void xor_states( struct brw_state_flags *result,
    result->cache = a->cache ^ b->cache;
 }
 
-static void
+void
 brw_clear_validated_bos(struct brw_context *brw)
 {
    int i;
diff --git a/src/mesa/drivers/dri/i965/brw_util.c b/src/mesa/drivers/dri/i965/brw_util.c
index ce21aa4869..bba9249d1b 100644
--- a/src/mesa/drivers/dri/i965/brw_util.c
+++ b/src/mesa/drivers/dri/i965/brw_util.c
@@ -35,7 +35,7 @@
 #include "brw_util.h"
 #include "brw_defines.h"
 
-GLuint brw_count_bits( GLuint val )
+GLuint brw_count_bits(uint64_t val)
 {
    GLuint i;
    for (i = 0; val ; val >>= 1)
diff --git a/src/mesa/drivers/dri/i965/brw_util.h b/src/mesa/drivers/dri/i965/brw_util.h
index 33e7cd87e4..04f3175d3e 100644
--- a/src/mesa/drivers/dri/i965/brw_util.h
+++ b/src/mesa/drivers/dri/i965/brw_util.h
@@ -35,7 +35,7 @@
 
 #include "main/mtypes.h"
 
-extern GLuint brw_count_bits( GLuint val );
+extern GLuint brw_count_bits(uint64_t val);
 extern GLuint brw_parameter_list_state_flags(struct gl_program_parameter_list *paramList);
 extern GLuint brw_translate_blend_factor( GLenum factor );
 extern GLuint brw_translate_blend_equation( GLenum mode );
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index f0c79efbd9..fd055e225e 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -56,7 +56,7 @@ static void do_vs_prog( struct brw_context *brw,
    c.prog_data.inputs_read = vp->program.Base.InputsRead;
 
    if (c.key.copy_edgeflag) {
-      c.prog_data.outputs_written |= 1<<VERT_RESULT_EDGE;
+      c.prog_data.outputs_written |= BITFIELD64_BIT(VERT_RESULT_EDGE);
       c.prog_data.inputs_read |= 1<<VERT_ATTRIB_EDGEFLAG;
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index f7b0726636..00efd3443d 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -147,7 +147,7 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
        mrf = 4;
 
    for (i = 0; i < VERT_RESULT_MAX; i++) {
-      if (c->prog_data.outputs_written & (1 << i)) {
+      if (c->prog_data.outputs_written & BITFIELD64_BIT(i)) {
 	 c->nr_outputs++;
          assert(i < Elements(c->regs[PROGRAM_OUTPUT]));
 	 if (i == VERT_RESULT_HPOS) {
@@ -1124,7 +1124,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
    /* Update the header for point size, user clipping flags, and -ve rhw
     * workaround.
     */
-   if ((c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) ||
+   if ((c->prog_data.outputs_written & BITFIELD64_BIT(VERT_RESULT_PSIZ)) ||
        c->key.nr_userclip || BRW_IS_965(p->brw))
    {
       struct brw_reg header1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
@@ -1134,7 +1134,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 
       brw_set_access_mode(p, BRW_ALIGN_16);	
 
-      if (c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) {
+      if (c->prog_data.outputs_written & BITFIELD64_BIT(VERT_RESULT_PSIZ)) {
 	 struct brw_reg psiz = c->regs[PROGRAM_OUTPUT][VERT_RESULT_PSIZ];
 	 brw_MUL(p, brw_writemask(header1, WRITEMASK_W), brw_swizzle1(psiz, 0), brw_imm_f(1<<11));
 	 brw_AND(p, brw_writemask(header1, WRITEMASK_W), header1, brw_imm_ud(0x7ff<<8));
@@ -1224,7 +1224,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
        */
       GLuint i, mrf = 0;
       for (i = c->first_overflow_output; i < VERT_RESULT_MAX; i++) {
-         if (c->prog_data.outputs_written & (1 << i)) {
+         if (c->prog_data.outputs_written & BITFIELD64_BIT(i)) {
             /* move from GRF to MRF */
             brw_MOV(p, brw_message_reg(4+mrf), c->regs[PROGRAM_OUTPUT][i]);
             mrf++;
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 114e6bd018..72749b3859 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -46,7 +46,7 @@
 #include "brw_state.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
-
+#include "brw_wm.h"
 
 static void
 dri_bo_release(dri_bo **bo)
@@ -66,8 +66,14 @@ static void brw_destroy_context( struct intel_context *intel )
 
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
-
-   _mesa_free(brw->wm.compile_data);
+   brw_clear_validated_bos(brw);
+   if (brw->wm.compile_data) {
+      _mesa_free(brw->wm.compile_data->instruction);
+      _mesa_free(brw->wm.compile_data->vreg);
+      _mesa_free(brw->wm.compile_data->refs);
+      _mesa_free(brw->wm.compile_data->prog_instructions);
+      _mesa_free(brw->wm.compile_data);
+   }
 
    for (i = 0; i < brw->state.nr_color_regions; i++)
       intel_region_release(&brw->state.color_regions[i]);
@@ -144,9 +150,6 @@ static void brw_new_batch( struct intel_context *intel )
 {
    struct brw_context *brw = brw_context(&intel->ctx);
 
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!brw->no_batch_wrap);
-
    brw->curbe.need_new_bo = GL_TRUE;
 
    /* Mark all context state as needing to be re-emitted.
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index d8971321f3..6895f64410 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -157,7 +157,6 @@ static void do_wm_prog( struct brw_context *brw,
 					  sizeof(*c->prog_instructions));
       c->vreg = _mesa_calloc(BRW_WM_MAX_VREG * sizeof(*c->vreg));
       c->refs = _mesa_calloc(BRW_WM_MAX_REF * sizeof(*c->refs));
-      c->vreg = _mesa_calloc(BRW_WM_MAX_VREG * sizeof(*c->vreg));
    } else {
       void *instruction = c->instruction;
       void *prog_instructions = c->prog_instructions;
@@ -231,7 +230,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
        ctx->Color.AlphaEnabled)
       lookup |= IZ_PS_KILL_ALPHATEST_BIT;
 
-   if (fp->program.Base.OutputsWritten & (1<<FRAG_RESULT_DEPTH))
+   if (fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH))
       lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
 
    /* _NEW_DEPTH */
@@ -347,7 +346,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
    key->nr_color_regions = brw->state.nr_color_regions;
 
    /* CACHE_NEW_VS_PROG */
-   key->vp_outputs_written = brw->vs.prog_data->outputs_written & DO_SETUP_BITS;
+   key->vp_outputs_written = brw->vs.prog_data->outputs_written;
 
    /* The unique fragment program ID */
    key->program_string_id = fp->id;
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index b3c05eb0ad..b9b987ea70 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -76,10 +76,10 @@ struct brw_wm_prog_key {
 
    GLushort tex_swizzles[BRW_MAX_TEX_UNIT];
 
-   GLuint program_string_id:32;
    GLushort origin_x, origin_y;
    GLushort drawable_height;
-   GLuint vp_outputs_written;
+   GLbitfield64 vp_outputs_written;
+   GLuint program_string_id:32;
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index 1c4f62ba48..3737faf26f 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -138,7 +138,6 @@ static struct prog_dst_register dst_reg(GLuint file, GLuint idx)
    reg.CondMask = COND_TR;
    reg.CondSwizzle = 0;
    reg.CondSrc = 0;
-   reg.pad = 0;
    return reg;
 }
 
@@ -986,7 +985,7 @@ static void emit_render_target_writes( struct brw_wm_compile *c )
    }
    else {
       /* if gl_FragData[0] is written, use it, else use gl_FragColor */
-      if (c->fp->program.Base.OutputsWritten & (1 << FRAG_RESULT_DATA0))
+      if (c->fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0))
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0);
       else 
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index 3ab446164c..e8c2cb66ec 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -371,7 +371,7 @@ static void prealloc_reg(struct brw_wm_compile *c)
 	  for (j = 0; j < 4; j++)
 	     set_reg(c, PROGRAM_PAYLOAD, fp_input, j, reg);
        }
-       if (c->key.vp_outputs_written & (1 << i)) {
+       if (c->key.vp_outputs_written & BITFIELD64_BIT(i)) {
 	  reg_index += 2;
        }
     }
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass2.c b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
index 6faea018fb..31303febf0 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass2.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
@@ -82,8 +82,8 @@ static void init_registers( struct brw_wm_compile *c )
    for (j = 0; j < c->nr_creg; j++) 
       prealloc_reg(c, &c->creg[j], i++);
 
-   for (j = 0; j < FRAG_ATTRIB_MAX; j++) {
-      if (c->key.vp_outputs_written & (1<<j)) {
+   for (j = 0; j < VERT_RESULT_MAX; j++) {
+      if (c->key.vp_outputs_written & BITFIELD64_BIT(j)) {
 	 int fp_index;
 
 	 if (j >= VERT_RESULT_VAR0)
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 361f91292b..f89ed9bce7 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -106,7 +106,7 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
 
    /* as far as we can tell */
    key->computes_depth =
-      (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) != 0;
+      (fp->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) != 0;
    /* BRW_NEW_DEPTH_BUFFER
     * Override for NULL depthbuffer case, required by the Pixel Shader Computed
     * Depth field.
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index e2f0a383e7..47035cc6fc 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -86,6 +86,9 @@ static GLuint translate_tex_format( gl_format mesa_format,
    case MESA_FORMAT_AL88:
       return BRW_SURFACEFORMAT_L8A8_UNORM;
 
+   case MESA_FORMAT_AL1616:
+      return BRW_SURFACEFORMAT_L16A16_UNORM;
+
    case MESA_FORMAT_RGB888:
       assert(0);		/* not supported for sampling */
       return BRW_SURFACEFORMAT_R8G8B8_UNORM;      
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
index ca6e2fa5b1..2eae9b66d8 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
@@ -80,7 +80,7 @@ intel_batchbuffer_reset(struct intel_batchbuffer *batch)
       batch->buf = NULL;
    }
 
-   if (!batch->buffer && intel->ttm == GL_TRUE)
+   if (!batch->buffer)
       batch->buffer = malloc (intel->maxBatchSize);
 
    batch->buf = dri_bo_alloc(intel->bufmgr, "batchbuffer",
@@ -212,7 +212,7 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
 
    batch->reserved_space = 0;
    /* Emit a flush if the bufmgr doesn't do it for us. */
-   if (intel->always_flush_cache || !intel->ttm) {
+   if (intel->always_flush_cache) {
       intel_batchbuffer_emit_mi_flush(batch);
       used = batch->ptr - batch->map;
    }
@@ -244,6 +244,9 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
    if (intel->vtbl.finish_batch)
       intel->vtbl.finish_batch(intel);
 
+   /* Check that we didn't just wrap our batchbuffer at a bad time. */
+   assert(!intel->no_batch_wrap);
+
    batch->reserved_space = BATCH_RESERVED;
 
    /* TODO: Just pass the relocation list and dma buffer up to the
diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c
index 817223da41..f14854602b 100644
--- a/src/mesa/drivers/dri/intel/intel_blit.c
+++ b/src/mesa/drivers/dri/intel/intel_blit.c
@@ -499,10 +499,11 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
 		  switch (irb->texformat) {
 		  case MESA_FORMAT_ARGB8888:
 		  case MESA_FORMAT_XRGB8888:
-		     clearVal = intel->ClearColor8888;
+		     clearVal = PACK_COLOR_8888(clear[3], clear[0],
+						clear[1], clear[2]);
 		     break;
 		  case MESA_FORMAT_RGB565:
-		     clearVal = intel->ClearColor565;
+		     clearVal = PACK_COLOR_565(clear[0], clear[1], clear[2]);
 		     break;
 		  case MESA_FORMAT_ARGB4444:
 		     clearVal = PACK_COLOR_4444(clear[3], clear[0],
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 2aeca6b81b..1434ae530b 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -176,9 +176,7 @@ intelGetString(GLcontext * ctx, GLenum name)
          break;
       }
 
-      (void) driGetRendererString(buffer, chipset, 
-				  (intel->ttm) ? DRIVER_DATE_GEM : DRIVER_DATE,
-				  0);
+      (void) driGetRendererString(buffer, chipset, DRIVER_DATE_GEM, 0);
       return (GLubyte *) buffer;
 
    default:
@@ -601,6 +599,7 @@ intelInitContext(struct intel_context *intel,
    __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
    intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
    int fthrottle_mode;
+   int bo_reuse_mode;
 
    if (!_mesa_initialize_context(&intel->ctx, mesaVis, shareCtx,
                                  functions, (void *) intel)) {
@@ -635,18 +634,14 @@ intelInitContext(struct intel_context *intel,
       intel->maxBatchSize = BATCH_SZ;
 
    intel->bufmgr = intelScreen->bufmgr;
-   intel->ttm = intelScreen->ttm;
-   if (intel->ttm) {
-      int bo_reuse_mode;
 
-      bo_reuse_mode = driQueryOptioni(&intel->optionCache, "bo_reuse");
-      switch (bo_reuse_mode) {
-      case DRI_CONF_BO_REUSE_DISABLED:
-	 break;
-      case DRI_CONF_BO_REUSE_ALL:
-	 intel_bufmgr_gem_enable_reuse(intel->bufmgr);
-	 break;
-      }
+   bo_reuse_mode = driQueryOptioni(&intel->optionCache, "bo_reuse");
+   switch (bo_reuse_mode) {
+   case DRI_CONF_BO_REUSE_DISABLED:
+      break;
+   case DRI_CONF_BO_REUSE_ALL:
+      intel_bufmgr_gem_enable_reuse(intel->bufmgr);
+      break;
    }
 
    /* This doesn't yet catch all non-conformant rendering, but it's a
@@ -733,12 +728,6 @@ intelInitContext(struct intel_context *intel,
    intel->RenderIndex = ~0;
 
    fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
-   intel->irqsEmitted = 0;
-
-   intel->do_irqs = (intel->intelScreen->irq_active &&
-                     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
-
-   intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
 
    if (intel->gen >= 4 && !intel->intelScreen->irq_active) {
       _mesa_printf("IRQs not active.  Exiting\n");
@@ -1058,21 +1047,6 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
       sarea->ctxOwner = me;
    }
 
-   /* If the last consumer of the texture memory wasn't us, notify the fake
-    * bufmgr and record the new owner.  We should have the memory shared
-    * between contexts of a single fake bufmgr, but this will at least make
-    * things correct for now.
-    */
-   if (!intel->ttm && sarea->texAge != intel->hHWContext) {
-      sarea->texAge = intel->hHWContext;
-      intel_bufmgr_fake_contended_lock_take(intel->bufmgr);
-      if (INTEL_DEBUG & DEBUG_BATCH)
-	 intel_decode_context_reset();
-      if (INTEL_DEBUG & DEBUG_BUFMGR)
-	 fprintf(stderr, "Lost Textures: sarea->texAge %x hw context %x\n",
-		 sarea->ctxOwner, intel->hHWContext);
-   }
-
    /* Drawable changed?
     */
    if (dPriv && intel->lastStamp != dPriv->lastStamp) {
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index eb7be7ddd0..481202c971 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -135,14 +135,6 @@ struct intel_context
                                 struct intel_region * draw_region,
                                 struct intel_region * depth_region);
 
-      void (*meta_draw_quad)(struct intel_context *intel,
-			     GLfloat x0, GLfloat x1,
-			     GLfloat y0, GLfloat y1,
-			     GLfloat z,
-			     GLuint color, /* ARGB32 */
-			     GLfloat s0, GLfloat s1,
-			     GLfloat t0, GLfloat t1);
-
       void (*meta_color_mask) (struct intel_context * intel, GLboolean);
 
       void (*meta_stencil_replace) (struct intel_context * intel,
@@ -189,12 +181,6 @@ struct intel_context
    struct intel_region *back_region;
    struct intel_region *depth_region;
 
-   /**
-    * This value indicates that the kernel memory manager is being used
-    * instead of the fake client-side memory manager.
-    */
-   GLboolean ttm;
-
    struct intel_batchbuffer *batch;
    drm_intel_bo *first_post_swapbuffers_batch;
    GLboolean no_batch_wrap;
@@ -217,10 +203,6 @@ struct intel_context
    char *prevLockFile;
    int prevLockLine;
 
-   GLuint ClearColor565;
-   GLuint ClearColor8888;
-
-
    /* Offsets of fields within the current vertex:
     */
    GLuint coloroffset;
@@ -237,6 +219,7 @@ struct intel_context
    GLboolean hw_stipple;
    GLboolean depth_buffer_is_float;
    GLboolean no_rast;
+   GLboolean no_hw;
    GLboolean always_flush_batch;
    GLboolean always_flush_cache;
 
@@ -302,13 +285,6 @@ struct intel_context
    GLboolean use_early_z;
    drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
 
-   int perf_boxes;
-
-   GLuint do_usleeps;
-   int do_irqs;
-   GLuint irqsEmitted;
-
-   GLboolean scissor;
    drm_clip_rect_t draw_rect;
    drm_clip_rect_t scissor_rect;
 
@@ -325,8 +301,6 @@ struct intel_context
 
    GLuint lastStamp;
 
-   GLboolean no_hw;
-
    /**
     * Configuration cache
     */
@@ -374,29 +348,6 @@ do {						\
 } while (0)
 
 /* ================================================================
- * Color packing:
- */
-
-#define INTEL_PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define INTEL_PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define INTEL_PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define INTEL_PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
-(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
- (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
-  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
-   0)))
-
-/* ================================================================
  * From linux kernel i386 header files, copes with odd sizes better
  * than COPY_DWORDS would:
  * XXX Put this in src/mesa/main/imports.h ???
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c
index 1682e115cc..f5fe543b5d 100644
--- a/src/mesa/drivers/dri/intel/intel_extensions.c
+++ b/src/mesa/drivers/dri/intel/intel_extensions.c
@@ -79,6 +79,7 @@ static const struct dri_extension card_extensions[] = {
    { "GL_ARB_half_float_pixel",           NULL },
    { "GL_ARB_map_buffer_range",           GL_ARB_map_buffer_range_functions },
    { "GL_ARB_multitexture",               NULL },
+   { "GL_ARB_pixel_buffer_object",      NULL },
    { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
    { "GL_ARB_point_sprite",               NULL },
    { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
@@ -104,6 +105,8 @@ static const struct dri_extension card_extensions[] = {
    { "GL_EXT_blend_logic_op",             NULL },
    { "GL_EXT_blend_subtract",             NULL },
    { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
+   { "GL_EXT_framebuffer_blit",         GL_EXT_framebuffer_blit_functions },
+   { "GL_EXT_framebuffer_object",       GL_EXT_framebuffer_object_functions },
    { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
    { "GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions },
    { "GL_EXT_packed_depth_stencil",       NULL },
@@ -176,14 +179,6 @@ static const struct dri_extension arb_oq_extensions[] = {
    { NULL, NULL }
 };
 
-
-static const struct dri_extension ttm_extensions[] = {
-   { "GL_ARB_pixel_buffer_object",      NULL },
-   { "GL_EXT_framebuffer_blit",         GL_EXT_framebuffer_blit_functions },
-   { "GL_EXT_framebuffer_object",       GL_EXT_framebuffer_object_functions },
-   { NULL, NULL }
-};
-
 static const struct dri_extension fragment_shader_extensions[] = {
    { "GL_ARB_fragment_shader",            NULL },
    { NULL, NULL }
@@ -202,9 +197,6 @@ intelInitExtensions(GLcontext *ctx)
     */
    driInitExtensions(ctx, card_extensions, GL_FALSE);
 
-   if (intel->ttm)
-      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
-
    if (IS_965(intel->intelScreen->deviceID))
       driInitExtensions(ctx, brw_extensions, GL_FALSE);
 
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
index abb3024bfb..6a565f80cf 100644
--- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
@@ -224,16 +224,12 @@ int intel_miptree_pitch_align (struct intel_context *intel,
    if (!mt->compressed) {
       int pitch_align;
 
-      if (intel->ttm) {
-	 /* XXX: Align pitch to multiple of 64 bytes for now to allow
-	  * render-to-texture to work in all cases. This should probably be
-	  * replaced at some point by some scheme to only do this when really
-	  * necessary.
-	  */
-	 pitch_align = 64;
-      } else {
-	 pitch_align = 4;
-      }
+      /* XXX: Align pitch to multiple of 64 bytes for now to allow
+       * render-to-texture to work in all cases. This should probably be
+       * replaced at some point by some scheme to only do this when really
+       * necessary.
+       */
+      pitch_align = 64;
 
       if (tiling == I915_TILING_X)
 	 pitch_align = 512;
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index 99330b6ddf..9572b67326 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -228,10 +228,9 @@ do_blit_bitmap( GLcontext *ctx,
    UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[3], tmpColor[3]);
 
    if (dst->cpp == 2)
-      color = INTEL_PACKCOLOR565(ubcolor[0], ubcolor[1], ubcolor[2]);
+      color = PACK_COLOR_565(ubcolor[0], ubcolor[1], ubcolor[2]);
    else
-      color = INTEL_PACKCOLOR8888(ubcolor[0], ubcolor[1],
-				  ubcolor[2], ubcolor[3]);
+      color = PACK_COLOR_8888(ubcolor[3], ubcolor[0], ubcolor[1], ubcolor[2]);
 
    if (!intel_check_blit_fragment_ops(ctx, tmpColor[3] == 1.0F))
       return GL_FALSE;
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index 80975163d4..d6b9dc4446 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -542,55 +542,18 @@ intel_recreate_static(struct intel_context *intel,
       region->buffer = NULL;
    }
 
-   if (intel->ttm) {
-      assert(region_desc->bo_handle != -1);
-      region->buffer = intel_bo_gem_create_from_name(intel->bufmgr,
-						     name,
-						     region_desc->bo_handle);
-
-      ret = dri_bo_get_tiling(region->buffer, &region->tiling,
-			      &region->bit_6_swizzle);
-      if (ret != 0) {
-	 fprintf(stderr, "Couldn't get tiling of buffer %d (%s): %s\n",
-		 region_desc->bo_handle, name, strerror(-ret));
-	 intel_region_release(&region);
-	 return NULL;
-      }
-   } else {
-      if (region->classic_map != NULL) {
-	 drmUnmap(region->classic_map,
-		  region->pitch * region->cpp * region->height);
-	 region->classic_map = NULL;
-      }
-      ret = drmMap(intel->driFd, region_desc->handle,
-		   region->pitch * region->cpp * region->height,
-		   &region->classic_map);
-      if (ret != 0) {
-	 fprintf(stderr, "Failed to drmMap %s buffer\n", name);
-	 free(region);
-	 return NULL;
-      }
-
-      region->buffer = intel_bo_fake_alloc_static(intel->bufmgr,
+   assert(region_desc->bo_handle != -1);
+   region->buffer = intel_bo_gem_create_from_name(intel->bufmgr,
 						  name,
-						  region_desc->offset,
-						  region->pitch * region->cpp *
-						  region->height,
-						  region->classic_map);
-
-      /* The sarea just gives us a boolean for whether it's tiled or not,
-       * instead of which tiling mode it is.  Guess.
-       */
-      if (region_desc->tiled) {
-	 if (intel->gen >= 4 && region_desc == &intelScreen->depth)
-	    region->tiling = I915_TILING_Y;
-	 else
-	    region->tiling = I915_TILING_X;
-      } else {
-	 region->tiling = I915_TILING_NONE;
-      }
+						  region_desc->bo_handle);
 
-      region->bit_6_swizzle = I915_BIT_6_SWIZZLE_NONE;
+   ret = dri_bo_get_tiling(region->buffer, &region->tiling,
+			   &region->bit_6_swizzle);
+   if (ret != 0) {
+      fprintf(stderr, "Couldn't get tiling of buffer %d (%s): %s\n",
+	      region_desc->bo_handle, name, strerror(-ret));
+      intel_region_release(&region);
+      return NULL;
    }
 
    assert(region->buffer != NULL);
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index 789135b49f..2c5a884a9b 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -605,7 +605,6 @@ intelFillInModes(__DRIscreenPrivate *psp,
 static GLboolean
 intel_init_bufmgr(intelScreenPrivate *intelScreen)
 {
-   GLboolean gem_disable = getenv("INTEL_NO_GEM") != NULL;
    int gem_kernel = 0;
    GLboolean gem_supported;
    struct drm_i915_getparam gp;
@@ -622,43 +621,24 @@ intel_init_bufmgr(intelScreenPrivate *intelScreen)
    /* If we've got a new enough DDX that's initializing GEM and giving us
     * object handles for the shared buffers, use that.
     */
-   intelScreen->ttm = GL_FALSE;
    if (intelScreen->driScrnPriv->dri2.enabled)
        gem_supported = GL_TRUE;
    else if (intelScreen->driScrnPriv->ddx_version.minor >= 9 &&
 	    gem_kernel &&
 	    intelScreen->front.bo_handle != -1)
        gem_supported = GL_TRUE;
-   else
-       gem_supported = GL_FALSE;
-
-   if (!gem_disable && gem_supported) {
-      intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
-      if (intelScreen->bufmgr != NULL)
-	 intelScreen->ttm = GL_TRUE;
+   else {
+      fprintf(stderr, "[%s:%u] Error initializing GEM.\n",
+	      __func__, __LINE__);
+      return GL_FALSE;
    }
+
+   intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
    /* Otherwise, use the classic buffer manager. */
    if (intelScreen->bufmgr == NULL) {
-      if (gem_disable) {
-	 _mesa_warning(NULL, "GEM disabled.  Using classic.");
-      } else {
-	 _mesa_warning(NULL,
-                       "Failed to initialize GEM.  Falling back to classic.");
-      }
-
-      if (intelScreen->tex.size == 0) {
-	 fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
-		 __func__, __LINE__);
-	 return GL_FALSE;
-      }
-
-      intelScreen->bufmgr =
-	 intel_bufmgr_fake_init(spriv->fd,
-				intelScreen->tex.offset,
-				intelScreen->tex.map,
-				intelScreen->tex.size,
-				(unsigned int * volatile)
-				&intelScreen->sarea->last_dispatch);
+      fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
+	      __func__, __LINE__);
+      return GL_FALSE;
    }
 
    if (intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences))
diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index a9b9e109a6..14ca0903b6 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -77,7 +77,6 @@ typedef struct
    GLboolean no_hw;
 
    GLboolean no_vbo;
-   int ttm;
    dri_bufmgr *bufmgr;
    GLboolean kernel_exec_fencing;
 
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
index 3607c7dded..2c89a66a95 100644
--- a/src/mesa/drivers/dri/intel/intel_span.c
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -613,15 +613,7 @@ intel_set_span_functions(struct intel_context *intel,
 			 struct gl_renderbuffer *rb)
 {
    struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
-   uint32_t tiling;
-
-   /* If in GEM mode, we need to do the tile address swizzling ourselves,
-    * instead of the fence registers handling it.
-    */
-   if (intel->ttm)
-      tiling = irb->region->tiling;
-   else
-      tiling = I915_TILING_NONE;
+   uint32_t tiling = irb->region->tiling;
 
    if (intel->intelScreen->kernel_exec_fencing) {
       switch (irb->texformat) {
@@ -673,6 +665,9 @@ intel_set_span_functions(struct intel_context *intel,
       return;
    }
 
+   /* If in GEM mode, we need to do the tile address swizzling ourselves,
+    * instead of the fence registers handling it.
+    */
    switch (irb->texformat) {
    case MESA_FORMAT_RGB565:
       switch (tiling) {
diff --git a/src/mesa/drivers/dri/intel/intel_state.c b/src/mesa/drivers/dri/intel/intel_state.c
index 4ee742377d..aefae53eb2 100644
--- a/src/mesa/drivers/dri/intel/intel_state.c
+++ b/src/mesa/drivers/dri/intel/intel_state.c
@@ -196,25 +196,6 @@ intel_translate_logic_op(GLenum opcode)
    }
 }
 
-
-static void
-intelClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLubyte clear[4];
-
-   CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
-
-   /* compute both 32 and 16-bit clear values */
-   intel->ClearColor8888 = INTEL_PACKCOLOR8888(clear[0], clear[1],
-                                               clear[2], clear[3]);
-   intel->ClearColor565 = INTEL_PACKCOLOR565(clear[0], clear[1], clear[2]);
-}
-
-
 /* Fallback to swrast for select and feedback.
  */
 static void
@@ -229,5 +210,4 @@ void
 intelInitStateFuncs(struct dd_function_table *functions)
 {
    functions->RenderMode = intelRenderMode;
-   functions->ClearColor = intelClearColor;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 4b5fe7be9f..767d04d2f4 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -221,6 +221,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y,
                              width, border);
 }
@@ -268,6 +270,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
                              width, height, border);
 }
@@ -292,6 +296,8 @@ intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
    if (!do_copy_texsubimage(intel_context(ctx), target,
                             intel_texture_image(texImage),
                             internalFormat, xoffset, 0, x, y, width, 1)) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width);
    }
 }
@@ -317,8 +323,8 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
                             internalFormat,
                             xoffset, yoffset, x, y, width, height)) {
 
-      DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__);
-
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage2D(ctx, target, level,
                                    xoffset, yoffset, x, y, width, height);
    }
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
index f37a545c7f..bfa3dba1f5 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_format.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -95,14 +95,20 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_COMPRESSED_LUMINANCE:
       return MESA_FORMAT_L8;
 
+   case GL_LUMINANCE12_ALPHA4:
+   case GL_LUMINANCE12_ALPHA12:
+   case GL_LUMINANCE16_ALPHA16:
+#ifndef I915
+      return MESA_FORMAT_AL1616;
+#else
+      /* FALLTHROUGH */
+#endif
+
    case 2:
    case GL_LUMINANCE_ALPHA:
    case GL_LUMINANCE4_ALPHA4:
    case GL_LUMINANCE6_ALPHA2:
    case GL_LUMINANCE8_ALPHA8:
-   case GL_LUMINANCE12_ALPHA4:
-   case GL_LUMINANCE12_ALPHA12:
-   case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
       return MESA_FORMAT_AL88;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vb.c b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
index 99f1a14e17..816682ec5f 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vb.c
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
@@ -207,19 +207,19 @@ INTERP_QUALIFIER void TAG(interp_extras)( GLcontext *ctx,
    LOCALVARS
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      assert(VB->ColorPtr[1]->stride == 4 * sizeof(GLfloat));
+   if (VB->BackfaceColorPtr) {
+      assert(VB->BackfaceColorPtr->stride == 4 * sizeof(GLfloat));
       
       INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		 GET_COLOR(VB->BackfaceColorPtr, dst),
+		 GET_COLOR(VB->BackfaceColorPtr, out),
+		 GET_COLOR(VB->BackfaceColorPtr, in) );
 
-      if (VB->SecondaryColorPtr[1]) {
+      if (VB->BackfaceSecondaryColorPtr) {
 	 INTERP_3F( t,
-		       GET_COLOR(VB->SecondaryColorPtr[1], dst),
-		       GET_COLOR(VB->SecondaryColorPtr[1], out),
-		       GET_COLOR(VB->SecondaryColorPtr[1], in) );
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, out),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, in) );
       }
    }
 
@@ -236,13 +236,13 @@ INTERP_QUALIFIER void TAG(copy_pv_extras)( GLcontext *ctx,
    LOCALVARS
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst), 
-		GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+      COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		GET_COLOR(VB->BackfaceColorPtr, src) );
 
-      if (VB->SecondaryColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->SecondaryColorPtr[1], dst), 
-		   GET_COLOR(VB->SecondaryColorPtr[1], src) );
+      if (VB->BackfaceSecondaryColorPtr) {
+	 COPY_4FV( GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		   GET_COLOR(VB->BackfaceSecondaryColorPtr, src) );
       }
    }
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h b/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
index 684f2acc89..6e5fa3520e 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
@@ -103,10 +103,10 @@ static void TAG(emit)( GLcontext *ctx,
 #if DO_TEX1
    {
       const GLuint t1 = GET_TEXSOURCE(1);
-      tc1 = VB->TexCoordPtr[t1]->data;
-      tc1_stride = VB->TexCoordPtr[t1]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
 #if DO_PTEX
-      tc1_size = VB->TexCoordPtr[t1]->size;
+      tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size;
 #endif
    }
 #endif
@@ -114,18 +114,18 @@ static void TAG(emit)( GLcontext *ctx,
 #if DO_TEX0
    {
       const GLuint t0 = GET_TEXSOURCE(0);
-      tc0 = VB->TexCoordPtr[t0]->data;
-      tc0_stride = VB->TexCoordPtr[t0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
 #if DO_PTEX
-      tc0_size = VB->TexCoordPtr[t0]->size;
+      tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size;
 #endif
    }
 #endif
 
 #if DO_SPEC
-   if (VB->SecondaryColorPtr[0]) {
-      spec = VB->SecondaryColorPtr[0]->data;
-      spec_stride = VB->SecondaryColorPtr[0]->stride;
+   if (VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
+      spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+      spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
    } else {
       spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
       spec_stride = 0;
@@ -133,9 +133,9 @@ static void TAG(emit)( GLcontext *ctx,
 #endif
 
 #if DO_FOG
-   if (VB->FogCoordPtr) {
-      fog = VB->FogCoordPtr->data;
-      fog_stride = VB->FogCoordPtr->stride;
+   if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+      fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+      fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
    } else {
       static GLfloat tmp[4] = {0, 0, 0, 0};
       fog = &tmp;
@@ -144,8 +144,8 @@ static void TAG(emit)( GLcontext *ctx,
 #endif
 
 #if DO_RGBA
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 #endif
 
    coord = VB->NdcPtr->data;
@@ -319,8 +319,8 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
@@ -328,12 +328,12 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
    /* No hardware support for projective texture.  Can fake it for
     * TEX0 only.
     */
-   if ((DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
+   if ((DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4)) {
       PTEX_FALLBACK();
       return GL_FALSE;
    }
 
-   if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
+   if (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4) {
       if (DO_TEX1) {
 	 PTEX_FALLBACK();
       }
diff --git a/src/mesa/drivers/dri/mach64/mach64_vbtmp.h b/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
index 938804af9e..60bfab8f6d 100644
--- a/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
+++ b/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
@@ -156,53 +156,53 @@ static void TAG(emit)( GLcontext *ctx,
 
    if (DO_TEX3) {
       const GLuint t3 = GET_TEXSOURCE(3);
-      tc3 = VB->TexCoordPtr[t3]->data;
-      tc3_stride = VB->TexCoordPtr[t3]->stride;
+      tc3 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->data;
+      tc3_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->stride;
       if (DO_PTEX)
-	 tc3_size = VB->TexCoordPtr[t3]->size;
+	 tc3_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->size;
    }
 
    if (DO_TEX2) {
       const GLuint t2 = GET_TEXSOURCE(2);
-      tc2 = VB->TexCoordPtr[t2]->data;
-      tc2_stride = VB->TexCoordPtr[t2]->stride;
+      tc2 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->data;
+      tc2_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->stride;
       if (DO_PTEX)
-	 tc2_size = VB->TexCoordPtr[t2]->size;
+	 tc2_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size;
    }
 
    if (DO_TEX1) {
       const GLuint t1 = GET_TEXSOURCE(1);
-      tc1 = VB->TexCoordPtr[t1]->data;
-      tc1_stride = VB->TexCoordPtr[t1]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
       if (DO_PTEX)
-	 tc1_size = VB->TexCoordPtr[t1]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size;
    }
 
    if (DO_TEX0) {
       const GLuint t0 = GET_TEXSOURCE(0);
-      tc0_stride = VB->TexCoordPtr[t0]->stride;
-      tc0 = VB->TexCoordPtr[t0]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
       if (DO_PTEX) 
-	 tc0_size = VB->TexCoordPtr[t0]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size;
    }
 
    if (DO_RGBA) {
-      col = VB->ColorPtr[0]->data;
-      col_stride = VB->ColorPtr[0]->stride;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
    }
 
    if (DO_SPEC) {
-      spec = VB->SecondaryColorPtr[0]->data;
-      spec_stride = VB->SecondaryColorPtr[0]->stride;
+      spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+      spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
    } else {
       spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
       spec_stride = 0;
    }
 
    if (DO_FOG) {
-      if (VB->FogCoordPtr) {
-	 fog = VB->FogCoordPtr->data;
-	 fog_stride = VB->FogCoordPtr->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+	 fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+	 fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
       } else {
 	 static GLfloat tmp[4] = {0, 0, 0, 0};
 	 fog = &tmp;
@@ -384,8 +384,8 @@ static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
 
    ASSERT(stride == 4);
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    /* Pack what's left into a 4-dword vertex.  Color is in a different
     * place, and there is no 'w' coordinate.
@@ -432,8 +432,8 @@ static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
    GLfloat *v = (GLfloat *)dest;
    int i;
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    if (start)
       STRIDE_4F(col, col_stride * start);
@@ -473,22 +473,22 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX2] = VB->AttribPtr[_TNL_ATTRIB_TEX3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX1] = VB->AttribPtr[_TNL_ATTRIB_TEX2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
    
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4) ||
-       (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4))
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4) ||
+       (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4))
       return GL_FALSE;
 
    return GL_TRUE;
@@ -501,14 +501,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX2] = VB->AttribPtr[_TNL_ATTRIB_TEX3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX1] = VB->AttribPtr[_TNL_ATTRIB_TEX2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
@@ -516,14 +516,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
    /* No hardware support for projective texture.  Can fake it for
     * TEX0 only.
     */
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4)) {
       PTEX_FALLBACK();
       return GL_FALSE;
    }
 
-   if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
+   if (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4) {
       if (DO_TEX1 || DO_TEX2 || DO_TEX3) {
 	 PTEX_FALLBACK();
       }
diff --git a/src/mesa/drivers/dri/r128/r128_tris.c b/src/mesa/drivers/dri/r128/r128_tris.c
index 5b91271d74..448e34e047 100644
--- a/src/mesa/drivers/dri/r128/r128_tris.c
+++ b/src/mesa/drivers/dri/r128/r128_tris.c
@@ -650,12 +650,12 @@ static void r128RenderStart( GLcontext *ctx )
    }
 
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(rmesa->tmu_source[0]) )) {
-      if ( VB->TexCoordPtr[rmesa->tmu_source[0]]->size > 2 )
+      if ( VB->AttribPtr[_TNL_ATTRIB_TEX0 + rmesa->tmu_source[0]]->size > 2 )
 	 fallback_projtex = GL_TRUE;
       EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_2F, R128_CCE_VC_FRMT_S_T, 8 );
    }
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(rmesa->tmu_source[1]) )) {
-      if ( VB->TexCoordPtr[rmesa->tmu_source[1]]->size > 2 )
+      if ( VB->AttribPtr[_TNL_ATTRIB_TEX0 + rmesa->tmu_source[1]]->size > 2 )
 	 fallback_projtex = GL_TRUE;
       EMIT_ATTR( _TNL_ATTRIB_TEX1, EMIT_2F, R128_CCE_VC_FRMT_S2_T2, 8 );
    }
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index 383a0c4b0d..249c0bbc11 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -90,12 +90,14 @@ static void r200_emit_vecfog(GLcontext *ctx, struct radeon_aos *aos,
 	aos->components = size;
 	aos->count = count;
 
+	radeon_bo_map(aos->bo, 1);
 	out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
 	for (i = 0; i < count; i++) {
 	  out[0] = r200ComputeFogBlendFactor( ctx, *(GLfloat *)data );
 	  out++;
 	  data += stride;
 	}
+	radeon_bo_unmap(aos->bo);
 }
 
 /* Emit any changed arrays to new GART memory, re-emit a packet to
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index 68bfeea701..e53fd72290 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -640,7 +640,7 @@ static void tex_emit(GLcontext *ctx, struct radeon_state_atom *atom)
    OUT_BATCH_TABLE(atom->cmd, 10);
 
    if (t && t->mt && !t->image_override) {
-     OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+     OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 		  RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
    } else if (!t) {
      /* workaround for old CS mechanism */
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index 240fb45078..4596912ddc 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -168,7 +168,7 @@ static void r200SetVertexFormat( GLcontext *ctx )
 
       for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    GLuint sz = VB->TexCoordPtr[i]->size;
+	    GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
 	    fmt_1 |= sz << (3 * i);
 	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_1F + sz - 1, 0 );
@@ -297,7 +297,7 @@ void r200_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
    radeonEmitState(&rmesa->radeon);
    r200EmitVertexAOS( rmesa,
 		      rmesa->radeon.swtcl.vertex_size,
-		      first_elem(&rmesa->radeon.dma.reserved)->bo,
+		      rmesa->radeon.swtcl.bo,
 		      current_offset);
 
 
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index c702910ef2..e7d48a7f29 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -509,25 +509,26 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
 	 prog to a not enabled output however, so just don't mess with it.
 	 We only need to change compsel. */
       GLuint out_compsel = 0;
-      GLuint vp_out = rmesa->curr_vp_hw->mesa_program.Base.OutputsWritten;
+      const GLbitfield64 vp_out =
+	 rmesa->curr_vp_hw->mesa_program.Base.OutputsWritten;
 
       vimap_rev = &rmesa->curr_vp_hw->inputmap_rev[0];
-      assert(vp_out & (1 << VERT_RESULT_HPOS));
+      assert(vp_out & BITFIELD64_BIT(VERT_RESULT_HPOS));
       out_compsel = R200_OUTPUT_XYZW;
-      if (vp_out & (1 << VERT_RESULT_COL0)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_COL0)) {
 	 out_compsel |= R200_OUTPUT_COLOR_0;
       }
-      if (vp_out & (1 << VERT_RESULT_COL1)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_COL1)) {
 	 out_compsel |= R200_OUTPUT_COLOR_1;
       }
-      if (vp_out & (1 << VERT_RESULT_FOGC)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_FOGC)) {
          out_compsel |= R200_OUTPUT_DISCRETE_FOG;
       }
-      if (vp_out & (1 << VERT_RESULT_PSIZ)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_PSIZ)) {
 	 out_compsel |= R200_OUTPUT_PT_SIZE;
       }
       for (i = VERT_RESULT_TEX0; i < VERT_RESULT_TEX6; i++) {
-	 if (vp_out & (1 << i)) {
+	 if (vp_out & BITFIELD64_BIT(i)) {
 	    out_compsel |= R200_OUTPUT_TEX_0 << (i - VERT_RESULT_TEX0);
 	 }
       }
diff --git a/src/mesa/drivers/dri/r200/r200_tex.c b/src/mesa/drivers/dri/r200/r200_tex.c
index 5a21a8b9c5..a417721553 100644
--- a/src/mesa/drivers/dri/r200/r200_tex.c
+++ b/src/mesa/drivers/dri/r200/r200_tex.c
@@ -385,16 +385,7 @@ static void r200TexParameter( GLcontext *ctx, GLenum target,
    case GL_TEXTURE_MAX_LEVEL:
    case GL_TEXTURE_MIN_LOD:
    case GL_TEXTURE_MAX_LOD:
-      /* This isn't the most efficient solution but there doesn't appear to
-       * be a nice alternative.  Since there's no LOD clamping,
-       * we just have to rely on loading the right subset of mipmap levels
-       * to simulate a clamped LOD.
-       */
-      if (t->mt) {
-         radeon_miptree_unreference(t->mt);
-	 t->mt = 0;
-	 t->validated = GL_FALSE;
-      }
+      t->validated = GL_FALSE;
       break;
 
    default:
@@ -413,7 +404,7 @@ static void r200DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 	      (void *)texObj,
 	      _mesa_lookup_enum_by_nr(texObj->Target));
    }
-   
+
    if (rmesa) {
       int i;
       radeon_firevertices(&rmesa->radeon);
@@ -425,11 +416,9 @@ static void r200DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 	 }
       }      
    }
-   
-   if (t->mt) {
-      radeon_miptree_unreference(t->mt);
-      t->mt = 0;
-   }
+
+   radeon_miptree_unreference(&t->mt);
+
    _mesa_delete_texture_object(ctx, texObj);
 }
 
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index 7d0afa1add..7782404a79 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -824,14 +824,10 @@ void r200SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
@@ -1423,10 +1419,9 @@ void set_re_cntl_d3d( GLcontext *ctx, int unit, GLboolean use_d3d )
  */
 static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
 {
-   int firstlevel = t->mt ? t->mt->firstLevel : 0;
-   const struct gl_texture_image *firstImage = t->base.Image[0][firstlevel];
+   const struct gl_texture_image *firstImage = t->base.Image[0][t->minLod];
    GLint log2Width, log2Height, log2Depth, texelBytes;
-   
+
    if ( t->bo ) {
        return;
    }
@@ -1454,9 +1449,9 @@ static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
 	 return;
       }
    }
-   
+
    t->pp_txfilter &= ~R200_MAX_MIP_LEVEL_MASK;
-   t->pp_txfilter |= (t->mt->lastLevel - t->mt->firstLevel) << R200_MAX_MIP_LEVEL_SHIFT;
+   t->pp_txfilter |= (t->maxLod - t->minLod) << R200_MAX_MIP_LEVEL_SHIFT;
 	
    t->pp_txformat &= ~(R200_TXFORMAT_WIDTH_MASK |
 		       R200_TXFORMAT_HEIGHT_MASK |
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.h b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
index 33db3ea0ff..03592884eb 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
@@ -191,7 +191,7 @@ struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register
 
 static inline void reset_srcreg(struct rc_src_register* reg)
 {
-	memset(reg, 0, sizeof(reg));
+	memset(reg, 0, sizeof(struct rc_src_register));
 	reg->Swizzle = RC_SWIZZLE_XYZW;
 }
 
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
index 0326d25233..ced66af1eb 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
@@ -560,23 +560,23 @@ static void sincos_constants(struct radeon_compiler* c, unsigned int *constants)
  * MAD dest, tmp.y, weight, tmp.x
  */
 static void sin_approx(
-	struct radeon_compiler* c, struct rc_instruction * before,
+	struct radeon_compiler* c, struct rc_instruction * inst,
 	struct rc_dst_register dst, struct rc_src_register src, const unsigned int* constants)
 {
 	unsigned int tempreg = rc_find_free_temporary(c);
 
-	emit2(c, before, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY),
+	emit2(c, inst->Prev, RC_OPCODE_MUL, 0, dstregtmpmask(tempreg, RC_MASK_XY),
 		swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
 		srcreg(RC_FILE_CONSTANT, constants[0]));
-	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X),
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_X),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
 		absolute(swizzle(src, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
-	emit3(c, before, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y),
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dstregtmpmask(tempreg, RC_MASK_Y),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X),
 		absolute(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)),
 		negate(swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X)));
-	emit3(c, before, RC_OPCODE_MAD, 0, dst,
+	emit3(c, inst->Prev, RC_OPCODE_MAD, 0, dst,
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y, RC_SWIZZLE_Y),
 		swizzle(srcreg(RC_FILE_CONSTANT, constants[0]), RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W, RC_SWIZZLE_W),
 		swizzle(srcreg(RC_FILE_TEMPORARY, tempreg), RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X, RC_SWIZZLE_X));
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
index 1e2a54f634..ad8db6e68e 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
@@ -46,14 +46,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "r300_context.h"
 #include "r300_ioctl.h"
-#include "radeon_reg.h"
 #include "r300_reg.h"
 #include "r300_cmdbuf.h"
 #include "r300_emit.h"
 #include "radeon_bocs_wrapper.h"
 #include "radeon_mipmap_tree.h"
 #include "r300_state.h"
-#include "radeon_reg.h"
 #include "radeon_queryobj.h"
 
 /** # of dwords reserved for additional instructions that may need to be written
@@ -171,7 +169,7 @@ static void emit_tex_offsets(GLcontext *ctx, struct radeon_state_atom * atom)
 		if (t && !t->image_override) {
 			BEGIN_BATCH_NO_AUTOSTATE(4);
 			OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1);
-			OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+			OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 					RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
 			END_BATCH();
 		} else if (!t) {
@@ -281,12 +279,27 @@ static void emit_cb_offset(GLcontext *ctx, struct radeon_state_atom * atom)
 		cbpitch |= R300_COLOR_FORMAT_ARGB8888;
 	else switch (rrb->base.Format) {
         case MESA_FORMAT_RGB565:
+		assert(_mesa_little_endian());
+		cbpitch |= R300_COLOR_FORMAT_RGB565;
+		break;
+        case MESA_FORMAT_RGB565_REV:
+		assert(!_mesa_little_endian());
 		cbpitch |= R300_COLOR_FORMAT_RGB565;
 		break;
         case MESA_FORMAT_ARGB4444:
+		assert(_mesa_little_endian());
+		cbpitch |= R300_COLOR_FORMAT_ARGB4444;
+		break;
+        case MESA_FORMAT_ARGB4444_REV:
+		assert(!_mesa_little_endian());
 		cbpitch |= R300_COLOR_FORMAT_ARGB4444;
 		break;
 	case MESA_FORMAT_ARGB1555:
+		assert(_mesa_little_endian());
+		cbpitch |= R300_COLOR_FORMAT_ARGB1555;
+		break;
+	case MESA_FORMAT_ARGB1555_REV:
+		assert(!_mesa_little_endian());
 		cbpitch |= R300_COLOR_FORMAT_ARGB1555;
 		break;
 	default:
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 6f66e970e4..5f07b95634 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -439,11 +439,11 @@ static void r300InitGLExtensions(GLcontext *ctx)
 	if (r300->options.stencil_two_side_disabled)
 		_mesa_disable_extension(ctx, "GL_EXT_stencil_two_side");
 
-	if (r300->options.s3tc_force_enabled) {
+	if (r300->options.s3tc_force_disabled) {
+		_mesa_disable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+	} else if (ctx->Mesa_DXTn || r300->options.s3tc_force_enabled) {
 		_mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 		_mesa_enable_extension(ctx, "GL_S3_s3tc");
-	} else if (r300->options.s3tc_force_disabled) {
-		_mesa_disable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 	}
 
 	if (!r300->radeon.radeonScreen->drmSupportsOcclusionQueries) {
diff --git a/src/mesa/drivers/dri/r300/r300_draw.c b/src/mesa/drivers/dri/r300/r300_draw.c
index e9968f9ffe..3dcd986e22 100644
--- a/src/mesa/drivers/dri/r300/r300_draw.c
+++ b/src/mesa/drivers/dri/r300/r300_draw.c
@@ -100,7 +100,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		GLubyte *in = (GLubyte *)src_ptr;
 
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo, &r300->ind_buf.bo_offset, size, 4);
-
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		out = (GLuint *)ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 
@@ -111,7 +111,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		if (i < mesa_ind_buf->count) {
 			*out++ = in[i];
 		}
-
+		radeon_bo_unmap(r300->ind_buf.bo);
 #if MESA_BIG_ENDIAN
 	} else { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */
 		GLushort *in = (GLushort *)src_ptr;
@@ -120,6 +120,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo,
 				     &r300->ind_buf.bo_offset, size, 4);
 
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		out = (GLuint *)ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 
@@ -130,6 +131,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		if (i < mesa_ind_buf->count) {
 			*out++ = in[i];
 		}
+		radeon_bo_unmap(r300->ind_buf.bo);
 #endif
 	}
 
@@ -173,10 +175,12 @@ static void r300SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo, &r300->ind_buf.bo_offset, size, 4);
 
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		dst_ptr = ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 		_mesa_memcpy(dst_ptr, src_ptr, size);
 
+		radeon_bo_unmap(r300->ind_buf.bo);
 		r300->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT);
 		r300->ind_buf.count = mesa_ind_buf->count;
 
@@ -242,6 +246,7 @@ static void r300ConvertAttrib(GLcontext *ctx, int count, const struct gl_client_
 	}
 
 	radeonAllocDmaRegion(&r300->radeon, &attr->bo, &attr->bo_offset, sizeof(GLfloat) * input->Size * count, 32);
+	radeon_bo_map(attr->bo, 1);
 	dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
 
 	radeon_print(RADEON_FALLBACKS, RADEON_IMPORTANT,
@@ -280,6 +285,7 @@ static void r300ConvertAttrib(GLcontext *ctx, int count, const struct gl_client_
 			break;
 	}
 
+	radeon_bo_unmap(attr->bo);
 	if (mapped_named_bo) {
 		ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
 	}
@@ -294,6 +300,8 @@ static void r300AlignDataToDword(GLcontext *ctx, const struct gl_client_array *i
 
 	radeonAllocDmaRegion(&r300->radeon, &attr->bo, &attr->bo_offset, size, 32);
 
+	radeon_bo_map(attr->bo, 1);
+
 	if (!input->BufferObj->Pointer) {
 		ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
 		mapped_named_bo = GL_TRUE;
@@ -317,6 +325,7 @@ static void r300AlignDataToDword(GLcontext *ctx, const struct gl_client_array *i
 		ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
 	}
 
+	radeon_bo_unmap(attr->bo);
 	attr->stride = dst_stride;
 }
 
@@ -527,6 +536,7 @@ static void r300AllocDmaRegions(GLcontext *ctx, const struct gl_client_array *in
 				}
 
 				radeonAllocDmaRegion(&r300->radeon, &vbuf->attribs[index].bo, &vbuf->attribs[index].bo_offset, size, 32);
+				radeon_bo_map(vbuf->attribs[index].bo, 1);
 				assert(vbuf->attribs[index].bo->ptr != NULL);
 				dst = (uint32_t *)ADD_POINTERS(vbuf->attribs[index].bo->ptr, vbuf->attribs[index].bo_offset);
 				switch (vbuf->attribs[index].dwords) {
@@ -536,6 +546,7 @@ static void r300AllocDmaRegions(GLcontext *ctx, const struct gl_client_array *in
 					case 4: radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); break;
 					default: assert(0); break;
 				}
+				radeon_bo_unmap(vbuf->attribs[index].bo);
 
 			}
 		}
diff --git a/src/mesa/drivers/dri/r300/r300_emit.h b/src/mesa/drivers/dri/r300/r300_emit.h
index 8e57e354d1..a456d8867c 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.h
+++ b/src/mesa/drivers/dri/r300/r300_emit.h
@@ -42,7 +42,6 @@
 #include "main/glheader.h"
 #include "r300_context.h"
 #include "r300_cmdbuf.h"
-#include "radeon_reg.h"
 
 static INLINE uint32_t cmdpacket0(struct radeon_screen *rscrn,
                                   int reg, int count)
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 3cd38753b8..4ae593cbe7 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -67,8 +67,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "vbo/vbo_split.h"
 #include "tnl/tnl.h"
 #include "tnl/t_vp_build.h"
-#include "radeon_reg.h"
-#include "radeon_macros.h"
 #include "r300_context.h"
 #include "r300_ioctl.h"
 #include "r300_state.h"
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index ac20c08e20..1fd32d497b 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -1741,7 +1741,8 @@ static void r300Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 		r300SetPolygonOffsetState(ctx, state);
 		break;
 	case GL_SCISSOR_TEST:
-		radeon_firevertices(&rmesa->radeon);
+		if (!rmesa->radeon.radeonScreen->kernel_mm)
+			radeon_firevertices(&rmesa->radeon);
 		rmesa->radeon.state.scissor.enabled = state;
 		radeonUpdateScissor( ctx );
 		break;
diff --git a/src/mesa/drivers/dri/r300/r300_swtcl.c b/src/mesa/drivers/dri/r300/r300_swtcl.c
index ee2c71e1a7..383c8a274b 100644
--- a/src/mesa/drivers/dri/r300/r300_swtcl.c
+++ b/src/mesa/drivers/dri/r300/r300_swtcl.c
@@ -124,7 +124,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 	}
 
 	if (ctx->Light.Enabled && ctx->Light.Model.TwoSide) {
-		VB->AttribPtr[VERT_ATTRIB_GENERIC0] = VB->ColorPtr[1];
+		VB->AttribPtr[VERT_ATTRIB_GENERIC0] = VB->BackfaceColorPtr;
 		OutputsWritten |= 1 << VERT_RESULT_BFC0;
 #if MESA_LITTLE_ENDIAN
 		EMIT_ATTR( _TNL_ATTRIB_GENERIC0, EMIT_4UB_4F_RGBA );
@@ -134,7 +134,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		ADD_ATTR(VERT_ATTRIB_GENERIC0, R300_DATA_TYPE_BYTE, SWTCL_OVM_COLOR2, SWIZZLE_XYZW, MASK_XYZW, 1);
 #endif
 		if (fp_reads & FRAG_BIT_COL1) {
-			VB->AttribPtr[VERT_ATTRIB_GENERIC1] = VB->SecondaryColorPtr[1];
+			VB->AttribPtr[VERT_ATTRIB_GENERIC1] = VB->BackfaceSecondaryColorPtr;
 			GLuint swiz = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ONE);
 			OutputsWritten |= 1 << VERT_RESULT_BFC1;
 #if MESA_LITTLE_ENDIAN
@@ -159,7 +159,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		int tex_id = rmesa->selected_fp->wpos_attr - FRAG_ATTRIB_TEX0;
 
 		VB->AttribPtr[VERT_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
-		VB->TexCoordPtr[tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
+		VB->AttribPtr[_TNL_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
 		RENDERINPUTS_SET(tnl->render_inputs_bitset, _TNL_ATTRIB_TEX0 + tex_id);
 	}
 
@@ -167,7 +167,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		int tex_id = rmesa->selected_fp->fog_attr - FRAG_ATTRIB_TEX0;
 
 		VB->AttribPtr[VERT_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
-		VB->TexCoordPtr[tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
+		VB->AttribPtr[_TNL_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
 		RENDERINPUTS_SET(tnl->render_inputs_bitset, _TNL_ATTRIB_TEX0 + tex_id);
 	}
 
@@ -180,7 +180,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		GLuint swiz, format, hw_format;
 		for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 			if (fp_reads & FRAG_BIT_TEX(i)) {
-				switch (VB->TexCoordPtr[i]->size) {
+				switch (VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size) {
 					case 1:
 						format = EMIT_1F;
 						hw_format = R300_DATA_TYPE_FLOAT_1;
@@ -665,11 +665,11 @@ void r300_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
 	r300EmitCacheFlush(rmesa);
 
 	radeonEmitState(&rmesa->radeon);
-    r300_emit_scissor(ctx);
+	r300_emit_scissor(ctx);
 	r300EmitVertexAOS(rmesa,
-			rmesa->radeon.swtcl.vertex_size,
-			first_elem(&rmesa->radeon.dma.reserved)->bo,
-			current_offset);
+			  rmesa->radeon.swtcl.vertex_size,
+			  rmesa->radeon.swtcl.bo,
+			  current_offset);
 
 	r300EmitVbufPrim(rmesa,
 		   rmesa->radeon.swtcl.hw_primitive,
diff --git a/src/mesa/drivers/dri/r300/r300_tex.c b/src/mesa/drivers/dri/r300/r300_tex.c
index 27b78a912f..726b3ff98e 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.c
+++ b/src/mesa/drivers/dri/r300/r300_tex.c
@@ -223,16 +223,7 @@ static void r300TexParameter(GLcontext * ctx, GLenum target,
 	case GL_TEXTURE_MAX_LEVEL:
 	case GL_TEXTURE_MIN_LOD:
 	case GL_TEXTURE_MAX_LOD:
-		/* This isn't the most efficient solution but there doesn't appear to
-		 * be a nice alternative.  Since there's no LOD clamping,
-		 * we just have to rely on loading the right subset of mipmap levels
-		 * to simulate a clamped LOD.
-		 */
-		if (t->mt) {
-			radeon_miptree_unreference(t->mt);
-			t->mt = 0;
-			t->validated = GL_FALSE;
-		}
+		t->validated = GL_FALSE;
 		break;
 
 	case GL_DEPTH_TEXTURE_MODE:
@@ -270,7 +261,11 @@ static void r300DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 
 	if (rmesa) {
 		int i;
-		radeon_firevertices(&rmesa->radeon);
+		struct radeon_bo *bo;
+		bo = !t->mt ? t->bo : t->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->radeon.cmdbuf.cs)) {
+			radeon_firevertices(&rmesa->radeon);
+		}
 
 		for(i = 0; i < R300_MAX_TEXTURE_UNITS; ++i)
 			if (rmesa->hw.textures[i] == t)
@@ -282,10 +277,8 @@ static void r300DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 		t->bo = NULL;
 	}
 
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
+	radeon_miptree_unreference(&t->mt);
+
 	_mesa_delete_texture_object(ctx, texObj);
 }
 
diff --git a/src/mesa/drivers/dri/r300/r300_texstate.c b/src/mesa/drivers/dri/r300/r300_texstate.c
index 44ca24daf8..e6f2c0c1a7 100644
--- a/src/mesa/drivers/dri/r300/r300_texstate.c
+++ b/src/mesa/drivers/dri/r300/r300_texstate.c
@@ -83,6 +83,7 @@ static const struct tx_table {
 	_ASSIGN(ARGB8888, R300_EASY_TX_FORMAT(W, Z, Y, X, W8Z8Y8X8)),
 	_ASSIGN(ARGB8888_REV, R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8)),
 #endif
+	_ASSIGN(XRGB8888, R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8)),
 	_ASSIGN(RGB888, R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8)),
 	_ASSIGN(RGB565, R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5)),
 	_ASSIGN(RGB565_REV, R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5)),
@@ -202,9 +203,7 @@ void r300SetDepthTexMode(struct gl_texture_object *tObj)
 static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 {
 	const struct gl_texture_image *firstImage;
-	int firstlevel = t->mt ? t->mt->firstLevel : 0;
-	    
-	firstImage = t->base.Image[0][firstlevel];
+	firstImage = t->base.Image[0][t->minLod];
 
 	if (!t->image_override
 	    && VALID_FORMAT(firstImage->TexFormat)) {
@@ -227,7 +226,7 @@ static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 	t->pp_txsize = (((R300_TX_WIDTHMASK_MASK & ((firstImage->Width - 1) << R300_TX_WIDTHMASK_SHIFT)))
 			| ((R300_TX_HEIGHTMASK_MASK & ((firstImage->Height - 1) << R300_TX_HEIGHTMASK_SHIFT)))
 			| ((R300_TX_DEPTHMASK_MASK & ((firstImage->DepthLog2) << R300_TX_DEPTHMASK_SHIFT)))
-			| ((R300_TX_MAX_MIP_LEVEL_MASK & ((t->mt->lastLevel - t->mt->firstLevel) << R300_TX_MAX_MIP_LEVEL_SHIFT))));
+			| ((R300_TX_MAX_MIP_LEVEL_MASK & ((t->maxLod - t->minLod) << R300_TX_MAX_MIP_LEVEL_SHIFT))));
 
 	t->tile_bits = 0;
 
@@ -238,7 +237,7 @@ static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 
 
 	if (t->base.Target == GL_TEXTURE_RECTANGLE_NV) {
-		unsigned int align = (64 / t->mt->bpp) - 1;
+		unsigned int align = (64 / _mesa_get_format_bytes(firstImage->TexFormat)) - 1;
 		t->pp_txsize |= R300_TX_SIZE_TXPITCH_EN;
 		if (!t->image_override)
 			t->pp_txpitch = ((firstImage->Width + align) & ~align) - 1;
@@ -437,14 +436,10 @@ void r300SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c
index dbd233729c..25314eff56 100644
--- a/src/mesa/drivers/dri/r600/r600_context.c
+++ b/src/mesa/drivers/dri/r600/r600_context.c
@@ -74,6 +74,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "utils.h"
 #include "xmlpool.h"		/* for symbolic values of enum-type options */
 
+//#define R600_ENABLE_GLSL_TEST 1
+
 #define need_GL_VERSION_2_0
 #define need_GL_ARB_occlusion_query
 #define need_GL_ARB_point_parameters
@@ -109,6 +111,7 @@ static const struct dri_extension card_extensions[] = {
   {"GL_ARB_texture_env_crossbar",	NULL},
   {"GL_ARB_texture_env_dot3",		NULL},
   {"GL_ARB_texture_mirrored_repeat",	NULL},
+  {"GL_ARB_texture_non_power_of_two",   NULL},
   {"GL_ARB_vertex_program",		GL_ARB_vertex_program_functions},
   {"GL_EXT_blend_equation_separate",	GL_EXT_blend_equation_separate_functions},
   {"GL_EXT_blend_func_separate",	GL_EXT_blend_func_separate_functions},
@@ -155,7 +158,11 @@ static const struct dri_extension mm_extensions[] = {
  * functions added by GL_ATI_separate_stencil.
  */
 static const struct dri_extension gl_20_extension[] = {
+#ifdef R600_ENABLE_GLSL_TEST
+    {"GL_ARB_shading_language_100",			GL_VERSION_2_0_functions },
+#else
   {"GL_VERSION_2_0",			GL_VERSION_2_0_functions },
+#endif /* R600_ENABLE_GLSL_TEST */
 };
 
 static const struct tnl_pipeline_stage *r600_pipeline[] = {
@@ -308,6 +315,26 @@ static void r600InitGLExtensions(GLcontext *ctx)
 	if (r600->radeon.radeonScreen->kernel_mm)
 	  driInitExtensions(ctx, mm_extensions, GL_FALSE);
 
+#ifdef R600_ENABLE_GLSL_TEST
+    driInitExtensions(ctx, gl_20_extension, GL_TRUE);
+    //_mesa_enable_2_0_extensions(ctx);
+    //1.5
+    ctx->Extensions.ARB_occlusion_query = GL_TRUE;
+    ctx->Extensions.ARB_vertex_buffer_object = GL_TRUE;
+    ctx->Extensions.EXT_shadow_funcs = GL_TRUE;
+    //2.0
+    ctx->Extensions.ARB_draw_buffers = GL_TRUE;
+    ctx->Extensions.ARB_point_sprite = GL_TRUE;
+    ctx->Extensions.ARB_shader_objects = GL_TRUE;
+    ctx->Extensions.ARB_vertex_shader = GL_TRUE;
+    ctx->Extensions.ARB_fragment_shader = GL_TRUE;
+    ctx->Extensions.EXT_blend_equation_separate = GL_TRUE;
+    ctx->Extensions.ATI_separate_stencil = GL_TRUE;
+
+    /* glsl compiler has problem if this is not GL_TRUE */
+    ctx->Shader.EmitCondCodes = GL_TRUE;
+#endif /* R600_ENABLE_GLSL_TEST */
+
 	if (driQueryOptionb
 	    (&r600->radeon.optionCache, "disable_stencil_two_side"))
 		_mesa_disable_extension(ctx, "GL_EXT_stencil_two_side");
diff --git a/src/mesa/drivers/dri/r600/r600_tex.c b/src/mesa/drivers/dri/r600/r600_tex.c
index 20965bb3c8..9d83a64e22 100644
--- a/src/mesa/drivers/dri/r600/r600_tex.c
+++ b/src/mesa/drivers/dri/r600/r600_tex.c
@@ -312,16 +312,7 @@ static void r600TexParameter(GLcontext * ctx, GLenum target,
 	case GL_TEXTURE_MAX_LEVEL:
 	case GL_TEXTURE_MIN_LOD:
 	case GL_TEXTURE_MAX_LOD:
-		/* This isn't the most efficient solution but there doesn't appear to
-		 * be a nice alternative.  Since there's no LOD clamping,
-		 * we just have to rely on loading the right subset of mipmap levels
-		 * to simulate a clamped LOD.
-		 */
-		if (t->mt) {
-			radeon_miptree_unreference(t->mt);
-			t->mt = 0;
-			t->validated = GL_FALSE;
-		}
+		t->validated = GL_FALSE;
 		break;
 
 	case GL_DEPTH_TEXTURE_MODE:
@@ -369,10 +360,8 @@ static void r600DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 		t->bo = NULL;
 	}
 
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
+	radeon_miptree_unreference(&t->mt);
+
 	_mesa_delete_texture_object(ctx, texObj);
 }
 
diff --git a/src/mesa/drivers/dri/r600/r600_texstate.c b/src/mesa/drivers/dri/r600/r600_texstate.c
index 27c8354923..4ec315b78c 100644
--- a/src/mesa/drivers/dri/r600/r600_texstate.c
+++ b/src/mesa/drivers/dri/r600/r600_texstate.c
@@ -649,7 +649,6 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 {
 	radeonTexObj *t = radeon_tex_obj(texObj);
 	const struct gl_texture_image *firstImage;
-	int firstlevel = t->mt ? t->mt->firstLevel : 0;
 	GLuint uTexelPitch, row_align;
 
 	if (rmesa->radeon.radeonScreen->driScreen->dri2.enabled &&
@@ -657,7 +656,7 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	    t->bo)
 		return;
 
-	firstImage = t->base.Image[0][firstlevel];
+	firstImage = t->base.Image[0][t->minLod];
 
 	if (!t->image_override) {
 		if (!r600GetTexFormat(texObj, firstImage->TexFormat)) {
@@ -692,7 +691,8 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	}
 
 	row_align = rmesa->radeon.texture_row_align - 1;
-	uTexelPitch = ((firstImage->Width * t->mt->bpp + row_align) & ~row_align) / t->mt->bpp;
+	uTexelPitch = (_mesa_format_row_stride(firstImage->TexFormat, firstImage->Width) + row_align) & ~row_align;
+	uTexelPitch = uTexelPitch / _mesa_get_format_bytes(firstImage->TexFormat);
 	uTexelPitch = (uTexelPitch + R700_TEXEL_PITCH_ALIGNMENT_MASK)
 		& ~R700_TEXEL_PITCH_ALIGNMENT_MASK;
 
@@ -706,10 +706,10 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	SETfield(t->SQ_TEX_RESOURCE1, firstImage->Height - 1,
 		 TEX_HEIGHT_shift, TEX_HEIGHT_mask);
 
-	if ((t->mt->lastLevel - t->mt->firstLevel) > 0) {
-		t->SQ_TEX_RESOURCE3 = t->mt->levels[0].size / 256;
-		SETfield(t->SQ_TEX_RESOURCE4, t->mt->firstLevel, BASE_LEVEL_shift, BASE_LEVEL_mask);
-		SETfield(t->SQ_TEX_RESOURCE5, t->mt->lastLevel, LAST_LEVEL_shift, LAST_LEVEL_mask);
+	if ((t->maxLod - t->minLod) > 0) {
+		t->SQ_TEX_RESOURCE3 = t->mt->levels[t->minLod].size / 256;
+		SETfield(t->SQ_TEX_RESOURCE4, 0, BASE_LEVEL_shift, BASE_LEVEL_mask);
+		SETfield(t->SQ_TEX_RESOURCE5, t->maxLod - t->minLod, LAST_LEVEL_shift, LAST_LEVEL_mask);
 	}
 }
 
@@ -808,9 +808,8 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	struct gl_texture_object *tObj =
 	    _mesa_lookup_texture(rmesa->radeon.glCtx, texname);
 	radeonTexObjPtr t = radeon_tex_obj(tObj);
-	int firstlevel = t->mt ? t->mt->firstLevel : 0;
 	const struct gl_texture_image *firstImage;
-	uint32_t pitch_val, size, row_align, bpp;
+	uint32_t pitch_val, size, row_align;
 
 	if (!tObj)
 		return;
@@ -820,13 +819,9 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	if (!offset)
 		return;
 
-	bpp = depth / 8;
-	if (bpp == 3) 
-		bpp = 4;
-
-	firstImage = t->base.Image[0][firstlevel];
+	firstImage = t->base.Image[0][t->minLod];
 	row_align = rmesa->radeon.texture_row_align - 1;
-	size = ((firstImage->Width * bpp + row_align) & ~row_align) * firstImage->Height;
+	size = ((_mesa_format_row_stride(firstImage->TexFormat, firstImage->Width) + row_align) & ~row_align) * firstImage->Height;
 	if (t->bo) {
 		radeon_bo_unref(t->bo);
 		t->bo = NULL;
@@ -949,14 +944,10 @@ void r600SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c
index e0d7d4fa6b..ba97d3e073 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.c
+++ b/src/mesa/drivers/dri/r600/r700_assembler.c
@@ -38,6 +38,9 @@
 
 #include "r700_assembler.h"
 
+#define USE_CF_FOR_CONTINUE_BREAK 1
+#define USE_CF_FOR_POP_AFTER      1
+
 BITS addrmode_PVSDST(PVSDST * pPVSDST)
 {
 	return pPVSDST->addrmode0 | ((BITS)pPVSDST->addrmode1 << 1);
@@ -337,12 +340,17 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm)
     switch (pAsm->D.dst.opcode)
     {
     case SQ_OP2_INST_ADD:
+    case SQ_OP2_INST_KILLE:
     case SQ_OP2_INST_KILLGT:
+    case SQ_OP2_INST_KILLGE:
+    case SQ_OP2_INST_KILLNE:
     case SQ_OP2_INST_MUL: 
     case SQ_OP2_INST_MAX:
     case SQ_OP2_INST_MIN:
     //case SQ_OP2_INST_MAX_DX10:
     //case SQ_OP2_INST_MIN_DX10:
+    case SQ_OP2_INST_SETE: 
+    case SQ_OP2_INST_SETNE:
     case SQ_OP2_INST_SETGT:
     case SQ_OP2_INST_SETGE:
     case SQ_OP2_INST_PRED_SETE:
@@ -358,6 +366,7 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm)
     case SQ_OP2_INST_MOVA_FLOOR:
     case SQ_OP2_INST_FRACT:
     case SQ_OP2_INST_FLOOR:
+    case SQ_OP2_INST_TRUNC:
     case SQ_OP2_INST_EXP_IEEE:
     case SQ_OP2_INST_LOG_CLAMPED:
     case SQ_OP2_INST_LOG_IEEE:
@@ -383,98 +392,115 @@ int Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700
     pAsm->pR700Shader = pShader;
     pAsm->currentShaderType = spt;
 
-	pAsm->cf_last_export_ptr   = NULL;
+    pAsm->cf_last_export_ptr   = NULL;
+
+    pAsm->cf_current_export_clause_ptr = NULL;
+    pAsm->cf_current_alu_clause_ptr    = NULL;
+    pAsm->cf_current_tex_clause_ptr    = NULL;
+    pAsm->cf_current_vtx_clause_ptr    = NULL;
+    pAsm->cf_current_cf_clause_ptr     = NULL;
 
-	pAsm->cf_current_export_clause_ptr = NULL;
-	pAsm->cf_current_alu_clause_ptr    = NULL;
-	pAsm->cf_current_tex_clause_ptr    = NULL;
-	pAsm->cf_current_vtx_clause_ptr    = NULL;
-	pAsm->cf_current_cf_clause_ptr     = NULL;
+    // No clause has been created yet
+    pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE;
 
-	// No clause has been created yet
-	pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE;
+    pAsm->number_of_colorandz_exports = 0;
+    pAsm->number_of_exports           = 0;
+    pAsm->number_of_export_opcodes    = 0;
 
-	pAsm->number_of_colorandz_exports = 0;
-	pAsm->number_of_exports           = 0;
-	pAsm->number_of_export_opcodes    = 0;
+    pAsm->alu_x_opcode = 0;
 
+    pAsm->D2.bits = 0;
 
-	pAsm->D.bits = 0;
-	pAsm->S[0].bits = 0;
-	pAsm->S[1].bits = 0;
-	pAsm->S[2].bits = 0;
+    pAsm->D.bits = 0;
+    pAsm->S[0].bits = 0;
+    pAsm->S[1].bits = 0;
+    pAsm->S[2].bits = 0;
 
-	pAsm->uLastPosUpdate = 0; 
+    pAsm->uLastPosUpdate = 0; 
 	
-	*(BITS *) &pAsm->fp_stOutFmt0 = 0;
+    *(BITS *) &pAsm->fp_stOutFmt0 = 0;
 
-	pAsm->uIIns = 0;
-	pAsm->uOIns = 0;
-	pAsm->number_used_registers = 0;
-	pAsm->uUsedConsts = 256; 
+    pAsm->uIIns = 0;
+    pAsm->uOIns = 0;
+    pAsm->number_used_registers = 0;
+    pAsm->uUsedConsts = 256; 
 
 
-	// Fragment programs
-	pAsm->uBoolConsts = 0;
-	pAsm->uIntConsts = 0;
-	pAsm->uInsts = 0;
-	pAsm->uConsts = 0;
+    // Fragment programs
+    pAsm->uBoolConsts = 0;
+    pAsm->uIntConsts = 0;
+    pAsm->uInsts = 0;
+    pAsm->uConsts = 0;
 
-	pAsm->FCSP = 0;
-	pAsm->fc_stack[0].type = FC_NONE;
+    pAsm->FCSP = 0;
+    pAsm->fc_stack[0].type = FC_NONE;
 
-	pAsm->branch_depth     = 0;
-	pAsm->max_branch_depth = 0;
+    pAsm->aArgSubst[0] =
+    pAsm->aArgSubst[1] =
+    pAsm->aArgSubst[2] =
+    pAsm->aArgSubst[3] = (-1);
 
-	pAsm->aArgSubst[0] =
-	pAsm->aArgSubst[1] =
-	pAsm->aArgSubst[2] =
-	pAsm->aArgSubst[3] = (-1);
+    pAsm->uOutputs = 0;
 
-	pAsm->uOutputs = 0;
+    for (i=0; i<NUMBER_OF_OUTPUT_COLORS; i++) 
+    {
+        pAsm->color_export_register_number[i] = (-1);
+    }
 
-	for (i=0; i<NUMBER_OF_OUTPUT_COLORS; i++) 
-	{
-		pAsm->color_export_register_number[i] = (-1);
-	}
 
+    pAsm->depth_export_register_number = (-1);
+    pAsm->stencil_export_register_number = (-1);
+    pAsm->coverage_to_mask_export_register_number = (-1);
+    pAsm->mask_export_register_number = (-1);
 
-	pAsm->depth_export_register_number = (-1);
-	pAsm->stencil_export_register_number = (-1);
-	pAsm->coverage_to_mask_export_register_number = (-1);
-	pAsm->mask_export_register_number = (-1);
+    pAsm->starting_export_register_number = 0;
+    pAsm->starting_vfetch_register_number = 0;
+    pAsm->starting_temp_register_number   = 0;
+    pAsm->uFirstHelpReg = 0;
 
-	pAsm->starting_export_register_number = 0;
-	pAsm->starting_vfetch_register_number = 0;
-	pAsm->starting_temp_register_number   = 0;
-	pAsm->uFirstHelpReg = 0;
+    pAsm->input_position_is_used = GL_FALSE;
+    pAsm->input_normal_is_used   = GL_FALSE;
 
+    for (i=0; i<NUMBER_OF_INPUT_COLORS; i++) 
+    {
+        pAsm->input_color_is_used[ i ] = GL_FALSE;
+    }
 
-	pAsm->input_position_is_used = GL_FALSE;
-	pAsm->input_normal_is_used   = GL_FALSE;
+    for (i=0; i<NUMBER_OF_TEXTURE_UNITS; i++) 
+    {
+        pAsm->input_texture_unit_is_used[ i ] = GL_FALSE;
+    }
 
+    for (i=0; i<VERT_ATTRIB_MAX; i++) 
+    {
+        pAsm->vfetch_instruction_ptr_array[ i ] = NULL;
+    }
 
-	for (i=0; i<NUMBER_OF_INPUT_COLORS; i++) 
-	{
-		pAsm->input_color_is_used[ i ] = GL_FALSE;
-	}
+    pAsm->number_of_inputs = 0;
 
-	for (i=0; i<NUMBER_OF_TEXTURE_UNITS; i++) 
-	{
-		pAsm->input_texture_unit_is_used[ i ] = GL_FALSE;
-	}
+    pAsm->is_tex = GL_FALSE;
+    pAsm->need_tex_barrier = GL_FALSE;
 
-	for (i=0; i<VERT_ATTRIB_MAX; i++) 
-	{
-		pAsm->vfetch_instruction_ptr_array[ i ] = NULL;
-	}
+    pAsm->subs              = NULL;
+    pAsm->unSubArraySize    = 0;
+    pAsm->unSubArrayPointer = 0;
+    pAsm->callers              = NULL;
+    pAsm->unCallerArraySize    = 0;
+    pAsm->unCallerArrayPointer = 0;
+
+    pAsm->CALLSP = 0;
+    pAsm->CALLSTACK[0].FCSP_BeforeEntry = 0;
+    pAsm->CALLSTACK[0].plstCFInstructions_local
+          = &(pAsm->pR700Shader->lstCFInstructions);
 
-	pAsm->number_of_inputs = 0;
+    pAsm->CALLSTACK[0].max = 0;
+    pAsm->CALLSTACK[0].current = 0;
 
-	pAsm->is_tex = GL_FALSE;
-	pAsm->need_tex_barrier = GL_FALSE;
+    SetActiveCFlist(pAsm->pR700Shader, pAsm->CALLSTACK[0].plstCFInstructions_local);
 
-	return 0;
+    pAsm->unCFflags = 0;
+
+    return 0;
 }
 
 GLboolean IsTex(gl_inst_opcode Opcode)
@@ -592,6 +618,31 @@ int check_current_clause(r700_AssemblerBase* pAsm,
     return GL_TRUE;
 }
 
+GLboolean add_cf_instruction(r700_AssemblerBase* pAsm)
+{
+    if(GL_FALSE == check_current_clause(pAsm, CF_OTHER_CLAUSE))
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr = 
+      (R700ControlFlowGenericClause*) CALLOC_STRUCT(R700ControlFlowGenericClause);
+
+    if (pAsm->cf_current_cf_clause_ptr != NULL) 
+	{
+		Init_R700ControlFlowGenericClause(pAsm->cf_current_cf_clause_ptr);
+		AddCFInstruction( pAsm->pR700Shader, 
+                          (R700ControlFlowInstruction *)pAsm->cf_current_cf_clause_ptr );
+	}
+	else 
+	{
+        radeon_error("Could not allocate a new VFetch CF instruction.\n");
+		return GL_FALSE;
+	}
+
+    return GL_TRUE;
+}
+
 GLboolean add_vfetch_instruction(r700_AssemblerBase*     pAsm,
 								 R700VertexInstruction*  vertex_instruction_ptr)
 {
@@ -1153,6 +1204,7 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm,
         case PROGRAM_LOCAL_PARAM:
         case PROGRAM_ENV_PARAM:
         case PROGRAM_STATE_VAR:
+        case PROGRAM_UNIFORM:
             if (1 == pILInst->SrcReg[src].RelAddr)
             {
                 setaddrmode_PVSSRC(&(pAsm->S[fld].src), ADDR_RELATIVE_A0);
@@ -1179,7 +1231,7 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm,
             }
             break;      
         default:
-            radeon_error("Invalid source argument type\n");
+            radeon_error("Invalid source argument type : %d \n", pILInst->SrcReg[src].File);
             return GL_FALSE;
         }
     } 
@@ -1315,7 +1367,7 @@ GLboolean tex_src(r700_AssemblerBase *pAsm)
                 case FRAG_ATTRIB_TEX0:
                 case FRAG_ATTRIB_TEX1:
                 case FRAG_ATTRIB_TEX2:
-	        case FRAG_ATTRIB_TEX3:
+                case FRAG_ATTRIB_TEX3:
                 case FRAG_ATTRIB_TEX4:
                 case FRAG_ATTRIB_TEX5:
                 case FRAG_ATTRIB_TEX6:
@@ -1331,10 +1383,17 @@ GLboolean tex_src(r700_AssemblerBase *pAsm)
                 case FRAG_ATTRIB_PNTC:
                     fprintf(stderr, "FRAG_ATTRIB_PNTC unsupported\n");
                     break;
-                case FRAG_ATTRIB_VAR0:
-                    fprintf(stderr, "FRAG_ATTRIB_VAR0 unsupported\n");
-                    break;
             }
+
+            if( (pILInst->SrcReg[0].Index >= FRAG_ATTRIB_VAR0) ||
+                (pILInst->SrcReg[0].Index < FRAG_ATTRIB_MAX) )
+            {
+				bValidTexCoord = GL_TRUE;
+                pAsm->S[0].src.reg   =
+                    pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index];
+                pAsm->S[0].src.rtype = SRC_REG_INPUT;
+            }
+
         break;
         }
     }
@@ -1517,6 +1576,10 @@ GLboolean assemble_alu_src(R700ALUInstruction*  alu_instruction_ptr,
         {
             src_sel = pSource->reg + CFILE_REGISTER_OFFSET;            
         }
+        else if (pSource->rtype == SRC_REC_LITERAL)
+        {
+            src_sel = SQ_ALU_SRC_LITERAL;            
+        }
         else
         {
             radeon_error("Source (%d) register type (%d) not one of TEMP, INPUT, or CONSTANT.\n",
@@ -1606,7 +1669,8 @@ GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
         return GL_FALSE;
     }
 
-    if ( pAsm->cf_current_alu_clause_ptr == NULL ||
+    if ( pAsm->alu_x_opcode != 0 ||
+         pAsm->cf_current_alu_clause_ptr == NULL ||
          ( (pAsm->cf_current_alu_clause_ptr != NULL) && 
            (pAsm->cf_current_alu_clause_ptr->m_Word1.f.count >= (GetCFMaxInstructions(pAsm->cf_current_alu_clause_ptr->m_ShaderInstType)-contiguous_slots_needed-1) )
          ) ) 
@@ -1636,9 +1700,17 @@ GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.kcache_addr0 = 0x0;
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.kcache_addr1 = 0x0;
 
-        //cf_current_alu_clause_ptr->m_Word1.f.count           = number_of_scalar_operations - 1;
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.count           = 0x0;
-        pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst         = SQ_CF_INST_ALU;
+
+        if(pAsm->alu_x_opcode != 0)
+        {
+            pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst = pAsm->alu_x_opcode;
+            pAsm->alu_x_opcode = 0;
+        }
+        else
+        {
+            pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst = SQ_CF_INST_ALU;
+        }
 
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.whole_quad_mode = 0x0;
 
@@ -2358,6 +2430,506 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
+GLboolean assemble_alu_instruction2(r700_AssemblerBase *pAsm)
+{
+    GLuint    number_of_scalar_operations;
+    GLboolean is_single_scalar_operation;
+    GLuint    scalar_channel_index;
+
+    PVSSRC * pcurrent_source;
+    int    current_source_index;
+    GLuint contiguous_slots_needed;
+
+    GLuint    uNumSrc = r700GetNumOperands(pAsm);
+    
+    GLboolean bSplitInst = GL_FALSE;
+
+    if (1 == pAsm->D.dst.math) 
+    {
+        is_single_scalar_operation = GL_TRUE;
+        number_of_scalar_operations = 1;
+    }
+    else 
+    {
+        is_single_scalar_operation = GL_FALSE;
+        number_of_scalar_operations = 4;
+    }
+
+    contiguous_slots_needed = 0;
+
+    if(GL_TRUE == is_reduction_opcode(&(pAsm->D)) ) 
+    {
+        contiguous_slots_needed = 4;
+    }
+
+    initialize(pAsm);    
+
+    for (scalar_channel_index=0;
+            scalar_channel_index < number_of_scalar_operations; 
+                scalar_channel_index++) 
+    {
+        R700ALUInstruction* alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
+        if (alu_instruction_ptr == NULL) 
+        {
+            return GL_FALSE;
+        }
+        Init_R700ALUInstruction(alu_instruction_ptr);
+        
+        //src 0
+        current_source_index = 0;
+        pcurrent_source = &(pAsm->S[0].src);
+
+        if (GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                         current_source_index,
+                                         pcurrent_source, 
+                                         scalar_channel_index) )     
+        {
+            return GL_FALSE;
+        }
+   
+        if (uNumSrc > 1) 
+        {            
+            // Process source 1            
+            current_source_index = 1;
+            pcurrent_source = &(pAsm->S[current_source_index].src);
+
+            if (GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                             current_source_index,
+                                             pcurrent_source, 
+                                             scalar_channel_index) ) 
+            {
+                return GL_FALSE;
+            }
+        }
+
+        //other bits
+        alu_instruction_ptr->m_Word0.f.index_mode = SQ_INDEX_LOOP;
+
+        if(   (is_single_scalar_operation == GL_TRUE) 
+           || (GL_TRUE == bSplitInst) )
+        {
+            alu_instruction_ptr->m_Word0.f.last = 1;
+        }
+        else 
+        {
+            alu_instruction_ptr->m_Word0.f.last = (scalar_channel_index == 3) ?  1 : 0;
+        }
+
+        alu_instruction_ptr->m_Word0.f.pred_sel = (pAsm->D.dst.pred_inv > 0) ? 1 : 0;
+        if(1 == pAsm->D.dst.predicated)
+        {
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x1;  
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x1; 
+        }
+        else
+        {
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x0;  
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x0; 
+        }
+       
+        // dst
+        if( (pAsm->D.dst.rtype == DST_REG_TEMPORARY) || 
+            (pAsm->D.dst.rtype == DST_REG_OUT) ) 
+        {
+            alu_instruction_ptr->m_Word1.f.dst_gpr  = pAsm->D.dst.reg;
+        }
+        else 
+        {
+            radeon_error("Only temp destination registers supported for ALU dest regs.\n");
+            return GL_FALSE;
+        }
+
+        alu_instruction_ptr->m_Word1.f.dst_rel  = SQ_ABSOLUTE;  //D.rtype
+
+        if ( is_single_scalar_operation == GL_TRUE ) 
+        {
+            // Override scalar_channel_index since only one scalar value will be written
+            if(pAsm->D.dst.writex) 
+            {
+                scalar_channel_index = 0;
+            }
+            else if(pAsm->D.dst.writey) 
+            {
+                scalar_channel_index = 1;
+            }
+            else if(pAsm->D.dst.writez) 
+            {
+                scalar_channel_index = 2;
+            }
+            else if(pAsm->D.dst.writew) 
+            {
+                scalar_channel_index = 3;
+            }
+        }
+
+        alu_instruction_ptr->m_Word1.f.dst_chan = scalar_channel_index;
+
+        alu_instruction_ptr->m_Word1.f.clamp    = pAsm->D2.dst2.SaturateMode;
+
+        if (pAsm->D.dst.op3) 
+        {            
+            //op3
+
+            alu_instruction_ptr->m_Word1_OP3.f.alu_inst = pAsm->D.dst.opcode;
+
+            //There's 3rd src for op3
+            current_source_index = 2;
+            pcurrent_source = &(pAsm->S[current_source_index].src);
+
+            if ( GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                              current_source_index,
+                                              pcurrent_source, 
+                                              scalar_channel_index) ) 
+            {
+                return GL_FALSE;
+            }
+        }
+        else 
+        {
+            //op2
+            if (pAsm->bR6xx)
+            {
+                alu_instruction_ptr->m_Word1_OP2.f6.alu_inst           = pAsm->D.dst.opcode;
+
+                alu_instruction_ptr->m_Word1_OP2.f6.src0_abs           = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f6.src1_abs           = 0x0;
+
+                //alu_instruction_ptr->m_Word1_OP2.f6.update_execute_mask = 0x0;
+                //alu_instruction_ptr->m_Word1_OP2.f6.update_pred         = 0x0;
+                switch (scalar_channel_index) 
+                {
+                    case 0: 
+                        alu_instruction_ptr->m_Word1_OP2.f6.write_mask = pAsm->D.dst.writex; 
+                        break;
+                    case 1: 
+                        alu_instruction_ptr->m_Word1_OP2.f6.write_mask = pAsm->D.dst.writey; 
+                        break;
+                    case 2: 
+                        alu_instruction_ptr->m_Word1_OP2.f6.write_mask = pAsm->D.dst.writez; 
+                        break;
+                    case 3: 
+                        alu_instruction_ptr->m_Word1_OP2.f6.write_mask = pAsm->D.dst.writew; 
+                        break;
+                    default: 
+                        alu_instruction_ptr->m_Word1_OP2.f6.write_mask = 1; //SQ_SEL_MASK;
+                        break;
+                }            
+                alu_instruction_ptr->m_Word1_OP2.f6.omod               = SQ_ALU_OMOD_OFF;
+            }
+            else
+            {
+                alu_instruction_ptr->m_Word1_OP2.f.alu_inst           = pAsm->D.dst.opcode;
+
+                alu_instruction_ptr->m_Word1_OP2.f.src0_abs           = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f.src1_abs           = 0x0;
+
+                //alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x0;
+                //alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x0;
+                switch (scalar_channel_index) 
+                {
+                    case 0: 
+                        alu_instruction_ptr->m_Word1_OP2.f.write_mask = pAsm->D.dst.writex; 
+                        break;
+                    case 1: 
+                        alu_instruction_ptr->m_Word1_OP2.f.write_mask = pAsm->D.dst.writey; 
+                        break;
+                    case 2: 
+                        alu_instruction_ptr->m_Word1_OP2.f.write_mask = pAsm->D.dst.writez; 
+                        break;
+                    case 3: 
+                        alu_instruction_ptr->m_Word1_OP2.f.write_mask = pAsm->D.dst.writew; 
+                        break;
+                    default: 
+                        alu_instruction_ptr->m_Word1_OP2.f.write_mask = 1; //SQ_SEL_MASK;
+                        break;
+                }            
+                alu_instruction_ptr->m_Word1_OP2.f.omod               = SQ_ALU_OMOD_OFF;
+            }
+        }
+
+        if(GL_FALSE == add_alu_instruction(pAsm, alu_instruction_ptr, contiguous_slots_needed) )
+        {
+            return GL_FALSE;
+        }
+
+        /*
+         * Judge the type of current instruction, is it vector or scalar 
+         * instruction.
+         */        
+        if (is_single_scalar_operation) 
+        {
+            if(GL_FALSE == check_scalar(pAsm, alu_instruction_ptr) )
+            {
+                return GL_FALSE;
+            }
+        }
+        else 
+        {
+            if(GL_FALSE == check_vector(pAsm, alu_instruction_ptr) )
+            {
+                return 1;
+            }
+        }
+
+        contiguous_slots_needed = 0;
+    }
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_alu_instruction_literal(r700_AssemblerBase *pAsm, GLfloat * pLiteral)
+{
+    R700ALUInstruction            * alu_instruction_ptr;
+    R700ALUInstructionHalfLiteral * alu_instruction_ptr_hl;
+    R700ALUInstructionFullLiteral * alu_instruction_ptr_fl;
+
+    GLuint    number_of_scalar_operations;
+    GLboolean is_single_scalar_operation;
+    GLuint    scalar_channel_index;
+
+    GLuint   contiguous_slots_needed;
+    GLuint   lastInstruction;
+    GLuint   not_masked[4];
+
+    GLuint    uNumSrc = r700GetNumOperands(pAsm);
+    
+    GLboolean bSplitInst = GL_FALSE;
+
+    number_of_scalar_operations = 0;
+    contiguous_slots_needed     = 0;
+
+    if(1 == pAsm->D.dst.writew)
+    {
+        lastInstruction = 3;
+        number_of_scalar_operations++;
+        not_masked[3] = 1;
+    }
+    else
+    {
+        not_masked[3] = 0;
+    }
+    if(1 == pAsm->D.dst.writez)
+    {
+        lastInstruction = 2;
+        number_of_scalar_operations++;
+        not_masked[2] = 1;
+    }
+    else
+    {
+        not_masked[2] = 0;
+    }
+    if(1 == pAsm->D.dst.writey)
+    {
+        lastInstruction = 1;
+        number_of_scalar_operations++;
+        not_masked[1] = 1;
+    }
+    else
+    {
+        not_masked[1] = 0;
+    }
+    if(1 == pAsm->D.dst.writex)
+    {
+        lastInstruction = 0;
+        number_of_scalar_operations++;
+        not_masked[0] = 1;
+    }
+    else
+    {
+        not_masked[0] = 0;
+    }
+    
+    if(GL_TRUE == is_reduction_opcode(&(pAsm->D)) ) 
+    {
+        contiguous_slots_needed = 4;
+    }
+    else
+    {
+        contiguous_slots_needed = number_of_scalar_operations;
+    }
+
+    if(1 == pAsm->D2.dst2.literal)
+    {
+        contiguous_slots_needed += 1;
+    }
+    else if(2 == pAsm->D2.dst2.literal)
+    {
+        contiguous_slots_needed += 2;
+    }
+
+    initialize(pAsm);    
+
+    for (scalar_channel_index=0; scalar_channel_index < 4; scalar_channel_index++) 
+    {
+        if(0 == not_masked[scalar_channel_index])
+        {
+            continue;
+        }
+
+        if(scalar_channel_index == lastInstruction)
+        {
+            switch (pAsm->D2.dst2.literal)
+            {
+            case 0:
+                alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
+                if (alu_instruction_ptr == NULL) 
+		        {
+			        return GL_FALSE;
+		        }
+                Init_R700ALUInstruction(alu_instruction_ptr);
+                break;
+            case 1:
+                alu_instruction_ptr_hl = (R700ALUInstructionHalfLiteral*) CALLOC_STRUCT(R700ALUInstructionHalfLiteral);
+                if (alu_instruction_ptr_hl == NULL) 
+		        {
+			        return GL_FALSE;
+		        }
+                Init_R700ALUInstructionHalfLiteral(alu_instruction_ptr_hl, pLiteral[0], pLiteral[1]);
+                alu_instruction_ptr = (R700ALUInstruction*)alu_instruction_ptr_hl;
+                break;
+            case 2:
+                alu_instruction_ptr_fl = (R700ALUInstructionFullLiteral*) CALLOC_STRUCT(R700ALUInstructionFullLiteral);
+                if (alu_instruction_ptr_fl == NULL) 
+		        {
+			        return GL_FALSE;
+		        }
+                Init_R700ALUInstructionFullLiteral(alu_instruction_ptr_fl, pLiteral[0], pLiteral[1], pLiteral[2], pLiteral[3]);
+                alu_instruction_ptr = (R700ALUInstruction*)alu_instruction_ptr_fl;
+                break;
+            default:
+                break;
+            };
+        }
+        else
+        {
+            alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
+            if (alu_instruction_ptr == NULL) 
+		    {
+			    return GL_FALSE;
+		    }
+            Init_R700ALUInstruction(alu_instruction_ptr);
+        }
+
+        //src 0
+        if (GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                         0,
+                                         &(pAsm->S[0].src), 
+                                         scalar_channel_index) )     
+        {
+            return GL_FALSE;
+        }
+   
+        if (uNumSrc > 1) 
+        {            
+            // Process source 1            
+            if (GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                             1,
+                                             &(pAsm->S[1].src), 
+                                             scalar_channel_index) ) 
+            {
+                return GL_FALSE;
+            }
+        }
+
+        //other bits
+        alu_instruction_ptr->m_Word0.f.index_mode = SQ_INDEX_LOOP;
+
+        if(scalar_channel_index == lastInstruction)
+        {
+            alu_instruction_ptr->m_Word0.f.last = 1;
+        }
+
+        alu_instruction_ptr->m_Word0.f.pred_sel = 0x0;
+        if(1 == pAsm->D.dst.predicated)
+        {            
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x1;  
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x1; 
+        }
+        else
+        {
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0;  
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0; 
+        }
+
+        // dst
+        if( (pAsm->D.dst.rtype == DST_REG_TEMPORARY) || 
+            (pAsm->D.dst.rtype == DST_REG_OUT) ) 
+        {
+            alu_instruction_ptr->m_Word1.f.dst_gpr  = pAsm->D.dst.reg;
+        }
+        else 
+        {
+            radeon_error("Only temp destination registers supported for ALU dest regs.\n");
+            return GL_FALSE;
+        }
+
+        alu_instruction_ptr->m_Word1.f.dst_rel  = SQ_ABSOLUTE;  //D.rtype
+
+        alu_instruction_ptr->m_Word1.f.dst_chan = scalar_channel_index;
+
+        alu_instruction_ptr->m_Word1.f.clamp    = pAsm->D2.dst2.SaturateMode;
+
+        if (pAsm->D.dst.op3) 
+        {            
+            //op3
+            alu_instruction_ptr->m_Word1_OP3.f.alu_inst = pAsm->D.dst.opcode;
+
+            //There's 3rd src for op3
+            if ( GL_FALSE == assemble_alu_src(alu_instruction_ptr,
+                                              2,
+                                              &(pAsm->S[2].src), 
+                                              scalar_channel_index) ) 
+            {
+                return GL_FALSE;
+            }
+        }
+        else 
+        {
+            //op2
+            if (pAsm->bR6xx)
+            {
+                alu_instruction_ptr->m_Word1_OP2.f6.alu_inst   = pAsm->D.dst.opcode;
+                alu_instruction_ptr->m_Word1_OP2.f6.src0_abs   = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f6.src1_abs   = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f6.write_mask = 1;           
+                alu_instruction_ptr->m_Word1_OP2.f6.omod       = SQ_ALU_OMOD_OFF;
+            }
+            else
+            {
+                alu_instruction_ptr->m_Word1_OP2.f.alu_inst    = pAsm->D.dst.opcode;
+                alu_instruction_ptr->m_Word1_OP2.f.src0_abs    = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f.src1_abs    = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f.write_mask  = 1;                        
+                alu_instruction_ptr->m_Word1_OP2.f.omod        = SQ_ALU_OMOD_OFF;
+            }
+        }
+
+        if(GL_FALSE == add_alu_instruction(pAsm, alu_instruction_ptr, contiguous_slots_needed) )
+        {
+            return GL_FALSE;
+        }
+  
+        if (1 == number_of_scalar_operations) 
+        {
+            if(GL_FALSE == check_scalar(pAsm, alu_instruction_ptr) )
+            {
+                return GL_FALSE;
+            }
+        }
+        else 
+        {
+            if(GL_FALSE == check_vector(pAsm, alu_instruction_ptr) )
+            {
+                return GL_FALSE;
+            }
+        }
+
+        contiguous_slots_needed -= 2;
+    }
+
+    return GL_TRUE;
+}
+
 GLboolean next_ins(r700_AssemblerBase *pAsm)
 {
     struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]);
@@ -2403,6 +2975,70 @@ GLboolean next_ins(r700_AssemblerBase *pAsm)
     
     //reset for next inst.
     pAsm->D.bits    = 0;
+    pAsm->D2.bits   = 0;
+    pAsm->S[0].bits = 0;
+    pAsm->S[1].bits = 0;
+    pAsm->S[2].bits = 0;
+    pAsm->is_tex = GL_FALSE;
+    pAsm->need_tex_barrier = GL_FALSE;
+
+    return GL_TRUE;
+}
+
+GLboolean next_ins2(r700_AssemblerBase *pAsm)
+{
+    struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]);
+
+    //ALU      
+    if( GL_FALSE == assemble_alu_instruction2(pAsm) ) 
+    {
+        radeon_error("Error assembling ALU instruction\n");
+        return GL_FALSE;
+    }
+     
+    if(pAsm->D.dst.rtype == DST_REG_OUT) 
+    {
+        if(pAsm->D.dst.op3) 
+        {        
+            // There is no mask for OP3 instructions, so all channels are written        
+            pAsm->pucOutMask[pAsm->D.dst.reg - pAsm->starting_export_register_number] = 0xF;
+        }
+        else 
+        {
+            pAsm->pucOutMask[pAsm->D.dst.reg - pAsm->starting_export_register_number] 
+               |= (unsigned char)pAsm->pILInst[pAsm->uiCurInst].DstReg.WriteMask;
+        }
+    }
+    
+    //reset for next inst.
+    pAsm->D.bits    = 0;
+    pAsm->D2.bits   = 0;
+    pAsm->S[0].bits = 0;
+    pAsm->S[1].bits = 0;
+    pAsm->S[2].bits = 0;
+    pAsm->is_tex = GL_FALSE;
+    pAsm->need_tex_barrier = GL_FALSE;
+
+    pAsm->D2.bits = 0;
+
+    return GL_TRUE;
+}
+
+/* not work yet */
+GLboolean next_ins_literal(r700_AssemblerBase *pAsm, GLfloat * pLiteral)
+{
+    struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]);
+
+    //ALU      
+    if( GL_FALSE == assemble_alu_instruction_literal(pAsm, pLiteral) ) 
+    {
+        radeon_error("Error assembling ALU instruction\n");
+        return GL_FALSE;
+    }
+    
+    //reset for next inst.
+    pAsm->D.bits    = 0;
+    pAsm->D2.bits   = 0;
     pAsm->S[0].bits = 0;
     pAsm->S[1].bits = 0;
     pAsm->S[2].bits = 0;
@@ -2910,13 +3546,12 @@ GLboolean assemble_FRC(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
  
-GLboolean assemble_KIL(r700_AssemblerBase *pAsm)
-{
-    /* TODO: doc says KILL has to be last(end) ALU clause */
-    
-    checkop1(pAsm);
+GLboolean assemble_KIL(r700_AssemblerBase *pAsm, GLuint opcode)
+{  
+    checkop2(pAsm);
 
-    pAsm->D.dst.opcode = SQ_OP2_INST_KILLGT;  
+    pAsm->D.dst.opcode = opcode;  
+    pAsm->D.dst.math = 1;
 
     setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
     pAsm->D.dst.rtype = DST_REG_TEMPORARY;
@@ -2926,24 +3561,24 @@ GLboolean assemble_KIL(r700_AssemblerBase *pAsm)
     pAsm->D.dst.writez = 0;
     pAsm->D.dst.writew = 0;
 
-    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
-    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
-    pAsm->S[0].src.reg = 0;
-
-    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_0);
-    noneg_PVSSRC(&(pAsm->S[0].src));
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+        return GL_FALSE;
+    }
 
-    if ( GL_FALSE == assemble_src(pAsm, 0, 1) )
+    if( GL_FALSE == assemble_src(pAsm, 1, -1) )
     {
         return GL_FALSE;
     }
   
-    if ( GL_FALSE == next_ins(pAsm) )
+    if ( GL_FALSE == next_ins2(pAsm) )
     {
         return GL_FALSE;
     }
 
+    /* Doc says KILL has to be last(end) ALU clause */
     pAsm->pR700Shader->killIsUsed = GL_TRUE;
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
     
     return GL_TRUE;
 }
@@ -3816,6 +4451,74 @@ GLboolean assemble_SCS(r700_AssemblerBase *pAsm)
 
     return GL_TRUE;
 }
+
+GLboolean assemble_LOGIC(r700_AssemblerBase *pAsm, BITS opcode) 
+{
+    if( GL_FALSE == checkop2(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = opcode;
+    pAsm->D.dst.math   = 1;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 1, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == next_ins(pAsm) ) 
+    {
+	    return GL_FALSE;
+    }
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_LOGIC_PRED(r700_AssemblerBase *pAsm, BITS opcode) 
+{
+    if( GL_FALSE == checkop2(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = opcode;
+    pAsm->D.dst.math   = 1;
+    pAsm->D.dst.predicated = 1;
+    pAsm->D2.dst2.SaturateMode = pAsm->pILInst[pAsm->uiCurInst].SaturateMode;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 1, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == next_ins2(pAsm) ) 
+    {
+	    return GL_FALSE;
+    }
+
+    return GL_TRUE;
+}
  
 GLboolean assemble_SGE(r700_AssemblerBase *pAsm) 
 {
@@ -4273,27 +4976,876 @@ GLboolean assemble_EXPORT(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
-GLboolean assemble_IF(r700_AssemblerBase *pAsm)
+static inline void decreaseCurrent(r700_AssemblerBase *pAsm, GLuint uReason)
+{
+    switch (uReason)
+    {
+    case FC_PUSH_VPM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current--;
+        break;
+    case FC_PUSH_WQM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 4;
+        break;
+    case FC_LOOP:
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 4;
+        break;
+    case FC_REP:
+        /* TODO : for 16 vp asic, should -= 2; */
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 1;
+        break;
+    };
+}
+
+static inline void checkStackDepth(r700_AssemblerBase *pAsm, GLuint uReason, GLboolean bCheckMaxOnly)
+{
+    if(GL_TRUE == bCheckMaxOnly)
+    {
+        switch (uReason)
+        {
+        case FC_PUSH_VPM:
+            if((pAsm->CALLSTACK[pAsm->CALLSP].current + 1)
+                    > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max =
+                    pAsm->CALLSTACK[pAsm->CALLSP].current + 1;
+            }
+            break;
+        case FC_PUSH_WQM:
+            if((pAsm->CALLSTACK[pAsm->CALLSP].current + 4)
+                    > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max =
+                    pAsm->CALLSTACK[pAsm->CALLSP].current + 4;
+            }
+            break;
+        }
+        return;
+    }
+
+    switch (uReason)
+    {
+    case FC_PUSH_VPM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current++;
+        break;
+    case FC_PUSH_WQM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 4;
+        break;
+    case FC_LOOP:
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 4;
+        break;
+    case FC_REP:
+        /* TODO : for 16 vp asic, should += 2; */
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 1;
+        break;
+    };
+
+    if(pAsm->CALLSTACK[pAsm->CALLSP].current
+         > pAsm->CALLSTACK[pAsm->CALLSP].max)
+    {
+        pAsm->CALLSTACK[pAsm->CALLSP].max =
+            pAsm->CALLSTACK[pAsm->CALLSP].current;
+    }
+}
+
+GLboolean jumpToOffest(r700_AssemblerBase *pAsm, GLuint pops, GLint offset)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = pops;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_JUMP;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex + offset;
+
+    return GL_TRUE;
+}
+
+GLboolean pops(r700_AssemblerBase *pAsm, GLuint pops)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = pops;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_IF(r700_AssemblerBase *pAsm, GLboolean bHasElse)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    if(GL_TRUE != bHasElse)
+    {
+        pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count = 1; 
+    }
+    else
+    {
+        pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count = 0;
+    }
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_JUMP;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->FCSP++;
+	pAsm->fc_stack[pAsm->FCSP].type  = FC_IF;
+    pAsm->fc_stack[pAsm->FCSP].mid   = NULL;
+    pAsm->fc_stack[pAsm->FCSP].midLen= 0;
+    pAsm->fc_stack[pAsm->FCSP].first = pAsm->cf_current_cf_clause_ptr;
+
+#ifndef USE_CF_FOR_POP_AFTER
+    if(GL_TRUE != bHasElse)
+    {
+        pAsm->alu_x_opcode = SQ_CF_INST_ALU_POP_AFTER;
+    }
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_FALSE); 
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_ELSE(r700_AssemblerBase *pAsm)
 {
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1; ///
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_ELSE;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[pAsm->FCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( (void *)pAsm->fc_stack[pAsm->FCSP].mid,
+                                                                                     0,
+                                                                                     sizeof(R700ControlFlowGenericClause *) );
+    pAsm->fc_stack[pAsm->FCSP].mid[0] = pAsm->cf_current_cf_clause_ptr;
+    //pAsm->fc_stack[pAsm->FCSP].unNumMid = 1;
+
+#ifndef USE_CF_FOR_POP_AFTER
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_POP_AFTER;
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode - 1; 
+
     return GL_TRUE;
 }
 
 GLboolean assemble_ENDIF(r700_AssemblerBase *pAsm)
 {
+#ifdef USE_CF_FOR_POP_AFTER
+    pops(pAsm, 1); 
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    if(NULL == pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        /* no else in between */
+        pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode;
+    }
+    else
+    {
+        pAsm->fc_stack[pAsm->FCSP].mid[0]->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode;
+    }
+
+    if(NULL != pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        FREE(pAsm->fc_stack[pAsm->FCSP].mid);
+    }
+
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_IF)
+    {
+        radeon_error("if/endif in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+    
+    pAsm->FCSP--;
+
+    decreaseCurrent(pAsm, FC_PUSH_VPM);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_BGNLOOP(r700_AssemblerBase *pAsm)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_START_NO_AL;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->FCSP++;
+	pAsm->fc_stack[pAsm->FCSP].type  = FC_LOOP;
+    pAsm->fc_stack[pAsm->FCSP].mid   = NULL;
+    pAsm->fc_stack[pAsm->FCSP].unNumMid = 0;
+    pAsm->fc_stack[pAsm->FCSP].midLen   = 0;
+    pAsm->fc_stack[pAsm->FCSP].first    = pAsm->cf_current_cf_clause_ptr;
+
+    checkStackDepth(pAsm, FC_LOOP, GL_FALSE);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_BRK(r700_AssemblerBase *pAsm)
+{
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+    unsigned int unFCSP;
+    for(unFCSP=pAsm->FCSP; unFCSP>0; unFCSP--)
+    {
+        if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+        {
+            break;
+        }
+    }
+    if(0 == FC_LOOP)
+    {
+        radeon_error("Break is not inside loop/endloop pair.\n");
+        return GL_FALSE;
+    }
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_BREAK;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+#endif //USE_CF_FOR_CONTINUE_BREAK
+    return GL_TRUE;
+}
+
+GLboolean assemble_CONT(r700_AssemblerBase *pAsm)
+{
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+    unsigned int unFCSP;
+    for(unFCSP=pAsm->FCSP; unFCSP>0; unFCSP--)
+    {
+        if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+        {
+            break;
+        }
+    }
+    if(0 == FC_LOOP)
+    {
+        radeon_error("Continue is not inside loop/endloop pair.\n");
+        return GL_FALSE;
+    }
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_CONTINUE;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+#endif /* USE_CF_FOR_CONTINUE_BREAK */
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_ENDLOOP(r700_AssemblerBase *pAsm)
+{
+    GLuint i;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_END;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr   = pAsm->fc_stack[pAsm->FCSP].first->m_uIndex + 1;
+    pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+    for(i=0; i<pAsm->fc_stack[pAsm->FCSP].unNumMid; i++)
+    {
+        pAsm->fc_stack[pAsm->FCSP].mid[i]->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex;
+    }
+    if(NULL != pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        FREE(pAsm->fc_stack[pAsm->FCSP].mid);
+    }
+#endif
+
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_LOOP)
+    {
+        radeon_error("loop/endloop in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+
+    GLuint unFCSP;
+    GLuint unIF = 0;
+    if((pAsm->unCFflags & HAS_CURRENT_LOOPRET) > 0)
+    {        
+        for(unFCSP=(pAsm->FCSP-1); unFCSP>pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry; unFCSP--)
+        {
+            if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+            {
+                breakLoopOnFlag(pAsm, unFCSP);
+                break;
+            }
+            else if(FC_IF == pAsm->fc_stack[unFCSP].type)
+            {
+                unIF++;
+            }
+        }
+        if(unFCSP <= pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry)
+        {            
+#ifdef USE_CF_FOR_POP_AFTER
+            returnOnFlag(pAsm, unIF); 
+#else
+            returnOnFlag(pAsm, 0);
+#endif /* USE_CF_FOR_POP_AFTER */
+            pAsm->unCFflags &= ~HAS_CURRENT_LOOPRET;
+        }
+    }
+
+    pAsm->FCSP--;
+
+    decreaseCurrent(pAsm, FC_LOOP);
+    
+    return GL_TRUE;
+}
+
+void add_return_inst(r700_AssemblerBase *pAsm)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+    //pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_RETURN;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+}
+
+GLboolean assemble_BGNSUB(r700_AssemblerBase *pAsm, GLint nILindex)
+{
+    /* Put in sub */
+    if( (pAsm->unSubArrayPointer + 1) > pAsm->unSubArraySize )
+    {
+        pAsm->subs = (SUB_OFFSET*)_mesa_realloc( (void *)pAsm->subs,
+                                  sizeof(SUB_OFFSET) * pAsm->unSubArraySize,
+                                  sizeof(SUB_OFFSET) * (pAsm->unSubArraySize + 10) );
+        if(NULL == pAsm->subs)
+        {
+            return GL_FALSE;
+        }
+        pAsm->unSubArraySize += 10;
+    }
+
+    pAsm->subs[pAsm->unSubArrayPointer].subIL_Offset = nILindex;
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.pHead=NULL;  
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.pTail=NULL;  
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.uNumOfNode=0;
+
+    pAsm->CALLSP++;
+    pAsm->CALLSTACK[pAsm->CALLSP].subDescIndex = pAsm->unSubArrayPointer;
+    pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry = pAsm->FCSP;
+    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local
+                   = &(pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local);
+    pAsm->CALLSTACK[pAsm->CALLSP].max = 0;
+    pAsm->CALLSTACK[pAsm->CALLSP].current = 0;
+    SetActiveCFlist(pAsm->pR700Shader, 
+                    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local);
+
+    pAsm->unSubArrayPointer++;
+
+    /* start sub */
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    pAsm->FCSP++;
+    pAsm->fc_stack[pAsm->FCSP].type  = FC_REP;
+
+    checkStackDepth(pAsm, FC_REP, GL_FALSE);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_ENDSUB(r700_AssemblerBase *pAsm)
+{
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_REP)
+    {
+        radeon_error("BGNSUB/ENDSUB in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+
+    /* copy max to sub structure */
+    pAsm->subs[pAsm->CALLSTACK[pAsm->CALLSP].subDescIndex].unStackDepthMax
+        = pAsm->CALLSTACK[pAsm->CALLSP].max;
+
+    decreaseCurrent(pAsm, FC_REP);
+
+    pAsm->CALLSP--;
+    SetActiveCFlist(pAsm->pR700Shader, 
+                    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local);
+    
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    pAsm->FCSP--;
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_RET(r700_AssemblerBase *pAsm)
+{
+    GLuint unIF = 0;
+
+    if(pAsm->CALLSP > 0)
+    {   /* in sub */
+        GLuint unFCSP;        
+        for(unFCSP=pAsm->FCSP; unFCSP>pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry; unFCSP--)
+        {
+            if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+            {
+                setRetInLoopFlag(pAsm, SQ_SEL_1);
+                breakLoopOnFlag(pAsm, unFCSP);
+                pAsm->unCFflags |= LOOPRET_FLAGS;
+
+                return GL_TRUE;
+            }
+            else if(FC_IF == pAsm->fc_stack[unFCSP].type)
+            {
+                unIF++;
+            }
+        }
+    }
+
+#ifdef USE_CF_FOR_POP_AFTER    
+    if(unIF > 0)
+    {
+        pops(pAsm, unIF);
+    }
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    add_return_inst(pAsm);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_CAL(r700_AssemblerBase *pAsm, 
+                       GLint nILindex,
+                       GLuint uiNumberInsts,
+                       struct prog_instruction *pILInst)
+{
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.call_count       = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_CALL;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    /* Put in caller */
+    if( (pAsm->unCallerArrayPointer + 1) > pAsm->unCallerArraySize )
+    {
+        pAsm->callers = (CALLER_POINTER*)_mesa_realloc( (void *)pAsm->callers, 
+                       sizeof(CALLER_POINTER) * pAsm->unCallerArraySize, 
+                       sizeof(CALLER_POINTER) * (pAsm->unCallerArraySize + 10) );
+        if(NULL == pAsm->callers)
+        {
+            return GL_FALSE;
+        }
+        pAsm->unCallerArraySize += 10;
+    }
+    
+    pAsm->callers[pAsm->unCallerArrayPointer].subIL_Offset = nILindex;
+    pAsm->callers[pAsm->unCallerArrayPointer].cf_ptr       = pAsm->cf_current_cf_clause_ptr; 
+
+    pAsm->unCallerArrayPointer++;
+
+    int j;
+    GLuint max;
+    GLuint unSubID;
+    GLboolean bRet;
+    for(j=0; j<pAsm->unSubArrayPointer; j++)
+    {
+        if(nILindex == pAsm->subs[j].subIL_Offset)
+        {   /* compiled before */
+
+            max = pAsm->subs[j].unStackDepthMax 
+                + pAsm->CALLSTACK[pAsm->CALLSP].current;
+            if(max > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max = max;
+            }
+            
+            pAsm->callers[pAsm->unCallerArrayPointer - 1].subDescIndex = j; 
+            return GL_TRUE;
+        }
+    }
+
+    pAsm->callers[pAsm->unCallerArrayPointer - 1].subDescIndex = pAsm->unSubArrayPointer;
+    unSubID = pAsm->unSubArrayPointer;
+
+    bRet = AssembleInstr(nILindex, uiNumberInsts, pILInst, pAsm);
+
+    if(GL_TRUE == bRet)
+    {
+        max = pAsm->subs[unSubID].unStackDepthMax 
+            + pAsm->CALLSTACK[pAsm->CALLSP].current;
+        if(max > pAsm->CALLSTACK[pAsm->CALLSP].max)
+        {
+            pAsm->CALLSTACK[pAsm->CALLSP].max = max;
+        }
+    }
+
+    return bRet;
+}
+
+GLboolean setRetInLoopFlag(r700_AssemblerBase *pAsm, GLuint flagValue)
+{
+    GLfloat fLiteral[2] = {0.1, 0.0};
+
+    pAsm->D.dst.opcode   = SQ_OP2_INST_MOV;
+    pAsm->D.dst.op3      = 0;
+    pAsm->D.dst.rtype    = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg      = pAsm->flag_reg_index;
+    pAsm->D.dst.writex   = 1;
+    pAsm->D.dst.writey   = 0;
+    pAsm->D.dst.writez   = 0;
+    pAsm->D.dst.writew   = 0;
+    pAsm->D2.dst2.literal      = 1;
+    pAsm->D2.dst2.SaturateMode = SATURATE_OFF;
+    pAsm->D.dst.predicated     = 0;
+    /* in reloc where dislink flag init inst, only one slot alu inst is handled. */
+    pAsm->D.dst.math           = 1; /* TODO : not math really, but one channel op, more generic alu assembler needed */
+#if 0
+    pAsm->S[0].src.rtype = SRC_REC_LITERAL;
+    //pAsm->S[0].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = SQ_SEL_X;
+    pAsm->S[0].src.swizzley = SQ_SEL_Y;
+    pAsm->S[0].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[0].src.swizzlew = SQ_SEL_W;
+
+    if( GL_FALSE == next_ins_literal(pAsm, &(fLiteral[0])) )
+    {
+        return GL_FALSE;
+    }
+#else
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = flagValue;
+    pAsm->S[0].src.swizzley = flagValue;
+    pAsm->S[0].src.swizzlez = flagValue;
+    pAsm->S[0].src.swizzlew = flagValue;
+
+    if( GL_FALSE == next_ins2(pAsm) )
+    {
+        return GL_FALSE;
+    }
+#endif
+
     return GL_TRUE;
 }
 
-GLboolean AssembleInstr(GLuint uiNumberInsts,
+GLboolean testFlag(r700_AssemblerBase *pAsm)
+{
+    GLfloat fLiteral[2] = {0.1, 0.0};
+
+    //Test flag
+    GLuint tmp = gethelpr(pAsm);
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+
+    pAsm->D.dst.opcode   = SQ_OP2_INST_PRED_SETE;
+    pAsm->D.dst.math     = 1;
+    pAsm->D.dst.rtype    = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg      = tmp;
+    pAsm->D.dst.writex   = 1;
+    pAsm->D.dst.writey   = 0;
+    pAsm->D.dst.writez   = 0;
+    pAsm->D.dst.writew   = 0;
+    pAsm->D2.dst2.literal      = 1;
+    pAsm->D2.dst2.SaturateMode = SATURATE_OFF;
+    pAsm->D.dst.predicated     = 1;
+
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = pAsm->flag_reg_index;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = SQ_SEL_X;
+    pAsm->S[0].src.swizzley = SQ_SEL_Y;
+    pAsm->S[0].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[0].src.swizzlew = SQ_SEL_W;
+#if 0
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    //pAsm->S[1].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[1].src));
+    pAsm->S[1].src.swizzlex = SQ_SEL_X;
+    pAsm->S[1].src.swizzley = SQ_SEL_Y;
+    pAsm->S[1].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[1].src.swizzlew = SQ_SEL_W;
+
+    if( GL_FALSE == next_ins_literal(pAsm, &(fLiteral[0])) )
+    {
+        return GL_FALSE;
+    }
+#else
+    pAsm->S[1].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[1].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[1].src));
+    pAsm->S[1].src.swizzlex = SQ_SEL_1;
+    pAsm->S[1].src.swizzley = SQ_SEL_1;
+    pAsm->S[1].src.swizzlez = SQ_SEL_1;
+    pAsm->S[1].src.swizzlew = SQ_SEL_1;
+
+    if( GL_FALSE == next_ins2(pAsm) )
+    {
+        return GL_FALSE;
+    }
+#endif
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+    return GL_TRUE;
+}
+
+GLboolean returnOnFlag(r700_AssemblerBase *pAsm, GLuint unIF)
+{
+    testFlag(pAsm);
+    jumpToOffest(pAsm, 1, 4);
+    setRetInLoopFlag(pAsm, SQ_SEL_0);
+    pops(pAsm, unIF + 1);
+    add_return_inst(pAsm);
+
+    return GL_TRUE;
+}
+
+GLboolean breakLoopOnFlag(r700_AssemblerBase *pAsm, GLuint unFCSP)
+{
+    testFlag(pAsm);
+ 
+    //break
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_BREAK;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    pops(pAsm, 1);
+               
+    return GL_TRUE;
+}
+
+GLboolean AssembleInstr(GLuint uiFirstInst,
+                        GLuint uiNumberInsts,
                         struct prog_instruction *pILInst, 
 						r700_AssemblerBase *pR700AsmCode)
 {
     GLuint i;
 
     pR700AsmCode->pILInst = pILInst;
-	for(i=0; i<uiNumberInsts; i++)
+	for(i=uiFirstInst; i<uiNumberInsts; i++)
     {
         pR700AsmCode->uiCurInst = i;
 
+#ifndef USE_CF_FOR_CONTINUE_BREAK
+        if(OPCODE_BRK == pILInst[i+1].Opcode)
+        {
+            switch(pILInst[i].Opcode)            
+            {
+            case OPCODE_SLE:
+                pILInst[i].Opcode = OPCODE_SGT;
+                break;
+            case OPCODE_SLT:
+                pILInst[i].Opcode = OPCODE_SGE;
+                break;
+            case OPCODE_SGE:
+                pILInst[i].Opcode = OPCODE_SLT;
+                break;
+            case OPCODE_SGT:
+                pILInst[i].Opcode = OPCODE_SLE;
+                break;
+            case OPCODE_SEQ:
+                pILInst[i].Opcode = OPCODE_SNE;
+                break;
+            case OPCODE_SNE:
+                pILInst[i].Opcode = OPCODE_SEQ;
+                break;
+            default:
+                break;
+            }
+        }
+#endif
+
         switch (pILInst[i].Opcode)
         {
         case OPCODE_ABS: 
@@ -4350,7 +5902,8 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
             if ( GL_FALSE == assemble_FLR(pR700AsmCode) ) 
                 return GL_FALSE;
             break;  
-        //case OP_FLR_INT: 
+        //case OP_FLR_INT: ;
+
         //    if ( GL_FALSE == assemble_FLR_INT() ) 
         //        return GL_FALSE;
         //    break;  
@@ -4361,8 +5914,10 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
             break;  
 
         case OPCODE_KIL: 
-            if ( GL_FALSE == assemble_KIL(pR700AsmCode) ) 
-                return GL_FALSE;
+        case OPCODE_KIL_NV: 
+            /* done at OPCODE_SE/SGT...etc. */
+            /* if ( GL_FALSE == assemble_KIL(pR700AsmCode) ) 
+                return GL_FALSE; */
             break;
         case OPCODE_LG2: 
             if ( GL_FALSE == assemble_LG2(pR700AsmCode) ) 
@@ -4422,16 +5977,340 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
         case OPCODE_SCS: 
             if ( GL_FALSE == assemble_SCS(pR700AsmCode) ) 
                 return GL_FALSE;
-            break;  
+            break; 
+            
+        case OPCODE_SEQ:
+            if(OPCODE_IF == pILInst[i+1].Opcode)
+            {
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_BRK == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_CONT == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif                
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+            {
+                if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else
+            {
+                if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            break;
+
+        case OPCODE_SGT: 
+            if(OPCODE_IF == pILInst[i+1].Opcode)
+            {
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_BRK == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_CONT == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif
+
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+            {
+                if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLGT) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else
+            {
+                if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGT) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            break;
 
         case OPCODE_SGE: 
-            if ( GL_FALSE == assemble_SGE(pR700AsmCode) ) 
-                return GL_FALSE;
-            break; 
+            if(OPCODE_IF == pILInst[i+1].Opcode)
+            {
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_BRK == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_CONT == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif
+
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+            {
+                if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLGE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else
+            {
+                if ( GL_FALSE == assemble_SGE(pR700AsmCode) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            break;
+        
+        /* NO LT, LE, TODO : use GE => LE, GT => LT : reverse 2 src order would be simpliest. Or use SQ_CF_COND_FALSE for SQ_CF_COND_ACTIVE.*/
         case OPCODE_SLT: 
-            if ( GL_FALSE == assemble_SLT(pR700AsmCode) ) 
-                return GL_FALSE;
-            break; 
+            {
+                struct prog_src_register SrcRegSave[2];
+                SrcRegSave[0] = pILInst[i].SrcReg[0];
+                SrcRegSave[1] = pILInst[i].SrcReg[1];
+                pILInst[i].SrcReg[0] = SrcRegSave[1];
+                pILInst[i].SrcReg[1] = SrcRegSave[0];
+                if(OPCODE_IF == pILInst[i+1].Opcode)
+                {
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if(OPCODE_BRK == pILInst[i+1].Opcode)
+                {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if(OPCODE_CONT == pILInst[i+1].Opcode)
+                {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif
+
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGT) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+                {
+                    if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLGT) ) 
+                    {
+                        return GL_FALSE;
+                    }
+                }
+                else
+                {
+                    if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGT) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                } 
+                pILInst[i].SrcReg[0] = SrcRegSave[0];
+                pILInst[i].SrcReg[1] = SrcRegSave[1];
+            }
+            break;
+
+        case OPCODE_SLE: 
+            {
+                struct prog_src_register SrcRegSave[2];
+                SrcRegSave[0] = pILInst[i].SrcReg[0];
+                SrcRegSave[1] = pILInst[i].SrcReg[1];
+                pILInst[i].SrcReg[0] = SrcRegSave[1];
+                pILInst[i].SrcReg[1] = SrcRegSave[0];
+                if(OPCODE_IF == pILInst[i+1].Opcode)
+                {
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if(OPCODE_BRK == pILInst[i+1].Opcode)
+                {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if(OPCODE_CONT == pILInst[i+1].Opcode)
+                {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                    pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif
+
+                    if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETGE) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+                {
+                    if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLGE) ) 
+                    {
+                        return GL_FALSE;
+                    }
+                }
+                else
+                {
+                    if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGE) ) 
+                    {
+                        pILInst[i].SrcReg[0] = SrcRegSave[0];
+                        pILInst[i].SrcReg[1] = SrcRegSave[1];
+                        return GL_FALSE;
+                    }
+                }
+                pILInst[i].SrcReg[0] = SrcRegSave[0];
+                pILInst[i].SrcReg[1] = SrcRegSave[1];
+            }
+            break;
+
+        case OPCODE_SNE: 
+            if(OPCODE_IF == pILInst[i+1].Opcode)
+            {
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETNE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_BRK == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_BREAK;
+#endif
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETNE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if(OPCODE_CONT == pILInst[i+1].Opcode)
+            {
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+#else
+                pR700AsmCode->alu_x_opcode = SQ_CF_INST_ALU_CONTINUE;
+#endif
+                if ( GL_FALSE == assemble_LOGIC_PRED(pR700AsmCode, SQ_OP2_INST_PRED_SETNE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else if((OPCODE_KIL == pILInst[i+1].Opcode)||(OPCODE_KIL_NV == pILInst[i+1].Opcode))
+            {
+                if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLNE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            else
+            {
+                if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETNE) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
+            break;
 
         //case OP_STP: 
         //    if ( GL_FALSE == assemble_STP(pR700AsmCode) ) 
@@ -4465,30 +6344,102 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;
 
+        case OPCODE_TRUNC:
+            if ( GL_FALSE == assemble_math_function(pR700AsmCode, SQ_OP2_INST_TRUNC) )
+                return GL_FALSE;
+            break;
+
         case OPCODE_XPD: 
             if ( GL_FALSE == assemble_XPD(pR700AsmCode) ) 
                 return GL_FALSE;
             break;  
 
         case OPCODE_IF   : 
-            if ( GL_FALSE == assemble_IF(pR700AsmCode) ) 
-                return GL_FALSE;
+            {                
+                GLboolean bHasElse = GL_FALSE;
+
+                if(pILInst[pILInst[i].BranchTarget - 1].Opcode == OPCODE_ELSE)
+                {
+                    bHasElse = GL_TRUE;
+                }
+
+                if ( GL_FALSE == assemble_IF(pR700AsmCode, bHasElse) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
             break;
+
         case OPCODE_ELSE : 
-            radeon_error("Not yet implemented instruction OPCODE_ELSE \n");
-            //if ( GL_FALSE == assemble_BAD("ELSE") ) 
+            if ( GL_FALSE == assemble_ELSE(pR700AsmCode) ) 
                 return GL_FALSE;
             break;
+
         case OPCODE_ENDIF: 
             if ( GL_FALSE == assemble_ENDIF(pR700AsmCode) ) 
                 return GL_FALSE;
             break;
 
+        case OPCODE_BGNLOOP:
+            if( GL_FALSE == assemble_BGNLOOP(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_BRK:
+            if( GL_FALSE == assemble_BRK(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_CONT:
+            if( GL_FALSE == assemble_CONT(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_ENDLOOP:
+            if( GL_FALSE == assemble_ENDLOOP(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_BGNSUB:
+            if( GL_FALSE == assemble_BGNSUB(pR700AsmCode, i) )
+            {
+                return GL_FALSE;
+            }
+            break;
+        
+        case OPCODE_RET:
+            if( GL_FALSE == assemble_RET(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+        
+        case OPCODE_CAL:
+            if( GL_FALSE == assemble_CAL(pR700AsmCode, 
+                                         pILInst[i].BranchTarget,                                         
+                                         uiNumberInsts,
+                                         pILInst) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
         //case OPCODE_EXPORT: 
         //    if ( GL_FALSE == assemble_EXPORT() ) 
         //        return GL_FALSE;
         //    break;
 
+        case OPCODE_ENDSUB:
+            return assemble_ENDSUB(pR700AsmCode);
+
         case OPCODE_END: 
 			//pR700AsmCode->uiCurInst = i;
 			//This is to remaind that if in later exoort there is depth/stencil
@@ -4505,6 +6456,123 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
     return GL_TRUE;
 }
 
+GLboolean InitShaderProgram(r700_AssemblerBase * pAsm)
+{
+    setRetInLoopFlag(pAsm, SQ_SEL_0);
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+    return GL_TRUE;
+}
+
+GLboolean RelocProgram(r700_AssemblerBase * pAsm)
+{
+    GLuint i;
+    GLuint unCFoffset;
+    TypedShaderList * plstCFmain;
+    TypedShaderList * plstCFsub;
+
+    R700ShaderInstruction *        pInst;
+    R700ControlFlowGenericClause * pCFInst;
+
+    plstCFmain = pAsm->CALLSTACK[0].plstCFInstructions_local;
+
+    /* remove flags init if they are not used */
+    if((pAsm->unCFflags & HAS_LOOPRET) == 0)
+    {
+        R700ControlFlowALUClause * pCF_ALU;
+        pInst = plstCFmain->pHead;
+        while(pInst)
+        {
+            if(SIT_CF_ALU == pInst->m_ShaderInstType)
+            {
+                pCF_ALU = (R700ControlFlowALUClause *)pInst;
+                if(0 == pCF_ALU->m_Word1.f.count)
+                {
+                    pCF_ALU->m_Word1.f.cf_inst = SQ_CF_INST_NOP;
+                }
+                else
+                {
+                    R700ALUInstruction * pALU = pCF_ALU->m_pLinkedALUInstruction;
+                    
+                    pALU->m_pLinkedALUClause = NULL;
+                    pALU = (R700ALUInstruction *)(pALU->pNextInst);
+                    pALU->m_pLinkedALUClause = pCF_ALU;
+                    pCF_ALU->m_pLinkedALUInstruction = pALU;
+
+                    pCF_ALU->m_Word1.f.count--;
+                }
+                break;
+            }
+            pInst = pInst->pNextInst;
+        };
+    }
+
+    if(pAsm->CALLSTACK[0].max > 0)
+    {
+        pAsm->pR700Shader->uStackSize = ((pAsm->CALLSTACK[0].max + 3)>>2) + 2;
+    }
+
+    if(0 == pAsm->unSubArrayPointer)
+    {
+        return GL_TRUE;
+    }
+
+    unCFoffset = plstCFmain->uNumOfNode;
+
+    /* Reloc subs */
+    for(i=0; i<pAsm->unSubArrayPointer; i++)
+    {
+        pAsm->subs[i].unCFoffset = unCFoffset;
+        plstCFsub = &(pAsm->subs[i].lstCFInstructions_local);
+
+        pInst = plstCFsub->pHead;
+
+        /* reloc instructions */
+        while(pInst)
+        {
+            if(SIT_CF_GENERIC == pInst->m_ShaderInstType)
+            {
+                pCFInst = (R700ControlFlowGenericClause *)pInst;
+
+                switch (pCFInst->m_Word1.f.cf_inst)
+                {
+                case SQ_CF_INST_POP:
+                case SQ_CF_INST_JUMP:
+                case SQ_CF_INST_ELSE:
+                case SQ_CF_INST_LOOP_END:
+                case SQ_CF_INST_LOOP_START:
+                case SQ_CF_INST_LOOP_START_NO_AL:
+                case SQ_CF_INST_LOOP_CONTINUE:
+                case SQ_CF_INST_LOOP_BREAK:
+                    pCFInst->m_Word0.f.addr += unCFoffset;
+                    break;
+                default:
+                    break;
+                }
+            }  
+            
+            pInst->m_uIndex += unCFoffset;
+
+            pInst = pInst->pNextInst;
+        };
+
+        /* Put sub into main */
+        plstCFmain->pTail->pNextInst = plstCFsub->pHead;
+        plstCFmain->pTail            = plstCFsub->pTail;
+        plstCFmain->uNumOfNode      += plstCFsub->uNumOfNode;
+
+        unCFoffset += plstCFsub->uNumOfNode;
+    }
+
+    /* reloc callers */
+    for(i=0; i<pAsm->unCallerArrayPointer; i++)
+    {
+        pAsm->callers[i].cf_ptr->m_Word0.f.addr
+            = pAsm->subs[pAsm->callers[i].subDescIndex].unCFoffset; 
+    }
+
+    return GL_TRUE;
+}
+
 GLboolean Process_Export(r700_AssemblerBase* pAsm,
                          GLuint type,
                          GLuint export_starting_index,
@@ -4800,6 +6868,25 @@ GLboolean Process_Vertex_Exports(r700_AssemblerBase *pR700AsmCode,
 		}
 	}
 
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+        unBit = 1 << i;
+        if(OutputsWritten & unBit)
+		{
+            if( GL_FALSE == Process_Export(pR700AsmCode,
+                                          SQ_EXPORT_PARAM, 
+                                          export_starting_index, 
+                                          1, 
+                                          pR700AsmCode->ucVP_OutputMap[i],
+                                          GL_FALSE) )
+            {
+                return GL_FALSE;
+            }
+
+            export_starting_index++;
+		}
+    }
+
     // At least one param should be exported
     if (export_count) 
     {
@@ -4833,6 +6920,16 @@ GLboolean Clean_Up_Assembler(r700_AssemblerBase *pR700AsmCode)
 {
     FREE(pR700AsmCode->pucOutMask);
     FREE(pR700AsmCode->pInstDeps);
+
+    if(NULL != pR700AsmCode->subs)
+    {
+        FREE(pR700AsmCode->subs);
+    }
+    if(NULL != pR700AsmCode->callers)
+    {
+        FREE(pR700AsmCode->callers);
+    }
+
     return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h
index c66db502a1..130fc89dae 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.h
+++ b/src/mesa/drivers/dri/r600/r700_assembler.h
@@ -72,7 +72,8 @@ typedef enum SrcRegisterType
     SRC_REG_INPUT          = 1,
     SRC_REG_CONSTANT       = 2,
     SRC_REG_ALT_TEMPORARY  = 3,
-    NUMBER_OF_SRC_REG_TYPE = 4
+    SRC_REC_LITERAL        = 4, 
+    NUMBER_OF_SRC_REG_TYPE = 5
 } SrcRegisterType;
 
 typedef enum DstRegisterType 
@@ -111,6 +112,12 @@ typedef struct PVSDSTtag
 	BITS addrmode1:1; //32
 } PVSDST;
 
+typedef struct PVSINSTtag
+{
+    BITS literal      :2; 
+    BITS SaturateMode :2; 
+} PVSINST;
+
 typedef struct PVSSRCtag 
 {
 	BITS rtype:4;            
@@ -148,6 +155,7 @@ typedef union PVSDWORDtag
 {
 	BITS    bits;
 	PVSDST  dst;
+    PVSINST dst2;
 	PVSSRC  src;
 	PVSMATH math;
 	float   f;
@@ -251,6 +259,8 @@ enum
     FC_IF = 1,
     FC_LOOP = 2,
     FC_REP = 3,
+    FC_PUSH_VPM = 4,
+    FC_PUSH_WQM = 5,
 
     COND_NONE = 0,
     COND_BOOL = 1,
@@ -263,22 +273,52 @@ enum
 
 typedef struct FC_LEVEL 
 {
-	unsigned int           first; ///< first fc instruction on level (if, rep, loop)
-	unsigned int*          mid; ///< middle instructions - else or all breaks on this level
-	unsigned int           midLen;
-	unsigned int           type;
-	unsigned int           cond;
-	unsigned int           inv;
-	unsigned int           bpush; ///< 1 if first instruction does branch stack push
-			 int           id; ///< id of bool or int variable
+    R700ControlFlowGenericClause *  first;
+    R700ControlFlowGenericClause ** mid;
+    unsigned int unNumMid;
+    unsigned int midLen;
+    unsigned int type;
+    unsigned int cond;
+    unsigned int inv;
+    int id; ///< id of bool or int variable
 } FC_LEVEL;
 
 typedef struct VTX_FETCH_METHOD 
 {
-	GLboolean bEnableMini;
-	GLuint mega_fetch_remainder;
+    GLboolean bEnableMini;
+    GLuint mega_fetch_remainder;
 } VTX_FETCH_METHOD;
 
+typedef struct SUB_OFFSET
+{
+    GLint  subIL_Offset;
+    GLuint unCFoffset;
+    GLuint unStackDepthMax;
+    TypedShaderList lstCFInstructions_local;
+} SUB_OFFSET;
+
+typedef struct CALLER_POINTER
+{
+    GLint  subIL_Offset;
+    GLint  subDescIndex;
+    R700ControlFlowGenericClause* cf_ptr;
+} CALLER_POINTER;
+
+#define SQ_MAX_CALL_DEPTH 0x00000020
+
+typedef struct CALL_LEVEL
+{
+    unsigned int      FCSP_BeforeEntry;
+    GLint             subDescIndex;
+    GLushort          current;
+    GLushort          max;
+    TypedShaderList * plstCFInstructions_local;
+} CALL_LEVEL;
+
+#define HAS_CURRENT_LOOPRET 0x1L
+#define HAS_LOOPRET         0x2L
+#define LOOPRET_FLAGS       HAS_LOOPRET | HAS_CURRENT_LOOPRET
+
 typedef struct r700_AssemblerBase 
 {
 	R700ControlFlowSXClause*      cf_last_export_ptr;
@@ -294,11 +334,14 @@ typedef struct r700_AssemblerBase
 	// No clause has been created yet
 	CF_CLAUSE_TYPE cf_current_clause_type;
 
+    BITS alu_x_opcode;
+
 	GLuint number_of_exports;
 	GLuint number_of_colorandz_exports;
 	GLuint number_of_export_opcodes;
 
 	PVSDWORD D;
+    PVSDWORD D2;
 	PVSDWORD S[3];
 
 	unsigned int uLastPosUpdate;
@@ -310,6 +353,8 @@ typedef struct r700_AssemblerBase
 	unsigned int number_used_registers;
 	unsigned int uUsedConsts; 
 
+    unsigned int flag_reg_index;
+
 	// Fragment programs
 	unsigned int uiFP_AttributeMap[FRAG_ATTRIB_MAX];
 	unsigned int uiFP_OutputMap[FRAG_RESULT_MAX];
@@ -330,9 +375,6 @@ typedef struct r700_AssemblerBase
 	unsigned int FCSP;
 	FC_LEVEL fc_stack[32];
 
-	unsigned int branch_depth;
-	unsigned int max_branch_depth;
-
 	//-----------------------------------------------------------------------------------
 	// ArgSubst used in Assemble_Source() function
 	//-----------------------------------------------------------------------------------
@@ -378,6 +420,18 @@ typedef struct r700_AssemblerBase
     GLboolean is_tex;
     /* we inserted helper intructions and need barrier on next TEX ins */ 
     GLboolean need_tex_barrier; 
+
+    SUB_OFFSET     * subs;
+    GLuint           unSubArraySize;
+    GLuint           unSubArrayPointer;
+    CALLER_POINTER * callers;
+    GLuint           unCallerArraySize;
+    GLuint           unCallerArrayPointer;
+    unsigned int     CALLSP;
+    CALL_LEVEL       CALLSTACK[SQ_MAX_CALL_DEPTH];
+
+    GLuint unCFflags;
+
 } r700_AssemblerBase;
 
 //Internal use
@@ -446,6 +500,10 @@ GLboolean assemble_alu_src(R700ALUInstruction*  alu_instruction_ptr,
 GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
                               R700ALUInstruction* alu_instruction_ptr,
                               GLuint              contiguous_slots_needed);
+
+GLboolean add_cf_instruction(r700_AssemblerBase* pAsm);
+void add_return_inst(r700_AssemblerBase *pAsm);
+
 void get_src_properties(R700ALUInstruction*  alu_instruction_ptr,
                         int                  source_index,
                         BITS*                psrc_sel,
@@ -467,6 +525,21 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
                        R700ALUInstruction* alu_instruction_ptr);
 GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm);
 GLboolean next_ins(r700_AssemblerBase *pAsm);
+
+GLboolean next_ins2(r700_AssemblerBase *pAsm);
+GLboolean assemble_alu_instruction2(r700_AssemblerBase *pAsm);
+
+/* TODO : merge next_ins/2/literal, assemble_alu_instruction/2/literal */
+GLboolean next_ins_literal(r700_AssemblerBase *pAsm, GLfloat * pLiteral);
+GLboolean assemble_alu_instruction_literal(r700_AssemblerBase *pAsm, GLfloat * pLiteral);
+
+GLboolean pops(r700_AssemblerBase *pAsm, GLuint pops);
+GLboolean jumpToOffest(r700_AssemblerBase *pAsm, GLuint pops, GLint offset);
+GLboolean setRetInLoopFlag(r700_AssemblerBase *pAsm, GLuint flagValue);
+GLboolean testFlag(r700_AssemblerBase *pAsm);
+GLboolean breakLoopOnFlag(r700_AssemblerBase *pAsm, GLuint unFCSP);
+GLboolean returnOnFlag(r700_AssemblerBase *pAsm, GLuint unIF);
+
 GLboolean assemble_math_function(r700_AssemblerBase* pAsm, BITS opcode);
 GLboolean assemble_ABS(r700_AssemblerBase *pAsm);
 GLboolean assemble_ADD(r700_AssemblerBase *pAsm);
@@ -481,7 +554,7 @@ GLboolean assemble_EXP(r700_AssemblerBase *pAsm);
 GLboolean assemble_FLR(r700_AssemblerBase *pAsm);
 GLboolean assemble_FLR_INT(r700_AssemblerBase *pAsm);
 GLboolean assemble_FRC(r700_AssemblerBase *pAsm);
-GLboolean assemble_KIL(r700_AssemblerBase *pAsm);
+GLboolean assemble_KIL(r700_AssemblerBase *pAsm, GLuint opcode);
 GLboolean assemble_LG2(r700_AssemblerBase *pAsm);
 GLboolean assemble_LRP(r700_AssemblerBase *pAsm);
 GLboolean assemble_LOG(r700_AssemblerBase *pAsm);
@@ -497,14 +570,32 @@ GLboolean assemble_RSQ(r700_AssemblerBase *pAsm);
 GLboolean assemble_SIN(r700_AssemblerBase *pAsm);
 GLboolean assemble_SCS(r700_AssemblerBase *pAsm);
 GLboolean assemble_SGE(r700_AssemblerBase *pAsm);
+
+GLboolean assemble_LOGIC(r700_AssemblerBase *pAsm, BITS opcode);
+GLboolean assemble_LOGIC_PRED(r700_AssemblerBase *pAsm, BITS opcode); 
+
 GLboolean assemble_SLT(r700_AssemblerBase *pAsm);
 GLboolean assemble_STP(r700_AssemblerBase *pAsm);
 GLboolean assemble_TEX(r700_AssemblerBase *pAsm);
 GLboolean assemble_XPD(r700_AssemblerBase *pAsm);
 GLboolean assemble_EXPORT(r700_AssemblerBase *pAsm);
-GLboolean assemble_IF(r700_AssemblerBase *pAsm);
+GLboolean assemble_IF(r700_AssemblerBase *pAsm, GLboolean bHasElse);
+GLboolean assemble_ELSE(r700_AssemblerBase *pAsm);
 GLboolean assemble_ENDIF(r700_AssemblerBase *pAsm);
 
+GLboolean assemble_BGNLOOP(r700_AssemblerBase *pAsm);
+GLboolean assemble_BRK(r700_AssemblerBase *pAsm);
+GLboolean assemble_COND(r700_AssemblerBase *pAsm);
+GLboolean assemble_ENDLOOP(r700_AssemblerBase *pAsm);
+
+GLboolean assemble_BGNSUB(r700_AssemblerBase *pAsm, GLint nILindex);
+GLboolean assemble_ENDSUB(r700_AssemblerBase *pAsm);
+GLboolean assemble_RET(r700_AssemblerBase *pAsm);
+GLboolean assemble_CAL(r700_AssemblerBase *pAsm, 
+                       GLint nILindex,
+                       GLuint uiNumberInsts,
+                       struct prog_instruction *pILInst);
+
 GLboolean Process_Export(r700_AssemblerBase* pAsm,
                          GLuint type, 
                          GLuint export_starting_index,
@@ -516,12 +607,16 @@ GLboolean Move_Depth_Exports_To_Correct_Channels(r700_AssemblerBase *pAsm,
 
 
 //Interface
-GLboolean AssembleInstr(GLuint uiNumberInsts,
+GLboolean AssembleInstr(GLuint uiFirstInst,
+                        GLuint uiNumberInsts,
                         struct prog_instruction *pILInst, 
 						r700_AssemblerBase *pR700AsmCode);
 GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode, GLbitfield OutputsWritten);  
 GLboolean Process_Vertex_Exports(r700_AssemblerBase *pR700AsmCode, GLbitfield OutputsWritten);
 
+GLboolean RelocProgram(r700_AssemblerBase * pAsm);
+GLboolean InitShaderProgram(r700_AssemblerBase * pAsm);
+
 int       Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700_Shader* pShader);
 GLboolean Clean_Up_Assembler(r700_AssemblerBase *pR700AsmCode);
 
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index ec76fbcb6d..8126777bf4 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -54,11 +54,15 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 	for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
 		if (ctx->Texture.Unit[i]._ReallyEnabled) {
 			radeonTexObj *t = r700->textures[i];
+			uint32_t offset;
 			if (t) {
-				if (!t->image_override)
+				if (!t->image_override) {
 					bo = t->mt->bo;
-				else
+					offset = get_base_teximage_offset(t);
+				} else {
 					bo = t->bo;
+					offset = 0;
+				}
 				if (bo) {
 
 					r700SyncSurf(context, bo,
@@ -77,7 +81,7 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 					R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6);
 					R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2,
 							     bo,
-							     0,
+							     offset,
 							     RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
 					R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3,
 							     bo,
@@ -442,68 +446,77 @@ static void r700SendRenderTargetState(GLcontext *ctx, struct radeon_state_atom *
 
 static void r700SendPSState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
-	context_t *context = R700_CONTEXT(ctx);
-	R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
-	struct radeon_bo * pbo;
-	BATCH_LOCALS(&context->radeon);
-	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+    context_t *context = R700_CONTEXT(ctx);
+    R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+    struct radeon_bo * pbo;
+    BATCH_LOCALS(&context->radeon);
+    radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
-	pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
+    pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
 
-	if (!pbo)
-		return;
+    if (!pbo)
+	    return;
 
-	r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+    r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
 
-        BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
-	R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
-	R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
-	R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
-			     pbo,
-			     r700->ps.SQ_PGM_START_PS.u32All,
-			     RADEON_GEM_DOMAIN_GTT, 0, 0);
-	END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+    R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
+    R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
+    R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
+		         pbo,
+		         r700->ps.SQ_PGM_START_PS.u32All,
+		         RADEON_GEM_DOMAIN_GTT, 0, 0);
+    END_BATCH();
 
-        BEGIN_BATCH_NO_AUTOSTATE(9);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
-        END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(9);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
+    END_BATCH();
 
-	COMMIT_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    R600_OUT_BATCH_REGVAL(SQ_LOOP_CONST_0, 0x01000FFF);
+    END_BATCH();
+
+    COMMIT_BATCH();
 
 }
 
 static void r700SendVSState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
-	context_t *context = R700_CONTEXT(ctx);
-	R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
-	struct radeon_bo * pbo;
-	BATCH_LOCALS(&context->radeon);
-	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+    context_t *context = R700_CONTEXT(ctx);
+    R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+    struct radeon_bo * pbo;
+    BATCH_LOCALS(&context->radeon);
+    radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
-	pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
+    pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
 
-	if (!pbo)
-		return;
+    if (!pbo)
+	    return;
 
-	r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+    r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
 
-        BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
-	R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
-	R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
-	R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
-			     pbo,
-			     r700->vs.SQ_PGM_START_VS.u32All,
-			     RADEON_GEM_DOMAIN_GTT, 0, 0);
-	END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+    R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
+    R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
+    R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
+		         pbo,
+		         r700->vs.SQ_PGM_START_VS.u32All,
+		         RADEON_GEM_DOMAIN_GTT, 0, 0);
+    END_BATCH();
 
-        BEGIN_BATCH_NO_AUTOSTATE(6);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
-        END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(6);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
+    END_BATCH();
 
-	COMMIT_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + 32*4), 0x0100000F);
+    //R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + (SQ_LOOP_CONST_vs<2)), 0x0100000F);
+    END_BATCH();
+
+    COMMIT_BATCH();
 }
 
 static void r700SendFSState(GLcontext *ctx, struct radeon_state_atom *atom)
@@ -1301,8 +1314,8 @@ void r600InitAtoms(context_t *context)
 	ALLOC_STATE(spi, always, (59 + R700_MAX_SHADER_EXPORTS), r700SendSPIState);
 	ALLOC_STATE(vpt, always, 16, r700SendViewportState);
 	ALLOC_STATE(fs, always, 18, r700SendFSState);
-	ALLOC_STATE(vs, always, 18, r700SendVSState);
-	ALLOC_STATE(ps, always, 21, r700SendPSState);
+	ALLOC_STATE(vs, always, 21, r700SendVSState);
+	ALLOC_STATE(ps, always, 24, r700SendPSState);
 	ALLOC_STATE(vs_consts, vs_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendVSConsts);
 	ALLOC_STATE(ps_consts, ps_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendPSConsts);
 	ALLOC_STATE(vtx, vtx, (6 + (VERT_ATTRIB_MAX * 18)), r700SendVTXState);
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c
index ccafd433bf..21ac46e7b8 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.c
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.c
@@ -73,11 +73,11 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
 		pAsm->uiFP_AttributeMap[FRAG_ATTRIB_COL1] = pAsm->number_used_registers++;
 	}
 
-        unBit = 1 << FRAG_ATTRIB_FOGC;
-        if(mesa_fp->Base.InputsRead & unBit)
-        {
-                pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FOGC] = pAsm->number_used_registers++;
-        }
+    unBit = 1 << FRAG_ATTRIB_FOGC;
+    if(mesa_fp->Base.InputsRead & unBit)
+    {
+            pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FOGC] = pAsm->number_used_registers++;
+    }
 
 	for(i=0; i<8; i++)
 	{
@@ -88,6 +88,62 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
 		}
 	}
 
+/* order has been taken care of */
+#if 1
+    for(i=FRAG_ATTRIB_VAR0; i<FRAG_ATTRIB_MAX; i++)
+    {
+        unBit = 1 << i;
+        if(mesa_fp->Base.InputsRead & unBit)
+		{
+            pAsm->uiFP_AttributeMap[i] = pAsm->number_used_registers++;
+        }
+    }
+#else
+    if( (mesa_fp->Base.InputsRead >> FRAG_ATTRIB_VAR0) > 0 )
+    {
+	    struct r700_vertex_program_cont *vpc =
+		       (struct r700_vertex_program_cont *)ctx->VertexProgram._Current;
+        struct gl_program_parameter_list * VsVarying = vpc->mesa_program.Base.Varying;
+        struct gl_program_parameter_list * PsVarying = mesa_fp->Base.Varying;
+        struct gl_program_parameter      * pVsParam;
+        struct gl_program_parameter      * pPsParam;
+        GLuint j, k;
+        GLuint unMaxVarying = 0;
+
+        for(i=0; i<VsVarying->NumParameters; i++)
+        {
+            pAsm->uiFP_AttributeMap[i + FRAG_ATTRIB_VAR0] = 0;
+        }
+
+        for(i=FRAG_ATTRIB_VAR0; i<FRAG_ATTRIB_MAX; i++)
+	    {
+            unBit = 1 << i;
+            if(mesa_fp->Base.InputsRead & unBit)
+		    {
+                j = i - FRAG_ATTRIB_VAR0;
+                pPsParam = PsVarying->Parameters + j;
+
+                for(k=0; k<VsVarying->NumParameters; k++)
+                {					
+                    pVsParam = VsVarying->Parameters + k;
+
+			        if( strcmp(pPsParam->Name, pVsParam->Name) == 0)
+                    {
+                        pAsm->uiFP_AttributeMap[i] = pAsm->number_used_registers + k;                  
+                        if(k > unMaxVarying)
+                        {
+                            unMaxVarying = k;
+                        }
+                        break;
+                    }
+                }
+		    }
+        }
+
+        pAsm->number_used_registers += unMaxVarying + 1;
+    }
+#endif
+
 /* Map temporary registers (GPRs) */
     pAsm->starting_temp_register_number = pAsm->number_used_registers;
 
@@ -127,6 +183,8 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
         pAsm->pucOutMask[ui] = 0x0;
     }
 
+    pAsm->flag_reg_index = pAsm->number_used_registers++;
+
     pAsm->uFirstHelpReg = pAsm->number_used_registers;
 }
 
@@ -247,8 +305,11 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
 	{
 		return GL_FALSE;
     }
+
+    InitShaderProgram(&(fp->r700AsmCode));
 	
-	if( GL_FALSE == AssembleInstr(mesa_fp->Base.NumInstructions,
+	if( GL_FALSE == AssembleInstr(0,
+                                  mesa_fp->Base.NumInstructions,
                                   &(mesa_fp->Base.Instructions[0]), 
                                   &(fp->r700AsmCode)) )
 	{
@@ -260,6 +321,11 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
         return GL_FALSE;
     }
 
+    if( GL_FALSE == RelocProgram(&(fp->r700AsmCode)) )
+    {
+        return GL_FALSE;
+    }
+
     fp->r700Shader.nRegs = (fp->r700AsmCode.number_used_registers == 0) ? 0 
                          : (fp->r700AsmCode.number_used_registers - 1);
 
@@ -459,6 +525,22 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 	    }
     }
 
+    for(i=FRAG_ATTRIB_VAR0; i<FRAG_ATTRIB_MAX; i++)
+	{
+		unBit = 1 << i;
+		if(mesa_fp->Base.InputsRead & unBit)
+		{
+            ui = pAsm->uiFP_AttributeMap[i];
+            SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
+            SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui,
+		             SEMANTIC_shift, SEMANTIC_mask);
+            if (r700->SPI_INTERP_CONTROL_0.u32All & FLAT_SHADE_ENA_bit)
+		        SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+            else
+		        CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+		}
+	}
+
     exportCount = (r700->ps.SQ_PGM_EXPORTS_PS.u32All & EXPORT_MODE_mask) / (1 << EXPORT_MODE_shift);
     if (r700->CB_SHADER_CONTROL.u32All != ((1 << exportCount) - 1))
     {
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index 47f89c91f8..eab27cbd84 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -526,6 +526,9 @@ static void r700ConvertAttrib(GLcontext *ctx, int count,
 
     radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, 
                          sizeof(GLfloat) * input->Size * count, 32);
+
+    radeon_bo_map(attr->bo, 1);
+
     dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
 
     assert(src_ptr != NULL);
@@ -559,6 +562,8 @@ static void r700ConvertAttrib(GLcontext *ctx, int count,
             break;
     }
 
+    radeon_bo_unmap(attr->bo);
+
     if (mapped_named_bo) 
     {
         ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
@@ -577,6 +582,8 @@ static void r700AlignDataToDword(GLcontext *ctx,
 
     radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32);
 
+    radeon_bo_map(attr->bo, 1);
+
     if (!input->BufferObj->Pointer) 
     {
         ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
@@ -596,6 +603,7 @@ static void r700AlignDataToDword(GLcontext *ctx,
         }
     }
 
+    radeon_bo_unmap(attr->bo);
     if (mapped_named_bo) 
     {
         ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
@@ -664,14 +672,18 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
 
                 radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, 
                                      &context->stream_desc[index].bo_offset, size, 32);
+
+                radeon_bo_map(context->stream_desc[index].bo, 1);
                 assert(context->stream_desc[index].bo->ptr != NULL);
+
+
                 dst = (uint32_t *)ADD_POINTERS(context->stream_desc[index].bo->ptr, 
                                                context->stream_desc[index].bo_offset);
 
                 switch (context->stream_desc[index].dwords) 
                 {
                 case 1:                     
-                    radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count);                         
+                    radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count);
                     break;
                 case 2: 
                     radeonEmitVec8(dst, input[i]->Ptr, input[i]->StrideB, local_count); 
@@ -686,6 +698,7 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
                     assert(0); 
                     break;
                 }
+		radeon_bo_unmap(context->stream_desc[index].bo);
             }
         }
 
@@ -757,6 +770,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
 
+	radeon_bo_map(context->ind_buf.bo, 1);
 	assert(context->ind_buf.bo->ptr != NULL);
 	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
 
@@ -770,6 +784,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
             *out++ = in[i];
         }
 
+	radeon_bo_unmap(context->ind_buf.bo);
 #if MESA_BIG_ENDIAN
     }
     else
@@ -780,6 +795,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
 
+	radeon_bo_map(context->ind_buf.bo, 1);
 	assert(context->ind_buf.bo->ptr != NULL);
 	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
 
@@ -792,6 +808,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
         {
             *out++ = in[i];
         }
+	radeon_bo_unmap(context->ind_buf.bo);
 #endif
     }
 
@@ -837,11 +854,13 @@ static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
+	radeon_bo_map(context->ind_buf.bo, 1);
 	assert(context->ind_buf.bo->ptr != NULL);
 	dst_ptr = ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
 
         _mesa_memcpy(dst_ptr, src_ptr, size);
 
+	radeon_bo_unmap(context->ind_buf.bo);
         context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT);
         context->ind_buf.count = mesa_ind_buf->count;
 
diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c
index 955ea4e4e1..2eed1acc2f 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.c
+++ b/src/mesa/drivers/dri/r600/r700_shader.c
@@ -159,13 +159,18 @@ void Init_R700_Shader(R700_Shader * pShader)
 	pShader->lstVTXInstructions.uNumOfNode=0;
 }
 
+void SetActiveCFlist(R700_Shader *pShader, TypedShaderList * plstCF)
+{
+    pShader->plstCFInstructions_active = plstCF;
+}
+
 void AddCFInstruction(R700_Shader *pShader, R700ControlFlowInstruction *pCFInst)
 {
     R700ControlFlowSXClause*  pSXClause; 
     R700ControlFlowSMXClause* pSMXClause;
 
-    pCFInst->m_uIndex = pShader->lstCFInstructions.uNumOfNode;
-    AddInstToList(&(pShader->lstCFInstructions), 
+    pCFInst->m_uIndex = pShader->plstCFInstructions_active->uNumOfNode;
+    AddInstToList(pShader->plstCFInstructions_active, 
                   (R700ShaderInstruction*)pCFInst);
     pShader->uShaderBinaryDWORDSize += GetInstructionSize(pCFInst->m_ShaderInstType);
 
diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h
index c6a058617e..0599ffd901 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.h
+++ b/src/mesa/drivers/dri/r600/r700_shader.h
@@ -109,6 +109,7 @@ typedef struct R700_Shader
     GLuint  uStackSize;
     GLuint  uMaxCallDepth;
 
+    TypedShaderList * plstCFInstructions_active;
 	TypedShaderList lstCFInstructions;
 	TypedShaderList lstALUInstructions;
 	TypedShaderList lstTEXInstructions;
@@ -132,13 +133,13 @@ void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruc
 void ResolveLinks(R700_Shader *pShader);
 void Assemble(R700_Shader *pShader);
 
-
 //Interface
 void Init_R700_Shader(R700_Shader * pShader);
 void AddCFInstruction(R700_Shader *pShader, R700ControlFlowInstruction *pCFInst);
 void AddVTXInstruction(R700_Shader *pShader, R700VertexInstruction *pVTXInst);
 void AddTEXInstruction(R700_Shader *pShader, R700TextureInstruction *pTEXInst);
 void AddALUInstruction(R700_Shader *pShader, R700ALUInstruction *pALUInst);
+void SetActiveCFlist(R700_Shader *pShader, TypedShaderList * plstCF);
 
 void LoadProgram(R700_Shader *pShader);
 void UpdateShaderRegisters(R700_Shader *pShader);
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index ffc6068bd8..c8f72d588b 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -111,6 +111,15 @@ unsigned int Map_Vertex_Output(r700_AssemblerBase       *pAsm,
 		}
 	}
 
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+		unBit = 1 << i;
+		if(mesa_vp->Base.OutputsWritten & unBit)
+		{
+			pAsm->ucVP_OutputMap[i] = unTotal++;
+		}
+	}
+
 	return (unTotal - unStart);
 }
 
@@ -235,6 +244,8 @@ void Map_Vertex_Program(GLcontext *ctx,
         pAsm->number_used_registers += mesa_vp->Base.NumTemporaries;
     }
 
+    pAsm->flag_reg_index = pAsm->number_used_registers++;
+
     pAsm->uFirstHelpReg = pAsm->number_used_registers;
 }
 
@@ -324,7 +335,10 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 		return NULL;
 	}
 
-	if(GL_FALSE == AssembleInstr(vp->mesa_program->Base.NumInstructions,
+    InitShaderProgram(&(vp->r700AsmCode));
+
+	if(GL_FALSE == AssembleInstr(0,
+                                 vp->mesa_program->Base.NumInstructions,
                                  &(vp->mesa_program->Base.Instructions[0]),
                                  &(vp->r700AsmCode)) )
 	{
@@ -336,6 +350,11 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
         return NULL;
     }
 
+    if( GL_FALSE == RelocProgram(&(vp->r700AsmCode)) )
+    {
+        return GL_FALSE;
+    }
+
     vp->r700Shader.nRegs = (vp->r700AsmCode.number_used_registers == 0) ? 0 
                          : (vp->r700AsmCode.number_used_registers - 1);
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
index 7141371633..46e30b905a 100644
--- a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
+++ b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
@@ -39,6 +39,7 @@
 #define RADEON_BO_FLAGS_MICRO_TILE  2
 
 struct radeon_bo_manager;
+struct radeon_cs;
 
 struct radeon_bo {
     uint32_t                    alignment;
@@ -74,6 +75,7 @@ struct radeon_bo_funcs {
     int (*bo_get_tiling)(struct radeon_bo *bo, uint32_t *tiling_flags,
 			  uint32_t *pitch);
     int (*bo_is_busy)(struct radeon_bo *bo, uint32_t *domain);
+    int (*bo_is_referenced_by_cs)(struct radeon_bo *bo, struct radeon_cs *cs);
 };
 
 struct radeon_bo_manager {
@@ -199,6 +201,15 @@ static inline int radeon_bo_is_static(struct radeon_bo *bo)
 	return 0;
 }
 
+static inline int _radeon_bo_is_referenced_by_cs(struct radeon_bo *bo,
+                                                 struct radeon_cs *cs,
+                                                 const char *file,
+                                                 const char *func,
+                                                 unsigned line)
+{
+    return bo->cref > 1;
+}
+
 #define radeon_bo_open(bom, h, s, a, d, f)\
     _radeon_bo_open(bom, h, s, a, d, f, __FILE__, __FUNCTION__, __LINE__)
 #define radeon_bo_ref(bo)\
@@ -215,5 +226,7 @@ static inline int radeon_bo_is_static(struct radeon_bo *bo)
     _radeon_bo_wait(bo, __FILE__, __func__, __LINE__)
 #define radeon_bo_is_busy(bo, domain) \
     _radeon_bo_is_busy(bo, domain, __FILE__, __func__, __LINE__)
+#define radeon_bo_is_referenced_by_cs(bo, cs) \
+    _radeon_bo_is_referenced_by_cs(bo, cs, __FILE__, __FUNCTION__, __LINE__)
 
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
index 8fac5c6c51..99d3ec7005 100644
--- a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
+++ b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
@@ -136,8 +136,13 @@ radeonBufferSubData(GLcontext * ctx,
                     const GLvoid * data,
                     struct gl_buffer_object *obj)
 {
+    radeonContextPtr radeon = RADEON_CONTEXT(ctx);
     struct radeon_buffer_object *radeon_obj = get_radeon_buffer_object(obj);
 
+    if (radeon_bo_is_referenced_by_cs(radeon_obj->bo, radeon->cmdbuf.cs)) {
+        radeon_firevertices(radeon);
+    }
+
     radeon_bo_map(radeon_obj->bo, GL_TRUE);
 
     _mesa_memcpy(radeon_obj->bo->ptr + offset, data, size);
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c
index 097ab7cf61..3b4366aa61 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common.c
@@ -257,7 +257,9 @@ void radeonScissor(GLcontext* ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
 	if (ctx->Scissor.Enabled) {
 		/* We don't pipeline cliprect changes */
-		radeon_firevertices(radeon);
+		if (!radeon->radeonScreen->kernel_mm) {
+			radeon_firevertices(radeon);
+		}
 		radeonUpdateScissor(ctx);
 	}
 }
@@ -1123,8 +1125,6 @@ void radeonFlush(GLcontext *ctx)
 	if (radeon->dma.flush)
 		radeon->dma.flush( ctx );
 
-	radeonEmitState(radeon);
-
 	if (radeon->cmdbuf.cs->cdw)
 		rcommonFlushCmdBuf(radeon, __FUNCTION__);
 
@@ -1147,9 +1147,6 @@ void radeonFlush(GLcontext *ctx)
 			}
 		}
 	}
-
-	make_empty_list(&radeon->query.not_flushed_head);
-
 }
 
 /* Make sure all commands have been sent to the hardware and have
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c
index fe99644907..71f70d724b 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c
@@ -262,10 +262,9 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 		else
 			radeon->texture_row_align = 32;
 		radeon->texture_rect_row_align = 64;
-		radeon->texture_compressed_row_align = 64;
+		radeon->texture_compressed_row_align = 32;
 	}
 
-	make_empty_list(&radeon->query.not_flushed_head);
 	radeon_init_dma(radeon);
 
 	return GL_TRUE;
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.h b/src/mesa/drivers/dri/radeon/radeon_common_context.h
index 0309345393..ad953ddbb5 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.h
@@ -208,6 +208,10 @@ struct radeon_tex_obj {
 	 * and so on.
 	 */
 	GLboolean validated;
+	/* Minimum LOD to be used during rendering */
+	unsigned minLod;
+	/* Miximum LOD to be used during rendering */
+	unsigned maxLod;
 
 	GLuint override_offset;
 	GLboolean image_override; /* Image overridden by GLX_EXT_tfp */
@@ -324,6 +328,7 @@ struct radeon_swtcl_info {
 	GLuint vertex_attr_count;
 
 	GLuint emit_prediction;
+        struct radeon_bo *bo;
 };
 
 #define RADEON_MAX_AOS_ARRAYS		16
@@ -502,7 +507,6 @@ struct radeon_context {
 
    struct {
 	struct radeon_query_object *current;
-	struct radeon_query_object not_flushed_head;
 	struct radeon_state_atom queryobj;
    } query;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_dma.c b/src/mesa/drivers/dri/radeon/radeon_dma.c
index c6edbae9a1..89a99974e2 100644
--- a/src/mesa/drivers/dri/radeon/radeon_dma.c
+++ b/src/mesa/drivers/dri/radeon/radeon_dma.c
@@ -151,6 +151,7 @@ void rcommon_emit_vector(GLcontext * ctx, struct radeon_aos *aos,
 	aos->components = size;
 	aos->count = count;
 
+	radeon_bo_map(aos->bo, 1);
 	out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
 	switch (size) {
 	case 1: radeonEmitVec4(out, data, stride, count); break;
@@ -161,6 +162,7 @@ void rcommon_emit_vector(GLcontext * ctx, struct radeon_aos *aos,
 		assert(0);
 		break;
 	}
+	radeon_bo_unmap(aos->bo);
 }
 
 void radeon_init_dma(radeonContextPtr rmesa)
@@ -183,10 +185,6 @@ void radeonRefillCurrentDmaRegion(radeonContextPtr rmesa, int size)
 			__FUNCTION__, size, rmesa->dma.minimum_size);
 
 
-	/* unmap old reserved bo */
-	if (!is_empty_list(&rmesa->dma.reserved))
-		radeon_bo_unmap(first_elem(&rmesa->dma.reserved)->bo);
-
 	if (is_empty_list(&rmesa->dma.free)
 	      || last_elem(&rmesa->dma.free)->bo->size < size) {
 		dma_bo = CALLOC_STRUCT(radeon_dma_bo);
@@ -223,8 +221,6 @@ again_alloc:
         /* Cmd buff have been flushed in radeon_revalidate_bos */
 		goto again_alloc;
 	}
-
-	radeon_bo_map(first_elem(&rmesa->dma.reserved)->bo, 1);
 }
 
 /* Allocates a region from rmesa->dma.current.  If there isn't enough
@@ -281,7 +277,6 @@ void radeonFreeDmaRegions(radeonContextPtr rmesa)
 
 	foreach_s(dma_bo, temp, &rmesa->dma.reserved) {
 		remove_from_list(dma_bo);
-		radeon_bo_unmap(dma_bo->bo);
 	        radeon_bo_unref(dma_bo->bo);
 		FREE(dma_bo);
 	}
@@ -367,9 +362,6 @@ void radeonReleaseDmaRegions(radeonContextPtr rmesa)
 		insert_at_tail(&rmesa->dma.free, dma_bo);
 	}
 
-	/* unmap the last dma region */
-	if (!is_empty_list(&rmesa->dma.reserved))
-		radeon_bo_unmap(first_elem(&rmesa->dma.reserved)->bo);
 	/* move reserved to wait list */
 	foreach_s(dma_bo, temp, &rmesa->dma.reserved) {
 		/* free objects that are too small to be used because of large request */
@@ -403,11 +395,12 @@ void rcommon_flush_last_swtcl_prim( GLcontext *ctx  )
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	struct radeon_dma *dma = &rmesa->dma;
 		
-
 	if (RADEON_DEBUG & RADEON_IOCTL)
 		fprintf(stderr, "%s\n", __FUNCTION__);
 	dma->flush = NULL;
 
+	radeon_bo_unmap(rmesa->swtcl.bo);
+
 	if (!is_empty_list(&dma->reserved)) {
 	    GLuint current_offset = dma->current_used;
 
@@ -422,6 +415,8 @@ void rcommon_flush_last_swtcl_prim( GLcontext *ctx  )
 	    }
 	    rmesa->swtcl.numverts = 0;
 	}
+	radeon_bo_unref(rmesa->swtcl.bo);
+	rmesa->swtcl.bo = NULL;
 }
 /* Alloc space in the current dma region.
  */
@@ -432,6 +427,7 @@ rcommonAllocDmaLowVerts( radeonContextPtr rmesa, int nverts, int vsize )
 	void *head;
 	if (RADEON_DEBUG & RADEON_IOCTL)
 		fprintf(stderr, "%s\n", __FUNCTION__);
+
 	if(is_empty_list(&rmesa->dma.reserved)
 	      ||rmesa->dma.current_vertexptr + bytes > first_elem(&rmesa->dma.reserved)->bo->size) {
 		if (rmesa->dma.flush) {
@@ -455,7 +451,13 @@ rcommonAllocDmaLowVerts( radeonContextPtr rmesa, int nverts, int vsize )
                 rmesa->swtcl.numverts * rmesa->swtcl.vertex_size * 4 ==
                 rmesa->dma.current_vertexptr );
 
-	head = (first_elem(&rmesa->dma.reserved)->bo->ptr + rmesa->dma.current_vertexptr);
+	if (!rmesa->swtcl.bo) {
+		rmesa->swtcl.bo = first_elem(&rmesa->dma.reserved)->bo;
+		radeon_bo_ref(rmesa->swtcl.bo);
+		radeon_bo_map(rmesa->swtcl.bo, 1);
+	}
+
+	head = (rmesa->swtcl.bo->ptr + rmesa->dma.current_vertexptr);
 	rmesa->dma.current_vertexptr += bytes;
 	rmesa->swtcl.numverts += nverts;
 	return head;
diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c
index bf69cd9337..7ec641ff18 100644
--- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
+++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
@@ -90,7 +90,7 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_R3_G3_B2:
    case GL_RGB4:
    case GL_RGB5:
-      rb->Format = MESA_FORMAT_RGB565;
+      rb->Format = _dri_texformat_rgb565;
       rb->DataType = GL_UNSIGNED_BYTE;
       cpp = 2;
       break;
@@ -99,7 +99,7 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      rb->Format = MESA_FORMAT_ARGB8888;
+      rb->Format = _dri_texformat_argb8888;
       rb->DataType = GL_UNSIGNED_BYTE;
       cpp = 4;
       break;
@@ -111,7 +111,7 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      rb->Format = MESA_FORMAT_ARGB8888;
+      rb->Format = _dri_texformat_argb8888;
       rb->DataType = GL_UNSIGNED_BYTE;
       cpp = 4;
       break;
@@ -261,14 +261,32 @@ radeon_create_renderbuffer(gl_format format, __DRIdrawablePrivate *driDrawPriv)
 
     switch (format) {
         case MESA_FORMAT_RGB565:
+	    assert(_mesa_little_endian());
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
+	    break;
+        case MESA_FORMAT_RGB565_REV:
+	    assert(!_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
             rrb->base._BaseFormat = GL_RGB;
 	    break;
         case MESA_FORMAT_XRGB8888:
+	    assert(_mesa_little_endian());
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
+	    break;
+        case MESA_FORMAT_XRGB8888_REV:
+	    assert(!_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
             rrb->base._BaseFormat = GL_RGB;
 	    break;
 	case MESA_FORMAT_ARGB8888:
+	    assert(_mesa_little_endian());
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGBA;
+	    break;
+	case MESA_FORMAT_ARGB8888_REV:
+	    assert(!_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
             rrb->base._BaseFormat = GL_RGBA;
 	    break;
@@ -359,21 +377,21 @@ radeon_update_wrapper(GLcontext *ctx, struct radeon_renderbuffer *rrb,
 	gl_format texFormat;
 
 restart:
-	if (texImage->TexFormat == MESA_FORMAT_ARGB8888) {
+	if (texImage->TexFormat == _dri_texformat_argb8888) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to RGBA8 texture OK\n");
 	}
-	else if (texImage->TexFormat == MESA_FORMAT_RGB565) {
+	else if (texImage->TexFormat == _dri_texformat_rgb565) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to RGB5 texture OK\n");
 	}
-	else if (texImage->TexFormat == MESA_FORMAT_ARGB1555) {
+	else if (texImage->TexFormat == _dri_texformat_argb1555) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to ARGB1555 texture OK\n");
 	}
-	else if (texImage->TexFormat == MESA_FORMAT_ARGB4444) {
+	else if (texImage->TexFormat == _dri_texformat_argb4444) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
-		DBG("Render to ARGB1555 texture OK\n");
+		DBG("Render to ARGB4444 texture OK\n");
 	}
 	else if (texImage->TexFormat == MESA_FORMAT_Z16) {
 		rrb->base.DataType = GL_UNSIGNED_SHORT;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
index 08e1c5d00d..d810e6080e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
@@ -76,12 +76,14 @@ static void emit_vecfog(GLcontext *ctx, struct radeon_aos *aos,
 
    /* Emit the data
     */
+   radeon_bo_map(aos->bo, 1);
    out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
    for (i = 0; i < count; i++) {
       out[0] = radeonComputeFogBlendFactor( ctx, *(GLfloat *)data );
       out++;
       data += stride;
    }
+   radeon_bo_unmap(aos->bo);
 }
 
 static void emit_s0_vec(uint32_t *out, GLvoid *data, int stride, int count)
@@ -151,6 +153,7 @@ static void emit_tex_vector(GLcontext *ctx, struct radeon_aos *aos,
 
    /* Emit the data
     */
+   radeon_bo_map(aos->bo, 1);
    out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
    switch (size) {
    case 1:
@@ -170,6 +173,7 @@ static void emit_tex_vector(GLcontext *ctx, struct radeon_aos *aos,
       exit(1);
       break;
    }
+   radeon_bo_unmap(aos->bo);
 }
 
 
@@ -196,12 +200,12 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.obj.buf) 
 	rcommon_emit_vector( ctx, 
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->ObjPtr->data,
-			     VB->ObjPtr->size,
-			     VB->ObjPtr->stride,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+			     VB->AttribPtr[_TNL_ATTRIB_POS]->size,
+			     VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
 			     count);
 
-      switch( VB->ObjPtr->size ) {
+      switch( VB->AttribPtr[_TNL_ATTRIB_POS]->size ) {
       case 4: vfmt |= RADEON_CP_VC_FRMT_W0;
       case 3: vfmt |= RADEON_CP_VC_FRMT_Z;
       case 2: vfmt |= RADEON_CP_VC_FRMT_XY;
@@ -216,9 +220,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.norm.buf)
 	 rcommon_emit_vector( ctx, 
 			      &(rmesa->tcl.aos[nr]),
-			      (char *)VB->NormalPtr->data,
+			      (char *)VB->AttribPtr[_TNL_ATTRIB_NORMAL]->data,
 			      3,
-			      VB->NormalPtr->stride,
+			      VB->AttribPtr[_TNL_ATTRIB_NORMAL]->stride,
 			      count);
 
       vfmt |= RADEON_CP_VC_FRMT_N0;
@@ -227,9 +231,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
    if (inputs & VERT_BIT_COLOR0) {
       int emitsize;
-      if (VB->ColorPtr[0]->size == 4 &&
-	  (VB->ColorPtr[0]->stride != 0 ||
-	   VB->ColorPtr[0]->data[0][3] != 1.0)) {
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size == 4 &&
+	  (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride != 0 ||
+	   VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data[0][3] != 1.0)) {
 	 vfmt |= RADEON_CP_VC_FRMT_FPCOLOR | RADEON_CP_VC_FRMT_FPALPHA;
 	 emitsize = 4;
       }
@@ -242,9 +246,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.rgba.buf)
 	rcommon_emit_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->ColorPtr[0]->data,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data,
 			     emitsize,
-			     VB->ColorPtr[0]->stride,
+			     VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride,
 			     count);
 
       nr++;
@@ -256,9 +260,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
 	rcommon_emit_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->SecondaryColorPtr[0]->data,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data,
 			     3,
-			     VB->SecondaryColorPtr[0]->stride,
+			     VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride,
 			     count);
       }
 
@@ -273,8 +277,8 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.fog.buf)
 	 emit_vecfog( ctx,
 		      &(rmesa->tcl.aos[nr]),
-		      (char *)VB->FogCoordPtr->data,
-		      VB->FogCoordPtr->stride,
+		      (char *)VB->AttribPtr[_TNL_ATTRIB_FOG]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_FOG]->stride,
 		      count);
 
       vfmt |= RADEON_CP_VC_FRMT_FPFOG;
@@ -290,24 +294,24 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 if (!rmesa->tcl.tex[unit].buf)
 	    emit_tex_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->TexCoordPtr[unit]->data,
-			     VB->TexCoordPtr[unit]->size,
-			     VB->TexCoordPtr[unit]->stride,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->data,
+			     VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size,
+			     VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->stride,
 			     count );
 	 nr++;
 
 	 vfmt |= RADEON_ST_BIT(unit);
          /* assume we need the 3rd coord if texgen is active for r/q OR at least
 	    3 coords are submitted. This may not be 100% correct */
-         if (VB->TexCoordPtr[unit]->size >= 3) {
+         if (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) {
 	    vtx |= RADEON_Q_BIT(unit);
 	    vfmt |= RADEON_Q_BIT(unit);
 	 }
 	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
 	    vtx |= RADEON_Q_BIT(unit);
-	 else if ((VB->TexCoordPtr[unit]->size >= 3) &&
+	 else if ((VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) &&
 	          ((ctx->Texture.Unit[unit]._ReallyEnabled & (TEXTURE_CUBE_BIT)) == 0)) {
-	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    GLuint swaptexmatcol = (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size - 3);
 	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
 		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
 	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
index 515783135d..d764ccb982 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
@@ -56,18 +56,18 @@ static void TAG(emit)( GLcontext *ctx,
 
    radeon_print(RADEON_SWRENDER, RADEON_VERBOSE, "%s\n", __FUNCTION__);
 
-   coord = (GLuint (*)[4])VB->ObjPtr->data;
-   coord_stride = VB->ObjPtr->stride;
+   coord = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_POS]->data;
+   coord_stride = VB->AttribPtr[_TNL_ATTRIB_POS]->stride;
 
    if (DO_TEX2) {
-      if (VB->TexCoordPtr[2]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX2]) {
 	 const GLuint t2 = GET_TEXSOURCE(2);
-	 tc2 = (GLuint (*)[4])VB->TexCoordPtr[t2]->data;
-	 tc2_stride = VB->TexCoordPtr[t2]->stride;
-	 if (DO_PTEX && VB->TexCoordPtr[t2]->size < 3) {
+	 tc2 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->data;
+	 tc2_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->stride;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size < 3) {
 	    fill_tex |= (1<<2);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t2]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size < 4) {
 	    rqcoordsnoswap |= (1<<2);
 	 }
       } else {
@@ -77,14 +77,14 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_TEX1) {
-      if (VB->TexCoordPtr[1]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]) {
 	 const GLuint t1 = GET_TEXSOURCE(1);
-	 tc1 = (GLuint (*)[4])VB->TexCoordPtr[t1]->data;
-	 tc1_stride = VB->TexCoordPtr[t1]->stride;
-	 if (DO_PTEX && VB->TexCoordPtr[t1]->size < 3) {
+	 tc1 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+	 tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size < 3) {
 	    fill_tex |= (1<<1);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t1]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size < 4) {
 	    rqcoordsnoswap |= (1<<1);
 	 }
       } else {
@@ -94,14 +94,14 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_TEX0) {
-      if (VB->TexCoordPtr[0]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]) {
 	 const GLuint t0 = GET_TEXSOURCE(0);
-	 tc0_stride = VB->TexCoordPtr[t0]->stride;
-	 tc0 = (GLuint (*)[4])VB->TexCoordPtr[t0]->data;
-	 if (DO_PTEX && VB->TexCoordPtr[t0]->size < 3) {
+	 tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
+	 tc0 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size < 3) {
 	    fill_tex |= (1<<0);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t0]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size < 4) {
 	    rqcoordsnoswap |= (1<<0);
 	 }
       } else {
@@ -112,9 +112,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_NORM) {
-      if (VB->NormalPtr) {
-	 norm_stride = VB->NormalPtr->stride;
-	 norm = (GLuint (*)[4])VB->NormalPtr->data;
+      if (VB->AttribPtr[_TNL_ATTRIB_NORMAL]) {
+	 norm_stride = VB->AttribPtr[_TNL_ATTRIB_NORMAL]->stride;
+	 norm = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_NORMAL]->data;
       } else {
 	 norm_stride = 0;
 	 norm = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_NORMAL];
@@ -122,9 +122,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_RGBA) {
-      if (VB->ColorPtr[0]) {
-	 col = VB->ColorPtr[0]->data;
-	 col_stride = VB->ColorPtr[0]->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR0]) {
+	 col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+	 col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
       } else {
 	 col = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
 	 col_stride = 0;
@@ -132,9 +132,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_SPEC_OR_FOG) {
-      if (VB->SecondaryColorPtr[0]) {
-	 spec = VB->SecondaryColorPtr[0]->data;
-	 spec_stride = VB->SecondaryColorPtr[0]->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
+	 spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+	 spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
       } else {
 	 spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
 	 spec_stride = 0;
@@ -142,9 +142,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_SPEC_OR_FOG) {
-      if (VB->FogCoordPtr) {
-	 fog = VB->FogCoordPtr->data;
-	 fog_stride = VB->FogCoordPtr->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+	 fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+	 fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
       } else {
 	 fog = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_FOG];
 	 fog_stride = 0;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index 78ec119302..98f96ff2a7 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -326,7 +326,7 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
    if (1) {
       req |= RADEON_CP_VC_FRMT_Z;
-      if (VB->ObjPtr->size == 4) {
+      if (VB->AttribPtr[_TNL_ATTRIB_POS]->size == 4) {
 	 req |= RADEON_CP_VC_FRMT_W0;
       }
    }
@@ -348,15 +348,15 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 req |= RADEON_ST_BIT(unit);
 	 /* assume we need the 3rd coord if texgen is active for r/q OR at least
 	    3 coords are submitted. This may not be 100% correct */
-	 if (VB->TexCoordPtr[unit]->size >= 3) {
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) {
 	    req |= RADEON_Q_BIT(unit);
 	    vtx |= RADEON_Q_BIT(unit);
 	 }
 	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
 	    vtx |= RADEON_Q_BIT(unit);
-	 else if ((VB->TexCoordPtr[unit]->size >= 3) &&
+	 else if ((VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) &&
 	          ((ctx->Texture.Unit[unit]._ReallyEnabled & (TEXTURE_CUBE_BIT)) == 0)) {
-	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    GLuint swaptexmatcol = (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size - 3);
 	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
 		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
 	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
@@ -390,19 +390,19 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
     * this, add more vertex code (for obj-2, obj-3) or preferably move
     * to maos.  
     */
-   if (VB->ObjPtr->size < 3 || 
-       (VB->ObjPtr->size == 3 && 
+   if (VB->AttribPtr[_TNL_ATTRIB_POS]->size < 3 ||
+       (VB->AttribPtr[_TNL_ATTRIB_POS]->size == 3 &&
 	(setup_tab[i].vertex_format & RADEON_CP_VC_FRMT_W0))) {
 
       _math_trans_4f( rmesa->tcl.ObjClean.data,
-		      VB->ObjPtr->data,
-		      VB->ObjPtr->stride,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
 		      GL_FLOAT,
-		      VB->ObjPtr->size,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->size,
 		      0,
 		      VB->Count );
 
-      switch (VB->ObjPtr->size) {
+      switch (VB->AttribPtr[_TNL_ATTRIB_POS]->size) {
       case 1:
 	    _mesa_vector4f_clean_elem(&rmesa->tcl.ObjClean, VB->Count, 1);
       case 2:
@@ -416,14 +416,14 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 break;
       }
 
-      VB->ObjPtr = &rmesa->tcl.ObjClean;
+      VB->AttribPtr[_TNL_ATTRIB_POS] = &rmesa->tcl.ObjClean;
    }
 
 
-
+   radeon_bo_map(rmesa->radeon.tcl.aos[0].bo, 1);
    setup_tab[i].emit( ctx, 0, VB->Count, 
 		      rmesa->radeon.tcl.aos[0].bo->ptr + rmesa->radeon.tcl.aos[0].offset);
-
+   radeon_bo_unmap(rmesa->radeon.tcl.aos[0].bo);
    //   rmesa->radeon.tcl.aos[0].size = setup_tab[i].vertex_size;
    rmesa->radeon.tcl.aos[0].stride = setup_tab[i].vertex_size;
    rmesa->tcl.vertex_format = setup_tab[i].vertex_format;
diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
index dadc72f4c1..f2f7b2a9fd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
+++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
@@ -1,4 +1,5 @@
 /*
+ * Copyright (C) 2009 Maciej Cencora.
  * Copyright (C) 2008 Nicolai Haehnle.
  *
  * All Rights Reserved.
@@ -32,50 +33,52 @@
 
 #include "main/simple_list.h"
 #include "main/texcompress.h"
-
-static GLuint radeon_compressed_texture_size(GLcontext *ctx,
-		GLsizei width, GLsizei height, GLsizei depth,
-		GLuint mesaFormat)
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "radeon_texture.h"
+
+static unsigned get_aligned_compressed_row_stride(
+		gl_format format,
+		unsigned width,
+		unsigned minStride)
 {
-	GLuint size = _mesa_format_image_size(mesaFormat, width, height, depth);
-
-	if (mesaFormat == MESA_FORMAT_RGB_DXT1 ||
-	    mesaFormat == MESA_FORMAT_RGBA_DXT1) {
-		if (width + 3 < 8)	/* width one block */
-			size = size * 4;
-		else if (width + 3 < 16)
-			size = size * 2;
-	} else {
-		/* DXT3/5, 16 bytes per block */
-	  //		WARN_ONCE("DXT 3/5 suffers from multitexturing problems!\n");
-		if (width + 3 < 8)
-			size = size * 2;
+	const unsigned blockSize = _mesa_get_format_bytes(format);
+	unsigned blockWidth, blockHeight, numXBlocks;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
+	numXBlocks = (width + blockWidth - 1) / blockWidth;
+
+	while (numXBlocks * blockSize < minStride)
+	{
+		++numXBlocks;
 	}
 
-	return size;
+	return numXBlocks * blockSize;
 }
 
+static unsigned get_compressed_image_size(
+		gl_format format,
+		unsigned rowStride,
+		unsigned height)
+{
+	unsigned blockWidth, blockHeight;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
 
-static int radeon_compressed_num_bytes(GLuint mesaFormat)
+	return rowStride * ((height + blockHeight - 1) / blockHeight);
+}
+
+static int find_next_power_of_two(GLuint value)
 {
-   int bytes = 0;
-   switch(mesaFormat) {
-     
-   case MESA_FORMAT_RGB_FXT1:
-   case MESA_FORMAT_RGBA_FXT1:
-   case MESA_FORMAT_RGB_DXT1:
-   case MESA_FORMAT_RGBA_DXT1:
-     bytes = 2;
-     break;
-     
-   case MESA_FORMAT_RGBA_DXT3:
-   case MESA_FORMAT_RGBA_DXT5:
-     bytes = 4;
-   default:
-     break;
-   }
-   
-   return bytes;
+	int i, tmp;
+
+	i = 0;
+	tmp = value - 1;
+	while (tmp) {
+		tmp >>= 1;
+		i++;
+	}
+	return (1 << i);
 }
 
 /**
@@ -90,28 +93,28 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree
 {
 	radeon_mipmap_level *lvl = &mt->levels[level];
 	uint32_t row_align;
+	GLuint height;
+
+	height = find_next_power_of_two(lvl->height);
 
 	/* Find image size in bytes */
-	if (mt->compressed) {
-		/* TODO: Is this correct? Need test cases for compressed textures! */
-		row_align = rmesa->texture_compressed_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = radeon_compressed_texture_size(mt->radeon->glCtx,
-							   lvl->width, lvl->height, lvl->depth, mt->compressed);
+	if (_mesa_is_format_compressed(mt->mesaFormat)) {
+		lvl->rowstride = get_aligned_compressed_row_stride(mt->mesaFormat, lvl->width, rmesa->texture_compressed_row_align);
+		lvl->size = get_compressed_image_size(mt->mesaFormat, lvl->rowstride, height);
 	} else if (mt->target == GL_TEXTURE_RECTANGLE_NV) {
 		row_align = rmesa->texture_rect_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = lvl->rowstride * lvl->height;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
+		lvl->size = lvl->rowstride * height;
 	} else if (mt->tilebits & RADEON_TXO_MICRO_TILE) {
 		/* tile pattern is 16 bytes x2. mipmaps stay 32 byte aligned,
 		 * though the actual offset may be different (if texture is less than
 		 * 32 bytes width) to the untiled case */
-		lvl->rowstride = (lvl->width * mt->bpp * 2 + 31) & ~31;
-		lvl->size = lvl->rowstride * ((lvl->height + 1) / 2) * lvl->depth;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) * 2 + 31) & ~31;
+		lvl->size = lvl->rowstride * ((height + 1) / 2) * lvl->depth;
 	} else {
 		row_align = rmesa->texture_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = lvl->rowstride * lvl->height * lvl->depth;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
+		lvl->size = lvl->rowstride * height * lvl->depth;
 	}
 	assert(lvl->size > 0);
 
@@ -123,7 +126,7 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree
 	if (RADEON_DEBUG & RADEON_TEXTURE)
 	  fprintf(stderr,
 		  "level %d, face %d: rs:%d %dx%d at %d\n",
-		  level, face, lvl->rowstride, lvl->width, lvl->height, lvl->faces[face].offset);
+		  level, face, lvl->rowstride, lvl->width, height, lvl->faces[face].offset);
 }
 
 static GLuint minify(GLuint size, GLuint levels)
@@ -138,17 +141,15 @@ static GLuint minify(GLuint size, GLuint levels)
 static void calculate_miptree_layout_r100(radeonContextPtr rmesa, radeon_mipmap_tree *mt)
 {
 	GLuint curOffset;
-	GLuint numLevels;
 	GLuint i;
 	GLuint face;
 
-	numLevels = mt->lastLevel - mt->firstLevel + 1;
-	assert(numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
+	assert(mt->numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
 
 	curOffset = 0;
 	for(face = 0; face < mt->faces; face++) {
 
-		for(i = 0; i < numLevels; i++) {
+		for(i = 0; i < mt->numLevels; i++) {
 			mt->levels[i].width = minify(mt->width0, i);
 			mt->levels[i].height = minify(mt->height0, i);
 			mt->levels[i].depth = minify(mt->depth0, i);
@@ -163,14 +164,12 @@ static void calculate_miptree_layout_r100(radeonContextPtr rmesa, radeon_mipmap_
 static void calculate_miptree_layout_r300(radeonContextPtr rmesa, radeon_mipmap_tree *mt)
 {
 	GLuint curOffset;
-	GLuint numLevels;
 	GLuint i;
 
-	numLevels = mt->lastLevel - mt->firstLevel + 1;
-	assert(numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
+	assert(mt->numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
 
 	curOffset = 0;
-	for(i = 0; i < numLevels; i++) {
+	for(i = 0; i < mt->numLevels; i++) {
 		GLuint face;
 
 		mt->levels[i].width = minify(mt->width0, i);
@@ -188,27 +187,22 @@ static void calculate_miptree_layout_r300(radeonContextPtr rmesa, radeon_mipmap_
 /**
  * Create a new mipmap tree, calculate its layout and allocate memory.
  */
-radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *t,
-		GLenum target, GLenum internal_format, GLuint firstLevel, GLuint lastLevel,
-		GLuint width0, GLuint height0, GLuint depth0,
-		GLuint bpp, GLuint tilebits, GLuint compressed)
+static radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa,
+		GLenum target, gl_format mesaFormat, GLuint baseLevel, GLuint numLevels,
+		GLuint width0, GLuint height0, GLuint depth0, GLuint tilebits)
 {
 	radeon_mipmap_tree *mt = CALLOC_STRUCT(_radeon_mipmap_tree);
 
-	mt->radeon = rmesa;
-	mt->internal_format = internal_format;
+	mt->mesaFormat = mesaFormat;
 	mt->refcount = 1;
-	mt->t = t;
 	mt->target = target;
 	mt->faces = (target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-	mt->firstLevel = firstLevel;
-	mt->lastLevel = lastLevel;
+	mt->baseLevel = baseLevel;
+	mt->numLevels = numLevels;
 	mt->width0 = width0;
 	mt->height0 = height0;
 	mt->depth0 = depth0;
-	mt->bpp = compressed ? radeon_compressed_num_bytes(compressed) : bpp;
 	mt->tilebits = tilebits;
-	mt->compressed = compressed;
 
 	if (rmesa->radeonScreen->chip_family >= CHIP_FAMILY_R300)
 		calculate_miptree_layout_r300(rmesa, mt);
@@ -223,53 +217,43 @@ radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *
 	return mt;
 }
 
-void radeon_miptree_reference(radeon_mipmap_tree *mt)
+void radeon_miptree_reference(radeon_mipmap_tree *mt, radeon_mipmap_tree **ptr)
 {
+	assert(!*ptr);
+
 	mt->refcount++;
 	assert(mt->refcount > 0);
+
+	*ptr = mt;
 }
 
-void radeon_miptree_unreference(radeon_mipmap_tree *mt)
+void radeon_miptree_unreference(radeon_mipmap_tree **ptr)
 {
+	radeon_mipmap_tree *mt = *ptr;
 	if (!mt)
 		return;
 
 	assert(mt->refcount > 0);
+
 	mt->refcount--;
 	if (!mt->refcount) {
 		radeon_bo_unref(mt->bo);
 		free(mt);
 	}
-}
 
+	*ptr = 0;
+}
 
 /**
- * Calculate first and last mip levels for the given texture object,
- * where the dimensions are taken from the given texture image at
- * the given level.
- *
- * Note: level is the OpenGL level number, which is not necessarily the same
- * as the first level that is actually present.
- *
- * The base level image of the given texture face must be non-null,
- * or this will fail.
+ * Calculate min and max LOD for the given texture object.
+ * @param[in] tObj texture object whose LOD values to calculate
+ * @param[out] pminLod minimal LOD
+ * @param[out] pmaxLod maximal LOD
  */
-static void calculate_first_last_level(struct gl_texture_object *tObj,
-				       GLuint *pfirstLevel, GLuint *plastLevel,
-				       GLuint face, GLuint level)
+static void calculate_min_max_lod(struct gl_texture_object *tObj,
+				       unsigned *pminLod, unsigned *pmaxLod)
 {
-	const struct gl_texture_image * const baseImage =
-		tObj->Image[face][level];
-
-	assert(baseImage);
-	
-	/* These must be signed values.  MinLod and MaxLod can be negative numbers,
-	* and having firstLevel and lastLevel as signed prevents the need for
-	* extra sign checks.
-	*/
-	int   firstLevel;
-	int   lastLevel;
-
+	int minLod, maxLod;
 	/* Yes, this looks overly complicated, but it's all needed.
 	*/
 	switch (tObj->Target) {
@@ -280,55 +264,46 @@ static void calculate_first_last_level(struct gl_texture_object *tObj,
 		if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) {
 			/* GL_NEAREST and GL_LINEAR only care about GL_TEXTURE_BASE_LEVEL.
 			*/
-			firstLevel = lastLevel = tObj->BaseLevel;
+			minLod = maxLod = tObj->BaseLevel;
 		} else {
-			firstLevel = tObj->BaseLevel + (GLint)(tObj->MinLod + 0.5);
-			firstLevel = MAX2(firstLevel, tObj->BaseLevel);
-			firstLevel = MIN2(firstLevel, level + baseImage->MaxLog2);
-			lastLevel = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5);
-			lastLevel = MAX2(lastLevel, tObj->BaseLevel);
-			lastLevel = MIN2(lastLevel, level + baseImage->MaxLog2);
-			lastLevel = MIN2(lastLevel, tObj->MaxLevel);
-			lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
+			minLod = tObj->BaseLevel + (GLint)(tObj->MinLod);
+			minLod = MAX2(minLod, tObj->BaseLevel);
+			minLod = MIN2(minLod, tObj->MaxLevel);
+			maxLod = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5);
+			maxLod = MIN2(maxLod, tObj->MaxLevel);
+			maxLod = MIN2(maxLod, tObj->Image[0][minLod]->MaxLog2 + minLod);
+			maxLod = MAX2(maxLod, minLod); /* need at least one level */
 		}
 		break;
 	case GL_TEXTURE_RECTANGLE_NV:
 	case GL_TEXTURE_4D_SGIS:
-		firstLevel = lastLevel = 0;
+		minLod = maxLod = 0;
 		break;
 	default:
 		return;
 	}
 
 	/* save these values */
-	*pfirstLevel = firstLevel;
-	*plastLevel = lastLevel;
+	*pminLod = minLod;
+	*pmaxLod = maxLod;
 }
 
-
 /**
  * Checks whether the given miptree can hold the given texture image at the
  * given face and level.
  */
 GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
-		struct gl_texture_image *texImage, GLuint face, GLuint level)
+		struct gl_texture_image *texImage, GLuint face, GLuint mtLevel)
 {
-	GLboolean isCompressed = _mesa_is_format_compressed(texImage->TexFormat);
 	radeon_mipmap_level *lvl;
 
-	if (face >= mt->faces || level < mt->firstLevel || level > mt->lastLevel)
-		return GL_FALSE;
-
-	if (texImage->InternalFormat != mt->internal_format ||
-	    isCompressed != mt->compressed)
+	if (face >= mt->faces || mtLevel > mt->numLevels)
 		return GL_FALSE;
 
-	if (!isCompressed &&
-	    !mt->compressed &&
-	    _mesa_get_format_bytes(texImage->TexFormat) != mt->bpp)
+	if (texImage->TexFormat != mt->mesaFormat)
 		return GL_FALSE;
 
-	lvl = &mt->levels[level - mt->firstLevel];
+	lvl = &mt->levels[mtLevel];
 	if (lvl->width != texImage->Width ||
 	    lvl->height != texImage->Height ||
 	    lvl->depth != texImage->Depth)
@@ -337,64 +312,72 @@ GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
 	return GL_TRUE;
 }
 
-
 /**
  * Checks whether the given miptree has the right format to store the given texture object.
  */
-GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj)
+static GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj)
 {
 	struct gl_texture_image *firstImage;
-	GLuint compressed;
-	GLuint numfaces = 1;
-	GLuint firstLevel, lastLevel;
-	GLuint texelBytes;
-
-	calculate_first_last_level(texObj, &firstLevel, &lastLevel, 0, texObj->BaseLevel);
-	if (texObj->Target == GL_TEXTURE_CUBE_MAP)
-		numfaces = 6;
-
-	firstImage = texObj->Image[0][firstLevel];
-	compressed = _mesa_is_format_compressed(firstImage->TexFormat) ? firstImage->TexFormat : 0;
-	texelBytes = _mesa_get_format_bytes(firstImage->TexFormat);
-
-	return (mt->firstLevel == firstLevel &&
-	        mt->lastLevel == lastLevel &&
-	        mt->width0 == firstImage->Width &&
-	        mt->height0 == firstImage->Height &&
-	        mt->depth0 == firstImage->Depth &&
-	        mt->compressed == compressed &&
-	        (!mt->compressed ? (mt->bpp == texelBytes) : 1));
-}
+	unsigned numLevels;
+	radeon_mipmap_level *mtBaseLevel;
+
+	if (texObj->BaseLevel < mt->baseLevel)
+		return GL_FALSE;
+
+	mtBaseLevel = &mt->levels[texObj->BaseLevel - mt->baseLevel];
+	firstImage = texObj->Image[0][texObj->BaseLevel];
+	numLevels = MIN2(texObj->MaxLevel - texObj->BaseLevel + 1, firstImage->MaxLog2 + 1);
+
+	if (RADEON_DEBUG & RADEON_TEXTURE) {
+		fprintf(stderr, "Checking if miptree %p matches texObj %p\n", mt, texObj);
+		fprintf(stderr, "target %d vs %d\n", mt->target, texObj->Target);
+		fprintf(stderr, "format %d vs %d\n", mt->mesaFormat, firstImage->TexFormat);
+		fprintf(stderr, "numLevels %d vs %d\n", mt->numLevels, numLevels);
+		fprintf(stderr, "width0 %d vs %d\n", mtBaseLevel->width, firstImage->Width);
+		fprintf(stderr, "height0 %d vs %d\n", mtBaseLevel->height, firstImage->Height);
+		fprintf(stderr, "depth0 %d vs %d\n", mtBaseLevel->depth, firstImage->Depth);
+		if (mt->target == texObj->Target &&
+	        mt->mesaFormat == firstImage->TexFormat &&
+	        mt->numLevels >= numLevels &&
+	        mtBaseLevel->width == firstImage->Width &&
+	        mtBaseLevel->height == firstImage->Height &&
+	        mtBaseLevel->depth == firstImage->Depth) {
+			fprintf(stderr, "MATCHED\n");
+		} else {
+			fprintf(stderr, "NOT MATCHED\n");
+		}
+	}
 
+	return (mt->target == texObj->Target &&
+	        mt->mesaFormat == firstImage->TexFormat &&
+	        mt->numLevels >= numLevels &&
+	        mtBaseLevel->width == firstImage->Width &&
+	        mtBaseLevel->height == firstImage->Height &&
+	        mtBaseLevel->depth == firstImage->Depth);
+}
 
 /**
- * Try to allocate a mipmap tree for the given texture that will fit the
- * given image in the given position.
+ * Try to allocate a mipmap tree for the given texture object.
+ * @param[in] rmesa radeon context
+ * @param[in] t radeon texture object
  */
-void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
-		radeon_texture_image *image, GLuint face, GLuint level)
+void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t)
 {
-	GLuint compressed = _mesa_is_format_compressed(image->base.TexFormat) ? image->base.TexFormat : 0;
-	GLuint numfaces = 1;
-	GLuint firstLevel, lastLevel;
-	GLuint texelBytes;
+	struct gl_texture_object *texObj = &t->base;
+	struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel];
+	GLuint numLevels;
 
 	assert(!t->mt);
 
-	calculate_first_last_level(&t->base, &firstLevel, &lastLevel, face, level);
-	if (t->base.Target == GL_TEXTURE_CUBE_MAP)
-		numfaces = 6;
-
-	if (level != firstLevel || face >= numfaces)
+	if (!texImg)
 		return;
 
-	texelBytes = _mesa_get_format_bytes(image->base.TexFormat);
+	numLevels = MIN2(texObj->MaxLevel - texObj->BaseLevel + 1, texImg->MaxLog2 + 1);
 
-	t->mt = radeon_miptree_create(rmesa, t, t->base.Target,
-		image->base.InternalFormat,
-		firstLevel, lastLevel,
-		image->base.Width, image->base.Height, image->base.Depth,
-		texelBytes, t->tile_bits, compressed);
+	t->mt = radeon_miptree_create(rmesa, t->base.Target,
+		texImg->TexFormat, texObj->BaseLevel,
+		numLevels, texImg->Width, texImg->Height,
+		texImg->Depth, t->tile_bits);
 }
 
 /* Although we use the image_offset[] array to store relative offsets
@@ -406,21 +389,244 @@ void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
 void
 radeon_miptree_depth_offsets(radeon_mipmap_tree *mt, GLuint level, GLuint *offsets)
 {
-     if (mt->target != GL_TEXTURE_3D || mt->faces == 1)
-        offsets[0] = 0;
-     else {
-	int i;
-	for (i = 0; i < 6; i++)
-		offsets[i] = mt->levels[level].faces[i].offset;
-     }
+	if (mt->target != GL_TEXTURE_3D || mt->faces == 1) {
+		offsets[0] = 0;
+	} else {
+		int i;
+		for (i = 0; i < 6; i++) {
+			offsets[i] = mt->levels[level].faces[i].offset;
+		}
+	}
 }
 
 GLuint
 radeon_miptree_image_offset(radeon_mipmap_tree *mt,
 			    GLuint face, GLuint level)
 {
-   if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
-      return (mt->levels[level].faces[face].offset);
-   else
-      return mt->levels[level].faces[0].offset;
+	if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
+		return (mt->levels[level].faces[face].offset);
+	else
+		return mt->levels[level].faces[0].offset;
+}
+
+/**
+ * Convert radeon miptree texture level to GL texture level
+ * @param[in] tObj texture object whom level is to be converted
+ * @param[in] level radeon miptree texture level
+ * @return GL texture level
+ */
+unsigned radeon_miptree_level_to_gl_level(struct gl_texture_object *tObj, unsigned level)
+{
+	return level + tObj->BaseLevel;
+}
+
+/**
+ * Convert GL texture level to radeon miptree texture level
+ * @param[in] tObj texture object whom level is to be converted
+ * @param[in] level GL texture level
+ * @return radeon miptree texture level
+ */
+unsigned radeon_gl_level_to_miptree_level(struct gl_texture_object *tObj, unsigned level)
+{
+	return level - tObj->BaseLevel;
+}
+
+/**
+ * Ensure that the given image is stored in the given miptree from now on.
+ */
+static void migrate_image_to_miptree(radeon_mipmap_tree *mt,
+									 radeon_texture_image *image,
+									 int face, int mtLevel)
+{
+	radeon_mipmap_level *dstlvl = &mt->levels[mtLevel];
+	unsigned char *dest;
+
+	assert(image->mt != mt);
+	assert(dstlvl->width == image->base.Width);
+	assert(dstlvl->height == image->base.Height);
+	assert(dstlvl->depth == image->base.Depth);
+
+	radeon_bo_map(mt->bo, GL_TRUE);
+	dest = mt->bo->ptr + dstlvl->faces[face].offset;
+
+	if (image->mt) {
+		/* Format etc. should match, so we really just need a memcpy().
+		 * In fact, that memcpy() could be done by the hardware in many
+		 * cases, provided that we have a proper memory manager.
+		 */
+		assert(mt->mesaFormat == image->base.TexFormat);
+
+		radeon_mipmap_level *srclvl = &image->mt->levels[image->mtlevel];
+
+		assert(srclvl->size == dstlvl->size);
+		assert(srclvl->rowstride == dstlvl->rowstride);
+
+		radeon_bo_map(image->mt->bo, GL_FALSE);
+
+		memcpy(dest,
+			image->mt->bo->ptr + srclvl->faces[face].offset,
+			dstlvl->size);
+		radeon_bo_unmap(image->mt->bo);
+
+		radeon_miptree_unreference(&image->mt);
+	} else {
+		/* need to confirm this value is correct */
+		if (_mesa_is_format_compressed(image->base.TexFormat)) {
+			unsigned size = _mesa_format_image_size(image->base.TexFormat,
+													image->base.Width,
+													image->base.Height,
+													image->base.Depth);
+			memcpy(dest, image->base.Data, size);
+		} else {
+			uint32_t srcrowstride;
+			uint32_t height;
+
+			height = image->base.Height * image->base.Depth;
+			srcrowstride = image->base.Width * _mesa_get_format_bytes(image->base.TexFormat);
+			copy_rows(dest, dstlvl->rowstride, image->base.Data, srcrowstride,
+					height, srcrowstride);
+		}
+
+		_mesa_free_texmemory(image->base.Data);
+		image->base.Data = 0;
+	}
+
+	radeon_bo_unmap(mt->bo);
+
+	radeon_miptree_reference(mt, &image->mt);
+	image->mtface = face;
+	image->mtlevel = mtLevel;
+}
+
+/**
+ * Filter matching miptrees, and select one with the most of data.
+ * @param[in] texObj radeon texture object
+ * @param[in] firstLevel first texture level to check
+ * @param[in] lastLevel last texture level to check
+ */
+static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj,
+														 unsigned firstLevel,
+														 unsigned lastLevel)
+{
+	const unsigned numLevels = lastLevel - firstLevel + 1;
+	unsigned *mtSizes = calloc(numLevels, sizeof(unsigned));
+	radeon_mipmap_tree **mts = calloc(numLevels, sizeof(radeon_mipmap_tree *));
+	unsigned mtCount = 0;
+	unsigned maxMtIndex = 0;
+	radeon_mipmap_tree *tmp;
+
+	for (unsigned level = firstLevel; level <= lastLevel; ++level) {
+		radeon_texture_image *img = get_radeon_texture_image(texObj->base.Image[0][level]);
+		unsigned found = 0;
+		// TODO: why this hack??
+		if (!img)
+			break;
+
+		if (!img->mt || !radeon_miptree_matches_texture(img->mt, &texObj->base))
+			continue;
+
+		for (int i = 0; i < mtCount; ++i) {
+			if (mts[i] == img->mt) {
+				found = 1;
+				mtSizes[i] += img->mt->levels[img->mtlevel].size;
+				break;
+			}
+		}
+
+		if (!found) {
+			mtSizes[mtCount] += img->mt->levels[img->mtlevel].size;
+			mts[mtCount] = img->mt;
+			mtCount++;
+		}
+	}
+
+	if (mtCount == 0) {
+		return NULL;
+	}
+
+	for (int i = 1; i < mtCount; ++i) {
+		if (mtSizes[i] > mtSizes[maxMtIndex]) {
+			maxMtIndex = i;
+		}
+	}
+
+	tmp = mts[maxMtIndex];
+	free(mtSizes);
+	free(mts);
+
+	return tmp;
+}
+
+/**
+ * Validate texture mipmap tree.
+ * If individual images are stored in different mipmap trees
+ * use the mipmap tree that has the most of the correct data.
+ */
+int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *texObj)
+{
+	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
+	radeonTexObj *t = radeon_tex_obj(texObj);
+
+	if (t->validated || t->image_override) {
+		return GL_TRUE;
+	}
+
+	if (texObj->Image[0][texObj->BaseLevel]->Border > 0)
+		return GL_FALSE;
+
+	_mesa_test_texobj_completeness(rmesa->glCtx, texObj);
+	if (!texObj->_Complete) {
+		return GL_FALSE;
+	}
+
+	calculate_min_max_lod(&t->base, &t->minLod, &t->maxLod);
+
+	if (RADEON_DEBUG & RADEON_TEXTURE)
+		fprintf(stderr, "%s: Validating texture %p now, minLod = %d, maxLod = %d\n",
+				__FUNCTION__, texObj ,t->minLod, t->maxLod);
+
+	radeon_mipmap_tree *dst_miptree;
+	dst_miptree = get_biggest_matching_miptree(t, t->minLod, t->maxLod);
+
+	if (!dst_miptree) {
+		radeon_miptree_unreference(&t->mt);
+		radeon_try_alloc_miptree(rmesa, t);
+		dst_miptree = t->mt;
+	}
+
+	const unsigned faces = texObj->Target == GL_TEXTURE_CUBE_MAP ? 6 : 1;
+	unsigned face, level;
+	radeon_texture_image *img;
+	/* Validate only the levels that will actually be used during rendering */
+	for (face = 0; face < faces; ++face) {
+		for (level = t->minLod; level <= t->maxLod; ++level) {
+			img = get_radeon_texture_image(texObj->Image[face][level]);
+
+			if (RADEON_DEBUG & RADEON_TEXTURE) {
+				fprintf(stderr, "Checking image level %d, face %d, mt %p ... ", level, face, img->mt);
+			}
+			
+			if (img->mt != dst_miptree) {
+				if (RADEON_DEBUG & RADEON_TEXTURE) {
+					fprintf(stderr, "MIGRATING\n");
+				}
+				migrate_image_to_miptree(dst_miptree, img, face, radeon_gl_level_to_miptree_level(texObj, level));
+			} else if (RADEON_DEBUG & RADEON_TEXTURE) {
+				fprintf(stderr, "OK\n");
+			}
+		}
+	}
+
+	t->validated = GL_TRUE;
+
+	return GL_TRUE;
 }
+
+uint32_t get_base_teximage_offset(radeonTexObj *texObj)
+{
+	if (!texObj->mt) {
+		return 0;
+	} else {
+		return radeon_miptree_image_offset(texObj->mt, 0, texObj->minLod);
+	}
+}
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
index db28252da3..28b8485095 100644
--- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
+++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
@@ -59,43 +59,38 @@ struct _radeon_mipmap_level {
  * changed.
  */
 struct _radeon_mipmap_tree {
-	radeonContextPtr radeon;
-	radeonTexObj *t;
 	struct radeon_bo *bo;
 	GLuint refcount;
 
 	GLuint totalsize; /** total size of the miptree, in bytes */
 
 	GLenum target; /** GL_TEXTURE_xxx */
-	GLenum internal_format;
+	GLenum mesaFormat; /** MESA_FORMAT_xxx */
 	GLuint faces; /** # of faces: 6 for cubemaps, 1 otherwise */
-	GLuint firstLevel; /** First mip level stored in this mipmap tree */
-	GLuint lastLevel; /** Last mip level stored in this mipmap tree */
+	GLuint baseLevel; /** gl_texture_object->baseLevel it was created for */
+	GLuint numLevels; /** Number of mip levels stored in this mipmap tree */
 
 	GLuint width0; /** Width of firstLevel image */
 	GLuint height0; /** Height of firstLevel image */
 	GLuint depth0; /** Depth of firstLevel image */
 
-	GLuint bpp; /** Bytes per texel */
 	GLuint tilebits; /** RADEON_TXO_xxx_TILE */
-	GLuint compressed; /** MESA_FORMAT_xxx indicating a compressed format, or 0 if uncompressed */
 
 	radeon_mipmap_level levels[RADEON_MIPTREE_MAX_TEXTURE_LEVELS];
 };
 
-radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *t,
-		GLenum target, GLenum internal_format, GLuint firstLevel, GLuint lastLevel,
-		GLuint width0, GLuint height0, GLuint depth0,
-		GLuint bpp, GLuint tilebits, GLuint compressed);
-void radeon_miptree_reference(radeon_mipmap_tree *mt);
-void radeon_miptree_unreference(radeon_mipmap_tree *mt);
+void radeon_miptree_reference(radeon_mipmap_tree *mt, radeon_mipmap_tree **ptr);
+void radeon_miptree_unreference(radeon_mipmap_tree **ptr);
 
 GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
 		struct gl_texture_image *texImage, GLuint face, GLuint level);
-GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj);
-void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
-			      radeon_texture_image *texImage, GLuint face, GLuint level);
+void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t);
 GLuint radeon_miptree_image_offset(radeon_mipmap_tree *mt,
 				   GLuint face, GLuint level);
 void radeon_miptree_depth_offsets(radeon_mipmap_tree *mt, GLuint level, GLuint *offsets);
+
+unsigned radeon_miptree_level_to_gl_level(struct gl_texture_object *tObj, unsigned level);
+unsigned radeon_gl_level_to_miptree_level(struct gl_texture_object *tObj, unsigned level);
+
+uint32_t get_base_teximage_offset(radeonTexObj *texObj);
 #endif /* __RADEON_MIPMAP_TREE_H_ */
diff --git a/src/mesa/drivers/dri/radeon/radeon_queryobj.c b/src/mesa/drivers/dri/radeon/radeon_queryobj.c
index 6539c36268..98117cdfc1 100644
--- a/src/mesa/drivers/dri/radeon/radeon_queryobj.c
+++ b/src/mesa/drivers/dri/radeon/radeon_queryobj.c
@@ -31,24 +31,11 @@
 #include "main/imports.h"
 #include "main/simple_list.h"
 
-static int radeonQueryIsFlushed(GLcontext *ctx, struct gl_query_object *q)
-{
-	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
-	struct radeon_query_object *tmp, *query = (struct radeon_query_object *)q;
-
-	foreach(tmp, &radeon->query.not_flushed_head) {
-		if (tmp == query) {
-			return 0;
-		}
-	}
-
-	return 1;
-}
-
 static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 {
 	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
 	struct radeon_query_object *query = (struct radeon_query_object *)q;
+        uint32_t *result;
 	int i;
 
 	radeon_print(RADEON_STATE, RADEON_VERBOSE,
@@ -56,6 +43,7 @@ static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 			__FUNCTION__, query->Base.Id, (int) query->Base.Result);
 
 	radeon_bo_map(query->bo, GL_FALSE);
+        result = query->bo->ptr;
 
 	query->Base.Result = 0;
 	if (IS_R600_CLASS(radeon->radeonScreen)) {
@@ -66,10 +54,11 @@ static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 		 * hw writes zpass end counts to qwords 1, 3, 5, 7.
 		 * then we substract. MSB is the valid bit.
 		 */
-		uint64_t *result = query->bo->ptr;
-		for (i = 0; i < 8; i += 2) {
-			uint64_t start = result[i];
-			uint64_t end = result[i + 1];
+		for (i = 0; i < 16; i += 4) {
+			uint64_t start = (uint64_t)LE32_TO_CPU(result[i]) |
+					 (uint64_t)LE32_TO_CPU(result[i + 1]) << 32;
+			uint64_t end = (uint64_t)LE32_TO_CPU(result[i + 2]) |
+				       (uint64_t)LE32_TO_CPU(result[i + 3]) << 32;
 			if ((start & 0x8000000000000000) && (end & 0x8000000000000000)) {
 				uint64_t query_count = end - start;
 				query->Base.Result += query_count;
@@ -79,10 +68,9 @@ static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 				     "%d start: %lx, end: %lx %ld\n", i, start, end, end - start);
 		}
 	} else {
-		uint32_t *result = query->bo->ptr;
 		for (i = 0; i < query->curr_offset/sizeof(uint32_t); ++i) {
-			query->Base.Result += result[i];
-			radeon_print(RADEON_STATE, RADEON_TRACE, "result[%d] = %d\n", i, result[i]);
+			query->Base.Result += LE32_TO_CPU(result[i]);
+			radeon_print(RADEON_STATE, RADEON_TRACE, "result[%d] = %d\n", i, LE32_TO_CPU(result[i]));
 		}
 	}
 
@@ -120,10 +108,11 @@ static void radeonDeleteQuery(GLcontext *ctx, struct gl_query_object *q)
 
 static void radeonWaitQuery(GLcontext *ctx, struct gl_query_object *q)
 {
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
 	struct radeon_query_object *query = (struct radeon_query_object *)q;
 
 	/* If the cmdbuf with packets for this query hasn't been flushed yet, do it now */
-	if (!radeonQueryIsFlushed(ctx, q))
+	if (radeon_bo_is_referenced_by_cs(query->bo, radeon->cmdbuf.cs))
 		ctx->Driver.Flush(ctx);
 
 	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s: query id %d, bo %p, offset %d\n", __FUNCTION__, q->Id, query->bo, query->curr_offset);
@@ -155,8 +144,6 @@ static void radeonBeginQuery(GLcontext *ctx, struct gl_query_object *q)
 
 	radeon->query.queryobj.dirty = GL_TRUE;
 	radeon->hw.is_dirty = GL_TRUE;
-	insert_at_tail(&radeon->query.not_flushed_head, query);
-
 }
 
 void radeonEmitQueryEnd(GLcontext *ctx)
@@ -204,7 +191,7 @@ static void radeonCheckQuery(GLcontext *ctx, struct gl_query_object *q)
 		uint32_t domain;
 
 		/* Need to perform a flush, as per ARB_occlusion_query spec */
-		if (!radeonQueryIsFlushed(ctx, q)) {
+		if (radeon_bo_is_referenced_by_cs(query->bo, radeon->cmdbuf.cs)) {
 			ctx->Driver.Flush(ctx);
 		}
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 7a124a8be6..be2d8365ef 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -390,12 +390,14 @@ static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id)
    screen->device_id = device_id;
    screen->chip_flags = 0;
    switch ( device_id ) {
+   case PCI_CHIP_RN50_515E:
+   case PCI_CHIP_RN50_5969:
+	return -1;
+
    case PCI_CHIP_RADEON_LY:
    case PCI_CHIP_RADEON_LZ:
    case PCI_CHIP_RADEON_QY:
    case PCI_CHIP_RADEON_QZ:
-   case PCI_CHIP_RN50_515E:
-   case PCI_CHIP_RN50_5969:
       screen->chip_family = CHIP_FAMILY_RV100;
       break;
 
@@ -1482,11 +1484,11 @@ radeonCreateBuffer( __DRIscreenPrivate *driScrnPriv,
     _mesa_initialize_framebuffer(&rfb->base, mesaVis);
 
     if (mesaVis->redBits == 5)
-        rgbFormat = MESA_FORMAT_RGB565;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_RGB565 : MESA_FORMAT_RGB565_REV;
     else if (mesaVis->alphaBits == 0)
-        rgbFormat = MESA_FORMAT_XRGB8888;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_XRGB8888 : MESA_FORMAT_XRGB8888_REV;
     else
-        rgbFormat = MESA_FORMAT_ARGB8888;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB8888_REV;
 
     /* front color renderbuffer */
     rfb->color_rb[0] = radeon_create_renderbuffer(rgbFormat, driDrawPriv);
diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c
index 2bc7d31254..37904dc8dc 100644
--- a/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -41,6 +41,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
 #include "main/glheader.h"
+#include "main/texformat.h"
 #include "swrast/swrast.h"
 
 #include "radeon_common.h"
@@ -400,6 +401,18 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5_REV
+
+#define TAG(x)    radeon##x##_RGB565_REV
+#define TAG2(x,y) radeon##x##_RGB565_REV##y
+#if defined(RADEON_R600)
+#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
+#else
+#define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
+#endif
+#include "spantmp2.h"
+
 /* 16 bit, ARGB1555 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -414,6 +427,18 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5
+
+#define TAG(x)    radeon##x##_ARGB1555_REV
+#define TAG2(x,y) radeon##x##_ARGB1555_REV##y
+#if defined(RADEON_R600)
+#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
+#else
+#define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
+#endif
+#include "spantmp2.h"
+
 /* 16 bit, RGBA4 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -428,6 +453,18 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4
+
+#define TAG(x)    radeon##x##_ARGB4444_REV
+#define TAG2(x,y) radeon##x##_ARGB4444_REV##y
+#if defined(RADEON_R600)
+#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
+#else
+#define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off)
+#endif
+#include "spantmp2.h"
+
 /* 32 bit, xRGB8888 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -472,6 +509,42 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 #endif
 #include "spantmp2.h"
 
+/* 32 bit, BGRx8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8
+
+#define TAG(x)    radeon##x##_BGRx8888
+#define TAG2(x,y) radeon##x##_BGRx8888##y
+#if defined(RADEON_R600)
+#define GET_VALUE(_x, _y) ((*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)) | 0x000000ff))
+#define PUT_VALUE(_x, _y, d) { \
+   GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off );		\
+   *_ptr = d;								\
+} while (0)
+#else
+#define GET_VALUE(_x, _y) ((*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) | 0x000000ff))
+#define PUT_VALUE(_x, _y, d) { \
+   GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
+   *_ptr = d;								\
+} while (0)
+#endif
+#include "spantmp2.h"
+
+/* 32 bit, BGRA8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8
+
+#define TAG(x)    radeon##x##_BGRA8888
+#define TAG2(x,y) radeon##x##_BGRA8888##y
+#if defined(RADEON_R600)
+#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off)
+#else
+#define GET_PTR(X,Y) radeon_ptr_4byte(rrb, (X) + x_off, (Y) + y_off)
+#endif
+#include "spantmp2.h"
+
 /* ================================================================
  * Depth buffer
  */
@@ -526,10 +599,10 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x000000ff;							\
    tmp |= ((d << 8) & 0xffffff00);					\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
@@ -544,26 +617,26 @@ do {									\
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xff000000;							\
    tmp |= ((d) & 0x00ffffff);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #else
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );	\
-   GLuint tmp = *_ptr;							\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xff000000;							\
    tmp |= ((d) & 0x00ffffff);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #endif
 
 #if defined(RADEON_R300)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
-    d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0xffffff00) >> 8; \
+    d = (LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))) & 0xffffff00) >> 8; \
   }while(0)
 #elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
@@ -573,11 +646,11 @@ do {									\
 #elif defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
-    d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \
+    d = LE32_TO_CPU(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))) & 0x00ffffff; \
   }while(0)
 #else
 #define READ_DEPTH( d, _x, _y )	\
-  d = *(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off)) & 0x00ffffff;
+  d = LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off))) & 0x00ffffff;
 #endif
 
 #define TAG(x) radeon##x##_z24
@@ -595,7 +668,7 @@ do {									\
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   *_ptr = d;								\
+   *_ptr = CPU_TO_LE32((((d) & 0xff000000) >> 24) | (((d) & 0x00ffffff) << 8));   \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
@@ -615,20 +688,21 @@ do {									\
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   *_ptr = d;								\
+   *_ptr = CPU_TO_LE32(d);						\
 } while (0)
 #else
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );	\
-   *_ptr = d;					\
+   *_ptr = CPU_TO_LE32(d);						\
 } while (0)
 #endif
 
 #if defined(RADEON_R300)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
-    d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)));	\
+    GLuint tmp = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)));	\
+    d = LE32_TO_CPU(((tmp & 0x000000ff) << 24) | ((tmp & 0xffffff00) >> 8));	\
   }while(0)
 #elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
@@ -639,11 +713,11 @@ do {									\
 #elif defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
-    d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off));	\
+    d = LE32_TO_CPU(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))); \
   }while(0)
 #else
 #define READ_DEPTH( d, _x, _y )	do {					\
-    d = *(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off )); \
+    d = LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))); \
   } while (0)
 #endif
 
@@ -660,10 +734,10 @@ do {									\
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xffffff00;							\
    tmp |= (d) & 0xff;							\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_STENCIL( _x, _y, d )					\
@@ -678,19 +752,19 @@ do {									\
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x00ffffff;							\
    tmp |= (((d) & 0xff) << 24);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #else
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x00ffffff;							\
    tmp |= (((d) & 0xff) << 24);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #endif
 
@@ -698,7 +772,7 @@ do {									\
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = tmp & 0x000000ff;						\
 } while (0)
 #elif defined(RADEON_R600)
@@ -712,14 +786,14 @@ do {									\
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = (tmp & 0xff000000) >> 24;					\
 } while (0)
 #else
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = (tmp & 0xff000000) >> 24;					\
 } while (0)
 #endif
@@ -848,14 +922,24 @@ static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb)
 {
 	if (rrb->base.Format == MESA_FORMAT_RGB565) {
 		radeonInitPointers_RGB565(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_RGB565_REV) {
+		radeonInitPointers_RGB565_REV(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_XRGB8888) {
 		radeonInitPointers_xRGB8888(&rrb->base);
+        } else if (rrb->base.Format == MESA_FORMAT_XRGB8888_REV) {
+		radeonInitPointers_BGRx8888(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_ARGB8888) {
 		radeonInitPointers_ARGB8888(&rrb->base);
+        } else if (rrb->base.Format == MESA_FORMAT_ARGB8888_REV) {
+		radeonInitPointers_BGRA8888(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_ARGB4444) {
 		radeonInitPointers_ARGB4444(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB4444_REV) {
+		radeonInitPointers_ARGB4444_REV(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_ARGB1555) {
 		radeonInitPointers_ARGB1555(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB1555_REV) {
+		radeonInitPointers_ARGB1555_REV(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_Z16) {
 		radeonInitDepthPointers_z16(&rrb->base);
 	} else if (rrb->base.Format == MESA_FORMAT_X8_Z24) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_state_init.c b/src/mesa/drivers/dri/radeon/radeon_state_init.c
index 2d19220d8a..dd82888254 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -645,11 +645,11 @@ static void tex_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom)
      OUT_BATCH(CP_PACKET0(RADEON_PP_TXOFFSET_0 + (24 * i), 0));
      if (t->mt && !t->image_override) {
         if ((ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_CUBE_BIT)) {
-            lvl = &t->mt->levels[0];
+            lvl = &t->mt->levels[t->minLod];
 	    OUT_BATCH_RELOC(lvl->faces[5].offset, t->mt->bo, lvl->faces[5].offset,
 			RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
         } else {
-           OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+           OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 		     RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
         }
       } else {
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index e61f59eaea..8bf1bfbc57 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -179,7 +179,7 @@ static void radeonSetVertexFormat( GLcontext *ctx )
 
       for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    GLuint sz = VB->TexCoordPtr[i]->size;
+	    GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
 	    switch (sz) {
 	    case 1:
@@ -309,7 +309,7 @@ void r100_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
    radeonEmitState(&rmesa->radeon);
    radeonEmitVertexAOS( rmesa,
 			rmesa->radeon.swtcl.vertex_size,
-			first_elem(&rmesa->radeon.dma.reserved)->bo,
+			rmesa->radeon.swtcl.bo,
 			current_offset);
 
 		      
diff --git a/src/mesa/drivers/dri/radeon/radeon_tex.c b/src/mesa/drivers/dri/radeon/radeon_tex.c
index 60981aada2..749ab75f20 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tex.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tex.c
@@ -348,17 +348,7 @@ static void radeonTexParameter( GLcontext *ctx, GLenum target,
    case GL_TEXTURE_MAX_LEVEL:
    case GL_TEXTURE_MIN_LOD:
    case GL_TEXTURE_MAX_LOD:
-
-      /* This isn't the most efficient solution but there doesn't appear to
-       * be a nice alternative.  Since there's no LOD clamping,
-       * we just have to rely on loading the right subset of mipmap levels
-       * to simulate a clamped LOD.
-       */
-      if (t->mt) {
-         radeon_miptree_unreference(t->mt);
-	 t->mt = 0;
-	 t->validated = GL_FALSE;
-      }
+      t->validated = GL_FALSE;
       break;
 
    default:
@@ -388,10 +378,8 @@ static void radeonDeleteTexture( GLcontext *ctx,
      }
    }
 
-   if (t->mt) {
-      radeon_miptree_unreference(t->mt);
-      t->mt = 0;
-   }
+   radeon_miptree_unreference(&t->mt);
+
    /* Free mipmap images and the texture object itself */
    _mesa_delete_texture_object(ctx, texObj);
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index 429977a8bc..3cbe3b4725 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -699,14 +699,10 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
@@ -718,8 +714,6 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 	t->tile_bits = 0;
 	t->image_override = GL_TRUE;
 	t->override_offset = 0;
-	t->pp_txpitch &= (1 << 13) -1;
-	pitch_val = rb->pitch;
 	switch (rb->cpp) {
 	case 4:
 		if (glx_texture_format == GLX_TEXTURE_FORMAT_RGB_EXT)
@@ -738,12 +732,17 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 		t->pp_txfilter |= tx_table[MESA_FORMAT_RGB565].filter;
 		break;
 	}
-        t->pp_txsize = ((rb->base.Width - 1) << RADEON_TEX_USIZE_SHIFT)
-		   | ((rb->base.Height - 1) << RADEON_TEX_VSIZE_SHIFT);
-        t->pp_txformat |= RADEON_TXFORMAT_NON_POWER2;
-	t->pp_txpitch = pitch_val;
-        t->pp_txpitch -= 32;
 
+	t->pp_txpitch &= (1 << 13) -1;
+	pitch_val = rb->pitch;
+
+        t->pp_txsize = ((rb->base.Width - 1) << RADEON_TEX_USIZE_SHIFT)
+		| ((rb->base.Height - 1) << RADEON_TEX_VSIZE_SHIFT);
+	if (target == GL_TEXTURE_RECTANGLE_NV) {
+		t->pp_txformat |= RADEON_TXFORMAT_NON_POWER2;
+		t->pp_txpitch = pitch_val;
+		t->pp_txpitch -= 32;
+	}
 	t->validated = GL_TRUE;
 	_mesa_unlock_texture(radeon->glCtx, texObj);
 	return;
@@ -1021,7 +1020,7 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 	return GL_TRUE;
    }
 
-   firstImage = t->base.Image[0][t->mt->firstLevel];   
+   firstImage = t->base.Image[0][t->minLod];
 
    if (firstImage->Border > 0) {
       fprintf(stderr, "%s: border\n", __FUNCTION__);
@@ -1049,9 +1048,9 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 	 return GL_FALSE;
       }
    }
-   
+
    t->pp_txfilter &= ~RADEON_MAX_MIP_LEVEL_MASK;
-   t->pp_txfilter |= (t->mt->lastLevel - t->mt->firstLevel) << RADEON_MAX_MIP_LEVEL_SHIFT;
+   t->pp_txfilter |= (t->maxLod - t->minLod) << RADEON_MAX_MIP_LEVEL_SHIFT;
 	
    t->pp_txformat &= ~(RADEON_TXFORMAT_WIDTH_MASK |
 		       RADEON_TXFORMAT_HEIGHT_MASK |
@@ -1060,9 +1059,9 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 		       RADEON_TXFORMAT_F5_HEIGHT_MASK);
    t->pp_txformat |= ((log2Width << RADEON_TXFORMAT_WIDTH_SHIFT) |
 		      (log2Height << RADEON_TXFORMAT_HEIGHT_SHIFT));
-   
+
    t->tile_bits = 0;
-   
+
    if (t->base.Target == GL_TEXTURE_CUBE_MAP) {
       ASSERT(log2Width == log2Height);
       t->pp_txformat |= ((log2Width << RADEON_TXFORMAT_F5_WIDTH_SHIFT) |
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c
index baa99b752b..1ee9e2792a 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.c
@@ -1,4 +1,5 @@
 /*
+ * Copyright (C) 2009 Maciej Cencora.
  * Copyright (C) 2008 Nicolai Haehnle.
  * Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
  *
@@ -46,7 +47,7 @@
 #include "radeon_mipmap_tree.h"
 
 
-static void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
+void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
 	GLuint numrows, GLuint rowsize)
 {
 	assert(rowsize <= dststride);
@@ -81,8 +82,7 @@ void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage)
 	radeon_texture_image* image = get_radeon_texture_image(timage);
 
 	if (image->mt) {
-		radeon_miptree_unreference(image->mt);
-		image->mt = 0;
+		radeon_miptree_unreference(&image->mt);
 		assert(!image->base.Data);
 	} else {
 		_mesa_free_texture_image_data(ctx, timage);
@@ -108,7 +108,7 @@ static void teximage_set_map_data(radeon_texture_image *image)
 	lvl = &image->mt->levels[image->mtlevel];
 
 	image->base.Data = image->mt->bo->ptr + lvl->faces[image->mtface].offset;
-	image->base.RowStride = lvl->rowstride / image->mt->bpp;
+	image->base.RowStride = lvl->rowstride / _mesa_get_format_bytes(image->base.TexFormat);
 }
 
 
@@ -174,7 +174,7 @@ void radeonMapTexture(GLcontext *ctx, struct gl_texture_object *texObj)
 
 	radeon_bo_map(t->mt->bo, GL_FALSE);
 	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level)
+		for(level = t->minLod; level <= t->maxLod; ++level)
 			teximage_set_map_data(get_radeon_texture_image(texObj->Image[face][level]));
 	}
 }
@@ -191,7 +191,7 @@ void radeonUnmapTexture(GLcontext *ctx, struct gl_texture_object *texObj)
 	  return;
 
 	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level)
+		for(level = t->minLod; level <= t->maxLod; ++level)
 			texObj->Image[face][level]->Data = 0;
 	}
 	radeon_bo_unmap(t->mt->bo);
@@ -240,8 +240,7 @@ static void radeon_generate_mipmap(GLcontext *ctx, GLenum target,
 			image->mtlevel = i;
 			image->mtface = face;
 
-			radeon_miptree_unreference(image->mt);
-			image->mt = NULL;
+			radeon_miptree_unreference(&image->mt);
 		}
 	}
 	
@@ -510,6 +509,137 @@ gl_format radeonChooseTextureFormat(GLcontext * ctx,
 	return MESA_FORMAT_NONE;		/* never get here */
 }
 
+static void teximage_assign_miptree(radeonContextPtr rmesa,
+	struct gl_texture_object *texObj,
+	struct gl_texture_image *texImage,
+	unsigned face,
+	unsigned level)
+{
+	radeonTexObj *t = radeon_tex_obj(texObj);
+	radeon_texture_image* image = get_radeon_texture_image(texImage);
+
+	/* Try using current miptree, or create new if there isn't any */
+	if (!t->mt || !radeon_miptree_matches_image(t->mt, texImage, face,
+					radeon_gl_level_to_miptree_level(texObj, level))) {
+		radeon_miptree_unreference(&t->mt);
+		radeon_try_alloc_miptree(rmesa, t);
+	}
+
+	/* Miptree alocation may have failed,
+	 * when there was no image for baselevel specified */
+	if (t->mt) {
+		image->mtface = face;
+		image->mtlevel = radeon_gl_level_to_miptree_level(texObj, level);
+		radeon_miptree_reference(t->mt, &image->mt);
+	}
+}
+
+static GLuint * allocate_image_offsets(GLcontext *ctx,
+	unsigned alignedWidth,
+	unsigned height,
+	unsigned depth)
+{
+	int i;
+	GLuint *offsets;
+
+	offsets = _mesa_malloc(depth * sizeof(GLuint)) ;
+	if (!offsets) {
+		_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTex[Sub]Image");
+		return NULL;
+	}
+
+	for (i = 0; i < depth; ++i) {
+		offsets[i] = alignedWidth * height * i;
+	}
+
+	return offsets;
+}
+
+/**
+ * Update a subregion of the given texture image.
+ */
+static void radeon_store_teximage(GLcontext* ctx, int dims,
+		GLint xoffset, GLint yoffset, GLint zoffset,
+		GLsizei width, GLsizei height, GLsizei depth,
+		GLsizei imageSize,
+		GLenum format, GLenum type,
+		const GLvoid * pixels,
+		const struct gl_pixelstore_attrib *packing,
+		struct gl_texture_object *texObj,
+		struct gl_texture_image *texImage,
+		int compressed)
+{
+	radeonTexObj *t = radeon_tex_obj(texObj);
+	radeon_texture_image* image = get_radeon_texture_image(texImage);
+
+	GLuint dstRowStride;
+	GLuint *dstImageOffsets;
+
+	if (image->mt) {
+		dstRowStride = image->mt->levels[image->mtlevel].rowstride;
+	} else if (t->bo) {
+		/* TFP case */
+		/* TODO */
+		assert(0);
+	} else {
+		dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
+	}
+
+	if (dims == 3) {
+		unsigned alignedWidth = dstRowStride/_mesa_get_format_bytes(texImage->TexFormat);
+		dstImageOffsets = allocate_image_offsets(ctx, alignedWidth, texImage->Height, texImage->Depth);
+		if (!dstImageOffsets) {
+			return;
+		}
+	} else {
+		dstImageOffsets = texImage->ImageOffsets;
+	}
+
+	radeon_teximage_map(image, GL_TRUE);
+
+	if (compressed) {
+		uint32_t srcRowStride, bytesPerRow, rows, block_width, block_height;
+		GLubyte *img_start;
+
+		_mesa_get_format_block_size(texImage->TexFormat, &block_width, &block_height);
+
+		if (!image->mt) {
+			dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
+			img_start = _mesa_compressed_image_address(xoffset, yoffset, 0,
+									texImage->TexFormat,
+									texImage->Width, texImage->Data);
+		}
+		else {
+			uint32_t offset;
+			offset = dstRowStride / _mesa_get_format_bytes(texImage->TexFormat) * yoffset / block_height + xoffset / block_width;
+			offset *= _mesa_get_format_bytes(texImage->TexFormat);
+			img_start = texImage->Data + offset;
+		}
+		srcRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
+		bytesPerRow = srcRowStride;
+		rows = (height + block_height - 1) / block_height;
+
+		copy_rows(img_start, dstRowStride, pixels, srcRowStride, rows, bytesPerRow);
+	}
+	else {
+		if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
+					texImage->TexFormat, texImage->Data,
+					xoffset, yoffset, zoffset,
+					dstRowStride,
+					dstImageOffsets,
+					width, height, depth,
+					format, type, pixels, packing)) {
+			_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
+		}
+	}
+
+	if (dims == 3) {
+		_mesa_free(dstImageOffsets);
+	}
+
+	radeon_teximage_unmap(image);
+}
+
 /**
  * All glTexImage calls go through this function.
  */
@@ -528,13 +658,17 @@ static void radeon_teximage(
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	radeonTexObj* t = radeon_tex_obj(texObj);
 	radeon_texture_image* image = get_radeon_texture_image(texImage);
-	GLuint dstRowStride;
 	GLint postConvWidth = width;
 	GLint postConvHeight = height;
-	GLuint texelBytes;
 	GLuint face = radeon_face_for_target(target);
 
-	radeon_firevertices(rmesa);
+	{
+		struct radeon_bo *bo;
+		bo = !image->mt ? image->bo : image->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->cmdbuf.cs)) {
+			radeon_firevertices(rmesa);
+		}
+	}
 
 	t->validated = GL_FALSE;
 
@@ -543,53 +677,30 @@ static void radeon_teximage(
 						  &postConvHeight);
 	}
 
-	if (_mesa_is_format_compressed(texImage->TexFormat)) {
-		texelBytes = 0;
-	} else {
-		texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
+	if (!_mesa_is_format_compressed(texImage->TexFormat)) {
+		GLuint texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 		/* Minimum pitch of 32 bytes */
 		if (postConvWidth * texelBytes < 32) {
-		  postConvWidth = 32 / texelBytes;
-		  texImage->RowStride = postConvWidth;
+			postConvWidth = 32 / texelBytes;
+			texImage->RowStride = postConvWidth;
 		}
-		if (!image->mt) {      
+		if (!image->mt) {
 			assert(texImage->RowStride == postConvWidth);
 		}
 	}
 
-	/* Allocate memory for image */
-	radeonFreeTexImageData(ctx, texImage); /* Mesa core only clears texImage->Data but not image->mt */
+	/* Mesa core only clears texImage->Data but not image->mt */
+	radeonFreeTexImageData(ctx, texImage);
 
-	if (t->mt &&
-	    t->mt->firstLevel == level &&
-	    t->mt->lastLevel == level &&
-	    t->mt->target != GL_TEXTURE_CUBE_MAP_ARB &&
-	    !radeon_miptree_matches_image(t->mt, texImage, face, level)) {
-	  radeon_miptree_unreference(t->mt);
-	  t->mt = NULL;
-	}
-
-	if (!t->mt)
-		radeon_try_alloc_miptree(rmesa, t, image, face, level);
-	if (t->mt && radeon_miptree_matches_image(t->mt, texImage, face, level)) {
-		radeon_mipmap_level *lvl;
-		image->mt = t->mt;
-		image->mtlevel = level - t->mt->firstLevel;
-		image->mtface = face;
-		radeon_miptree_reference(t->mt);
-		lvl = &image->mt->levels[image->mtlevel];
-		dstRowStride = lvl->rowstride;
-	} else {
-		int size;
-		if (_mesa_is_format_compressed(texImage->TexFormat)) {
-			size = _mesa_format_image_size(texImage->TexFormat,
-						       texImage->Width,
-						       texImage->Height,
-						       texImage->Depth);
-		} else {
-			size = texImage->Width * texImage->Height * texImage->Depth * _mesa_get_format_bytes(texImage->TexFormat);
+	if (!t->bo) {
+		teximage_assign_miptree(rmesa, texObj, texImage, face, level);
+		if (!t->mt) {
+			int size = _mesa_format_image_size(texImage->TexFormat,
+								texImage->Width,
+								texImage->Height,
+								texImage->Depth);
+			texImage->Data = _mesa_alloc_texmemory(size);
 		}
-		texImage->Data = _mesa_alloc_texmemory(size);
 	}
 
 	/* Upload texture image; note that the spec allows pixels to be NULL */
@@ -603,65 +714,16 @@ static void radeon_teximage(
 	}
 
 	if (pixels) {
-		radeon_teximage_map(image, GL_TRUE);
-		if (compressed) {
-			if (image->mt) {
-				uint32_t srcRowStride, bytesPerRow, rows;
-				srcRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
-				bytesPerRow = srcRowStride;
-				rows = (height + 3) / 4;
-				copy_rows(texImage->Data, image->mt->levels[level].rowstride,
-					  pixels, srcRowStride, rows, bytesPerRow);
-			} else {
-				memcpy(texImage->Data, pixels, imageSize);
-			}
-		} else {
-			GLuint dstRowStride;
-			GLuint *dstImageOffsets;
-
-			if (image->mt) {
-				radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel];
-				dstRowStride = lvl->rowstride;
-			} else {
-				dstRowStride = texImage->Width * _mesa_get_format_bytes(texImage->TexFormat);
-			}
-
-			if (dims == 3) {
-				int i;
-
-				dstImageOffsets = _mesa_malloc(depth * sizeof(GLuint)) ;
-				if (!dstImageOffsets)
-					_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
-
-				for (i = 0; i < depth; ++i) {
-					dstImageOffsets[i] = dstRowStride/_mesa_get_format_bytes(texImage->TexFormat) * height * i;
-				}
-			} else {
-				dstImageOffsets = texImage->ImageOffsets;
-			}
-
-			if (!_mesa_texstore(ctx, dims,
-					    texImage->_BaseFormat,
-					    texImage->TexFormat,
-					    texImage->Data, 0, 0, 0, /* dstX/Y/Zoffset */
-					    dstRowStride,
-					    dstImageOffsets,
-					    width, height, depth,
-					    format, type, pixels, packing)) {
-				_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
-			}
-
-			if (dims == 3)
-				_mesa_free(dstImageOffsets);
-		}
+		radeon_store_teximage(ctx, dims,
+			0, 0, 0,
+			width, height, depth,
+			imageSize, format, type,
+			pixels, packing,
+			texObj, texImage,
+			compressed);
 	}
 
 	_mesa_unmap_teximage_pbo(ctx, packing);
-
-	if (pixels)
-	  radeon_teximage_unmap(image);
-
-
 }
 
 void radeonTexImage1D(GLcontext * ctx, GLenum target, GLint level,
@@ -714,7 +776,7 @@ void radeonTexImage3D(GLcontext * ctx, GLenum target, GLint level,
 }
 
 /**
- * Update a subregion of the given texture image.
+ * All glTexSubImage calls go through this function.
  */
 static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int level,
 		GLint xoffset, GLint yoffset, GLint zoffset,
@@ -731,66 +793,34 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve
 	radeonTexObj* t = radeon_tex_obj(texObj);
 	radeon_texture_image* image = get_radeon_texture_image(texImage);
 
-	radeon_firevertices(rmesa);
+	{
+		struct radeon_bo *bo;
+		bo = !image->mt ? image->bo : image->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->cmdbuf.cs)) {
+			radeon_firevertices(rmesa);
+		}
+	}
 
 	t->validated = GL_FALSE;
 	if (compressed) {
 		pixels = _mesa_validate_pbo_compressed_teximage(
-			ctx, imageSize, pixels, packing, "glCompressedTexImage");
+			ctx, imageSize, pixels, packing, "glCompressedTexSubImage");
 	} else {
 		pixels = _mesa_validate_pbo_teximage(ctx, dims,
-			width, height, depth, format, type, pixels, packing, "glTexSubImage1D");
+			width, height, depth, format, type, pixels, packing, "glTexSubImage");
 	}
 
 	if (pixels) {
-		GLint dstRowStride;
-		radeon_teximage_map(image, GL_TRUE);
-
-		if (image->mt) {
-			radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel];
-			dstRowStride = lvl->rowstride;
-		} else {
-			dstRowStride = texImage->RowStride * _mesa_get_format_bytes(texImage->TexFormat);
-		}
-
-		if (compressed) {
-			uint32_t srcRowStride, bytesPerRow, rows;
-			GLubyte *img_start;
-			if (!image->mt) {
-				dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
-				img_start = _mesa_compressed_image_address(xoffset, yoffset, 0,
-									   texImage->TexFormat,
-									   texImage->Width, texImage->Data);
-			}
-			else {
-				uint32_t blocks_x = dstRowStride / (image->mt->bpp * 4);
-				img_start = texImage->Data + image->mt->bpp * 4 * (blocks_x * (yoffset / 4) + xoffset / 4);
-			}
-			srcRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
-			bytesPerRow = srcRowStride;
-			rows = (height + 3) / 4;
-
-			copy_rows(img_start, dstRowStride,  pixels, srcRowStride, rows,  bytesPerRow);
-			
-		}
-		else {
-			if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
-					    texImage->TexFormat, texImage->Data,
-					    xoffset, yoffset, zoffset,
-					    dstRowStride,
-					    texImage->ImageOffsets,
-					    width, height, depth,
-					    format, type, pixels, packing)) {
-				_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
-			}
-		}
+		radeon_store_teximage(ctx, dims,
+			xoffset, yoffset, zoffset,
+			width, height, depth,
+			imageSize, format, type,
+			pixels, packing,
+			texObj, texImage,
+			compressed);
 	}
 
-	radeon_teximage_unmap(image);
-
 	_mesa_unmap_teximage_pbo(ctx, packing);
-
-
 }
 
 void radeonTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
@@ -846,143 +876,6 @@ void radeonTexSubImage3D(GLcontext * ctx, GLenum target, GLint level,
 		format, type, pixels, packing, texObj, texImage, 0);
 }
 
-
-
-/**
- * Ensure that the given image is stored in the given miptree from now on.
- */
-static void migrate_image_to_miptree(radeon_mipmap_tree *mt, radeon_texture_image *image, int face, int level)
-{
-	radeon_mipmap_level *dstlvl = &mt->levels[level - mt->firstLevel];
-	unsigned char *dest;
-
-	assert(image->mt != mt);
-	assert(dstlvl->width == image->base.Width);
-	assert(dstlvl->height == image->base.Height);
-	assert(dstlvl->depth == image->base.Depth);
-
-
-	radeon_bo_map(mt->bo, GL_TRUE);
-	dest = mt->bo->ptr + dstlvl->faces[face].offset;
-
-	if (image->mt) {
-		/* Format etc. should match, so we really just need a memcpy().
-		 * In fact, that memcpy() could be done by the hardware in many
-		 * cases, provided that we have a proper memory manager.
-		 */
-		radeon_mipmap_level *srclvl = &image->mt->levels[image->mtlevel-image->mt->firstLevel];
-
-		assert(srclvl->size == dstlvl->size);
-		assert(srclvl->rowstride == dstlvl->rowstride);
-
-		radeon_bo_map(image->mt->bo, GL_FALSE);
-
-		memcpy(dest,
-			image->mt->bo->ptr + srclvl->faces[face].offset,
-			dstlvl->size);
-		radeon_bo_unmap(image->mt->bo);
-
-		radeon_miptree_unreference(image->mt);
-	} else {
-		uint32_t srcrowstride;
-		uint32_t height;
-		/* need to confirm this value is correct */
-		if (mt->compressed) {
-			height = (image->base.Height + 3) / 4;
-			srcrowstride = _mesa_format_row_stride(image->base.TexFormat, image->base.Width);
-		} else {
-			height = image->base.Height * image->base.Depth;
-			srcrowstride = image->base.Width * _mesa_get_format_bytes(image->base.TexFormat);
-		}
-
-//		if (mt->tilebits)
-//			WARN_ONCE("%s: tiling not supported yet", __FUNCTION__);
-
-		copy_rows(dest, dstlvl->rowstride, image->base.Data, srcrowstride,
-			  height, srcrowstride);
-
-		_mesa_free_texmemory(image->base.Data);
-		image->base.Data = 0;
-	}
-
-	radeon_bo_unmap(mt->bo);
-
-	image->mt = mt;
-	image->mtface = face;
-	image->mtlevel = level;
-	radeon_miptree_reference(image->mt);
-}
-
-int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *texObj)
-{
-	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-	radeonTexObj *t = radeon_tex_obj(texObj);
-	radeon_texture_image *baseimage = get_radeon_texture_image(texObj->Image[0][texObj->BaseLevel]);
-	int face, level;
-
-	if (t->validated || t->image_override)
-		return GL_TRUE;
-
-	if (RADEON_DEBUG & RADEON_TEXTURE)
-		fprintf(stderr, "%s: Validating texture %p now\n", __FUNCTION__, texObj);
-
-	if (baseimage->base.Border > 0)
-		return GL_FALSE;
-
-	/* Ensure a matching miptree exists.
-	 *
-	 * Differing mipmap trees can result when the app uses TexImage to
-	 * change texture dimensions.
-	 *
-	 * Prefer to use base image's miptree if it
-	 * exists, since that most likely contains more valid data (remember
-	 * that the base level is usually significantly larger than the rest
-	 * of the miptree, so cubemaps are the only possible exception).
-	 */
-	if (baseimage->mt &&
-	    baseimage->mt != t->mt &&
-	    radeon_miptree_matches_texture(baseimage->mt, &t->base)) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = baseimage->mt;
-		radeon_miptree_reference(t->mt);
-	} else if (t->mt && !radeon_miptree_matches_texture(t->mt, &t->base)) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
-
-	if (!t->mt) {
-		if (RADEON_DEBUG & RADEON_TEXTURE)
-			fprintf(stderr, " Allocate new miptree\n");
-		radeon_try_alloc_miptree(rmesa, t, baseimage, 0, texObj->BaseLevel);
-		if (!t->mt) {
-			_mesa_problem(ctx, "radeon_validate_texture failed to alloc miptree");
-			return GL_FALSE;
-		}
-	}
-
-	/* Ensure all images are stored in the single main miptree */
-	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level) {
-			radeon_texture_image *image = get_radeon_texture_image(texObj->Image[face][level]);
-			if (RADEON_DEBUG & RADEON_TEXTURE)
-				fprintf(stderr, " face %i, level %i... %p vs %p ", face, level, t->mt, image->mt);
-			if (t->mt == image->mt || (!image->mt && !image->base.Data)) {
-				if (RADEON_DEBUG & RADEON_TEXTURE)
-					fprintf(stderr, "OK\n");
-
-				continue;
-			}
-
-			if (RADEON_DEBUG & RADEON_TEXTURE)
-				fprintf(stderr, "migrating\n");
-			migrate_image_to_miptree(t->mt, image, face, level);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-
 /**
  * Need to map texture image into memory before copying image data,
  * then unmap it.
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.h b/src/mesa/drivers/dri/radeon/radeon_texture.h
index 8995546d77..906daf12d0 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.h
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.h
@@ -33,7 +33,8 @@
 
 #include "main/formats.h"
 
-
+void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
+	GLuint numrows, GLuint rowsize);
 struct gl_texture_image *radeonNewTextureImage(GLcontext *ctx);
 void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage);
 
diff --git a/src/mesa/drivers/dri/savage/savagerender.c b/src/mesa/drivers/dri/savage/savagerender.c
index 32c74f9467..8221edf387 100644
--- a/src/mesa/drivers/dri/savage/savagerender.c
+++ b/src/mesa/drivers/dri/savage/savagerender.c
@@ -252,13 +252,13 @@ static GLboolean run_texnorm_stage( GLcontext *ctx,
          const GLboolean normalizeS = (texObj->WrapS == GL_REPEAT);
          const GLboolean normalizeT = (reallyEnabled & TEXTURE_2D_BIT) &&
             (texObj->WrapT == GL_REPEAT);
-         const GLfloat *in = (GLfloat *)VB->TexCoordPtr[i]->data;
-         const GLint instride = VB->TexCoordPtr[i]->stride;
+         const GLfloat *in = (GLfloat *)VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->data;
+         const GLint instride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->stride;
          GLfloat (*out)[4] = store->texcoord[i].data;
          GLint j;
 
          if (!ctx->Texture.Unit[i]._ReallyEnabled ||
-             VB->TexCoordPtr[i]->size == 4)
+             VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size == 4)
             /* Never try to normalize homogenous tex coords! */
             continue;
 
@@ -297,7 +297,7 @@ static GLboolean run_texnorm_stage( GLcontext *ctx,
          }
 
          if (normalizeS || normalizeT)
-            VB->AttribPtr[VERT_ATTRIB_TEX0+i] = VB->TexCoordPtr[i] = &store->texcoord[i];
+            VB->AttribPtr[_TNL_ATTRIB_TEX0 + i] = &store->texcoord[i];
       }
    }
 
diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c
index c04763b40e..e9529d1939 100644
--- a/src/mesa/drivers/dri/savage/savagetris.c
+++ b/src/mesa/drivers/dri/savage/savagetris.c
@@ -879,13 +879,13 @@ static GLboolean savageCheckPTexHack( GLcontext *ctx )
 
    RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->TexCoordPtr[0]->size == 4) {
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4) {
       if (!RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_ATTRIB_TEX1, _TNL_LAST_TEX ))
 	 return GL_TRUE; /* apply ptex hack */
       else
 	 FALLBACK(ctx, SAVAGE_FALLBACK_PROJ_TEXTURE, GL_TRUE);
    }
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->TexCoordPtr[1]->size == 4)
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
       FALLBACK(ctx, SAVAGE_FALLBACK_PROJ_TEXTURE, GL_TRUE);
 
    return GL_FALSE; /* don't apply ptex hack */
@@ -976,13 +976,13 @@ static INLINE GLuint savageChooseVertexFormat_s3d( GLcontext *ctx )
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
       if (imesa->ptexHack)
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_3F_XYW, SAVAGE_EMIT_STQ0, SAVAGE_SKIP_ST0);
-      else if (VB->TexCoordPtr[0]->size == 4)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[0]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size >= 2)
 	 /* The chromium menu emits some 3D tex coords even though no
 	  * 3D texture is enabled. Ignore the 3rd coordinate. */
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_2F, SAVAGE_EMIT_ST0, SAVAGE_SKIP_ST0 );
-      else if (VB->TexCoordPtr[0]->size == 1) {
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 1) {
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_1F, SAVAGE_EMIT_S0, SAVAGE_SKIP_S0 );
 	 EMIT_PAD( 4 );
       } else
@@ -1025,9 +1025,9 @@ static INLINE GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
       if (imesa->ptexHack)
 	 NEED_ATTR( SAVAGE_EMIT_STQ0, SAVAGE_SKIP_ST0);
-      else if (VB->TexCoordPtr[0]->size == 4)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[0]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size >= 2)
 	 /* The chromium menu emits some 3D tex coords even though no
 	  * 3D texture is enabled. Ignore the 3rd coordinate. */
 	 NEED_ATTR( SAVAGE_EMIT_ST0, SAVAGE_SKIP_ST0 );
@@ -1035,10 +1035,10 @@ static INLINE GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
 	 NEED_ATTR( SAVAGE_EMIT_S0, SAVAGE_SKIP_S0 );
    }
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 )) {
-      if (VB->TexCoordPtr[1]->size == 4)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
 	 /* projective textures are not supported by the hardware */
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[1]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size >= 2)
 	 NEED_ATTR( SAVAGE_EMIT_ST1, SAVAGE_SKIP_ST1 );
       else
 	 NEED_ATTR( SAVAGE_EMIT_S1, SAVAGE_SKIP_S1 );
diff --git a/src/mesa/drivers/dri/sis/sis_tris.c b/src/mesa/drivers/dri/sis/sis_tris.c
index 76d12d07b3..3cf10007b5 100644
--- a/src/mesa/drivers/dri/sis/sis_tris.c
+++ b/src/mesa/drivers/dri/sis/sis_tris.c
@@ -903,14 +903,14 @@ static void sisRenderStart( GLcontext *ctx )
 
    /* projective textures are not supported by the hardware */
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
-      if (VB->TexCoordPtr[0]->size > 2)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size > 2)
 	 tex_fallback = GL_TRUE;
       EMIT_ATTR(_TNL_ATTRIB_TEX0, EMIT_2F);
       AGPParseSet |= SiS_PS_HAS_UV0;
    }
    /* Will only hit tex1 on SiS300 */
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 )) {
-      if (VB->TexCoordPtr[1]->size > 2)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size > 2)
 	 tex_fallback = GL_TRUE;
       EMIT_ATTR(_TNL_ATTRIB_TEX1, EMIT_2F);
       AGPParseSet |= SiS_PS_HAS_UV1;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vb.c b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
index 4928802232..c200ba3255 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vb.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
@@ -69,11 +69,11 @@ static void interp_extras( GLcontext *ctx,
 
    /*fprintf(stderr, "%s\n", __FUNCTION__);*/
 
-   if (VB->ColorPtr[1]) {
+   if (VB->BackfaceColorPtr) {
       INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		 GET_COLOR(VB->BackfaceColorPtr, dst),
+		 GET_COLOR(VB->BackfaceColorPtr, out),
+		 GET_COLOR(VB->BackfaceColorPtr, in) );
    }
 
    if (VB->EdgeFlag) {
@@ -88,9 +88,9 @@ static void copy_pv_extras( GLcontext *ctx, GLuint dst, GLuint src )
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst), 
-		     GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+      COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		GET_COLOR(VB->BackfaceColorPtr, src) );
    }
 
    setup_tab[TDFX_CONTEXT(ctx)->SetupIndex].copy_pv(ctx, dst, src);
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h b/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
index 9b780761f4..19baf7d0d2 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
@@ -58,32 +58,32 @@ static void TAG(emit)( GLcontext *ctx,
 /*     fprintf(stderr, "%s\n", __FUNCTION__); */
 
    if (IND & TDFX_TEX0_BIT) {
-      tc0_stride = VB->TexCoordPtr[tmu0_source]->stride;
-      tc0 = VB->TexCoordPtr[tmu0_source]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->data;
       u0scale = fxMesa->sScale0;
       v0scale = fxMesa->tScale0;
       if (IND & TDFX_PTEX_BIT)
-	 tc0_size = VB->TexCoordPtr[tmu0_source]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->size;
    }
 
    if (IND & TDFX_TEX1_BIT) {
-      tc1 = VB->TexCoordPtr[tmu1_source]->data;
-      tc1_stride = VB->TexCoordPtr[tmu1_source]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->stride;
       u1scale = fxMesa->sScale1;
       v1scale = fxMesa->tScale1;
       if (IND & TDFX_PTEX_BIT)
-	 tc1_size = VB->TexCoordPtr[tmu1_source]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->size;
    }
    
    if (IND & TDFX_RGBA_BIT) {
-      col = VB->ColorPtr[0]->data;
-      col_stride = VB->ColorPtr[0]->stride;
-      col_size = VB->ColorPtr[0]->size;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
+      col_size = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
    }
    
    if (IND & TDFX_FOGC_BIT) {
-      fog = VB->FogCoordPtr->data;
-      fog_stride = VB->FogCoordPtr->stride;
+      fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+      fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
    }
 
    {
@@ -168,14 +168,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
       if (IND & TDFX_TEX1_BIT) {
-	 if (VB->TexCoordPtr[0] == 0)
-	    VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+	    VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 	 
-	 if (VB->TexCoordPtr[1]->size == 4)
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
 	    return GL_FALSE;
       }
 
-      if (VB->TexCoordPtr[0]->size == 4)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 return GL_FALSE;
    }
 
diff --git a/src/mesa/drivers/dri/unichrome/via_tris.c b/src/mesa/drivers/dri/unichrome/via_tris.c
index 79e67620c9..ab457d41dc 100644
--- a/src/mesa/drivers/dri/unichrome/via_tris.c
+++ b/src/mesa/drivers/dri/unichrome/via_tris.c
@@ -832,13 +832,13 @@ static GLboolean viaCheckPTexHack( GLcontext *ctx )
 
    RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->TexCoordPtr[0]->size == 4) {
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4) {
       if (!RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_ATTRIB_TEX1, _TNL_LAST_TEX ))
 	 ptexHack = GL_TRUE; 
       else
 	 fallback = GL_TRUE;
    }
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->TexCoordPtr[1]->size == 4)
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
       fallback = GL_TRUE;
 
    FALLBACK(VIA_CONTEXT(ctx), VIA_FALLBACK_PROJ_TEXTURE, fallback);
diff --git a/src/mesa/drivers/glide/fxvb.c b/src/mesa/drivers/glide/fxvb.c
index 1dc5f9891a..cc9ad0e8b8 100644
--- a/src/mesa/drivers/glide/fxvb.c
+++ b/src/mesa/drivers/glide/fxvb.c
@@ -104,24 +104,24 @@ static void interp_extras( GLcontext *ctx,
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      /* If stride is zero, ColorPtr[1] is constant across the VB, so
+   if (VB->BackfaceColorPtr) {
+      /* If stride is zero, BackfaceColorPtr is constant across the VB, so
        * there is no point interpolating between two values as they will
        * be identical.  This case is handled in t_dd_tritmp.h
        */
-      if (VB->ColorPtr[1]->stride) {
-	 assert(VB->ColorPtr[1]->stride == 4 * sizeof(GLfloat));
+      if (VB->BackfaceColorPtr->stride) {
+	 assert(VB->BackfaceColorPtr->stride == 4 * sizeof(GLfloat));
 	 INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		    GET_COLOR(VB->BackfaceColorPtr, dst),
+		    GET_COLOR(VB->BackfaceColorPtr, out),
+		    GET_COLOR(VB->BackfaceColorPtr, in) );
       }
 
-      if (VB->SecondaryColorPtr[1]) {
+      if (VB->BackfaceSecondaryColorPtr) {
 	 INTERP_3F( t,
-		    GET_COLOR(VB->SecondaryColorPtr[1], dst),
-		    GET_COLOR(VB->SecondaryColorPtr[1], out),
-		    GET_COLOR(VB->SecondaryColorPtr[1], in) );
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, out),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, in) );
       }
    }
 
@@ -137,13 +137,13 @@ static void copy_pv_extras( GLcontext *ctx, GLuint dst, GLuint src )
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst),
-		   GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+	 COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		   GET_COLOR(VB->BackfaceColorPtr, src) );
 
-	 if (VB->SecondaryColorPtr[1]) {
-	    COPY_3FV( GET_COLOR(VB->SecondaryColorPtr[1], dst),
-		      GET_COLOR(VB->SecondaryColorPtr[1], src) );
+	 if (VB->BackfaceSecondaryColorPtr) {
+	    COPY_3FV( GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		      GET_COLOR(VB->BackfaceSecondaryColorPtr, src) );
 	 }
    }
 
diff --git a/src/mesa/drivers/glide/fxvbtmp.h b/src/mesa/drivers/glide/fxvbtmp.h
index f7970c78e2..f7893c1573 100644
--- a/src/mesa/drivers/glide/fxvbtmp.h
+++ b/src/mesa/drivers/glide/fxvbtmp.h
@@ -62,37 +62,37 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (IND & SETUP_TMU0) {
-      tc0 = VB->TexCoordPtr[tmu0_source]->data;
-      tc0_stride = VB->TexCoordPtr[tmu0_source]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->stride;
       u0scale = fxMesa->s0scale;
       v0scale = fxMesa->t0scale;
       if (IND & SETUP_PTEX)
-	 tc0_size = VB->TexCoordPtr[tmu0_source]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->size;
    }
 
    if (IND & SETUP_TMU1) {
-      tc1 = VB->TexCoordPtr[tmu1_source]->data;
-      tc1_stride = VB->TexCoordPtr[tmu1_source]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->stride;
       u1scale = fxMesa->s1scale; /* wrong if tmu1_source == 0, possible? */
       v1scale = fxMesa->t1scale;
       if (IND & SETUP_PTEX)
-	 tc1_size = VB->TexCoordPtr[tmu1_source]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->size;
    }
    
    if (IND & SETUP_RGBA) {
-      col = VB->ColorPtr[0]->data;
-      col_stride = VB->ColorPtr[0]->stride;
-      col_size = VB->ColorPtr[0]->size;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
+      col_size = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
    }
 
    if (IND & SETUP_SPEC) {
-      spec = VB->SecondaryColorPtr[0]->data;
-      spec_stride = VB->SecondaryColorPtr[0]->stride;
+      spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+      spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
    }
 
    if (IND & SETUP_FOGC) {
-      fog = VB->FogCoordPtr->data;
-      fog_stride = VB->FogCoordPtr->stride;
+      fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+      fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
    }
 
    if (start) {
@@ -220,14 +220,15 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
       if (IND & SETUP_TMU1) {
-	 if (VB->TexCoordPtr[0] == 0)
-	    VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+	    VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 	 
-	 if (VB->TexCoordPtr[1]->size == 4)
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
 	    return GL_FALSE;
       }
 
-      if (VB->TexCoordPtr[0] && VB->TexCoordPtr[0]->size == 4)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0] &&
+	  VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 return GL_FALSE;
    }
 
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
index c99ba0bba5..0b373814fe 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
@@ -189,9 +189,9 @@
 		GLfloat		ex,ey,fx,fy,cc;							\
 		/* Get vars for later */							\
 		VB		= &TNL_CONTEXT(ctx)->vb;					\
-		vbcolor	= (GLchan (*)[4])VB->ColorPtr[1]->data;		\
-		if (VB->SecondaryColorPtr[1]) {						\
-			vbspec = (GLchan (*)[4])VB->SecondaryColorPtr[1]->data;	\
+		vbcolor	= (GLchan (*)[4])VB->BackfaceColorPtr->data;	\
+		if (VB->BackfaceSecondaryColorPtr) {			\
+			vbspec = (GLchan (*)[4])VB->BackfaceSecondaryColorPtr->data;	\
 		} else {													\
 			vbspec = NULL;											\
 		}															\
@@ -241,33 +241,33 @@
 	DWORD					dwColor;
 
 #define GLD_SETUP_3D_VERTEX(v)					\
-	p4f				= VB->ObjPtr->data;			\
+	p4f = VB->AttribPtr[_TNL_ATTRIB_POS]->data;		\
 	pV->Position.x	= p4f[##v][0];				\
 	pV->Position.y	= p4f[##v][1];				\
 	pV->Position.z	= p4f[##v][2];
 
 #define GLD_SETUP_SMOOTH_COLOUR_3D(v)															\
-	p4f			= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;										\
 	pV->Diffuse	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 
 #define GLD_SETUP_GET_FLAT_COLOUR_3D(v)													\
-	p4f		= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;	\
 	dwColor	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 #define GLD_SETUP_USE_FLAT_COLOUR_3D			\
 	pV->Diffuse = dwColor;
 
 #define GLD_SETUP_TEX0_3D(v)						\
-	if (VB->TexCoordPtr[0]) {						\
-		tc				= VB->TexCoordPtr[0]->data;	\
+	if (VB->AttribPtr[_TNL_ATTRIB_TEX0]) {				\
+		tc = VB->AttribPtr[_TNL_ATTRIB_TEX0]->data;		\
 		pV->TexUnit0.x	= tc[##v][0];				\
 		pV->TexUnit0.y	= tc[##v][1];				\
 	}
 
 #define GLD_SETUP_TEX1_3D(v)						\
-	if (VB->TexCoordPtr[1]) {						\
-		tc				= VB->TexCoordPtr[1]->data;	\
+	if (VB->AttribPtr[_TNL_ATTRIB_TEX1]) {				\
+		tc = VB->AttribPtr[_TNL_ATTRIB_TEX1]->data;		\
 		pV->TexUnit1.x	= tc[##v][0];				\
 		pV->TexUnit1.y	= tc[##v][1];				\
 	}
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_vb_d3d_render_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_vb_d3d_render_dx7.c
index a85620dde8..c39775cad3 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_vb_d3d_render_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_vb_d3d_render_dx7.c
@@ -151,7 +151,7 @@ static GLboolean gld_d3d_render_stage_run(
 #if 0
    // For debugging: Useful to see if an app passes colour data in
    // an unusual format.
-   switch (VB->ColorPtr[0]->Type) {
+   switch (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->Type) {
    case GL_FLOAT:
 	   ddlogMessage(GLDLOG_SYSTEM, "ColorPtr: GL_FLOAT\n");
 	   break;
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
index a5b5462f03..990922580a 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
@@ -189,9 +189,9 @@
 		GLfloat		ex,ey,fx,fy,cc;							\
 		/* Get vars for later */							\
 		VB		= &TNL_CONTEXT(ctx)->vb;					\
-		vbcolor	= (GLchan (*)[4])VB->ColorPtr[1]->data;		\
-		if (VB->SecondaryColorPtr[1]) {						\
-			vbspec = (GLchan (*)[4])VB->SecondaryColorPtr[1]->data;	\
+		vbcolor	= (GLchan (*)[4])VB->BackfaceColorPtr->data;	\
+		if (VB->BackfaceSecondaryColorPtr) {			\
+			vbspec = (GLchan (*)[4])VB->BackfaceSecondaryColorPtr->data;	\
 		} else {													\
 			vbspec = NULL;											\
 		}															\
@@ -241,33 +241,33 @@
 	DWORD					dwColor;
 
 #define GLD_SETUP_3D_VERTEX(v)					\
-	p4f				= VB->ObjPtr->data;			\
+	p4f = VB->AttribPtr[_TNL_ATTRIB_POS]->data;		\
 	pV->Position.x	= p4f[##v][0];				\
 	pV->Position.y	= p4f[##v][1];				\
 	pV->Position.z	= p4f[##v][2];
 
 #define GLD_SETUP_SMOOTH_COLOUR_3D(v)															\
-	p4f			= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;	\
 	pV->Diffuse	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 
 #define GLD_SETUP_GET_FLAT_COLOUR_3D(v)													\
-	p4f		= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;	\
 	dwColor	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 #define GLD_SETUP_USE_FLAT_COLOUR_3D			\
 	pV->Diffuse = dwColor;
 
 #define GLD_SETUP_TEX0_3D(v)						\
-	if (VB->TexCoordPtr[0]) {						\
-		tc				= VB->TexCoordPtr[0]->data;	\
+	if (VB->AttribPtr[_TNL_ATTRIB_TEX0]) {				\
+		tc = VB->TnlAttribPtr[_TNL_ATTRIB_TEX0]->data;		\
 		pV->TexUnit0.x	= tc[##v][0];				\
 		pV->TexUnit0.y	= tc[##v][1];				\
 	}
 
 #define GLD_SETUP_TEX1_3D(v)						\
-	if (VB->TexCoordPtr[1]) {						\
-		tc				= VB->TexCoordPtr[1]->data;	\
+	if (VB->TnlAttribPtr[_TNL_ATTRIB_TEX1]) {			\
+		tc = VB->TnlAttribPtr[_TNL_ATTRIB_TEX1]->data;		\
 		pV->TexUnit1.x	= tc[##v][0];				\
 		pV->TexUnit1.y	= tc[##v][1];				\
 	}
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_vb_d3d_render_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_vb_d3d_render_dx8.c
index cafbf4f5c5..265c81fb4a 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_vb_d3d_render_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_vb_d3d_render_dx8.c
@@ -149,7 +149,7 @@ static GLboolean gld_d3d_render_stage_run(
 #if 0
    // For debugging: Useful to see if an app passes colour data in
    // an unusual format.
-   switch (VB->ColorPtr[0]->Type) {
+   switch (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->Type) {
    case GL_FLOAT:
 	   ddlogMessage(GLDLOG_SYSTEM, "ColorPtr: GL_FLOAT\n");
 	   break;
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
index 403a9d5f86..fd4dd4ed75 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
@@ -189,9 +189,9 @@
 		GLfloat		ex,ey,fx,fy,cc;							\
 		/* Get vars for later */							\
 		VB		= &TNL_CONTEXT(ctx)->vb;					\
-		vbcolor	= (GLchan (*)[4])VB->ColorPtr[1]->data;		\
-		if (VB->SecondaryColorPtr[1]) {						\
-			vbspec = (GLchan (*)[4])VB->SecondaryColorPtr[1]->data;	\
+		vbcolor	= (GLchan (*)[4])VB->BackfaceColorPtr->data;	\
+		if (VB->BackfaceSecondaryColorPtr) {			\
+			vbspec = (GLchan (*)[4])VB->BackfaceSecondaryColorPtr->data;	\
 		} else {													\
 			vbspec = NULL;											\
 		}															\
@@ -241,33 +241,33 @@
 	DWORD					dwColor;
 
 #define GLD_SETUP_3D_VERTEX(v)					\
-	p4f				= VB->ObjPtr->data;			\
+	p4f = VB->AttribPtr[_TNL_ATTRIB_POS]->data;		\
 	pV->Position.x	= p4f[##v][0];				\
 	pV->Position.y	= p4f[##v][1];				\
 	pV->Position.z	= p4f[##v][2];
 
 #define GLD_SETUP_SMOOTH_COLOUR_3D(v)															\
-	p4f			= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;										\
 	pV->Diffuse	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 
 #define GLD_SETUP_GET_FLAT_COLOUR_3D(v)													\
-	p4f		= (GLfloat (*)[4])VB->ColorPtr[0]->data;										\
+	p4f = (GLfloat (*)[4])VB->AttribPtr[_TNL_ATTRIB_COLOR00]->data;	\
 	dwColor	= D3DCOLOR_COLORVALUE(p4f[##v][0], p4f[##v][1], p4f[##v][2], p4f[##v][3]);
 
 #define GLD_SETUP_USE_FLAT_COLOUR_3D			\
 	pV->Diffuse = dwColor;
 
 #define GLD_SETUP_TEX0_3D(v)						\
-	if (VB->TexCoordPtr[0]) {						\
-		tc				= VB->TexCoordPtr[0]->data;	\
+	if (VB->AttribPtr[_TNL_ATTRIB_TEX0]) {				\
+		tc = VB->AttribPtr[_TNL_ATTRIB_TEX0]->data;		\
 		pV->TexUnit0.x	= tc[##v][0];				\
 		pV->TexUnit0.y	= tc[##v][1];				\
 	}
 
 #define GLD_SETUP_TEX1_3D(v)						\
-	if (VB->TexCoordPtr[1]) {						\
-		tc				= VB->TexCoordPtr[1]->data;	\
+	if (VB->AttribPtr[_TNL_ATTRIB_TEX1]) {				\
+		tc = VB->AttribPtr[_TNL_ATTRIB_TEX1]->data;		\
 		pV->TexUnit1.x	= tc[##v][0];				\
 		pV->TexUnit1.y	= tc[##v][1];				\
 	}
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_vb_d3d_render_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_vb_d3d_render_dx9.c
index 4fa6bcaf1a..91a68b3f2d 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_vb_d3d_render_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_vb_d3d_render_dx9.c
@@ -149,7 +149,7 @@ static GLboolean gld_d3d_render_stage_run(
 #if 0
    // For debugging: Useful to see if an app passes colour data in
    // an unusual format.
-   switch (VB->ColorPtr[0]->Type) {
+   switch (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->Type) {
    case GL_FLOAT:
 	   ddlogMessage(GLDLOG_SYSTEM, "ColorPtr: GL_FLOAT\n");
 	   break;
diff --git a/src/mesa/glapi/ARB_framebuffer_object.xml b/src/mesa/glapi/ARB_framebuffer_object.xml
index b700e7e77d..e6bdcd6e50 100644
--- a/src/mesa/glapi/ARB_framebuffer_object.xml
+++ b/src/mesa/glapi/ARB_framebuffer_object.xml
@@ -12,10 +12,10 @@
 
     <enum name="RENDERBUFFER"                             value="0x8D41"/>
 
-    <enum name="STENCIL_INDEX1_EXT"                       value="0x8D46"/>
-    <enum name="STENCIL_INDEX4_EXT"                       value="0x8D47"/>
-    <enum name="STENCIL_INDEX8_EXT"                       value="0x8D48"/>
-    <enum name="STENCIL_INDEX16_EXT"                      value="0x8D49"/>
+    <enum name="STENCIL_INDEX1"                           value="0x8D46"/>
+    <enum name="STENCIL_INDEX4"                           value="0x8D47"/>
+    <enum name="STENCIL_INDEX8"                           value="0x8D48"/>
+    <enum name="STENCIL_INDEX16"                          value="0x8D49"/>
 
     <enum name="RENDERBUFFER_WIDTH"              value="0x8D42"/>
     <enum name="RENDERBUFFER_HEIGHT"             value="0x8D43"/>
@@ -93,33 +93,35 @@
     <enum name="STENCIL_ATTACHMENT"                        value="0x8D20"/>
     <enum name="DEPTH_STENCIL_ATTACHMENT"                  value="0x821A"/>
 
-    <enum name="MAX_SAMPLES"                               value="0x8D57"/>
+    <enum name="MAX_SAMPLES"                     count="1" value="0x8D57">
+        <size name="Get" mode="get"/>
+    </enum>
 
     <enum name="FRAMEBUFFER_COMPLETE"                      value="0x8CD5"/>
     <enum name="FRAMEBUFFER_INCOMPLETE_ATTACHMENT"         value="0x8CD6"/>
     <enum name="FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT" value="0x8CD7"/>
-    <enum name="FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT"    value="0x8CDB"/>
-    <enum name="FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT"    value="0x8CDC"/>
+    <enum name="FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER"        value="0x8CDB"/>
+    <enum name="FRAMEBUFFER_INCOMPLETE_READ_BUFFER"        value="0x8CDC"/>
     <enum name="FRAMEBUFFER_UNSUPPORTED"                   value="0x8CDD"/>
     <enum name="FRAMEBUFFER_INCOMPLETE_MULTISAMPLE"        value="0x8D56"/>
     <enum name="FRAMEBUFFER_UNDEFINED"                     value="0x8219"/>
 
-    <enum name="FRAMEBUFFER_BINDING_EXT"         count="1" value="0x8CA6">
+    <enum name="FRAMEBUFFER_BINDING"             count="1" value="0x8CA6">
         <size name="Get" mode="get"/>
     </enum>
-    <enum name="DRAW_FRAMEBUFFER_BINDING_EXT"    count="1" value="0x8CA6">
+    <enum name="DRAW_FRAMEBUFFER_BINDING"        count="1" value="0x8CA6">
         <size name="Get" mode="get"/>
     </enum>
-    <enum name="READ_FRAMEBUFFER_BINDING_EXT"    count="1" value="0x8CAA">
+    <enum name="READ_FRAMEBUFFER_BINDING"        count="1" value="0x8CAA">
         <size name="Get" mode="get"/>
     </enum>
-    <enum name="RENDERBUFFER_BINDING_EXT"        count="1" value="0x8CA7">
+    <enum name="RENDERBUFFER_BINDING"            count="1" value="0x8CA7">
         <size name="Get" mode="get"/>
     </enum>
-    <enum name="MAX_COLOR_ATTACHMENTS_EXT"       count="1" value="0x8CDF">
+    <enum name="MAX_COLOR_ATTACHMENTS"           count="1" value="0x8CDF">
         <size name="Get" mode="get"/>
     </enum>
-    <enum name="MAX_RENDERBUFFER_SIZE_EXT"       count="1" value="0x84E8">
+    <enum name="MAX_RENDERBUFFER_SIZE"           count="1" value="0x84E8">
         <size name="Get" mode="get"/>
     </enum>
 
@@ -131,7 +133,10 @@
 
     <enum name="DEPTH24_STENCIL8"                         value="0x88F0"/>
 
-    <enum name="TEXTURE_STENCIL_SIZE"                     value="0x88F1"/>
+    <enum name="TEXTURE_STENCIL_SIZE"          count="1"  value="0x88F1">
+        <size name="GetTexLevelParameterfv" mode="get"/>
+        <size name="GetTexLevelParameteriv" mode="get"/>
+    </enum>
 
 
 
@@ -168,6 +173,7 @@
         <param name="internalformat" type="GLenum"/>
         <param name="width" type="GLsizei"/>
         <param name="height" type="GLsizei"/>
+        <glx rop="4331"/>
     </function>
 
     <function name="GetRenderbufferParameteriv" alias="GetRenderbufferParameterivEXT">
diff --git a/src/mesa/glapi/EXT_framebuffer_object.xml b/src/mesa/glapi/EXT_framebuffer_object.xml
index 5559b48b11..4f418f7b64 100644
--- a/src/mesa/glapi/EXT_framebuffer_object.xml
+++ b/src/mesa/glapi/EXT_framebuffer_object.xml
@@ -192,4 +192,44 @@
     </function>
 </category>
 
+<category name="GL_EXT_framebuffer_blit" number="316">
+    <enum name="READ_FRAMEBUFFER_EXT" value="0x8CA8"/>
+    <enum name="DRAW_FRAMEBUFFER_EXT" value="0x8CA9"/>
+    <enum name="DRAW_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CA6">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="READ_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CAA">
+        <size name="Get" mode="get"/>
+    </enum>
+    <function name="BlitFramebufferEXT" offset="assign" static_dispatch="false">
+        <param name="srcX0" type="GLint"/>
+        <param name="srcY0" type="GLint"/>
+        <param name="srcX1" type="GLint"/>
+        <param name="srcY1" type="GLint"/>
+        <param name="dstX0" type="GLint"/>
+        <param name="dstY0" type="GLint"/>
+        <param name="dstX1" type="GLint"/>
+        <param name="dstY1" type="GLint"/>
+        <param name="mask" type="GLbitfield"/>
+        <param name="filter" type="GLenum"/>
+        <glx rop="4330"/>
+    </function>
+</category>
+
+<category name="GL_EXT_framebuffer_multisample" number="317">
+    <enum name="RENDERBUFFER_SAMPLES_EXT"                  value="0x8CAB"/>
+    <enum name="FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT"    value="0x8D56"/>
+    <enum name="MAX_SAMPLES_EXT"                 count="1" value="0x8D57">
+        <size name="Get" mode="get"/>
+    </enum>
+
+    <function name="RenderbufferStorageMultisampleEXT" alias="RenderbufferStorageMultisample">
+        <param name="target" type="GLenum"/>
+        <param name="samples" type="GLsizei"/>
+        <param name="internalformat" type="GLenum"/>
+        <param name="width" type="GLsizei"/>
+        <param name="height" type="GLsizei"/>
+    </function>
+</category>
+
 </OpenGLAPI>
diff --git a/src/mesa/glapi/EXT_packed_depth_stencil.xml b/src/mesa/glapi/EXT_packed_depth_stencil.xml
new file mode 100644
index 0000000000..5be810302f
--- /dev/null
+++ b/src/mesa/glapi/EXT_packed_depth_stencil.xml
@@ -0,0 +1,18 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+<category name="GL_EXT_packed_depth_stencil" number="312">
+    <!-- These enums are shared with GL_NV_packed_depth_stencil. -->
+    <enum name="DEPTH_STENCIL_EXT"                        value="0x84F9"/>
+    <enum name="UNSIGNED_INT_24_8_EXT"                    value="0x84FA"/>
+
+    <enum name="DEPTH24_STENCIL8_EXT"                     value="0x88F0"/>
+
+    <enum name="TEXTURE_STENCIL_SIZE_EXT"      count="1"  value="0x88F1">
+        <size name="GetTexLevelParameterfv" mode="get"/>
+        <size name="GetTexLevelParameteriv" mode="get"/>
+    </enum>
+</category>
+
+</OpenGLAPI>
diff --git a/src/mesa/glapi/EXT_texture_array.xml b/src/mesa/glapi/EXT_texture_array.xml
index e5bd9f3c69..b5b8bd406f 100644
--- a/src/mesa/glapi/EXT_texture_array.xml
+++ b/src/mesa/glapi/EXT_texture_array.xml
@@ -35,6 +35,7 @@
         <param name="texture" type="GLuint"/>
         <param name="level" type="GLint"/>
         <param name="layer" type="GLint"/>
+	<glx rop="237"/>
     </function>
 </category>
 
diff --git a/src/mesa/glapi/Makefile b/src/mesa/glapi/Makefile
index fb6be1a3a9..71bef68ea5 100644
--- a/src/mesa/glapi/Makefile
+++ b/src/mesa/glapi/Makefile
@@ -47,16 +47,19 @@ SERVER_OUTPUTS = \
 	$(SERVER_GLAPI_FILES)
 
 API_XML = gl_API.xml \
-	EXT_framebuffer_object.xml \
 	ARB_copy_buffer.xml \
 	ARB_depth_clamp.xml \
+	ARB_draw_elements_base_vertex.xml \
 	ARB_framebuffer_object.xml \
 	ARB_map_buffer_range.xml \
 	ARB_seamless_cube_map.xml \
 	ARB_sync.xml \
 	ARB_vertex_array_object.xml \
 	APPLE_vertex_array_object.xml \
-	EXT_provoking_vertex.xml
+	EXT_framebuffer_object.xml \
+	EXT_packed_depth_stencil.xml \
+	EXT_provoking_vertex.xml \
+	EXT_texture_array.xml
 
 COMMON = gl_XML.py glX_XML.py license.py $(API_XML) typeexpr.py
 COMMON_GLX = $(COMMON) glX_API.xml glX_XML.py glX_proto_common.py
diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml
index f8a311a7ce..0346a6dd9d 100644
--- a/src/mesa/glapi/gl_API.xml
+++ b/src/mesa/glapi/gl_API.xml
@@ -12285,30 +12285,9 @@
 
 <xi:include href="EXT_framebuffer_object.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
-<xi:include href="EXT_provoking_vertex.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+<xi:include href="EXT_packed_depth_stencil.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
-<category name="GL_EXT_framebuffer_blit" number="316">
-    <enum name="READ_FRAMEBUFFER_EXT" value="0x8CA8"/>
-    <enum name="DRAW_FRAMEBUFFER_EXT" value="0x8CA9"/>
-    <enum name="DRAW_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CA6">
-        <size name="Get" mode="get"/>
-    </enum>
-    <enum name="READ_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CAA">
-        <size name="Get" mode="get"/>
-    </enum>
-    <function name="BlitFramebufferEXT" offset="assign" static_dispatch="false">
-        <param name="srcX0" type="GLint"/>
-        <param name="srcY0" type="GLint"/>
-        <param name="srcX1" type="GLint"/>
-        <param name="srcY1" type="GLint"/>
-        <param name="dstX0" type="GLint"/>
-        <param name="dstY0" type="GLint"/>
-        <param name="dstX1" type="GLint"/>
-        <param name="dstY1" type="GLint"/>
-        <param name="mask" type="GLbitfield"/>
-        <param name="filter" type="GLenum"/>
-    </function>
-</category>
+<xi:include href="EXT_provoking_vertex.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
 <category name="GL_APPLE_flush_buffer_range" number="321">
     <enum name="BUFFER_SERIALIZED_MODIFY_APPLE" count="1" value="0x8A12">
diff --git a/src/mesa/glapi/glapitemp.h b/src/mesa/glapi/glapitemp.h
index b910386298..00279fa3e6 100644
--- a/src/mesa/glapi/glapitemp.h
+++ b/src/mesa/glapi/glapitemp.h
@@ -4011,6 +4011,11 @@ KEYWORD1 void KEYWORD2 NAME(RenderbufferStorageMultisample)(GLenum target, GLsiz
    DISPATCH(RenderbufferStorageMultisample, (target, samples, internalformat, width, height), (F, "glRenderbufferStorageMultisample(0x%x, %d, 0x%x, %d, %d);\n", target, samples, internalformat, width, height));
 }
 
+KEYWORD1 void KEYWORD2 NAME(RenderbufferStorageMultisampleEXT)(GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height)
+{
+   DISPATCH(RenderbufferStorageMultisample, (target, samples, internalformat, width, height), (F, "glRenderbufferStorageMultisampleEXT(0x%x, %d, 0x%x, %d, %d);\n", target, samples, internalformat, width, height));
+}
+
 KEYWORD1 void KEYWORD2 NAME(FramebufferTextureARB)(GLenum target, GLenum attachment, GLuint texture, GLint level)
 {
    DISPATCH(FramebufferTextureARB, (target, attachment, texture, level), (F, "glFramebufferTextureARB(0x%x, 0x%x, %d, %d);\n", target, attachment, texture, level));
@@ -6855,6 +6860,7 @@ static _glapi_proc UNUSED_TABLE_NAME[] = {
    TABLE_ENTRY(GetAttribLocation),
    TABLE_ENTRY(DrawBuffers),
    TABLE_ENTRY(DrawBuffersATI),
+   TABLE_ENTRY(RenderbufferStorageMultisampleEXT),
    TABLE_ENTRY(PointParameterf),
    TABLE_ENTRY(PointParameterfARB),
    TABLE_ENTRY(PointParameterfv),
diff --git a/src/mesa/glapi/glprocs.h b/src/mesa/glapi/glprocs.h
index aaff49eb63..fbb30ea4a3 100644
--- a/src/mesa/glapi/glprocs.h
+++ b/src/mesa/glapi/glprocs.h
@@ -1053,6 +1053,7 @@ static const char gl_string_table[] =
     "glGetAttribLocation\0"
     "glDrawBuffers\0"
     "glDrawBuffersATI\0"
+    "glRenderbufferStorageMultisampleEXT\0"
     "glSampleMaskEXT\0"
     "glSamplePatternEXT\0"
     "glPointParameterf\0"
@@ -2263,104 +2264,105 @@ static const glprocs_table_t static_functions[] = {
     NAME_FUNC_OFFSET(17979, glGetAttribLocationARB, glGetAttribLocationARB, NULL, _gloffset_GetAttribLocationARB),
     NAME_FUNC_OFFSET(17999, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
     NAME_FUNC_OFFSET(18013, glDrawBuffersARB, glDrawBuffersARB, NULL, _gloffset_DrawBuffersARB),
-    NAME_FUNC_OFFSET(18030, gl_dispatch_stub_587, gl_dispatch_stub_587, NULL, _gloffset_SampleMaskSGIS),
-    NAME_FUNC_OFFSET(18046, gl_dispatch_stub_588, gl_dispatch_stub_588, NULL, _gloffset_SamplePatternSGIS),
-    NAME_FUNC_OFFSET(18065, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(18083, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(18104, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
-    NAME_FUNC_OFFSET(18126, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(18145, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(18167, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
-    NAME_FUNC_OFFSET(18190, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
-    NAME_FUNC_OFFSET(18209, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
-    NAME_FUNC_OFFSET(18229, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
-    NAME_FUNC_OFFSET(18248, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
-    NAME_FUNC_OFFSET(18268, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
-    NAME_FUNC_OFFSET(18287, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
-    NAME_FUNC_OFFSET(18307, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
-    NAME_FUNC_OFFSET(18326, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
-    NAME_FUNC_OFFSET(18346, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
-    NAME_FUNC_OFFSET(18365, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
-    NAME_FUNC_OFFSET(18385, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
-    NAME_FUNC_OFFSET(18405, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
-    NAME_FUNC_OFFSET(18426, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
-    NAME_FUNC_OFFSET(18446, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
-    NAME_FUNC_OFFSET(18467, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
-    NAME_FUNC_OFFSET(18487, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
-    NAME_FUNC_OFFSET(18508, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
-    NAME_FUNC_OFFSET(18532, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
-    NAME_FUNC_OFFSET(18550, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
-    NAME_FUNC_OFFSET(18570, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
-    NAME_FUNC_OFFSET(18588, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
-    NAME_FUNC_OFFSET(18600, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
-    NAME_FUNC_OFFSET(18613, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
-    NAME_FUNC_OFFSET(18625, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
-    NAME_FUNC_OFFSET(18638, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
-    NAME_FUNC_OFFSET(18658, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
-    NAME_FUNC_OFFSET(18682, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
-    NAME_FUNC_OFFSET(18696, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
-    NAME_FUNC_OFFSET(18713, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
-    NAME_FUNC_OFFSET(18728, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
-    NAME_FUNC_OFFSET(18746, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
-    NAME_FUNC_OFFSET(18760, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
-    NAME_FUNC_OFFSET(18777, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
-    NAME_FUNC_OFFSET(18792, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
-    NAME_FUNC_OFFSET(18810, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
-    NAME_FUNC_OFFSET(18824, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
-    NAME_FUNC_OFFSET(18841, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
-    NAME_FUNC_OFFSET(18856, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
-    NAME_FUNC_OFFSET(18874, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
-    NAME_FUNC_OFFSET(18888, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
-    NAME_FUNC_OFFSET(18905, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
-    NAME_FUNC_OFFSET(18920, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
-    NAME_FUNC_OFFSET(18938, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
-    NAME_FUNC_OFFSET(18952, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
-    NAME_FUNC_OFFSET(18969, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
-    NAME_FUNC_OFFSET(18984, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
-    NAME_FUNC_OFFSET(19002, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
-    NAME_FUNC_OFFSET(19016, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
-    NAME_FUNC_OFFSET(19033, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
-    NAME_FUNC_OFFSET(19048, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
-    NAME_FUNC_OFFSET(19066, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
-    NAME_FUNC_OFFSET(19080, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
-    NAME_FUNC_OFFSET(19097, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
-    NAME_FUNC_OFFSET(19112, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
-    NAME_FUNC_OFFSET(19130, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
-    NAME_FUNC_OFFSET(19144, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
-    NAME_FUNC_OFFSET(19161, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
-    NAME_FUNC_OFFSET(19176, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
-    NAME_FUNC_OFFSET(19194, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
-    NAME_FUNC_OFFSET(19211, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
-    NAME_FUNC_OFFSET(19231, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
-    NAME_FUNC_OFFSET(19248, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
-    NAME_FUNC_OFFSET(19274, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
-    NAME_FUNC_OFFSET(19303, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
-    NAME_FUNC_OFFSET(19318, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
-    NAME_FUNC_OFFSET(19336, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
-    NAME_FUNC_OFFSET(19355, gl_dispatch_stub_758, gl_dispatch_stub_758, NULL, _gloffset_DeleteVertexArraysAPPLE),
-    NAME_FUNC_OFFSET(19376, gl_dispatch_stub_760, gl_dispatch_stub_760, NULL, _gloffset_IsVertexArrayAPPLE),
-    NAME_FUNC_OFFSET(19392, gl_dispatch_stub_768, gl_dispatch_stub_768, NULL, _gloffset_BlendEquationSeparateEXT),
-    NAME_FUNC_OFFSET(19416, gl_dispatch_stub_768, gl_dispatch_stub_768, NULL, _gloffset_BlendEquationSeparateEXT),
-    NAME_FUNC_OFFSET(19443, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
-    NAME_FUNC_OFFSET(19461, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
-    NAME_FUNC_OFFSET(19480, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
-    NAME_FUNC_OFFSET(19505, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
-    NAME_FUNC_OFFSET(19526, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
-    NAME_FUNC_OFFSET(19548, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
-    NAME_FUNC_OFFSET(19574, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
-    NAME_FUNC_OFFSET(19597, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
-    NAME_FUNC_OFFSET(19620, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
-    NAME_FUNC_OFFSET(19643, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
-    NAME_FUNC_OFFSET(19661, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
-    NAME_FUNC_OFFSET(19680, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
-    NAME_FUNC_OFFSET(19697, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
-    NAME_FUNC_OFFSET(19735, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
-    NAME_FUNC_OFFSET(19764, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
-    NAME_FUNC_OFFSET(19780, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
-    NAME_FUNC_OFFSET(19797, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
-    NAME_FUNC_OFFSET(19819, gl_dispatch_stub_786, gl_dispatch_stub_786, NULL, _gloffset_BlitFramebufferEXT),
-    NAME_FUNC_OFFSET(19837, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
-    NAME_FUNC_OFFSET(19863, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT),
+    NAME_FUNC_OFFSET(18030, glRenderbufferStorageMultisample, glRenderbufferStorageMultisample, NULL, _gloffset_RenderbufferStorageMultisample),
+    NAME_FUNC_OFFSET(18066, gl_dispatch_stub_587, gl_dispatch_stub_587, NULL, _gloffset_SampleMaskSGIS),
+    NAME_FUNC_OFFSET(18082, gl_dispatch_stub_588, gl_dispatch_stub_588, NULL, _gloffset_SamplePatternSGIS),
+    NAME_FUNC_OFFSET(18101, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18119, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18140, glPointParameterfEXT, glPointParameterfEXT, NULL, _gloffset_PointParameterfEXT),
+    NAME_FUNC_OFFSET(18162, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18181, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18203, glPointParameterfvEXT, glPointParameterfvEXT, NULL, _gloffset_PointParameterfvEXT),
+    NAME_FUNC_OFFSET(18226, glSecondaryColor3bEXT, glSecondaryColor3bEXT, NULL, _gloffset_SecondaryColor3bEXT),
+    NAME_FUNC_OFFSET(18245, glSecondaryColor3bvEXT, glSecondaryColor3bvEXT, NULL, _gloffset_SecondaryColor3bvEXT),
+    NAME_FUNC_OFFSET(18265, glSecondaryColor3dEXT, glSecondaryColor3dEXT, NULL, _gloffset_SecondaryColor3dEXT),
+    NAME_FUNC_OFFSET(18284, glSecondaryColor3dvEXT, glSecondaryColor3dvEXT, NULL, _gloffset_SecondaryColor3dvEXT),
+    NAME_FUNC_OFFSET(18304, glSecondaryColor3fEXT, glSecondaryColor3fEXT, NULL, _gloffset_SecondaryColor3fEXT),
+    NAME_FUNC_OFFSET(18323, glSecondaryColor3fvEXT, glSecondaryColor3fvEXT, NULL, _gloffset_SecondaryColor3fvEXT),
+    NAME_FUNC_OFFSET(18343, glSecondaryColor3iEXT, glSecondaryColor3iEXT, NULL, _gloffset_SecondaryColor3iEXT),
+    NAME_FUNC_OFFSET(18362, glSecondaryColor3ivEXT, glSecondaryColor3ivEXT, NULL, _gloffset_SecondaryColor3ivEXT),
+    NAME_FUNC_OFFSET(18382, glSecondaryColor3sEXT, glSecondaryColor3sEXT, NULL, _gloffset_SecondaryColor3sEXT),
+    NAME_FUNC_OFFSET(18401, glSecondaryColor3svEXT, glSecondaryColor3svEXT, NULL, _gloffset_SecondaryColor3svEXT),
+    NAME_FUNC_OFFSET(18421, glSecondaryColor3ubEXT, glSecondaryColor3ubEXT, NULL, _gloffset_SecondaryColor3ubEXT),
+    NAME_FUNC_OFFSET(18441, glSecondaryColor3ubvEXT, glSecondaryColor3ubvEXT, NULL, _gloffset_SecondaryColor3ubvEXT),
+    NAME_FUNC_OFFSET(18462, glSecondaryColor3uiEXT, glSecondaryColor3uiEXT, NULL, _gloffset_SecondaryColor3uiEXT),
+    NAME_FUNC_OFFSET(18482, glSecondaryColor3uivEXT, glSecondaryColor3uivEXT, NULL, _gloffset_SecondaryColor3uivEXT),
+    NAME_FUNC_OFFSET(18503, glSecondaryColor3usEXT, glSecondaryColor3usEXT, NULL, _gloffset_SecondaryColor3usEXT),
+    NAME_FUNC_OFFSET(18523, glSecondaryColor3usvEXT, glSecondaryColor3usvEXT, NULL, _gloffset_SecondaryColor3usvEXT),
+    NAME_FUNC_OFFSET(18544, glSecondaryColorPointerEXT, glSecondaryColorPointerEXT, NULL, _gloffset_SecondaryColorPointerEXT),
+    NAME_FUNC_OFFSET(18568, glMultiDrawArraysEXT, glMultiDrawArraysEXT, NULL, _gloffset_MultiDrawArraysEXT),
+    NAME_FUNC_OFFSET(18586, glMultiDrawElementsEXT, glMultiDrawElementsEXT, NULL, _gloffset_MultiDrawElementsEXT),
+    NAME_FUNC_OFFSET(18606, glFogCoordPointerEXT, glFogCoordPointerEXT, NULL, _gloffset_FogCoordPointerEXT),
+    NAME_FUNC_OFFSET(18624, glFogCoorddEXT, glFogCoorddEXT, NULL, _gloffset_FogCoorddEXT),
+    NAME_FUNC_OFFSET(18636, glFogCoorddvEXT, glFogCoorddvEXT, NULL, _gloffset_FogCoorddvEXT),
+    NAME_FUNC_OFFSET(18649, glFogCoordfEXT, glFogCoordfEXT, NULL, _gloffset_FogCoordfEXT),
+    NAME_FUNC_OFFSET(18661, glFogCoordfvEXT, glFogCoordfvEXT, NULL, _gloffset_FogCoordfvEXT),
+    NAME_FUNC_OFFSET(18674, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
+    NAME_FUNC_OFFSET(18694, glBlendFuncSeparateEXT, glBlendFuncSeparateEXT, NULL, _gloffset_BlendFuncSeparateEXT),
+    NAME_FUNC_OFFSET(18718, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
+    NAME_FUNC_OFFSET(18732, glWindowPos2dMESA, glWindowPos2dMESA, NULL, _gloffset_WindowPos2dMESA),
+    NAME_FUNC_OFFSET(18749, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
+    NAME_FUNC_OFFSET(18764, glWindowPos2dvMESA, glWindowPos2dvMESA, NULL, _gloffset_WindowPos2dvMESA),
+    NAME_FUNC_OFFSET(18782, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
+    NAME_FUNC_OFFSET(18796, glWindowPos2fMESA, glWindowPos2fMESA, NULL, _gloffset_WindowPos2fMESA),
+    NAME_FUNC_OFFSET(18813, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
+    NAME_FUNC_OFFSET(18828, glWindowPos2fvMESA, glWindowPos2fvMESA, NULL, _gloffset_WindowPos2fvMESA),
+    NAME_FUNC_OFFSET(18846, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
+    NAME_FUNC_OFFSET(18860, glWindowPos2iMESA, glWindowPos2iMESA, NULL, _gloffset_WindowPos2iMESA),
+    NAME_FUNC_OFFSET(18877, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
+    NAME_FUNC_OFFSET(18892, glWindowPos2ivMESA, glWindowPos2ivMESA, NULL, _gloffset_WindowPos2ivMESA),
+    NAME_FUNC_OFFSET(18910, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
+    NAME_FUNC_OFFSET(18924, glWindowPos2sMESA, glWindowPos2sMESA, NULL, _gloffset_WindowPos2sMESA),
+    NAME_FUNC_OFFSET(18941, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
+    NAME_FUNC_OFFSET(18956, glWindowPos2svMESA, glWindowPos2svMESA, NULL, _gloffset_WindowPos2svMESA),
+    NAME_FUNC_OFFSET(18974, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
+    NAME_FUNC_OFFSET(18988, glWindowPos3dMESA, glWindowPos3dMESA, NULL, _gloffset_WindowPos3dMESA),
+    NAME_FUNC_OFFSET(19005, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
+    NAME_FUNC_OFFSET(19020, glWindowPos3dvMESA, glWindowPos3dvMESA, NULL, _gloffset_WindowPos3dvMESA),
+    NAME_FUNC_OFFSET(19038, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
+    NAME_FUNC_OFFSET(19052, glWindowPos3fMESA, glWindowPos3fMESA, NULL, _gloffset_WindowPos3fMESA),
+    NAME_FUNC_OFFSET(19069, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
+    NAME_FUNC_OFFSET(19084, glWindowPos3fvMESA, glWindowPos3fvMESA, NULL, _gloffset_WindowPos3fvMESA),
+    NAME_FUNC_OFFSET(19102, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
+    NAME_FUNC_OFFSET(19116, glWindowPos3iMESA, glWindowPos3iMESA, NULL, _gloffset_WindowPos3iMESA),
+    NAME_FUNC_OFFSET(19133, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
+    NAME_FUNC_OFFSET(19148, glWindowPos3ivMESA, glWindowPos3ivMESA, NULL, _gloffset_WindowPos3ivMESA),
+    NAME_FUNC_OFFSET(19166, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
+    NAME_FUNC_OFFSET(19180, glWindowPos3sMESA, glWindowPos3sMESA, NULL, _gloffset_WindowPos3sMESA),
+    NAME_FUNC_OFFSET(19197, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
+    NAME_FUNC_OFFSET(19212, glWindowPos3svMESA, glWindowPos3svMESA, NULL, _gloffset_WindowPos3svMESA),
+    NAME_FUNC_OFFSET(19230, glBindProgramNV, glBindProgramNV, NULL, _gloffset_BindProgramNV),
+    NAME_FUNC_OFFSET(19247, glDeleteProgramsNV, glDeleteProgramsNV, NULL, _gloffset_DeleteProgramsNV),
+    NAME_FUNC_OFFSET(19267, glGenProgramsNV, glGenProgramsNV, NULL, _gloffset_GenProgramsNV),
+    NAME_FUNC_OFFSET(19284, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
+    NAME_FUNC_OFFSET(19310, glGetVertexAttribPointervNV, glGetVertexAttribPointervNV, NULL, _gloffset_GetVertexAttribPointervNV),
+    NAME_FUNC_OFFSET(19339, glIsProgramNV, glIsProgramNV, NULL, _gloffset_IsProgramNV),
+    NAME_FUNC_OFFSET(19354, glPointParameteriNV, glPointParameteriNV, NULL, _gloffset_PointParameteriNV),
+    NAME_FUNC_OFFSET(19372, glPointParameterivNV, glPointParameterivNV, NULL, _gloffset_PointParameterivNV),
+    NAME_FUNC_OFFSET(19391, gl_dispatch_stub_758, gl_dispatch_stub_758, NULL, _gloffset_DeleteVertexArraysAPPLE),
+    NAME_FUNC_OFFSET(19412, gl_dispatch_stub_760, gl_dispatch_stub_760, NULL, _gloffset_IsVertexArrayAPPLE),
+    NAME_FUNC_OFFSET(19428, gl_dispatch_stub_768, gl_dispatch_stub_768, NULL, _gloffset_BlendEquationSeparateEXT),
+    NAME_FUNC_OFFSET(19452, gl_dispatch_stub_768, gl_dispatch_stub_768, NULL, _gloffset_BlendEquationSeparateEXT),
+    NAME_FUNC_OFFSET(19479, glBindFramebufferEXT, glBindFramebufferEXT, NULL, _gloffset_BindFramebufferEXT),
+    NAME_FUNC_OFFSET(19497, glBindRenderbufferEXT, glBindRenderbufferEXT, NULL, _gloffset_BindRenderbufferEXT),
+    NAME_FUNC_OFFSET(19516, glCheckFramebufferStatusEXT, glCheckFramebufferStatusEXT, NULL, _gloffset_CheckFramebufferStatusEXT),
+    NAME_FUNC_OFFSET(19541, glDeleteFramebuffersEXT, glDeleteFramebuffersEXT, NULL, _gloffset_DeleteFramebuffersEXT),
+    NAME_FUNC_OFFSET(19562, glDeleteRenderbuffersEXT, glDeleteRenderbuffersEXT, NULL, _gloffset_DeleteRenderbuffersEXT),
+    NAME_FUNC_OFFSET(19584, glFramebufferRenderbufferEXT, glFramebufferRenderbufferEXT, NULL, _gloffset_FramebufferRenderbufferEXT),
+    NAME_FUNC_OFFSET(19610, glFramebufferTexture1DEXT, glFramebufferTexture1DEXT, NULL, _gloffset_FramebufferTexture1DEXT),
+    NAME_FUNC_OFFSET(19633, glFramebufferTexture2DEXT, glFramebufferTexture2DEXT, NULL, _gloffset_FramebufferTexture2DEXT),
+    NAME_FUNC_OFFSET(19656, glFramebufferTexture3DEXT, glFramebufferTexture3DEXT, NULL, _gloffset_FramebufferTexture3DEXT),
+    NAME_FUNC_OFFSET(19679, glGenFramebuffersEXT, glGenFramebuffersEXT, NULL, _gloffset_GenFramebuffersEXT),
+    NAME_FUNC_OFFSET(19697, glGenRenderbuffersEXT, glGenRenderbuffersEXT, NULL, _gloffset_GenRenderbuffersEXT),
+    NAME_FUNC_OFFSET(19716, glGenerateMipmapEXT, glGenerateMipmapEXT, NULL, _gloffset_GenerateMipmapEXT),
+    NAME_FUNC_OFFSET(19733, glGetFramebufferAttachmentParameterivEXT, glGetFramebufferAttachmentParameterivEXT, NULL, _gloffset_GetFramebufferAttachmentParameterivEXT),
+    NAME_FUNC_OFFSET(19771, glGetRenderbufferParameterivEXT, glGetRenderbufferParameterivEXT, NULL, _gloffset_GetRenderbufferParameterivEXT),
+    NAME_FUNC_OFFSET(19800, glIsFramebufferEXT, glIsFramebufferEXT, NULL, _gloffset_IsFramebufferEXT),
+    NAME_FUNC_OFFSET(19816, glIsRenderbufferEXT, glIsRenderbufferEXT, NULL, _gloffset_IsRenderbufferEXT),
+    NAME_FUNC_OFFSET(19833, glRenderbufferStorageEXT, glRenderbufferStorageEXT, NULL, _gloffset_RenderbufferStorageEXT),
+    NAME_FUNC_OFFSET(19855, gl_dispatch_stub_786, gl_dispatch_stub_786, NULL, _gloffset_BlitFramebufferEXT),
+    NAME_FUNC_OFFSET(19873, glFramebufferTextureLayerEXT, glFramebufferTextureLayerEXT, NULL, _gloffset_FramebufferTextureLayerEXT),
+    NAME_FUNC_OFFSET(19899, glProvokingVertexEXT, glProvokingVertexEXT, NULL, _gloffset_ProvokingVertexEXT),
     NAME_FUNC_OFFSET(-1, NULL, NULL, NULL, 0)
 };
 
diff --git a/src/mesa/main/buffers.c b/src/mesa/main/buffers.c
index 7f77c5d772..97f0659758 100644
--- a/src/mesa/main/buffers.c
+++ b/src/mesa/main/buffers.c
@@ -382,8 +382,6 @@ _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
     */
    if (n == 1) {
       GLuint count = 0, destMask0 = destMask[0];
-      /* init to -1 to help catch errors */
-      //fb->_ColorDrawBufferIndexes[0] = -1;
       while (destMask0) {
          GLint bufIndex = _mesa_ffs(destMask0) - 1;
          if (fb->_ColorDrawBufferIndexes[count] != bufIndex) {
diff --git a/src/mesa/main/colormac.h b/src/mesa/main/colormac.h
index 7ae781ae23..905f4e2283 100644
--- a/src/mesa/main/colormac.h
+++ b/src/mesa/main/colormac.h
@@ -210,6 +210,12 @@ do {						\
 #define PACK_COLOR_88_REV( L, A )					\
    (((A) << 8) | (L))
 
+#define PACK_COLOR_1616( L, A )						\
+   (((L) << 16) | (A))
+
+#define PACK_COLOR_1616_REV( L, A )					\
+   (((A) << 16) | (L))
+
 #define PACK_COLOR_332( R, G, B )					\
    (((R) & 0xe0) | (((G) & 0xe0) >> 3) | (((B) & 0xc0) >> 6))
 
diff --git a/src/mesa/main/config.h b/src/mesa/main/config.h
index a1465e02ee..08dd36c1e5 100644
--- a/src/mesa/main/config.h
+++ b/src/mesa/main/config.h
@@ -205,7 +205,7 @@
 #define MAX_VARYING                    16    /**< number of float[4] vectors */
 #define MAX_SAMPLERS                   MAX_TEXTURE_IMAGE_UNITS
 #define MAX_PROGRAM_INPUTS             32
-#define MAX_PROGRAM_OUTPUTS            32
+#define MAX_PROGRAM_OUTPUTS            64
 /*@}*/
 
 /** For GL_ARB_vertex_program */
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index 46a1814a79..ff1ab4bca1 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -619,9 +619,11 @@ _mesa_init_constants(GLcontext *ctx)
    ASSERT(MAX_NV_VERTEX_PROGRAM_INPUTS <= VERT_ATTRIB_MAX);
    ASSERT(MAX_NV_VERTEX_PROGRAM_OUTPUTS <= VERT_RESULT_MAX);
 
-   /* check that we don't exceed various 32-bit bitfields */
-   ASSERT(VERT_RESULT_MAX <= 32);
-   ASSERT(FRAG_ATTRIB_MAX <= 32);
+   /* check that we don't exceed the size of various bitfields */
+   ASSERT(VERT_RESULT_MAX <=
+	  (8 * sizeof(ctx->VertexProgram._Current->Base.OutputsWritten)));
+   ASSERT(FRAG_ATTRIB_MAX <=
+	  (8 * sizeof(ctx->FragmentProgram._Current->Base.InputsRead)));
 }
 
 
diff --git a/src/mesa/main/enums.c b/src/mesa/main/enums.c
index cfbcd306f8..7e10afcf15 100644
--- a/src/mesa/main/enums.c
+++ b/src/mesa/main/enums.c
@@ -384,6 +384,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_DELETE_STATUS\0"
    "GL_DEPTH\0"
    "GL_DEPTH24_STENCIL8\0"
+   "GL_DEPTH24_STENCIL8_EXT\0"
    "GL_DEPTH_ATTACHMENT\0"
    "GL_DEPTH_ATTACHMENT_EXT\0"
    "GL_DEPTH_BIAS\0"
@@ -409,6 +410,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_DEPTH_SCALE\0"
    "GL_DEPTH_STENCIL\0"
    "GL_DEPTH_STENCIL_ATTACHMENT\0"
+   "GL_DEPTH_STENCIL_EXT\0"
    "GL_DEPTH_STENCIL_NV\0"
    "GL_DEPTH_STENCIL_TO_BGRA_NV\0"
    "GL_DEPTH_STENCIL_TO_RGBA_NV\0"
@@ -478,6 +480,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_DRAW_BUFFER9_ARB\0"
    "GL_DRAW_BUFFER9_ATI\0"
    "GL_DRAW_FRAMEBUFFER\0"
+   "GL_DRAW_FRAMEBUFFER_BINDING\0"
    "GL_DRAW_FRAMEBUFFER_BINDING_EXT\0"
    "GL_DRAW_FRAMEBUFFER_EXT\0"
    "GL_DRAW_PIXEL_TOKEN\0"
@@ -594,6 +597,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT\0"
    "GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL\0"
    "GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT\0"
+   "GL_FRAMEBUFFER_BINDING\0"
    "GL_FRAMEBUFFER_BINDING_EXT\0"
    "GL_FRAMEBUFFER_COMPLETE\0"
    "GL_FRAMEBUFFER_COMPLETE_EXT\0"
@@ -602,6 +606,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT\0"
+   "GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER\0"
    "GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT\0"
@@ -610,6 +615,8 @@ LONGSTRING static const char enum_string_table[] =
    "GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT\0"
    "GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE\0"
+   "GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT\0"
+   "GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER\0"
    "GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT\0"
    "GL_FRAMEBUFFER_STATUS_ERROR_EXT\0"
    "GL_FRAMEBUFFER_UNDEFINED\0"
@@ -901,6 +908,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_MAX_CLIPMAP_DEPTH_SGIX\0"
    "GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX\0"
    "GL_MAX_CLIP_PLANES\0"
+   "GL_MAX_COLOR_ATTACHMENTS\0"
    "GL_MAX_COLOR_ATTACHMENTS_EXT\0"
    "GL_MAX_COLOR_MATRIX_STACK_DEPTH\0"
    "GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI\0"
@@ -961,8 +969,10 @@ LONGSTRING static const char enum_string_table[] =
    "GL_MAX_PROJECTION_STACK_DEPTH\0"
    "GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB\0"
    "GL_MAX_RECTANGLE_TEXTURE_SIZE_NV\0"
+   "GL_MAX_RENDERBUFFER_SIZE\0"
    "GL_MAX_RENDERBUFFER_SIZE_EXT\0"
    "GL_MAX_SAMPLES\0"
+   "GL_MAX_SAMPLES_EXT\0"
    "GL_MAX_SERVER_WAIT_TIMEOUT\0"
    "GL_MAX_SHININESS_NV\0"
    "GL_MAX_SPOT_EXPONENT_NV\0"
@@ -1344,6 +1354,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_RASTER_POSITION_UNCLIPPED_IBM\0"
    "GL_READ_BUFFER\0"
    "GL_READ_FRAMEBUFFER\0"
+   "GL_READ_FRAMEBUFFER_BINDING\0"
    "GL_READ_FRAMEBUFFER_BINDING_EXT\0"
    "GL_READ_FRAMEBUFFER_EXT\0"
    "GL_READ_ONLY\0"
@@ -1362,6 +1373,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_RENDER\0"
    "GL_RENDERBUFFER\0"
    "GL_RENDERBUFFER_ALPHA_SIZE\0"
+   "GL_RENDERBUFFER_BINDING\0"
    "GL_RENDERBUFFER_BINDING_EXT\0"
    "GL_RENDERBUFFER_BLUE_SIZE\0"
    "GL_RENDERBUFFER_DEPTH_SIZE\0"
@@ -1373,6 +1385,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_RENDERBUFFER_INTERNAL_FORMAT_EXT\0"
    "GL_RENDERBUFFER_RED_SIZE\0"
    "GL_RENDERBUFFER_SAMPLES\0"
+   "GL_RENDERBUFFER_SAMPLES_EXT\0"
    "GL_RENDERBUFFER_STENCIL_SIZE\0"
    "GL_RENDERBUFFER_WIDTH\0"
    "GL_RENDERBUFFER_WIDTH_EXT\0"
@@ -1558,9 +1571,13 @@ LONGSTRING static const char enum_string_table[] =
    "GL_STENCIL_FAIL\0"
    "GL_STENCIL_FUNC\0"
    "GL_STENCIL_INDEX\0"
+   "GL_STENCIL_INDEX1\0"
+   "GL_STENCIL_INDEX16\0"
    "GL_STENCIL_INDEX16_EXT\0"
    "GL_STENCIL_INDEX1_EXT\0"
+   "GL_STENCIL_INDEX4\0"
    "GL_STENCIL_INDEX4_EXT\0"
+   "GL_STENCIL_INDEX8\0"
    "GL_STENCIL_INDEX8_EXT\0"
    "GL_STENCIL_INDEX_EXT\0"
    "GL_STENCIL_PASS_DEPTH_FAIL\0"
@@ -1778,6 +1795,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_TEXTURE_RESIDENT\0"
    "GL_TEXTURE_STACK_DEPTH\0"
    "GL_TEXTURE_STENCIL_SIZE\0"
+   "GL_TEXTURE_STENCIL_SIZE_EXT\0"
    "GL_TEXTURE_STORAGE_HINT_APPLE\0"
    "GL_TEXTURE_TOO_LARGE_EXT\0"
    "GL_TEXTURE_UNSIGNED_REMAP_MODE_NV\0"
@@ -1823,6 +1841,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_UNSIGNED_INT\0"
    "GL_UNSIGNED_INT_10_10_10_2\0"
    "GL_UNSIGNED_INT_24_8\0"
+   "GL_UNSIGNED_INT_24_8_EXT\0"
    "GL_UNSIGNED_INT_24_8_NV\0"
    "GL_UNSIGNED_INT_2_10_10_10_REV\0"
    "GL_UNSIGNED_INT_8_8_8_8\0"
@@ -1919,7 +1938,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_ZOOM_Y\0"
    ;
 
-static const enum_elt all_enums[1881] =
+static const enum_elt all_enums[1900] =
 {
    {     0, 0x00000600 }, /* GL_2D */
    {     6, 0x00001407 }, /* GL_2_BYTES */
@@ -2269,1623 +2288,1642 @@ static const enum_elt all_enums[1881] =
    {  7144, 0x00008B80 }, /* GL_DELETE_STATUS */
    {  7161, 0x00001801 }, /* GL_DEPTH */
    {  7170, 0x000088F0 }, /* GL_DEPTH24_STENCIL8 */
-   {  7190, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT */
-   {  7210, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT_EXT */
-   {  7234, 0x00000D1F }, /* GL_DEPTH_BIAS */
-   {  7248, 0x00000D56 }, /* GL_DEPTH_BITS */
-   {  7262, 0x00008891 }, /* GL_DEPTH_BOUNDS_EXT */
-   {  7282, 0x00008890 }, /* GL_DEPTH_BOUNDS_TEST_EXT */
-   {  7307, 0x00000100 }, /* GL_DEPTH_BUFFER_BIT */
-   {  7327, 0x0000864F }, /* GL_DEPTH_CLAMP */
-   {  7342, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
-   {  7360, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
-   {  7381, 0x00001902 }, /* GL_DEPTH_COMPONENT */
-   {  7400, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
-   {  7421, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
-   {  7446, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
-   {  7472, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
-   {  7493, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
-   {  7518, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
-   {  7544, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
-   {  7565, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
-   {  7590, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
-   {  7616, 0x00000B74 }, /* GL_DEPTH_FUNC */
-   {  7630, 0x00000B70 }, /* GL_DEPTH_RANGE */
-   {  7645, 0x00000D1E }, /* GL_DEPTH_SCALE */
-   {  7660, 0x000084F9 }, /* GL_DEPTH_STENCIL */
-   {  7677, 0x0000821A }, /* GL_DEPTH_STENCIL_ATTACHMENT */
-   {  7705, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
-   {  7725, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-   {  7753, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-   {  7781, 0x00000B71 }, /* GL_DEPTH_TEST */
-   {  7795, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
-   {  7817, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
-   {  7843, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
-   {  7862, 0x00001201 }, /* GL_DIFFUSE */
-   {  7873, 0x00000BD0 }, /* GL_DITHER */
-   {  7883, 0x00000A02 }, /* GL_DOMAIN */
-   {  7893, 0x00001100 }, /* GL_DONT_CARE */
-   {  7906, 0x000086AE }, /* GL_DOT3_RGB */
-   {  7918, 0x000086AF }, /* GL_DOT3_RGBA */
-   {  7931, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
-   {  7948, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
-   {  7965, 0x000086AE }, /* GL_DOT3_RGB_ARB */
-   {  7981, 0x00008740 }, /* GL_DOT3_RGB_EXT */
-   {  7997, 0x0000140A }, /* GL_DOUBLE */
-   {  8007, 0x00000C32 }, /* GL_DOUBLEBUFFER */
-   {  8023, 0x00000C01 }, /* GL_DRAW_BUFFER */
-   {  8038, 0x00008825 }, /* GL_DRAW_BUFFER0 */
-   {  8054, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
-   {  8074, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
-   {  8094, 0x00008826 }, /* GL_DRAW_BUFFER1 */
-   {  8110, 0x0000882F }, /* GL_DRAW_BUFFER10 */
-   {  8127, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
-   {  8148, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
-   {  8169, 0x00008830 }, /* GL_DRAW_BUFFER11 */
-   {  8186, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
-   {  8207, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
-   {  8228, 0x00008831 }, /* GL_DRAW_BUFFER12 */
-   {  8245, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
-   {  8266, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
-   {  8287, 0x00008832 }, /* GL_DRAW_BUFFER13 */
-   {  8304, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
-   {  8325, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
-   {  8346, 0x00008833 }, /* GL_DRAW_BUFFER14 */
-   {  8363, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
-   {  8384, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
-   {  8405, 0x00008834 }, /* GL_DRAW_BUFFER15 */
-   {  8422, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
-   {  8443, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
-   {  8464, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
-   {  8484, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
-   {  8504, 0x00008827 }, /* GL_DRAW_BUFFER2 */
-   {  8520, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
-   {  8540, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
-   {  8560, 0x00008828 }, /* GL_DRAW_BUFFER3 */
-   {  8576, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
-   {  8596, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
-   {  8616, 0x00008829 }, /* GL_DRAW_BUFFER4 */
-   {  8632, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
-   {  8652, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
-   {  8672, 0x0000882A }, /* GL_DRAW_BUFFER5 */
-   {  8688, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
-   {  8708, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
-   {  8728, 0x0000882B }, /* GL_DRAW_BUFFER6 */
-   {  8744, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
-   {  8764, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
-   {  8784, 0x0000882C }, /* GL_DRAW_BUFFER7 */
-   {  8800, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
-   {  8820, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
-   {  8840, 0x0000882D }, /* GL_DRAW_BUFFER8 */
-   {  8856, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
-   {  8876, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
-   {  8896, 0x0000882E }, /* GL_DRAW_BUFFER9 */
-   {  8912, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
-   {  8932, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
-   {  8952, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER */
-   {  8972, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-   {  9004, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
-   {  9028, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
-   {  9048, 0x00000304 }, /* GL_DST_ALPHA */
-   {  9061, 0x00000306 }, /* GL_DST_COLOR */
-   {  9074, 0x0000877A }, /* GL_DU8DV8_ATI */
-   {  9088, 0x00008779 }, /* GL_DUDV_ATI */
-   {  9100, 0x000088EA }, /* GL_DYNAMIC_COPY */
-   {  9116, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
-   {  9136, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
-   {  9152, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
-   {  9172, 0x000088E9 }, /* GL_DYNAMIC_READ */
-   {  9188, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
-   {  9208, 0x00000B43 }, /* GL_EDGE_FLAG */
-   {  9221, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
-   {  9240, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-   {  9274, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
-   {  9312, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
-   {  9339, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-   {  9365, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
-   {  9389, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-   {  9421, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
-   {  9457, 0x00001600 }, /* GL_EMISSION */
-   {  9469, 0x00002000 }, /* GL_ENABLE_BIT */
-   {  9483, 0x00000202 }, /* GL_EQUAL */
-   {  9492, 0x00001509 }, /* GL_EQUIV */
-   {  9501, 0x00010000 }, /* GL_EVAL_BIT */
-   {  9513, 0x00000800 }, /* GL_EXP */
-   {  9520, 0x00000801 }, /* GL_EXP2 */
-   {  9528, 0x00001F03 }, /* GL_EXTENSIONS */
-   {  9542, 0x00002400 }, /* GL_EYE_LINEAR */
-   {  9556, 0x00002502 }, /* GL_EYE_PLANE */
-   {  9569, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
-   {  9594, 0x0000855B }, /* GL_EYE_RADIAL_NV */
-   {  9611, 0x00000000 }, /* GL_FALSE */
-   {  9620, 0x00001101 }, /* GL_FASTEST */
-   {  9631, 0x00001C01 }, /* GL_FEEDBACK */
-   {  9643, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
-   {  9670, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
-   {  9694, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
-   {  9718, 0x00001B02 }, /* GL_FILL */
-   {  9726, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION */
-   {  9753, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */
-   {  9784, 0x00001D00 }, /* GL_FLAT */
-   {  9792, 0x00001406 }, /* GL_FLOAT */
-   {  9801, 0x00008B5A }, /* GL_FLOAT_MAT2 */
-   {  9815, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
-   {  9833, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */
-   {  9849, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */
-   {  9865, 0x00008B5B }, /* GL_FLOAT_MAT3 */
-   {  9879, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
-   {  9897, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */
-   {  9913, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */
-   {  9929, 0x00008B5C }, /* GL_FLOAT_MAT4 */
-   {  9943, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
-   {  9961, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */
-   {  9977, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */
-   {  9993, 0x00008B50 }, /* GL_FLOAT_VEC2 */
-   { 10007, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
-   { 10025, 0x00008B51 }, /* GL_FLOAT_VEC3 */
-   { 10039, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
-   { 10057, 0x00008B52 }, /* GL_FLOAT_VEC4 */
-   { 10071, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
-   { 10089, 0x00000B60 }, /* GL_FOG */
-   { 10096, 0x00000080 }, /* GL_FOG_BIT */
-   { 10107, 0x00000B66 }, /* GL_FOG_COLOR */
-   { 10120, 0x00008451 }, /* GL_FOG_COORD */
-   { 10133, 0x00008451 }, /* GL_FOG_COORDINATE */
-   { 10151, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
-   { 10175, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-   { 10214, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
-   { 10257, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-   { 10289, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-   { 10320, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-   { 10349, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
-   { 10374, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
-   { 10393, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
-   { 10427, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
-   { 10454, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
-   { 10480, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
-   { 10504, 0x00008450 }, /* GL_FOG_COORD_SRC */
-   { 10521, 0x00000B62 }, /* GL_FOG_DENSITY */
-   { 10536, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
-   { 10560, 0x00000B64 }, /* GL_FOG_END */
-   { 10571, 0x00000C54 }, /* GL_FOG_HINT */
-   { 10583, 0x00000B61 }, /* GL_FOG_INDEX */
-   { 10596, 0x00000B65 }, /* GL_FOG_MODE */
-   { 10608, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
-   { 10627, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
-   { 10652, 0x00000B63 }, /* GL_FOG_START */
-   { 10665, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
-   { 10683, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
-   { 10707, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
-   { 10726, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
-   { 10749, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-   { 10784, 0x00008D40 }, /* GL_FRAMEBUFFER */
-   { 10799, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
-   { 10836, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
-   { 10872, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
-   { 10913, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
-   { 10954, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
-   { 10991, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
-   { 11028, 0x00008DA7 }, /* GL_FRAMEBUFFER_ATTACHMENT_LAYERED_ARB */
-   { 11066, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
-   { 11104, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
-   { 11146, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
-   { 11184, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
-   { 11226, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
-   { 11261, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
-   { 11300, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
-   { 11349, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
-   { 11397, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
-   { 11449, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
-   { 11489, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
-   { 11533, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
-   { 11573, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
-   { 11617, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
-   { 11644, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */
-   { 11668, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
-   { 11696, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */
-   { 11719, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
-   { 11738, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
-   { 11775, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
-   { 11816, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-   { 11857, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-   { 11899, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-   { 11950, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-   { 11988, 0x00008DA9 }, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_ARB */
-   { 12030, 0x00008DA8 }, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_ARB */
-   { 12074, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
-   { 12119, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
-   { 12168, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
-   { 12206, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-   { 12248, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-   { 12280, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */
-   { 12305, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */
-   { 12332, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
-   { 12363, 0x00000404 }, /* GL_FRONT */
-   { 12372, 0x00000408 }, /* GL_FRONT_AND_BACK */
-   { 12390, 0x00000B46 }, /* GL_FRONT_FACE */
-   { 12404, 0x00000400 }, /* GL_FRONT_LEFT */
-   { 12418, 0x00000401 }, /* GL_FRONT_RIGHT */
-   { 12433, 0x00008006 }, /* GL_FUNC_ADD */
-   { 12445, 0x00008006 }, /* GL_FUNC_ADD_EXT */
-   { 12461, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
-   { 12486, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
-   { 12515, 0x0000800A }, /* GL_FUNC_SUBTRACT */
-   { 12532, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
-   { 12553, 0x00008191 }, /* GL_GENERATE_MIPMAP */
-   { 12572, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
-   { 12596, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
-   { 12625, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
-   { 12649, 0x00008DDB }, /* GL_GEOMETRY_INPUT_TYPE_ARB */
-   { 12676, 0x00008DDC }, /* GL_GEOMETRY_OUTPUT_TYPE_ARB */
-   { 12704, 0x00008DD9 }, /* GL_GEOMETRY_SHADER_ARB */
-   { 12727, 0x00008DDA }, /* GL_GEOMETRY_VERTICES_OUT_ARB */
-   { 12756, 0x00000206 }, /* GL_GEQUAL */
-   { 12766, 0x00000204 }, /* GL_GREATER */
-   { 12777, 0x00001904 }, /* GL_GREEN */
-   { 12786, 0x00000D19 }, /* GL_GREEN_BIAS */
-   { 12800, 0x00000D53 }, /* GL_GREEN_BITS */
-   { 12814, 0x00000D18 }, /* GL_GREEN_SCALE */
-   { 12829, 0x00008000 }, /* GL_HINT_BIT */
-   { 12841, 0x00008024 }, /* GL_HISTOGRAM */
-   { 12854, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
-   { 12878, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
-   { 12906, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
-   { 12929, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
-   { 12956, 0x00008024 }, /* GL_HISTOGRAM_EXT */
-   { 12973, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
-   { 12993, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
-   { 13017, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
-   { 13041, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
-   { 13069, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-   { 13097, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
-   { 13129, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
-   { 13151, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
-   { 13177, 0x0000802D }, /* GL_HISTOGRAM_SINK */
-   { 13195, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
-   { 13217, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
-   { 13236, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
-   { 13259, 0x0000862A }, /* GL_IDENTITY_NV */
-   { 13274, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
-   { 13294, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-   { 13334, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-   { 13372, 0x00001E02 }, /* GL_INCR */
-   { 13380, 0x00008507 }, /* GL_INCR_WRAP */
-   { 13393, 0x00008507 }, /* GL_INCR_WRAP_EXT */
-   { 13410, 0x00008222 }, /* GL_INDEX */
-   { 13419, 0x00008077 }, /* GL_INDEX_ARRAY */
-   { 13434, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-   { 13464, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
-   { 13498, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
-   { 13521, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
-   { 13543, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
-   { 13563, 0x00000D51 }, /* GL_INDEX_BITS */
-   { 13577, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
-   { 13598, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
-   { 13616, 0x00000C30 }, /* GL_INDEX_MODE */
-   { 13630, 0x00000D13 }, /* GL_INDEX_OFFSET */
-   { 13646, 0x00000D12 }, /* GL_INDEX_SHIFT */
-   { 13661, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
-   { 13680, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
-   { 13699, 0x00001404 }, /* GL_INT */
-   { 13706, 0x00008049 }, /* GL_INTENSITY */
-   { 13719, 0x0000804C }, /* GL_INTENSITY12 */
-   { 13734, 0x0000804C }, /* GL_INTENSITY12_EXT */
-   { 13753, 0x0000804D }, /* GL_INTENSITY16 */
-   { 13768, 0x0000804D }, /* GL_INTENSITY16_EXT */
-   { 13787, 0x0000804A }, /* GL_INTENSITY4 */
-   { 13801, 0x0000804A }, /* GL_INTENSITY4_EXT */
-   { 13819, 0x0000804B }, /* GL_INTENSITY8 */
-   { 13833, 0x0000804B }, /* GL_INTENSITY8_EXT */
-   { 13851, 0x00008049 }, /* GL_INTENSITY_EXT */
-   { 13868, 0x00008575 }, /* GL_INTERPOLATE */
-   { 13883, 0x00008575 }, /* GL_INTERPOLATE_ARB */
-   { 13902, 0x00008575 }, /* GL_INTERPOLATE_EXT */
-   { 13921, 0x00008B53 }, /* GL_INT_VEC2 */
-   { 13933, 0x00008B53 }, /* GL_INT_VEC2_ARB */
-   { 13949, 0x00008B54 }, /* GL_INT_VEC3 */
-   { 13961, 0x00008B54 }, /* GL_INT_VEC3_ARB */
-   { 13977, 0x00008B55 }, /* GL_INT_VEC4 */
-   { 13989, 0x00008B55 }, /* GL_INT_VEC4_ARB */
-   { 14005, 0x00000500 }, /* GL_INVALID_ENUM */
-   { 14021, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */
-   { 14054, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
-   { 14091, 0x00000502 }, /* GL_INVALID_OPERATION */
-   { 14112, 0x00000501 }, /* GL_INVALID_VALUE */
-   { 14129, 0x0000862B }, /* GL_INVERSE_NV */
-   { 14143, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
-   { 14167, 0x0000150A }, /* GL_INVERT */
-   { 14177, 0x00001E00 }, /* GL_KEEP */
-   { 14185, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION */
-   { 14211, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */
-   { 14241, 0x00000406 }, /* GL_LEFT */
-   { 14249, 0x00000203 }, /* GL_LEQUAL */
-   { 14259, 0x00000201 }, /* GL_LESS */
-   { 14267, 0x00004000 }, /* GL_LIGHT0 */
-   { 14277, 0x00004001 }, /* GL_LIGHT1 */
-   { 14287, 0x00004002 }, /* GL_LIGHT2 */
-   { 14297, 0x00004003 }, /* GL_LIGHT3 */
-   { 14307, 0x00004004 }, /* GL_LIGHT4 */
-   { 14317, 0x00004005 }, /* GL_LIGHT5 */
-   { 14327, 0x00004006 }, /* GL_LIGHT6 */
-   { 14337, 0x00004007 }, /* GL_LIGHT7 */
-   { 14347, 0x00000B50 }, /* GL_LIGHTING */
-   { 14359, 0x00000040 }, /* GL_LIGHTING_BIT */
-   { 14375, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
-   { 14398, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-   { 14427, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
-   { 14460, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-   { 14488, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
-   { 14512, 0x00001B01 }, /* GL_LINE */
-   { 14520, 0x00002601 }, /* GL_LINEAR */
-   { 14530, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
-   { 14552, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-   { 14582, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-   { 14613, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
-   { 14637, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
-   { 14662, 0x00000001 }, /* GL_LINES */
-   { 14671, 0x0000000A }, /* GL_LINES_ADJACENCY_ARB */
-   { 14694, 0x00000004 }, /* GL_LINE_BIT */
-   { 14706, 0x00000002 }, /* GL_LINE_LOOP */
-   { 14719, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
-   { 14739, 0x00000B20 }, /* GL_LINE_SMOOTH */
-   { 14754, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
-   { 14774, 0x00000B24 }, /* GL_LINE_STIPPLE */
-   { 14790, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
-   { 14814, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
-   { 14837, 0x00000003 }, /* GL_LINE_STRIP */
-   { 14851, 0x0000000B }, /* GL_LINE_STRIP_ADJACENCY_ARB */
-   { 14879, 0x00000702 }, /* GL_LINE_TOKEN */
-   { 14893, 0x00000B21 }, /* GL_LINE_WIDTH */
-   { 14907, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
-   { 14933, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
-   { 14953, 0x00008B82 }, /* GL_LINK_STATUS */
-   { 14968, 0x00000B32 }, /* GL_LIST_BASE */
-   { 14981, 0x00020000 }, /* GL_LIST_BIT */
-   { 14993, 0x00000B33 }, /* GL_LIST_INDEX */
-   { 15007, 0x00000B30 }, /* GL_LIST_MODE */
-   { 15020, 0x00000101 }, /* GL_LOAD */
-   { 15028, 0x00000BF1 }, /* GL_LOGIC_OP */
-   { 15040, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
-   { 15057, 0x00008CA1 }, /* GL_LOWER_LEFT */
-   { 15071, 0x00001909 }, /* GL_LUMINANCE */
-   { 15084, 0x00008041 }, /* GL_LUMINANCE12 */
-   { 15099, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
-   { 15122, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
-   { 15149, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
-   { 15171, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
-   { 15197, 0x00008041 }, /* GL_LUMINANCE12_EXT */
-   { 15216, 0x00008042 }, /* GL_LUMINANCE16 */
-   { 15231, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
-   { 15254, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
-   { 15281, 0x00008042 }, /* GL_LUMINANCE16_EXT */
-   { 15300, 0x0000803F }, /* GL_LUMINANCE4 */
-   { 15314, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
-   { 15335, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
-   { 15360, 0x0000803F }, /* GL_LUMINANCE4_EXT */
-   { 15378, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
-   { 15399, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
-   { 15424, 0x00008040 }, /* GL_LUMINANCE8 */
-   { 15438, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
-   { 15459, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
-   { 15484, 0x00008040 }, /* GL_LUMINANCE8_EXT */
-   { 15502, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
-   { 15521, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
-   { 15537, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
-   { 15557, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
-   { 15579, 0x00000D91 }, /* GL_MAP1_INDEX */
-   { 15593, 0x00000D92 }, /* GL_MAP1_NORMAL */
-   { 15608, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
-   { 15632, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
-   { 15656, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
-   { 15680, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
-   { 15704, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
-   { 15721, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
-   { 15738, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-   { 15766, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-   { 15795, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-   { 15824, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-   { 15853, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-   { 15882, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-   { 15911, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-   { 15940, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-   { 15968, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-   { 15996, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-   { 16024, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-   { 16052, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-   { 16080, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-   { 16108, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-   { 16136, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-   { 16164, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-   { 16192, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
-   { 16208, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
-   { 16228, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
-   { 16250, 0x00000DB1 }, /* GL_MAP2_INDEX */
-   { 16264, 0x00000DB2 }, /* GL_MAP2_NORMAL */
-   { 16279, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
-   { 16303, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
-   { 16327, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
-   { 16351, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
-   { 16375, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
-   { 16392, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
-   { 16409, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-   { 16437, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-   { 16466, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-   { 16495, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-   { 16524, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-   { 16553, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-   { 16582, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-   { 16611, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-   { 16639, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-   { 16667, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-   { 16695, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-   { 16723, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-   { 16751, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-   { 16779, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
-   { 16807, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-   { 16835, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-   { 16863, 0x00000D10 }, /* GL_MAP_COLOR */
-   { 16876, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */
-   { 16902, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */
-   { 16931, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */
-   { 16959, 0x00000001 }, /* GL_MAP_READ_BIT */
-   { 16975, 0x00000D11 }, /* GL_MAP_STENCIL */
-   { 16990, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */
-   { 17016, 0x00000002 }, /* GL_MAP_WRITE_BIT */
-   { 17033, 0x000088C0 }, /* GL_MATRIX0_ARB */
-   { 17048, 0x00008630 }, /* GL_MATRIX0_NV */
-   { 17062, 0x000088CA }, /* GL_MATRIX10_ARB */
-   { 17078, 0x000088CB }, /* GL_MATRIX11_ARB */
-   { 17094, 0x000088CC }, /* GL_MATRIX12_ARB */
-   { 17110, 0x000088CD }, /* GL_MATRIX13_ARB */
-   { 17126, 0x000088CE }, /* GL_MATRIX14_ARB */
-   { 17142, 0x000088CF }, /* GL_MATRIX15_ARB */
-   { 17158, 0x000088D0 }, /* GL_MATRIX16_ARB */
-   { 17174, 0x000088D1 }, /* GL_MATRIX17_ARB */
-   { 17190, 0x000088D2 }, /* GL_MATRIX18_ARB */
-   { 17206, 0x000088D3 }, /* GL_MATRIX19_ARB */
-   { 17222, 0x000088C1 }, /* GL_MATRIX1_ARB */
-   { 17237, 0x00008631 }, /* GL_MATRIX1_NV */
-   { 17251, 0x000088D4 }, /* GL_MATRIX20_ARB */
-   { 17267, 0x000088D5 }, /* GL_MATRIX21_ARB */
-   { 17283, 0x000088D6 }, /* GL_MATRIX22_ARB */
-   { 17299, 0x000088D7 }, /* GL_MATRIX23_ARB */
-   { 17315, 0x000088D8 }, /* GL_MATRIX24_ARB */
-   { 17331, 0x000088D9 }, /* GL_MATRIX25_ARB */
-   { 17347, 0x000088DA }, /* GL_MATRIX26_ARB */
-   { 17363, 0x000088DB }, /* GL_MATRIX27_ARB */
-   { 17379, 0x000088DC }, /* GL_MATRIX28_ARB */
-   { 17395, 0x000088DD }, /* GL_MATRIX29_ARB */
-   { 17411, 0x000088C2 }, /* GL_MATRIX2_ARB */
-   { 17426, 0x00008632 }, /* GL_MATRIX2_NV */
-   { 17440, 0x000088DE }, /* GL_MATRIX30_ARB */
-   { 17456, 0x000088DF }, /* GL_MATRIX31_ARB */
-   { 17472, 0x000088C3 }, /* GL_MATRIX3_ARB */
-   { 17487, 0x00008633 }, /* GL_MATRIX3_NV */
-   { 17501, 0x000088C4 }, /* GL_MATRIX4_ARB */
-   { 17516, 0x00008634 }, /* GL_MATRIX4_NV */
-   { 17530, 0x000088C5 }, /* GL_MATRIX5_ARB */
-   { 17545, 0x00008635 }, /* GL_MATRIX5_NV */
-   { 17559, 0x000088C6 }, /* GL_MATRIX6_ARB */
-   { 17574, 0x00008636 }, /* GL_MATRIX6_NV */
-   { 17588, 0x000088C7 }, /* GL_MATRIX7_ARB */
-   { 17603, 0x00008637 }, /* GL_MATRIX7_NV */
-   { 17617, 0x000088C8 }, /* GL_MATRIX8_ARB */
-   { 17632, 0x000088C9 }, /* GL_MATRIX9_ARB */
-   { 17647, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
-   { 17673, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-   { 17707, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-   { 17738, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-   { 17771, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-   { 17802, 0x00000BA0 }, /* GL_MATRIX_MODE */
-   { 17817, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
-   { 17839, 0x00008008 }, /* GL_MAX */
-   { 17846, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
-   { 17869, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-   { 17901, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
-   { 17927, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-   { 17960, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-   { 17986, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 18020, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
-   { 18039, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
-   { 18068, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-   { 18100, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
-   { 18136, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-   { 18172, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
-   { 18212, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
-   { 18238, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
-   { 18268, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
-   { 18293, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
-   { 18322, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-   { 18351, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
-   { 18384, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
-   { 18404, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
-   { 18428, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
-   { 18452, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
-   { 18476, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
-   { 18501, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
-   { 18519, 0x00008008 }, /* GL_MAX_EXT */
-   { 18530, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-   { 18565, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
-   { 18604, 0x00008DE0 }, /* GL_MAX_GEOMETRY_OUTPUT_VERTICES_ARB */
-   { 18640, 0x00008C29 }, /* GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_ARB */
-   { 18680, 0x00008DE1 }, /* GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_ARB */
-   { 18724, 0x00008DDF }, /* GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_ARB */
-   { 18763, 0x00008DDD }, /* GL_MAX_GEOMETRY_VARYING_COMPONENTS_ARB */
-   { 18802, 0x00000D31 }, /* GL_MAX_LIGHTS */
-   { 18816, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
-   { 18836, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-   { 18874, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-   { 18903, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
-   { 18927, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
-   { 18955, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
-   { 18978, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 19015, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 19051, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-   { 19078, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-   { 19107, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-   { 19141, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
-   { 19177, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-   { 19204, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-   { 19236, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-   { 19272, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-   { 19301, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-   { 19330, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
-   { 19358, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-   { 19396, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 19440, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 19483, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 19517, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 19556, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 19593, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 19631, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 19674, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 19717, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-   { 19747, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-   { 19778, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 19814, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 19850, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
-   { 19880, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-   { 19914, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
-   { 19947, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
-   { 19976, 0x00008D57 }, /* GL_MAX_SAMPLES */
-   { 19991, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */
-   { 20018, 0x00008504 }, /* GL_MAX_SHININESS_NV */
-   { 20038, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
-   { 20062, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
-   { 20084, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
-   { 20110, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-   { 20137, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
-   { 20168, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
-   { 20192, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 20226, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
-   { 20246, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
-   { 20273, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
-   { 20294, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
-   { 20319, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
-   { 20344, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
-   { 20379, 0x00008B4B }, /* GL_MAX_VARYING_COMPONENTS */
-   { 20405, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
-   { 20427, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
-   { 20453, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
-   { 20475, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
-   { 20501, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-   { 20535, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
-   { 20573, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-   { 20606, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
-   { 20643, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
-   { 20667, 0x00008DDE }, /* GL_MAX_VERTEX_VARYING_COMPONENTS_ARB */
-   { 20704, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
-   { 20725, 0x00008007 }, /* GL_MIN */
-   { 20732, 0x0000802E }, /* GL_MINMAX */
-   { 20742, 0x0000802E }, /* GL_MINMAX_EXT */
-   { 20756, 0x0000802F }, /* GL_MINMAX_FORMAT */
-   { 20773, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
-   { 20794, 0x00008030 }, /* GL_MINMAX_SINK */
-   { 20809, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
-   { 20828, 0x00008007 }, /* GL_MIN_EXT */
-   { 20839, 0x00008370 }, /* GL_MIRRORED_REPEAT */
-   { 20858, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
-   { 20881, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
-   { 20904, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
-   { 20924, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
-   { 20944, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-   { 20974, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
-   { 21002, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-   { 21030, 0x00001700 }, /* GL_MODELVIEW */
-   { 21043, 0x00001700 }, /* GL_MODELVIEW0_ARB */
-   { 21061, 0x0000872A }, /* GL_MODELVIEW10_ARB */
-   { 21080, 0x0000872B }, /* GL_MODELVIEW11_ARB */
-   { 21099, 0x0000872C }, /* GL_MODELVIEW12_ARB */
-   { 21118, 0x0000872D }, /* GL_MODELVIEW13_ARB */
-   { 21137, 0x0000872E }, /* GL_MODELVIEW14_ARB */
-   { 21156, 0x0000872F }, /* GL_MODELVIEW15_ARB */
-   { 21175, 0x00008730 }, /* GL_MODELVIEW16_ARB */
-   { 21194, 0x00008731 }, /* GL_MODELVIEW17_ARB */
-   { 21213, 0x00008732 }, /* GL_MODELVIEW18_ARB */
-   { 21232, 0x00008733 }, /* GL_MODELVIEW19_ARB */
-   { 21251, 0x0000850A }, /* GL_MODELVIEW1_ARB */
-   { 21269, 0x00008734 }, /* GL_MODELVIEW20_ARB */
-   { 21288, 0x00008735 }, /* GL_MODELVIEW21_ARB */
-   { 21307, 0x00008736 }, /* GL_MODELVIEW22_ARB */
-   { 21326, 0x00008737 }, /* GL_MODELVIEW23_ARB */
-   { 21345, 0x00008738 }, /* GL_MODELVIEW24_ARB */
-   { 21364, 0x00008739 }, /* GL_MODELVIEW25_ARB */
-   { 21383, 0x0000873A }, /* GL_MODELVIEW26_ARB */
-   { 21402, 0x0000873B }, /* GL_MODELVIEW27_ARB */
-   { 21421, 0x0000873C }, /* GL_MODELVIEW28_ARB */
-   { 21440, 0x0000873D }, /* GL_MODELVIEW29_ARB */
-   { 21459, 0x00008722 }, /* GL_MODELVIEW2_ARB */
-   { 21477, 0x0000873E }, /* GL_MODELVIEW30_ARB */
-   { 21496, 0x0000873F }, /* GL_MODELVIEW31_ARB */
-   { 21515, 0x00008723 }, /* GL_MODELVIEW3_ARB */
-   { 21533, 0x00008724 }, /* GL_MODELVIEW4_ARB */
-   { 21551, 0x00008725 }, /* GL_MODELVIEW5_ARB */
-   { 21569, 0x00008726 }, /* GL_MODELVIEW6_ARB */
-   { 21587, 0x00008727 }, /* GL_MODELVIEW7_ARB */
-   { 21605, 0x00008728 }, /* GL_MODELVIEW8_ARB */
-   { 21623, 0x00008729 }, /* GL_MODELVIEW9_ARB */
-   { 21641, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
-   { 21661, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
-   { 21688, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
-   { 21713, 0x00002100 }, /* GL_MODULATE */
-   { 21725, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
-   { 21745, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
-   { 21772, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
-   { 21797, 0x00000103 }, /* GL_MULT */
-   { 21805, 0x0000809D }, /* GL_MULTISAMPLE */
-   { 21820, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
-   { 21840, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
-   { 21859, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
-   { 21878, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
-   { 21902, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
-   { 21925, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-   { 21955, 0x00002A25 }, /* GL_N3F_V3F */
-   { 21966, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
-   { 21986, 0x0000150E }, /* GL_NAND */
-   { 21994, 0x00002600 }, /* GL_NEAREST */
-   { 22005, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-   { 22036, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-   { 22068, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
-   { 22093, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
-   { 22119, 0x00000200 }, /* GL_NEVER */
-   { 22128, 0x00001102 }, /* GL_NICEST */
-   { 22138, 0x00000000 }, /* GL_NONE */
-   { 22146, 0x00001505 }, /* GL_NOOP */
-   { 22154, 0x00001508 }, /* GL_NOR */
-   { 22161, 0x00000BA1 }, /* GL_NORMALIZE */
-   { 22174, 0x00008075 }, /* GL_NORMAL_ARRAY */
-   { 22190, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
-   { 22221, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
-   { 22256, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
-   { 22280, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
-   { 22303, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
-   { 22324, 0x00008511 }, /* GL_NORMAL_MAP */
-   { 22338, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
-   { 22356, 0x00008511 }, /* GL_NORMAL_MAP_NV */
-   { 22373, 0x00000205 }, /* GL_NOTEQUAL */
-   { 22385, 0x00000000 }, /* GL_NO_ERROR */
-   { 22397, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
-   { 22431, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
-   { 22469, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
-   { 22501, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
-   { 22543, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
-   { 22573, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
-   { 22613, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
-   { 22644, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
-   { 22673, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
-   { 22701, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
-   { 22731, 0x00002401 }, /* GL_OBJECT_LINEAR */
-   { 22748, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
-   { 22774, 0x00002501 }, /* GL_OBJECT_PLANE */
-   { 22790, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
-   { 22825, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
-   { 22847, 0x00009112 }, /* GL_OBJECT_TYPE */
-   { 22862, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
-   { 22881, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
-   { 22911, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
-   { 22932, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
-   { 22960, 0x00000001 }, /* GL_ONE */
-   { 22967, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
-   { 22995, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
-   { 23027, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
-   { 23055, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
-   { 23087, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
-   { 23110, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
-   { 23133, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
-   { 23156, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
-   { 23179, 0x00008598 }, /* GL_OPERAND0_ALPHA */
-   { 23197, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
-   { 23219, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
-   { 23241, 0x00008590 }, /* GL_OPERAND0_RGB */
-   { 23257, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
-   { 23277, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
-   { 23297, 0x00008599 }, /* GL_OPERAND1_ALPHA */
-   { 23315, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
-   { 23337, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
-   { 23359, 0x00008591 }, /* GL_OPERAND1_RGB */
-   { 23375, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
-   { 23395, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
-   { 23415, 0x0000859A }, /* GL_OPERAND2_ALPHA */
-   { 23433, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
-   { 23455, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
-   { 23477, 0x00008592 }, /* GL_OPERAND2_RGB */
-   { 23493, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
-   { 23513, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
-   { 23533, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
-   { 23554, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
-   { 23573, 0x00001507 }, /* GL_OR */
-   { 23579, 0x00000A01 }, /* GL_ORDER */
-   { 23588, 0x0000150D }, /* GL_OR_INVERTED */
-   { 23603, 0x0000150B }, /* GL_OR_REVERSE */
-   { 23617, 0x00000505 }, /* GL_OUT_OF_MEMORY */
-   { 23634, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
-   { 23652, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
-   { 23673, 0x00008758 }, /* GL_PACK_INVERT_MESA */
-   { 23693, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
-   { 23711, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
-   { 23730, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
-   { 23750, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
-   { 23770, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
-   { 23788, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
-   { 23807, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
-   { 23832, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
-   { 23856, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
-   { 23877, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
-   { 23899, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
-   { 23921, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
-   { 23946, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
-   { 23970, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
-   { 23991, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
-   { 24013, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
-   { 24035, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
-   { 24057, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
-   { 24088, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
-   { 24108, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-   { 24133, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
-   { 24153, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-   { 24178, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
-   { 24198, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-   { 24223, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
-   { 24243, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-   { 24268, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
-   { 24288, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-   { 24313, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
-   { 24333, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-   { 24358, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
-   { 24378, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-   { 24403, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
-   { 24423, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-   { 24448, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
-   { 24468, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-   { 24493, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
-   { 24513, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-   { 24538, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
-   { 24556, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */
-   { 24577, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */
-   { 24606, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
-   { 24639, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
-   { 24664, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */
-   { 24687, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
-   { 24718, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
-   { 24753, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
-   { 24780, 0x00001B00 }, /* GL_POINT */
-   { 24789, 0x00000000 }, /* GL_POINTS */
-   { 24799, 0x00000002 }, /* GL_POINT_BIT */
-   { 24812, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
-   { 24842, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
-   { 24876, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
-   { 24910, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
-   { 24945, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
-   { 24974, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
-   { 25007, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
-   { 25040, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
-   { 25074, 0x00000B11 }, /* GL_POINT_SIZE */
-   { 25088, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
-   { 25114, 0x00008127 }, /* GL_POINT_SIZE_MAX */
-   { 25132, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
-   { 25154, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
-   { 25176, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
-   { 25199, 0x00008126 }, /* GL_POINT_SIZE_MIN */
-   { 25217, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
-   { 25239, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
-   { 25261, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
-   { 25284, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
-   { 25304, 0x00000B10 }, /* GL_POINT_SMOOTH */
-   { 25320, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
-   { 25341, 0x00008861 }, /* GL_POINT_SPRITE */
-   { 25357, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
-   { 25377, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
-   { 25406, 0x00008861 }, /* GL_POINT_SPRITE_NV */
-   { 25425, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
-   { 25451, 0x00000701 }, /* GL_POINT_TOKEN */
-   { 25466, 0x00000009 }, /* GL_POLYGON */
-   { 25477, 0x00000008 }, /* GL_POLYGON_BIT */
-   { 25492, 0x00000B40 }, /* GL_POLYGON_MODE */
-   { 25508, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
-   { 25531, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
-   { 25556, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
-   { 25579, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
-   { 25602, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
-   { 25626, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
-   { 25650, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
-   { 25668, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
-   { 25691, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
-   { 25710, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
-   { 25733, 0x00000703 }, /* GL_POLYGON_TOKEN */
-   { 25750, 0x00001203 }, /* GL_POSITION */
-   { 25762, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-   { 25794, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
-   { 25830, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-   { 25863, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
-   { 25900, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-   { 25931, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
-   { 25966, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-   { 25998, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
-   { 26034, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 26067, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-   { 26099, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
-   { 26135, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-   { 26168, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
-   { 26205, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-   { 26235, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
-   { 26269, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-   { 26300, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
-   { 26335, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-   { 26366, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
-   { 26401, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-   { 26433, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
-   { 26469, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-   { 26499, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
-   { 26533, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-   { 26564, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
-   { 26599, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-   { 26631, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-   { 26662, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
-   { 26697, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-   { 26729, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
-   { 26765, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
-   { 26794, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
-   { 26827, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
-   { 26857, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
-   { 26891, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-   { 26930, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-   { 26963, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-   { 27003, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-   { 27037, 0x00008578 }, /* GL_PREVIOUS */
-   { 27049, 0x00008578 }, /* GL_PREVIOUS_ARB */
-   { 27065, 0x00008578 }, /* GL_PREVIOUS_EXT */
-   { 27081, 0x00008577 }, /* GL_PRIMARY_COLOR */
-   { 27098, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
-   { 27119, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
-   { 27140, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 27173, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 27205, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
-   { 27228, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
-   { 27251, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
-   { 27281, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
-   { 27310, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
-   { 27338, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
-   { 27360, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-   { 27388, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-   { 27416, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
-   { 27438, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
-   { 27459, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 27499, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 27538, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 27568, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 27603, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 27636, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 27670, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 27709, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 27748, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
-   { 27770, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
-   { 27796, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
-   { 27820, 0x00008642 }, /* GL_PROGRAM_POINT_SIZE_ARB */
-   { 27846, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
-   { 27869, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
-   { 27891, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
-   { 27912, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
-   { 27933, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
-   { 27960, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 27992, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 28024, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-   { 28059, 0x00001701 }, /* GL_PROJECTION */
-   { 28073, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
-   { 28094, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
-   { 28120, 0x00008E4F }, /* GL_PROVOKING_VERTEX */
-   { 28140, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */
-   { 28164, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
-   { 28185, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
-   { 28204, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
-   { 28227, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 28266, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-   { 28304, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
-   { 28324, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-   { 28354, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
-   { 28378, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
-   { 28398, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-   { 28428, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
-   { 28452, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
-   { 28472, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-   { 28505, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
-   { 28531, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
-   { 28561, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-   { 28592, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
-   { 28622, 0x00002003 }, /* GL_Q */
-   { 28627, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
-   { 28652, 0x00000007 }, /* GL_QUADS */
-   { 28661, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
-   { 28705, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */
-   { 28753, 0x00008614 }, /* GL_QUAD_MESH_SUN */
-   { 28770, 0x00000008 }, /* GL_QUAD_STRIP */
-   { 28784, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
-   { 28806, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
-   { 28832, 0x00008866 }, /* GL_QUERY_RESULT */
-   { 28848, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
-   { 28868, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
-   { 28894, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
-   { 28924, 0x00002002 }, /* GL_R */
-   { 28929, 0x00002A10 }, /* GL_R3_G3_B2 */
-   { 28941, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-   { 28974, 0x00000C02 }, /* GL_READ_BUFFER */
-   { 28989, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */
-   { 29009, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-   { 29041, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
-   { 29065, 0x000088B8 }, /* GL_READ_ONLY */
-   { 29078, 0x000088B8 }, /* GL_READ_ONLY_ARB */
-   { 29095, 0x000088BA }, /* GL_READ_WRITE */
-   { 29109, 0x000088BA }, /* GL_READ_WRITE_ARB */
-   { 29127, 0x00001903 }, /* GL_RED */
-   { 29134, 0x00008016 }, /* GL_REDUCE */
-   { 29144, 0x00008016 }, /* GL_REDUCE_EXT */
-   { 29158, 0x00000D15 }, /* GL_RED_BIAS */
-   { 29170, 0x00000D52 }, /* GL_RED_BITS */
-   { 29182, 0x00000D14 }, /* GL_RED_SCALE */
-   { 29195, 0x00008512 }, /* GL_REFLECTION_MAP */
-   { 29213, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
-   { 29235, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
-   { 29256, 0x00001C00 }, /* GL_RENDER */
-   { 29266, 0x00008D41 }, /* GL_RENDERBUFFER */
-   { 29282, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */
-   { 29309, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
-   { 29337, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */
-   { 29363, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */
-   { 29390, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
-   { 29410, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */
-   { 29437, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */
-   { 29460, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
-   { 29487, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
-   { 29519, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
-   { 29555, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */
-   { 29580, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */
-   { 29604, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */
-   { 29633, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */
-   { 29655, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
-   { 29681, 0x00001F01 }, /* GL_RENDERER */
-   { 29693, 0x00000C40 }, /* GL_RENDER_MODE */
-   { 29708, 0x00002901 }, /* GL_REPEAT */
-   { 29718, 0x00001E01 }, /* GL_REPLACE */
-   { 29729, 0x00008062 }, /* GL_REPLACE_EXT */
-   { 29744, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
-   { 29767, 0x0000803A }, /* GL_RESCALE_NORMAL */
-   { 29785, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
-   { 29807, 0x00000102 }, /* GL_RETURN */
-   { 29817, 0x00001907 }, /* GL_RGB */
-   { 29824, 0x00008052 }, /* GL_RGB10 */
-   { 29833, 0x00008059 }, /* GL_RGB10_A2 */
-   { 29845, 0x00008059 }, /* GL_RGB10_A2_EXT */
-   { 29861, 0x00008052 }, /* GL_RGB10_EXT */
-   { 29874, 0x00008053 }, /* GL_RGB12 */
-   { 29883, 0x00008053 }, /* GL_RGB12_EXT */
-   { 29896, 0x00008054 }, /* GL_RGB16 */
-   { 29905, 0x00008054 }, /* GL_RGB16_EXT */
-   { 29918, 0x0000804E }, /* GL_RGB2_EXT */
-   { 29930, 0x0000804F }, /* GL_RGB4 */
-   { 29938, 0x0000804F }, /* GL_RGB4_EXT */
-   { 29950, 0x000083A1 }, /* GL_RGB4_S3TC */
-   { 29963, 0x00008050 }, /* GL_RGB5 */
-   { 29971, 0x00008057 }, /* GL_RGB5_A1 */
-   { 29982, 0x00008057 }, /* GL_RGB5_A1_EXT */
-   { 29997, 0x00008050 }, /* GL_RGB5_EXT */
-   { 30009, 0x00008051 }, /* GL_RGB8 */
-   { 30017, 0x00008051 }, /* GL_RGB8_EXT */
-   { 30029, 0x00001908 }, /* GL_RGBA */
-   { 30037, 0x0000805A }, /* GL_RGBA12 */
-   { 30047, 0x0000805A }, /* GL_RGBA12_EXT */
-   { 30061, 0x0000805B }, /* GL_RGBA16 */
-   { 30071, 0x0000805B }, /* GL_RGBA16_EXT */
-   { 30085, 0x00008055 }, /* GL_RGBA2 */
-   { 30094, 0x00008055 }, /* GL_RGBA2_EXT */
-   { 30107, 0x00008056 }, /* GL_RGBA4 */
-   { 30116, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
-   { 30135, 0x00008056 }, /* GL_RGBA4_EXT */
-   { 30148, 0x000083A3 }, /* GL_RGBA4_S3TC */
-   { 30162, 0x00008058 }, /* GL_RGBA8 */
-   { 30171, 0x00008058 }, /* GL_RGBA8_EXT */
-   { 30184, 0x00008F97 }, /* GL_RGBA8_SNORM */
-   { 30199, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
-   { 30217, 0x00000C31 }, /* GL_RGBA_MODE */
-   { 30230, 0x000083A2 }, /* GL_RGBA_S3TC */
-   { 30243, 0x00008F93 }, /* GL_RGBA_SNORM */
-   { 30257, 0x000083A0 }, /* GL_RGB_S3TC */
-   { 30269, 0x00008573 }, /* GL_RGB_SCALE */
-   { 30282, 0x00008573 }, /* GL_RGB_SCALE_ARB */
-   { 30299, 0x00008573 }, /* GL_RGB_SCALE_EXT */
-   { 30316, 0x00000407 }, /* GL_RIGHT */
-   { 30325, 0x00002000 }, /* GL_S */
-   { 30330, 0x00008B5D }, /* GL_SAMPLER_1D */
-   { 30344, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
-   { 30365, 0x00008B5E }, /* GL_SAMPLER_2D */
-   { 30379, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
-   { 30400, 0x00008B5F }, /* GL_SAMPLER_3D */
-   { 30414, 0x00008B60 }, /* GL_SAMPLER_CUBE */
-   { 30430, 0x000080A9 }, /* GL_SAMPLES */
-   { 30441, 0x000086B4 }, /* GL_SAMPLES_3DFX */
-   { 30457, 0x000080A9 }, /* GL_SAMPLES_ARB */
-   { 30472, 0x00008914 }, /* GL_SAMPLES_PASSED */
-   { 30490, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
-   { 30512, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-   { 30540, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
-   { 30572, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
-   { 30595, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
-   { 30622, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
-   { 30640, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
-   { 30663, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
-   { 30685, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
-   { 30704, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
-   { 30727, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
-   { 30753, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
-   { 30783, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
-   { 30808, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
-   { 30837, 0x00080000 }, /* GL_SCISSOR_BIT */
-   { 30852, 0x00000C10 }, /* GL_SCISSOR_BOX */
-   { 30867, 0x00000C11 }, /* GL_SCISSOR_TEST */
-   { 30883, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
-   { 30908, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-   { 30948, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
-   { 30992, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-   { 31025, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-   { 31055, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-   { 31087, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-   { 31117, 0x00001C02 }, /* GL_SELECT */
-   { 31127, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
-   { 31155, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
-   { 31180, 0x00008012 }, /* GL_SEPARABLE_2D */
-   { 31196, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
-   { 31223, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
-   { 31254, 0x0000150F }, /* GL_SET */
-   { 31261, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
-   { 31282, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
-   { 31306, 0x00008B4F }, /* GL_SHADER_TYPE */
-   { 31321, 0x00000B54 }, /* GL_SHADE_MODEL */
-   { 31336, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
-   { 31364, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
-   { 31387, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-   { 31417, 0x00001601 }, /* GL_SHININESS */
-   { 31430, 0x00001402 }, /* GL_SHORT */
-   { 31439, 0x00009119 }, /* GL_SIGNALED */
-   { 31451, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */
-   { 31472, 0x000081F9 }, /* GL_SINGLE_COLOR */
-   { 31488, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
-   { 31508, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
-   { 31527, 0x00008C46 }, /* GL_SLUMINANCE */
-   { 31541, 0x00008C47 }, /* GL_SLUMINANCE8 */
-   { 31556, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */
-   { 31578, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */
-   { 31598, 0x00001D01 }, /* GL_SMOOTH */
-   { 31608, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
-   { 31641, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
-   { 31668, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
-   { 31701, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
-   { 31728, 0x00008588 }, /* GL_SOURCE0_ALPHA */
-   { 31745, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
-   { 31766, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
-   { 31787, 0x00008580 }, /* GL_SOURCE0_RGB */
-   { 31802, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
-   { 31821, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
-   { 31840, 0x00008589 }, /* GL_SOURCE1_ALPHA */
-   { 31857, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
-   { 31878, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
-   { 31899, 0x00008581 }, /* GL_SOURCE1_RGB */
-   { 31914, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
-   { 31933, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
-   { 31952, 0x0000858A }, /* GL_SOURCE2_ALPHA */
-   { 31969, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
-   { 31990, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
-   { 32011, 0x00008582 }, /* GL_SOURCE2_RGB */
-   { 32026, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
-   { 32045, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
-   { 32064, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
-   { 32084, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
-   { 32102, 0x00001202 }, /* GL_SPECULAR */
-   { 32114, 0x00002402 }, /* GL_SPHERE_MAP */
-   { 32128, 0x00001206 }, /* GL_SPOT_CUTOFF */
-   { 32143, 0x00001204 }, /* GL_SPOT_DIRECTION */
-   { 32161, 0x00001205 }, /* GL_SPOT_EXPONENT */
-   { 32178, 0x00008588 }, /* GL_SRC0_ALPHA */
-   { 32192, 0x00008580 }, /* GL_SRC0_RGB */
-   { 32204, 0x00008589 }, /* GL_SRC1_ALPHA */
-   { 32218, 0x00008581 }, /* GL_SRC1_RGB */
-   { 32230, 0x0000858A }, /* GL_SRC2_ALPHA */
-   { 32244, 0x00008582 }, /* GL_SRC2_RGB */
-   { 32256, 0x00000302 }, /* GL_SRC_ALPHA */
-   { 32269, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
-   { 32291, 0x00000300 }, /* GL_SRC_COLOR */
-   { 32304, 0x00008C40 }, /* GL_SRGB */
-   { 32312, 0x00008C41 }, /* GL_SRGB8 */
-   { 32321, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */
-   { 32337, 0x00008C42 }, /* GL_SRGB_ALPHA */
-   { 32351, 0x00000503 }, /* GL_STACK_OVERFLOW */
-   { 32369, 0x00000504 }, /* GL_STACK_UNDERFLOW */
-   { 32388, 0x000088E6 }, /* GL_STATIC_COPY */
-   { 32403, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
-   { 32422, 0x000088E4 }, /* GL_STATIC_DRAW */
-   { 32437, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
-   { 32456, 0x000088E5 }, /* GL_STATIC_READ */
-   { 32471, 0x000088E5 }, /* GL_STATIC_READ_ARB */
-   { 32490, 0x00001802 }, /* GL_STENCIL */
-   { 32501, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */
-   { 32523, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
-   { 32549, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
-   { 32570, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
-   { 32595, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
-   { 32616, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
-   { 32641, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-   { 32673, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
-   { 32709, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-   { 32741, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
-   { 32777, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
-   { 32797, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
-   { 32824, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
-   { 32850, 0x00000D57 }, /* GL_STENCIL_BITS */
-   { 32866, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
-   { 32888, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
-   { 32911, 0x00000B94 }, /* GL_STENCIL_FAIL */
-   { 32927, 0x00000B92 }, /* GL_STENCIL_FUNC */
-   { 32943, 0x00001901 }, /* GL_STENCIL_INDEX */
-   { 32960, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
-   { 32983, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
-   { 33005, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
-   { 33027, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
-   { 33049, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
-   { 33070, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
-   { 33097, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
-   { 33124, 0x00000B97 }, /* GL_STENCIL_REF */
-   { 33139, 0x00000B90 }, /* GL_STENCIL_TEST */
-   { 33155, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
-   { 33184, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
-   { 33206, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
-   { 33227, 0x00000C33 }, /* GL_STEREO */
-   { 33237, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */
-   { 33261, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */
-   { 33286, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */
-   { 33310, 0x000088E2 }, /* GL_STREAM_COPY */
-   { 33325, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
-   { 33344, 0x000088E0 }, /* GL_STREAM_DRAW */
-   { 33359, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
-   { 33378, 0x000088E1 }, /* GL_STREAM_READ */
-   { 33393, 0x000088E1 }, /* GL_STREAM_READ_ARB */
-   { 33412, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
-   { 33429, 0x000084E7 }, /* GL_SUBTRACT */
-   { 33441, 0x000084E7 }, /* GL_SUBTRACT_ARB */
-   { 33457, 0x00009113 }, /* GL_SYNC_CONDITION */
-   { 33475, 0x00009116 }, /* GL_SYNC_FENCE */
-   { 33489, 0x00009115 }, /* GL_SYNC_FLAGS */
-   { 33503, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */
-   { 33530, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
-   { 33560, 0x00009114 }, /* GL_SYNC_STATUS */
-   { 33575, 0x00002001 }, /* GL_T */
-   { 33580, 0x00002A2A }, /* GL_T2F_C3F_V3F */
-   { 33595, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
-   { 33614, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
-   { 33630, 0x00002A2B }, /* GL_T2F_N3F_V3F */
-   { 33645, 0x00002A27 }, /* GL_T2F_V3F */
-   { 33656, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
-   { 33675, 0x00002A28 }, /* GL_T4F_V4F */
-   { 33686, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
-   { 33709, 0x00001702 }, /* GL_TEXTURE */
-   { 33720, 0x000084C0 }, /* GL_TEXTURE0 */
-   { 33732, 0x000084C0 }, /* GL_TEXTURE0_ARB */
-   { 33748, 0x000084C1 }, /* GL_TEXTURE1 */
-   { 33760, 0x000084CA }, /* GL_TEXTURE10 */
-   { 33773, 0x000084CA }, /* GL_TEXTURE10_ARB */
-   { 33790, 0x000084CB }, /* GL_TEXTURE11 */
-   { 33803, 0x000084CB }, /* GL_TEXTURE11_ARB */
-   { 33820, 0x000084CC }, /* GL_TEXTURE12 */
-   { 33833, 0x000084CC }, /* GL_TEXTURE12_ARB */
-   { 33850, 0x000084CD }, /* GL_TEXTURE13 */
-   { 33863, 0x000084CD }, /* GL_TEXTURE13_ARB */
-   { 33880, 0x000084CE }, /* GL_TEXTURE14 */
-   { 33893, 0x000084CE }, /* GL_TEXTURE14_ARB */
-   { 33910, 0x000084CF }, /* GL_TEXTURE15 */
-   { 33923, 0x000084CF }, /* GL_TEXTURE15_ARB */
-   { 33940, 0x000084D0 }, /* GL_TEXTURE16 */
-   { 33953, 0x000084D0 }, /* GL_TEXTURE16_ARB */
-   { 33970, 0x000084D1 }, /* GL_TEXTURE17 */
-   { 33983, 0x000084D1 }, /* GL_TEXTURE17_ARB */
-   { 34000, 0x000084D2 }, /* GL_TEXTURE18 */
-   { 34013, 0x000084D2 }, /* GL_TEXTURE18_ARB */
-   { 34030, 0x000084D3 }, /* GL_TEXTURE19 */
-   { 34043, 0x000084D3 }, /* GL_TEXTURE19_ARB */
-   { 34060, 0x000084C1 }, /* GL_TEXTURE1_ARB */
-   { 34076, 0x000084C2 }, /* GL_TEXTURE2 */
-   { 34088, 0x000084D4 }, /* GL_TEXTURE20 */
-   { 34101, 0x000084D4 }, /* GL_TEXTURE20_ARB */
-   { 34118, 0x000084D5 }, /* GL_TEXTURE21 */
-   { 34131, 0x000084D5 }, /* GL_TEXTURE21_ARB */
-   { 34148, 0x000084D6 }, /* GL_TEXTURE22 */
-   { 34161, 0x000084D6 }, /* GL_TEXTURE22_ARB */
-   { 34178, 0x000084D7 }, /* GL_TEXTURE23 */
-   { 34191, 0x000084D7 }, /* GL_TEXTURE23_ARB */
-   { 34208, 0x000084D8 }, /* GL_TEXTURE24 */
-   { 34221, 0x000084D8 }, /* GL_TEXTURE24_ARB */
-   { 34238, 0x000084D9 }, /* GL_TEXTURE25 */
-   { 34251, 0x000084D9 }, /* GL_TEXTURE25_ARB */
-   { 34268, 0x000084DA }, /* GL_TEXTURE26 */
-   { 34281, 0x000084DA }, /* GL_TEXTURE26_ARB */
-   { 34298, 0x000084DB }, /* GL_TEXTURE27 */
-   { 34311, 0x000084DB }, /* GL_TEXTURE27_ARB */
-   { 34328, 0x000084DC }, /* GL_TEXTURE28 */
-   { 34341, 0x000084DC }, /* GL_TEXTURE28_ARB */
-   { 34358, 0x000084DD }, /* GL_TEXTURE29 */
-   { 34371, 0x000084DD }, /* GL_TEXTURE29_ARB */
-   { 34388, 0x000084C2 }, /* GL_TEXTURE2_ARB */
-   { 34404, 0x000084C3 }, /* GL_TEXTURE3 */
-   { 34416, 0x000084DE }, /* GL_TEXTURE30 */
-   { 34429, 0x000084DE }, /* GL_TEXTURE30_ARB */
-   { 34446, 0x000084DF }, /* GL_TEXTURE31 */
-   { 34459, 0x000084DF }, /* GL_TEXTURE31_ARB */
-   { 34476, 0x000084C3 }, /* GL_TEXTURE3_ARB */
-   { 34492, 0x000084C4 }, /* GL_TEXTURE4 */
-   { 34504, 0x000084C4 }, /* GL_TEXTURE4_ARB */
-   { 34520, 0x000084C5 }, /* GL_TEXTURE5 */
-   { 34532, 0x000084C5 }, /* GL_TEXTURE5_ARB */
-   { 34548, 0x000084C6 }, /* GL_TEXTURE6 */
-   { 34560, 0x000084C6 }, /* GL_TEXTURE6_ARB */
-   { 34576, 0x000084C7 }, /* GL_TEXTURE7 */
-   { 34588, 0x000084C7 }, /* GL_TEXTURE7_ARB */
-   { 34604, 0x000084C8 }, /* GL_TEXTURE8 */
-   { 34616, 0x000084C8 }, /* GL_TEXTURE8_ARB */
-   { 34632, 0x000084C9 }, /* GL_TEXTURE9 */
-   { 34644, 0x000084C9 }, /* GL_TEXTURE9_ARB */
-   { 34660, 0x00000DE0 }, /* GL_TEXTURE_1D */
-   { 34674, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
-   { 34698, 0x00000DE1 }, /* GL_TEXTURE_2D */
-   { 34712, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
-   { 34736, 0x0000806F }, /* GL_TEXTURE_3D */
-   { 34750, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
-   { 34772, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
-   { 34798, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
-   { 34820, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
-   { 34842, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-   { 34874, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
-   { 34896, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-   { 34928, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
-   { 34950, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
-   { 34978, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
-   { 35010, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-   { 35043, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
-   { 35075, 0x00040000 }, /* GL_TEXTURE_BIT */
-   { 35090, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
-   { 35111, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
-   { 35136, 0x00001005 }, /* GL_TEXTURE_BORDER */
-   { 35154, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
-   { 35178, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-   { 35209, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-   { 35239, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-   { 35269, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-   { 35304, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-   { 35335, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 35373, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
-   { 35400, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-   { 35432, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
-   { 35466, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
-   { 35490, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
-   { 35518, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
-   { 35542, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
-   { 35570, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-   { 35603, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
-   { 35627, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
-   { 35649, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
-   { 35671, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
-   { 35697, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
-   { 35731, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-   { 35764, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
-   { 35801, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
-   { 35829, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
-   { 35861, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
-   { 35884, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-   { 35922, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
-   { 35964, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-   { 35995, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-   { 36023, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-   { 36053, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-   { 36081, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
-   { 36101, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
-   { 36125, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-   { 36156, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
-   { 36191, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-   { 36222, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
-   { 36257, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-   { 36288, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
-   { 36323, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-   { 36354, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
-   { 36389, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-   { 36420, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
-   { 36455, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-   { 36486, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
-   { 36521, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
-   { 36550, 0x00008071 }, /* GL_TEXTURE_DEPTH */
-   { 36567, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
-   { 36589, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
-   { 36615, 0x00002300 }, /* GL_TEXTURE_ENV */
-   { 36630, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
-   { 36651, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
-   { 36671, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
-   { 36697, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
-   { 36717, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
-   { 36734, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
-   { 36751, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
-   { 36768, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
-   { 36785, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
-   { 36810, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
-   { 36832, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
-   { 36858, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
-   { 36876, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
-   { 36902, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
-   { 36928, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
-   { 36958, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
-   { 36985, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
-   { 37010, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
-   { 37030, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
-   { 37054, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-   { 37081, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-   { 37108, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-   { 37135, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
-   { 37161, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
-   { 37191, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
-   { 37213, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
-   { 37231, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 37261, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-   { 37289, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-   { 37317, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-   { 37345, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
-   { 37366, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
-   { 37385, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
-   { 37407, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
-   { 37426, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
-   { 37446, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
-   { 37476, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */
-   { 37507, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
-   { 37532, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
-   { 37556, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
-   { 37576, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
-   { 37600, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
-   { 37620, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
-   { 37643, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */
-   { 37667, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */
-   { 37697, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
-   { 37722, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
-   { 37756, 0x00001000 }, /* GL_TEXTURE_WIDTH */
-   { 37773, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
-   { 37791, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
-   { 37809, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
-   { 37827, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */
-   { 37846, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */
-   { 37865, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
-   { 37885, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
-   { 37904, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-   { 37933, 0x00001000 }, /* GL_TRANSFORM_BIT */
-   { 37950, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
-   { 37976, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
-   { 38006, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-   { 38038, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-   { 38068, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
-   { 38102, 0x0000862C }, /* GL_TRANSPOSE_NV */
-   { 38118, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-   { 38149, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
-   { 38184, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-   { 38212, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
-   { 38244, 0x00000004 }, /* GL_TRIANGLES */
-   { 38257, 0x0000000C }, /* GL_TRIANGLES_ADJACENCY_ARB */
-   { 38284, 0x00000006 }, /* GL_TRIANGLE_FAN */
-   { 38300, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
-   { 38321, 0x00000005 }, /* GL_TRIANGLE_STRIP */
-   { 38339, 0x0000000D }, /* GL_TRIANGLE_STRIP_ADJACENCY_ARB */
-   { 38371, 0x00000001 }, /* GL_TRUE */
-   { 38379, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
-   { 38399, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
-   { 38422, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
-   { 38442, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
-   { 38463, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
-   { 38485, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
-   { 38507, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
-   { 38527, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
-   { 38548, 0x00009118 }, /* GL_UNSIGNALED */
-   { 38562, 0x00001401 }, /* GL_UNSIGNED_BYTE */
-   { 38579, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-   { 38606, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
-   { 38629, 0x00001405 }, /* GL_UNSIGNED_INT */
-   { 38645, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
-   { 38672, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */
-   { 38693, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
-   { 38717, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-   { 38748, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
-   { 38772, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-   { 38800, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */
-   { 38823, 0x00001403 }, /* GL_UNSIGNED_SHORT */
-   { 38841, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-   { 38871, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-   { 38897, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-   { 38927, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-   { 38953, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
-   { 38977, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-   { 39005, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-   { 39033, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
-   { 39060, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-   { 39092, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
-   { 39123, 0x00008CA2 }, /* GL_UPPER_LEFT */
-   { 39137, 0x00002A20 }, /* GL_V2F */
-   { 39144, 0x00002A21 }, /* GL_V3F */
-   { 39151, 0x00008B83 }, /* GL_VALIDATE_STATUS */
-   { 39170, 0x00001F00 }, /* GL_VENDOR */
-   { 39180, 0x00001F02 }, /* GL_VERSION */
-   { 39191, 0x00008074 }, /* GL_VERTEX_ARRAY */
-   { 39207, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */
-   { 39231, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
-   { 39261, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-   { 39292, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
-   { 39327, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
-   { 39351, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
-   { 39372, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
-   { 39395, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
-   { 39416, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-   { 39443, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-   { 39471, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-   { 39499, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-   { 39527, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-   { 39555, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-   { 39583, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-   { 39611, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-   { 39638, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-   { 39665, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-   { 39692, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-   { 39719, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-   { 39746, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-   { 39773, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-   { 39800, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-   { 39827, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-   { 39854, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-   { 39892, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
-   { 39934, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-   { 39965, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
-   { 40000, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-   { 40034, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
-   { 40072, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-   { 40103, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
-   { 40138, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-   { 40166, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
-   { 40198, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-   { 40228, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
-   { 40262, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
-   { 40290, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
-   { 40322, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
-   { 40342, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
-   { 40364, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
-   { 40393, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
-   { 40414, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-   { 40443, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
-   { 40476, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
-   { 40508, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-   { 40535, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
-   { 40566, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
-   { 40596, 0x00008B31 }, /* GL_VERTEX_SHADER */
-   { 40613, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
-   { 40634, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
-   { 40661, 0x00000BA2 }, /* GL_VIEWPORT */
-   { 40673, 0x00000800 }, /* GL_VIEWPORT_BIT */
-   { 40689, 0x0000911D }, /* GL_WAIT_FAILED */
-   { 40704, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
-   { 40724, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-   { 40755, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
-   { 40790, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-   { 40818, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-   { 40843, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-   { 40870, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-   { 40895, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
-   { 40919, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
-   { 40938, 0x000088B9 }, /* GL_WRITE_ONLY */
-   { 40952, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
-   { 40970, 0x00001506 }, /* GL_XOR */
-   { 40977, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
-   { 40996, 0x00008757 }, /* GL_YCBCR_MESA */
-   { 41010, 0x00000000 }, /* GL_ZERO */
-   { 41018, 0x00000D16 }, /* GL_ZOOM_X */
-   { 41028, 0x00000D17 }, /* GL_ZOOM_Y */
+   {  7190, 0x000088F0 }, /* GL_DEPTH24_STENCIL8_EXT */
+   {  7214, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT */
+   {  7234, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT_EXT */
+   {  7258, 0x00000D1F }, /* GL_DEPTH_BIAS */
+   {  7272, 0x00000D56 }, /* GL_DEPTH_BITS */
+   {  7286, 0x00008891 }, /* GL_DEPTH_BOUNDS_EXT */
+   {  7306, 0x00008890 }, /* GL_DEPTH_BOUNDS_TEST_EXT */
+   {  7331, 0x00000100 }, /* GL_DEPTH_BUFFER_BIT */
+   {  7351, 0x0000864F }, /* GL_DEPTH_CLAMP */
+   {  7366, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
+   {  7384, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
+   {  7405, 0x00001902 }, /* GL_DEPTH_COMPONENT */
+   {  7424, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
+   {  7445, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
+   {  7470, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
+   {  7496, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
+   {  7517, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
+   {  7542, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
+   {  7568, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
+   {  7589, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
+   {  7614, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
+   {  7640, 0x00000B74 }, /* GL_DEPTH_FUNC */
+   {  7654, 0x00000B70 }, /* GL_DEPTH_RANGE */
+   {  7669, 0x00000D1E }, /* GL_DEPTH_SCALE */
+   {  7684, 0x000084F9 }, /* GL_DEPTH_STENCIL */
+   {  7701, 0x0000821A }, /* GL_DEPTH_STENCIL_ATTACHMENT */
+   {  7729, 0x000084F9 }, /* GL_DEPTH_STENCIL_EXT */
+   {  7750, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
+   {  7770, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+   {  7798, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+   {  7826, 0x00000B71 }, /* GL_DEPTH_TEST */
+   {  7840, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
+   {  7862, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
+   {  7888, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
+   {  7907, 0x00001201 }, /* GL_DIFFUSE */
+   {  7918, 0x00000BD0 }, /* GL_DITHER */
+   {  7928, 0x00000A02 }, /* GL_DOMAIN */
+   {  7938, 0x00001100 }, /* GL_DONT_CARE */
+   {  7951, 0x000086AE }, /* GL_DOT3_RGB */
+   {  7963, 0x000086AF }, /* GL_DOT3_RGBA */
+   {  7976, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
+   {  7993, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
+   {  8010, 0x000086AE }, /* GL_DOT3_RGB_ARB */
+   {  8026, 0x00008740 }, /* GL_DOT3_RGB_EXT */
+   {  8042, 0x0000140A }, /* GL_DOUBLE */
+   {  8052, 0x00000C32 }, /* GL_DOUBLEBUFFER */
+   {  8068, 0x00000C01 }, /* GL_DRAW_BUFFER */
+   {  8083, 0x00008825 }, /* GL_DRAW_BUFFER0 */
+   {  8099, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
+   {  8119, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
+   {  8139, 0x00008826 }, /* GL_DRAW_BUFFER1 */
+   {  8155, 0x0000882F }, /* GL_DRAW_BUFFER10 */
+   {  8172, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
+   {  8193, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
+   {  8214, 0x00008830 }, /* GL_DRAW_BUFFER11 */
+   {  8231, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
+   {  8252, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
+   {  8273, 0x00008831 }, /* GL_DRAW_BUFFER12 */
+   {  8290, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
+   {  8311, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
+   {  8332, 0x00008832 }, /* GL_DRAW_BUFFER13 */
+   {  8349, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
+   {  8370, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
+   {  8391, 0x00008833 }, /* GL_DRAW_BUFFER14 */
+   {  8408, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
+   {  8429, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
+   {  8450, 0x00008834 }, /* GL_DRAW_BUFFER15 */
+   {  8467, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
+   {  8488, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
+   {  8509, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
+   {  8529, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
+   {  8549, 0x00008827 }, /* GL_DRAW_BUFFER2 */
+   {  8565, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
+   {  8585, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
+   {  8605, 0x00008828 }, /* GL_DRAW_BUFFER3 */
+   {  8621, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
+   {  8641, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
+   {  8661, 0x00008829 }, /* GL_DRAW_BUFFER4 */
+   {  8677, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
+   {  8697, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
+   {  8717, 0x0000882A }, /* GL_DRAW_BUFFER5 */
+   {  8733, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
+   {  8753, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
+   {  8773, 0x0000882B }, /* GL_DRAW_BUFFER6 */
+   {  8789, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
+   {  8809, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
+   {  8829, 0x0000882C }, /* GL_DRAW_BUFFER7 */
+   {  8845, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
+   {  8865, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
+   {  8885, 0x0000882D }, /* GL_DRAW_BUFFER8 */
+   {  8901, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
+   {  8921, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
+   {  8941, 0x0000882E }, /* GL_DRAW_BUFFER9 */
+   {  8957, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
+   {  8977, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
+   {  8997, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER */
+   {  9017, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING */
+   {  9045, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
+   {  9077, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
+   {  9101, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
+   {  9121, 0x00000304 }, /* GL_DST_ALPHA */
+   {  9134, 0x00000306 }, /* GL_DST_COLOR */
+   {  9147, 0x0000877A }, /* GL_DU8DV8_ATI */
+   {  9161, 0x00008779 }, /* GL_DUDV_ATI */
+   {  9173, 0x000088EA }, /* GL_DYNAMIC_COPY */
+   {  9189, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
+   {  9209, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
+   {  9225, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
+   {  9245, 0x000088E9 }, /* GL_DYNAMIC_READ */
+   {  9261, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
+   {  9281, 0x00000B43 }, /* GL_EDGE_FLAG */
+   {  9294, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
+   {  9313, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+   {  9347, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
+   {  9385, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
+   {  9412, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+   {  9438, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
+   {  9462, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+   {  9494, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
+   {  9530, 0x00001600 }, /* GL_EMISSION */
+   {  9542, 0x00002000 }, /* GL_ENABLE_BIT */
+   {  9556, 0x00000202 }, /* GL_EQUAL */
+   {  9565, 0x00001509 }, /* GL_EQUIV */
+   {  9574, 0x00010000 }, /* GL_EVAL_BIT */
+   {  9586, 0x00000800 }, /* GL_EXP */
+   {  9593, 0x00000801 }, /* GL_EXP2 */
+   {  9601, 0x00001F03 }, /* GL_EXTENSIONS */
+   {  9615, 0x00002400 }, /* GL_EYE_LINEAR */
+   {  9629, 0x00002502 }, /* GL_EYE_PLANE */
+   {  9642, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
+   {  9667, 0x0000855B }, /* GL_EYE_RADIAL_NV */
+   {  9684, 0x00000000 }, /* GL_FALSE */
+   {  9693, 0x00001101 }, /* GL_FASTEST */
+   {  9704, 0x00001C01 }, /* GL_FEEDBACK */
+   {  9716, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
+   {  9743, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
+   {  9767, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
+   {  9791, 0x00001B02 }, /* GL_FILL */
+   {  9799, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION */
+   {  9826, 0x00008E4D }, /* GL_FIRST_VERTEX_CONVENTION_EXT */
+   {  9857, 0x00001D00 }, /* GL_FLAT */
+   {  9865, 0x00001406 }, /* GL_FLOAT */
+   {  9874, 0x00008B5A }, /* GL_FLOAT_MAT2 */
+   {  9888, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
+   {  9906, 0x00008B65 }, /* GL_FLOAT_MAT2x3 */
+   {  9922, 0x00008B66 }, /* GL_FLOAT_MAT2x4 */
+   {  9938, 0x00008B5B }, /* GL_FLOAT_MAT3 */
+   {  9952, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
+   {  9970, 0x00008B67 }, /* GL_FLOAT_MAT3x2 */
+   {  9986, 0x00008B68 }, /* GL_FLOAT_MAT3x4 */
+   { 10002, 0x00008B5C }, /* GL_FLOAT_MAT4 */
+   { 10016, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
+   { 10034, 0x00008B69 }, /* GL_FLOAT_MAT4x2 */
+   { 10050, 0x00008B6A }, /* GL_FLOAT_MAT4x3 */
+   { 10066, 0x00008B50 }, /* GL_FLOAT_VEC2 */
+   { 10080, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
+   { 10098, 0x00008B51 }, /* GL_FLOAT_VEC3 */
+   { 10112, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
+   { 10130, 0x00008B52 }, /* GL_FLOAT_VEC4 */
+   { 10144, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
+   { 10162, 0x00000B60 }, /* GL_FOG */
+   { 10169, 0x00000080 }, /* GL_FOG_BIT */
+   { 10180, 0x00000B66 }, /* GL_FOG_COLOR */
+   { 10193, 0x00008451 }, /* GL_FOG_COORD */
+   { 10206, 0x00008451 }, /* GL_FOG_COORDINATE */
+   { 10224, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
+   { 10248, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+   { 10287, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
+   { 10330, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+   { 10362, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+   { 10393, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+   { 10422, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
+   { 10447, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
+   { 10466, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
+   { 10500, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
+   { 10527, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
+   { 10553, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
+   { 10577, 0x00008450 }, /* GL_FOG_COORD_SRC */
+   { 10594, 0x00000B62 }, /* GL_FOG_DENSITY */
+   { 10609, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
+   { 10633, 0x00000B64 }, /* GL_FOG_END */
+   { 10644, 0x00000C54 }, /* GL_FOG_HINT */
+   { 10656, 0x00000B61 }, /* GL_FOG_INDEX */
+   { 10669, 0x00000B65 }, /* GL_FOG_MODE */
+   { 10681, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
+   { 10700, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
+   { 10725, 0x00000B63 }, /* GL_FOG_START */
+   { 10738, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
+   { 10756, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
+   { 10780, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
+   { 10799, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
+   { 10822, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+   { 10857, 0x00008D40 }, /* GL_FRAMEBUFFER */
+   { 10872, 0x00008215 }, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
+   { 10909, 0x00008214 }, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
+   { 10945, 0x00008210 }, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
+   { 10986, 0x00008211 }, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
+   { 11027, 0x00008216 }, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
+   { 11064, 0x00008213 }, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
+   { 11101, 0x00008DA7 }, /* GL_FRAMEBUFFER_ATTACHMENT_LAYERED_ARB */
+   { 11139, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
+   { 11177, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
+   { 11219, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
+   { 11257, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
+   { 11299, 0x00008212 }, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
+   { 11334, 0x00008217 }, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
+   { 11373, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
+   { 11422, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
+   { 11470, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
+   { 11522, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
+   { 11562, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
+   { 11606, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
+   { 11646, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
+   { 11690, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING */
+   { 11713, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
+   { 11740, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE */
+   { 11764, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
+   { 11792, 0x00008218 }, /* GL_FRAMEBUFFER_DEFAULT */
+   { 11815, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
+   { 11834, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
+   { 11871, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
+   { 11912, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+   { 11953, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER */
+   { 11991, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
+   { 12033, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+   { 12084, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+   { 12122, 0x00008DA9 }, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_ARB */
+   { 12164, 0x00008DA8 }, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_ARB */
+   { 12208, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
+   { 12253, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
+   { 12302, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
+   { 12340, 0x00008D56 }, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT */
+   { 12382, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER */
+   { 12420, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
+   { 12462, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+   { 12494, 0x00008219 }, /* GL_FRAMEBUFFER_UNDEFINED */
+   { 12519, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED */
+   { 12546, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
+   { 12577, 0x00000404 }, /* GL_FRONT */
+   { 12586, 0x00000408 }, /* GL_FRONT_AND_BACK */
+   { 12604, 0x00000B46 }, /* GL_FRONT_FACE */
+   { 12618, 0x00000400 }, /* GL_FRONT_LEFT */
+   { 12632, 0x00000401 }, /* GL_FRONT_RIGHT */
+   { 12647, 0x00008006 }, /* GL_FUNC_ADD */
+   { 12659, 0x00008006 }, /* GL_FUNC_ADD_EXT */
+   { 12675, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
+   { 12700, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
+   { 12729, 0x0000800A }, /* GL_FUNC_SUBTRACT */
+   { 12746, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
+   { 12767, 0x00008191 }, /* GL_GENERATE_MIPMAP */
+   { 12786, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
+   { 12810, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
+   { 12839, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
+   { 12863, 0x00008DDB }, /* GL_GEOMETRY_INPUT_TYPE_ARB */
+   { 12890, 0x00008DDC }, /* GL_GEOMETRY_OUTPUT_TYPE_ARB */
+   { 12918, 0x00008DD9 }, /* GL_GEOMETRY_SHADER_ARB */
+   { 12941, 0x00008DDA }, /* GL_GEOMETRY_VERTICES_OUT_ARB */
+   { 12970, 0x00000206 }, /* GL_GEQUAL */
+   { 12980, 0x00000204 }, /* GL_GREATER */
+   { 12991, 0x00001904 }, /* GL_GREEN */
+   { 13000, 0x00000D19 }, /* GL_GREEN_BIAS */
+   { 13014, 0x00000D53 }, /* GL_GREEN_BITS */
+   { 13028, 0x00000D18 }, /* GL_GREEN_SCALE */
+   { 13043, 0x00008000 }, /* GL_HINT_BIT */
+   { 13055, 0x00008024 }, /* GL_HISTOGRAM */
+   { 13068, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
+   { 13092, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
+   { 13120, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
+   { 13143, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
+   { 13170, 0x00008024 }, /* GL_HISTOGRAM_EXT */
+   { 13187, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
+   { 13207, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
+   { 13231, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
+   { 13255, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
+   { 13283, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+   { 13311, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
+   { 13343, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
+   { 13365, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
+   { 13391, 0x0000802D }, /* GL_HISTOGRAM_SINK */
+   { 13409, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
+   { 13431, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
+   { 13450, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
+   { 13473, 0x0000862A }, /* GL_IDENTITY_NV */
+   { 13488, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
+   { 13508, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+   { 13548, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+   { 13586, 0x00001E02 }, /* GL_INCR */
+   { 13594, 0x00008507 }, /* GL_INCR_WRAP */
+   { 13607, 0x00008507 }, /* GL_INCR_WRAP_EXT */
+   { 13624, 0x00008222 }, /* GL_INDEX */
+   { 13633, 0x00008077 }, /* GL_INDEX_ARRAY */
+   { 13648, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+   { 13678, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
+   { 13712, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
+   { 13735, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
+   { 13757, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
+   { 13777, 0x00000D51 }, /* GL_INDEX_BITS */
+   { 13791, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
+   { 13812, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
+   { 13830, 0x00000C30 }, /* GL_INDEX_MODE */
+   { 13844, 0x00000D13 }, /* GL_INDEX_OFFSET */
+   { 13860, 0x00000D12 }, /* GL_INDEX_SHIFT */
+   { 13875, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
+   { 13894, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
+   { 13913, 0x00001404 }, /* GL_INT */
+   { 13920, 0x00008049 }, /* GL_INTENSITY */
+   { 13933, 0x0000804C }, /* GL_INTENSITY12 */
+   { 13948, 0x0000804C }, /* GL_INTENSITY12_EXT */
+   { 13967, 0x0000804D }, /* GL_INTENSITY16 */
+   { 13982, 0x0000804D }, /* GL_INTENSITY16_EXT */
+   { 14001, 0x0000804A }, /* GL_INTENSITY4 */
+   { 14015, 0x0000804A }, /* GL_INTENSITY4_EXT */
+   { 14033, 0x0000804B }, /* GL_INTENSITY8 */
+   { 14047, 0x0000804B }, /* GL_INTENSITY8_EXT */
+   { 14065, 0x00008049 }, /* GL_INTENSITY_EXT */
+   { 14082, 0x00008575 }, /* GL_INTERPOLATE */
+   { 14097, 0x00008575 }, /* GL_INTERPOLATE_ARB */
+   { 14116, 0x00008575 }, /* GL_INTERPOLATE_EXT */
+   { 14135, 0x00008B53 }, /* GL_INT_VEC2 */
+   { 14147, 0x00008B53 }, /* GL_INT_VEC2_ARB */
+   { 14163, 0x00008B54 }, /* GL_INT_VEC3 */
+   { 14175, 0x00008B54 }, /* GL_INT_VEC3_ARB */
+   { 14191, 0x00008B55 }, /* GL_INT_VEC4 */
+   { 14203, 0x00008B55 }, /* GL_INT_VEC4_ARB */
+   { 14219, 0x00000500 }, /* GL_INVALID_ENUM */
+   { 14235, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION */
+   { 14268, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
+   { 14305, 0x00000502 }, /* GL_INVALID_OPERATION */
+   { 14326, 0x00000501 }, /* GL_INVALID_VALUE */
+   { 14343, 0x0000862B }, /* GL_INVERSE_NV */
+   { 14357, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
+   { 14381, 0x0000150A }, /* GL_INVERT */
+   { 14391, 0x00001E00 }, /* GL_KEEP */
+   { 14399, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION */
+   { 14425, 0x00008E4E }, /* GL_LAST_VERTEX_CONVENTION_EXT */
+   { 14455, 0x00000406 }, /* GL_LEFT */
+   { 14463, 0x00000203 }, /* GL_LEQUAL */
+   { 14473, 0x00000201 }, /* GL_LESS */
+   { 14481, 0x00004000 }, /* GL_LIGHT0 */
+   { 14491, 0x00004001 }, /* GL_LIGHT1 */
+   { 14501, 0x00004002 }, /* GL_LIGHT2 */
+   { 14511, 0x00004003 }, /* GL_LIGHT3 */
+   { 14521, 0x00004004 }, /* GL_LIGHT4 */
+   { 14531, 0x00004005 }, /* GL_LIGHT5 */
+   { 14541, 0x00004006 }, /* GL_LIGHT6 */
+   { 14551, 0x00004007 }, /* GL_LIGHT7 */
+   { 14561, 0x00000B50 }, /* GL_LIGHTING */
+   { 14573, 0x00000040 }, /* GL_LIGHTING_BIT */
+   { 14589, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
+   { 14612, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+   { 14641, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
+   { 14674, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+   { 14702, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
+   { 14726, 0x00001B01 }, /* GL_LINE */
+   { 14734, 0x00002601 }, /* GL_LINEAR */
+   { 14744, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
+   { 14766, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+   { 14796, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+   { 14827, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
+   { 14851, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
+   { 14876, 0x00000001 }, /* GL_LINES */
+   { 14885, 0x0000000A }, /* GL_LINES_ADJACENCY_ARB */
+   { 14908, 0x00000004 }, /* GL_LINE_BIT */
+   { 14920, 0x00000002 }, /* GL_LINE_LOOP */
+   { 14933, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
+   { 14953, 0x00000B20 }, /* GL_LINE_SMOOTH */
+   { 14968, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
+   { 14988, 0x00000B24 }, /* GL_LINE_STIPPLE */
+   { 15004, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
+   { 15028, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
+   { 15051, 0x00000003 }, /* GL_LINE_STRIP */
+   { 15065, 0x0000000B }, /* GL_LINE_STRIP_ADJACENCY_ARB */
+   { 15093, 0x00000702 }, /* GL_LINE_TOKEN */
+   { 15107, 0x00000B21 }, /* GL_LINE_WIDTH */
+   { 15121, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
+   { 15147, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
+   { 15167, 0x00008B82 }, /* GL_LINK_STATUS */
+   { 15182, 0x00000B32 }, /* GL_LIST_BASE */
+   { 15195, 0x00020000 }, /* GL_LIST_BIT */
+   { 15207, 0x00000B33 }, /* GL_LIST_INDEX */
+   { 15221, 0x00000B30 }, /* GL_LIST_MODE */
+   { 15234, 0x00000101 }, /* GL_LOAD */
+   { 15242, 0x00000BF1 }, /* GL_LOGIC_OP */
+   { 15254, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
+   { 15271, 0x00008CA1 }, /* GL_LOWER_LEFT */
+   { 15285, 0x00001909 }, /* GL_LUMINANCE */
+   { 15298, 0x00008041 }, /* GL_LUMINANCE12 */
+   { 15313, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
+   { 15336, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
+   { 15363, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
+   { 15385, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
+   { 15411, 0x00008041 }, /* GL_LUMINANCE12_EXT */
+   { 15430, 0x00008042 }, /* GL_LUMINANCE16 */
+   { 15445, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
+   { 15468, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
+   { 15495, 0x00008042 }, /* GL_LUMINANCE16_EXT */
+   { 15514, 0x0000803F }, /* GL_LUMINANCE4 */
+   { 15528, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
+   { 15549, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
+   { 15574, 0x0000803F }, /* GL_LUMINANCE4_EXT */
+   { 15592, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
+   { 15613, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
+   { 15638, 0x00008040 }, /* GL_LUMINANCE8 */
+   { 15652, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
+   { 15673, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
+   { 15698, 0x00008040 }, /* GL_LUMINANCE8_EXT */
+   { 15716, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
+   { 15735, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
+   { 15751, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
+   { 15771, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
+   { 15793, 0x00000D91 }, /* GL_MAP1_INDEX */
+   { 15807, 0x00000D92 }, /* GL_MAP1_NORMAL */
+   { 15822, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
+   { 15846, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
+   { 15870, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
+   { 15894, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
+   { 15918, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
+   { 15935, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
+   { 15952, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+   { 15980, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+   { 16009, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+   { 16038, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+   { 16067, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+   { 16096, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+   { 16125, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+   { 16154, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+   { 16182, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+   { 16210, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+   { 16238, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+   { 16266, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+   { 16294, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+   { 16322, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+   { 16350, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+   { 16378, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+   { 16406, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
+   { 16422, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
+   { 16442, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
+   { 16464, 0x00000DB1 }, /* GL_MAP2_INDEX */
+   { 16478, 0x00000DB2 }, /* GL_MAP2_NORMAL */
+   { 16493, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
+   { 16517, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
+   { 16541, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
+   { 16565, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
+   { 16589, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
+   { 16606, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
+   { 16623, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+   { 16651, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+   { 16680, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+   { 16709, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+   { 16738, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+   { 16767, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+   { 16796, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+   { 16825, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+   { 16853, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+   { 16881, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+   { 16909, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+   { 16937, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+   { 16965, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+   { 16993, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
+   { 17021, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+   { 17049, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+   { 17077, 0x00000D10 }, /* GL_MAP_COLOR */
+   { 17090, 0x00000010 }, /* GL_MAP_FLUSH_EXPLICIT_BIT */
+   { 17116, 0x00000008 }, /* GL_MAP_INVALIDATE_BUFFER_BIT */
+   { 17145, 0x00000004 }, /* GL_MAP_INVALIDATE_RANGE_BIT */
+   { 17173, 0x00000001 }, /* GL_MAP_READ_BIT */
+   { 17189, 0x00000D11 }, /* GL_MAP_STENCIL */
+   { 17204, 0x00000020 }, /* GL_MAP_UNSYNCHRONIZED_BIT */
+   { 17230, 0x00000002 }, /* GL_MAP_WRITE_BIT */
+   { 17247, 0x000088C0 }, /* GL_MATRIX0_ARB */
+   { 17262, 0x00008630 }, /* GL_MATRIX0_NV */
+   { 17276, 0x000088CA }, /* GL_MATRIX10_ARB */
+   { 17292, 0x000088CB }, /* GL_MATRIX11_ARB */
+   { 17308, 0x000088CC }, /* GL_MATRIX12_ARB */
+   { 17324, 0x000088CD }, /* GL_MATRIX13_ARB */
+   { 17340, 0x000088CE }, /* GL_MATRIX14_ARB */
+   { 17356, 0x000088CF }, /* GL_MATRIX15_ARB */
+   { 17372, 0x000088D0 }, /* GL_MATRIX16_ARB */
+   { 17388, 0x000088D1 }, /* GL_MATRIX17_ARB */
+   { 17404, 0x000088D2 }, /* GL_MATRIX18_ARB */
+   { 17420, 0x000088D3 }, /* GL_MATRIX19_ARB */
+   { 17436, 0x000088C1 }, /* GL_MATRIX1_ARB */
+   { 17451, 0x00008631 }, /* GL_MATRIX1_NV */
+   { 17465, 0x000088D4 }, /* GL_MATRIX20_ARB */
+   { 17481, 0x000088D5 }, /* GL_MATRIX21_ARB */
+   { 17497, 0x000088D6 }, /* GL_MATRIX22_ARB */
+   { 17513, 0x000088D7 }, /* GL_MATRIX23_ARB */
+   { 17529, 0x000088D8 }, /* GL_MATRIX24_ARB */
+   { 17545, 0x000088D9 }, /* GL_MATRIX25_ARB */
+   { 17561, 0x000088DA }, /* GL_MATRIX26_ARB */
+   { 17577, 0x000088DB }, /* GL_MATRIX27_ARB */
+   { 17593, 0x000088DC }, /* GL_MATRIX28_ARB */
+   { 17609, 0x000088DD }, /* GL_MATRIX29_ARB */
+   { 17625, 0x000088C2 }, /* GL_MATRIX2_ARB */
+   { 17640, 0x00008632 }, /* GL_MATRIX2_NV */
+   { 17654, 0x000088DE }, /* GL_MATRIX30_ARB */
+   { 17670, 0x000088DF }, /* GL_MATRIX31_ARB */
+   { 17686, 0x000088C3 }, /* GL_MATRIX3_ARB */
+   { 17701, 0x00008633 }, /* GL_MATRIX3_NV */
+   { 17715, 0x000088C4 }, /* GL_MATRIX4_ARB */
+   { 17730, 0x00008634 }, /* GL_MATRIX4_NV */
+   { 17744, 0x000088C5 }, /* GL_MATRIX5_ARB */
+   { 17759, 0x00008635 }, /* GL_MATRIX5_NV */
+   { 17773, 0x000088C6 }, /* GL_MATRIX6_ARB */
+   { 17788, 0x00008636 }, /* GL_MATRIX6_NV */
+   { 17802, 0x000088C7 }, /* GL_MATRIX7_ARB */
+   { 17817, 0x00008637 }, /* GL_MATRIX7_NV */
+   { 17831, 0x000088C8 }, /* GL_MATRIX8_ARB */
+   { 17846, 0x000088C9 }, /* GL_MATRIX9_ARB */
+   { 17861, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
+   { 17887, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+   { 17921, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+   { 17952, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+   { 17985, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+   { 18016, 0x00000BA0 }, /* GL_MATRIX_MODE */
+   { 18031, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
+   { 18053, 0x00008008 }, /* GL_MAX */
+   { 18060, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
+   { 18083, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+   { 18115, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
+   { 18141, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+   { 18174, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+   { 18200, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 18234, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
+   { 18253, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS */
+   { 18278, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+   { 18307, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+   { 18339, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
+   { 18375, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+   { 18411, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
+   { 18451, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
+   { 18477, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
+   { 18507, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
+   { 18532, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
+   { 18561, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+   { 18590, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
+   { 18623, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
+   { 18643, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
+   { 18667, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
+   { 18691, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
+   { 18715, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
+   { 18740, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
+   { 18758, 0x00008008 }, /* GL_MAX_EXT */
+   { 18769, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+   { 18804, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
+   { 18843, 0x00008DE0 }, /* GL_MAX_GEOMETRY_OUTPUT_VERTICES_ARB */
+   { 18879, 0x00008C29 }, /* GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_ARB */
+   { 18919, 0x00008DE1 }, /* GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_ARB */
+   { 18963, 0x00008DDF }, /* GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_ARB */
+   { 19002, 0x00008DDD }, /* GL_MAX_GEOMETRY_VARYING_COMPONENTS_ARB */
+   { 19041, 0x00000D31 }, /* GL_MAX_LIGHTS */
+   { 19055, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
+   { 19075, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+   { 19113, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+   { 19142, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
+   { 19166, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
+   { 19194, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
+   { 19217, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 19254, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 19290, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+   { 19317, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+   { 19346, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+   { 19380, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
+   { 19416, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+   { 19443, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+   { 19475, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+   { 19511, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+   { 19540, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+   { 19569, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
+   { 19597, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+   { 19635, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 19679, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 19722, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 19756, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 19795, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 19832, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 19870, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 19913, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 19956, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+   { 19986, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+   { 20017, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 20053, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 20089, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
+   { 20119, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+   { 20153, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
+   { 20186, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE */
+   { 20211, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+   { 20240, 0x00008D57 }, /* GL_MAX_SAMPLES */
+   { 20255, 0x00008D57 }, /* GL_MAX_SAMPLES_EXT */
+   { 20274, 0x00009111 }, /* GL_MAX_SERVER_WAIT_TIMEOUT */
+   { 20301, 0x00008504 }, /* GL_MAX_SHININESS_NV */
+   { 20321, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
+   { 20345, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
+   { 20367, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
+   { 20393, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+   { 20420, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
+   { 20451, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
+   { 20475, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 20509, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
+   { 20529, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
+   { 20556, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
+   { 20577, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
+   { 20602, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
+   { 20627, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
+   { 20662, 0x00008B4B }, /* GL_MAX_VARYING_COMPONENTS */
+   { 20688, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
+   { 20710, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
+   { 20736, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
+   { 20758, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
+   { 20784, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+   { 20818, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
+   { 20856, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+   { 20889, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
+   { 20926, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
+   { 20950, 0x00008DDE }, /* GL_MAX_VERTEX_VARYING_COMPONENTS_ARB */
+   { 20987, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
+   { 21008, 0x00008007 }, /* GL_MIN */
+   { 21015, 0x0000802E }, /* GL_MINMAX */
+   { 21025, 0x0000802E }, /* GL_MINMAX_EXT */
+   { 21039, 0x0000802F }, /* GL_MINMAX_FORMAT */
+   { 21056, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
+   { 21077, 0x00008030 }, /* GL_MINMAX_SINK */
+   { 21092, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
+   { 21111, 0x00008007 }, /* GL_MIN_EXT */
+   { 21122, 0x00008370 }, /* GL_MIRRORED_REPEAT */
+   { 21141, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
+   { 21164, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
+   { 21187, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
+   { 21207, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
+   { 21227, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+   { 21257, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
+   { 21285, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+   { 21313, 0x00001700 }, /* GL_MODELVIEW */
+   { 21326, 0x00001700 }, /* GL_MODELVIEW0_ARB */
+   { 21344, 0x0000872A }, /* GL_MODELVIEW10_ARB */
+   { 21363, 0x0000872B }, /* GL_MODELVIEW11_ARB */
+   { 21382, 0x0000872C }, /* GL_MODELVIEW12_ARB */
+   { 21401, 0x0000872D }, /* GL_MODELVIEW13_ARB */
+   { 21420, 0x0000872E }, /* GL_MODELVIEW14_ARB */
+   { 21439, 0x0000872F }, /* GL_MODELVIEW15_ARB */
+   { 21458, 0x00008730 }, /* GL_MODELVIEW16_ARB */
+   { 21477, 0x00008731 }, /* GL_MODELVIEW17_ARB */
+   { 21496, 0x00008732 }, /* GL_MODELVIEW18_ARB */
+   { 21515, 0x00008733 }, /* GL_MODELVIEW19_ARB */
+   { 21534, 0x0000850A }, /* GL_MODELVIEW1_ARB */
+   { 21552, 0x00008734 }, /* GL_MODELVIEW20_ARB */
+   { 21571, 0x00008735 }, /* GL_MODELVIEW21_ARB */
+   { 21590, 0x00008736 }, /* GL_MODELVIEW22_ARB */
+   { 21609, 0x00008737 }, /* GL_MODELVIEW23_ARB */
+   { 21628, 0x00008738 }, /* GL_MODELVIEW24_ARB */
+   { 21647, 0x00008739 }, /* GL_MODELVIEW25_ARB */
+   { 21666, 0x0000873A }, /* GL_MODELVIEW26_ARB */
+   { 21685, 0x0000873B }, /* GL_MODELVIEW27_ARB */
+   { 21704, 0x0000873C }, /* GL_MODELVIEW28_ARB */
+   { 21723, 0x0000873D }, /* GL_MODELVIEW29_ARB */
+   { 21742, 0x00008722 }, /* GL_MODELVIEW2_ARB */
+   { 21760, 0x0000873E }, /* GL_MODELVIEW30_ARB */
+   { 21779, 0x0000873F }, /* GL_MODELVIEW31_ARB */
+   { 21798, 0x00008723 }, /* GL_MODELVIEW3_ARB */
+   { 21816, 0x00008724 }, /* GL_MODELVIEW4_ARB */
+   { 21834, 0x00008725 }, /* GL_MODELVIEW5_ARB */
+   { 21852, 0x00008726 }, /* GL_MODELVIEW6_ARB */
+   { 21870, 0x00008727 }, /* GL_MODELVIEW7_ARB */
+   { 21888, 0x00008728 }, /* GL_MODELVIEW8_ARB */
+   { 21906, 0x00008729 }, /* GL_MODELVIEW9_ARB */
+   { 21924, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
+   { 21944, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
+   { 21971, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
+   { 21996, 0x00002100 }, /* GL_MODULATE */
+   { 22008, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
+   { 22028, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
+   { 22055, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
+   { 22080, 0x00000103 }, /* GL_MULT */
+   { 22088, 0x0000809D }, /* GL_MULTISAMPLE */
+   { 22103, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
+   { 22123, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
+   { 22142, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
+   { 22161, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
+   { 22185, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
+   { 22208, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+   { 22238, 0x00002A25 }, /* GL_N3F_V3F */
+   { 22249, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
+   { 22269, 0x0000150E }, /* GL_NAND */
+   { 22277, 0x00002600 }, /* GL_NEAREST */
+   { 22288, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+   { 22319, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+   { 22351, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
+   { 22376, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
+   { 22402, 0x00000200 }, /* GL_NEVER */
+   { 22411, 0x00001102 }, /* GL_NICEST */
+   { 22421, 0x00000000 }, /* GL_NONE */
+   { 22429, 0x00001505 }, /* GL_NOOP */
+   { 22437, 0x00001508 }, /* GL_NOR */
+   { 22444, 0x00000BA1 }, /* GL_NORMALIZE */
+   { 22457, 0x00008075 }, /* GL_NORMAL_ARRAY */
+   { 22473, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+   { 22504, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
+   { 22539, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
+   { 22563, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
+   { 22586, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
+   { 22607, 0x00008511 }, /* GL_NORMAL_MAP */
+   { 22621, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
+   { 22639, 0x00008511 }, /* GL_NORMAL_MAP_NV */
+   { 22656, 0x00000205 }, /* GL_NOTEQUAL */
+   { 22668, 0x00000000 }, /* GL_NO_ERROR */
+   { 22680, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+   { 22714, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
+   { 22752, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
+   { 22784, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
+   { 22826, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
+   { 22856, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
+   { 22896, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
+   { 22927, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
+   { 22956, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
+   { 22984, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
+   { 23014, 0x00002401 }, /* GL_OBJECT_LINEAR */
+   { 23031, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
+   { 23057, 0x00002501 }, /* GL_OBJECT_PLANE */
+   { 23073, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
+   { 23108, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
+   { 23130, 0x00009112 }, /* GL_OBJECT_TYPE */
+   { 23145, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
+   { 23164, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
+   { 23194, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
+   { 23215, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
+   { 23243, 0x00000001 }, /* GL_ONE */
+   { 23250, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+   { 23278, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
+   { 23310, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
+   { 23338, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
+   { 23370, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
+   { 23393, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
+   { 23416, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
+   { 23439, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
+   { 23462, 0x00008598 }, /* GL_OPERAND0_ALPHA */
+   { 23480, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
+   { 23502, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
+   { 23524, 0x00008590 }, /* GL_OPERAND0_RGB */
+   { 23540, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
+   { 23560, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
+   { 23580, 0x00008599 }, /* GL_OPERAND1_ALPHA */
+   { 23598, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
+   { 23620, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
+   { 23642, 0x00008591 }, /* GL_OPERAND1_RGB */
+   { 23658, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
+   { 23678, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
+   { 23698, 0x0000859A }, /* GL_OPERAND2_ALPHA */
+   { 23716, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
+   { 23738, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
+   { 23760, 0x00008592 }, /* GL_OPERAND2_RGB */
+   { 23776, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
+   { 23796, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
+   { 23816, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
+   { 23837, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
+   { 23856, 0x00001507 }, /* GL_OR */
+   { 23862, 0x00000A01 }, /* GL_ORDER */
+   { 23871, 0x0000150D }, /* GL_OR_INVERTED */
+   { 23886, 0x0000150B }, /* GL_OR_REVERSE */
+   { 23900, 0x00000505 }, /* GL_OUT_OF_MEMORY */
+   { 23917, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
+   { 23935, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
+   { 23956, 0x00008758 }, /* GL_PACK_INVERT_MESA */
+   { 23976, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
+   { 23994, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
+   { 24013, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
+   { 24033, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
+   { 24053, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
+   { 24071, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
+   { 24090, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
+   { 24115, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
+   { 24139, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
+   { 24160, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
+   { 24182, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
+   { 24204, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
+   { 24229, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
+   { 24253, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
+   { 24274, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
+   { 24296, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
+   { 24318, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
+   { 24340, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
+   { 24371, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
+   { 24391, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+   { 24416, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
+   { 24436, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+   { 24461, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
+   { 24481, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+   { 24506, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
+   { 24526, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+   { 24551, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
+   { 24571, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+   { 24596, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
+   { 24616, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+   { 24641, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
+   { 24661, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+   { 24686, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
+   { 24706, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+   { 24731, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
+   { 24751, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+   { 24776, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
+   { 24796, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+   { 24821, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
+   { 24839, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER */
+   { 24860, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING */
+   { 24889, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
+   { 24922, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
+   { 24947, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER */
+   { 24970, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
+   { 25001, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
+   { 25036, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
+   { 25063, 0x00001B00 }, /* GL_POINT */
+   { 25072, 0x00000000 }, /* GL_POINTS */
+   { 25082, 0x00000002 }, /* GL_POINT_BIT */
+   { 25095, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
+   { 25125, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
+   { 25159, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
+   { 25193, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
+   { 25228, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
+   { 25257, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
+   { 25290, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
+   { 25323, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
+   { 25357, 0x00000B11 }, /* GL_POINT_SIZE */
+   { 25371, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
+   { 25397, 0x00008127 }, /* GL_POINT_SIZE_MAX */
+   { 25415, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
+   { 25437, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
+   { 25459, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
+   { 25482, 0x00008126 }, /* GL_POINT_SIZE_MIN */
+   { 25500, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
+   { 25522, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
+   { 25544, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
+   { 25567, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
+   { 25587, 0x00000B10 }, /* GL_POINT_SMOOTH */
+   { 25603, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
+   { 25624, 0x00008861 }, /* GL_POINT_SPRITE */
+   { 25640, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
+   { 25660, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
+   { 25689, 0x00008861 }, /* GL_POINT_SPRITE_NV */
+   { 25708, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
+   { 25734, 0x00000701 }, /* GL_POINT_TOKEN */
+   { 25749, 0x00000009 }, /* GL_POLYGON */
+   { 25760, 0x00000008 }, /* GL_POLYGON_BIT */
+   { 25775, 0x00000B40 }, /* GL_POLYGON_MODE */
+   { 25791, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
+   { 25814, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
+   { 25839, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
+   { 25862, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
+   { 25885, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
+   { 25909, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
+   { 25933, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
+   { 25951, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
+   { 25974, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
+   { 25993, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
+   { 26016, 0x00000703 }, /* GL_POLYGON_TOKEN */
+   { 26033, 0x00001203 }, /* GL_POSITION */
+   { 26045, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+   { 26077, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
+   { 26113, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+   { 26146, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
+   { 26183, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+   { 26214, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
+   { 26249, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+   { 26281, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
+   { 26317, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 26350, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+   { 26382, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
+   { 26418, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+   { 26451, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
+   { 26488, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+   { 26518, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
+   { 26552, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+   { 26583, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
+   { 26618, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+   { 26649, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
+   { 26684, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+   { 26716, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
+   { 26752, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+   { 26782, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
+   { 26816, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+   { 26847, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
+   { 26882, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+   { 26914, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+   { 26945, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
+   { 26980, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+   { 27012, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
+   { 27048, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
+   { 27077, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
+   { 27110, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
+   { 27140, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
+   { 27174, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+   { 27213, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+   { 27246, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+   { 27286, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+   { 27320, 0x00008578 }, /* GL_PREVIOUS */
+   { 27332, 0x00008578 }, /* GL_PREVIOUS_ARB */
+   { 27348, 0x00008578 }, /* GL_PREVIOUS_EXT */
+   { 27364, 0x00008577 }, /* GL_PRIMARY_COLOR */
+   { 27381, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
+   { 27402, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
+   { 27423, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 27456, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 27488, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
+   { 27511, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
+   { 27534, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
+   { 27564, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
+   { 27593, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
+   { 27621, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
+   { 27643, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+   { 27671, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+   { 27699, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
+   { 27721, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
+   { 27742, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 27782, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 27821, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 27851, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 27886, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 27919, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 27953, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 27992, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 28031, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
+   { 28053, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
+   { 28079, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
+   { 28103, 0x00008642 }, /* GL_PROGRAM_POINT_SIZE_ARB */
+   { 28129, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
+   { 28152, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
+   { 28174, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
+   { 28195, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
+   { 28216, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
+   { 28243, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 28275, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 28307, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+   { 28342, 0x00001701 }, /* GL_PROJECTION */
+   { 28356, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
+   { 28377, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
+   { 28403, 0x00008E4F }, /* GL_PROVOKING_VERTEX */
+   { 28423, 0x00008E4F }, /* GL_PROVOKING_VERTEX_EXT */
+   { 28447, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
+   { 28468, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
+   { 28487, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
+   { 28510, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 28549, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+   { 28587, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
+   { 28607, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+   { 28637, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
+   { 28661, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
+   { 28681, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+   { 28711, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
+   { 28735, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
+   { 28755, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+   { 28788, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
+   { 28814, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
+   { 28844, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+   { 28875, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
+   { 28905, 0x00002003 }, /* GL_Q */
+   { 28910, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
+   { 28935, 0x00000007 }, /* GL_QUADS */
+   { 28944, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
+   { 28988, 0x00008E4C }, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION_EXT */
+   { 29036, 0x00008614 }, /* GL_QUAD_MESH_SUN */
+   { 29053, 0x00000008 }, /* GL_QUAD_STRIP */
+   { 29067, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
+   { 29089, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
+   { 29115, 0x00008866 }, /* GL_QUERY_RESULT */
+   { 29131, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
+   { 29151, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
+   { 29177, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
+   { 29207, 0x00002002 }, /* GL_R */
+   { 29212, 0x00002A10 }, /* GL_R3_G3_B2 */
+   { 29224, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+   { 29257, 0x00000C02 }, /* GL_READ_BUFFER */
+   { 29272, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER */
+   { 29292, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING */
+   { 29320, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
+   { 29352, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
+   { 29376, 0x000088B8 }, /* GL_READ_ONLY */
+   { 29389, 0x000088B8 }, /* GL_READ_ONLY_ARB */
+   { 29406, 0x000088BA }, /* GL_READ_WRITE */
+   { 29420, 0x000088BA }, /* GL_READ_WRITE_ARB */
+   { 29438, 0x00001903 }, /* GL_RED */
+   { 29445, 0x00008016 }, /* GL_REDUCE */
+   { 29455, 0x00008016 }, /* GL_REDUCE_EXT */
+   { 29469, 0x00000D15 }, /* GL_RED_BIAS */
+   { 29481, 0x00000D52 }, /* GL_RED_BITS */
+   { 29493, 0x00000D14 }, /* GL_RED_SCALE */
+   { 29506, 0x00008512 }, /* GL_REFLECTION_MAP */
+   { 29524, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
+   { 29546, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
+   { 29567, 0x00001C00 }, /* GL_RENDER */
+   { 29577, 0x00008D41 }, /* GL_RENDERBUFFER */
+   { 29593, 0x00008D53 }, /* GL_RENDERBUFFER_ALPHA_SIZE */
+   { 29620, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING */
+   { 29644, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
+   { 29672, 0x00008D52 }, /* GL_RENDERBUFFER_BLUE_SIZE */
+   { 29698, 0x00008D54 }, /* GL_RENDERBUFFER_DEPTH_SIZE */
+   { 29725, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
+   { 29745, 0x00008D51 }, /* GL_RENDERBUFFER_GREEN_SIZE */
+   { 29772, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT */
+   { 29795, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
+   { 29822, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
+   { 29854, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
+   { 29890, 0x00008D50 }, /* GL_RENDERBUFFER_RED_SIZE */
+   { 29915, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES */
+   { 29939, 0x00008CAB }, /* GL_RENDERBUFFER_SAMPLES_EXT */
+   { 29967, 0x00008D55 }, /* GL_RENDERBUFFER_STENCIL_SIZE */
+   { 29996, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH */
+   { 30018, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
+   { 30044, 0x00001F01 }, /* GL_RENDERER */
+   { 30056, 0x00000C40 }, /* GL_RENDER_MODE */
+   { 30071, 0x00002901 }, /* GL_REPEAT */
+   { 30081, 0x00001E01 }, /* GL_REPLACE */
+   { 30092, 0x00008062 }, /* GL_REPLACE_EXT */
+   { 30107, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
+   { 30130, 0x0000803A }, /* GL_RESCALE_NORMAL */
+   { 30148, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
+   { 30170, 0x00000102 }, /* GL_RETURN */
+   { 30180, 0x00001907 }, /* GL_RGB */
+   { 30187, 0x00008052 }, /* GL_RGB10 */
+   { 30196, 0x00008059 }, /* GL_RGB10_A2 */
+   { 30208, 0x00008059 }, /* GL_RGB10_A2_EXT */
+   { 30224, 0x00008052 }, /* GL_RGB10_EXT */
+   { 30237, 0x00008053 }, /* GL_RGB12 */
+   { 30246, 0x00008053 }, /* GL_RGB12_EXT */
+   { 30259, 0x00008054 }, /* GL_RGB16 */
+   { 30268, 0x00008054 }, /* GL_RGB16_EXT */
+   { 30281, 0x0000804E }, /* GL_RGB2_EXT */
+   { 30293, 0x0000804F }, /* GL_RGB4 */
+   { 30301, 0x0000804F }, /* GL_RGB4_EXT */
+   { 30313, 0x000083A1 }, /* GL_RGB4_S3TC */
+   { 30326, 0x00008050 }, /* GL_RGB5 */
+   { 30334, 0x00008057 }, /* GL_RGB5_A1 */
+   { 30345, 0x00008057 }, /* GL_RGB5_A1_EXT */
+   { 30360, 0x00008050 }, /* GL_RGB5_EXT */
+   { 30372, 0x00008051 }, /* GL_RGB8 */
+   { 30380, 0x00008051 }, /* GL_RGB8_EXT */
+   { 30392, 0x00001908 }, /* GL_RGBA */
+   { 30400, 0x0000805A }, /* GL_RGBA12 */
+   { 30410, 0x0000805A }, /* GL_RGBA12_EXT */
+   { 30424, 0x0000805B }, /* GL_RGBA16 */
+   { 30434, 0x0000805B }, /* GL_RGBA16_EXT */
+   { 30448, 0x00008055 }, /* GL_RGBA2 */
+   { 30457, 0x00008055 }, /* GL_RGBA2_EXT */
+   { 30470, 0x00008056 }, /* GL_RGBA4 */
+   { 30479, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
+   { 30498, 0x00008056 }, /* GL_RGBA4_EXT */
+   { 30511, 0x000083A3 }, /* GL_RGBA4_S3TC */
+   { 30525, 0x00008058 }, /* GL_RGBA8 */
+   { 30534, 0x00008058 }, /* GL_RGBA8_EXT */
+   { 30547, 0x00008F97 }, /* GL_RGBA8_SNORM */
+   { 30562, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
+   { 30580, 0x00000C31 }, /* GL_RGBA_MODE */
+   { 30593, 0x000083A2 }, /* GL_RGBA_S3TC */
+   { 30606, 0x00008F93 }, /* GL_RGBA_SNORM */
+   { 30620, 0x000083A0 }, /* GL_RGB_S3TC */
+   { 30632, 0x00008573 }, /* GL_RGB_SCALE */
+   { 30645, 0x00008573 }, /* GL_RGB_SCALE_ARB */
+   { 30662, 0x00008573 }, /* GL_RGB_SCALE_EXT */
+   { 30679, 0x00000407 }, /* GL_RIGHT */
+   { 30688, 0x00002000 }, /* GL_S */
+   { 30693, 0x00008B5D }, /* GL_SAMPLER_1D */
+   { 30707, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
+   { 30728, 0x00008B5E }, /* GL_SAMPLER_2D */
+   { 30742, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
+   { 30763, 0x00008B5F }, /* GL_SAMPLER_3D */
+   { 30777, 0x00008B60 }, /* GL_SAMPLER_CUBE */
+   { 30793, 0x000080A9 }, /* GL_SAMPLES */
+   { 30804, 0x000086B4 }, /* GL_SAMPLES_3DFX */
+   { 30820, 0x000080A9 }, /* GL_SAMPLES_ARB */
+   { 30835, 0x00008914 }, /* GL_SAMPLES_PASSED */
+   { 30853, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
+   { 30875, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+   { 30903, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
+   { 30935, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
+   { 30958, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
+   { 30985, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
+   { 31003, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
+   { 31026, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
+   { 31048, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
+   { 31067, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
+   { 31090, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
+   { 31116, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
+   { 31146, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
+   { 31171, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
+   { 31200, 0x00080000 }, /* GL_SCISSOR_BIT */
+   { 31215, 0x00000C10 }, /* GL_SCISSOR_BOX */
+   { 31230, 0x00000C11 }, /* GL_SCISSOR_TEST */
+   { 31246, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
+   { 31271, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+   { 31311, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
+   { 31355, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+   { 31388, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+   { 31418, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+   { 31450, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+   { 31480, 0x00001C02 }, /* GL_SELECT */
+   { 31490, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
+   { 31518, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
+   { 31543, 0x00008012 }, /* GL_SEPARABLE_2D */
+   { 31559, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
+   { 31586, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
+   { 31617, 0x0000150F }, /* GL_SET */
+   { 31624, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
+   { 31645, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
+   { 31669, 0x00008B4F }, /* GL_SHADER_TYPE */
+   { 31684, 0x00000B54 }, /* GL_SHADE_MODEL */
+   { 31699, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
+   { 31727, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
+   { 31750, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+   { 31780, 0x00001601 }, /* GL_SHININESS */
+   { 31793, 0x00001402 }, /* GL_SHORT */
+   { 31802, 0x00009119 }, /* GL_SIGNALED */
+   { 31814, 0x00008F9C }, /* GL_SIGNED_NORMALIZED */
+   { 31835, 0x000081F9 }, /* GL_SINGLE_COLOR */
+   { 31851, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
+   { 31871, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
+   { 31890, 0x00008C46 }, /* GL_SLUMINANCE */
+   { 31904, 0x00008C47 }, /* GL_SLUMINANCE8 */
+   { 31919, 0x00008C45 }, /* GL_SLUMINANCE8_ALPHA8 */
+   { 31941, 0x00008C44 }, /* GL_SLUMINANCE_ALPHA */
+   { 31961, 0x00001D01 }, /* GL_SMOOTH */
+   { 31971, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
+   { 32004, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
+   { 32031, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
+   { 32064, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
+   { 32091, 0x00008588 }, /* GL_SOURCE0_ALPHA */
+   { 32108, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
+   { 32129, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
+   { 32150, 0x00008580 }, /* GL_SOURCE0_RGB */
+   { 32165, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
+   { 32184, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
+   { 32203, 0x00008589 }, /* GL_SOURCE1_ALPHA */
+   { 32220, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
+   { 32241, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
+   { 32262, 0x00008581 }, /* GL_SOURCE1_RGB */
+   { 32277, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
+   { 32296, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
+   { 32315, 0x0000858A }, /* GL_SOURCE2_ALPHA */
+   { 32332, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
+   { 32353, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
+   { 32374, 0x00008582 }, /* GL_SOURCE2_RGB */
+   { 32389, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
+   { 32408, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
+   { 32427, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
+   { 32447, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
+   { 32465, 0x00001202 }, /* GL_SPECULAR */
+   { 32477, 0x00002402 }, /* GL_SPHERE_MAP */
+   { 32491, 0x00001206 }, /* GL_SPOT_CUTOFF */
+   { 32506, 0x00001204 }, /* GL_SPOT_DIRECTION */
+   { 32524, 0x00001205 }, /* GL_SPOT_EXPONENT */
+   { 32541, 0x00008588 }, /* GL_SRC0_ALPHA */
+   { 32555, 0x00008580 }, /* GL_SRC0_RGB */
+   { 32567, 0x00008589 }, /* GL_SRC1_ALPHA */
+   { 32581, 0x00008581 }, /* GL_SRC1_RGB */
+   { 32593, 0x0000858A }, /* GL_SRC2_ALPHA */
+   { 32607, 0x00008582 }, /* GL_SRC2_RGB */
+   { 32619, 0x00000302 }, /* GL_SRC_ALPHA */
+   { 32632, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
+   { 32654, 0x00000300 }, /* GL_SRC_COLOR */
+   { 32667, 0x00008C40 }, /* GL_SRGB */
+   { 32675, 0x00008C41 }, /* GL_SRGB8 */
+   { 32684, 0x00008C43 }, /* GL_SRGB8_ALPHA8 */
+   { 32700, 0x00008C42 }, /* GL_SRGB_ALPHA */
+   { 32714, 0x00000503 }, /* GL_STACK_OVERFLOW */
+   { 32732, 0x00000504 }, /* GL_STACK_UNDERFLOW */
+   { 32751, 0x000088E6 }, /* GL_STATIC_COPY */
+   { 32766, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
+   { 32785, 0x000088E4 }, /* GL_STATIC_DRAW */
+   { 32800, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
+   { 32819, 0x000088E5 }, /* GL_STATIC_READ */
+   { 32834, 0x000088E5 }, /* GL_STATIC_READ_ARB */
+   { 32853, 0x00001802 }, /* GL_STENCIL */
+   { 32864, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT */
+   { 32886, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
+   { 32912, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
+   { 32933, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
+   { 32958, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
+   { 32979, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
+   { 33004, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+   { 33036, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
+   { 33072, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+   { 33104, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
+   { 33140, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
+   { 33160, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
+   { 33187, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
+   { 33213, 0x00000D57 }, /* GL_STENCIL_BITS */
+   { 33229, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
+   { 33251, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
+   { 33274, 0x00000B94 }, /* GL_STENCIL_FAIL */
+   { 33290, 0x00000B92 }, /* GL_STENCIL_FUNC */
+   { 33306, 0x00001901 }, /* GL_STENCIL_INDEX */
+   { 33323, 0x00008D46 }, /* GL_STENCIL_INDEX1 */
+   { 33341, 0x00008D49 }, /* GL_STENCIL_INDEX16 */
+   { 33360, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
+   { 33383, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
+   { 33405, 0x00008D47 }, /* GL_STENCIL_INDEX4 */
+   { 33423, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
+   { 33445, 0x00008D48 }, /* GL_STENCIL_INDEX8 */
+   { 33463, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
+   { 33485, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
+   { 33506, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
+   { 33533, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
+   { 33560, 0x00000B97 }, /* GL_STENCIL_REF */
+   { 33575, 0x00000B90 }, /* GL_STENCIL_TEST */
+   { 33591, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+   { 33620, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
+   { 33642, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
+   { 33663, 0x00000C33 }, /* GL_STEREO */
+   { 33673, 0x000085BE }, /* GL_STORAGE_CACHED_APPLE */
+   { 33697, 0x000085BD }, /* GL_STORAGE_PRIVATE_APPLE */
+   { 33722, 0x000085BF }, /* GL_STORAGE_SHARED_APPLE */
+   { 33746, 0x000088E2 }, /* GL_STREAM_COPY */
+   { 33761, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
+   { 33780, 0x000088E0 }, /* GL_STREAM_DRAW */
+   { 33795, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
+   { 33814, 0x000088E1 }, /* GL_STREAM_READ */
+   { 33829, 0x000088E1 }, /* GL_STREAM_READ_ARB */
+   { 33848, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
+   { 33865, 0x000084E7 }, /* GL_SUBTRACT */
+   { 33877, 0x000084E7 }, /* GL_SUBTRACT_ARB */
+   { 33893, 0x00009113 }, /* GL_SYNC_CONDITION */
+   { 33911, 0x00009116 }, /* GL_SYNC_FENCE */
+   { 33925, 0x00009115 }, /* GL_SYNC_FLAGS */
+   { 33939, 0x00000001 }, /* GL_SYNC_FLUSH_COMMANDS_BIT */
+   { 33966, 0x00009117 }, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
+   { 33996, 0x00009114 }, /* GL_SYNC_STATUS */
+   { 34011, 0x00002001 }, /* GL_T */
+   { 34016, 0x00002A2A }, /* GL_T2F_C3F_V3F */
+   { 34031, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
+   { 34050, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
+   { 34066, 0x00002A2B }, /* GL_T2F_N3F_V3F */
+   { 34081, 0x00002A27 }, /* GL_T2F_V3F */
+   { 34092, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
+   { 34111, 0x00002A28 }, /* GL_T4F_V4F */
+   { 34122, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
+   { 34145, 0x00001702 }, /* GL_TEXTURE */
+   { 34156, 0x000084C0 }, /* GL_TEXTURE0 */
+   { 34168, 0x000084C0 }, /* GL_TEXTURE0_ARB */
+   { 34184, 0x000084C1 }, /* GL_TEXTURE1 */
+   { 34196, 0x000084CA }, /* GL_TEXTURE10 */
+   { 34209, 0x000084CA }, /* GL_TEXTURE10_ARB */
+   { 34226, 0x000084CB }, /* GL_TEXTURE11 */
+   { 34239, 0x000084CB }, /* GL_TEXTURE11_ARB */
+   { 34256, 0x000084CC }, /* GL_TEXTURE12 */
+   { 34269, 0x000084CC }, /* GL_TEXTURE12_ARB */
+   { 34286, 0x000084CD }, /* GL_TEXTURE13 */
+   { 34299, 0x000084CD }, /* GL_TEXTURE13_ARB */
+   { 34316, 0x000084CE }, /* GL_TEXTURE14 */
+   { 34329, 0x000084CE }, /* GL_TEXTURE14_ARB */
+   { 34346, 0x000084CF }, /* GL_TEXTURE15 */
+   { 34359, 0x000084CF }, /* GL_TEXTURE15_ARB */
+   { 34376, 0x000084D0 }, /* GL_TEXTURE16 */
+   { 34389, 0x000084D0 }, /* GL_TEXTURE16_ARB */
+   { 34406, 0x000084D1 }, /* GL_TEXTURE17 */
+   { 34419, 0x000084D1 }, /* GL_TEXTURE17_ARB */
+   { 34436, 0x000084D2 }, /* GL_TEXTURE18 */
+   { 34449, 0x000084D2 }, /* GL_TEXTURE18_ARB */
+   { 34466, 0x000084D3 }, /* GL_TEXTURE19 */
+   { 34479, 0x000084D3 }, /* GL_TEXTURE19_ARB */
+   { 34496, 0x000084C1 }, /* GL_TEXTURE1_ARB */
+   { 34512, 0x000084C2 }, /* GL_TEXTURE2 */
+   { 34524, 0x000084D4 }, /* GL_TEXTURE20 */
+   { 34537, 0x000084D4 }, /* GL_TEXTURE20_ARB */
+   { 34554, 0x000084D5 }, /* GL_TEXTURE21 */
+   { 34567, 0x000084D5 }, /* GL_TEXTURE21_ARB */
+   { 34584, 0x000084D6 }, /* GL_TEXTURE22 */
+   { 34597, 0x000084D6 }, /* GL_TEXTURE22_ARB */
+   { 34614, 0x000084D7 }, /* GL_TEXTURE23 */
+   { 34627, 0x000084D7 }, /* GL_TEXTURE23_ARB */
+   { 34644, 0x000084D8 }, /* GL_TEXTURE24 */
+   { 34657, 0x000084D8 }, /* GL_TEXTURE24_ARB */
+   { 34674, 0x000084D9 }, /* GL_TEXTURE25 */
+   { 34687, 0x000084D9 }, /* GL_TEXTURE25_ARB */
+   { 34704, 0x000084DA }, /* GL_TEXTURE26 */
+   { 34717, 0x000084DA }, /* GL_TEXTURE26_ARB */
+   { 34734, 0x000084DB }, /* GL_TEXTURE27 */
+   { 34747, 0x000084DB }, /* GL_TEXTURE27_ARB */
+   { 34764, 0x000084DC }, /* GL_TEXTURE28 */
+   { 34777, 0x000084DC }, /* GL_TEXTURE28_ARB */
+   { 34794, 0x000084DD }, /* GL_TEXTURE29 */
+   { 34807, 0x000084DD }, /* GL_TEXTURE29_ARB */
+   { 34824, 0x000084C2 }, /* GL_TEXTURE2_ARB */
+   { 34840, 0x000084C3 }, /* GL_TEXTURE3 */
+   { 34852, 0x000084DE }, /* GL_TEXTURE30 */
+   { 34865, 0x000084DE }, /* GL_TEXTURE30_ARB */
+   { 34882, 0x000084DF }, /* GL_TEXTURE31 */
+   { 34895, 0x000084DF }, /* GL_TEXTURE31_ARB */
+   { 34912, 0x000084C3 }, /* GL_TEXTURE3_ARB */
+   { 34928, 0x000084C4 }, /* GL_TEXTURE4 */
+   { 34940, 0x000084C4 }, /* GL_TEXTURE4_ARB */
+   { 34956, 0x000084C5 }, /* GL_TEXTURE5 */
+   { 34968, 0x000084C5 }, /* GL_TEXTURE5_ARB */
+   { 34984, 0x000084C6 }, /* GL_TEXTURE6 */
+   { 34996, 0x000084C6 }, /* GL_TEXTURE6_ARB */
+   { 35012, 0x000084C7 }, /* GL_TEXTURE7 */
+   { 35024, 0x000084C7 }, /* GL_TEXTURE7_ARB */
+   { 35040, 0x000084C8 }, /* GL_TEXTURE8 */
+   { 35052, 0x000084C8 }, /* GL_TEXTURE8_ARB */
+   { 35068, 0x000084C9 }, /* GL_TEXTURE9 */
+   { 35080, 0x000084C9 }, /* GL_TEXTURE9_ARB */
+   { 35096, 0x00000DE0 }, /* GL_TEXTURE_1D */
+   { 35110, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
+   { 35134, 0x00000DE1 }, /* GL_TEXTURE_2D */
+   { 35148, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
+   { 35172, 0x0000806F }, /* GL_TEXTURE_3D */
+   { 35186, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
+   { 35208, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
+   { 35234, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
+   { 35256, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
+   { 35278, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+   { 35310, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
+   { 35332, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+   { 35364, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
+   { 35386, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
+   { 35414, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
+   { 35446, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+   { 35479, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
+   { 35511, 0x00040000 }, /* GL_TEXTURE_BIT */
+   { 35526, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
+   { 35547, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
+   { 35572, 0x00001005 }, /* GL_TEXTURE_BORDER */
+   { 35590, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
+   { 35614, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+   { 35645, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+   { 35675, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+   { 35705, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+   { 35740, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+   { 35771, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 35809, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
+   { 35836, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+   { 35868, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+   { 35902, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
+   { 35926, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
+   { 35954, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
+   { 35978, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
+   { 36006, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+   { 36039, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
+   { 36063, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
+   { 36085, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
+   { 36107, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
+   { 36133, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
+   { 36167, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+   { 36200, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
+   { 36237, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
+   { 36265, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
+   { 36297, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
+   { 36320, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+   { 36358, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
+   { 36400, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+   { 36431, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+   { 36459, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+   { 36489, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+   { 36517, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
+   { 36537, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
+   { 36561, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+   { 36592, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
+   { 36627, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+   { 36658, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
+   { 36693, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+   { 36724, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
+   { 36759, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+   { 36790, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
+   { 36825, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+   { 36856, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
+   { 36891, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+   { 36922, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
+   { 36957, 0x000088F4 }, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
+   { 36986, 0x00008071 }, /* GL_TEXTURE_DEPTH */
+   { 37003, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
+   { 37025, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
+   { 37051, 0x00002300 }, /* GL_TEXTURE_ENV */
+   { 37066, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
+   { 37087, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
+   { 37107, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
+   { 37133, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
+   { 37153, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
+   { 37170, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
+   { 37187, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
+   { 37204, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
+   { 37221, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
+   { 37246, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
+   { 37268, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
+   { 37294, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
+   { 37312, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
+   { 37338, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
+   { 37364, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
+   { 37394, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
+   { 37421, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
+   { 37446, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
+   { 37466, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
+   { 37490, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+   { 37517, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+   { 37544, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+   { 37571, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
+   { 37597, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
+   { 37627, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
+   { 37649, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
+   { 37667, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 37697, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+   { 37725, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+   { 37753, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+   { 37781, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
+   { 37802, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
+   { 37821, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
+   { 37843, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
+   { 37862, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
+   { 37882, 0x000085B7 }, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
+   { 37912, 0x000085B8 }, /* GL_TEXTURE_RANGE_POINTER_APPLE */
+   { 37943, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
+   { 37968, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
+   { 37992, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
+   { 38012, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
+   { 38036, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
+   { 38056, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
+   { 38079, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE */
+   { 38103, 0x000088F1 }, /* GL_TEXTURE_STENCIL_SIZE_EXT */
+   { 38131, 0x000085BC }, /* GL_TEXTURE_STORAGE_HINT_APPLE */
+   { 38161, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
+   { 38186, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+   { 38220, 0x00001000 }, /* GL_TEXTURE_WIDTH */
+   { 38237, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
+   { 38255, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
+   { 38273, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
+   { 38291, 0x0000911B }, /* GL_TIMEOUT_EXPIRED */
+   { 38310, 0xFFFFFFFFFFFFFFFF }, /* GL_TIMEOUT_IGNORED */
+   { 38329, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
+   { 38349, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
+   { 38368, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+   { 38397, 0x00001000 }, /* GL_TRANSFORM_BIT */
+   { 38414, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
+   { 38440, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
+   { 38470, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+   { 38502, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+   { 38532, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
+   { 38566, 0x0000862C }, /* GL_TRANSPOSE_NV */
+   { 38582, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+   { 38613, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
+   { 38648, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+   { 38676, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
+   { 38708, 0x00000004 }, /* GL_TRIANGLES */
+   { 38721, 0x0000000C }, /* GL_TRIANGLES_ADJACENCY_ARB */
+   { 38748, 0x00000006 }, /* GL_TRIANGLE_FAN */
+   { 38764, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
+   { 38785, 0x00000005 }, /* GL_TRIANGLE_STRIP */
+   { 38803, 0x0000000D }, /* GL_TRIANGLE_STRIP_ADJACENCY_ARB */
+   { 38835, 0x00000001 }, /* GL_TRUE */
+   { 38843, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
+   { 38863, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
+   { 38886, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
+   { 38906, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
+   { 38927, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
+   { 38949, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
+   { 38971, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
+   { 38991, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
+   { 39012, 0x00009118 }, /* GL_UNSIGNALED */
+   { 39026, 0x00001401 }, /* GL_UNSIGNED_BYTE */
+   { 39043, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+   { 39070, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
+   { 39093, 0x00001405 }, /* GL_UNSIGNED_INT */
+   { 39109, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
+   { 39136, 0x000084FA }, /* GL_UNSIGNED_INT_24_8 */
+   { 39157, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_EXT */
+   { 39182, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
+   { 39206, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+   { 39237, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
+   { 39261, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+   { 39289, 0x00008C17 }, /* GL_UNSIGNED_NORMALIZED */
+   { 39312, 0x00001403 }, /* GL_UNSIGNED_SHORT */
+   { 39330, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+   { 39360, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+   { 39386, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+   { 39416, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+   { 39442, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
+   { 39466, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+   { 39494, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+   { 39522, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
+   { 39549, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+   { 39581, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
+   { 39612, 0x00008CA2 }, /* GL_UPPER_LEFT */
+   { 39626, 0x00002A20 }, /* GL_V2F */
+   { 39633, 0x00002A21 }, /* GL_V3F */
+   { 39640, 0x00008B83 }, /* GL_VALIDATE_STATUS */
+   { 39659, 0x00001F00 }, /* GL_VENDOR */
+   { 39669, 0x00001F02 }, /* GL_VERSION */
+   { 39680, 0x00008074 }, /* GL_VERTEX_ARRAY */
+   { 39696, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING */
+   { 39720, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
+   { 39750, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+   { 39781, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
+   { 39816, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
+   { 39840, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
+   { 39861, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
+   { 39884, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
+   { 39905, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+   { 39932, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+   { 39960, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+   { 39988, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+   { 40016, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+   { 40044, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+   { 40072, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+   { 40100, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+   { 40127, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+   { 40154, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+   { 40181, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+   { 40208, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+   { 40235, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+   { 40262, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+   { 40289, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+   { 40316, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+   { 40343, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+   { 40381, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
+   { 40423, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+   { 40454, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
+   { 40489, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+   { 40523, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
+   { 40561, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+   { 40592, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
+   { 40627, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+   { 40655, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
+   { 40687, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+   { 40717, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
+   { 40751, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+   { 40779, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
+   { 40811, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
+   { 40831, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
+   { 40853, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
+   { 40882, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
+   { 40903, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+   { 40932, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
+   { 40965, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
+   { 40997, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+   { 41024, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
+   { 41055, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
+   { 41085, 0x00008B31 }, /* GL_VERTEX_SHADER */
+   { 41102, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
+   { 41123, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
+   { 41150, 0x00000BA2 }, /* GL_VIEWPORT */
+   { 41162, 0x00000800 }, /* GL_VIEWPORT_BIT */
+   { 41178, 0x0000911D }, /* GL_WAIT_FAILED */
+   { 41193, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
+   { 41213, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+   { 41244, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
+   { 41279, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+   { 41307, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+   { 41332, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+   { 41359, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+   { 41384, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
+   { 41408, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
+   { 41427, 0x000088B9 }, /* GL_WRITE_ONLY */
+   { 41441, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
+   { 41459, 0x00001506 }, /* GL_XOR */
+   { 41466, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
+   { 41485, 0x00008757 }, /* GL_YCBCR_MESA */
+   { 41499, 0x00000000 }, /* GL_ZERO */
+   { 41507, 0x00000D16 }, /* GL_ZOOM_X */
+   { 41517, 0x00000D17 }, /* GL_ZOOM_Y */
 };
 
 static const unsigned reduced_enums[1364] =
 {
-       476, /* GL_FALSE */
-       701, /* GL_LINES */
-       704, /* GL_LINE_LOOP */
-       711, /* GL_LINE_STRIP */
-      1765, /* GL_TRIANGLES */
-      1769, /* GL_TRIANGLE_STRIP */
-      1767, /* GL_TRIANGLE_FAN */
-      1292, /* GL_QUADS */
-      1296, /* GL_QUAD_STRIP */
-      1177, /* GL_POLYGON */
-       702, /* GL_LINES_ADJACENCY_ARB */
-       712, /* GL_LINE_STRIP_ADJACENCY_ARB */
-      1766, /* GL_TRIANGLES_ADJACENCY_ARB */
-      1770, /* GL_TRIANGLE_STRIP_ADJACENCY_ARB */
-      1189, /* GL_POLYGON_STIPPLE_BIT */
-      1138, /* GL_PIXEL_MODE_BIT */
-       688, /* GL_LIGHTING_BIT */
-       506, /* GL_FOG_BIT */
+       479, /* GL_FALSE */
+       708, /* GL_LINES */
+       711, /* GL_LINE_LOOP */
+       718, /* GL_LINE_STRIP */
+      1783, /* GL_TRIANGLES */
+      1787, /* GL_TRIANGLE_STRIP */
+      1785, /* GL_TRIANGLE_FAN */
+      1302, /* GL_QUADS */
+      1306, /* GL_QUAD_STRIP */
+      1187, /* GL_POLYGON */
+       709, /* GL_LINES_ADJACENCY_ARB */
+       719, /* GL_LINE_STRIP_ADJACENCY_ARB */
+      1784, /* GL_TRIANGLES_ADJACENCY_ARB */
+      1788, /* GL_TRIANGLE_STRIP_ADJACENCY_ARB */
+      1199, /* GL_POLYGON_STIPPLE_BIT */
+      1148, /* GL_PIXEL_MODE_BIT */
+       695, /* GL_LIGHTING_BIT */
+       509, /* GL_FOG_BIT */
          8, /* GL_ACCUM */
-       722, /* GL_LOAD */
-      1348, /* GL_RETURN */
-      1010, /* GL_MULT */
+       729, /* GL_LOAD */
+      1361, /* GL_RETURN */
+      1020, /* GL_MULT */
         23, /* GL_ADD */
-      1026, /* GL_NEVER */
-       678, /* GL_LESS */
-       466, /* GL_EQUAL */
-       677, /* GL_LEQUAL */
-       599, /* GL_GREATER */
-      1041, /* GL_NOTEQUAL */
-       598, /* GL_GEQUAL */
+      1036, /* GL_NEVER */
+       685, /* GL_LESS */
+       469, /* GL_EQUAL */
+       684, /* GL_LEQUAL */
+       606, /* GL_GREATER */
+      1051, /* GL_NOTEQUAL */
+       605, /* GL_GEQUAL */
         47, /* GL_ALWAYS */
-      1489, /* GL_SRC_COLOR */
-      1071, /* GL_ONE_MINUS_SRC_COLOR */
-      1487, /* GL_SRC_ALPHA */
-      1070, /* GL_ONE_MINUS_SRC_ALPHA */
-       445, /* GL_DST_ALPHA */
-      1068, /* GL_ONE_MINUS_DST_ALPHA */
-       446, /* GL_DST_COLOR */
-      1069, /* GL_ONE_MINUS_DST_COLOR */
-      1488, /* GL_SRC_ALPHA_SATURATE */
-       582, /* GL_FRONT_LEFT */
-       583, /* GL_FRONT_RIGHT */
+      1502, /* GL_SRC_COLOR */
+      1081, /* GL_ONE_MINUS_SRC_COLOR */
+      1500, /* GL_SRC_ALPHA */
+      1080, /* GL_ONE_MINUS_SRC_ALPHA */
+       448, /* GL_DST_ALPHA */
+      1078, /* GL_ONE_MINUS_DST_ALPHA */
+       449, /* GL_DST_COLOR */
+      1079, /* GL_ONE_MINUS_DST_COLOR */
+      1501, /* GL_SRC_ALPHA_SATURATE */
+       589, /* GL_FRONT_LEFT */
+       590, /* GL_FRONT_RIGHT */
         69, /* GL_BACK_LEFT */
         70, /* GL_BACK_RIGHT */
-       579, /* GL_FRONT */
+       586, /* GL_FRONT */
         68, /* GL_BACK */
-       676, /* GL_LEFT */
-      1390, /* GL_RIGHT */
-       580, /* GL_FRONT_AND_BACK */
+       683, /* GL_LEFT */
+      1403, /* GL_RIGHT */
+       587, /* GL_FRONT_AND_BACK */
         63, /* GL_AUX0 */
         64, /* GL_AUX1 */
         65, /* GL_AUX2 */
         66, /* GL_AUX3 */
-       665, /* GL_INVALID_ENUM */
-       669, /* GL_INVALID_VALUE */
-       668, /* GL_INVALID_OPERATION */
-      1494, /* GL_STACK_OVERFLOW */
-      1495, /* GL_STACK_UNDERFLOW */
-      1096, /* GL_OUT_OF_MEMORY */
-       666, /* GL_INVALID_FRAMEBUFFER_OPERATION */
+       672, /* GL_INVALID_ENUM */
+       676, /* GL_INVALID_VALUE */
+       675, /* GL_INVALID_OPERATION */
+      1507, /* GL_STACK_OVERFLOW */
+      1508, /* GL_STACK_UNDERFLOW */
+      1106, /* GL_OUT_OF_MEMORY */
+       673, /* GL_INVALID_FRAMEBUFFER_OPERATION */
          0, /* GL_2D */
          2, /* GL_3D */
          3, /* GL_3D_COLOR */
          4, /* GL_3D_COLOR_TEXTURE */
          6, /* GL_4D_COLOR_TEXTURE */
-      1116, /* GL_PASS_THROUGH_TOKEN */
-      1176, /* GL_POINT_TOKEN */
-       713, /* GL_LINE_TOKEN */
-      1190, /* GL_POLYGON_TOKEN */
+      1126, /* GL_PASS_THROUGH_TOKEN */
+      1186, /* GL_POINT_TOKEN */
+       720, /* GL_LINE_TOKEN */
+      1200, /* GL_POLYGON_TOKEN */
         74, /* GL_BITMAP_TOKEN */
-       444, /* GL_DRAW_PIXEL_TOKEN */
+       447, /* GL_DRAW_PIXEL_TOKEN */
        301, /* GL_COPY_PIXEL_TOKEN */
-       705, /* GL_LINE_RESET_TOKEN */
-       469, /* GL_EXP */
-       470, /* GL_EXP2 */
+       712, /* GL_LINE_RESET_TOKEN */
+       472, /* GL_EXP */
+       473, /* GL_EXP2 */
        337, /* GL_CW */
        125, /* GL_CCW */
        146, /* GL_COEFF */
-      1093, /* GL_ORDER */
-       382, /* GL_DOMAIN */
+      1103, /* GL_ORDER */
+       384, /* GL_DOMAIN */
        311, /* GL_CURRENT_COLOR */
        314, /* GL_CURRENT_INDEX */
        320, /* GL_CURRENT_NORMAL */
@@ -3896,519 +3934,519 @@ static const unsigned reduced_enums[1364] =
        328, /* GL_CURRENT_RASTER_POSITION */
        329, /* GL_CURRENT_RASTER_POSITION_VALID */
        326, /* GL_CURRENT_RASTER_DISTANCE */
-      1169, /* GL_POINT_SMOOTH */
-      1158, /* GL_POINT_SIZE */
-      1168, /* GL_POINT_SIZE_RANGE */
-      1159, /* GL_POINT_SIZE_GRANULARITY */
-       706, /* GL_LINE_SMOOTH */
-       714, /* GL_LINE_WIDTH */
-       716, /* GL_LINE_WIDTH_RANGE */
-       715, /* GL_LINE_WIDTH_GRANULARITY */
-       708, /* GL_LINE_STIPPLE */
-       709, /* GL_LINE_STIPPLE_PATTERN */
-       710, /* GL_LINE_STIPPLE_REPEAT */
-       721, /* GL_LIST_MODE */
-       891, /* GL_MAX_LIST_NESTING */
-       718, /* GL_LIST_BASE */
-       720, /* GL_LIST_INDEX */
-      1179, /* GL_POLYGON_MODE */
-      1186, /* GL_POLYGON_SMOOTH */
-      1188, /* GL_POLYGON_STIPPLE */
-       455, /* GL_EDGE_FLAG */
+      1179, /* GL_POINT_SMOOTH */
+      1168, /* GL_POINT_SIZE */
+      1178, /* GL_POINT_SIZE_RANGE */
+      1169, /* GL_POINT_SIZE_GRANULARITY */
+       713, /* GL_LINE_SMOOTH */
+       721, /* GL_LINE_WIDTH */
+       723, /* GL_LINE_WIDTH_RANGE */
+       722, /* GL_LINE_WIDTH_GRANULARITY */
+       715, /* GL_LINE_STIPPLE */
+       716, /* GL_LINE_STIPPLE_PATTERN */
+       717, /* GL_LINE_STIPPLE_REPEAT */
+       728, /* GL_LIST_MODE */
+       899, /* GL_MAX_LIST_NESTING */
+       725, /* GL_LIST_BASE */
+       727, /* GL_LIST_INDEX */
+      1189, /* GL_POLYGON_MODE */
+      1196, /* GL_POLYGON_SMOOTH */
+      1198, /* GL_POLYGON_STIPPLE */
+       458, /* GL_EDGE_FLAG */
        304, /* GL_CULL_FACE */
        305, /* GL_CULL_FACE_MODE */
-       581, /* GL_FRONT_FACE */
-       687, /* GL_LIGHTING */
-       692, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-       693, /* GL_LIGHT_MODEL_TWO_SIDE */
-       689, /* GL_LIGHT_MODEL_AMBIENT */
-      1436, /* GL_SHADE_MODEL */
+       588, /* GL_FRONT_FACE */
+       694, /* GL_LIGHTING */
+       699, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+       700, /* GL_LIGHT_MODEL_TWO_SIDE */
+       696, /* GL_LIGHT_MODEL_AMBIENT */
+      1449, /* GL_SHADE_MODEL */
        193, /* GL_COLOR_MATERIAL_FACE */
        194, /* GL_COLOR_MATERIAL_PARAMETER */
        192, /* GL_COLOR_MATERIAL */
-       505, /* GL_FOG */
-       527, /* GL_FOG_INDEX */
-       523, /* GL_FOG_DENSITY */
-       531, /* GL_FOG_START */
-       525, /* GL_FOG_END */
-       528, /* GL_FOG_MODE */
-       507, /* GL_FOG_COLOR */
-       369, /* GL_DEPTH_RANGE */
-       376, /* GL_DEPTH_TEST */
-       379, /* GL_DEPTH_WRITEMASK */
-       357, /* GL_DEPTH_CLEAR_VALUE */
-       368, /* GL_DEPTH_FUNC */
+       508, /* GL_FOG */
+       530, /* GL_FOG_INDEX */
+       526, /* GL_FOG_DENSITY */
+       534, /* GL_FOG_START */
+       528, /* GL_FOG_END */
+       531, /* GL_FOG_MODE */
+       510, /* GL_FOG_COLOR */
+       370, /* GL_DEPTH_RANGE */
+       378, /* GL_DEPTH_TEST */
+       381, /* GL_DEPTH_WRITEMASK */
+       358, /* GL_DEPTH_CLEAR_VALUE */
+       369, /* GL_DEPTH_FUNC */
         12, /* GL_ACCUM_CLEAR_VALUE */
-      1530, /* GL_STENCIL_TEST */
-      1518, /* GL_STENCIL_CLEAR_VALUE */
-      1520, /* GL_STENCIL_FUNC */
-      1532, /* GL_STENCIL_VALUE_MASK */
-      1519, /* GL_STENCIL_FAIL */
-      1527, /* GL_STENCIL_PASS_DEPTH_FAIL */
-      1528, /* GL_STENCIL_PASS_DEPTH_PASS */
-      1529, /* GL_STENCIL_REF */
-      1533, /* GL_STENCIL_WRITEMASK */
-       855, /* GL_MATRIX_MODE */
-      1031, /* GL_NORMALIZE */
-      1861, /* GL_VIEWPORT */
-      1005, /* GL_MODELVIEW_STACK_DEPTH */
-      1270, /* GL_PROJECTION_STACK_DEPTH */
-      1740, /* GL_TEXTURE_STACK_DEPTH */
-      1003, /* GL_MODELVIEW_MATRIX */
-      1269, /* GL_PROJECTION_MATRIX */
-      1723, /* GL_TEXTURE_MATRIX */
+      1547, /* GL_STENCIL_TEST */
+      1531, /* GL_STENCIL_CLEAR_VALUE */
+      1533, /* GL_STENCIL_FUNC */
+      1549, /* GL_STENCIL_VALUE_MASK */
+      1532, /* GL_STENCIL_FAIL */
+      1544, /* GL_STENCIL_PASS_DEPTH_FAIL */
+      1545, /* GL_STENCIL_PASS_DEPTH_PASS */
+      1546, /* GL_STENCIL_REF */
+      1550, /* GL_STENCIL_WRITEMASK */
+       862, /* GL_MATRIX_MODE */
+      1041, /* GL_NORMALIZE */
+      1880, /* GL_VIEWPORT */
+      1015, /* GL_MODELVIEW_STACK_DEPTH */
+      1280, /* GL_PROJECTION_STACK_DEPTH */
+      1757, /* GL_TEXTURE_STACK_DEPTH */
+      1013, /* GL_MODELVIEW_MATRIX */
+      1279, /* GL_PROJECTION_MATRIX */
+      1740, /* GL_TEXTURE_MATRIX */
         61, /* GL_ATTRIB_STACK_DEPTH */
        136, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
         43, /* GL_ALPHA_TEST */
         44, /* GL_ALPHA_TEST_FUNC */
         45, /* GL_ALPHA_TEST_REF */
-       381, /* GL_DITHER */
+       383, /* GL_DITHER */
         78, /* GL_BLEND_DST */
         87, /* GL_BLEND_SRC */
         75, /* GL_BLEND */
-       724, /* GL_LOGIC_OP_MODE */
-       639, /* GL_INDEX_LOGIC_OP */
+       731, /* GL_LOGIC_OP_MODE */
+       646, /* GL_INDEX_LOGIC_OP */
        191, /* GL_COLOR_LOGIC_OP */
         67, /* GL_AUX_BUFFERS */
-       392, /* GL_DRAW_BUFFER */
-      1306, /* GL_READ_BUFFER */
-      1417, /* GL_SCISSOR_BOX */
-      1418, /* GL_SCISSOR_TEST */
-       638, /* GL_INDEX_CLEAR_VALUE */
-       643, /* GL_INDEX_WRITEMASK */
+       394, /* GL_DRAW_BUFFER */
+      1316, /* GL_READ_BUFFER */
+      1430, /* GL_SCISSOR_BOX */
+      1431, /* GL_SCISSOR_TEST */
+       645, /* GL_INDEX_CLEAR_VALUE */
+       650, /* GL_INDEX_WRITEMASK */
        188, /* GL_COLOR_CLEAR_VALUE */
        230, /* GL_COLOR_WRITEMASK */
-       640, /* GL_INDEX_MODE */
-      1383, /* GL_RGBA_MODE */
-       391, /* GL_DOUBLEBUFFER */
-      1534, /* GL_STEREO */
-      1341, /* GL_RENDER_MODE */
-      1117, /* GL_PERSPECTIVE_CORRECTION_HINT */
-      1170, /* GL_POINT_SMOOTH_HINT */
-       707, /* GL_LINE_SMOOTH_HINT */
-      1187, /* GL_POLYGON_SMOOTH_HINT */
-       526, /* GL_FOG_HINT */
-      1704, /* GL_TEXTURE_GEN_S */
-      1705, /* GL_TEXTURE_GEN_T */
-      1703, /* GL_TEXTURE_GEN_R */
-      1702, /* GL_TEXTURE_GEN_Q */
-      1130, /* GL_PIXEL_MAP_I_TO_I */
-      1136, /* GL_PIXEL_MAP_S_TO_S */
-      1132, /* GL_PIXEL_MAP_I_TO_R */
-      1128, /* GL_PIXEL_MAP_I_TO_G */
-      1126, /* GL_PIXEL_MAP_I_TO_B */
-      1124, /* GL_PIXEL_MAP_I_TO_A */
-      1134, /* GL_PIXEL_MAP_R_TO_R */
-      1122, /* GL_PIXEL_MAP_G_TO_G */
-      1120, /* GL_PIXEL_MAP_B_TO_B */
-      1118, /* GL_PIXEL_MAP_A_TO_A */
-      1131, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-      1137, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-      1133, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-      1129, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-      1127, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-      1125, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-      1135, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-      1123, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-      1121, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-      1119, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-      1779, /* GL_UNPACK_SWAP_BYTES */
-      1774, /* GL_UNPACK_LSB_FIRST */
-      1775, /* GL_UNPACK_ROW_LENGTH */
-      1778, /* GL_UNPACK_SKIP_ROWS */
-      1777, /* GL_UNPACK_SKIP_PIXELS */
-      1772, /* GL_UNPACK_ALIGNMENT */
-      1105, /* GL_PACK_SWAP_BYTES */
-      1100, /* GL_PACK_LSB_FIRST */
-      1101, /* GL_PACK_ROW_LENGTH */
-      1104, /* GL_PACK_SKIP_ROWS */
-      1103, /* GL_PACK_SKIP_PIXELS */
-      1097, /* GL_PACK_ALIGNMENT */
-       802, /* GL_MAP_COLOR */
-       807, /* GL_MAP_STENCIL */
-       642, /* GL_INDEX_SHIFT */
-       641, /* GL_INDEX_OFFSET */
-      1319, /* GL_RED_SCALE */
-      1317, /* GL_RED_BIAS */
-      1879, /* GL_ZOOM_X */
-      1880, /* GL_ZOOM_Y */
-       603, /* GL_GREEN_SCALE */
-       601, /* GL_GREEN_BIAS */
+       647, /* GL_INDEX_MODE */
+      1396, /* GL_RGBA_MODE */
+       393, /* GL_DOUBLEBUFFER */
+      1551, /* GL_STEREO */
+      1354, /* GL_RENDER_MODE */
+      1127, /* GL_PERSPECTIVE_CORRECTION_HINT */
+      1180, /* GL_POINT_SMOOTH_HINT */
+       714, /* GL_LINE_SMOOTH_HINT */
+      1197, /* GL_POLYGON_SMOOTH_HINT */
+       529, /* GL_FOG_HINT */
+      1721, /* GL_TEXTURE_GEN_S */
+      1722, /* GL_TEXTURE_GEN_T */
+      1720, /* GL_TEXTURE_GEN_R */
+      1719, /* GL_TEXTURE_GEN_Q */
+      1140, /* GL_PIXEL_MAP_I_TO_I */
+      1146, /* GL_PIXEL_MAP_S_TO_S */
+      1142, /* GL_PIXEL_MAP_I_TO_R */
+      1138, /* GL_PIXEL_MAP_I_TO_G */
+      1136, /* GL_PIXEL_MAP_I_TO_B */
+      1134, /* GL_PIXEL_MAP_I_TO_A */
+      1144, /* GL_PIXEL_MAP_R_TO_R */
+      1132, /* GL_PIXEL_MAP_G_TO_G */
+      1130, /* GL_PIXEL_MAP_B_TO_B */
+      1128, /* GL_PIXEL_MAP_A_TO_A */
+      1141, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+      1147, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+      1143, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+      1139, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+      1137, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+      1135, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+      1145, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+      1133, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+      1131, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+      1129, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+      1797, /* GL_UNPACK_SWAP_BYTES */
+      1792, /* GL_UNPACK_LSB_FIRST */
+      1793, /* GL_UNPACK_ROW_LENGTH */
+      1796, /* GL_UNPACK_SKIP_ROWS */
+      1795, /* GL_UNPACK_SKIP_PIXELS */
+      1790, /* GL_UNPACK_ALIGNMENT */
+      1115, /* GL_PACK_SWAP_BYTES */
+      1110, /* GL_PACK_LSB_FIRST */
+      1111, /* GL_PACK_ROW_LENGTH */
+      1114, /* GL_PACK_SKIP_ROWS */
+      1113, /* GL_PACK_SKIP_PIXELS */
+      1107, /* GL_PACK_ALIGNMENT */
+       809, /* GL_MAP_COLOR */
+       814, /* GL_MAP_STENCIL */
+       649, /* GL_INDEX_SHIFT */
+       648, /* GL_INDEX_OFFSET */
+      1330, /* GL_RED_SCALE */
+      1328, /* GL_RED_BIAS */
+      1898, /* GL_ZOOM_X */
+      1899, /* GL_ZOOM_Y */
+       610, /* GL_GREEN_SCALE */
+       608, /* GL_GREEN_BIAS */
         93, /* GL_BLUE_SCALE */
         91, /* GL_BLUE_BIAS */
         42, /* GL_ALPHA_SCALE */
         40, /* GL_ALPHA_BIAS */
-       370, /* GL_DEPTH_SCALE */
-       350, /* GL_DEPTH_BIAS */
-       881, /* GL_MAX_EVAL_ORDER */
-       890, /* GL_MAX_LIGHTS */
-       864, /* GL_MAX_CLIP_PLANES */
-       936, /* GL_MAX_TEXTURE_SIZE */
-       896, /* GL_MAX_PIXEL_MAP_TABLE */
-       860, /* GL_MAX_ATTRIB_STACK_DEPTH */
-       893, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-       894, /* GL_MAX_NAME_STACK_DEPTH */
-       922, /* GL_MAX_PROJECTION_STACK_DEPTH */
-       937, /* GL_MAX_TEXTURE_STACK_DEPTH */
-       953, /* GL_MAX_VIEWPORT_DIMS */
-       861, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-      1544, /* GL_SUBPIXEL_BITS */
-       637, /* GL_INDEX_BITS */
-      1318, /* GL_RED_BITS */
-       602, /* GL_GREEN_BITS */
+       371, /* GL_DEPTH_SCALE */
+       351, /* GL_DEPTH_BIAS */
+       889, /* GL_MAX_EVAL_ORDER */
+       898, /* GL_MAX_LIGHTS */
+       871, /* GL_MAX_CLIP_PLANES */
+       946, /* GL_MAX_TEXTURE_SIZE */
+       904, /* GL_MAX_PIXEL_MAP_TABLE */
+       867, /* GL_MAX_ATTRIB_STACK_DEPTH */
+       901, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+       902, /* GL_MAX_NAME_STACK_DEPTH */
+       930, /* GL_MAX_PROJECTION_STACK_DEPTH */
+       947, /* GL_MAX_TEXTURE_STACK_DEPTH */
+       963, /* GL_MAX_VIEWPORT_DIMS */
+       868, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+      1561, /* GL_SUBPIXEL_BITS */
+       644, /* GL_INDEX_BITS */
+      1329, /* GL_RED_BITS */
+       609, /* GL_GREEN_BITS */
         92, /* GL_BLUE_BITS */
         41, /* GL_ALPHA_BITS */
-       351, /* GL_DEPTH_BITS */
-      1516, /* GL_STENCIL_BITS */
+       352, /* GL_DEPTH_BITS */
+      1529, /* GL_STENCIL_BITS */
         14, /* GL_ACCUM_RED_BITS */
         13, /* GL_ACCUM_GREEN_BITS */
         10, /* GL_ACCUM_BLUE_BITS */
          9, /* GL_ACCUM_ALPHA_BITS */
-      1019, /* GL_NAME_STACK_DEPTH */
+      1029, /* GL_NAME_STACK_DEPTH */
         62, /* GL_AUTO_NORMAL */
-       748, /* GL_MAP1_COLOR_4 */
-       751, /* GL_MAP1_INDEX */
-       752, /* GL_MAP1_NORMAL */
-       753, /* GL_MAP1_TEXTURE_COORD_1 */
-       754, /* GL_MAP1_TEXTURE_COORD_2 */
-       755, /* GL_MAP1_TEXTURE_COORD_3 */
-       756, /* GL_MAP1_TEXTURE_COORD_4 */
-       757, /* GL_MAP1_VERTEX_3 */
-       758, /* GL_MAP1_VERTEX_4 */
-       775, /* GL_MAP2_COLOR_4 */
-       778, /* GL_MAP2_INDEX */
-       779, /* GL_MAP2_NORMAL */
-       780, /* GL_MAP2_TEXTURE_COORD_1 */
-       781, /* GL_MAP2_TEXTURE_COORD_2 */
-       782, /* GL_MAP2_TEXTURE_COORD_3 */
-       783, /* GL_MAP2_TEXTURE_COORD_4 */
-       784, /* GL_MAP2_VERTEX_3 */
-       785, /* GL_MAP2_VERTEX_4 */
-       749, /* GL_MAP1_GRID_DOMAIN */
-       750, /* GL_MAP1_GRID_SEGMENTS */
-       776, /* GL_MAP2_GRID_DOMAIN */
-       777, /* GL_MAP2_GRID_SEGMENTS */
-      1627, /* GL_TEXTURE_1D */
-      1629, /* GL_TEXTURE_2D */
-       479, /* GL_FEEDBACK_BUFFER_POINTER */
-       480, /* GL_FEEDBACK_BUFFER_SIZE */
-       481, /* GL_FEEDBACK_BUFFER_TYPE */
-      1427, /* GL_SELECTION_BUFFER_POINTER */
-      1428, /* GL_SELECTION_BUFFER_SIZE */
-      1745, /* GL_TEXTURE_WIDTH */
-      1709, /* GL_TEXTURE_HEIGHT */
-      1664, /* GL_TEXTURE_COMPONENTS */
-      1648, /* GL_TEXTURE_BORDER_COLOR */
-      1647, /* GL_TEXTURE_BORDER */
-       383, /* GL_DONT_CARE */
-       477, /* GL_FASTEST */
-      1027, /* GL_NICEST */
+       755, /* GL_MAP1_COLOR_4 */
+       758, /* GL_MAP1_INDEX */
+       759, /* GL_MAP1_NORMAL */
+       760, /* GL_MAP1_TEXTURE_COORD_1 */
+       761, /* GL_MAP1_TEXTURE_COORD_2 */
+       762, /* GL_MAP1_TEXTURE_COORD_3 */
+       763, /* GL_MAP1_TEXTURE_COORD_4 */
+       764, /* GL_MAP1_VERTEX_3 */
+       765, /* GL_MAP1_VERTEX_4 */
+       782, /* GL_MAP2_COLOR_4 */
+       785, /* GL_MAP2_INDEX */
+       786, /* GL_MAP2_NORMAL */
+       787, /* GL_MAP2_TEXTURE_COORD_1 */
+       788, /* GL_MAP2_TEXTURE_COORD_2 */
+       789, /* GL_MAP2_TEXTURE_COORD_3 */
+       790, /* GL_MAP2_TEXTURE_COORD_4 */
+       791, /* GL_MAP2_VERTEX_3 */
+       792, /* GL_MAP2_VERTEX_4 */
+       756, /* GL_MAP1_GRID_DOMAIN */
+       757, /* GL_MAP1_GRID_SEGMENTS */
+       783, /* GL_MAP2_GRID_DOMAIN */
+       784, /* GL_MAP2_GRID_SEGMENTS */
+      1644, /* GL_TEXTURE_1D */
+      1646, /* GL_TEXTURE_2D */
+       482, /* GL_FEEDBACK_BUFFER_POINTER */
+       483, /* GL_FEEDBACK_BUFFER_SIZE */
+       484, /* GL_FEEDBACK_BUFFER_TYPE */
+      1440, /* GL_SELECTION_BUFFER_POINTER */
+      1441, /* GL_SELECTION_BUFFER_SIZE */
+      1763, /* GL_TEXTURE_WIDTH */
+      1726, /* GL_TEXTURE_HEIGHT */
+      1681, /* GL_TEXTURE_COMPONENTS */
+      1665, /* GL_TEXTURE_BORDER_COLOR */
+      1664, /* GL_TEXTURE_BORDER */
+       385, /* GL_DONT_CARE */
+       480, /* GL_FASTEST */
+      1037, /* GL_NICEST */
         48, /* GL_AMBIENT */
-       380, /* GL_DIFFUSE */
-      1476, /* GL_SPECULAR */
-      1191, /* GL_POSITION */
-      1479, /* GL_SPOT_DIRECTION */
-      1480, /* GL_SPOT_EXPONENT */
-      1478, /* GL_SPOT_CUTOFF */
+       382, /* GL_DIFFUSE */
+      1489, /* GL_SPECULAR */
+      1201, /* GL_POSITION */
+      1492, /* GL_SPOT_DIRECTION */
+      1493, /* GL_SPOT_EXPONENT */
+      1491, /* GL_SPOT_CUTOFF */
        275, /* GL_CONSTANT_ATTENUATION */
-       696, /* GL_LINEAR_ATTENUATION */
-      1291, /* GL_QUADRATIC_ATTENUATION */
+       703, /* GL_LINEAR_ATTENUATION */
+      1301, /* GL_QUADRATIC_ATTENUATION */
        244, /* GL_COMPILE */
        245, /* GL_COMPILE_AND_EXECUTE */
        120, /* GL_BYTE */
-      1781, /* GL_UNSIGNED_BYTE */
-      1441, /* GL_SHORT */
-      1792, /* GL_UNSIGNED_SHORT */
-       645, /* GL_INT */
-      1784, /* GL_UNSIGNED_INT */
-       486, /* GL_FLOAT */
+      1799, /* GL_UNSIGNED_BYTE */
+      1454, /* GL_SHORT */
+      1811, /* GL_UNSIGNED_SHORT */
+       652, /* GL_INT */
+      1802, /* GL_UNSIGNED_INT */
+       489, /* GL_FLOAT */
          1, /* GL_2_BYTES */
          5, /* GL_3_BYTES */
          7, /* GL_4_BYTES */
-       390, /* GL_DOUBLE */
+       392, /* GL_DOUBLE */
        132, /* GL_CLEAR */
         50, /* GL_AND */
         52, /* GL_AND_REVERSE */
        299, /* GL_COPY */
         51, /* GL_AND_INVERTED */
-      1029, /* GL_NOOP */
-      1875, /* GL_XOR */
-      1092, /* GL_OR */
-      1030, /* GL_NOR */
-       467, /* GL_EQUIV */
-       672, /* GL_INVERT */
-      1095, /* GL_OR_REVERSE */
+      1039, /* GL_NOOP */
+      1894, /* GL_XOR */
+      1102, /* GL_OR */
+      1040, /* GL_NOR */
+       470, /* GL_EQUIV */
+       679, /* GL_INVERT */
+      1105, /* GL_OR_REVERSE */
        300, /* GL_COPY_INVERTED */
-      1094, /* GL_OR_INVERTED */
-      1020, /* GL_NAND */
-      1432, /* GL_SET */
-       464, /* GL_EMISSION */
-      1440, /* GL_SHININESS */
+      1104, /* GL_OR_INVERTED */
+      1030, /* GL_NAND */
+      1445, /* GL_SET */
+       467, /* GL_EMISSION */
+      1453, /* GL_SHININESS */
         49, /* GL_AMBIENT_AND_DIFFUSE */
        190, /* GL_COLOR_INDEXES */
-       970, /* GL_MODELVIEW */
-      1268, /* GL_PROJECTION */
-      1562, /* GL_TEXTURE */
+       980, /* GL_MODELVIEW */
+      1278, /* GL_PROJECTION */
+      1579, /* GL_TEXTURE */
        147, /* GL_COLOR */
        346, /* GL_DEPTH */
-      1502, /* GL_STENCIL */
+      1515, /* GL_STENCIL */
        189, /* GL_COLOR_INDEX */
-      1521, /* GL_STENCIL_INDEX */
-       358, /* GL_DEPTH_COMPONENT */
-      1314, /* GL_RED */
-       600, /* GL_GREEN */
+      1534, /* GL_STENCIL_INDEX */
+       359, /* GL_DEPTH_COMPONENT */
+      1325, /* GL_RED */
+       607, /* GL_GREEN */
         90, /* GL_BLUE */
         31, /* GL_ALPHA */
-      1349, /* GL_RGB */
-      1368, /* GL_RGBA */
-       726, /* GL_LUMINANCE */
-       747, /* GL_LUMINANCE_ALPHA */
+      1362, /* GL_RGB */
+      1381, /* GL_RGBA */
+       733, /* GL_LUMINANCE */
+       754, /* GL_LUMINANCE_ALPHA */
         73, /* GL_BITMAP */
-      1147, /* GL_POINT */
-       694, /* GL_LINE */
-       482, /* GL_FILL */
-      1323, /* GL_RENDER */
-       478, /* GL_FEEDBACK */
-      1426, /* GL_SELECT */
-       485, /* GL_FLAT */
-      1451, /* GL_SMOOTH */
-       673, /* GL_KEEP */
-      1343, /* GL_REPLACE */
-       627, /* GL_INCR */
+      1157, /* GL_POINT */
+       701, /* GL_LINE */
+       485, /* GL_FILL */
+      1334, /* GL_RENDER */
+       481, /* GL_FEEDBACK */
+      1439, /* GL_SELECT */
+       488, /* GL_FLAT */
+      1464, /* GL_SMOOTH */
+       680, /* GL_KEEP */
+      1356, /* GL_REPLACE */
+       634, /* GL_INCR */
        342, /* GL_DECR */
-      1807, /* GL_VENDOR */
-      1340, /* GL_RENDERER */
-      1808, /* GL_VERSION */
-       471, /* GL_EXTENSIONS */
-      1391, /* GL_S */
-      1553, /* GL_T */
-      1303, /* GL_R */
-      1290, /* GL_Q */
-      1006, /* GL_MODULATE */
+      1826, /* GL_VENDOR */
+      1353, /* GL_RENDERER */
+      1827, /* GL_VERSION */
+       474, /* GL_EXTENSIONS */
+      1404, /* GL_S */
+      1570, /* GL_T */
+      1313, /* GL_R */
+      1300, /* GL_Q */
+      1016, /* GL_MODULATE */
        341, /* GL_DECAL */
-      1699, /* GL_TEXTURE_ENV_MODE */
-      1698, /* GL_TEXTURE_ENV_COLOR */
-      1697, /* GL_TEXTURE_ENV */
-       472, /* GL_EYE_LINEAR */
-      1053, /* GL_OBJECT_LINEAR */
-      1477, /* GL_SPHERE_MAP */
-      1701, /* GL_TEXTURE_GEN_MODE */
-      1055, /* GL_OBJECT_PLANE */
-       473, /* GL_EYE_PLANE */
-      1021, /* GL_NEAREST */
-       695, /* GL_LINEAR */
-      1025, /* GL_NEAREST_MIPMAP_NEAREST */
-       700, /* GL_LINEAR_MIPMAP_NEAREST */
-      1024, /* GL_NEAREST_MIPMAP_LINEAR */
-       699, /* GL_LINEAR_MIPMAP_LINEAR */
-      1722, /* GL_TEXTURE_MAG_FILTER */
-      1730, /* GL_TEXTURE_MIN_FILTER */
-      1747, /* GL_TEXTURE_WRAP_S */
-      1748, /* GL_TEXTURE_WRAP_T */
+      1716, /* GL_TEXTURE_ENV_MODE */
+      1715, /* GL_TEXTURE_ENV_COLOR */
+      1714, /* GL_TEXTURE_ENV */
+       475, /* GL_EYE_LINEAR */
+      1063, /* GL_OBJECT_LINEAR */
+      1490, /* GL_SPHERE_MAP */
+      1718, /* GL_TEXTURE_GEN_MODE */
+      1065, /* GL_OBJECT_PLANE */
+       476, /* GL_EYE_PLANE */
+      1031, /* GL_NEAREST */
+       702, /* GL_LINEAR */
+      1035, /* GL_NEAREST_MIPMAP_NEAREST */
+       707, /* GL_LINEAR_MIPMAP_NEAREST */
+      1034, /* GL_NEAREST_MIPMAP_LINEAR */
+       706, /* GL_LINEAR_MIPMAP_LINEAR */
+      1739, /* GL_TEXTURE_MAG_FILTER */
+      1747, /* GL_TEXTURE_MIN_FILTER */
+      1765, /* GL_TEXTURE_WRAP_S */
+      1766, /* GL_TEXTURE_WRAP_T */
        126, /* GL_CLAMP */
-      1342, /* GL_REPEAT */
-      1185, /* GL_POLYGON_OFFSET_UNITS */
-      1184, /* GL_POLYGON_OFFSET_POINT */
-      1183, /* GL_POLYGON_OFFSET_LINE */
-      1304, /* GL_R3_G3_B2 */
-      1804, /* GL_V2F */
-      1805, /* GL_V3F */
+      1355, /* GL_REPEAT */
+      1195, /* GL_POLYGON_OFFSET_UNITS */
+      1194, /* GL_POLYGON_OFFSET_POINT */
+      1193, /* GL_POLYGON_OFFSET_LINE */
+      1314, /* GL_R3_G3_B2 */
+      1823, /* GL_V2F */
+      1824, /* GL_V3F */
        123, /* GL_C4UB_V2F */
        124, /* GL_C4UB_V3F */
        121, /* GL_C3F_V3F */
-      1018, /* GL_N3F_V3F */
+      1028, /* GL_N3F_V3F */
        122, /* GL_C4F_N3F_V3F */
-      1558, /* GL_T2F_V3F */
-      1560, /* GL_T4F_V4F */
-      1556, /* GL_T2F_C4UB_V3F */
-      1554, /* GL_T2F_C3F_V3F */
-      1557, /* GL_T2F_N3F_V3F */
-      1555, /* GL_T2F_C4F_N3F_V3F */
-      1559, /* GL_T4F_C4F_N3F_V4F */
+      1575, /* GL_T2F_V3F */
+      1577, /* GL_T4F_V4F */
+      1573, /* GL_T2F_C4UB_V3F */
+      1571, /* GL_T2F_C3F_V3F */
+      1574, /* GL_T2F_N3F_V3F */
+      1572, /* GL_T2F_C4F_N3F_V3F */
+      1576, /* GL_T4F_C4F_N3F_V4F */
        139, /* GL_CLIP_PLANE0 */
        140, /* GL_CLIP_PLANE1 */
        141, /* GL_CLIP_PLANE2 */
        142, /* GL_CLIP_PLANE3 */
        143, /* GL_CLIP_PLANE4 */
        144, /* GL_CLIP_PLANE5 */
-       679, /* GL_LIGHT0 */
-       680, /* GL_LIGHT1 */
-       681, /* GL_LIGHT2 */
-       682, /* GL_LIGHT3 */
-       683, /* GL_LIGHT4 */
-       684, /* GL_LIGHT5 */
-       685, /* GL_LIGHT6 */
-       686, /* GL_LIGHT7 */
-       604, /* GL_HINT_BIT */
+       686, /* GL_LIGHT0 */
+       687, /* GL_LIGHT1 */
+       688, /* GL_LIGHT2 */
+       689, /* GL_LIGHT3 */
+       690, /* GL_LIGHT4 */
+       691, /* GL_LIGHT5 */
+       692, /* GL_LIGHT6 */
+       693, /* GL_LIGHT7 */
+       611, /* GL_HINT_BIT */
        277, /* GL_CONSTANT_COLOR */
-      1066, /* GL_ONE_MINUS_CONSTANT_COLOR */
+      1076, /* GL_ONE_MINUS_CONSTANT_COLOR */
        272, /* GL_CONSTANT_ALPHA */
-      1064, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+      1074, /* GL_ONE_MINUS_CONSTANT_ALPHA */
         76, /* GL_BLEND_COLOR */
-       584, /* GL_FUNC_ADD */
-       954, /* GL_MIN */
-       857, /* GL_MAX */
+       591, /* GL_FUNC_ADD */
+       964, /* GL_MIN */
+       864, /* GL_MAX */
         81, /* GL_BLEND_EQUATION */
-       588, /* GL_FUNC_SUBTRACT */
-       586, /* GL_FUNC_REVERSE_SUBTRACT */
+       595, /* GL_FUNC_SUBTRACT */
+       593, /* GL_FUNC_REVERSE_SUBTRACT */
        280, /* GL_CONVOLUTION_1D */
        281, /* GL_CONVOLUTION_2D */
-      1429, /* GL_SEPARABLE_2D */
+      1442, /* GL_SEPARABLE_2D */
        284, /* GL_CONVOLUTION_BORDER_MODE */
        288, /* GL_CONVOLUTION_FILTER_SCALE */
        286, /* GL_CONVOLUTION_FILTER_BIAS */
-      1315, /* GL_REDUCE */
+      1326, /* GL_REDUCE */
        290, /* GL_CONVOLUTION_FORMAT */
        294, /* GL_CONVOLUTION_WIDTH */
        292, /* GL_CONVOLUTION_HEIGHT */
-       872, /* GL_MAX_CONVOLUTION_WIDTH */
-       870, /* GL_MAX_CONVOLUTION_HEIGHT */
-      1224, /* GL_POST_CONVOLUTION_RED_SCALE */
-      1220, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-      1215, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-      1211, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-      1222, /* GL_POST_CONVOLUTION_RED_BIAS */
-      1218, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-      1213, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-      1209, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-       605, /* GL_HISTOGRAM */
-      1274, /* GL_PROXY_HISTOGRAM */
-       621, /* GL_HISTOGRAM_WIDTH */
-       611, /* GL_HISTOGRAM_FORMAT */
-       617, /* GL_HISTOGRAM_RED_SIZE */
-       613, /* GL_HISTOGRAM_GREEN_SIZE */
-       608, /* GL_HISTOGRAM_BLUE_SIZE */
-       606, /* GL_HISTOGRAM_ALPHA_SIZE */
-       615, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-       619, /* GL_HISTOGRAM_SINK */
-       955, /* GL_MINMAX */
-       957, /* GL_MINMAX_FORMAT */
-       959, /* GL_MINMAX_SINK */
-      1561, /* GL_TABLE_TOO_LARGE_EXT */
-      1783, /* GL_UNSIGNED_BYTE_3_3_2 */
-      1794, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-      1796, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-      1789, /* GL_UNSIGNED_INT_8_8_8_8 */
-      1785, /* GL_UNSIGNED_INT_10_10_10_2 */
-      1182, /* GL_POLYGON_OFFSET_FILL */
-      1181, /* GL_POLYGON_OFFSET_FACTOR */
-      1180, /* GL_POLYGON_OFFSET_BIAS */
-      1346, /* GL_RESCALE_NORMAL */
+       880, /* GL_MAX_CONVOLUTION_WIDTH */
+       878, /* GL_MAX_CONVOLUTION_HEIGHT */
+      1234, /* GL_POST_CONVOLUTION_RED_SCALE */
+      1230, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+      1225, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+      1221, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+      1232, /* GL_POST_CONVOLUTION_RED_BIAS */
+      1228, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+      1223, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+      1219, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+       612, /* GL_HISTOGRAM */
+      1284, /* GL_PROXY_HISTOGRAM */
+       628, /* GL_HISTOGRAM_WIDTH */
+       618, /* GL_HISTOGRAM_FORMAT */
+       624, /* GL_HISTOGRAM_RED_SIZE */
+       620, /* GL_HISTOGRAM_GREEN_SIZE */
+       615, /* GL_HISTOGRAM_BLUE_SIZE */
+       613, /* GL_HISTOGRAM_ALPHA_SIZE */
+       622, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+       626, /* GL_HISTOGRAM_SINK */
+       965, /* GL_MINMAX */
+       967, /* GL_MINMAX_FORMAT */
+       969, /* GL_MINMAX_SINK */
+      1578, /* GL_TABLE_TOO_LARGE_EXT */
+      1801, /* GL_UNSIGNED_BYTE_3_3_2 */
+      1813, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+      1815, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+      1808, /* GL_UNSIGNED_INT_8_8_8_8 */
+      1803, /* GL_UNSIGNED_INT_10_10_10_2 */
+      1192, /* GL_POLYGON_OFFSET_FILL */
+      1191, /* GL_POLYGON_OFFSET_FACTOR */
+      1190, /* GL_POLYGON_OFFSET_BIAS */
+      1359, /* GL_RESCALE_NORMAL */
         36, /* GL_ALPHA4 */
         38, /* GL_ALPHA8 */
         32, /* GL_ALPHA12 */
         34, /* GL_ALPHA16 */
-       737, /* GL_LUMINANCE4 */
-       743, /* GL_LUMINANCE8 */
-       727, /* GL_LUMINANCE12 */
-       733, /* GL_LUMINANCE16 */
-       738, /* GL_LUMINANCE4_ALPHA4 */
-       741, /* GL_LUMINANCE6_ALPHA2 */
-       744, /* GL_LUMINANCE8_ALPHA8 */
-       730, /* GL_LUMINANCE12_ALPHA4 */
-       728, /* GL_LUMINANCE12_ALPHA12 */
-       734, /* GL_LUMINANCE16_ALPHA16 */
-       646, /* GL_INTENSITY */
-       651, /* GL_INTENSITY4 */
-       653, /* GL_INTENSITY8 */
-       647, /* GL_INTENSITY12 */
-       649, /* GL_INTENSITY16 */
-      1358, /* GL_RGB2_EXT */
-      1359, /* GL_RGB4 */
-      1362, /* GL_RGB5 */
-      1366, /* GL_RGB8 */
-      1350, /* GL_RGB10 */
-      1354, /* GL_RGB12 */
-      1356, /* GL_RGB16 */
-      1373, /* GL_RGBA2 */
-      1375, /* GL_RGBA4 */
-      1363, /* GL_RGB5_A1 */
-      1379, /* GL_RGBA8 */
-      1351, /* GL_RGB10_A2 */
-      1369, /* GL_RGBA12 */
-      1371, /* GL_RGBA16 */
-      1737, /* GL_TEXTURE_RED_SIZE */
-      1707, /* GL_TEXTURE_GREEN_SIZE */
-      1645, /* GL_TEXTURE_BLUE_SIZE */
-      1632, /* GL_TEXTURE_ALPHA_SIZE */
-      1720, /* GL_TEXTURE_LUMINANCE_SIZE */
-      1711, /* GL_TEXTURE_INTENSITY_SIZE */
-      1344, /* GL_REPLACE_EXT */
-      1278, /* GL_PROXY_TEXTURE_1D */
-      1281, /* GL_PROXY_TEXTURE_2D */
-      1743, /* GL_TEXTURE_TOO_LARGE_EXT */
-      1732, /* GL_TEXTURE_PRIORITY */
-      1739, /* GL_TEXTURE_RESIDENT */
-      1635, /* GL_TEXTURE_BINDING_1D */
-      1637, /* GL_TEXTURE_BINDING_2D */
-      1639, /* GL_TEXTURE_BINDING_3D */
-      1102, /* GL_PACK_SKIP_IMAGES */
-      1098, /* GL_PACK_IMAGE_HEIGHT */
-      1776, /* GL_UNPACK_SKIP_IMAGES */
-      1773, /* GL_UNPACK_IMAGE_HEIGHT */
-      1631, /* GL_TEXTURE_3D */
-      1284, /* GL_PROXY_TEXTURE_3D */
-      1694, /* GL_TEXTURE_DEPTH */
-      1746, /* GL_TEXTURE_WRAP_R */
-       858, /* GL_MAX_3D_TEXTURE_SIZE */
-      1809, /* GL_VERTEX_ARRAY */
-      1032, /* GL_NORMAL_ARRAY */
+       744, /* GL_LUMINANCE4 */
+       750, /* GL_LUMINANCE8 */
+       734, /* GL_LUMINANCE12 */
+       740, /* GL_LUMINANCE16 */
+       745, /* GL_LUMINANCE4_ALPHA4 */
+       748, /* GL_LUMINANCE6_ALPHA2 */
+       751, /* GL_LUMINANCE8_ALPHA8 */
+       737, /* GL_LUMINANCE12_ALPHA4 */
+       735, /* GL_LUMINANCE12_ALPHA12 */
+       741, /* GL_LUMINANCE16_ALPHA16 */
+       653, /* GL_INTENSITY */
+       658, /* GL_INTENSITY4 */
+       660, /* GL_INTENSITY8 */
+       654, /* GL_INTENSITY12 */
+       656, /* GL_INTENSITY16 */
+      1371, /* GL_RGB2_EXT */
+      1372, /* GL_RGB4 */
+      1375, /* GL_RGB5 */
+      1379, /* GL_RGB8 */
+      1363, /* GL_RGB10 */
+      1367, /* GL_RGB12 */
+      1369, /* GL_RGB16 */
+      1386, /* GL_RGBA2 */
+      1388, /* GL_RGBA4 */
+      1376, /* GL_RGB5_A1 */
+      1392, /* GL_RGBA8 */
+      1364, /* GL_RGB10_A2 */
+      1382, /* GL_RGBA12 */
+      1384, /* GL_RGBA16 */
+      1754, /* GL_TEXTURE_RED_SIZE */
+      1724, /* GL_TEXTURE_GREEN_SIZE */
+      1662, /* GL_TEXTURE_BLUE_SIZE */
+      1649, /* GL_TEXTURE_ALPHA_SIZE */
+      1737, /* GL_TEXTURE_LUMINANCE_SIZE */
+      1728, /* GL_TEXTURE_INTENSITY_SIZE */
+      1357, /* GL_REPLACE_EXT */
+      1288, /* GL_PROXY_TEXTURE_1D */
+      1291, /* GL_PROXY_TEXTURE_2D */
+      1761, /* GL_TEXTURE_TOO_LARGE_EXT */
+      1749, /* GL_TEXTURE_PRIORITY */
+      1756, /* GL_TEXTURE_RESIDENT */
+      1652, /* GL_TEXTURE_BINDING_1D */
+      1654, /* GL_TEXTURE_BINDING_2D */
+      1656, /* GL_TEXTURE_BINDING_3D */
+      1112, /* GL_PACK_SKIP_IMAGES */
+      1108, /* GL_PACK_IMAGE_HEIGHT */
+      1794, /* GL_UNPACK_SKIP_IMAGES */
+      1791, /* GL_UNPACK_IMAGE_HEIGHT */
+      1648, /* GL_TEXTURE_3D */
+      1294, /* GL_PROXY_TEXTURE_3D */
+      1711, /* GL_TEXTURE_DEPTH */
+      1764, /* GL_TEXTURE_WRAP_R */
+       865, /* GL_MAX_3D_TEXTURE_SIZE */
+      1828, /* GL_VERTEX_ARRAY */
+      1042, /* GL_NORMAL_ARRAY */
        148, /* GL_COLOR_ARRAY */
-       631, /* GL_INDEX_ARRAY */
-      1672, /* GL_TEXTURE_COORD_ARRAY */
-       456, /* GL_EDGE_FLAG_ARRAY */
-      1815, /* GL_VERTEX_ARRAY_SIZE */
-      1817, /* GL_VERTEX_ARRAY_TYPE */
-      1816, /* GL_VERTEX_ARRAY_STRIDE */
-      1037, /* GL_NORMAL_ARRAY_TYPE */
-      1036, /* GL_NORMAL_ARRAY_STRIDE */
+       638, /* GL_INDEX_ARRAY */
+      1689, /* GL_TEXTURE_COORD_ARRAY */
+       459, /* GL_EDGE_FLAG_ARRAY */
+      1834, /* GL_VERTEX_ARRAY_SIZE */
+      1836, /* GL_VERTEX_ARRAY_TYPE */
+      1835, /* GL_VERTEX_ARRAY_STRIDE */
+      1047, /* GL_NORMAL_ARRAY_TYPE */
+      1046, /* GL_NORMAL_ARRAY_STRIDE */
        152, /* GL_COLOR_ARRAY_SIZE */
        154, /* GL_COLOR_ARRAY_TYPE */
        153, /* GL_COLOR_ARRAY_STRIDE */
-       636, /* GL_INDEX_ARRAY_TYPE */
-       635, /* GL_INDEX_ARRAY_STRIDE */
-      1676, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-      1678, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-      1677, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-       460, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-      1814, /* GL_VERTEX_ARRAY_POINTER */
-      1035, /* GL_NORMAL_ARRAY_POINTER */
+       643, /* GL_INDEX_ARRAY_TYPE */
+       642, /* GL_INDEX_ARRAY_STRIDE */
+      1693, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+      1695, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+      1694, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+       463, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+      1833, /* GL_VERTEX_ARRAY_POINTER */
+      1045, /* GL_NORMAL_ARRAY_POINTER */
        151, /* GL_COLOR_ARRAY_POINTER */
-       634, /* GL_INDEX_ARRAY_POINTER */
-      1675, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-       459, /* GL_EDGE_FLAG_ARRAY_POINTER */
-      1011, /* GL_MULTISAMPLE */
-      1403, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-      1405, /* GL_SAMPLE_ALPHA_TO_ONE */
-      1410, /* GL_SAMPLE_COVERAGE */
-      1407, /* GL_SAMPLE_BUFFERS */
-      1398, /* GL_SAMPLES */
-      1414, /* GL_SAMPLE_COVERAGE_VALUE */
-      1412, /* GL_SAMPLE_COVERAGE_INVERT */
+       641, /* GL_INDEX_ARRAY_POINTER */
+      1692, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+       462, /* GL_EDGE_FLAG_ARRAY_POINTER */
+      1021, /* GL_MULTISAMPLE */
+      1416, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+      1418, /* GL_SAMPLE_ALPHA_TO_ONE */
+      1423, /* GL_SAMPLE_COVERAGE */
+      1420, /* GL_SAMPLE_BUFFERS */
+      1411, /* GL_SAMPLES */
+      1427, /* GL_SAMPLE_COVERAGE_VALUE */
+      1425, /* GL_SAMPLE_COVERAGE_INVERT */
        195, /* GL_COLOR_MATRIX */
        197, /* GL_COLOR_MATRIX_STACK_DEPTH */
-       866, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-      1207, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-      1203, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-      1198, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-      1194, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-      1205, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-      1201, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-      1196, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-      1192, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-      1655, /* GL_TEXTURE_COLOR_TABLE_SGI */
-      1285, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-      1657, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+       874, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+      1217, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+      1213, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+      1208, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+      1204, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+      1215, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+      1211, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+      1206, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+      1202, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+      1672, /* GL_TEXTURE_COLOR_TABLE_SGI */
+      1295, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+      1674, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
         80, /* GL_BLEND_DST_RGB */
         89, /* GL_BLEND_SRC_RGB */
         79, /* GL_BLEND_DST_ALPHA */
         88, /* GL_BLEND_SRC_ALPHA */
        201, /* GL_COLOR_TABLE */
-      1217, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-      1200, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-      1273, /* GL_PROXY_COLOR_TABLE */
-      1277, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-      1276, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+      1227, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+      1210, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+      1283, /* GL_PROXY_COLOR_TABLE */
+      1287, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+      1286, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
        225, /* GL_COLOR_TABLE_SCALE */
        205, /* GL_COLOR_TABLE_BIAS */
        210, /* GL_COLOR_TABLE_FORMAT */
@@ -4421,380 +4459,380 @@ static const unsigned reduced_enums[1364] =
        216, /* GL_COLOR_TABLE_INTENSITY_SIZE */
         71, /* GL_BGR */
         72, /* GL_BGRA */
-       880, /* GL_MAX_ELEMENTS_VERTICES */
-       879, /* GL_MAX_ELEMENTS_INDICES */
-      1710, /* GL_TEXTURE_INDEX_SIZE_EXT */
+       888, /* GL_MAX_ELEMENTS_VERTICES */
+       887, /* GL_MAX_ELEMENTS_INDICES */
+      1727, /* GL_TEXTURE_INDEX_SIZE_EXT */
        145, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
-      1164, /* GL_POINT_SIZE_MIN */
-      1160, /* GL_POINT_SIZE_MAX */
-      1154, /* GL_POINT_FADE_THRESHOLD_SIZE */
-      1150, /* GL_POINT_DISTANCE_ATTENUATION */
+      1174, /* GL_POINT_SIZE_MIN */
+      1170, /* GL_POINT_SIZE_MAX */
+      1164, /* GL_POINT_FADE_THRESHOLD_SIZE */
+      1160, /* GL_POINT_DISTANCE_ATTENUATION */
        127, /* GL_CLAMP_TO_BORDER */
        130, /* GL_CLAMP_TO_EDGE */
-      1731, /* GL_TEXTURE_MIN_LOD */
-      1729, /* GL_TEXTURE_MAX_LOD */
-      1634, /* GL_TEXTURE_BASE_LEVEL */
-      1728, /* GL_TEXTURE_MAX_LEVEL */
-       624, /* GL_IGNORE_BORDER_HP */
+      1748, /* GL_TEXTURE_MIN_LOD */
+      1746, /* GL_TEXTURE_MAX_LOD */
+      1651, /* GL_TEXTURE_BASE_LEVEL */
+      1745, /* GL_TEXTURE_MAX_LEVEL */
+       631, /* GL_IGNORE_BORDER_HP */
        276, /* GL_CONSTANT_BORDER_HP */
-      1345, /* GL_REPLICATE_BORDER_HP */
+      1358, /* GL_REPLICATE_BORDER_HP */
        282, /* GL_CONVOLUTION_BORDER_COLOR */
-      1061, /* GL_OCCLUSION_TEST_HP */
-      1062, /* GL_OCCLUSION_TEST_RESULT_HP */
-       697, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-      1649, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-      1651, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-      1653, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-      1654, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1652, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-      1650, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-       862, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-       863, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1227, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-      1229, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-      1226, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-      1228, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-      1718, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-      1719, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-      1717, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-       590, /* GL_GENERATE_MIPMAP */
-       591, /* GL_GENERATE_MIPMAP_HINT */
-       529, /* GL_FOG_OFFSET_SGIX */
-       530, /* GL_FOG_OFFSET_VALUE_SGIX */
-      1663, /* GL_TEXTURE_COMPARE_SGIX */
-      1662, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-      1714, /* GL_TEXTURE_LEQUAL_R_SGIX */
-      1706, /* GL_TEXTURE_GEQUAL_R_SGIX */
-       359, /* GL_DEPTH_COMPONENT16 */
-       362, /* GL_DEPTH_COMPONENT24 */
-       365, /* GL_DEPTH_COMPONENT32 */
+      1071, /* GL_OCCLUSION_TEST_HP */
+      1072, /* GL_OCCLUSION_TEST_RESULT_HP */
+       704, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+      1666, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+      1668, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+      1670, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+      1671, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1669, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+      1667, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+       869, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+       870, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1237, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+      1239, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+      1236, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+      1238, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+      1735, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+      1736, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+      1734, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+       597, /* GL_GENERATE_MIPMAP */
+       598, /* GL_GENERATE_MIPMAP_HINT */
+       532, /* GL_FOG_OFFSET_SGIX */
+       533, /* GL_FOG_OFFSET_VALUE_SGIX */
+      1680, /* GL_TEXTURE_COMPARE_SGIX */
+      1679, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+      1731, /* GL_TEXTURE_LEQUAL_R_SGIX */
+      1723, /* GL_TEXTURE_GEQUAL_R_SGIX */
+       360, /* GL_DEPTH_COMPONENT16 */
+       363, /* GL_DEPTH_COMPONENT24 */
+       366, /* GL_DEPTH_COMPONENT32 */
        306, /* GL_CULL_VERTEX_EXT */
        308, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
        307, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
-      1872, /* GL_WRAP_BORDER_SUN */
-      1656, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-       690, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-      1444, /* GL_SINGLE_COLOR */
-      1430, /* GL_SEPARATE_SPECULAR_COLOR */
-      1439, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-       540, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
-       541, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
-       549, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
-       543, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
-       539, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
-       538, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
-       542, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
-       550, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
-       561, /* GL_FRAMEBUFFER_DEFAULT */
-       576, /* GL_FRAMEBUFFER_UNDEFINED */
-       372, /* GL_DEPTH_STENCIL_ATTACHMENT */
-       630, /* GL_INDEX */
-      1782, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-      1797, /* GL_UNSIGNED_SHORT_5_6_5 */
-      1798, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-      1795, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-      1793, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-      1790, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-      1788, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-      1726, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-      1727, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-      1725, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-       962, /* GL_MIRRORED_REPEAT */
-      1386, /* GL_RGB_S3TC */
-      1361, /* GL_RGB4_S3TC */
-      1384, /* GL_RGBA_S3TC */
-      1378, /* GL_RGBA4_S3TC */
-      1382, /* GL_RGBA_DXT5_S3TC */
-      1376, /* GL_RGBA4_DXT5_S3TC */
+      1891, /* GL_WRAP_BORDER_SUN */
+      1673, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+       697, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+      1457, /* GL_SINGLE_COLOR */
+      1443, /* GL_SEPARATE_SPECULAR_COLOR */
+      1452, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+       543, /* GL_FRAMEBUFFER_ATTACHMENT_COLOR_ENCODING */
+       544, /* GL_FRAMEBUFFER_ATTACHMENT_COMPONENT_TYPE */
+       552, /* GL_FRAMEBUFFER_ATTACHMENT_RED_SIZE */
+       546, /* GL_FRAMEBUFFER_ATTACHMENT_GREEN_SIZE */
+       542, /* GL_FRAMEBUFFER_ATTACHMENT_BLUE_SIZE */
+       541, /* GL_FRAMEBUFFER_ATTACHMENT_ALPHA_SIZE */
+       545, /* GL_FRAMEBUFFER_ATTACHMENT_DEPTH_SIZE */
+       553, /* GL_FRAMEBUFFER_ATTACHMENT_STENCIL_SIZE */
+       565, /* GL_FRAMEBUFFER_DEFAULT */
+       583, /* GL_FRAMEBUFFER_UNDEFINED */
+       373, /* GL_DEPTH_STENCIL_ATTACHMENT */
+       637, /* GL_INDEX */
+      1800, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+      1816, /* GL_UNSIGNED_SHORT_5_6_5 */
+      1817, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+      1814, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+      1812, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+      1809, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+      1807, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+      1743, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+      1744, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+      1742, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+       972, /* GL_MIRRORED_REPEAT */
+      1399, /* GL_RGB_S3TC */
+      1374, /* GL_RGB4_S3TC */
+      1397, /* GL_RGBA_S3TC */
+      1391, /* GL_RGBA4_S3TC */
+      1395, /* GL_RGBA_DXT5_S3TC */
+      1389, /* GL_RGBA4_DXT5_S3TC */
        264, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
        259, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
        260, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
        261, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
-      1023, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-      1022, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-       698, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-       516, /* GL_FOG_COORDINATE_SOURCE */
-       508, /* GL_FOG_COORD */
-       532, /* GL_FRAGMENT_DEPTH */
+      1033, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+      1032, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+       705, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+       519, /* GL_FOG_COORDINATE_SOURCE */
+       511, /* GL_FOG_COORD */
+       535, /* GL_FRAGMENT_DEPTH */
        312, /* GL_CURRENT_FOG_COORD */
-       515, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-       514, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-       513, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-       510, /* GL_FOG_COORDINATE_ARRAY */
+       518, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+       517, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+       516, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+       513, /* GL_FOG_COORDINATE_ARRAY */
        199, /* GL_COLOR_SUM */
        332, /* GL_CURRENT_SECONDARY_COLOR */
-      1423, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-      1425, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-      1424, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-      1422, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-      1419, /* GL_SECONDARY_COLOR_ARRAY */
+      1436, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+      1438, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+      1437, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+      1435, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+      1432, /* GL_SECONDARY_COLOR_ARRAY */
        330, /* GL_CURRENT_RASTER_SECONDARY_COLOR */
         28, /* GL_ALIASED_POINT_SIZE_RANGE */
         27, /* GL_ALIASED_LINE_WIDTH_RANGE */
-      1563, /* GL_TEXTURE0 */
-      1565, /* GL_TEXTURE1 */
-      1587, /* GL_TEXTURE2 */
-      1609, /* GL_TEXTURE3 */
-      1615, /* GL_TEXTURE4 */
-      1617, /* GL_TEXTURE5 */
-      1619, /* GL_TEXTURE6 */
-      1621, /* GL_TEXTURE7 */
-      1623, /* GL_TEXTURE8 */
-      1625, /* GL_TEXTURE9 */
-      1566, /* GL_TEXTURE10 */
-      1568, /* GL_TEXTURE11 */
-      1570, /* GL_TEXTURE12 */
-      1572, /* GL_TEXTURE13 */
-      1574, /* GL_TEXTURE14 */
-      1576, /* GL_TEXTURE15 */
-      1578, /* GL_TEXTURE16 */
-      1580, /* GL_TEXTURE17 */
-      1582, /* GL_TEXTURE18 */
-      1584, /* GL_TEXTURE19 */
-      1588, /* GL_TEXTURE20 */
-      1590, /* GL_TEXTURE21 */
-      1592, /* GL_TEXTURE22 */
-      1594, /* GL_TEXTURE23 */
-      1596, /* GL_TEXTURE24 */
-      1598, /* GL_TEXTURE25 */
-      1600, /* GL_TEXTURE26 */
-      1602, /* GL_TEXTURE27 */
-      1604, /* GL_TEXTURE28 */
-      1606, /* GL_TEXTURE29 */
-      1610, /* GL_TEXTURE30 */
-      1612, /* GL_TEXTURE31 */
+      1580, /* GL_TEXTURE0 */
+      1582, /* GL_TEXTURE1 */
+      1604, /* GL_TEXTURE2 */
+      1626, /* GL_TEXTURE3 */
+      1632, /* GL_TEXTURE4 */
+      1634, /* GL_TEXTURE5 */
+      1636, /* GL_TEXTURE6 */
+      1638, /* GL_TEXTURE7 */
+      1640, /* GL_TEXTURE8 */
+      1642, /* GL_TEXTURE9 */
+      1583, /* GL_TEXTURE10 */
+      1585, /* GL_TEXTURE11 */
+      1587, /* GL_TEXTURE12 */
+      1589, /* GL_TEXTURE13 */
+      1591, /* GL_TEXTURE14 */
+      1593, /* GL_TEXTURE15 */
+      1595, /* GL_TEXTURE16 */
+      1597, /* GL_TEXTURE17 */
+      1599, /* GL_TEXTURE18 */
+      1601, /* GL_TEXTURE19 */
+      1605, /* GL_TEXTURE20 */
+      1607, /* GL_TEXTURE21 */
+      1609, /* GL_TEXTURE22 */
+      1611, /* GL_TEXTURE23 */
+      1613, /* GL_TEXTURE24 */
+      1615, /* GL_TEXTURE25 */
+      1617, /* GL_TEXTURE26 */
+      1619, /* GL_TEXTURE27 */
+      1621, /* GL_TEXTURE28 */
+      1623, /* GL_TEXTURE29 */
+      1627, /* GL_TEXTURE30 */
+      1629, /* GL_TEXTURE31 */
         18, /* GL_ACTIVE_TEXTURE */
        133, /* GL_CLIENT_ACTIVE_TEXTURE */
-       938, /* GL_MAX_TEXTURE_UNITS */
-      1758, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-      1761, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-      1763, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-      1755, /* GL_TRANSPOSE_COLOR_MATRIX */
-      1545, /* GL_SUBTRACT */
-       925, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+       948, /* GL_MAX_TEXTURE_UNITS */
+      1776, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+      1779, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+      1781, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+      1773, /* GL_TRANSPOSE_COLOR_MATRIX */
+      1562, /* GL_SUBTRACT */
+       933, /* GL_MAX_RENDERBUFFER_SIZE */
        247, /* GL_COMPRESSED_ALPHA */
        251, /* GL_COMPRESSED_LUMINANCE */
        252, /* GL_COMPRESSED_LUMINANCE_ALPHA */
        249, /* GL_COMPRESSED_INTENSITY */
        255, /* GL_COMPRESSED_RGB */
        256, /* GL_COMPRESSED_RGBA */
-      1670, /* GL_TEXTURE_COMPRESSION_HINT */
-      1735, /* GL_TEXTURE_RECTANGLE_ARB */
-      1642, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-      1288, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-       923, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-       371, /* GL_DEPTH_STENCIL */
-      1786, /* GL_UNSIGNED_INT_24_8 */
-       934, /* GL_MAX_TEXTURE_LOD_BIAS */
-      1724, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-       935, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-      1700, /* GL_TEXTURE_FILTER_CONTROL */
-      1715, /* GL_TEXTURE_LOD_BIAS */
+      1687, /* GL_TEXTURE_COMPRESSION_HINT */
+      1752, /* GL_TEXTURE_RECTANGLE_ARB */
+      1659, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+      1298, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+       931, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+       372, /* GL_DEPTH_STENCIL */
+      1804, /* GL_UNSIGNED_INT_24_8 */
+       944, /* GL_MAX_TEXTURE_LOD_BIAS */
+      1741, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+       945, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+      1717, /* GL_TEXTURE_FILTER_CONTROL */
+      1732, /* GL_TEXTURE_LOD_BIAS */
        232, /* GL_COMBINE4 */
-       928, /* GL_MAX_SHININESS_NV */
-       929, /* GL_MAX_SPOT_EXPONENT_NV */
-       628, /* GL_INCR_WRAP */
+       938, /* GL_MAX_SHININESS_NV */
+       939, /* GL_MAX_SPOT_EXPONENT_NV */
+       635, /* GL_INCR_WRAP */
        343, /* GL_DECR_WRAP */
-       982, /* GL_MODELVIEW1_ARB */
-      1038, /* GL_NORMAL_MAP */
-      1320, /* GL_REFLECTION_MAP */
-      1679, /* GL_TEXTURE_CUBE_MAP */
-      1640, /* GL_TEXTURE_BINDING_CUBE_MAP */
-      1687, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-      1681, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-      1689, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-      1683, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-      1691, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-      1685, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-      1286, /* GL_PROXY_TEXTURE_CUBE_MAP */
-       874, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-      1017, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-       524, /* GL_FOG_DISTANCE_MODE_NV */
-       475, /* GL_EYE_RADIAL_NV */
-       474, /* GL_EYE_PLANE_ABSOLUTE_NV */
+       992, /* GL_MODELVIEW1_ARB */
+      1048, /* GL_NORMAL_MAP */
+      1331, /* GL_REFLECTION_MAP */
+      1696, /* GL_TEXTURE_CUBE_MAP */
+      1657, /* GL_TEXTURE_BINDING_CUBE_MAP */
+      1704, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+      1698, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+      1706, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+      1700, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+      1708, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+      1702, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+      1296, /* GL_PROXY_TEXTURE_CUBE_MAP */
+       882, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+      1027, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+       527, /* GL_FOG_DISTANCE_MODE_NV */
+       478, /* GL_EYE_RADIAL_NV */
+       477, /* GL_EYE_PLANE_ABSOLUTE_NV */
        231, /* GL_COMBINE */
        238, /* GL_COMBINE_RGB */
        233, /* GL_COMBINE_ALPHA */
-      1387, /* GL_RGB_SCALE */
+      1400, /* GL_RGB_SCALE */
         24, /* GL_ADD_SIGNED */
-       656, /* GL_INTERPOLATE */
+       663, /* GL_INTERPOLATE */
        271, /* GL_CONSTANT */
-      1233, /* GL_PRIMARY_COLOR */
-      1230, /* GL_PREVIOUS */
-      1459, /* GL_SOURCE0_RGB */
-      1465, /* GL_SOURCE1_RGB */
-      1471, /* GL_SOURCE2_RGB */
-      1475, /* GL_SOURCE3_RGB_NV */
-      1456, /* GL_SOURCE0_ALPHA */
-      1462, /* GL_SOURCE1_ALPHA */
-      1468, /* GL_SOURCE2_ALPHA */
-      1474, /* GL_SOURCE3_ALPHA_NV */
-      1075, /* GL_OPERAND0_RGB */
-      1081, /* GL_OPERAND1_RGB */
-      1087, /* GL_OPERAND2_RGB */
-      1091, /* GL_OPERAND3_RGB_NV */
-      1072, /* GL_OPERAND0_ALPHA */
-      1078, /* GL_OPERAND1_ALPHA */
-      1084, /* GL_OPERAND2_ALPHA */
-      1090, /* GL_OPERAND3_ALPHA_NV */
-      1810, /* GL_VERTEX_ARRAY_BINDING */
-      1733, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
-      1734, /* GL_TEXTURE_RANGE_POINTER_APPLE */
-      1876, /* GL_YCBCR_422_APPLE */
-      1799, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-      1801, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-      1742, /* GL_TEXTURE_STORAGE_HINT_APPLE */
-      1536, /* GL_STORAGE_PRIVATE_APPLE */
-      1535, /* GL_STORAGE_CACHED_APPLE */
-      1537, /* GL_STORAGE_SHARED_APPLE */
-      1446, /* GL_SLICE_ACCUM_SUN */
-      1295, /* GL_QUAD_MESH_SUN */
-      1768, /* GL_TRIANGLE_MESH_SUN */
-      1849, /* GL_VERTEX_PROGRAM_ARB */
-      1860, /* GL_VERTEX_STATE_PROGRAM_NV */
-      1836, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-      1842, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-      1844, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-      1846, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+      1243, /* GL_PRIMARY_COLOR */
+      1240, /* GL_PREVIOUS */
+      1472, /* GL_SOURCE0_RGB */
+      1478, /* GL_SOURCE1_RGB */
+      1484, /* GL_SOURCE2_RGB */
+      1488, /* GL_SOURCE3_RGB_NV */
+      1469, /* GL_SOURCE0_ALPHA */
+      1475, /* GL_SOURCE1_ALPHA */
+      1481, /* GL_SOURCE2_ALPHA */
+      1487, /* GL_SOURCE3_ALPHA_NV */
+      1085, /* GL_OPERAND0_RGB */
+      1091, /* GL_OPERAND1_RGB */
+      1097, /* GL_OPERAND2_RGB */
+      1101, /* GL_OPERAND3_RGB_NV */
+      1082, /* GL_OPERAND0_ALPHA */
+      1088, /* GL_OPERAND1_ALPHA */
+      1094, /* GL_OPERAND2_ALPHA */
+      1100, /* GL_OPERAND3_ALPHA_NV */
+      1829, /* GL_VERTEX_ARRAY_BINDING */
+      1750, /* GL_TEXTURE_RANGE_LENGTH_APPLE */
+      1751, /* GL_TEXTURE_RANGE_POINTER_APPLE */
+      1895, /* GL_YCBCR_422_APPLE */
+      1818, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+      1820, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+      1760, /* GL_TEXTURE_STORAGE_HINT_APPLE */
+      1553, /* GL_STORAGE_PRIVATE_APPLE */
+      1552, /* GL_STORAGE_CACHED_APPLE */
+      1554, /* GL_STORAGE_SHARED_APPLE */
+      1459, /* GL_SLICE_ACCUM_SUN */
+      1305, /* GL_QUAD_MESH_SUN */
+      1786, /* GL_TRIANGLE_MESH_SUN */
+      1868, /* GL_VERTEX_PROGRAM_ARB */
+      1879, /* GL_VERTEX_STATE_PROGRAM_NV */
+      1855, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+      1861, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+      1863, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+      1865, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
        334, /* GL_CURRENT_VERTEX_ATTRIB */
-      1246, /* GL_PROGRAM_LENGTH_ARB */
-      1261, /* GL_PROGRAM_STRING_ARB */
-      1004, /* GL_MODELVIEW_PROJECTION_NV */
-       623, /* GL_IDENTITY_NV */
-       670, /* GL_INVERSE_NV */
-      1760, /* GL_TRANSPOSE_NV */
-       671, /* GL_INVERSE_TRANSPOSE_NV */
-       909, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-       908, /* GL_MAX_PROGRAM_MATRICES_ARB */
-       811, /* GL_MATRIX0_NV */
-       823, /* GL_MATRIX1_NV */
-       835, /* GL_MATRIX2_NV */
-       839, /* GL_MATRIX3_NV */
-       841, /* GL_MATRIX4_NV */
-       843, /* GL_MATRIX5_NV */
-       845, /* GL_MATRIX6_NV */
-       847, /* GL_MATRIX7_NV */
+      1256, /* GL_PROGRAM_LENGTH_ARB */
+      1271, /* GL_PROGRAM_STRING_ARB */
+      1014, /* GL_MODELVIEW_PROJECTION_NV */
+       630, /* GL_IDENTITY_NV */
+       677, /* GL_INVERSE_NV */
+      1778, /* GL_TRANSPOSE_NV */
+       678, /* GL_INVERSE_TRANSPOSE_NV */
+       917, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+       916, /* GL_MAX_PROGRAM_MATRICES_ARB */
+       818, /* GL_MATRIX0_NV */
+       830, /* GL_MATRIX1_NV */
+       842, /* GL_MATRIX2_NV */
+       846, /* GL_MATRIX3_NV */
+       848, /* GL_MATRIX4_NV */
+       850, /* GL_MATRIX5_NV */
+       852, /* GL_MATRIX6_NV */
+       854, /* GL_MATRIX7_NV */
        318, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
        315, /* GL_CURRENT_MATRIX_ARB */
-      1852, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-      1855, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-      1258, /* GL_PROGRAM_PARAMETER_NV */
-      1840, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-      1263, /* GL_PROGRAM_TARGET_NV */
-      1260, /* GL_PROGRAM_RESIDENT_NV */
-      1752, /* GL_TRACK_MATRIX_NV */
-      1753, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-      1850, /* GL_VERTEX_PROGRAM_BINDING_NV */
-      1240, /* GL_PROGRAM_ERROR_POSITION_ARB */
-       355, /* GL_DEPTH_CLAMP */
-      1818, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-      1825, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-      1826, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-      1827, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-      1828, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-      1829, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-      1830, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-      1831, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-      1832, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-      1833, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-      1819, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-      1820, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-      1821, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-      1822, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-      1823, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-      1824, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-       759, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-       766, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-       767, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-       768, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-       769, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-       770, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-       771, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-       772, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-       773, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-       774, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-       760, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-       761, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-       762, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-       763, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-       764, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-       765, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-       786, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-       793, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-       794, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-       795, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-       796, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-       797, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-       798, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-      1239, /* GL_PROGRAM_BINDING_ARB */
-       800, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-       801, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-       787, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-       788, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-       789, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-       790, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-       791, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-       792, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-      1668, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-      1665, /* GL_TEXTURE_COMPRESSED */
-      1043, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+      1871, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+      1874, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+      1268, /* GL_PROGRAM_PARAMETER_NV */
+      1859, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+      1273, /* GL_PROGRAM_TARGET_NV */
+      1270, /* GL_PROGRAM_RESIDENT_NV */
+      1770, /* GL_TRACK_MATRIX_NV */
+      1771, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+      1869, /* GL_VERTEX_PROGRAM_BINDING_NV */
+      1250, /* GL_PROGRAM_ERROR_POSITION_ARB */
+       356, /* GL_DEPTH_CLAMP */
+      1837, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+      1844, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+      1845, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+      1846, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+      1847, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+      1848, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+      1849, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+      1850, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+      1851, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+      1852, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+      1838, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+      1839, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+      1840, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+      1841, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+      1842, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+      1843, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+       766, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+       773, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+       774, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+       775, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+       776, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+       777, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+       778, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+       779, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+       780, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+       781, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+       767, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+       768, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+       769, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+       770, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+       771, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+       772, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+       793, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+       800, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+       801, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+       802, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+       803, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+       804, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+       805, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+      1249, /* GL_PROGRAM_BINDING_ARB */
+       807, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+       808, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+       794, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+       795, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+       796, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+       797, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+       798, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+       799, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+      1685, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+      1682, /* GL_TEXTURE_COMPRESSED */
+      1053, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
        269, /* GL_COMPRESSED_TEXTURE_FORMATS */
-       951, /* GL_MAX_VERTEX_UNITS_ARB */
+       961, /* GL_MAX_VERTEX_UNITS_ARB */
         22, /* GL_ACTIVE_VERTEX_UNITS_ARB */
-      1871, /* GL_WEIGHT_SUM_UNITY_ARB */
-      1848, /* GL_VERTEX_BLEND_ARB */
+      1890, /* GL_WEIGHT_SUM_UNITY_ARB */
+      1867, /* GL_VERTEX_BLEND_ARB */
        336, /* GL_CURRENT_WEIGHT_ARB */
-      1870, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-      1869, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-      1868, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-      1867, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-      1864, /* GL_WEIGHT_ARRAY_ARB */
-       384, /* GL_DOT3_RGB */
-       385, /* GL_DOT3_RGBA */
+      1889, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+      1888, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+      1887, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+      1886, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+      1883, /* GL_WEIGHT_ARRAY_ARB */
+       386, /* GL_DOT3_RGB */
+       387, /* GL_DOT3_RGBA */
        263, /* GL_COMPRESSED_RGB_FXT1_3DFX */
        258, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
-      1012, /* GL_MULTISAMPLE_3DFX */
-      1408, /* GL_SAMPLE_BUFFERS_3DFX */
-      1399, /* GL_SAMPLES_3DFX */
-       993, /* GL_MODELVIEW2_ARB */
-       996, /* GL_MODELVIEW3_ARB */
-       997, /* GL_MODELVIEW4_ARB */
-       998, /* GL_MODELVIEW5_ARB */
-       999, /* GL_MODELVIEW6_ARB */
-      1000, /* GL_MODELVIEW7_ARB */
-      1001, /* GL_MODELVIEW8_ARB */
-      1002, /* GL_MODELVIEW9_ARB */
-       972, /* GL_MODELVIEW10_ARB */
-       973, /* GL_MODELVIEW11_ARB */
-       974, /* GL_MODELVIEW12_ARB */
-       975, /* GL_MODELVIEW13_ARB */
-       976, /* GL_MODELVIEW14_ARB */
-       977, /* GL_MODELVIEW15_ARB */
-       978, /* GL_MODELVIEW16_ARB */
-       979, /* GL_MODELVIEW17_ARB */
-       980, /* GL_MODELVIEW18_ARB */
-       981, /* GL_MODELVIEW19_ARB */
-       983, /* GL_MODELVIEW20_ARB */
-       984, /* GL_MODELVIEW21_ARB */
-       985, /* GL_MODELVIEW22_ARB */
-       986, /* GL_MODELVIEW23_ARB */
-       987, /* GL_MODELVIEW24_ARB */
-       988, /* GL_MODELVIEW25_ARB */
-       989, /* GL_MODELVIEW26_ARB */
-       990, /* GL_MODELVIEW27_ARB */
-       991, /* GL_MODELVIEW28_ARB */
-       992, /* GL_MODELVIEW29_ARB */
-       994, /* GL_MODELVIEW30_ARB */
-       995, /* GL_MODELVIEW31_ARB */
-       389, /* GL_DOT3_RGB_EXT */
-       387, /* GL_DOT3_RGBA_EXT */
-       966, /* GL_MIRROR_CLAMP_EXT */
-       969, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-      1007, /* GL_MODULATE_ADD_ATI */
-      1008, /* GL_MODULATE_SIGNED_ADD_ATI */
-      1009, /* GL_MODULATE_SUBTRACT_ATI */
-      1877, /* GL_YCBCR_MESA */
-      1099, /* GL_PACK_INVERT_MESA */
+      1022, /* GL_MULTISAMPLE_3DFX */
+      1421, /* GL_SAMPLE_BUFFERS_3DFX */
+      1412, /* GL_SAMPLES_3DFX */
+      1003, /* GL_MODELVIEW2_ARB */
+      1006, /* GL_MODELVIEW3_ARB */
+      1007, /* GL_MODELVIEW4_ARB */
+      1008, /* GL_MODELVIEW5_ARB */
+      1009, /* GL_MODELVIEW6_ARB */
+      1010, /* GL_MODELVIEW7_ARB */
+      1011, /* GL_MODELVIEW8_ARB */
+      1012, /* GL_MODELVIEW9_ARB */
+       982, /* GL_MODELVIEW10_ARB */
+       983, /* GL_MODELVIEW11_ARB */
+       984, /* GL_MODELVIEW12_ARB */
+       985, /* GL_MODELVIEW13_ARB */
+       986, /* GL_MODELVIEW14_ARB */
+       987, /* GL_MODELVIEW15_ARB */
+       988, /* GL_MODELVIEW16_ARB */
+       989, /* GL_MODELVIEW17_ARB */
+       990, /* GL_MODELVIEW18_ARB */
+       991, /* GL_MODELVIEW19_ARB */
+       993, /* GL_MODELVIEW20_ARB */
+       994, /* GL_MODELVIEW21_ARB */
+       995, /* GL_MODELVIEW22_ARB */
+       996, /* GL_MODELVIEW23_ARB */
+       997, /* GL_MODELVIEW24_ARB */
+       998, /* GL_MODELVIEW25_ARB */
+       999, /* GL_MODELVIEW26_ARB */
+      1000, /* GL_MODELVIEW27_ARB */
+      1001, /* GL_MODELVIEW28_ARB */
+      1002, /* GL_MODELVIEW29_ARB */
+      1004, /* GL_MODELVIEW30_ARB */
+      1005, /* GL_MODELVIEW31_ARB */
+       391, /* GL_DOT3_RGB_EXT */
+       389, /* GL_DOT3_RGBA_EXT */
+       976, /* GL_MIRROR_CLAMP_EXT */
+       979, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+      1017, /* GL_MODULATE_ADD_ATI */
+      1018, /* GL_MODULATE_SIGNED_ADD_ATI */
+      1019, /* GL_MODULATE_SUBTRACT_ATI */
+      1896, /* GL_YCBCR_MESA */
+      1109, /* GL_PACK_INVERT_MESA */
        339, /* GL_DEBUG_OBJECT_MESA */
        340, /* GL_DEBUG_PRINT_MESA */
        338, /* GL_DEBUG_ASSERT_MESA */
@@ -4804,293 +4842,293 @@ static const unsigned reduced_enums[1364] =
        117, /* GL_BUMP_ROT_MATRIX_SIZE_ATI */
        115, /* GL_BUMP_NUM_TEX_UNITS_ATI */
        119, /* GL_BUMP_TEX_UNITS_ATI */
-       448, /* GL_DUDV_ATI */
-       447, /* GL_DU8DV8_ATI */
+       451, /* GL_DUDV_ATI */
+       450, /* GL_DU8DV8_ATI */
        114, /* GL_BUMP_ENVMAP_ATI */
        118, /* GL_BUMP_TARGET_ATI */
-      1507, /* GL_STENCIL_BACK_FUNC */
-      1505, /* GL_STENCIL_BACK_FAIL */
-      1509, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-      1511, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-       533, /* GL_FRAGMENT_PROGRAM_ARB */
-      1237, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-      1266, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-      1265, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-      1249, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-      1255, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-      1254, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       898, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-       921, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-       920, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-       911, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-       917, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-       916, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       876, /* GL_MAX_DRAW_BUFFERS */
-       393, /* GL_DRAW_BUFFER0 */
-       396, /* GL_DRAW_BUFFER1 */
-       417, /* GL_DRAW_BUFFER2 */
-       420, /* GL_DRAW_BUFFER3 */
-       423, /* GL_DRAW_BUFFER4 */
-       426, /* GL_DRAW_BUFFER5 */
-       429, /* GL_DRAW_BUFFER6 */
-       432, /* GL_DRAW_BUFFER7 */
-       435, /* GL_DRAW_BUFFER8 */
-       438, /* GL_DRAW_BUFFER9 */
-       397, /* GL_DRAW_BUFFER10 */
-       400, /* GL_DRAW_BUFFER11 */
-       403, /* GL_DRAW_BUFFER12 */
-       406, /* GL_DRAW_BUFFER13 */
-       409, /* GL_DRAW_BUFFER14 */
-       412, /* GL_DRAW_BUFFER15 */
+      1520, /* GL_STENCIL_BACK_FUNC */
+      1518, /* GL_STENCIL_BACK_FAIL */
+      1522, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+      1524, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+       536, /* GL_FRAGMENT_PROGRAM_ARB */
+      1247, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+      1276, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+      1275, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+      1259, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+      1265, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+      1264, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       906, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+       929, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+       928, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+       919, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+       925, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+       924, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       884, /* GL_MAX_DRAW_BUFFERS */
+       395, /* GL_DRAW_BUFFER0 */
+       398, /* GL_DRAW_BUFFER1 */
+       419, /* GL_DRAW_BUFFER2 */
+       422, /* GL_DRAW_BUFFER3 */
+       425, /* GL_DRAW_BUFFER4 */
+       428, /* GL_DRAW_BUFFER5 */
+       431, /* GL_DRAW_BUFFER6 */
+       434, /* GL_DRAW_BUFFER7 */
+       437, /* GL_DRAW_BUFFER8 */
+       440, /* GL_DRAW_BUFFER9 */
+       399, /* GL_DRAW_BUFFER10 */
+       402, /* GL_DRAW_BUFFER11 */
+       405, /* GL_DRAW_BUFFER12 */
+       408, /* GL_DRAW_BUFFER13 */
+       411, /* GL_DRAW_BUFFER14 */
+       414, /* GL_DRAW_BUFFER15 */
         82, /* GL_BLEND_EQUATION_ALPHA */
-       856, /* GL_MATRIX_PALETTE_ARB */
-       892, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-       895, /* GL_MAX_PALETTE_MATRICES_ARB */
+       863, /* GL_MATRIX_PALETTE_ARB */
+       900, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+       903, /* GL_MAX_PALETTE_MATRICES_ARB */
        321, /* GL_CURRENT_PALETTE_MATRIX_ARB */
-       850, /* GL_MATRIX_INDEX_ARRAY_ARB */
+       857, /* GL_MATRIX_INDEX_ARRAY_ARB */
        316, /* GL_CURRENT_MATRIX_INDEX_ARB */
-       852, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-       854, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-       853, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-       851, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-      1695, /* GL_TEXTURE_DEPTH_SIZE */
-       377, /* GL_DEPTH_TEXTURE_MODE */
-      1660, /* GL_TEXTURE_COMPARE_MODE */
-      1658, /* GL_TEXTURE_COMPARE_FUNC */
+       859, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+       861, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+       860, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+       858, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+      1712, /* GL_TEXTURE_DEPTH_SIZE */
+       379, /* GL_DEPTH_TEXTURE_MODE */
+      1677, /* GL_TEXTURE_COMPARE_MODE */
+      1675, /* GL_TEXTURE_COMPARE_FUNC */
        242, /* GL_COMPARE_R_TO_TEXTURE */
-      1171, /* GL_POINT_SPRITE */
+      1181, /* GL_POINT_SPRITE */
        296, /* GL_COORD_REPLACE */
-      1175, /* GL_POINT_SPRITE_R_MODE_NV */
-      1297, /* GL_QUERY_COUNTER_BITS */
+      1185, /* GL_POINT_SPRITE_R_MODE_NV */
+      1307, /* GL_QUERY_COUNTER_BITS */
        323, /* GL_CURRENT_QUERY */
-      1299, /* GL_QUERY_RESULT */
-      1301, /* GL_QUERY_RESULT_AVAILABLE */
-       945, /* GL_MAX_VERTEX_ATTRIBS */
-      1838, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-       375, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-       374, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-       930, /* GL_MAX_TEXTURE_COORDS */
-       932, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-      1242, /* GL_PROGRAM_ERROR_STRING_ARB */
-      1244, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-      1243, /* GL_PROGRAM_FORMAT_ARB */
-      1744, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
-       353, /* GL_DEPTH_BOUNDS_TEST_EXT */
-       352, /* GL_DEPTH_BOUNDS_EXT */
+      1309, /* GL_QUERY_RESULT */
+      1311, /* GL_QUERY_RESULT_AVAILABLE */
+       955, /* GL_MAX_VERTEX_ATTRIBS */
+      1857, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+       377, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+       376, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+       940, /* GL_MAX_TEXTURE_COORDS */
+       942, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+      1252, /* GL_PROGRAM_ERROR_STRING_ARB */
+      1254, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+      1253, /* GL_PROGRAM_FORMAT_ARB */
+      1762, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+       354, /* GL_DEPTH_BOUNDS_TEST_EXT */
+       353, /* GL_DEPTH_BOUNDS_EXT */
         53, /* GL_ARRAY_BUFFER */
-       461, /* GL_ELEMENT_ARRAY_BUFFER */
+       464, /* GL_ELEMENT_ARRAY_BUFFER */
         54, /* GL_ARRAY_BUFFER_BINDING */
-       462, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-      1812, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-      1033, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+       465, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+      1831, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+      1043, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
        149, /* GL_COLOR_ARRAY_BUFFER_BINDING */
-       632, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-      1673, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-       457, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-      1420, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-       511, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-      1865, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-      1834, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-      1245, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-       904, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-      1251, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-       913, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-      1264, /* GL_PROGRAM_TEMPORARIES_ARB */
-       919, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-      1253, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-       915, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-      1257, /* GL_PROGRAM_PARAMETERS_ARB */
-       918, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-      1252, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-       914, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-      1238, /* GL_PROGRAM_ATTRIBS_ARB */
-       899, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-      1250, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-       912, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-      1236, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-       897, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-      1248, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       910, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       905, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-       901, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-      1267, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-      1757, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-      1310, /* GL_READ_ONLY */
-      1873, /* GL_WRITE_ONLY */
-      1312, /* GL_READ_WRITE */
+       639, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+      1690, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+       460, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+      1433, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+       514, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+      1884, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+      1853, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+      1255, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+       912, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+      1261, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+       921, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+      1274, /* GL_PROGRAM_TEMPORARIES_ARB */
+       927, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+      1263, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+       923, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+      1267, /* GL_PROGRAM_PARAMETERS_ARB */
+       926, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+      1262, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+       922, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+      1248, /* GL_PROGRAM_ATTRIBS_ARB */
+       907, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+      1260, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+       920, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+      1246, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+       905, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+      1258, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       918, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       913, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+       909, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+      1277, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+      1775, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+      1321, /* GL_READ_ONLY */
+      1892, /* GL_WRITE_ONLY */
+      1323, /* GL_READ_WRITE */
        102, /* GL_BUFFER_ACCESS */
        105, /* GL_BUFFER_MAPPED */
        107, /* GL_BUFFER_MAP_POINTER */
-      1751, /* GL_TIME_ELAPSED_EXT */
-       810, /* GL_MATRIX0_ARB */
-       822, /* GL_MATRIX1_ARB */
-       834, /* GL_MATRIX2_ARB */
-       838, /* GL_MATRIX3_ARB */
-       840, /* GL_MATRIX4_ARB */
-       842, /* GL_MATRIX5_ARB */
-       844, /* GL_MATRIX6_ARB */
-       846, /* GL_MATRIX7_ARB */
-       848, /* GL_MATRIX8_ARB */
-       849, /* GL_MATRIX9_ARB */
-       812, /* GL_MATRIX10_ARB */
-       813, /* GL_MATRIX11_ARB */
-       814, /* GL_MATRIX12_ARB */
-       815, /* GL_MATRIX13_ARB */
-       816, /* GL_MATRIX14_ARB */
-       817, /* GL_MATRIX15_ARB */
-       818, /* GL_MATRIX16_ARB */
-       819, /* GL_MATRIX17_ARB */
-       820, /* GL_MATRIX18_ARB */
-       821, /* GL_MATRIX19_ARB */
-       824, /* GL_MATRIX20_ARB */
-       825, /* GL_MATRIX21_ARB */
-       826, /* GL_MATRIX22_ARB */
-       827, /* GL_MATRIX23_ARB */
-       828, /* GL_MATRIX24_ARB */
-       829, /* GL_MATRIX25_ARB */
-       830, /* GL_MATRIX26_ARB */
-       831, /* GL_MATRIX27_ARB */
-       832, /* GL_MATRIX28_ARB */
-       833, /* GL_MATRIX29_ARB */
-       836, /* GL_MATRIX30_ARB */
-       837, /* GL_MATRIX31_ARB */
-      1540, /* GL_STREAM_DRAW */
-      1542, /* GL_STREAM_READ */
-      1538, /* GL_STREAM_COPY */
-      1498, /* GL_STATIC_DRAW */
-      1500, /* GL_STATIC_READ */
-      1496, /* GL_STATIC_COPY */
-       451, /* GL_DYNAMIC_DRAW */
-       453, /* GL_DYNAMIC_READ */
-       449, /* GL_DYNAMIC_COPY */
-      1139, /* GL_PIXEL_PACK_BUFFER */
-      1143, /* GL_PIXEL_UNPACK_BUFFER */
-      1140, /* GL_PIXEL_PACK_BUFFER_BINDING */
-      1144, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
+      1769, /* GL_TIME_ELAPSED_EXT */
+       817, /* GL_MATRIX0_ARB */
+       829, /* GL_MATRIX1_ARB */
+       841, /* GL_MATRIX2_ARB */
+       845, /* GL_MATRIX3_ARB */
+       847, /* GL_MATRIX4_ARB */
+       849, /* GL_MATRIX5_ARB */
+       851, /* GL_MATRIX6_ARB */
+       853, /* GL_MATRIX7_ARB */
+       855, /* GL_MATRIX8_ARB */
+       856, /* GL_MATRIX9_ARB */
+       819, /* GL_MATRIX10_ARB */
+       820, /* GL_MATRIX11_ARB */
+       821, /* GL_MATRIX12_ARB */
+       822, /* GL_MATRIX13_ARB */
+       823, /* GL_MATRIX14_ARB */
+       824, /* GL_MATRIX15_ARB */
+       825, /* GL_MATRIX16_ARB */
+       826, /* GL_MATRIX17_ARB */
+       827, /* GL_MATRIX18_ARB */
+       828, /* GL_MATRIX19_ARB */
+       831, /* GL_MATRIX20_ARB */
+       832, /* GL_MATRIX21_ARB */
+       833, /* GL_MATRIX22_ARB */
+       834, /* GL_MATRIX23_ARB */
+       835, /* GL_MATRIX24_ARB */
+       836, /* GL_MATRIX25_ARB */
+       837, /* GL_MATRIX26_ARB */
+       838, /* GL_MATRIX27_ARB */
+       839, /* GL_MATRIX28_ARB */
+       840, /* GL_MATRIX29_ARB */
+       843, /* GL_MATRIX30_ARB */
+       844, /* GL_MATRIX31_ARB */
+      1557, /* GL_STREAM_DRAW */
+      1559, /* GL_STREAM_READ */
+      1555, /* GL_STREAM_COPY */
+      1511, /* GL_STATIC_DRAW */
+      1513, /* GL_STATIC_READ */
+      1509, /* GL_STATIC_COPY */
+       454, /* GL_DYNAMIC_DRAW */
+       456, /* GL_DYNAMIC_READ */
+       452, /* GL_DYNAMIC_COPY */
+      1149, /* GL_PIXEL_PACK_BUFFER */
+      1153, /* GL_PIXEL_UNPACK_BUFFER */
+      1150, /* GL_PIXEL_PACK_BUFFER_BINDING */
+      1154, /* GL_PIXEL_UNPACK_BUFFER_BINDING */
        347, /* GL_DEPTH24_STENCIL8 */
-      1741, /* GL_TEXTURE_STENCIL_SIZE */
-      1693, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
-       900, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-       903, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-       907, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-       906, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-       859, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-      1531, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+      1758, /* GL_TEXTURE_STENCIL_SIZE */
+      1710, /* GL_TEXTURE_CUBE_MAP_SEAMLESS */
+       908, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+       911, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+       915, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+       914, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+       866, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+      1548, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
         17, /* GL_ACTIVE_STENCIL_FACE_EXT */
-       967, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-      1401, /* GL_SAMPLES_PASSED */
+       977, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+      1414, /* GL_SAMPLES_PASSED */
        109, /* GL_BUFFER_SERIALIZED_MODIFY_APPLE */
        104, /* GL_BUFFER_FLUSHING_UNMAP_APPLE */
-       534, /* GL_FRAGMENT_SHADER */
-      1858, /* GL_VERTEX_SHADER */
-      1256, /* GL_PROGRAM_OBJECT_ARB */
-      1433, /* GL_SHADER_OBJECT_ARB */
-       883, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-       949, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-       943, /* GL_MAX_VARYING_FLOATS */
-       947, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-       868, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-      1059, /* GL_OBJECT_TYPE_ARB */
-      1435, /* GL_SHADER_TYPE */
-       499, /* GL_FLOAT_VEC2 */
-       501, /* GL_FLOAT_VEC3 */
-       503, /* GL_FLOAT_VEC4 */
-       659, /* GL_INT_VEC2 */
-       661, /* GL_INT_VEC3 */
-       663, /* GL_INT_VEC4 */
+       537, /* GL_FRAGMENT_SHADER */
+      1877, /* GL_VERTEX_SHADER */
+      1266, /* GL_PROGRAM_OBJECT_ARB */
+      1446, /* GL_SHADER_OBJECT_ARB */
+       891, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+       959, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+       953, /* GL_MAX_VARYING_FLOATS */
+       957, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+       876, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+      1069, /* GL_OBJECT_TYPE_ARB */
+      1448, /* GL_SHADER_TYPE */
+       502, /* GL_FLOAT_VEC2 */
+       504, /* GL_FLOAT_VEC3 */
+       506, /* GL_FLOAT_VEC4 */
+       666, /* GL_INT_VEC2 */
+       668, /* GL_INT_VEC3 */
+       670, /* GL_INT_VEC4 */
         94, /* GL_BOOL */
         96, /* GL_BOOL_VEC2 */
         98, /* GL_BOOL_VEC3 */
        100, /* GL_BOOL_VEC4 */
-       487, /* GL_FLOAT_MAT2 */
-       491, /* GL_FLOAT_MAT3 */
-       495, /* GL_FLOAT_MAT4 */
-      1392, /* GL_SAMPLER_1D */
-      1394, /* GL_SAMPLER_2D */
-      1396, /* GL_SAMPLER_3D */
-      1397, /* GL_SAMPLER_CUBE */
-      1393, /* GL_SAMPLER_1D_SHADOW */
-      1395, /* GL_SAMPLER_2D_SHADOW */
-       489, /* GL_FLOAT_MAT2x3 */
-       490, /* GL_FLOAT_MAT2x4 */
-       493, /* GL_FLOAT_MAT3x2 */
-       494, /* GL_FLOAT_MAT3x4 */
-       497, /* GL_FLOAT_MAT4x2 */
-       498, /* GL_FLOAT_MAT4x3 */
+       490, /* GL_FLOAT_MAT2 */
+       494, /* GL_FLOAT_MAT3 */
+       498, /* GL_FLOAT_MAT4 */
+      1405, /* GL_SAMPLER_1D */
+      1407, /* GL_SAMPLER_2D */
+      1409, /* GL_SAMPLER_3D */
+      1410, /* GL_SAMPLER_CUBE */
+      1406, /* GL_SAMPLER_1D_SHADOW */
+      1408, /* GL_SAMPLER_2D_SHADOW */
+       492, /* GL_FLOAT_MAT2x3 */
+       493, /* GL_FLOAT_MAT2x4 */
+       496, /* GL_FLOAT_MAT3x2 */
+       497, /* GL_FLOAT_MAT3x4 */
+       500, /* GL_FLOAT_MAT4x2 */
+       501, /* GL_FLOAT_MAT4x3 */
        345, /* GL_DELETE_STATUS */
        246, /* GL_COMPILE_STATUS */
-       717, /* GL_LINK_STATUS */
-      1806, /* GL_VALIDATE_STATUS */
-       644, /* GL_INFO_LOG_LENGTH */
+       724, /* GL_LINK_STATUS */
+      1825, /* GL_VALIDATE_STATUS */
+       651, /* GL_INFO_LOG_LENGTH */
         56, /* GL_ATTACHED_SHADERS */
         20, /* GL_ACTIVE_UNIFORMS */
         21, /* GL_ACTIVE_UNIFORM_MAX_LENGTH */
-      1434, /* GL_SHADER_SOURCE_LENGTH */
+      1447, /* GL_SHADER_SOURCE_LENGTH */
         15, /* GL_ACTIVE_ATTRIBUTES */
         16, /* GL_ACTIVE_ATTRIBUTE_MAX_LENGTH */
-       536, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-      1437, /* GL_SHADING_LANGUAGE_VERSION */
+       539, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+      1450, /* GL_SHADING_LANGUAGE_VERSION */
        322, /* GL_CURRENT_PROGRAM */
-      1108, /* GL_PALETTE4_RGB8_OES */
-      1110, /* GL_PALETTE4_RGBA8_OES */
-      1106, /* GL_PALETTE4_R5_G6_B5_OES */
-      1109, /* GL_PALETTE4_RGBA4_OES */
-      1107, /* GL_PALETTE4_RGB5_A1_OES */
-      1113, /* GL_PALETTE8_RGB8_OES */
-      1115, /* GL_PALETTE8_RGBA8_OES */
-      1111, /* GL_PALETTE8_R5_G6_B5_OES */
-      1114, /* GL_PALETTE8_RGBA4_OES */
-      1112, /* GL_PALETTE8_RGB5_A1_OES */
-       626, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-       625, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-      1791, /* GL_UNSIGNED_NORMALIZED */
-      1628, /* GL_TEXTURE_1D_ARRAY_EXT */
-      1279, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-      1630, /* GL_TEXTURE_2D_ARRAY_EXT */
-      1282, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-      1636, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-      1638, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-       886, /* GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_ARB */
-      1490, /* GL_SRGB */
-      1491, /* GL_SRGB8 */
-      1493, /* GL_SRGB_ALPHA */
-      1492, /* GL_SRGB8_ALPHA8 */
-      1450, /* GL_SLUMINANCE_ALPHA */
-      1449, /* GL_SLUMINANCE8_ALPHA8 */
-      1447, /* GL_SLUMINANCE */
-      1448, /* GL_SLUMINANCE8 */
+      1118, /* GL_PALETTE4_RGB8_OES */
+      1120, /* GL_PALETTE4_RGBA8_OES */
+      1116, /* GL_PALETTE4_R5_G6_B5_OES */
+      1119, /* GL_PALETTE4_RGBA4_OES */
+      1117, /* GL_PALETTE4_RGB5_A1_OES */
+      1123, /* GL_PALETTE8_RGB8_OES */
+      1125, /* GL_PALETTE8_RGBA8_OES */
+      1121, /* GL_PALETTE8_R5_G6_B5_OES */
+      1124, /* GL_PALETTE8_RGBA4_OES */
+      1122, /* GL_PALETTE8_RGB5_A1_OES */
+       633, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+       632, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+      1810, /* GL_UNSIGNED_NORMALIZED */
+      1645, /* GL_TEXTURE_1D_ARRAY_EXT */
+      1289, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+      1647, /* GL_TEXTURE_2D_ARRAY_EXT */
+      1292, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+      1653, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+      1655, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+       894, /* GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS_ARB */
+      1503, /* GL_SRGB */
+      1504, /* GL_SRGB8 */
+      1506, /* GL_SRGB_ALPHA */
+      1505, /* GL_SRGB8_ALPHA8 */
+      1463, /* GL_SLUMINANCE_ALPHA */
+      1462, /* GL_SLUMINANCE8_ALPHA8 */
+      1460, /* GL_SLUMINANCE */
+      1461, /* GL_SLUMINANCE8 */
        267, /* GL_COMPRESSED_SRGB */
        268, /* GL_COMPRESSED_SRGB_ALPHA */
        265, /* GL_COMPRESSED_SLUMINANCE */
        266, /* GL_COMPRESSED_SLUMINANCE_ALPHA */
-      1173, /* GL_POINT_SPRITE_COORD_ORIGIN */
-       725, /* GL_LOWER_LEFT */
-      1803, /* GL_UPPER_LEFT */
-      1513, /* GL_STENCIL_BACK_REF */
-      1514, /* GL_STENCIL_BACK_VALUE_MASK */
-      1515, /* GL_STENCIL_BACK_WRITEMASK */
-       442, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-      1326, /* GL_RENDERBUFFER_BINDING_EXT */
-      1307, /* GL_READ_FRAMEBUFFER */
-       441, /* GL_DRAW_FRAMEBUFFER */
-      1308, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-      1336, /* GL_RENDERBUFFER_SAMPLES */
-       547, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
-       545, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
-       556, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
-       552, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
-       554, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
-       559, /* GL_FRAMEBUFFER_COMPLETE */
-       563, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
-       571, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
-       567, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-       565, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-       568, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-       566, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-       574, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-       577, /* GL_FRAMEBUFFER_UNSUPPORTED */
-       575, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-       865, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+      1183, /* GL_POINT_SPRITE_COORD_ORIGIN */
+       732, /* GL_LOWER_LEFT */
+      1822, /* GL_UPPER_LEFT */
+      1526, /* GL_STENCIL_BACK_REF */
+      1527, /* GL_STENCIL_BACK_VALUE_MASK */
+      1528, /* GL_STENCIL_BACK_WRITEMASK */
+       444, /* GL_DRAW_FRAMEBUFFER_BINDING */
+      1337, /* GL_RENDERBUFFER_BINDING */
+      1317, /* GL_READ_FRAMEBUFFER */
+       443, /* GL_DRAW_FRAMEBUFFER */
+      1318, /* GL_READ_FRAMEBUFFER_BINDING */
+      1348, /* GL_RENDERBUFFER_SAMPLES */
+       550, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE */
+       548, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME */
+       559, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL */
+       555, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE */
+       557, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
+       563, /* GL_FRAMEBUFFER_COMPLETE */
+       567, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT */
+       576, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT */
+       572, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+       569, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+       573, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+       570, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER */
+       580, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER */
+       584, /* GL_FRAMEBUFFER_UNSUPPORTED */
+       582, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+       872, /* GL_MAX_COLOR_ATTACHMENTS */
        155, /* GL_COLOR_ATTACHMENT0 */
        157, /* GL_COLOR_ATTACHMENT1 */
        171, /* GL_COLOR_ATTACHMENT2 */
@@ -5107,69 +5145,69 @@ static const unsigned reduced_enums[1364] =
        164, /* GL_COLOR_ATTACHMENT13 */
        166, /* GL_COLOR_ATTACHMENT14 */
        168, /* GL_COLOR_ATTACHMENT15 */
-       348, /* GL_DEPTH_ATTACHMENT */
-      1503, /* GL_STENCIL_ATTACHMENT */
-       537, /* GL_FRAMEBUFFER */
-      1324, /* GL_RENDERBUFFER */
-      1338, /* GL_RENDERBUFFER_WIDTH */
-      1331, /* GL_RENDERBUFFER_HEIGHT */
-      1333, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
-      1526, /* GL_STENCIL_INDEX_EXT */
-      1523, /* GL_STENCIL_INDEX1_EXT */
-      1524, /* GL_STENCIL_INDEX4_EXT */
-      1525, /* GL_STENCIL_INDEX8_EXT */
-      1522, /* GL_STENCIL_INDEX16_EXT */
-      1335, /* GL_RENDERBUFFER_RED_SIZE */
-      1330, /* GL_RENDERBUFFER_GREEN_SIZE */
-      1327, /* GL_RENDERBUFFER_BLUE_SIZE */
-      1325, /* GL_RENDERBUFFER_ALPHA_SIZE */
-      1328, /* GL_RENDERBUFFER_DEPTH_SIZE */
-      1337, /* GL_RENDERBUFFER_STENCIL_SIZE */
-       573, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
-       926, /* GL_MAX_SAMPLES */
-       544, /* GL_FRAMEBUFFER_ATTACHMENT_LAYERED_ARB */
-       570, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_ARB */
-       569, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_ARB */
-       596, /* GL_GEOMETRY_SHADER_ARB */
-       597, /* GL_GEOMETRY_VERTICES_OUT_ARB */
-       594, /* GL_GEOMETRY_INPUT_TYPE_ARB */
-       595, /* GL_GEOMETRY_OUTPUT_TYPE_ARB */
-       889, /* GL_MAX_GEOMETRY_VARYING_COMPONENTS_ARB */
-       952, /* GL_MAX_VERTEX_VARYING_COMPONENTS_ARB */
-       888, /* GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_ARB */
-       885, /* GL_MAX_GEOMETRY_OUTPUT_VERTICES_ARB */
-       887, /* GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_ARB */
-      1293, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
-       483, /* GL_FIRST_VERTEX_CONVENTION */
-       674, /* GL_LAST_VERTEX_CONVENTION */
-      1271, /* GL_PROVOKING_VERTEX */
+       349, /* GL_DEPTH_ATTACHMENT */
+      1516, /* GL_STENCIL_ATTACHMENT */
+       540, /* GL_FRAMEBUFFER */
+      1335, /* GL_RENDERBUFFER */
+      1351, /* GL_RENDERBUFFER_WIDTH */
+      1343, /* GL_RENDERBUFFER_HEIGHT */
+      1345, /* GL_RENDERBUFFER_INTERNAL_FORMAT */
+      1543, /* GL_STENCIL_INDEX_EXT */
+      1535, /* GL_STENCIL_INDEX1 */
+      1539, /* GL_STENCIL_INDEX4 */
+      1541, /* GL_STENCIL_INDEX8 */
+      1536, /* GL_STENCIL_INDEX16 */
+      1347, /* GL_RENDERBUFFER_RED_SIZE */
+      1342, /* GL_RENDERBUFFER_GREEN_SIZE */
+      1339, /* GL_RENDERBUFFER_BLUE_SIZE */
+      1336, /* GL_RENDERBUFFER_ALPHA_SIZE */
+      1340, /* GL_RENDERBUFFER_DEPTH_SIZE */
+      1350, /* GL_RENDERBUFFER_STENCIL_SIZE */
+       578, /* GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE */
+       935, /* GL_MAX_SAMPLES */
+       547, /* GL_FRAMEBUFFER_ATTACHMENT_LAYERED_ARB */
+       575, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS_ARB */
+       574, /* GL_FRAMEBUFFER_INCOMPLETE_LAYER_COUNT_ARB */
+       603, /* GL_GEOMETRY_SHADER_ARB */
+       604, /* GL_GEOMETRY_VERTICES_OUT_ARB */
+       601, /* GL_GEOMETRY_INPUT_TYPE_ARB */
+       602, /* GL_GEOMETRY_OUTPUT_TYPE_ARB */
+       897, /* GL_MAX_GEOMETRY_VARYING_COMPONENTS_ARB */
+       962, /* GL_MAX_VERTEX_VARYING_COMPONENTS_ARB */
+       896, /* GL_MAX_GEOMETRY_UNIFORM_COMPONENTS_ARB */
+       893, /* GL_MAX_GEOMETRY_OUTPUT_VERTICES_ARB */
+       895, /* GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS_ARB */
+      1303, /* GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
+       486, /* GL_FIRST_VERTEX_CONVENTION */
+       681, /* GL_LAST_VERTEX_CONVENTION */
+      1281, /* GL_PROVOKING_VERTEX */
        302, /* GL_COPY_READ_BUFFER */
        303, /* GL_COPY_WRITE_BUFFER */
-      1385, /* GL_RGBA_SNORM */
-      1381, /* GL_RGBA8_SNORM */
-      1443, /* GL_SIGNED_NORMALIZED */
-       927, /* GL_MAX_SERVER_WAIT_TIMEOUT */
-      1058, /* GL_OBJECT_TYPE */
-      1547, /* GL_SYNC_CONDITION */
-      1552, /* GL_SYNC_STATUS */
-      1549, /* GL_SYNC_FLAGS */
-      1548, /* GL_SYNC_FENCE */
-      1551, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
-      1780, /* GL_UNSIGNALED */
-      1442, /* GL_SIGNALED */
+      1398, /* GL_RGBA_SNORM */
+      1394, /* GL_RGBA8_SNORM */
+      1456, /* GL_SIGNED_NORMALIZED */
+       937, /* GL_MAX_SERVER_WAIT_TIMEOUT */
+      1068, /* GL_OBJECT_TYPE */
+      1564, /* GL_SYNC_CONDITION */
+      1569, /* GL_SYNC_STATUS */
+      1566, /* GL_SYNC_FLAGS */
+      1565, /* GL_SYNC_FENCE */
+      1568, /* GL_SYNC_GPU_COMMANDS_COMPLETE */
+      1798, /* GL_UNSIGNALED */
+      1455, /* GL_SIGNALED */
         46, /* GL_ALREADY_SIGNALED */
-      1749, /* GL_TIMEOUT_EXPIRED */
+      1767, /* GL_TIMEOUT_EXPIRED */
        270, /* GL_CONDITION_SATISFIED */
-      1863, /* GL_WAIT_FAILED */
-       468, /* GL_EVAL_BIT */
-      1305, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-       719, /* GL_LIST_BIT */
-      1644, /* GL_TEXTURE_BIT */
-      1416, /* GL_SCISSOR_BIT */
+      1882, /* GL_WAIT_FAILED */
+       471, /* GL_EVAL_BIT */
+      1315, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+       726, /* GL_LIST_BIT */
+      1661, /* GL_TEXTURE_BIT */
+      1429, /* GL_SCISSOR_BIT */
         29, /* GL_ALL_ATTRIB_BITS */
-      1014, /* GL_MULTISAMPLE_BIT */
+      1024, /* GL_MULTISAMPLE_BIT */
         30, /* GL_ALL_CLIENT_ATTRIB_BITS */
-      1750, /* GL_TIMEOUT_IGNORED */
+      1768, /* GL_TIMEOUT_IGNORED */
 };
 
 typedef int (*cfunc)(const void *, const void *);
diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
index fabc6cb27c..062fa2583c 100644
--- a/src/mesa/main/extensions.c
+++ b/src/mesa/main/extensions.c
@@ -105,8 +105,9 @@ static const struct {
    { ON,  "GL_EXT_copy_texture",               F(EXT_copy_texture) },
    { OFF, "GL_EXT_depth_bounds_test",          F(EXT_depth_bounds_test) },
    { ON,  "GL_EXT_draw_range_elements",        F(EXT_draw_range_elements) },
-   { OFF, "GL_EXT_framebuffer_object",         F(EXT_framebuffer_object) },
    { OFF, "GL_EXT_framebuffer_blit",           F(EXT_framebuffer_blit) },
+   { OFF, "GL_EXT_framebuffer_multisample",    F(EXT_framebuffer_multisample) },
+   { OFF, "GL_EXT_framebuffer_object",         F(EXT_framebuffer_object) },
    { OFF, "GL_EXT_fog_coord",                  F(EXT_fog_coord) },
    { OFF, "GL_EXT_gpu_program_parameters",     F(EXT_gpu_program_parameters) },
    { OFF, "GL_EXT_histogram",                  F(EXT_histogram) },
@@ -170,6 +171,7 @@ static const struct {
    { OFF, "GL_NV_fragment_program",            F(NV_fragment_program) },
    { OFF, "GL_NV_fragment_program_option",     F(NV_fragment_program_option) },
    { ON,  "GL_NV_light_max_exponent",          F(NV_light_max_exponent) },
+   { OFF, "GL_NV_packed_depth_stencil",        F(EXT_packed_depth_stencil) },
    { OFF, "GL_NV_point_sprite",                F(NV_point_sprite) },
    { OFF, "GL_NV_texture_env_combine4",        F(NV_texture_env_combine4) },
    { OFF, "GL_NV_texture_rectangle",           F(NV_texture_rectangle) },
@@ -278,6 +280,9 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
 #if FEATURE_EXT_framebuffer_blit
    ctx->Extensions.EXT_framebuffer_blit = GL_TRUE;
 #endif
+#if FEATURE_ARB_framebuffer_object
+   ctx->Extensions.EXT_framebuffer_multisample = GL_TRUE;
+#endif
    ctx->Extensions.EXT_histogram = GL_TRUE;
    /*ctx->Extensions.EXT_multi_draw_arrays = GL_TRUE;*/
    ctx->Extensions.EXT_packed_depth_stencil = GL_TRUE;
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 356476e35a..5cfa898031 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -438,7 +438,7 @@ static struct ureg register_input( struct tnl_program *p, GLuint input )
  */
 static struct ureg register_output( struct tnl_program *p, GLuint output )
 {
-   p->program->Base.OutputsWritten |= (1<<output);
+   p->program->Base.OutputsWritten |= BITFIELD64_BIT(output);
    return make_ureg(PROGRAM_OUTPUT, output);
 }
 
@@ -523,7 +523,6 @@ static void emit_dst( struct prog_dst_register *dst,
    dst->CondMask = COND_TR;  /* always pass cond test */
    dst->CondSwizzle = SWIZZLE_NOOP;
    dst->CondSrc = 0;
-   dst->pad = 0;
    /* Check that bitfield sizes aren't exceeded */
    ASSERT(dst->Index == reg.idx);
 }
diff --git a/src/mesa/main/formats.c b/src/mesa/main/formats.c
index 7d0e2d2914..329b795074 100644
--- a/src/mesa/main/formats.c
+++ b/src/mesa/main/formats.c
@@ -133,6 +133,15 @@ static struct gl_format_info format_info[MESA_FORMAT_COUNT] =
       1, 1, 4                      /* BlockWidth/Height,Bytes */
    },
    {
+      MESA_FORMAT_XRGB8888_REV,    /* Name */
+      "MESA_FORMAT_XRGB8888_REV",  /* StrName */
+      GL_RGB,                      /* BaseFormat */
+      GL_UNSIGNED_NORMALIZED,      /* DataType */
+      8, 8, 8, 0,                  /* Red/Green/Blue/AlphaBits */
+      0, 0, 0, 0, 0,               /* Lum/Int/Index/Depth/StencilBits */
+      1, 1, 4                      /* BlockWidth/Height,Bytes */
+   },
+   {
       MESA_FORMAT_RGB888,          /* Name */
       "MESA_FORMAT_RGB888",        /* StrName */
       GL_RGB,                      /* BaseFormat */
@@ -232,6 +241,24 @@ static struct gl_format_info format_info[MESA_FORMAT_COUNT] =
       1, 1, 2                      /* BlockWidth/Height,Bytes */
    },
    {
+      MESA_FORMAT_AL1616,          /* Name */
+      "MESA_FORMAT_AL1616",        /* StrName */
+      GL_LUMINANCE_ALPHA,          /* BaseFormat */
+      GL_UNSIGNED_NORMALIZED,      /* DataType */
+      0, 0, 0, 16,                 /* Red/Green/Blue/AlphaBits */
+      16, 0, 0, 0, 0,              /* Lum/Int/Index/Depth/StencilBits */
+      1, 1, 4                      /* BlockWidth/Height,Bytes */
+   },
+   {
+      MESA_FORMAT_AL1616_REV,      /* Name */
+      "MESA_FORMAT_AL1616_REV",    /* StrName */
+      GL_LUMINANCE_ALPHA,          /* BaseFormat */
+      GL_UNSIGNED_NORMALIZED,      /* DataType */
+      0, 0, 0, 16,                 /* Red/Green/Blue/AlphaBits */
+      16, 0, 0, 0, 0,              /* Lum/Int/Index/Depth/StencilBits */
+      1, 1, 4                      /* BlockWidth/Height,Bytes */
+   },
+   {
       MESA_FORMAT_RGB332,          /* Name */
       "MESA_FORMAT_RGB332",        /* StrName */
       GL_RGB,                      /* BaseFormat */
@@ -974,6 +1001,13 @@ _mesa_format_to_type_and_comps(gl_format format,
       *datatype = GL_UNSIGNED_BYTE;
       *comps = 2;
       return;
+
+   case MESA_FORMAT_AL1616:
+   case MESA_FORMAT_AL1616_REV:
+      *datatype = GL_UNSIGNED_SHORT;
+      *comps = 2;
+      return;
+
    case MESA_FORMAT_RGB332:
       *datatype = GL_UNSIGNED_BYTE_3_3_2;
       *comps = 3;
diff --git a/src/mesa/main/formats.h b/src/mesa/main/formats.h
index fa6359f5c8..0eeeb8b2ba 100644
--- a/src/mesa/main/formats.h
+++ b/src/mesa/main/formats.h
@@ -55,6 +55,7 @@ typedef enum
    MESA_FORMAT_ARGB8888,	/* AAAA AAAA RRRR RRRR GGGG GGGG BBBB BBBB */
    MESA_FORMAT_ARGB8888_REV,	/* BBBB BBBB GGGG GGGG RRRR RRRR AAAA AAAA */
    MESA_FORMAT_XRGB8888,	/* xxxx xxxx RRRR RRRR GGGG GGGG BBBB BBBB */
+   MESA_FORMAT_XRGB8888_REV,	/* BBBB BBBB GGGG GGGG RRRR RRRR xxxx xxxx */
    MESA_FORMAT_RGB888,		/*           RRRR RRRR GGGG GGGG BBBB BBBB */
    MESA_FORMAT_BGR888,		/*           BBBB BBBB GGGG GGGG RRRR RRRR */
    MESA_FORMAT_RGB565,		/*                     RRRR RGGG GGGB BBBB */
@@ -66,6 +67,8 @@ typedef enum
    MESA_FORMAT_ARGB1555_REV,	/*                     GGGB BBBB ARRR RRGG */
    MESA_FORMAT_AL88,		/*                     AAAA AAAA LLLL LLLL */
    MESA_FORMAT_AL88_REV,	/*                     LLLL LLLL AAAA AAAA */
+   MESA_FORMAT_AL1616,          /* AAAA AAAA AAAA AAAA LLLL LLLL LLLL LLLL */
+   MESA_FORMAT_AL1616_REV,      /* LLLL LLLL LLLL LLLL AAAA AAAA AAAA AAAA */
    MESA_FORMAT_RGB332,		/*                               RRRG GGBB */
    MESA_FORMAT_A8,		/*                               AAAA AAAA */
    MESA_FORMAT_L8,		/*                               LLLL LLLL */
@@ -137,7 +140,7 @@ typedef enum
    MESA_FORMAT_SIGNED_RGBA_16,
    /*@}*/
 
-   MESA_FORMAT_COUNT,
+   MESA_FORMAT_COUNT
 } gl_format;
 
 
diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c
index 46ffb929b6..c9e00cf752 100644
--- a/src/mesa/main/imports.c
+++ b/src/mesa/main/imports.c
@@ -629,11 +629,15 @@ _mesa_ffsll(int64_t val)
 unsigned int
 _mesa_bitcount(unsigned int n)
 {
+#if defined(__GNUC__)
+   return __builtin_popcount(n);
+#else
    unsigned int bits;
    for (bits = 0; n > 0; n = n >> 1) {
       bits += (n & 1);
    }
    return bits;
+#endif
 }
 
 
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index ccf21e97f4..19db4f5d30 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -86,6 +86,31 @@
 
 
 /**
+ * \name 64-bit extension of GLbitfield.
+ */
+/*@{*/
+typedef GLuint64 GLbitfield64;
+
+#define BITFIELD64_ONE         1ULL
+#define BITFIELD64_ALLONES     ~0ULL
+
+/** Set a single bit */
+#define BITFIELD64_BIT(b)      (BITFIELD64_ONE << (b))
+
+/** Set a mask of the least significant \c b bits */
+#define BITFIELD64_MASK(b)     (((b) >= 64) ? BITFIELD64_ALLONES : \
+				(BITFIELD64_BIT(b) - 1))
+
+/**
+ * Set all bits from l (low bit) to h (high bit), inclusive.
+ *
+ * \note \C BITFIELD_64_RANGE(0, 63) return 64 set bits.
+ */
+#define BITFIELD64_RANGE(l, h) (BITFIELD64_MASK((h) + 1) & ~BITFIELD64_MASK(l))
+/*@}*/
+
+
+/**
  * \name Some forward type declarations
  */
 /*@{*/
@@ -1751,7 +1776,7 @@ struct gl_program
    struct prog_instruction *Instructions;
 
    GLbitfield InputsRead;     /**< Bitmask of which input regs are read */
-   GLbitfield OutputsWritten; /**< Bitmask of which output regs are written to */
+   GLbitfield64 OutputsWritten; /**< Bitmask of which output regs are written */
    GLbitfield InputFlags[MAX_PROGRAM_INPUTS];   /**< PROG_PARAM_BIT_x flags */
    GLbitfield OutputFlags[MAX_PROGRAM_OUTPUTS]; /**< PROG_PARAM_BIT_x flags */
    GLbitfield TexturesUsed[MAX_TEXTURE_UNITS];  /**< TEXTURE_x_BIT bitmask */
@@ -2540,9 +2565,10 @@ struct gl_extensions
    GLboolean EXT_copy_texture;
    GLboolean EXT_depth_bounds_test;
    GLboolean EXT_draw_range_elements;
-   GLboolean EXT_framebuffer_object;
    GLboolean EXT_fog_coord;
    GLboolean EXT_framebuffer_blit;
+   GLboolean EXT_framebuffer_multisample;
+   GLboolean EXT_framebuffer_object;
    GLboolean EXT_gpu_program_parameters;
    GLboolean EXT_histogram;
    GLboolean EXT_multi_draw_arrays;
diff --git a/src/mesa/main/remap_helper.h b/src/mesa/main/remap_helper.h
index afd3103a4f..40ed4b27d7 100644
--- a/src/mesa/main/remap_helper.h
+++ b/src/mesa/main/remap_helper.h
@@ -1724,2618 +1724,2619 @@ static const char _mesa_function_pool[] =
    /* _mesa_function_pool[11550]: RenderbufferStorageMultisample (will be remapped) */
    "iiiii\0"
    "glRenderbufferStorageMultisample\0"
+   "glRenderbufferStorageMultisampleEXT\0"
    "\0"
-   /* _mesa_function_pool[11590]: TexCoord4fv (offset 121) */
+   /* _mesa_function_pool[11626]: TexCoord4fv (offset 121) */
    "p\0"
    "glTexCoord4fv\0"
    "\0"
-   /* _mesa_function_pool[11607]: Tangent3sEXT (dynamic) */
+   /* _mesa_function_pool[11643]: Tangent3sEXT (dynamic) */
    "iii\0"
    "glTangent3sEXT\0"
    "\0"
-   /* _mesa_function_pool[11627]: GlobalAlphaFactorfSUN (dynamic) */
+   /* _mesa_function_pool[11663]: GlobalAlphaFactorfSUN (dynamic) */
    "f\0"
    "glGlobalAlphaFactorfSUN\0"
    "\0"
-   /* _mesa_function_pool[11654]: MultiTexCoord3iARB (offset 396) */
+   /* _mesa_function_pool[11690]: MultiTexCoord3iARB (offset 396) */
    "iiii\0"
    "glMultiTexCoord3i\0"
    "glMultiTexCoord3iARB\0"
    "\0"
-   /* _mesa_function_pool[11699]: IsProgram (will be remapped) */
+   /* _mesa_function_pool[11735]: IsProgram (will be remapped) */
    "i\0"
    "glIsProgram\0"
    "\0"
-   /* _mesa_function_pool[11714]: TexCoordPointerListIBM (dynamic) */
+   /* _mesa_function_pool[11750]: TexCoordPointerListIBM (dynamic) */
    "iiipi\0"
    "glTexCoordPointerListIBM\0"
    "\0"
-   /* _mesa_function_pool[11746]: GlobalAlphaFactorusSUN (dynamic) */
+   /* _mesa_function_pool[11782]: GlobalAlphaFactorusSUN (dynamic) */
    "i\0"
    "glGlobalAlphaFactorusSUN\0"
    "\0"
-   /* _mesa_function_pool[11774]: VertexAttrib2dvNV (will be remapped) */
+   /* _mesa_function_pool[11810]: VertexAttrib2dvNV (will be remapped) */
    "ip\0"
    "glVertexAttrib2dvNV\0"
    "\0"
-   /* _mesa_function_pool[11798]: FramebufferRenderbufferEXT (will be remapped) */
+   /* _mesa_function_pool[11834]: FramebufferRenderbufferEXT (will be remapped) */
    "iiii\0"
    "glFramebufferRenderbuffer\0"
    "glFramebufferRenderbufferEXT\0"
    "\0"
-   /* _mesa_function_pool[11859]: VertexAttrib1dvNV (will be remapped) */
+   /* _mesa_function_pool[11895]: VertexAttrib1dvNV (will be remapped) */
    "ip\0"
    "glVertexAttrib1dvNV\0"
    "\0"
-   /* _mesa_function_pool[11883]: GenTextures (offset 328) */
+   /* _mesa_function_pool[11919]: GenTextures (offset 328) */
    "ip\0"
    "glGenTextures\0"
    "glGenTexturesEXT\0"
    "\0"
-   /* _mesa_function_pool[11918]: FramebufferTextureARB (will be remapped) */
+   /* _mesa_function_pool[11954]: FramebufferTextureARB (will be remapped) */
    "iiii\0"
    "glFramebufferTextureARB\0"
    "\0"
-   /* _mesa_function_pool[11948]: SetFenceNV (will be remapped) */
+   /* _mesa_function_pool[11984]: SetFenceNV (will be remapped) */
    "ii\0"
    "glSetFenceNV\0"
    "\0"
-   /* _mesa_function_pool[11965]: FramebufferTexture1DEXT (will be remapped) */
+   /* _mesa_function_pool[12001]: FramebufferTexture1DEXT (will be remapped) */
    "iiiii\0"
    "glFramebufferTexture1D\0"
    "glFramebufferTexture1DEXT\0"
    "\0"
-   /* _mesa_function_pool[12021]: GetCombinerOutputParameterivNV (will be remapped) */
+   /* _mesa_function_pool[12057]: GetCombinerOutputParameterivNV (will be remapped) */
    "iiip\0"
    "glGetCombinerOutputParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[12060]: MultiModeDrawArraysIBM (will be remapped) */
+   /* _mesa_function_pool[12096]: MultiModeDrawArraysIBM (will be remapped) */
    "pppii\0"
    "glMultiModeDrawArraysIBM\0"
    "\0"
-   /* _mesa_function_pool[12092]: PixelTexGenParameterivSGIS (will be remapped) */
+   /* _mesa_function_pool[12128]: PixelTexGenParameterivSGIS (will be remapped) */
    "ip\0"
    "glPixelTexGenParameterivSGIS\0"
    "\0"
-   /* _mesa_function_pool[12125]: TextureNormalEXT (dynamic) */
+   /* _mesa_function_pool[12161]: TextureNormalEXT (dynamic) */
    "i\0"
    "glTextureNormalEXT\0"
    "\0"
-   /* _mesa_function_pool[12147]: IndexPointerListIBM (dynamic) */
+   /* _mesa_function_pool[12183]: IndexPointerListIBM (dynamic) */
    "iipi\0"
    "glIndexPointerListIBM\0"
    "\0"
-   /* _mesa_function_pool[12175]: WeightfvARB (dynamic) */
+   /* _mesa_function_pool[12211]: WeightfvARB (dynamic) */
    "ip\0"
    "glWeightfvARB\0"
    "\0"
-   /* _mesa_function_pool[12193]: RasterPos2sv (offset 69) */
+   /* _mesa_function_pool[12229]: RasterPos2sv (offset 69) */
    "p\0"
    "glRasterPos2sv\0"
    "\0"
-   /* _mesa_function_pool[12211]: Color4ubv (offset 36) */
+   /* _mesa_function_pool[12247]: Color4ubv (offset 36) */
    "p\0"
    "glColor4ubv\0"
    "\0"
-   /* _mesa_function_pool[12226]: DrawBuffer (offset 202) */
+   /* _mesa_function_pool[12262]: DrawBuffer (offset 202) */
    "i\0"
    "glDrawBuffer\0"
    "\0"
-   /* _mesa_function_pool[12242]: TexCoord2fv (offset 105) */
+   /* _mesa_function_pool[12278]: TexCoord2fv (offset 105) */
    "p\0"
    "glTexCoord2fv\0"
    "\0"
-   /* _mesa_function_pool[12259]: WindowPos4fMESA (will be remapped) */
+   /* _mesa_function_pool[12295]: WindowPos4fMESA (will be remapped) */
    "ffff\0"
    "glWindowPos4fMESA\0"
    "\0"
-   /* _mesa_function_pool[12283]: TexCoord1sv (offset 101) */
+   /* _mesa_function_pool[12319]: TexCoord1sv (offset 101) */
    "p\0"
    "glTexCoord1sv\0"
    "\0"
-   /* _mesa_function_pool[12300]: WindowPos3dvMESA (will be remapped) */
+   /* _mesa_function_pool[12336]: WindowPos3dvMESA (will be remapped) */
    "p\0"
    "glWindowPos3dv\0"
    "glWindowPos3dvARB\0"
    "glWindowPos3dvMESA\0"
    "\0"
-   /* _mesa_function_pool[12355]: DepthFunc (offset 245) */
+   /* _mesa_function_pool[12391]: DepthFunc (offset 245) */
    "i\0"
    "glDepthFunc\0"
    "\0"
-   /* _mesa_function_pool[12370]: PixelMapusv (offset 253) */
+   /* _mesa_function_pool[12406]: PixelMapusv (offset 253) */
    "iip\0"
    "glPixelMapusv\0"
    "\0"
-   /* _mesa_function_pool[12389]: GetQueryObjecti64vEXT (will be remapped) */
+   /* _mesa_function_pool[12425]: GetQueryObjecti64vEXT (will be remapped) */
    "iip\0"
    "glGetQueryObjecti64vEXT\0"
    "\0"
-   /* _mesa_function_pool[12418]: MultiTexCoord1dARB (offset 376) */
+   /* _mesa_function_pool[12454]: MultiTexCoord1dARB (offset 376) */
    "id\0"
    "glMultiTexCoord1d\0"
    "glMultiTexCoord1dARB\0"
    "\0"
-   /* _mesa_function_pool[12461]: PointParameterivNV (will be remapped) */
+   /* _mesa_function_pool[12497]: PointParameterivNV (will be remapped) */
    "ip\0"
    "glPointParameteriv\0"
    "glPointParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[12505]: BlendFunc (offset 241) */
+   /* _mesa_function_pool[12541]: BlendFunc (offset 241) */
    "ii\0"
    "glBlendFunc\0"
    "\0"
-   /* _mesa_function_pool[12521]: Uniform2fvARB (will be remapped) */
+   /* _mesa_function_pool[12557]: Uniform2fvARB (will be remapped) */
    "iip\0"
    "glUniform2fv\0"
    "glUniform2fvARB\0"
    "\0"
-   /* _mesa_function_pool[12555]: BufferParameteriAPPLE (will be remapped) */
+   /* _mesa_function_pool[12591]: BufferParameteriAPPLE (will be remapped) */
    "iii\0"
    "glBufferParameteriAPPLE\0"
    "\0"
-   /* _mesa_function_pool[12584]: MultiTexCoord3dvARB (offset 393) */
+   /* _mesa_function_pool[12620]: MultiTexCoord3dvARB (offset 393) */
    "ip\0"
    "glMultiTexCoord3dv\0"
    "glMultiTexCoord3dvARB\0"
    "\0"
-   /* _mesa_function_pool[12629]: ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[12665]: ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN (dynamic) */
    "pppp\0"
    "glReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[12685]: DeleteObjectARB (will be remapped) */
+   /* _mesa_function_pool[12721]: DeleteObjectARB (will be remapped) */
    "i\0"
    "glDeleteObjectARB\0"
    "\0"
-   /* _mesa_function_pool[12706]: MatrixIndexPointerARB (dynamic) */
+   /* _mesa_function_pool[12742]: MatrixIndexPointerARB (dynamic) */
    "iiip\0"
    "glMatrixIndexPointerARB\0"
    "\0"
-   /* _mesa_function_pool[12736]: ProgramNamedParameter4dvNV (will be remapped) */
+   /* _mesa_function_pool[12772]: ProgramNamedParameter4dvNV (will be remapped) */
    "iipp\0"
    "glProgramNamedParameter4dvNV\0"
    "\0"
-   /* _mesa_function_pool[12771]: Tangent3fvEXT (dynamic) */
+   /* _mesa_function_pool[12807]: Tangent3fvEXT (dynamic) */
    "p\0"
    "glTangent3fvEXT\0"
    "\0"
-   /* _mesa_function_pool[12790]: Flush (offset 217) */
+   /* _mesa_function_pool[12826]: Flush (offset 217) */
    "\0"
    "glFlush\0"
    "\0"
-   /* _mesa_function_pool[12800]: Color4uiv (offset 38) */
+   /* _mesa_function_pool[12836]: Color4uiv (offset 38) */
    "p\0"
    "glColor4uiv\0"
    "\0"
-   /* _mesa_function_pool[12815]: GenVertexArrays (will be remapped) */
+   /* _mesa_function_pool[12851]: GenVertexArrays (will be remapped) */
    "ip\0"
    "glGenVertexArrays\0"
    "\0"
-   /* _mesa_function_pool[12837]: RasterPos3sv (offset 77) */
+   /* _mesa_function_pool[12873]: RasterPos3sv (offset 77) */
    "p\0"
    "glRasterPos3sv\0"
    "\0"
-   /* _mesa_function_pool[12855]: BindFramebufferEXT (will be remapped) */
+   /* _mesa_function_pool[12891]: BindFramebufferEXT (will be remapped) */
    "ii\0"
    "glBindFramebuffer\0"
    "glBindFramebufferEXT\0"
    "\0"
-   /* _mesa_function_pool[12898]: ReferencePlaneSGIX (dynamic) */
+   /* _mesa_function_pool[12934]: ReferencePlaneSGIX (dynamic) */
    "p\0"
    "glReferencePlaneSGIX\0"
    "\0"
-   /* _mesa_function_pool[12922]: PushAttrib (offset 219) */
+   /* _mesa_function_pool[12958]: PushAttrib (offset 219) */
    "i\0"
    "glPushAttrib\0"
    "\0"
-   /* _mesa_function_pool[12938]: RasterPos2i (offset 66) */
+   /* _mesa_function_pool[12974]: RasterPos2i (offset 66) */
    "ii\0"
    "glRasterPos2i\0"
    "\0"
-   /* _mesa_function_pool[12956]: ValidateProgramARB (will be remapped) */
+   /* _mesa_function_pool[12992]: ValidateProgramARB (will be remapped) */
    "i\0"
    "glValidateProgram\0"
    "glValidateProgramARB\0"
    "\0"
-   /* _mesa_function_pool[12998]: TexParameteriv (offset 181) */
+   /* _mesa_function_pool[13034]: TexParameteriv (offset 181) */
    "iip\0"
    "glTexParameteriv\0"
    "\0"
-   /* _mesa_function_pool[13020]: UnlockArraysEXT (will be remapped) */
+   /* _mesa_function_pool[13056]: UnlockArraysEXT (will be remapped) */
    "\0"
    "glUnlockArraysEXT\0"
    "\0"
-   /* _mesa_function_pool[13040]: TexCoord2fColor3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[13076]: TexCoord2fColor3fVertex3fSUN (dynamic) */
    "ffffffff\0"
    "glTexCoord2fColor3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[13081]: WindowPos3fvMESA (will be remapped) */
+   /* _mesa_function_pool[13117]: WindowPos3fvMESA (will be remapped) */
    "p\0"
    "glWindowPos3fv\0"
    "glWindowPos3fvARB\0"
    "glWindowPos3fvMESA\0"
    "\0"
-   /* _mesa_function_pool[13136]: RasterPos2f (offset 64) */
+   /* _mesa_function_pool[13172]: RasterPos2f (offset 64) */
    "ff\0"
    "glRasterPos2f\0"
    "\0"
-   /* _mesa_function_pool[13154]: VertexAttrib1svNV (will be remapped) */
+   /* _mesa_function_pool[13190]: VertexAttrib1svNV (will be remapped) */
    "ip\0"
    "glVertexAttrib1svNV\0"
    "\0"
-   /* _mesa_function_pool[13178]: RasterPos2d (offset 62) */
+   /* _mesa_function_pool[13214]: RasterPos2d (offset 62) */
    "dd\0"
    "glRasterPos2d\0"
    "\0"
-   /* _mesa_function_pool[13196]: RasterPos3fv (offset 73) */
+   /* _mesa_function_pool[13232]: RasterPos3fv (offset 73) */
    "p\0"
    "glRasterPos3fv\0"
    "\0"
-   /* _mesa_function_pool[13214]: CopyTexSubImage3D (offset 373) */
+   /* _mesa_function_pool[13250]: CopyTexSubImage3D (offset 373) */
    "iiiiiiiii\0"
    "glCopyTexSubImage3D\0"
    "glCopyTexSubImage3DEXT\0"
    "\0"
-   /* _mesa_function_pool[13268]: VertexAttrib2dARB (will be remapped) */
+   /* _mesa_function_pool[13304]: VertexAttrib2dARB (will be remapped) */
    "idd\0"
    "glVertexAttrib2d\0"
    "glVertexAttrib2dARB\0"
    "\0"
-   /* _mesa_function_pool[13310]: Color4ub (offset 35) */
+   /* _mesa_function_pool[13346]: Color4ub (offset 35) */
    "iiii\0"
    "glColor4ub\0"
    "\0"
-   /* _mesa_function_pool[13327]: GetInteger64v (will be remapped) */
+   /* _mesa_function_pool[13363]: GetInteger64v (will be remapped) */
    "ip\0"
    "glGetInteger64v\0"
    "\0"
-   /* _mesa_function_pool[13347]: TextureColorMaskSGIS (dynamic) */
+   /* _mesa_function_pool[13383]: TextureColorMaskSGIS (dynamic) */
    "iiii\0"
    "glTextureColorMaskSGIS\0"
    "\0"
-   /* _mesa_function_pool[13376]: RasterPos2s (offset 68) */
+   /* _mesa_function_pool[13412]: RasterPos2s (offset 68) */
    "ii\0"
    "glRasterPos2s\0"
    "\0"
-   /* _mesa_function_pool[13394]: GetColorTable (offset 343) */
+   /* _mesa_function_pool[13430]: GetColorTable (offset 343) */
    "iiip\0"
    "glGetColorTable\0"
    "glGetColorTableSGI\0"
    "glGetColorTableEXT\0"
    "\0"
-   /* _mesa_function_pool[13454]: SelectBuffer (offset 195) */
+   /* _mesa_function_pool[13490]: SelectBuffer (offset 195) */
    "ip\0"
    "glSelectBuffer\0"
    "\0"
-   /* _mesa_function_pool[13473]: Indexiv (offset 49) */
+   /* _mesa_function_pool[13509]: Indexiv (offset 49) */
    "p\0"
    "glIndexiv\0"
    "\0"
-   /* _mesa_function_pool[13486]: TexCoord3i (offset 114) */
+   /* _mesa_function_pool[13522]: TexCoord3i (offset 114) */
    "iii\0"
    "glTexCoord3i\0"
    "\0"
-   /* _mesa_function_pool[13504]: CopyColorTable (offset 342) */
+   /* _mesa_function_pool[13540]: CopyColorTable (offset 342) */
    "iiiii\0"
    "glCopyColorTable\0"
    "glCopyColorTableSGI\0"
    "\0"
-   /* _mesa_function_pool[13548]: GetHistogramParameterfv (offset 362) */
+   /* _mesa_function_pool[13584]: GetHistogramParameterfv (offset 362) */
    "iip\0"
    "glGetHistogramParameterfv\0"
    "glGetHistogramParameterfvEXT\0"
    "\0"
-   /* _mesa_function_pool[13608]: Frustum (offset 289) */
+   /* _mesa_function_pool[13644]: Frustum (offset 289) */
    "dddddd\0"
    "glFrustum\0"
    "\0"
-   /* _mesa_function_pool[13626]: GetString (offset 275) */
+   /* _mesa_function_pool[13662]: GetString (offset 275) */
    "i\0"
    "glGetString\0"
    "\0"
-   /* _mesa_function_pool[13641]: ColorPointervINTEL (dynamic) */
+   /* _mesa_function_pool[13677]: ColorPointervINTEL (dynamic) */
    "iip\0"
    "glColorPointervINTEL\0"
    "\0"
-   /* _mesa_function_pool[13667]: TexEnvf (offset 184) */
+   /* _mesa_function_pool[13703]: TexEnvf (offset 184) */
    "iif\0"
    "glTexEnvf\0"
    "\0"
-   /* _mesa_function_pool[13682]: TexCoord3d (offset 110) */
+   /* _mesa_function_pool[13718]: TexCoord3d (offset 110) */
    "ddd\0"
    "glTexCoord3d\0"
    "\0"
-   /* _mesa_function_pool[13700]: AlphaFragmentOp1ATI (will be remapped) */
+   /* _mesa_function_pool[13736]: AlphaFragmentOp1ATI (will be remapped) */
    "iiiiii\0"
    "glAlphaFragmentOp1ATI\0"
    "\0"
-   /* _mesa_function_pool[13730]: TexCoord3f (offset 112) */
+   /* _mesa_function_pool[13766]: TexCoord3f (offset 112) */
    "fff\0"
    "glTexCoord3f\0"
    "\0"
-   /* _mesa_function_pool[13748]: MultiTexCoord3ivARB (offset 397) */
+   /* _mesa_function_pool[13784]: MultiTexCoord3ivARB (offset 397) */
    "ip\0"
    "glMultiTexCoord3iv\0"
    "glMultiTexCoord3ivARB\0"
    "\0"
-   /* _mesa_function_pool[13793]: MultiTexCoord2sARB (offset 390) */
+   /* _mesa_function_pool[13829]: MultiTexCoord2sARB (offset 390) */
    "iii\0"
    "glMultiTexCoord2s\0"
    "glMultiTexCoord2sARB\0"
    "\0"
-   /* _mesa_function_pool[13837]: VertexAttrib1dvARB (will be remapped) */
+   /* _mesa_function_pool[13873]: VertexAttrib1dvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib1dv\0"
    "glVertexAttrib1dvARB\0"
    "\0"
-   /* _mesa_function_pool[13880]: DeleteTextures (offset 327) */
+   /* _mesa_function_pool[13916]: DeleteTextures (offset 327) */
    "ip\0"
    "glDeleteTextures\0"
    "glDeleteTexturesEXT\0"
    "\0"
-   /* _mesa_function_pool[13921]: TexCoordPointerEXT (will be remapped) */
+   /* _mesa_function_pool[13957]: TexCoordPointerEXT (will be remapped) */
    "iiiip\0"
    "glTexCoordPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[13949]: TexSubImage4DSGIS (dynamic) */
+   /* _mesa_function_pool[13985]: TexSubImage4DSGIS (dynamic) */
    "iiiiiiiiiiiip\0"
    "glTexSubImage4DSGIS\0"
    "\0"
-   /* _mesa_function_pool[13984]: TexCoord3s (offset 116) */
+   /* _mesa_function_pool[14020]: TexCoord3s (offset 116) */
    "iii\0"
    "glTexCoord3s\0"
    "\0"
-   /* _mesa_function_pool[14002]: GetTexLevelParameteriv (offset 285) */
+   /* _mesa_function_pool[14038]: GetTexLevelParameteriv (offset 285) */
    "iiip\0"
    "glGetTexLevelParameteriv\0"
    "\0"
-   /* _mesa_function_pool[14033]: CombinerStageParameterfvNV (dynamic) */
+   /* _mesa_function_pool[14069]: CombinerStageParameterfvNV (dynamic) */
    "iip\0"
    "glCombinerStageParameterfvNV\0"
    "\0"
-   /* _mesa_function_pool[14067]: StopInstrumentsSGIX (dynamic) */
+   /* _mesa_function_pool[14103]: StopInstrumentsSGIX (dynamic) */
    "i\0"
    "glStopInstrumentsSGIX\0"
    "\0"
-   /* _mesa_function_pool[14092]: TexCoord4fColor4fNormal3fVertex4fSUN (dynamic) */
+   /* _mesa_function_pool[14128]: TexCoord4fColor4fNormal3fVertex4fSUN (dynamic) */
    "fffffffffffffff\0"
    "glTexCoord4fColor4fNormal3fVertex4fSUN\0"
    "\0"
-   /* _mesa_function_pool[14148]: ClearAccum (offset 204) */
+   /* _mesa_function_pool[14184]: ClearAccum (offset 204) */
    "ffff\0"
    "glClearAccum\0"
    "\0"
-   /* _mesa_function_pool[14167]: DeformSGIX (dynamic) */
+   /* _mesa_function_pool[14203]: DeformSGIX (dynamic) */
    "i\0"
    "glDeformSGIX\0"
    "\0"
-   /* _mesa_function_pool[14183]: GetVertexAttribfvARB (will be remapped) */
+   /* _mesa_function_pool[14219]: GetVertexAttribfvARB (will be remapped) */
    "iip\0"
    "glGetVertexAttribfv\0"
    "glGetVertexAttribfvARB\0"
    "\0"
-   /* _mesa_function_pool[14231]: SecondaryColor3ivEXT (will be remapped) */
+   /* _mesa_function_pool[14267]: SecondaryColor3ivEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3iv\0"
    "glSecondaryColor3ivEXT\0"
    "\0"
-   /* _mesa_function_pool[14277]: TexCoord4iv (offset 123) */
+   /* _mesa_function_pool[14313]: TexCoord4iv (offset 123) */
    "p\0"
    "glTexCoord4iv\0"
    "\0"
-   /* _mesa_function_pool[14294]: UniformMatrix4x2fv (will be remapped) */
+   /* _mesa_function_pool[14330]: UniformMatrix4x2fv (will be remapped) */
    "iiip\0"
    "glUniformMatrix4x2fv\0"
    "\0"
-   /* _mesa_function_pool[14321]: GetDetailTexFuncSGIS (dynamic) */
+   /* _mesa_function_pool[14357]: GetDetailTexFuncSGIS (dynamic) */
    "ip\0"
    "glGetDetailTexFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[14348]: GetCombinerStageParameterfvNV (dynamic) */
+   /* _mesa_function_pool[14384]: GetCombinerStageParameterfvNV (dynamic) */
    "iip\0"
    "glGetCombinerStageParameterfvNV\0"
    "\0"
-   /* _mesa_function_pool[14385]: PolygonOffset (offset 319) */
+   /* _mesa_function_pool[14421]: PolygonOffset (offset 319) */
    "ff\0"
    "glPolygonOffset\0"
    "\0"
-   /* _mesa_function_pool[14405]: BindVertexArray (will be remapped) */
+   /* _mesa_function_pool[14441]: BindVertexArray (will be remapped) */
    "i\0"
    "glBindVertexArray\0"
    "\0"
-   /* _mesa_function_pool[14426]: Color4ubVertex2fvSUN (dynamic) */
+   /* _mesa_function_pool[14462]: Color4ubVertex2fvSUN (dynamic) */
    "pp\0"
    "glColor4ubVertex2fvSUN\0"
    "\0"
-   /* _mesa_function_pool[14453]: Rectd (offset 86) */
+   /* _mesa_function_pool[14489]: Rectd (offset 86) */
    "dddd\0"
    "glRectd\0"
    "\0"
-   /* _mesa_function_pool[14467]: TexFilterFuncSGIS (dynamic) */
+   /* _mesa_function_pool[14503]: TexFilterFuncSGIS (dynamic) */
    "iiip\0"
    "glTexFilterFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[14493]: SampleMaskSGIS (will be remapped) */
+   /* _mesa_function_pool[14529]: SampleMaskSGIS (will be remapped) */
    "fi\0"
    "glSampleMaskSGIS\0"
    "glSampleMaskEXT\0"
    "\0"
-   /* _mesa_function_pool[14530]: GetAttribLocationARB (will be remapped) */
+   /* _mesa_function_pool[14566]: GetAttribLocationARB (will be remapped) */
    "ip\0"
    "glGetAttribLocation\0"
    "glGetAttribLocationARB\0"
    "\0"
-   /* _mesa_function_pool[14577]: RasterPos3i (offset 74) */
+   /* _mesa_function_pool[14613]: RasterPos3i (offset 74) */
    "iii\0"
    "glRasterPos3i\0"
    "\0"
-   /* _mesa_function_pool[14596]: VertexAttrib4ubvARB (will be remapped) */
+   /* _mesa_function_pool[14632]: VertexAttrib4ubvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4ubv\0"
    "glVertexAttrib4ubvARB\0"
    "\0"
-   /* _mesa_function_pool[14641]: DetailTexFuncSGIS (dynamic) */
+   /* _mesa_function_pool[14677]: DetailTexFuncSGIS (dynamic) */
    "iip\0"
    "glDetailTexFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[14666]: Normal3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[14702]: Normal3fVertex3fSUN (dynamic) */
    "ffffff\0"
    "glNormal3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[14696]: CopyTexImage2D (offset 324) */
+   /* _mesa_function_pool[14732]: CopyTexImage2D (offset 324) */
    "iiiiiiii\0"
    "glCopyTexImage2D\0"
    "glCopyTexImage2DEXT\0"
    "\0"
-   /* _mesa_function_pool[14743]: GetBufferPointervARB (will be remapped) */
+   /* _mesa_function_pool[14779]: GetBufferPointervARB (will be remapped) */
    "iip\0"
    "glGetBufferPointerv\0"
    "glGetBufferPointervARB\0"
    "\0"
-   /* _mesa_function_pool[14791]: ProgramEnvParameter4fARB (will be remapped) */
+   /* _mesa_function_pool[14827]: ProgramEnvParameter4fARB (will be remapped) */
    "iiffff\0"
    "glProgramEnvParameter4fARB\0"
    "glProgramParameter4fNV\0"
    "\0"
-   /* _mesa_function_pool[14849]: Uniform3ivARB (will be remapped) */
+   /* _mesa_function_pool[14885]: Uniform3ivARB (will be remapped) */
    "iip\0"
    "glUniform3iv\0"
    "glUniform3ivARB\0"
    "\0"
-   /* _mesa_function_pool[14883]: Lightfv (offset 160) */
+   /* _mesa_function_pool[14919]: Lightfv (offset 160) */
    "iip\0"
    "glLightfv\0"
    "\0"
-   /* _mesa_function_pool[14898]: ClearDepth (offset 208) */
+   /* _mesa_function_pool[14934]: ClearDepth (offset 208) */
    "d\0"
    "glClearDepth\0"
    "\0"
-   /* _mesa_function_pool[14914]: GetFenceivNV (will be remapped) */
+   /* _mesa_function_pool[14950]: GetFenceivNV (will be remapped) */
    "iip\0"
    "glGetFenceivNV\0"
    "\0"
-   /* _mesa_function_pool[14934]: WindowPos4dvMESA (will be remapped) */
+   /* _mesa_function_pool[14970]: WindowPos4dvMESA (will be remapped) */
    "p\0"
    "glWindowPos4dvMESA\0"
    "\0"
-   /* _mesa_function_pool[14956]: ColorSubTable (offset 346) */
+   /* _mesa_function_pool[14992]: ColorSubTable (offset 346) */
    "iiiiip\0"
    "glColorSubTable\0"
    "glColorSubTableEXT\0"
    "\0"
-   /* _mesa_function_pool[14999]: Color4fv (offset 30) */
+   /* _mesa_function_pool[15035]: Color4fv (offset 30) */
    "p\0"
    "glColor4fv\0"
    "\0"
-   /* _mesa_function_pool[15013]: MultiTexCoord4ivARB (offset 405) */
+   /* _mesa_function_pool[15049]: MultiTexCoord4ivARB (offset 405) */
    "ip\0"
    "glMultiTexCoord4iv\0"
    "glMultiTexCoord4ivARB\0"
    "\0"
-   /* _mesa_function_pool[15058]: ProgramLocalParameters4fvEXT (will be remapped) */
+   /* _mesa_function_pool[15094]: ProgramLocalParameters4fvEXT (will be remapped) */
    "iiip\0"
    "glProgramLocalParameters4fvEXT\0"
    "\0"
-   /* _mesa_function_pool[15095]: ColorPointer (offset 308) */
+   /* _mesa_function_pool[15131]: ColorPointer (offset 308) */
    "iiip\0"
    "glColorPointer\0"
    "\0"
-   /* _mesa_function_pool[15116]: Rects (offset 92) */
+   /* _mesa_function_pool[15152]: Rects (offset 92) */
    "iiii\0"
    "glRects\0"
    "\0"
-   /* _mesa_function_pool[15130]: GetMapAttribParameterfvNV (dynamic) */
+   /* _mesa_function_pool[15166]: GetMapAttribParameterfvNV (dynamic) */
    "iiip\0"
    "glGetMapAttribParameterfvNV\0"
    "\0"
-   /* _mesa_function_pool[15164]: Lightiv (offset 162) */
+   /* _mesa_function_pool[15200]: Lightiv (offset 162) */
    "iip\0"
    "glLightiv\0"
    "\0"
-   /* _mesa_function_pool[15179]: VertexAttrib4sARB (will be remapped) */
+   /* _mesa_function_pool[15215]: VertexAttrib4sARB (will be remapped) */
    "iiiii\0"
    "glVertexAttrib4s\0"
    "glVertexAttrib4sARB\0"
    "\0"
-   /* _mesa_function_pool[15223]: GetQueryObjectuivARB (will be remapped) */
+   /* _mesa_function_pool[15259]: GetQueryObjectuivARB (will be remapped) */
    "iip\0"
    "glGetQueryObjectuiv\0"
    "glGetQueryObjectuivARB\0"
    "\0"
-   /* _mesa_function_pool[15271]: GetTexParameteriv (offset 283) */
+   /* _mesa_function_pool[15307]: GetTexParameteriv (offset 283) */
    "iip\0"
    "glGetTexParameteriv\0"
    "\0"
-   /* _mesa_function_pool[15296]: MapParameterivNV (dynamic) */
+   /* _mesa_function_pool[15332]: MapParameterivNV (dynamic) */
    "iip\0"
    "glMapParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[15320]: GenRenderbuffersEXT (will be remapped) */
+   /* _mesa_function_pool[15356]: GenRenderbuffersEXT (will be remapped) */
    "ip\0"
    "glGenRenderbuffers\0"
    "glGenRenderbuffersEXT\0"
    "\0"
-   /* _mesa_function_pool[15365]: VertexAttrib2dvARB (will be remapped) */
+   /* _mesa_function_pool[15401]: VertexAttrib2dvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib2dv\0"
    "glVertexAttrib2dvARB\0"
    "\0"
-   /* _mesa_function_pool[15408]: EdgeFlagPointerEXT (will be remapped) */
+   /* _mesa_function_pool[15444]: EdgeFlagPointerEXT (will be remapped) */
    "iip\0"
    "glEdgeFlagPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[15434]: VertexAttribs2svNV (will be remapped) */
+   /* _mesa_function_pool[15470]: VertexAttribs2svNV (will be remapped) */
    "iip\0"
    "glVertexAttribs2svNV\0"
    "\0"
-   /* _mesa_function_pool[15460]: WeightbvARB (dynamic) */
+   /* _mesa_function_pool[15496]: WeightbvARB (dynamic) */
    "ip\0"
    "glWeightbvARB\0"
    "\0"
-   /* _mesa_function_pool[15478]: VertexAttrib2fvARB (will be remapped) */
+   /* _mesa_function_pool[15514]: VertexAttrib2fvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib2fv\0"
    "glVertexAttrib2fvARB\0"
    "\0"
-   /* _mesa_function_pool[15521]: GetBufferParameterivARB (will be remapped) */
+   /* _mesa_function_pool[15557]: GetBufferParameterivARB (will be remapped) */
    "iip\0"
    "glGetBufferParameteriv\0"
    "glGetBufferParameterivARB\0"
    "\0"
-   /* _mesa_function_pool[15575]: Rectdv (offset 87) */
+   /* _mesa_function_pool[15611]: Rectdv (offset 87) */
    "pp\0"
    "glRectdv\0"
    "\0"
-   /* _mesa_function_pool[15588]: ListParameteriSGIX (dynamic) */
+   /* _mesa_function_pool[15624]: ListParameteriSGIX (dynamic) */
    "iii\0"
    "glListParameteriSGIX\0"
    "\0"
-   /* _mesa_function_pool[15614]: ReplacementCodeuiColor4fNormal3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[15650]: ReplacementCodeuiColor4fNormal3fVertex3fSUN (dynamic) */
    "iffffffffff\0"
    "glReplacementCodeuiColor4fNormal3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[15673]: InstrumentsBufferSGIX (dynamic) */
+   /* _mesa_function_pool[15709]: InstrumentsBufferSGIX (dynamic) */
    "ip\0"
    "glInstrumentsBufferSGIX\0"
    "\0"
-   /* _mesa_function_pool[15701]: VertexAttrib4NivARB (will be remapped) */
+   /* _mesa_function_pool[15737]: VertexAttrib4NivARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4Niv\0"
    "glVertexAttrib4NivARB\0"
    "\0"
-   /* _mesa_function_pool[15746]: GetAttachedShaders (will be remapped) */
+   /* _mesa_function_pool[15782]: GetAttachedShaders (will be remapped) */
    "iipp\0"
    "glGetAttachedShaders\0"
    "\0"
-   /* _mesa_function_pool[15773]: GenVertexArraysAPPLE (will be remapped) */
+   /* _mesa_function_pool[15809]: GenVertexArraysAPPLE (will be remapped) */
    "ip\0"
    "glGenVertexArraysAPPLE\0"
    "\0"
-   /* _mesa_function_pool[15800]: Materialiv (offset 172) */
+   /* _mesa_function_pool[15836]: Materialiv (offset 172) */
    "iip\0"
    "glMaterialiv\0"
    "\0"
-   /* _mesa_function_pool[15818]: PushClientAttrib (offset 335) */
+   /* _mesa_function_pool[15854]: PushClientAttrib (offset 335) */
    "i\0"
    "glPushClientAttrib\0"
    "\0"
-   /* _mesa_function_pool[15840]: ProgramEnvParameters4fvEXT (will be remapped) */
+   /* _mesa_function_pool[15876]: ProgramEnvParameters4fvEXT (will be remapped) */
    "iiip\0"
    "glProgramEnvParameters4fvEXT\0"
    "\0"
-   /* _mesa_function_pool[15875]: TexCoord2fColor4fNormal3fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[15911]: TexCoord2fColor4fNormal3fVertex3fvSUN (dynamic) */
    "pppp\0"
    "glTexCoord2fColor4fNormal3fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[15921]: WindowPos2iMESA (will be remapped) */
+   /* _mesa_function_pool[15957]: WindowPos2iMESA (will be remapped) */
    "ii\0"
    "glWindowPos2i\0"
    "glWindowPos2iARB\0"
    "glWindowPos2iMESA\0"
    "\0"
-   /* _mesa_function_pool[15974]: SecondaryColor3fvEXT (will be remapped) */
+   /* _mesa_function_pool[16010]: SecondaryColor3fvEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3fv\0"
    "glSecondaryColor3fvEXT\0"
    "\0"
-   /* _mesa_function_pool[16020]: PolygonMode (offset 174) */
+   /* _mesa_function_pool[16056]: PolygonMode (offset 174) */
    "ii\0"
    "glPolygonMode\0"
    "\0"
-   /* _mesa_function_pool[16038]: CompressedTexSubImage1DARB (will be remapped) */
+   /* _mesa_function_pool[16074]: CompressedTexSubImage1DARB (will be remapped) */
    "iiiiiip\0"
    "glCompressedTexSubImage1D\0"
    "glCompressedTexSubImage1DARB\0"
    "\0"
-   /* _mesa_function_pool[16102]: GetVertexAttribivNV (will be remapped) */
+   /* _mesa_function_pool[16138]: GetVertexAttribivNV (will be remapped) */
    "iip\0"
    "glGetVertexAttribivNV\0"
    "\0"
-   /* _mesa_function_pool[16129]: GetProgramStringARB (will be remapped) */
+   /* _mesa_function_pool[16165]: GetProgramStringARB (will be remapped) */
    "iip\0"
    "glGetProgramStringARB\0"
    "\0"
-   /* _mesa_function_pool[16156]: TexBumpParameterfvATI (will be remapped) */
+   /* _mesa_function_pool[16192]: TexBumpParameterfvATI (will be remapped) */
    "ip\0"
    "glTexBumpParameterfvATI\0"
    "\0"
-   /* _mesa_function_pool[16184]: CompileShaderARB (will be remapped) */
+   /* _mesa_function_pool[16220]: CompileShaderARB (will be remapped) */
    "i\0"
    "glCompileShader\0"
    "glCompileShaderARB\0"
    "\0"
-   /* _mesa_function_pool[16222]: DeleteShader (will be remapped) */
+   /* _mesa_function_pool[16258]: DeleteShader (will be remapped) */
    "i\0"
    "glDeleteShader\0"
    "\0"
-   /* _mesa_function_pool[16240]: DisableClientState (offset 309) */
+   /* _mesa_function_pool[16276]: DisableClientState (offset 309) */
    "i\0"
    "glDisableClientState\0"
    "\0"
-   /* _mesa_function_pool[16264]: TexGeni (offset 192) */
+   /* _mesa_function_pool[16300]: TexGeni (offset 192) */
    "iii\0"
    "glTexGeni\0"
    "\0"
-   /* _mesa_function_pool[16279]: TexGenf (offset 190) */
+   /* _mesa_function_pool[16315]: TexGenf (offset 190) */
    "iif\0"
    "glTexGenf\0"
    "\0"
-   /* _mesa_function_pool[16294]: Uniform3fARB (will be remapped) */
+   /* _mesa_function_pool[16330]: Uniform3fARB (will be remapped) */
    "ifff\0"
    "glUniform3f\0"
    "glUniform3fARB\0"
    "\0"
-   /* _mesa_function_pool[16327]: TexGend (offset 188) */
+   /* _mesa_function_pool[16363]: TexGend (offset 188) */
    "iid\0"
    "glTexGend\0"
    "\0"
-   /* _mesa_function_pool[16342]: ListParameterfvSGIX (dynamic) */
+   /* _mesa_function_pool[16378]: ListParameterfvSGIX (dynamic) */
    "iip\0"
    "glListParameterfvSGIX\0"
    "\0"
-   /* _mesa_function_pool[16369]: GetPolygonStipple (offset 274) */
+   /* _mesa_function_pool[16405]: GetPolygonStipple (offset 274) */
    "p\0"
    "glGetPolygonStipple\0"
    "\0"
-   /* _mesa_function_pool[16392]: Tangent3dvEXT (dynamic) */
+   /* _mesa_function_pool[16428]: Tangent3dvEXT (dynamic) */
    "p\0"
    "glTangent3dvEXT\0"
    "\0"
-   /* _mesa_function_pool[16411]: GetVertexAttribfvNV (will be remapped) */
+   /* _mesa_function_pool[16447]: GetVertexAttribfvNV (will be remapped) */
    "iip\0"
    "glGetVertexAttribfvNV\0"
    "\0"
-   /* _mesa_function_pool[16438]: WindowPos3sMESA (will be remapped) */
+   /* _mesa_function_pool[16474]: WindowPos3sMESA (will be remapped) */
    "iii\0"
    "glWindowPos3s\0"
    "glWindowPos3sARB\0"
    "glWindowPos3sMESA\0"
    "\0"
-   /* _mesa_function_pool[16492]: VertexAttrib2svNV (will be remapped) */
+   /* _mesa_function_pool[16528]: VertexAttrib2svNV (will be remapped) */
    "ip\0"
    "glVertexAttrib2svNV\0"
    "\0"
-   /* _mesa_function_pool[16516]: VertexAttribs1fvNV (will be remapped) */
+   /* _mesa_function_pool[16552]: VertexAttribs1fvNV (will be remapped) */
    "iip\0"
    "glVertexAttribs1fvNV\0"
    "\0"
-   /* _mesa_function_pool[16542]: TexCoord2fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[16578]: TexCoord2fVertex3fvSUN (dynamic) */
    "pp\0"
    "glTexCoord2fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[16571]: WindowPos4sMESA (will be remapped) */
+   /* _mesa_function_pool[16607]: WindowPos4sMESA (will be remapped) */
    "iiii\0"
    "glWindowPos4sMESA\0"
    "\0"
-   /* _mesa_function_pool[16595]: VertexAttrib4NuivARB (will be remapped) */
+   /* _mesa_function_pool[16631]: VertexAttrib4NuivARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4Nuiv\0"
    "glVertexAttrib4NuivARB\0"
    "\0"
-   /* _mesa_function_pool[16642]: ClientActiveTextureARB (offset 375) */
+   /* _mesa_function_pool[16678]: ClientActiveTextureARB (offset 375) */
    "i\0"
    "glClientActiveTexture\0"
    "glClientActiveTextureARB\0"
    "\0"
-   /* _mesa_function_pool[16692]: PixelTexGenSGIX (will be remapped) */
+   /* _mesa_function_pool[16728]: PixelTexGenSGIX (will be remapped) */
    "i\0"
    "glPixelTexGenSGIX\0"
    "\0"
-   /* _mesa_function_pool[16713]: ReplacementCodeusvSUN (dynamic) */
+   /* _mesa_function_pool[16749]: ReplacementCodeusvSUN (dynamic) */
    "p\0"
    "glReplacementCodeusvSUN\0"
    "\0"
-   /* _mesa_function_pool[16740]: Uniform4fARB (will be remapped) */
+   /* _mesa_function_pool[16776]: Uniform4fARB (will be remapped) */
    "iffff\0"
    "glUniform4f\0"
    "glUniform4fARB\0"
    "\0"
-   /* _mesa_function_pool[16774]: Color4sv (offset 34) */
+   /* _mesa_function_pool[16810]: Color4sv (offset 34) */
    "p\0"
    "glColor4sv\0"
    "\0"
-   /* _mesa_function_pool[16788]: FlushMappedBufferRange (will be remapped) */
+   /* _mesa_function_pool[16824]: FlushMappedBufferRange (will be remapped) */
    "iii\0"
    "glFlushMappedBufferRange\0"
    "\0"
-   /* _mesa_function_pool[16818]: IsProgramNV (will be remapped) */
+   /* _mesa_function_pool[16854]: IsProgramNV (will be remapped) */
    "i\0"
    "glIsProgramARB\0"
    "glIsProgramNV\0"
    "\0"
-   /* _mesa_function_pool[16850]: FlushMappedBufferRangeAPPLE (will be remapped) */
+   /* _mesa_function_pool[16886]: FlushMappedBufferRangeAPPLE (will be remapped) */
    "iii\0"
    "glFlushMappedBufferRangeAPPLE\0"
    "\0"
-   /* _mesa_function_pool[16885]: PixelZoom (offset 246) */
+   /* _mesa_function_pool[16921]: PixelZoom (offset 246) */
    "ff\0"
    "glPixelZoom\0"
    "\0"
-   /* _mesa_function_pool[16901]: ReplacementCodePointerSUN (dynamic) */
+   /* _mesa_function_pool[16937]: ReplacementCodePointerSUN (dynamic) */
    "iip\0"
    "glReplacementCodePointerSUN\0"
    "\0"
-   /* _mesa_function_pool[16934]: ProgramEnvParameter4dARB (will be remapped) */
+   /* _mesa_function_pool[16970]: ProgramEnvParameter4dARB (will be remapped) */
    "iidddd\0"
    "glProgramEnvParameter4dARB\0"
    "glProgramParameter4dNV\0"
    "\0"
-   /* _mesa_function_pool[16992]: ColorTableParameterfv (offset 340) */
+   /* _mesa_function_pool[17028]: ColorTableParameterfv (offset 340) */
    "iip\0"
    "glColorTableParameterfv\0"
    "glColorTableParameterfvSGI\0"
    "\0"
-   /* _mesa_function_pool[17048]: FragmentLightModelfSGIX (dynamic) */
+   /* _mesa_function_pool[17084]: FragmentLightModelfSGIX (dynamic) */
    "if\0"
    "glFragmentLightModelfSGIX\0"
    "\0"
-   /* _mesa_function_pool[17078]: Binormal3bvEXT (dynamic) */
+   /* _mesa_function_pool[17114]: Binormal3bvEXT (dynamic) */
    "p\0"
    "glBinormal3bvEXT\0"
    "\0"
-   /* _mesa_function_pool[17098]: PixelMapuiv (offset 252) */
+   /* _mesa_function_pool[17134]: PixelMapuiv (offset 252) */
    "iip\0"
    "glPixelMapuiv\0"
    "\0"
-   /* _mesa_function_pool[17117]: Color3dv (offset 12) */
+   /* _mesa_function_pool[17153]: Color3dv (offset 12) */
    "p\0"
    "glColor3dv\0"
    "\0"
-   /* _mesa_function_pool[17131]: IsTexture (offset 330) */
+   /* _mesa_function_pool[17167]: IsTexture (offset 330) */
    "i\0"
    "glIsTexture\0"
    "glIsTextureEXT\0"
    "\0"
-   /* _mesa_function_pool[17161]: VertexWeightfvEXT (dynamic) */
+   /* _mesa_function_pool[17197]: VertexWeightfvEXT (dynamic) */
    "p\0"
    "glVertexWeightfvEXT\0"
    "\0"
-   /* _mesa_function_pool[17184]: VertexAttrib1dARB (will be remapped) */
+   /* _mesa_function_pool[17220]: VertexAttrib1dARB (will be remapped) */
    "id\0"
    "glVertexAttrib1d\0"
    "glVertexAttrib1dARB\0"
    "\0"
-   /* _mesa_function_pool[17225]: ImageTransformParameterivHP (dynamic) */
+   /* _mesa_function_pool[17261]: ImageTransformParameterivHP (dynamic) */
    "iip\0"
    "glImageTransformParameterivHP\0"
    "\0"
-   /* _mesa_function_pool[17260]: TexCoord4i (offset 122) */
+   /* _mesa_function_pool[17296]: TexCoord4i (offset 122) */
    "iiii\0"
    "glTexCoord4i\0"
    "\0"
-   /* _mesa_function_pool[17279]: DeleteQueriesARB (will be remapped) */
+   /* _mesa_function_pool[17315]: DeleteQueriesARB (will be remapped) */
    "ip\0"
    "glDeleteQueries\0"
    "glDeleteQueriesARB\0"
    "\0"
-   /* _mesa_function_pool[17318]: Color4ubVertex2fSUN (dynamic) */
+   /* _mesa_function_pool[17354]: Color4ubVertex2fSUN (dynamic) */
    "iiiiff\0"
    "glColor4ubVertex2fSUN\0"
    "\0"
-   /* _mesa_function_pool[17348]: FragmentColorMaterialSGIX (dynamic) */
+   /* _mesa_function_pool[17384]: FragmentColorMaterialSGIX (dynamic) */
    "ii\0"
    "glFragmentColorMaterialSGIX\0"
    "\0"
-   /* _mesa_function_pool[17380]: CurrentPaletteMatrixARB (dynamic) */
+   /* _mesa_function_pool[17416]: CurrentPaletteMatrixARB (dynamic) */
    "i\0"
    "glCurrentPaletteMatrixARB\0"
    "\0"
-   /* _mesa_function_pool[17409]: GetMapdv (offset 266) */
+   /* _mesa_function_pool[17445]: GetMapdv (offset 266) */
    "iip\0"
    "glGetMapdv\0"
    "\0"
-   /* _mesa_function_pool[17425]: SamplePatternSGIS (will be remapped) */
+   /* _mesa_function_pool[17461]: SamplePatternSGIS (will be remapped) */
    "i\0"
    "glSamplePatternSGIS\0"
    "glSamplePatternEXT\0"
    "\0"
-   /* _mesa_function_pool[17467]: PixelStoref (offset 249) */
+   /* _mesa_function_pool[17503]: PixelStoref (offset 249) */
    "if\0"
    "glPixelStoref\0"
    "\0"
-   /* _mesa_function_pool[17485]: IsQueryARB (will be remapped) */
+   /* _mesa_function_pool[17521]: IsQueryARB (will be remapped) */
    "i\0"
    "glIsQuery\0"
    "glIsQueryARB\0"
    "\0"
-   /* _mesa_function_pool[17511]: ReplacementCodeuiColor4ubVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[17547]: ReplacementCodeuiColor4ubVertex3fSUN (dynamic) */
    "iiiiifff\0"
    "glReplacementCodeuiColor4ubVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[17560]: PixelStorei (offset 250) */
+   /* _mesa_function_pool[17596]: PixelStorei (offset 250) */
    "ii\0"
    "glPixelStorei\0"
    "\0"
-   /* _mesa_function_pool[17578]: VertexAttrib4usvARB (will be remapped) */
+   /* _mesa_function_pool[17614]: VertexAttrib4usvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4usv\0"
    "glVertexAttrib4usvARB\0"
    "\0"
-   /* _mesa_function_pool[17623]: LinkProgramARB (will be remapped) */
+   /* _mesa_function_pool[17659]: LinkProgramARB (will be remapped) */
    "i\0"
    "glLinkProgram\0"
    "glLinkProgramARB\0"
    "\0"
-   /* _mesa_function_pool[17657]: VertexAttrib2fNV (will be remapped) */
+   /* _mesa_function_pool[17693]: VertexAttrib2fNV (will be remapped) */
    "iff\0"
    "glVertexAttrib2fNV\0"
    "\0"
-   /* _mesa_function_pool[17681]: ShaderSourceARB (will be remapped) */
+   /* _mesa_function_pool[17717]: ShaderSourceARB (will be remapped) */
    "iipp\0"
    "glShaderSource\0"
    "glShaderSourceARB\0"
    "\0"
-   /* _mesa_function_pool[17720]: FragmentMaterialiSGIX (dynamic) */
+   /* _mesa_function_pool[17756]: FragmentMaterialiSGIX (dynamic) */
    "iii\0"
    "glFragmentMaterialiSGIX\0"
    "\0"
-   /* _mesa_function_pool[17749]: EvalCoord2dv (offset 233) */
+   /* _mesa_function_pool[17785]: EvalCoord2dv (offset 233) */
    "p\0"
    "glEvalCoord2dv\0"
    "\0"
-   /* _mesa_function_pool[17767]: VertexAttrib3svARB (will be remapped) */
+   /* _mesa_function_pool[17803]: VertexAttrib3svARB (will be remapped) */
    "ip\0"
    "glVertexAttrib3sv\0"
    "glVertexAttrib3svARB\0"
    "\0"
-   /* _mesa_function_pool[17810]: ColorMaterial (offset 151) */
+   /* _mesa_function_pool[17846]: ColorMaterial (offset 151) */
    "ii\0"
    "glColorMaterial\0"
    "\0"
-   /* _mesa_function_pool[17830]: CompressedTexSubImage3DARB (will be remapped) */
+   /* _mesa_function_pool[17866]: CompressedTexSubImage3DARB (will be remapped) */
    "iiiiiiiiiip\0"
    "glCompressedTexSubImage3D\0"
    "glCompressedTexSubImage3DARB\0"
    "\0"
-   /* _mesa_function_pool[17898]: WindowPos2ivMESA (will be remapped) */
+   /* _mesa_function_pool[17934]: WindowPos2ivMESA (will be remapped) */
    "p\0"
    "glWindowPos2iv\0"
    "glWindowPos2ivARB\0"
    "glWindowPos2ivMESA\0"
    "\0"
-   /* _mesa_function_pool[17953]: IsFramebufferEXT (will be remapped) */
+   /* _mesa_function_pool[17989]: IsFramebufferEXT (will be remapped) */
    "i\0"
    "glIsFramebuffer\0"
    "glIsFramebufferEXT\0"
    "\0"
-   /* _mesa_function_pool[17991]: Uniform4ivARB (will be remapped) */
+   /* _mesa_function_pool[18027]: Uniform4ivARB (will be remapped) */
    "iip\0"
    "glUniform4iv\0"
    "glUniform4ivARB\0"
    "\0"
-   /* _mesa_function_pool[18025]: GetVertexAttribdvARB (will be remapped) */
+   /* _mesa_function_pool[18061]: GetVertexAttribdvARB (will be remapped) */
    "iip\0"
    "glGetVertexAttribdv\0"
    "glGetVertexAttribdvARB\0"
    "\0"
-   /* _mesa_function_pool[18073]: TexBumpParameterivATI (will be remapped) */
+   /* _mesa_function_pool[18109]: TexBumpParameterivATI (will be remapped) */
    "ip\0"
    "glTexBumpParameterivATI\0"
    "\0"
-   /* _mesa_function_pool[18101]: GetSeparableFilter (offset 359) */
+   /* _mesa_function_pool[18137]: GetSeparableFilter (offset 359) */
    "iiippp\0"
    "glGetSeparableFilter\0"
    "glGetSeparableFilterEXT\0"
    "\0"
-   /* _mesa_function_pool[18154]: Binormal3dEXT (dynamic) */
+   /* _mesa_function_pool[18190]: Binormal3dEXT (dynamic) */
    "ddd\0"
    "glBinormal3dEXT\0"
    "\0"
-   /* _mesa_function_pool[18175]: SpriteParameteriSGIX (dynamic) */
+   /* _mesa_function_pool[18211]: SpriteParameteriSGIX (dynamic) */
    "ii\0"
    "glSpriteParameteriSGIX\0"
    "\0"
-   /* _mesa_function_pool[18202]: RequestResidentProgramsNV (will be remapped) */
+   /* _mesa_function_pool[18238]: RequestResidentProgramsNV (will be remapped) */
    "ip\0"
    "glRequestResidentProgramsNV\0"
    "\0"
-   /* _mesa_function_pool[18234]: TagSampleBufferSGIX (dynamic) */
+   /* _mesa_function_pool[18270]: TagSampleBufferSGIX (dynamic) */
    "\0"
    "glTagSampleBufferSGIX\0"
    "\0"
-   /* _mesa_function_pool[18258]: ReplacementCodeusSUN (dynamic) */
+   /* _mesa_function_pool[18294]: ReplacementCodeusSUN (dynamic) */
    "i\0"
    "glReplacementCodeusSUN\0"
    "\0"
-   /* _mesa_function_pool[18284]: FeedbackBuffer (offset 194) */
+   /* _mesa_function_pool[18320]: FeedbackBuffer (offset 194) */
    "iip\0"
    "glFeedbackBuffer\0"
    "\0"
-   /* _mesa_function_pool[18306]: RasterPos2iv (offset 67) */
+   /* _mesa_function_pool[18342]: RasterPos2iv (offset 67) */
    "p\0"
    "glRasterPos2iv\0"
    "\0"
-   /* _mesa_function_pool[18324]: TexImage1D (offset 182) */
+   /* _mesa_function_pool[18360]: TexImage1D (offset 182) */
    "iiiiiiip\0"
    "glTexImage1D\0"
    "\0"
-   /* _mesa_function_pool[18347]: ListParameterivSGIX (dynamic) */
+   /* _mesa_function_pool[18383]: ListParameterivSGIX (dynamic) */
    "iip\0"
    "glListParameterivSGIX\0"
    "\0"
-   /* _mesa_function_pool[18374]: MultiDrawElementsEXT (will be remapped) */
+   /* _mesa_function_pool[18410]: MultiDrawElementsEXT (will be remapped) */
    "ipipi\0"
    "glMultiDrawElements\0"
    "glMultiDrawElementsEXT\0"
    "\0"
-   /* _mesa_function_pool[18424]: Color3s (offset 17) */
+   /* _mesa_function_pool[18460]: Color3s (offset 17) */
    "iii\0"
    "glColor3s\0"
    "\0"
-   /* _mesa_function_pool[18439]: Uniform1ivARB (will be remapped) */
+   /* _mesa_function_pool[18475]: Uniform1ivARB (will be remapped) */
    "iip\0"
    "glUniform1iv\0"
    "glUniform1ivARB\0"
    "\0"
-   /* _mesa_function_pool[18473]: WindowPos2sMESA (will be remapped) */
+   /* _mesa_function_pool[18509]: WindowPos2sMESA (will be remapped) */
    "ii\0"
    "glWindowPos2s\0"
    "glWindowPos2sARB\0"
    "glWindowPos2sMESA\0"
    "\0"
-   /* _mesa_function_pool[18526]: WeightusvARB (dynamic) */
+   /* _mesa_function_pool[18562]: WeightusvARB (dynamic) */
    "ip\0"
    "glWeightusvARB\0"
    "\0"
-   /* _mesa_function_pool[18545]: TexCoordPointer (offset 320) */
+   /* _mesa_function_pool[18581]: TexCoordPointer (offset 320) */
    "iiip\0"
    "glTexCoordPointer\0"
    "\0"
-   /* _mesa_function_pool[18569]: FogCoordPointerEXT (will be remapped) */
+   /* _mesa_function_pool[18605]: FogCoordPointerEXT (will be remapped) */
    "iip\0"
    "glFogCoordPointer\0"
    "glFogCoordPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[18613]: IndexMaterialEXT (dynamic) */
+   /* _mesa_function_pool[18649]: IndexMaterialEXT (dynamic) */
    "ii\0"
    "glIndexMaterialEXT\0"
    "\0"
-   /* _mesa_function_pool[18636]: Color3i (offset 15) */
+   /* _mesa_function_pool[18672]: Color3i (offset 15) */
    "iii\0"
    "glColor3i\0"
    "\0"
-   /* _mesa_function_pool[18651]: FrontFace (offset 157) */
+   /* _mesa_function_pool[18687]: FrontFace (offset 157) */
    "i\0"
    "glFrontFace\0"
    "\0"
-   /* _mesa_function_pool[18666]: EvalCoord2d (offset 232) */
+   /* _mesa_function_pool[18702]: EvalCoord2d (offset 232) */
    "dd\0"
    "glEvalCoord2d\0"
    "\0"
-   /* _mesa_function_pool[18684]: SecondaryColor3ubvEXT (will be remapped) */
+   /* _mesa_function_pool[18720]: SecondaryColor3ubvEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3ubv\0"
    "glSecondaryColor3ubvEXT\0"
    "\0"
-   /* _mesa_function_pool[18732]: EvalCoord2f (offset 234) */
+   /* _mesa_function_pool[18768]: EvalCoord2f (offset 234) */
    "ff\0"
    "glEvalCoord2f\0"
    "\0"
-   /* _mesa_function_pool[18750]: VertexAttrib4dvARB (will be remapped) */
+   /* _mesa_function_pool[18786]: VertexAttrib4dvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4dv\0"
    "glVertexAttrib4dvARB\0"
    "\0"
-   /* _mesa_function_pool[18793]: BindAttribLocationARB (will be remapped) */
+   /* _mesa_function_pool[18829]: BindAttribLocationARB (will be remapped) */
    "iip\0"
    "glBindAttribLocation\0"
    "glBindAttribLocationARB\0"
    "\0"
-   /* _mesa_function_pool[18843]: Color3b (offset 9) */
+   /* _mesa_function_pool[18879]: Color3b (offset 9) */
    "iii\0"
    "glColor3b\0"
    "\0"
-   /* _mesa_function_pool[18858]: MultiTexCoord2dARB (offset 384) */
+   /* _mesa_function_pool[18894]: MultiTexCoord2dARB (offset 384) */
    "idd\0"
    "glMultiTexCoord2d\0"
    "glMultiTexCoord2dARB\0"
    "\0"
-   /* _mesa_function_pool[18902]: ExecuteProgramNV (will be remapped) */
+   /* _mesa_function_pool[18938]: ExecuteProgramNV (will be remapped) */
    "iip\0"
    "glExecuteProgramNV\0"
    "\0"
-   /* _mesa_function_pool[18926]: Color3f (offset 13) */
+   /* _mesa_function_pool[18962]: Color3f (offset 13) */
    "fff\0"
    "glColor3f\0"
    "\0"
-   /* _mesa_function_pool[18941]: LightEnviSGIX (dynamic) */
+   /* _mesa_function_pool[18977]: LightEnviSGIX (dynamic) */
    "ii\0"
    "glLightEnviSGIX\0"
    "\0"
-   /* _mesa_function_pool[18961]: Color3d (offset 11) */
+   /* _mesa_function_pool[18997]: Color3d (offset 11) */
    "ddd\0"
    "glColor3d\0"
    "\0"
-   /* _mesa_function_pool[18976]: Normal3dv (offset 55) */
+   /* _mesa_function_pool[19012]: Normal3dv (offset 55) */
    "p\0"
    "glNormal3dv\0"
    "\0"
-   /* _mesa_function_pool[18991]: Lightf (offset 159) */
+   /* _mesa_function_pool[19027]: Lightf (offset 159) */
    "iif\0"
    "glLightf\0"
    "\0"
-   /* _mesa_function_pool[19005]: ReplacementCodeuiSUN (dynamic) */
+   /* _mesa_function_pool[19041]: ReplacementCodeuiSUN (dynamic) */
    "i\0"
    "glReplacementCodeuiSUN\0"
    "\0"
-   /* _mesa_function_pool[19031]: MatrixMode (offset 293) */
+   /* _mesa_function_pool[19067]: MatrixMode (offset 293) */
    "i\0"
    "glMatrixMode\0"
    "\0"
-   /* _mesa_function_pool[19047]: GetPixelMapusv (offset 273) */
+   /* _mesa_function_pool[19083]: GetPixelMapusv (offset 273) */
    "ip\0"
    "glGetPixelMapusv\0"
    "\0"
-   /* _mesa_function_pool[19068]: Lighti (offset 161) */
+   /* _mesa_function_pool[19104]: Lighti (offset 161) */
    "iii\0"
    "glLighti\0"
    "\0"
-   /* _mesa_function_pool[19082]: VertexAttribPointerNV (will be remapped) */
+   /* _mesa_function_pool[19118]: VertexAttribPointerNV (will be remapped) */
    "iiiip\0"
    "glVertexAttribPointerNV\0"
    "\0"
-   /* _mesa_function_pool[19113]: GetFramebufferAttachmentParameterivEXT (will be remapped) */
+   /* _mesa_function_pool[19149]: GetFramebufferAttachmentParameterivEXT (will be remapped) */
    "iiip\0"
    "glGetFramebufferAttachmentParameteriv\0"
    "glGetFramebufferAttachmentParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[19198]: PixelTransformParameterfEXT (dynamic) */
+   /* _mesa_function_pool[19234]: PixelTransformParameterfEXT (dynamic) */
    "iif\0"
    "glPixelTransformParameterfEXT\0"
    "\0"
-   /* _mesa_function_pool[19233]: MultiTexCoord4dvARB (offset 401) */
+   /* _mesa_function_pool[19269]: MultiTexCoord4dvARB (offset 401) */
    "ip\0"
    "glMultiTexCoord4dv\0"
    "glMultiTexCoord4dvARB\0"
    "\0"
-   /* _mesa_function_pool[19278]: PixelTransformParameteriEXT (dynamic) */
+   /* _mesa_function_pool[19314]: PixelTransformParameteriEXT (dynamic) */
    "iii\0"
    "glPixelTransformParameteriEXT\0"
    "\0"
-   /* _mesa_function_pool[19313]: GetDoublev (offset 260) */
+   /* _mesa_function_pool[19349]: GetDoublev (offset 260) */
    "ip\0"
    "glGetDoublev\0"
    "\0"
-   /* _mesa_function_pool[19330]: MultMatrixd (offset 295) */
+   /* _mesa_function_pool[19366]: MultMatrixd (offset 295) */
    "p\0"
    "glMultMatrixd\0"
    "\0"
-   /* _mesa_function_pool[19347]: MultMatrixf (offset 294) */
+   /* _mesa_function_pool[19383]: MultMatrixf (offset 294) */
    "p\0"
    "glMultMatrixf\0"
    "\0"
-   /* _mesa_function_pool[19364]: TexCoord2fColor4ubVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[19400]: TexCoord2fColor4ubVertex3fSUN (dynamic) */
    "ffiiiifff\0"
    "glTexCoord2fColor4ubVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[19407]: Uniform1iARB (will be remapped) */
+   /* _mesa_function_pool[19443]: Uniform1iARB (will be remapped) */
    "ii\0"
    "glUniform1i\0"
    "glUniform1iARB\0"
    "\0"
-   /* _mesa_function_pool[19438]: VertexAttribPointerARB (will be remapped) */
+   /* _mesa_function_pool[19474]: VertexAttribPointerARB (will be remapped) */
    "iiiiip\0"
    "glVertexAttribPointer\0"
    "glVertexAttribPointerARB\0"
    "\0"
-   /* _mesa_function_pool[19493]: SharpenTexFuncSGIS (dynamic) */
+   /* _mesa_function_pool[19529]: SharpenTexFuncSGIS (dynamic) */
    "iip\0"
    "glSharpenTexFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[19519]: MultiTexCoord4fvARB (offset 403) */
+   /* _mesa_function_pool[19555]: MultiTexCoord4fvARB (offset 403) */
    "ip\0"
    "glMultiTexCoord4fv\0"
    "glMultiTexCoord4fvARB\0"
    "\0"
-   /* _mesa_function_pool[19564]: UniformMatrix2x3fv (will be remapped) */
+   /* _mesa_function_pool[19600]: UniformMatrix2x3fv (will be remapped) */
    "iiip\0"
    "glUniformMatrix2x3fv\0"
    "\0"
-   /* _mesa_function_pool[19591]: TrackMatrixNV (will be remapped) */
+   /* _mesa_function_pool[19627]: TrackMatrixNV (will be remapped) */
    "iiii\0"
    "glTrackMatrixNV\0"
    "\0"
-   /* _mesa_function_pool[19613]: CombinerParameteriNV (will be remapped) */
+   /* _mesa_function_pool[19649]: CombinerParameteriNV (will be remapped) */
    "ii\0"
    "glCombinerParameteriNV\0"
    "\0"
-   /* _mesa_function_pool[19640]: DeleteAsyncMarkersSGIX (dynamic) */
+   /* _mesa_function_pool[19676]: DeleteAsyncMarkersSGIX (dynamic) */
    "ii\0"
    "glDeleteAsyncMarkersSGIX\0"
    "\0"
-   /* _mesa_function_pool[19669]: IsAsyncMarkerSGIX (dynamic) */
+   /* _mesa_function_pool[19705]: IsAsyncMarkerSGIX (dynamic) */
    "i\0"
    "glIsAsyncMarkerSGIX\0"
    "\0"
-   /* _mesa_function_pool[19692]: FrameZoomSGIX (dynamic) */
+   /* _mesa_function_pool[19728]: FrameZoomSGIX (dynamic) */
    "i\0"
    "glFrameZoomSGIX\0"
    "\0"
-   /* _mesa_function_pool[19711]: Normal3fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[19747]: Normal3fVertex3fvSUN (dynamic) */
    "pp\0"
    "glNormal3fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[19738]: RasterPos4sv (offset 85) */
+   /* _mesa_function_pool[19774]: RasterPos4sv (offset 85) */
    "p\0"
    "glRasterPos4sv\0"
    "\0"
-   /* _mesa_function_pool[19756]: VertexAttrib4NsvARB (will be remapped) */
+   /* _mesa_function_pool[19792]: VertexAttrib4NsvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4Nsv\0"
    "glVertexAttrib4NsvARB\0"
    "\0"
-   /* _mesa_function_pool[19801]: VertexAttrib3fvARB (will be remapped) */
+   /* _mesa_function_pool[19837]: VertexAttrib3fvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib3fv\0"
    "glVertexAttrib3fvARB\0"
    "\0"
-   /* _mesa_function_pool[19844]: ClearColor (offset 206) */
+   /* _mesa_function_pool[19880]: ClearColor (offset 206) */
    "ffff\0"
    "glClearColor\0"
    "\0"
-   /* _mesa_function_pool[19863]: GetSynciv (will be remapped) */
+   /* _mesa_function_pool[19899]: GetSynciv (will be remapped) */
    "iiipp\0"
    "glGetSynciv\0"
    "\0"
-   /* _mesa_function_pool[19882]: DeleteFramebuffersEXT (will be remapped) */
+   /* _mesa_function_pool[19918]: DeleteFramebuffersEXT (will be remapped) */
    "ip\0"
    "glDeleteFramebuffers\0"
    "glDeleteFramebuffersEXT\0"
    "\0"
-   /* _mesa_function_pool[19931]: GlobalAlphaFactorsSUN (dynamic) */
+   /* _mesa_function_pool[19967]: GlobalAlphaFactorsSUN (dynamic) */
    "i\0"
    "glGlobalAlphaFactorsSUN\0"
    "\0"
-   /* _mesa_function_pool[19958]: TexEnviv (offset 187) */
+   /* _mesa_function_pool[19994]: TexEnviv (offset 187) */
    "iip\0"
    "glTexEnviv\0"
    "\0"
-   /* _mesa_function_pool[19974]: TexSubImage3D (offset 372) */
+   /* _mesa_function_pool[20010]: TexSubImage3D (offset 372) */
    "iiiiiiiiiip\0"
    "glTexSubImage3D\0"
    "glTexSubImage3DEXT\0"
    "\0"
-   /* _mesa_function_pool[20022]: Tangent3fEXT (dynamic) */
+   /* _mesa_function_pool[20058]: Tangent3fEXT (dynamic) */
    "fff\0"
    "glTangent3fEXT\0"
    "\0"
-   /* _mesa_function_pool[20042]: SecondaryColor3uivEXT (will be remapped) */
+   /* _mesa_function_pool[20078]: SecondaryColor3uivEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3uiv\0"
    "glSecondaryColor3uivEXT\0"
    "\0"
-   /* _mesa_function_pool[20090]: MatrixIndexubvARB (dynamic) */
+   /* _mesa_function_pool[20126]: MatrixIndexubvARB (dynamic) */
    "ip\0"
    "glMatrixIndexubvARB\0"
    "\0"
-   /* _mesa_function_pool[20114]: Color4fNormal3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[20150]: Color4fNormal3fVertex3fSUN (dynamic) */
    "ffffffffff\0"
    "glColor4fNormal3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[20155]: PixelTexGenParameterfSGIS (will be remapped) */
+   /* _mesa_function_pool[20191]: PixelTexGenParameterfSGIS (will be remapped) */
    "if\0"
    "glPixelTexGenParameterfSGIS\0"
    "\0"
-   /* _mesa_function_pool[20187]: CreateShader (will be remapped) */
+   /* _mesa_function_pool[20223]: CreateShader (will be remapped) */
    "i\0"
    "glCreateShader\0"
    "\0"
-   /* _mesa_function_pool[20205]: GetColorTableParameterfv (offset 344) */
+   /* _mesa_function_pool[20241]: GetColorTableParameterfv (offset 344) */
    "iip\0"
    "glGetColorTableParameterfv\0"
    "glGetColorTableParameterfvSGI\0"
    "glGetColorTableParameterfvEXT\0"
    "\0"
-   /* _mesa_function_pool[20297]: FragmentLightModelfvSGIX (dynamic) */
+   /* _mesa_function_pool[20333]: FragmentLightModelfvSGIX (dynamic) */
    "ip\0"
    "glFragmentLightModelfvSGIX\0"
    "\0"
-   /* _mesa_function_pool[20328]: Bitmap (offset 8) */
+   /* _mesa_function_pool[20364]: Bitmap (offset 8) */
    "iiffffp\0"
    "glBitmap\0"
    "\0"
-   /* _mesa_function_pool[20346]: MultiTexCoord3fARB (offset 394) */
+   /* _mesa_function_pool[20382]: MultiTexCoord3fARB (offset 394) */
    "ifff\0"
    "glMultiTexCoord3f\0"
    "glMultiTexCoord3fARB\0"
    "\0"
-   /* _mesa_function_pool[20391]: GetTexLevelParameterfv (offset 284) */
+   /* _mesa_function_pool[20427]: GetTexLevelParameterfv (offset 284) */
    "iiip\0"
    "glGetTexLevelParameterfv\0"
    "\0"
-   /* _mesa_function_pool[20422]: GetPixelTexGenParameterfvSGIS (will be remapped) */
+   /* _mesa_function_pool[20458]: GetPixelTexGenParameterfvSGIS (will be remapped) */
    "ip\0"
    "glGetPixelTexGenParameterfvSGIS\0"
    "\0"
-   /* _mesa_function_pool[20458]: GenFramebuffersEXT (will be remapped) */
+   /* _mesa_function_pool[20494]: GenFramebuffersEXT (will be remapped) */
    "ip\0"
    "glGenFramebuffers\0"
    "glGenFramebuffersEXT\0"
    "\0"
-   /* _mesa_function_pool[20501]: GetProgramParameterdvNV (will be remapped) */
+   /* _mesa_function_pool[20537]: GetProgramParameterdvNV (will be remapped) */
    "iiip\0"
    "glGetProgramParameterdvNV\0"
    "\0"
-   /* _mesa_function_pool[20533]: Vertex2sv (offset 133) */
+   /* _mesa_function_pool[20569]: Vertex2sv (offset 133) */
    "p\0"
    "glVertex2sv\0"
    "\0"
-   /* _mesa_function_pool[20548]: GetIntegerv (offset 263) */
+   /* _mesa_function_pool[20584]: GetIntegerv (offset 263) */
    "ip\0"
    "glGetIntegerv\0"
    "\0"
-   /* _mesa_function_pool[20566]: IsVertexArrayAPPLE (will be remapped) */
+   /* _mesa_function_pool[20602]: IsVertexArrayAPPLE (will be remapped) */
    "i\0"
    "glIsVertexArray\0"
    "glIsVertexArrayAPPLE\0"
    "\0"
-   /* _mesa_function_pool[20606]: FragmentLightfvSGIX (dynamic) */
+   /* _mesa_function_pool[20642]: FragmentLightfvSGIX (dynamic) */
    "iip\0"
    "glFragmentLightfvSGIX\0"
    "\0"
-   /* _mesa_function_pool[20633]: DetachShader (will be remapped) */
+   /* _mesa_function_pool[20669]: DetachShader (will be remapped) */
    "ii\0"
    "glDetachShader\0"
    "\0"
-   /* _mesa_function_pool[20652]: VertexAttrib4NubARB (will be remapped) */
+   /* _mesa_function_pool[20688]: VertexAttrib4NubARB (will be remapped) */
    "iiiii\0"
    "glVertexAttrib4Nub\0"
    "glVertexAttrib4NubARB\0"
    "\0"
-   /* _mesa_function_pool[20700]: GetProgramEnvParameterfvARB (will be remapped) */
+   /* _mesa_function_pool[20736]: GetProgramEnvParameterfvARB (will be remapped) */
    "iip\0"
    "glGetProgramEnvParameterfvARB\0"
    "\0"
-   /* _mesa_function_pool[20735]: GetTrackMatrixivNV (will be remapped) */
+   /* _mesa_function_pool[20771]: GetTrackMatrixivNV (will be remapped) */
    "iiip\0"
    "glGetTrackMatrixivNV\0"
    "\0"
-   /* _mesa_function_pool[20762]: VertexAttrib3svNV (will be remapped) */
+   /* _mesa_function_pool[20798]: VertexAttrib3svNV (will be remapped) */
    "ip\0"
    "glVertexAttrib3svNV\0"
    "\0"
-   /* _mesa_function_pool[20786]: Uniform4fvARB (will be remapped) */
+   /* _mesa_function_pool[20822]: Uniform4fvARB (will be remapped) */
    "iip\0"
    "glUniform4fv\0"
    "glUniform4fvARB\0"
    "\0"
-   /* _mesa_function_pool[20820]: MultTransposeMatrixfARB (will be remapped) */
+   /* _mesa_function_pool[20856]: MultTransposeMatrixfARB (will be remapped) */
    "p\0"
    "glMultTransposeMatrixf\0"
    "glMultTransposeMatrixfARB\0"
    "\0"
-   /* _mesa_function_pool[20872]: GetTexEnviv (offset 277) */
+   /* _mesa_function_pool[20908]: GetTexEnviv (offset 277) */
    "iip\0"
    "glGetTexEnviv\0"
    "\0"
-   /* _mesa_function_pool[20891]: ColorFragmentOp1ATI (will be remapped) */
+   /* _mesa_function_pool[20927]: ColorFragmentOp1ATI (will be remapped) */
    "iiiiiii\0"
    "glColorFragmentOp1ATI\0"
    "\0"
-   /* _mesa_function_pool[20922]: GetUniformfvARB (will be remapped) */
+   /* _mesa_function_pool[20958]: GetUniformfvARB (will be remapped) */
    "iip\0"
    "glGetUniformfv\0"
    "glGetUniformfvARB\0"
    "\0"
-   /* _mesa_function_pool[20960]: PopClientAttrib (offset 334) */
+   /* _mesa_function_pool[20996]: PopClientAttrib (offset 334) */
    "\0"
    "glPopClientAttrib\0"
    "\0"
-   /* _mesa_function_pool[20980]: ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[21016]: ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN (dynamic) */
    "iffffffffffff\0"
    "glReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[21051]: DetachObjectARB (will be remapped) */
+   /* _mesa_function_pool[21087]: DetachObjectARB (will be remapped) */
    "ii\0"
    "glDetachObjectARB\0"
    "\0"
-   /* _mesa_function_pool[21073]: VertexBlendARB (dynamic) */
+   /* _mesa_function_pool[21109]: VertexBlendARB (dynamic) */
    "i\0"
    "glVertexBlendARB\0"
    "\0"
-   /* _mesa_function_pool[21093]: WindowPos3iMESA (will be remapped) */
+   /* _mesa_function_pool[21129]: WindowPos3iMESA (will be remapped) */
    "iii\0"
    "glWindowPos3i\0"
    "glWindowPos3iARB\0"
    "glWindowPos3iMESA\0"
    "\0"
-   /* _mesa_function_pool[21147]: SeparableFilter2D (offset 360) */
+   /* _mesa_function_pool[21183]: SeparableFilter2D (offset 360) */
    "iiiiiipp\0"
    "glSeparableFilter2D\0"
    "glSeparableFilter2DEXT\0"
    "\0"
-   /* _mesa_function_pool[21200]: ProgramParameteriARB (will be remapped) */
+   /* _mesa_function_pool[21236]: ProgramParameteriARB (will be remapped) */
    "iii\0"
    "glProgramParameteriARB\0"
    "\0"
-   /* _mesa_function_pool[21228]: Map1d (offset 220) */
+   /* _mesa_function_pool[21264]: Map1d (offset 220) */
    "iddiip\0"
    "glMap1d\0"
    "\0"
-   /* _mesa_function_pool[21244]: Map1f (offset 221) */
+   /* _mesa_function_pool[21280]: Map1f (offset 221) */
    "iffiip\0"
    "glMap1f\0"
    "\0"
-   /* _mesa_function_pool[21260]: CompressedTexImage2DARB (will be remapped) */
+   /* _mesa_function_pool[21296]: CompressedTexImage2DARB (will be remapped) */
    "iiiiiiip\0"
    "glCompressedTexImage2D\0"
    "glCompressedTexImage2DARB\0"
    "\0"
-   /* _mesa_function_pool[21319]: ArrayElement (offset 306) */
+   /* _mesa_function_pool[21355]: ArrayElement (offset 306) */
    "i\0"
    "glArrayElement\0"
    "glArrayElementEXT\0"
    "\0"
-   /* _mesa_function_pool[21355]: TexImage2D (offset 183) */
+   /* _mesa_function_pool[21391]: TexImage2D (offset 183) */
    "iiiiiiiip\0"
    "glTexImage2D\0"
    "\0"
-   /* _mesa_function_pool[21379]: DepthBoundsEXT (will be remapped) */
+   /* _mesa_function_pool[21415]: DepthBoundsEXT (will be remapped) */
    "dd\0"
    "glDepthBoundsEXT\0"
    "\0"
-   /* _mesa_function_pool[21400]: ProgramParameters4fvNV (will be remapped) */
+   /* _mesa_function_pool[21436]: ProgramParameters4fvNV (will be remapped) */
    "iiip\0"
    "glProgramParameters4fvNV\0"
    "\0"
-   /* _mesa_function_pool[21431]: DeformationMap3fSGIX (dynamic) */
+   /* _mesa_function_pool[21467]: DeformationMap3fSGIX (dynamic) */
    "iffiiffiiffiip\0"
    "glDeformationMap3fSGIX\0"
    "\0"
-   /* _mesa_function_pool[21470]: GetProgramivNV (will be remapped) */
+   /* _mesa_function_pool[21506]: GetProgramivNV (will be remapped) */
    "iip\0"
    "glGetProgramivNV\0"
    "\0"
-   /* _mesa_function_pool[21492]: GetMinmaxParameteriv (offset 366) */
+   /* _mesa_function_pool[21528]: GetMinmaxParameteriv (offset 366) */
    "iip\0"
    "glGetMinmaxParameteriv\0"
    "glGetMinmaxParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[21546]: PixelTransferf (offset 247) */
+   /* _mesa_function_pool[21582]: PixelTransferf (offset 247) */
    "if\0"
    "glPixelTransferf\0"
    "\0"
-   /* _mesa_function_pool[21567]: CopyTexImage1D (offset 323) */
+   /* _mesa_function_pool[21603]: CopyTexImage1D (offset 323) */
    "iiiiiii\0"
    "glCopyTexImage1D\0"
    "glCopyTexImage1DEXT\0"
    "\0"
-   /* _mesa_function_pool[21613]: PushMatrix (offset 298) */
+   /* _mesa_function_pool[21649]: PushMatrix (offset 298) */
    "\0"
    "glPushMatrix\0"
    "\0"
-   /* _mesa_function_pool[21628]: Fogiv (offset 156) */
+   /* _mesa_function_pool[21664]: Fogiv (offset 156) */
    "ip\0"
    "glFogiv\0"
    "\0"
-   /* _mesa_function_pool[21640]: TexCoord1dv (offset 95) */
+   /* _mesa_function_pool[21676]: TexCoord1dv (offset 95) */
    "p\0"
    "glTexCoord1dv\0"
    "\0"
-   /* _mesa_function_pool[21657]: AlphaFragmentOp3ATI (will be remapped) */
+   /* _mesa_function_pool[21693]: AlphaFragmentOp3ATI (will be remapped) */
    "iiiiiiiiiiii\0"
    "glAlphaFragmentOp3ATI\0"
    "\0"
-   /* _mesa_function_pool[21693]: PixelTransferi (offset 248) */
+   /* _mesa_function_pool[21729]: PixelTransferi (offset 248) */
    "ii\0"
    "glPixelTransferi\0"
    "\0"
-   /* _mesa_function_pool[21714]: GetVertexAttribdvNV (will be remapped) */
+   /* _mesa_function_pool[21750]: GetVertexAttribdvNV (will be remapped) */
    "iip\0"
    "glGetVertexAttribdvNV\0"
    "\0"
-   /* _mesa_function_pool[21741]: VertexAttrib3fvNV (will be remapped) */
+   /* _mesa_function_pool[21777]: VertexAttrib3fvNV (will be remapped) */
    "ip\0"
    "glVertexAttrib3fvNV\0"
    "\0"
-   /* _mesa_function_pool[21765]: Rotatef (offset 300) */
+   /* _mesa_function_pool[21801]: Rotatef (offset 300) */
    "ffff\0"
    "glRotatef\0"
    "\0"
-   /* _mesa_function_pool[21781]: GetFinalCombinerInputParameterivNV (will be remapped) */
+   /* _mesa_function_pool[21817]: GetFinalCombinerInputParameterivNV (will be remapped) */
    "iip\0"
    "glGetFinalCombinerInputParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[21823]: Vertex3i (offset 138) */
+   /* _mesa_function_pool[21859]: Vertex3i (offset 138) */
    "iii\0"
    "glVertex3i\0"
    "\0"
-   /* _mesa_function_pool[21839]: Vertex3f (offset 136) */
+   /* _mesa_function_pool[21875]: Vertex3f (offset 136) */
    "fff\0"
    "glVertex3f\0"
    "\0"
-   /* _mesa_function_pool[21855]: Clear (offset 203) */
+   /* _mesa_function_pool[21891]: Clear (offset 203) */
    "i\0"
    "glClear\0"
    "\0"
-   /* _mesa_function_pool[21866]: Vertex3d (offset 134) */
+   /* _mesa_function_pool[21902]: Vertex3d (offset 134) */
    "ddd\0"
    "glVertex3d\0"
    "\0"
-   /* _mesa_function_pool[21882]: GetMapParameterivNV (dynamic) */
+   /* _mesa_function_pool[21918]: GetMapParameterivNV (dynamic) */
    "iip\0"
    "glGetMapParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[21909]: Uniform4iARB (will be remapped) */
+   /* _mesa_function_pool[21945]: Uniform4iARB (will be remapped) */
    "iiiii\0"
    "glUniform4i\0"
    "glUniform4iARB\0"
    "\0"
-   /* _mesa_function_pool[21943]: ReadBuffer (offset 254) */
+   /* _mesa_function_pool[21979]: ReadBuffer (offset 254) */
    "i\0"
    "glReadBuffer\0"
    "\0"
-   /* _mesa_function_pool[21959]: ConvolutionParameteri (offset 352) */
+   /* _mesa_function_pool[21995]: ConvolutionParameteri (offset 352) */
    "iii\0"
    "glConvolutionParameteri\0"
    "glConvolutionParameteriEXT\0"
    "\0"
-   /* _mesa_function_pool[22015]: Ortho (offset 296) */
+   /* _mesa_function_pool[22051]: Ortho (offset 296) */
    "dddddd\0"
    "glOrtho\0"
    "\0"
-   /* _mesa_function_pool[22031]: Binormal3sEXT (dynamic) */
+   /* _mesa_function_pool[22067]: Binormal3sEXT (dynamic) */
    "iii\0"
    "glBinormal3sEXT\0"
    "\0"
-   /* _mesa_function_pool[22052]: ListBase (offset 6) */
+   /* _mesa_function_pool[22088]: ListBase (offset 6) */
    "i\0"
    "glListBase\0"
    "\0"
-   /* _mesa_function_pool[22066]: Vertex3s (offset 140) */
+   /* _mesa_function_pool[22102]: Vertex3s (offset 140) */
    "iii\0"
    "glVertex3s\0"
    "\0"
-   /* _mesa_function_pool[22082]: ConvolutionParameterf (offset 350) */
+   /* _mesa_function_pool[22118]: ConvolutionParameterf (offset 350) */
    "iif\0"
    "glConvolutionParameterf\0"
    "glConvolutionParameterfEXT\0"
    "\0"
-   /* _mesa_function_pool[22138]: GetColorTableParameteriv (offset 345) */
+   /* _mesa_function_pool[22174]: GetColorTableParameteriv (offset 345) */
    "iip\0"
    "glGetColorTableParameteriv\0"
    "glGetColorTableParameterivSGI\0"
    "glGetColorTableParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[22230]: ProgramEnvParameter4dvARB (will be remapped) */
+   /* _mesa_function_pool[22266]: ProgramEnvParameter4dvARB (will be remapped) */
    "iip\0"
    "glProgramEnvParameter4dvARB\0"
    "glProgramParameter4dvNV\0"
    "\0"
-   /* _mesa_function_pool[22287]: ShadeModel (offset 177) */
+   /* _mesa_function_pool[22323]: ShadeModel (offset 177) */
    "i\0"
    "glShadeModel\0"
    "\0"
-   /* _mesa_function_pool[22303]: VertexAttribs2fvNV (will be remapped) */
+   /* _mesa_function_pool[22339]: VertexAttribs2fvNV (will be remapped) */
    "iip\0"
    "glVertexAttribs2fvNV\0"
    "\0"
-   /* _mesa_function_pool[22329]: Rectiv (offset 91) */
+   /* _mesa_function_pool[22365]: Rectiv (offset 91) */
    "pp\0"
    "glRectiv\0"
    "\0"
-   /* _mesa_function_pool[22342]: UseProgramObjectARB (will be remapped) */
+   /* _mesa_function_pool[22378]: UseProgramObjectARB (will be remapped) */
    "i\0"
    "glUseProgram\0"
    "glUseProgramObjectARB\0"
    "\0"
-   /* _mesa_function_pool[22380]: GetMapParameterfvNV (dynamic) */
+   /* _mesa_function_pool[22416]: GetMapParameterfvNV (dynamic) */
    "iip\0"
    "glGetMapParameterfvNV\0"
    "\0"
-   /* _mesa_function_pool[22407]: PassTexCoordATI (will be remapped) */
+   /* _mesa_function_pool[22443]: PassTexCoordATI (will be remapped) */
    "iii\0"
    "glPassTexCoordATI\0"
    "\0"
-   /* _mesa_function_pool[22430]: DeleteProgram (will be remapped) */
+   /* _mesa_function_pool[22466]: DeleteProgram (will be remapped) */
    "i\0"
    "glDeleteProgram\0"
    "\0"
-   /* _mesa_function_pool[22449]: Tangent3ivEXT (dynamic) */
+   /* _mesa_function_pool[22485]: Tangent3ivEXT (dynamic) */
    "p\0"
    "glTangent3ivEXT\0"
    "\0"
-   /* _mesa_function_pool[22468]: Tangent3dEXT (dynamic) */
+   /* _mesa_function_pool[22504]: Tangent3dEXT (dynamic) */
    "ddd\0"
    "glTangent3dEXT\0"
    "\0"
-   /* _mesa_function_pool[22488]: SecondaryColor3dvEXT (will be remapped) */
+   /* _mesa_function_pool[22524]: SecondaryColor3dvEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3dv\0"
    "glSecondaryColor3dvEXT\0"
    "\0"
-   /* _mesa_function_pool[22534]: Vertex2fv (offset 129) */
+   /* _mesa_function_pool[22570]: Vertex2fv (offset 129) */
    "p\0"
    "glVertex2fv\0"
    "\0"
-   /* _mesa_function_pool[22549]: MultiDrawArraysEXT (will be remapped) */
+   /* _mesa_function_pool[22585]: MultiDrawArraysEXT (will be remapped) */
    "ippi\0"
    "glMultiDrawArrays\0"
    "glMultiDrawArraysEXT\0"
    "\0"
-   /* _mesa_function_pool[22594]: BindRenderbufferEXT (will be remapped) */
+   /* _mesa_function_pool[22630]: BindRenderbufferEXT (will be remapped) */
    "ii\0"
    "glBindRenderbuffer\0"
    "glBindRenderbufferEXT\0"
    "\0"
-   /* _mesa_function_pool[22639]: MultiTexCoord4dARB (offset 400) */
+   /* _mesa_function_pool[22675]: MultiTexCoord4dARB (offset 400) */
    "idddd\0"
    "glMultiTexCoord4d\0"
    "glMultiTexCoord4dARB\0"
    "\0"
-   /* _mesa_function_pool[22685]: FramebufferTextureFaceARB (will be remapped) */
+   /* _mesa_function_pool[22721]: FramebufferTextureFaceARB (will be remapped) */
    "iiiii\0"
    "glFramebufferTextureFaceARB\0"
    "\0"
-   /* _mesa_function_pool[22720]: Vertex3sv (offset 141) */
+   /* _mesa_function_pool[22756]: Vertex3sv (offset 141) */
    "p\0"
    "glVertex3sv\0"
    "\0"
-   /* _mesa_function_pool[22735]: SecondaryColor3usEXT (will be remapped) */
+   /* _mesa_function_pool[22771]: SecondaryColor3usEXT (will be remapped) */
    "iii\0"
    "glSecondaryColor3us\0"
    "glSecondaryColor3usEXT\0"
    "\0"
-   /* _mesa_function_pool[22783]: ProgramLocalParameter4fvARB (will be remapped) */
+   /* _mesa_function_pool[22819]: ProgramLocalParameter4fvARB (will be remapped) */
    "iip\0"
    "glProgramLocalParameter4fvARB\0"
    "\0"
-   /* _mesa_function_pool[22818]: DeleteProgramsNV (will be remapped) */
+   /* _mesa_function_pool[22854]: DeleteProgramsNV (will be remapped) */
    "ip\0"
    "glDeleteProgramsARB\0"
    "glDeleteProgramsNV\0"
    "\0"
-   /* _mesa_function_pool[22861]: EvalMesh1 (offset 236) */
+   /* _mesa_function_pool[22897]: EvalMesh1 (offset 236) */
    "iii\0"
    "glEvalMesh1\0"
    "\0"
-   /* _mesa_function_pool[22878]: GetCombinerOutputParameterfvNV (will be remapped) */
+   /* _mesa_function_pool[22914]: GetCombinerOutputParameterfvNV (will be remapped) */
    "iiip\0"
    "glGetCombinerOutputParameterfvNV\0"
    "\0"
-   /* _mesa_function_pool[22917]: MultiTexCoord1sARB (offset 382) */
+   /* _mesa_function_pool[22953]: MultiTexCoord1sARB (offset 382) */
    "ii\0"
    "glMultiTexCoord1s\0"
    "glMultiTexCoord1sARB\0"
    "\0"
-   /* _mesa_function_pool[22960]: ReplacementCodeuiColor3fVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[22996]: ReplacementCodeuiColor3fVertex3fSUN (dynamic) */
    "iffffff\0"
    "glReplacementCodeuiColor3fVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[23007]: GetVertexAttribPointervNV (will be remapped) */
+   /* _mesa_function_pool[23043]: GetVertexAttribPointervNV (will be remapped) */
    "iip\0"
    "glGetVertexAttribPointerv\0"
    "glGetVertexAttribPointervARB\0"
    "glGetVertexAttribPointervNV\0"
    "\0"
-   /* _mesa_function_pool[23095]: MultiTexCoord1dvARB (offset 377) */
+   /* _mesa_function_pool[23131]: MultiTexCoord1dvARB (offset 377) */
    "ip\0"
    "glMultiTexCoord1dv\0"
    "glMultiTexCoord1dvARB\0"
    "\0"
-   /* _mesa_function_pool[23140]: Uniform2iARB (will be remapped) */
+   /* _mesa_function_pool[23176]: Uniform2iARB (will be remapped) */
    "iii\0"
    "glUniform2i\0"
    "glUniform2iARB\0"
    "\0"
-   /* _mesa_function_pool[23172]: Vertex2iv (offset 131) */
+   /* _mesa_function_pool[23208]: Vertex2iv (offset 131) */
    "p\0"
    "glVertex2iv\0"
    "\0"
-   /* _mesa_function_pool[23187]: GetProgramStringNV (will be remapped) */
+   /* _mesa_function_pool[23223]: GetProgramStringNV (will be remapped) */
    "iip\0"
    "glGetProgramStringNV\0"
    "\0"
-   /* _mesa_function_pool[23213]: ColorPointerEXT (will be remapped) */
+   /* _mesa_function_pool[23249]: ColorPointerEXT (will be remapped) */
    "iiiip\0"
    "glColorPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[23238]: LineWidth (offset 168) */
+   /* _mesa_function_pool[23274]: LineWidth (offset 168) */
    "f\0"
    "glLineWidth\0"
    "\0"
-   /* _mesa_function_pool[23253]: MapBufferARB (will be remapped) */
+   /* _mesa_function_pool[23289]: MapBufferARB (will be remapped) */
    "ii\0"
    "glMapBuffer\0"
    "glMapBufferARB\0"
    "\0"
-   /* _mesa_function_pool[23284]: MultiDrawElementsBaseVertex (will be remapped) */
+   /* _mesa_function_pool[23320]: MultiDrawElementsBaseVertex (will be remapped) */
    "ipipip\0"
    "glMultiDrawElementsBaseVertex\0"
    "\0"
-   /* _mesa_function_pool[23322]: Binormal3svEXT (dynamic) */
+   /* _mesa_function_pool[23358]: Binormal3svEXT (dynamic) */
    "p\0"
    "glBinormal3svEXT\0"
    "\0"
-   /* _mesa_function_pool[23342]: ApplyTextureEXT (dynamic) */
+   /* _mesa_function_pool[23378]: ApplyTextureEXT (dynamic) */
    "i\0"
    "glApplyTextureEXT\0"
    "\0"
-   /* _mesa_function_pool[23363]: TexGendv (offset 189) */
+   /* _mesa_function_pool[23399]: TexGendv (offset 189) */
    "iip\0"
    "glTexGendv\0"
    "\0"
-   /* _mesa_function_pool[23379]: TextureMaterialEXT (dynamic) */
+   /* _mesa_function_pool[23415]: TextureMaterialEXT (dynamic) */
    "ii\0"
    "glTextureMaterialEXT\0"
    "\0"
-   /* _mesa_function_pool[23404]: TextureLightEXT (dynamic) */
+   /* _mesa_function_pool[23440]: TextureLightEXT (dynamic) */
    "i\0"
    "glTextureLightEXT\0"
    "\0"
-   /* _mesa_function_pool[23425]: ResetMinmax (offset 370) */
+   /* _mesa_function_pool[23461]: ResetMinmax (offset 370) */
    "i\0"
    "glResetMinmax\0"
    "glResetMinmaxEXT\0"
    "\0"
-   /* _mesa_function_pool[23459]: SpriteParameterfSGIX (dynamic) */
+   /* _mesa_function_pool[23495]: SpriteParameterfSGIX (dynamic) */
    "if\0"
    "glSpriteParameterfSGIX\0"
    "\0"
-   /* _mesa_function_pool[23486]: EnableClientState (offset 313) */
+   /* _mesa_function_pool[23522]: EnableClientState (offset 313) */
    "i\0"
    "glEnableClientState\0"
    "\0"
-   /* _mesa_function_pool[23509]: VertexAttrib4sNV (will be remapped) */
+   /* _mesa_function_pool[23545]: VertexAttrib4sNV (will be remapped) */
    "iiiii\0"
    "glVertexAttrib4sNV\0"
    "\0"
-   /* _mesa_function_pool[23535]: GetConvolutionParameterfv (offset 357) */
+   /* _mesa_function_pool[23571]: GetConvolutionParameterfv (offset 357) */
    "iip\0"
    "glGetConvolutionParameterfv\0"
    "glGetConvolutionParameterfvEXT\0"
    "\0"
-   /* _mesa_function_pool[23599]: VertexAttribs4dvNV (will be remapped) */
+   /* _mesa_function_pool[23635]: VertexAttribs4dvNV (will be remapped) */
    "iip\0"
    "glVertexAttribs4dvNV\0"
    "\0"
-   /* _mesa_function_pool[23625]: VertexAttrib4dARB (will be remapped) */
+   /* _mesa_function_pool[23661]: VertexAttrib4dARB (will be remapped) */
    "idddd\0"
    "glVertexAttrib4d\0"
    "glVertexAttrib4dARB\0"
    "\0"
-   /* _mesa_function_pool[23669]: GetTexBumpParameterfvATI (will be remapped) */
+   /* _mesa_function_pool[23705]: GetTexBumpParameterfvATI (will be remapped) */
    "ip\0"
    "glGetTexBumpParameterfvATI\0"
    "\0"
-   /* _mesa_function_pool[23700]: ProgramNamedParameter4dNV (will be remapped) */
+   /* _mesa_function_pool[23736]: ProgramNamedParameter4dNV (will be remapped) */
    "iipdddd\0"
    "glProgramNamedParameter4dNV\0"
    "\0"
-   /* _mesa_function_pool[23737]: GetMaterialfv (offset 269) */
+   /* _mesa_function_pool[23773]: GetMaterialfv (offset 269) */
    "iip\0"
    "glGetMaterialfv\0"
    "\0"
-   /* _mesa_function_pool[23758]: VertexWeightfEXT (dynamic) */
+   /* _mesa_function_pool[23794]: VertexWeightfEXT (dynamic) */
    "f\0"
    "glVertexWeightfEXT\0"
    "\0"
-   /* _mesa_function_pool[23780]: Binormal3fEXT (dynamic) */
+   /* _mesa_function_pool[23816]: Binormal3fEXT (dynamic) */
    "fff\0"
    "glBinormal3fEXT\0"
    "\0"
-   /* _mesa_function_pool[23801]: CallList (offset 2) */
+   /* _mesa_function_pool[23837]: CallList (offset 2) */
    "i\0"
    "glCallList\0"
    "\0"
-   /* _mesa_function_pool[23815]: Materialfv (offset 170) */
+   /* _mesa_function_pool[23851]: Materialfv (offset 170) */
    "iip\0"
    "glMaterialfv\0"
    "\0"
-   /* _mesa_function_pool[23833]: TexCoord3fv (offset 113) */
+   /* _mesa_function_pool[23869]: TexCoord3fv (offset 113) */
    "p\0"
    "glTexCoord3fv\0"
    "\0"
-   /* _mesa_function_pool[23850]: FogCoordfvEXT (will be remapped) */
+   /* _mesa_function_pool[23886]: FogCoordfvEXT (will be remapped) */
    "p\0"
    "glFogCoordfv\0"
    "glFogCoordfvEXT\0"
    "\0"
-   /* _mesa_function_pool[23882]: MultiTexCoord1ivARB (offset 381) */
+   /* _mesa_function_pool[23918]: MultiTexCoord1ivARB (offset 381) */
    "ip\0"
    "glMultiTexCoord1iv\0"
    "glMultiTexCoord1ivARB\0"
    "\0"
-   /* _mesa_function_pool[23927]: SecondaryColor3ubEXT (will be remapped) */
+   /* _mesa_function_pool[23963]: SecondaryColor3ubEXT (will be remapped) */
    "iii\0"
    "glSecondaryColor3ub\0"
    "glSecondaryColor3ubEXT\0"
    "\0"
-   /* _mesa_function_pool[23975]: MultiTexCoord2ivARB (offset 389) */
+   /* _mesa_function_pool[24011]: MultiTexCoord2ivARB (offset 389) */
    "ip\0"
    "glMultiTexCoord2iv\0"
    "glMultiTexCoord2ivARB\0"
    "\0"
-   /* _mesa_function_pool[24020]: FogFuncSGIS (dynamic) */
+   /* _mesa_function_pool[24056]: FogFuncSGIS (dynamic) */
    "ip\0"
    "glFogFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[24038]: CopyTexSubImage2D (offset 326) */
+   /* _mesa_function_pool[24074]: CopyTexSubImage2D (offset 326) */
    "iiiiiiii\0"
    "glCopyTexSubImage2D\0"
    "glCopyTexSubImage2DEXT\0"
    "\0"
-   /* _mesa_function_pool[24091]: GetObjectParameterivARB (will be remapped) */
+   /* _mesa_function_pool[24127]: GetObjectParameterivARB (will be remapped) */
    "iip\0"
    "glGetObjectParameterivARB\0"
    "\0"
-   /* _mesa_function_pool[24122]: Color3iv (offset 16) */
+   /* _mesa_function_pool[24158]: Color3iv (offset 16) */
    "p\0"
    "glColor3iv\0"
    "\0"
-   /* _mesa_function_pool[24136]: TexCoord4fVertex4fSUN (dynamic) */
+   /* _mesa_function_pool[24172]: TexCoord4fVertex4fSUN (dynamic) */
    "ffffffff\0"
    "glTexCoord4fVertex4fSUN\0"
    "\0"
-   /* _mesa_function_pool[24170]: DrawElements (offset 311) */
+   /* _mesa_function_pool[24206]: DrawElements (offset 311) */
    "iiip\0"
    "glDrawElements\0"
    "\0"
-   /* _mesa_function_pool[24191]: BindVertexArrayAPPLE (will be remapped) */
+   /* _mesa_function_pool[24227]: BindVertexArrayAPPLE (will be remapped) */
    "i\0"
    "glBindVertexArrayAPPLE\0"
    "\0"
-   /* _mesa_function_pool[24217]: GetProgramLocalParameterdvARB (will be remapped) */
+   /* _mesa_function_pool[24253]: GetProgramLocalParameterdvARB (will be remapped) */
    "iip\0"
    "glGetProgramLocalParameterdvARB\0"
    "\0"
-   /* _mesa_function_pool[24254]: GetHistogramParameteriv (offset 363) */
+   /* _mesa_function_pool[24290]: GetHistogramParameteriv (offset 363) */
    "iip\0"
    "glGetHistogramParameteriv\0"
    "glGetHistogramParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[24314]: MultiTexCoord1iARB (offset 380) */
+   /* _mesa_function_pool[24350]: MultiTexCoord1iARB (offset 380) */
    "ii\0"
    "glMultiTexCoord1i\0"
    "glMultiTexCoord1iARB\0"
    "\0"
-   /* _mesa_function_pool[24357]: GetConvolutionFilter (offset 356) */
+   /* _mesa_function_pool[24393]: GetConvolutionFilter (offset 356) */
    "iiip\0"
    "glGetConvolutionFilter\0"
    "glGetConvolutionFilterEXT\0"
    "\0"
-   /* _mesa_function_pool[24412]: GetProgramivARB (will be remapped) */
+   /* _mesa_function_pool[24448]: GetProgramivARB (will be remapped) */
    "iip\0"
    "glGetProgramivARB\0"
    "\0"
-   /* _mesa_function_pool[24435]: BlendFuncSeparateEXT (will be remapped) */
+   /* _mesa_function_pool[24471]: BlendFuncSeparateEXT (will be remapped) */
    "iiii\0"
    "glBlendFuncSeparate\0"
    "glBlendFuncSeparateEXT\0"
    "glBlendFuncSeparateINGR\0"
    "\0"
-   /* _mesa_function_pool[24508]: MapBufferRange (will be remapped) */
+   /* _mesa_function_pool[24544]: MapBufferRange (will be remapped) */
    "iiii\0"
    "glMapBufferRange\0"
    "\0"
-   /* _mesa_function_pool[24531]: ProgramParameters4dvNV (will be remapped) */
+   /* _mesa_function_pool[24567]: ProgramParameters4dvNV (will be remapped) */
    "iiip\0"
    "glProgramParameters4dvNV\0"
    "\0"
-   /* _mesa_function_pool[24562]: TexCoord2fColor3fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[24598]: TexCoord2fColor3fVertex3fvSUN (dynamic) */
    "ppp\0"
    "glTexCoord2fColor3fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[24599]: EvalPoint2 (offset 239) */
+   /* _mesa_function_pool[24635]: EvalPoint2 (offset 239) */
    "ii\0"
    "glEvalPoint2\0"
    "\0"
-   /* _mesa_function_pool[24616]: EvalPoint1 (offset 237) */
+   /* _mesa_function_pool[24652]: EvalPoint1 (offset 237) */
    "i\0"
    "glEvalPoint1\0"
    "\0"
-   /* _mesa_function_pool[24632]: Binormal3dvEXT (dynamic) */
+   /* _mesa_function_pool[24668]: Binormal3dvEXT (dynamic) */
    "p\0"
    "glBinormal3dvEXT\0"
    "\0"
-   /* _mesa_function_pool[24652]: PopMatrix (offset 297) */
+   /* _mesa_function_pool[24688]: PopMatrix (offset 297) */
    "\0"
    "glPopMatrix\0"
    "\0"
-   /* _mesa_function_pool[24666]: FinishFenceNV (will be remapped) */
+   /* _mesa_function_pool[24702]: FinishFenceNV (will be remapped) */
    "i\0"
    "glFinishFenceNV\0"
    "\0"
-   /* _mesa_function_pool[24685]: GetFogFuncSGIS (dynamic) */
+   /* _mesa_function_pool[24721]: GetFogFuncSGIS (dynamic) */
    "p\0"
    "glGetFogFuncSGIS\0"
    "\0"
-   /* _mesa_function_pool[24705]: GetUniformLocationARB (will be remapped) */
+   /* _mesa_function_pool[24741]: GetUniformLocationARB (will be remapped) */
    "ip\0"
    "glGetUniformLocation\0"
    "glGetUniformLocationARB\0"
    "\0"
-   /* _mesa_function_pool[24754]: SecondaryColor3fEXT (will be remapped) */
+   /* _mesa_function_pool[24790]: SecondaryColor3fEXT (will be remapped) */
    "fff\0"
    "glSecondaryColor3f\0"
    "glSecondaryColor3fEXT\0"
    "\0"
-   /* _mesa_function_pool[24800]: GetTexGeniv (offset 280) */
+   /* _mesa_function_pool[24836]: GetTexGeniv (offset 280) */
    "iip\0"
    "glGetTexGeniv\0"
    "\0"
-   /* _mesa_function_pool[24819]: CombinerInputNV (will be remapped) */
+   /* _mesa_function_pool[24855]: CombinerInputNV (will be remapped) */
    "iiiiii\0"
    "glCombinerInputNV\0"
    "\0"
-   /* _mesa_function_pool[24845]: VertexAttrib3sARB (will be remapped) */
+   /* _mesa_function_pool[24881]: VertexAttrib3sARB (will be remapped) */
    "iiii\0"
    "glVertexAttrib3s\0"
    "glVertexAttrib3sARB\0"
    "\0"
-   /* _mesa_function_pool[24888]: ReplacementCodeuiNormal3fVertex3fvSUN (dynamic) */
+   /* _mesa_function_pool[24924]: ReplacementCodeuiNormal3fVertex3fvSUN (dynamic) */
    "ppp\0"
    "glReplacementCodeuiNormal3fVertex3fvSUN\0"
    "\0"
-   /* _mesa_function_pool[24933]: Map2d (offset 222) */
+   /* _mesa_function_pool[24969]: Map2d (offset 222) */
    "iddiiddiip\0"
    "glMap2d\0"
    "\0"
-   /* _mesa_function_pool[24953]: Map2f (offset 223) */
+   /* _mesa_function_pool[24989]: Map2f (offset 223) */
    "iffiiffiip\0"
    "glMap2f\0"
    "\0"
-   /* _mesa_function_pool[24973]: ProgramStringARB (will be remapped) */
+   /* _mesa_function_pool[25009]: ProgramStringARB (will be remapped) */
    "iiip\0"
    "glProgramStringARB\0"
    "\0"
-   /* _mesa_function_pool[24998]: Vertex4s (offset 148) */
+   /* _mesa_function_pool[25034]: Vertex4s (offset 148) */
    "iiii\0"
    "glVertex4s\0"
    "\0"
-   /* _mesa_function_pool[25015]: TexCoord4fVertex4fvSUN (dynamic) */
+   /* _mesa_function_pool[25051]: TexCoord4fVertex4fvSUN (dynamic) */
    "pp\0"
    "glTexCoord4fVertex4fvSUN\0"
    "\0"
-   /* _mesa_function_pool[25044]: VertexAttrib3sNV (will be remapped) */
+   /* _mesa_function_pool[25080]: VertexAttrib3sNV (will be remapped) */
    "iiii\0"
    "glVertexAttrib3sNV\0"
    "\0"
-   /* _mesa_function_pool[25069]: VertexAttrib1fNV (will be remapped) */
+   /* _mesa_function_pool[25105]: VertexAttrib1fNV (will be remapped) */
    "if\0"
    "glVertexAttrib1fNV\0"
    "\0"
-   /* _mesa_function_pool[25092]: Vertex4f (offset 144) */
+   /* _mesa_function_pool[25128]: Vertex4f (offset 144) */
    "ffff\0"
    "glVertex4f\0"
    "\0"
-   /* _mesa_function_pool[25109]: EvalCoord1d (offset 228) */
+   /* _mesa_function_pool[25145]: EvalCoord1d (offset 228) */
    "d\0"
    "glEvalCoord1d\0"
    "\0"
-   /* _mesa_function_pool[25126]: Vertex4d (offset 142) */
+   /* _mesa_function_pool[25162]: Vertex4d (offset 142) */
    "dddd\0"
    "glVertex4d\0"
    "\0"
-   /* _mesa_function_pool[25143]: RasterPos4dv (offset 79) */
+   /* _mesa_function_pool[25179]: RasterPos4dv (offset 79) */
    "p\0"
    "glRasterPos4dv\0"
    "\0"
-   /* _mesa_function_pool[25161]: FragmentLightfSGIX (dynamic) */
+   /* _mesa_function_pool[25197]: FragmentLightfSGIX (dynamic) */
    "iif\0"
    "glFragmentLightfSGIX\0"
    "\0"
-   /* _mesa_function_pool[25187]: GetCompressedTexImageARB (will be remapped) */
+   /* _mesa_function_pool[25223]: GetCompressedTexImageARB (will be remapped) */
    "iip\0"
    "glGetCompressedTexImage\0"
    "glGetCompressedTexImageARB\0"
    "\0"
-   /* _mesa_function_pool[25243]: GetTexGenfv (offset 279) */
+   /* _mesa_function_pool[25279]: GetTexGenfv (offset 279) */
    "iip\0"
    "glGetTexGenfv\0"
    "\0"
-   /* _mesa_function_pool[25262]: Vertex4i (offset 146) */
+   /* _mesa_function_pool[25298]: Vertex4i (offset 146) */
    "iiii\0"
    "glVertex4i\0"
    "\0"
-   /* _mesa_function_pool[25279]: VertexWeightPointerEXT (dynamic) */
+   /* _mesa_function_pool[25315]: VertexWeightPointerEXT (dynamic) */
    "iiip\0"
    "glVertexWeightPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[25310]: GetHistogram (offset 361) */
+   /* _mesa_function_pool[25346]: GetHistogram (offset 361) */
    "iiiip\0"
    "glGetHistogram\0"
    "glGetHistogramEXT\0"
    "\0"
-   /* _mesa_function_pool[25350]: ActiveStencilFaceEXT (will be remapped) */
+   /* _mesa_function_pool[25386]: ActiveStencilFaceEXT (will be remapped) */
    "i\0"
    "glActiveStencilFaceEXT\0"
    "\0"
-   /* _mesa_function_pool[25376]: StencilFuncSeparateATI (will be remapped) */
+   /* _mesa_function_pool[25412]: StencilFuncSeparateATI (will be remapped) */
    "iiii\0"
    "glStencilFuncSeparateATI\0"
    "\0"
-   /* _mesa_function_pool[25407]: Materialf (offset 169) */
+   /* _mesa_function_pool[25443]: Materialf (offset 169) */
    "iif\0"
    "glMaterialf\0"
    "\0"
-   /* _mesa_function_pool[25424]: GetShaderSourceARB (will be remapped) */
+   /* _mesa_function_pool[25460]: GetShaderSourceARB (will be remapped) */
    "iipp\0"
    "glGetShaderSource\0"
    "glGetShaderSourceARB\0"
    "\0"
-   /* _mesa_function_pool[25469]: IglooInterfaceSGIX (dynamic) */
+   /* _mesa_function_pool[25505]: IglooInterfaceSGIX (dynamic) */
    "ip\0"
    "glIglooInterfaceSGIX\0"
    "\0"
-   /* _mesa_function_pool[25494]: Materiali (offset 171) */
+   /* _mesa_function_pool[25530]: Materiali (offset 171) */
    "iii\0"
    "glMateriali\0"
    "\0"
-   /* _mesa_function_pool[25511]: VertexAttrib4dNV (will be remapped) */
+   /* _mesa_function_pool[25547]: VertexAttrib4dNV (will be remapped) */
    "idddd\0"
    "glVertexAttrib4dNV\0"
    "\0"
-   /* _mesa_function_pool[25537]: MultiModeDrawElementsIBM (will be remapped) */
+   /* _mesa_function_pool[25573]: MultiModeDrawElementsIBM (will be remapped) */
    "ppipii\0"
    "glMultiModeDrawElementsIBM\0"
    "\0"
-   /* _mesa_function_pool[25572]: Indexsv (offset 51) */
+   /* _mesa_function_pool[25608]: Indexsv (offset 51) */
    "p\0"
    "glIndexsv\0"
    "\0"
-   /* _mesa_function_pool[25585]: MultiTexCoord4svARB (offset 407) */
+   /* _mesa_function_pool[25621]: MultiTexCoord4svARB (offset 407) */
    "ip\0"
    "glMultiTexCoord4sv\0"
    "glMultiTexCoord4svARB\0"
    "\0"
-   /* _mesa_function_pool[25630]: LightModelfv (offset 164) */
+   /* _mesa_function_pool[25666]: LightModelfv (offset 164) */
    "ip\0"
    "glLightModelfv\0"
    "\0"
-   /* _mesa_function_pool[25649]: TexCoord2dv (offset 103) */
+   /* _mesa_function_pool[25685]: TexCoord2dv (offset 103) */
    "p\0"
    "glTexCoord2dv\0"
    "\0"
-   /* _mesa_function_pool[25666]: GenQueriesARB (will be remapped) */
+   /* _mesa_function_pool[25702]: GenQueriesARB (will be remapped) */
    "ip\0"
    "glGenQueries\0"
    "glGenQueriesARB\0"
    "\0"
-   /* _mesa_function_pool[25699]: EvalCoord1dv (offset 229) */
+   /* _mesa_function_pool[25735]: EvalCoord1dv (offset 229) */
    "p\0"
    "glEvalCoord1dv\0"
    "\0"
-   /* _mesa_function_pool[25717]: ReplacementCodeuiVertex3fSUN (dynamic) */
+   /* _mesa_function_pool[25753]: ReplacementCodeuiVertex3fSUN (dynamic) */
    "ifff\0"
    "glReplacementCodeuiVertex3fSUN\0"
    "\0"
-   /* _mesa_function_pool[25754]: Translated (offset 303) */
+   /* _mesa_function_pool[25790]: Translated (offset 303) */
    "ddd\0"
    "glTranslated\0"
    "\0"
-   /* _mesa_function_pool[25772]: Translatef (offset 304) */
+   /* _mesa_function_pool[25808]: Translatef (offset 304) */
    "fff\0"
    "glTranslatef\0"
    "\0"
-   /* _mesa_function_pool[25790]: StencilMask (offset 209) */
+   /* _mesa_function_pool[25826]: StencilMask (offset 209) */
    "i\0"
    "glStencilMask\0"
    "\0"
-   /* _mesa_function_pool[25807]: Tangent3iEXT (dynamic) */
+   /* _mesa_function_pool[25843]: Tangent3iEXT (dynamic) */
    "iii\0"
    "glTangent3iEXT\0"
    "\0"
-   /* _mesa_function_pool[25827]: GetLightiv (offset 265) */
+   /* _mesa_function_pool[25863]: GetLightiv (offset 265) */
    "iip\0"
    "glGetLightiv\0"
    "\0"
-   /* _mesa_function_pool[25845]: DrawMeshArraysSUN (dynamic) */
+   /* _mesa_function_pool[25881]: DrawMeshArraysSUN (dynamic) */
    "iiii\0"
    "glDrawMeshArraysSUN\0"
    "\0"
-   /* _mesa_function_pool[25871]: IsList (offset 287) */
+   /* _mesa_function_pool[25907]: IsList (offset 287) */
    "i\0"
    "glIsList\0"
    "\0"
-   /* _mesa_function_pool[25883]: IsSync (will be remapped) */
+   /* _mesa_function_pool[25919]: IsSync (will be remapped) */
    "i\0"
    "glIsSync\0"
    "\0"
-   /* _mesa_function_pool[25895]: RenderMode (offset 196) */
+   /* _mesa_function_pool[25931]: RenderMode (offset 196) */
    "i\0"
    "glRenderMode\0"
    "\0"
-   /* _mesa_function_pool[25911]: GetMapControlPointsNV (dynamic) */
+   /* _mesa_function_pool[25947]: GetMapControlPointsNV (dynamic) */
    "iiiiiip\0"
    "glGetMapControlPointsNV\0"
    "\0"
-   /* _mesa_function_pool[25944]: DrawBuffersARB (will be remapped) */
+   /* _mesa_function_pool[25980]: DrawBuffersARB (will be remapped) */
    "ip\0"
    "glDrawBuffers\0"
    "glDrawBuffersARB\0"
    "glDrawBuffersATI\0"
    "\0"
-   /* _mesa_function_pool[25996]: ProgramLocalParameter4fARB (will be remapped) */
+   /* _mesa_function_pool[26032]: ProgramLocalParameter4fARB (will be remapped) */
    "iiffff\0"
    "glProgramLocalParameter4fARB\0"
    "\0"
-   /* _mesa_function_pool[26033]: SpriteParameterivSGIX (dynamic) */
+   /* _mesa_function_pool[26069]: SpriteParameterivSGIX (dynamic) */
    "ip\0"
    "glSpriteParameterivSGIX\0"
    "\0"
-   /* _mesa_function_pool[26061]: ProvokingVertexEXT (will be remapped) */
+   /* _mesa_function_pool[26097]: ProvokingVertexEXT (will be remapped) */
    "i\0"
    "glProvokingVertexEXT\0"
    "glProvokingVertex\0"
    "\0"
-   /* _mesa_function_pool[26103]: MultiTexCoord1fARB (offset 378) */
+   /* _mesa_function_pool[26139]: MultiTexCoord1fARB (offset 378) */
    "if\0"
    "glMultiTexCoord1f\0"
    "glMultiTexCoord1fARB\0"
    "\0"
-   /* _mesa_function_pool[26146]: LoadName (offset 198) */
+   /* _mesa_function_pool[26182]: LoadName (offset 198) */
    "i\0"
    "glLoadName\0"
    "\0"
-   /* _mesa_function_pool[26160]: VertexAttribs4ubvNV (will be remapped) */
+   /* _mesa_function_pool[26196]: VertexAttribs4ubvNV (will be remapped) */
    "iip\0"
    "glVertexAttribs4ubvNV\0"
    "\0"
-   /* _mesa_function_pool[26187]: WeightsvARB (dynamic) */
+   /* _mesa_function_pool[26223]: WeightsvARB (dynamic) */
    "ip\0"
    "glWeightsvARB\0"
    "\0"
-   /* _mesa_function_pool[26205]: Uniform1fvARB (will be remapped) */
+   /* _mesa_function_pool[26241]: Uniform1fvARB (will be remapped) */
    "iip\0"
    "glUniform1fv\0"
    "glUniform1fvARB\0"
    "\0"
-   /* _mesa_function_pool[26239]: CopyTexSubImage1D (offset 325) */
+   /* _mesa_function_pool[26275]: CopyTexSubImage1D (offset 325) */
    "iiiiii\0"
    "glCopyTexSubImage1D\0"
    "glCopyTexSubImage1DEXT\0"
    "\0"
-   /* _mesa_function_pool[26290]: CullFace (offset 152) */
+   /* _mesa_function_pool[26326]: CullFace (offset 152) */
    "i\0"
    "glCullFace\0"
    "\0"
-   /* _mesa_function_pool[26304]: BindTexture (offset 307) */
+   /* _mesa_function_pool[26340]: BindTexture (offset 307) */
    "ii\0"
    "glBindTexture\0"
    "glBindTextureEXT\0"
    "\0"
-   /* _mesa_function_pool[26339]: BeginFragmentShaderATI (will be remapped) */
+   /* _mesa_function_pool[26375]: BeginFragmentShaderATI (will be remapped) */
    "\0"
    "glBeginFragmentShaderATI\0"
    "\0"
-   /* _mesa_function_pool[26366]: MultiTexCoord4fARB (offset 402) */
+   /* _mesa_function_pool[26402]: MultiTexCoord4fARB (offset 402) */
    "iffff\0"
    "glMultiTexCoord4f\0"
    "glMultiTexCoord4fARB\0"
    "\0"
-   /* _mesa_function_pool[26412]: VertexAttribs3svNV (will be remapped) */
+   /* _mesa_function_pool[26448]: VertexAttribs3svNV (will be remapped) */
    "iip\0"
    "glVertexAttribs3svNV\0"
    "\0"
-   /* _mesa_function_pool[26438]: StencilFunc (offset 243) */
+   /* _mesa_function_pool[26474]: StencilFunc (offset 243) */
    "iii\0"
    "glStencilFunc\0"
    "\0"
-   /* _mesa_function_pool[26457]: CopyPixels (offset 255) */
+   /* _mesa_function_pool[26493]: CopyPixels (offset 255) */
    "iiiii\0"
    "glCopyPixels\0"
    "\0"
-   /* _mesa_function_pool[26477]: Rectsv (offset 93) */
+   /* _mesa_function_pool[26513]: Rectsv (offset 93) */
    "pp\0"
    "glRectsv\0"
    "\0"
-   /* _mesa_function_pool[26490]: ReplacementCodeuivSUN (dynamic) */
+   /* _mesa_function_pool[26526]: ReplacementCodeuivSUN (dynamic) */
    "p\0"
    "glReplacementCodeuivSUN\0"
    "\0"
-   /* _mesa_function_pool[26517]: EnableVertexAttribArrayARB (will be remapped) */
+   /* _mesa_function_pool[26553]: EnableVertexAttribArrayARB (will be remapped) */
    "i\0"
    "glEnableVertexAttribArray\0"
    "glEnableVertexAttribArrayARB\0"
    "\0"
-   /* _mesa_function_pool[26575]: NormalPointervINTEL (dynamic) */
+   /* _mesa_function_pool[26611]: NormalPointervINTEL (dynamic) */
    "ip\0"
    "glNormalPointervINTEL\0"
    "\0"
-   /* _mesa_function_pool[26601]: CopyConvolutionFilter2D (offset 355) */
+   /* _mesa_function_pool[26637]: CopyConvolutionFilter2D (offset 355) */
    "iiiiii\0"
    "glCopyConvolutionFilter2D\0"
    "glCopyConvolutionFilter2DEXT\0"
    "\0"
-   /* _mesa_function_pool[26664]: WindowPos3ivMESA (will be remapped) */
+   /* _mesa_function_pool[26700]: WindowPos3ivMESA (will be remapped) */
    "p\0"
    "glWindowPos3iv\0"
    "glWindowPos3ivARB\0"
    "glWindowPos3ivMESA\0"
    "\0"
-   /* _mesa_function_pool[26719]: CopyBufferSubData (will be remapped) */
+   /* _mesa_function_pool[26755]: CopyBufferSubData (will be remapped) */
    "iiiii\0"
    "glCopyBufferSubData\0"
    "\0"
-   /* _mesa_function_pool[26746]: NormalPointer (offset 318) */
+   /* _mesa_function_pool[26782]: NormalPointer (offset 318) */
    "iip\0"
    "glNormalPointer\0"
    "\0"
-   /* _mesa_function_pool[26767]: TexParameterfv (offset 179) */
+   /* _mesa_function_pool[26803]: TexParameterfv (offset 179) */
    "iip\0"
    "glTexParameterfv\0"
    "\0"
-   /* _mesa_function_pool[26789]: IsBufferARB (will be remapped) */
+   /* _mesa_function_pool[26825]: IsBufferARB (will be remapped) */
    "i\0"
    "glIsBuffer\0"
    "glIsBufferARB\0"
    "\0"
-   /* _mesa_function_pool[26817]: WindowPos4iMESA (will be remapped) */
+   /* _mesa_function_pool[26853]: WindowPos4iMESA (will be remapped) */
    "iiii\0"
    "glWindowPos4iMESA\0"
    "\0"
-   /* _mesa_function_pool[26841]: VertexAttrib4uivARB (will be remapped) */
+   /* _mesa_function_pool[26877]: VertexAttrib4uivARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4uiv\0"
    "glVertexAttrib4uivARB\0"
    "\0"
-   /* _mesa_function_pool[26886]: Tangent3bvEXT (dynamic) */
+   /* _mesa_function_pool[26922]: Tangent3bvEXT (dynamic) */
    "p\0"
    "glTangent3bvEXT\0"
    "\0"
-   /* _mesa_function_pool[26905]: UniformMatrix3x4fv (will be remapped) */
+   /* _mesa_function_pool[26941]: UniformMatrix3x4fv (will be remapped) */
    "iiip\0"
    "glUniformMatrix3x4fv\0"
    "\0"
-   /* _mesa_function_pool[26932]: ClipPlane (offset 150) */
+   /* _mesa_function_pool[26968]: ClipPlane (offset 150) */
    "ip\0"
    "glClipPlane\0"
    "\0"
-   /* _mesa_function_pool[26948]: Recti (offset 90) */
+   /* _mesa_function_pool[26984]: Recti (offset 90) */
    "iiii\0"
    "glRecti\0"
    "\0"
-   /* _mesa_function_pool[26962]: DrawRangeElementsBaseVertex (will be remapped) */
+   /* _mesa_function_pool[26998]: DrawRangeElementsBaseVertex (will be remapped) */
    "iiiiipi\0"
    "glDrawRangeElementsBaseVertex\0"
    "\0"
-   /* _mesa_function_pool[27001]: TexCoordPointervINTEL (dynamic) */
+   /* _mesa_function_pool[27037]: TexCoordPointervINTEL (dynamic) */
    "iip\0"
    "glTexCoordPointervINTEL\0"
    "\0"
-   /* _mesa_function_pool[27030]: DeleteBuffersARB (will be remapped) */
+   /* _mesa_function_pool[27066]: DeleteBuffersARB (will be remapped) */
    "ip\0"
    "glDeleteBuffers\0"
    "glDeleteBuffersARB\0"
    "\0"
-   /* _mesa_function_pool[27069]: WindowPos4fvMESA (will be remapped) */
+   /* _mesa_function_pool[27105]: WindowPos4fvMESA (will be remapped) */
    "p\0"
    "glWindowPos4fvMESA\0"
    "\0"
-   /* _mesa_function_pool[27091]: GetPixelMapuiv (offset 272) */
+   /* _mesa_function_pool[27127]: GetPixelMapuiv (offset 272) */
    "ip\0"
    "glGetPixelMapuiv\0"
    "\0"
-   /* _mesa_function_pool[27112]: Rectf (offset 88) */
+   /* _mesa_function_pool[27148]: Rectf (offset 88) */
    "ffff\0"
    "glRectf\0"
    "\0"
-   /* _mesa_function_pool[27126]: VertexAttrib1sNV (will be remapped) */
+   /* _mesa_function_pool[27162]: VertexAttrib1sNV (will be remapped) */
    "ii\0"
    "glVertexAttrib1sNV\0"
    "\0"
-   /* _mesa_function_pool[27149]: Indexfv (offset 47) */
+   /* _mesa_function_pool[27185]: Indexfv (offset 47) */
    "p\0"
    "glIndexfv\0"
    "\0"
-   /* _mesa_function_pool[27162]: SecondaryColor3svEXT (will be remapped) */
+   /* _mesa_function_pool[27198]: SecondaryColor3svEXT (will be remapped) */
    "p\0"
    "glSecondaryColor3sv\0"
    "glSecondaryColor3svEXT\0"
    "\0"
-   /* _mesa_function_pool[27208]: LoadTransposeMatrixfARB (will be remapped) */
+   /* _mesa_function_pool[27244]: LoadTransposeMatrixfARB (will be remapped) */
    "p\0"
    "glLoadTransposeMatrixf\0"
    "glLoadTransposeMatrixfARB\0"
    "\0"
-   /* _mesa_function_pool[27260]: GetPointerv (offset 329) */
+   /* _mesa_function_pool[27296]: GetPointerv (offset 329) */
    "ip\0"
    "glGetPointerv\0"
    "glGetPointervEXT\0"
    "\0"
-   /* _mesa_function_pool[27295]: Tangent3bEXT (dynamic) */
+   /* _mesa_function_pool[27331]: Tangent3bEXT (dynamic) */
    "iii\0"
    "glTangent3bEXT\0"
    "\0"
-   /* _mesa_function_pool[27315]: CombinerParameterfNV (will be remapped) */
+   /* _mesa_function_pool[27351]: CombinerParameterfNV (will be remapped) */
    "if\0"
    "glCombinerParameterfNV\0"
    "\0"
-   /* _mesa_function_pool[27342]: IndexMask (offset 212) */
+   /* _mesa_function_pool[27378]: IndexMask (offset 212) */
    "i\0"
    "glIndexMask\0"
    "\0"
-   /* _mesa_function_pool[27357]: BindProgramNV (will be remapped) */
+   /* _mesa_function_pool[27393]: BindProgramNV (will be remapped) */
    "ii\0"
    "glBindProgramARB\0"
    "glBindProgramNV\0"
    "\0"
-   /* _mesa_function_pool[27394]: VertexAttrib4svARB (will be remapped) */
+   /* _mesa_function_pool[27430]: VertexAttrib4svARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4sv\0"
    "glVertexAttrib4svARB\0"
    "\0"
-   /* _mesa_function_pool[27437]: GetFloatv (offset 262) */
+   /* _mesa_function_pool[27473]: GetFloatv (offset 262) */
    "ip\0"
    "glGetFloatv\0"
    "\0"
-   /* _mesa_function_pool[27453]: CreateDebugObjectMESA (dynamic) */
+   /* _mesa_function_pool[27489]: CreateDebugObjectMESA (dynamic) */
    "\0"
    "glCreateDebugObjectMESA\0"
    "\0"
-   /* _mesa_function_pool[27479]: GetShaderiv (will be remapped) */
+   /* _mesa_function_pool[27515]: GetShaderiv (will be remapped) */
    "iip\0"
    "glGetShaderiv\0"
    "\0"
-   /* _mesa_function_pool[27498]: ClientWaitSync (will be remapped) */
+   /* _mesa_function_pool[27534]: ClientWaitSync (will be remapped) */
    "iii\0"
    "glClientWaitSync\0"
    "\0"
-   /* _mesa_function_pool[27520]: TexCoord4s (offset 124) */
+   /* _mesa_function_pool[27556]: TexCoord4s (offset 124) */
    "iiii\0"
    "glTexCoord4s\0"
    "\0"
-   /* _mesa_function_pool[27539]: TexCoord3sv (offset 117) */
+   /* _mesa_function_pool[27575]: TexCoord3sv (offset 117) */
    "p\0"
    "glTexCoord3sv\0"
    "\0"
-   /* _mesa_function_pool[27556]: BindFragmentShaderATI (will be remapped) */
+   /* _mesa_function_pool[27592]: BindFragmentShaderATI (will be remapped) */
    "i\0"
    "glBindFragmentShaderATI\0"
    "\0"
-   /* _mesa_function_pool[27583]: PopAttrib (offset 218) */
+   /* _mesa_function_pool[27619]: PopAttrib (offset 218) */
    "\0"
    "glPopAttrib\0"
    "\0"
-   /* _mesa_function_pool[27597]: Fogfv (offset 154) */
+   /* _mesa_function_pool[27633]: Fogfv (offset 154) */
    "ip\0"
    "glFogfv\0"
    "\0"
-   /* _mesa_function_pool[27609]: UnmapBufferARB (will be remapped) */
+   /* _mesa_function_pool[27645]: UnmapBufferARB (will be remapped) */
    "i\0"
    "glUnmapBuffer\0"
    "glUnmapBufferARB\0"
    "\0"
-   /* _mesa_function_pool[27643]: InitNames (offset 197) */
+   /* _mesa_function_pool[27679]: InitNames (offset 197) */
    "\0"
    "glInitNames\0"
    "\0"
-   /* _mesa_function_pool[27657]: Normal3sv (offset 61) */
+   /* _mesa_function_pool[27693]: Normal3sv (offset 61) */
    "p\0"
    "glNormal3sv\0"
    "\0"
-   /* _mesa_function_pool[27672]: Minmax (offset 368) */
+   /* _mesa_function_pool[27708]: Minmax (offset 368) */
    "iii\0"
    "glMinmax\0"
    "glMinmaxEXT\0"
    "\0"
-   /* _mesa_function_pool[27698]: TexCoord4d (offset 118) */
+   /* _mesa_function_pool[27734]: TexCoord4d (offset 118) */
    "dddd\0"
    "glTexCoord4d\0"
    "\0"
-   /* _mesa_function_pool[27717]: DeformationMap3dSGIX (dynamic) */
+   /* _mesa_function_pool[27753]: DeformationMap3dSGIX (dynamic) */
    "iddiiddiiddiip\0"
    "glDeformationMap3dSGIX\0"
    "\0"
-   /* _mesa_function_pool[27756]: TexCoord4f (offset 120) */
+   /* _mesa_function_pool[27792]: TexCoord4f (offset 120) */
    "ffff\0"
    "glTexCoord4f\0"
    "\0"
-   /* _mesa_function_pool[27775]: FogCoorddvEXT (will be remapped) */
+   /* _mesa_function_pool[27811]: FogCoorddvEXT (will be remapped) */
    "p\0"
    "glFogCoorddv\0"
    "glFogCoorddvEXT\0"
    "\0"
-   /* _mesa_function_pool[27807]: FinishTextureSUNX (dynamic) */
+   /* _mesa_function_pool[27843]: FinishTextureSUNX (dynamic) */
    "\0"
    "glFinishTextureSUNX\0"
    "\0"
-   /* _mesa_function_pool[27829]: GetFragmentLightfvSGIX (dynamic) */
+   /* _mesa_function_pool[27865]: GetFragmentLightfvSGIX (dynamic) */
    "iip\0"
    "glGetFragmentLightfvSGIX\0"
    "\0"
-   /* _mesa_function_pool[27859]: Binormal3fvEXT (dynamic) */
+   /* _mesa_function_pool[27895]: Binormal3fvEXT (dynamic) */
    "p\0"
    "glBinormal3fvEXT\0"
    "\0"
-   /* _mesa_function_pool[27879]: GetBooleanv (offset 258) */
+   /* _mesa_function_pool[27915]: GetBooleanv (offset 258) */
    "ip\0"
    "glGetBooleanv\0"
    "\0"
-   /* _mesa_function_pool[27897]: ColorFragmentOp3ATI (will be remapped) */
+   /* _mesa_function_pool[27933]: ColorFragmentOp3ATI (will be remapped) */
    "iiiiiiiiiiiii\0"
    "glColorFragmentOp3ATI\0"
    "\0"
-   /* _mesa_function_pool[27934]: Hint (offset 158) */
+   /* _mesa_function_pool[27970]: Hint (offset 158) */
    "ii\0"
    "glHint\0"
    "\0"
-   /* _mesa_function_pool[27945]: Color4dv (offset 28) */
+   /* _mesa_function_pool[27981]: Color4dv (offset 28) */
    "p\0"
    "glColor4dv\0"
    "\0"
-   /* _mesa_function_pool[27959]: VertexAttrib2svARB (will be remapped) */
+   /* _mesa_function_pool[27995]: VertexAttrib2svARB (will be remapped) */
    "ip\0"
    "glVertexAttrib2sv\0"
    "glVertexAttrib2svARB\0"
    "\0"
-   /* _mesa_function_pool[28002]: AreProgramsResidentNV (will be remapped) */
+   /* _mesa_function_pool[28038]: AreProgramsResidentNV (will be remapped) */
    "ipp\0"
    "glAreProgramsResidentNV\0"
    "\0"
-   /* _mesa_function_pool[28031]: WindowPos3svMESA (will be remapped) */
+   /* _mesa_function_pool[28067]: WindowPos3svMESA (will be remapped) */
    "p\0"
    "glWindowPos3sv\0"
    "glWindowPos3svARB\0"
    "glWindowPos3svMESA\0"
    "\0"
-   /* _mesa_function_pool[28086]: CopyColorSubTable (offset 347) */
+   /* _mesa_function_pool[28122]: CopyColorSubTable (offset 347) */
    "iiiii\0"
    "glCopyColorSubTable\0"
    "glCopyColorSubTableEXT\0"
    "\0"
-   /* _mesa_function_pool[28136]: WeightdvARB (dynamic) */
+   /* _mesa_function_pool[28172]: WeightdvARB (dynamic) */
    "ip\0"
    "glWeightdvARB\0"
    "\0"
-   /* _mesa_function_pool[28154]: DeleteRenderbuffersEXT (will be remapped) */
+   /* _mesa_function_pool[28190]: DeleteRenderbuffersEXT (will be remapped) */
    "ip\0"
    "glDeleteRenderbuffers\0"
    "glDeleteRenderbuffersEXT\0"
    "\0"
-   /* _mesa_function_pool[28205]: VertexAttrib4NubvARB (will be remapped) */
+   /* _mesa_function_pool[28241]: VertexAttrib4NubvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib4Nubv\0"
    "glVertexAttrib4NubvARB\0"
    "\0"
-   /* _mesa_function_pool[28252]: VertexAttrib3dvNV (will be remapped) */
+   /* _mesa_function_pool[28288]: VertexAttrib3dvNV (will be remapped) */
    "ip\0"
    "glVertexAttrib3dvNV\0"
    "\0"
-   /* _mesa_function_pool[28276]: GetObjectParameterfvARB (will be remapped) */
+   /* _mesa_function_pool[28312]: GetObjectParameterfvARB (will be remapped) */
    "iip\0"
    "glGetObjectParameterfvARB\0"
    "\0"
-   /* _mesa_function_pool[28307]: Vertex4iv (offset 147) */
+   /* _mesa_function_pool[28343]: Vertex4iv (offset 147) */
    "p\0"
    "glVertex4iv\0"
    "\0"
-   /* _mesa_function_pool[28322]: GetProgramEnvParameterdvARB (will be remapped) */
+   /* _mesa_function_pool[28358]: GetProgramEnvParameterdvARB (will be remapped) */
    "iip\0"
    "glGetProgramEnvParameterdvARB\0"
    "\0"
-   /* _mesa_function_pool[28357]: TexCoord4dv (offset 119) */
+   /* _mesa_function_pool[28393]: TexCoord4dv (offset 119) */
    "p\0"
    "glTexCoord4dv\0"
    "\0"
-   /* _mesa_function_pool[28374]: LockArraysEXT (will be remapped) */
+   /* _mesa_function_pool[28410]: LockArraysEXT (will be remapped) */
    "ii\0"
    "glLockArraysEXT\0"
    "\0"
-   /* _mesa_function_pool[28394]: Begin (offset 7) */
+   /* _mesa_function_pool[28430]: Begin (offset 7) */
    "i\0"
    "glBegin\0"
    "\0"
-   /* _mesa_function_pool[28405]: LightModeli (offset 165) */
+   /* _mesa_function_pool[28441]: LightModeli (offset 165) */
    "ii\0"
    "glLightModeli\0"
    "\0"
-   /* _mesa_function_pool[28423]: Rectfv (offset 89) */
+   /* _mesa_function_pool[28459]: Rectfv (offset 89) */
    "pp\0"
    "glRectfv\0"
    "\0"
-   /* _mesa_function_pool[28436]: LightModelf (offset 163) */
+   /* _mesa_function_pool[28472]: LightModelf (offset 163) */
    "if\0"
    "glLightModelf\0"
    "\0"
-   /* _mesa_function_pool[28454]: GetTexParameterfv (offset 282) */
+   /* _mesa_function_pool[28490]: GetTexParameterfv (offset 282) */
    "iip\0"
    "glGetTexParameterfv\0"
    "\0"
-   /* _mesa_function_pool[28479]: GetLightfv (offset 264) */
+   /* _mesa_function_pool[28515]: GetLightfv (offset 264) */
    "iip\0"
    "glGetLightfv\0"
    "\0"
-   /* _mesa_function_pool[28497]: PixelTransformParameterivEXT (dynamic) */
+   /* _mesa_function_pool[28533]: PixelTransformParameterivEXT (dynamic) */
    "iip\0"
    "glPixelTransformParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[28533]: BinormalPointerEXT (dynamic) */
+   /* _mesa_function_pool[28569]: BinormalPointerEXT (dynamic) */
    "iip\0"
    "glBinormalPointerEXT\0"
    "\0"
-   /* _mesa_function_pool[28559]: VertexAttrib1dNV (will be remapped) */
+   /* _mesa_function_pool[28595]: VertexAttrib1dNV (will be remapped) */
    "id\0"
    "glVertexAttrib1dNV\0"
    "\0"
-   /* _mesa_function_pool[28582]: GetCombinerInputParameterivNV (will be remapped) */
+   /* _mesa_function_pool[28618]: GetCombinerInputParameterivNV (will be remapped) */
    "iiiip\0"
    "glGetCombinerInputParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[28621]: Disable (offset 214) */
+   /* _mesa_function_pool[28657]: Disable (offset 214) */
    "i\0"
    "glDisable\0"
    "\0"
-   /* _mesa_function_pool[28634]: MultiTexCoord2fvARB (offset 387) */
+   /* _mesa_function_pool[28670]: MultiTexCoord2fvARB (offset 387) */
    "ip\0"
    "glMultiTexCoord2fv\0"
    "glMultiTexCoord2fvARB\0"
    "\0"
-   /* _mesa_function_pool[28679]: GetRenderbufferParameterivEXT (will be remapped) */
+   /* _mesa_function_pool[28715]: GetRenderbufferParameterivEXT (will be remapped) */
    "iip\0"
    "glGetRenderbufferParameteriv\0"
    "glGetRenderbufferParameterivEXT\0"
    "\0"
-   /* _mesa_function_pool[28745]: CombinerParameterivNV (will be remapped) */
+   /* _mesa_function_pool[28781]: CombinerParameterivNV (will be remapped) */
    "ip\0"
    "glCombinerParameterivNV\0"
    "\0"
-   /* _mesa_function_pool[28773]: GenFragmentShadersATI (will be remapped) */
+   /* _mesa_function_pool[28809]: GenFragmentShadersATI (will be remapped) */
    "i\0"
    "glGenFragmentShadersATI\0"
    "\0"
-   /* _mesa_function_pool[28800]: DrawArrays (offset 310) */
+   /* _mesa_function_pool[28836]: DrawArrays (offset 310) */
    "iii\0"
    "glDrawArrays\0"
    "glDrawArraysEXT\0"
    "\0"
-   /* _mesa_function_pool[28834]: WeightuivARB (dynamic) */
+   /* _mesa_function_pool[28870]: WeightuivARB (dynamic) */
    "ip\0"
    "glWeightuivARB\0"
    "\0"
-   /* _mesa_function_pool[28853]: VertexAttrib2sARB (will be remapped) */
+   /* _mesa_function_pool[28889]: VertexAttrib2sARB (will be remapped) */
    "iii\0"
    "glVertexAttrib2s\0"
    "glVertexAttrib2sARB\0"
    "\0"
-   /* _mesa_function_pool[28895]: ColorMask (offset 210) */
+   /* _mesa_function_pool[28931]: ColorMask (offset 210) */
    "iiii\0"
    "glColorMask\0"
    "\0"
-   /* _mesa_function_pool[28913]: GenAsyncMarkersSGIX (dynamic) */
+   /* _mesa_function_pool[28949]: GenAsyncMarkersSGIX (dynamic) */
    "i\0"
    "glGenAsyncMarkersSGIX\0"
    "\0"
-   /* _mesa_function_pool[28938]: Tangent3svEXT (dynamic) */
+   /* _mesa_function_pool[28974]: Tangent3svEXT (dynamic) */
    "p\0"
    "glTangent3svEXT\0"
    "\0"
-   /* _mesa_function_pool[28957]: GetListParameterivSGIX (dynamic) */
+   /* _mesa_function_pool[28993]: GetListParameterivSGIX (dynamic) */
    "iip\0"
    "glGetListParameterivSGIX\0"
    "\0"
-   /* _mesa_function_pool[28987]: BindBufferARB (will be remapped) */
+   /* _mesa_function_pool[29023]: BindBufferARB (will be remapped) */
    "ii\0"
    "glBindBuffer\0"
    "glBindBufferARB\0"
    "\0"
-   /* _mesa_function_pool[29020]: GetInfoLogARB (will be remapped) */
+   /* _mesa_function_pool[29056]: GetInfoLogARB (will be remapped) */
    "iipp\0"
    "glGetInfoLogARB\0"
    "\0"
-   /* _mesa_function_pool[29042]: RasterPos4iv (offset 83) */
+   /* _mesa_function_pool[29078]: RasterPos4iv (offset 83) */
    "p\0"
    "glRasterPos4iv\0"
    "\0"
-   /* _mesa_function_pool[29060]: Enable (offset 215) */
+   /* _mesa_function_pool[29096]: Enable (offset 215) */
    "i\0"
    "glEnable\0"
    "\0"
-   /* _mesa_function_pool[29072]: LineStipple (offset 167) */
+   /* _mesa_function_pool[29108]: LineStipple (offset 167) */
    "ii\0"
    "glLineStipple\0"
    "\0"
-   /* _mesa_function_pool[29090]: VertexAttribs4svNV (will be remapped) */
+   /* _mesa_function_pool[29126]: VertexAttribs4svNV (will be remapped) */
    "iip\0"
    "glVertexAttribs4svNV\0"
    "\0"
-   /* _mesa_function_pool[29116]: EdgeFlagPointerListIBM (dynamic) */
+   /* _mesa_function_pool[29152]: EdgeFlagPointerListIBM (dynamic) */
    "ipi\0"
    "glEdgeFlagPointerListIBM\0"
    "\0"
-   /* _mesa_function_pool[29146]: UniformMatrix3x2fv (will be remapped) */
+   /* _mesa_function_pool[29182]: UniformMatrix3x2fv (will be remapped) */
    "iiip\0"
    "glUniformMatrix3x2fv\0"
    "\0"
-   /* _mesa_function_pool[29173]: GetMinmaxParameterfv (offset 365) */
+   /* _mesa_function_pool[29209]: GetMinmaxParameterfv (offset 365) */
    "iip\0"
    "glGetMinmaxParameterfv\0"
    "glGetMinmaxParameterfvEXT\0"
    "\0"
-   /* _mesa_function_pool[29227]: VertexAttrib1fvARB (will be remapped) */
+   /* _mesa_function_pool[29263]: VertexAttrib1fvARB (will be remapped) */
    "ip\0"
    "glVertexAttrib1fv\0"
    "glVertexAttrib1fvARB\0"
    "\0"
-   /* _mesa_function_pool[29270]: GenBuffersARB (will be remapped) */
+   /* _mesa_function_pool[29306]: GenBuffersARB (will be remapped) */
    "ip\0"
    "glGenBuffers\0"
    "glGenBuffersARB\0"
    "\0"
-   /* _mesa_function_pool[29303]: VertexAttribs1svNV (will be remapped) */
+   /* _mesa_function_pool[29339]: VertexAttribs1svNV (will be remapped) */
    "iip\0"
    "glVertexAttribs1svNV\0"
    "\0"
-   /* _mesa_function_pool[29329]: Vertex3fv (offset 137) */
+   /* _mesa_function_pool[29365]: Vertex3fv (offset 137) */
    "p\0"
    "glVertex3fv\0"
    "\0"
-   /* _mesa_function_pool[29344]: GetTexBumpParameterivATI (will be remapped) */
+   /* _mesa_function_pool[29380]: GetTexBumpParameterivATI (will be remapped) */
    "ip\0"
    "glGetTexBumpParameterivATI\0"
    "\0"
-   /* _mesa_function_pool[29375]: Binormal3bEXT (dynamic) */
+   /* _mesa_function_pool[29411]: Binormal3bEXT (dynamic) */
    "iii\0"
    "glBinormal3bEXT\0"
    "\0"
-   /* _mesa_function_pool[29396]: FragmentMaterialivSGIX (dynamic) */
+   /* _mesa_function_pool[29432]: FragmentMaterialivSGIX (dynamic) */
    "iip\0"
    "glFragmentMaterialivSGIX\0"
    "\0"
-   /* _mesa_function_pool[29426]: IsRenderbufferEXT (will be remapped) */
+   /* _mesa_function_pool[29462]: IsRenderbufferEXT (will be remapped) */
    "i\0"
    "glIsRenderbuffer\0"
    "glIsRenderbufferEXT\0"
    "\0"
-   /* _mesa_function_pool[29466]: GenProgramsNV (will be remapped) */
+   /* _mesa_function_pool[29502]: GenProgramsNV (will be remapped) */
    "ip\0"
    "glGenProgramsARB\0"
    "glGenProgramsNV\0"
    "\0"
-   /* _mesa_function_pool[29503]: VertexAttrib4dvNV (will be remapped) */
+   /* _mesa_function_pool[29539]: VertexAttrib4dvNV (will be remapped) */
    "ip\0"
    "glVertexAttrib4dvNV\0"
    "\0"
-   /* _mesa_function_pool[29527]: EndFragmentShaderATI (will be remapped) */
+   /* _mesa_function_pool[29563]: EndFragmentShaderATI (will be remapped) */
    "\0"
    "glEndFragmentShaderATI\0"
    "\0"
-   /* _mesa_function_pool[29552]: Binormal3iEXT (dynamic) */
+   /* _mesa_function_pool[29588]: Binormal3iEXT (dynamic) */
    "iii\0"
    "glBinormal3iEXT\0"
    "\0"
-   /* _mesa_function_pool[29573]: WindowPos2fMESA (will be remapped) */
+   /* _mesa_function_pool[29609]: WindowPos2fMESA (will be remapped) */
    "ff\0"
    "glWindowPos2f\0"
    "glWindowPos2fARB\0"
@@ -4350,393 +4351,393 @@ static const struct {
 } MESA_remap_table_functions[] = {
    {  1461, AttachShader_remap_index },
    {  8710, CreateProgram_remap_index },
-   { 20187, CreateShader_remap_index },
-   { 22430, DeleteProgram_remap_index },
-   { 16222, DeleteShader_remap_index },
-   { 20633, DetachShader_remap_index },
-   { 15746, GetAttachedShaders_remap_index },
+   { 20223, CreateShader_remap_index },
+   { 22466, DeleteProgram_remap_index },
+   { 16258, DeleteShader_remap_index },
+   { 20669, DetachShader_remap_index },
+   { 15782, GetAttachedShaders_remap_index },
    {  4244, GetProgramInfoLog_remap_index },
    {   361, GetProgramiv_remap_index },
    {  5547, GetShaderInfoLog_remap_index },
-   { 27479, GetShaderiv_remap_index },
-   { 11699, IsProgram_remap_index },
+   { 27515, GetShaderiv_remap_index },
+   { 11735, IsProgram_remap_index },
    { 10734, IsShader_remap_index },
    {  8814, StencilFuncSeparate_remap_index },
    {  3487, StencilMaskSeparate_remap_index },
    {  6629, StencilOpSeparate_remap_index },
-   { 19564, UniformMatrix2x3fv_remap_index },
+   { 19600, UniformMatrix2x3fv_remap_index },
    {  2615, UniformMatrix2x4fv_remap_index },
-   { 29146, UniformMatrix3x2fv_remap_index },
-   { 26905, UniformMatrix3x4fv_remap_index },
-   { 14294, UniformMatrix4x2fv_remap_index },
+   { 29182, UniformMatrix3x2fv_remap_index },
+   { 26941, UniformMatrix3x4fv_remap_index },
+   { 14330, UniformMatrix4x2fv_remap_index },
    {  2937, UniformMatrix4x3fv_remap_index },
    {  8728, LoadTransposeMatrixdARB_remap_index },
-   { 27208, LoadTransposeMatrixfARB_remap_index },
+   { 27244, LoadTransposeMatrixfARB_remap_index },
    {  4817, MultTransposeMatrixdARB_remap_index },
-   { 20820, MultTransposeMatrixfARB_remap_index },
+   { 20856, MultTransposeMatrixfARB_remap_index },
    {   172, SampleCoverageARB_remap_index },
    {  4971, CompressedTexImage1DARB_remap_index },
-   { 21260, CompressedTexImage2DARB_remap_index },
+   { 21296, CompressedTexImage2DARB_remap_index },
    {  3550, CompressedTexImage3DARB_remap_index },
-   { 16038, CompressedTexSubImage1DARB_remap_index },
+   { 16074, CompressedTexSubImage1DARB_remap_index },
    {  1880, CompressedTexSubImage2DARB_remap_index },
-   { 17830, CompressedTexSubImage3DARB_remap_index },
-   { 25187, GetCompressedTexImageARB_remap_index },
+   { 17866, CompressedTexSubImage3DARB_remap_index },
+   { 25223, GetCompressedTexImageARB_remap_index },
    {  3395, DisableVertexAttribArrayARB_remap_index },
-   { 26517, EnableVertexAttribArrayARB_remap_index },
-   { 28322, GetProgramEnvParameterdvARB_remap_index },
-   { 20700, GetProgramEnvParameterfvARB_remap_index },
-   { 24217, GetProgramLocalParameterdvARB_remap_index },
+   { 26553, EnableVertexAttribArrayARB_remap_index },
+   { 28358, GetProgramEnvParameterdvARB_remap_index },
+   { 20736, GetProgramEnvParameterfvARB_remap_index },
+   { 24253, GetProgramLocalParameterdvARB_remap_index },
    {  7071, GetProgramLocalParameterfvARB_remap_index },
-   { 16129, GetProgramStringARB_remap_index },
-   { 24412, GetProgramivARB_remap_index },
-   { 18025, GetVertexAttribdvARB_remap_index },
-   { 14183, GetVertexAttribfvARB_remap_index },
+   { 16165, GetProgramStringARB_remap_index },
+   { 24448, GetProgramivARB_remap_index },
+   { 18061, GetVertexAttribdvARB_remap_index },
+   { 14219, GetVertexAttribfvARB_remap_index },
    {  8623, GetVertexAttribivARB_remap_index },
-   { 16934, ProgramEnvParameter4dARB_remap_index },
-   { 22230, ProgramEnvParameter4dvARB_remap_index },
-   { 14791, ProgramEnvParameter4fARB_remap_index },
+   { 16970, ProgramEnvParameter4dARB_remap_index },
+   { 22266, ProgramEnvParameter4dvARB_remap_index },
+   { 14827, ProgramEnvParameter4fARB_remap_index },
    {  7934, ProgramEnvParameter4fvARB_remap_index },
    {  3513, ProgramLocalParameter4dARB_remap_index },
    { 11445, ProgramLocalParameter4dvARB_remap_index },
-   { 25996, ProgramLocalParameter4fARB_remap_index },
-   { 22783, ProgramLocalParameter4fvARB_remap_index },
-   { 24973, ProgramStringARB_remap_index },
-   { 17184, VertexAttrib1dARB_remap_index },
-   { 13837, VertexAttrib1dvARB_remap_index },
+   { 26032, ProgramLocalParameter4fARB_remap_index },
+   { 22819, ProgramLocalParameter4fvARB_remap_index },
+   { 25009, ProgramStringARB_remap_index },
+   { 17220, VertexAttrib1dARB_remap_index },
+   { 13873, VertexAttrib1dvARB_remap_index },
    {  3688, VertexAttrib1fARB_remap_index },
-   { 29227, VertexAttrib1fvARB_remap_index },
+   { 29263, VertexAttrib1fvARB_remap_index },
    {  6155, VertexAttrib1sARB_remap_index },
    {  2054, VertexAttrib1svARB_remap_index },
-   { 13268, VertexAttrib2dARB_remap_index },
-   { 15365, VertexAttrib2dvARB_remap_index },
+   { 13304, VertexAttrib2dARB_remap_index },
+   { 15401, VertexAttrib2dvARB_remap_index },
    {  1480, VertexAttrib2fARB_remap_index },
-   { 15478, VertexAttrib2fvARB_remap_index },
-   { 28853, VertexAttrib2sARB_remap_index },
-   { 27959, VertexAttrib2svARB_remap_index },
+   { 15514, VertexAttrib2fvARB_remap_index },
+   { 28889, VertexAttrib2sARB_remap_index },
+   { 27995, VertexAttrib2svARB_remap_index },
    {  9932, VertexAttrib3dARB_remap_index },
    {  7637, VertexAttrib3dvARB_remap_index },
    {  1567, VertexAttrib3fARB_remap_index },
-   { 19801, VertexAttrib3fvARB_remap_index },
-   { 24845, VertexAttrib3sARB_remap_index },
-   { 17767, VertexAttrib3svARB_remap_index },
+   { 19837, VertexAttrib3fvARB_remap_index },
+   { 24881, VertexAttrib3sARB_remap_index },
+   { 17803, VertexAttrib3svARB_remap_index },
    {  4270, VertexAttrib4NbvARB_remap_index },
-   { 15701, VertexAttrib4NivARB_remap_index },
-   { 19756, VertexAttrib4NsvARB_remap_index },
-   { 20652, VertexAttrib4NubARB_remap_index },
-   { 28205, VertexAttrib4NubvARB_remap_index },
-   { 16595, VertexAttrib4NuivARB_remap_index },
+   { 15737, VertexAttrib4NivARB_remap_index },
+   { 19792, VertexAttrib4NsvARB_remap_index },
+   { 20688, VertexAttrib4NubARB_remap_index },
+   { 28241, VertexAttrib4NubvARB_remap_index },
+   { 16631, VertexAttrib4NuivARB_remap_index },
    {  2810, VertexAttrib4NusvARB_remap_index },
    {  9555, VertexAttrib4bvARB_remap_index },
-   { 23625, VertexAttrib4dARB_remap_index },
-   { 18750, VertexAttrib4dvARB_remap_index },
+   { 23661, VertexAttrib4dARB_remap_index },
+   { 18786, VertexAttrib4dvARB_remap_index },
    { 10039, VertexAttrib4fARB_remap_index },
    { 10409, VertexAttrib4fvARB_remap_index },
    {  9007, VertexAttrib4ivARB_remap_index },
-   { 15179, VertexAttrib4sARB_remap_index },
-   { 27394, VertexAttrib4svARB_remap_index },
-   { 14596, VertexAttrib4ubvARB_remap_index },
-   { 26841, VertexAttrib4uivARB_remap_index },
-   { 17578, VertexAttrib4usvARB_remap_index },
-   { 19438, VertexAttribPointerARB_remap_index },
-   { 28987, BindBufferARB_remap_index },
+   { 15215, VertexAttrib4sARB_remap_index },
+   { 27430, VertexAttrib4svARB_remap_index },
+   { 14632, VertexAttrib4ubvARB_remap_index },
+   { 26877, VertexAttrib4uivARB_remap_index },
+   { 17614, VertexAttrib4usvARB_remap_index },
+   { 19474, VertexAttribPointerARB_remap_index },
+   { 29023, BindBufferARB_remap_index },
    {  5862, BufferDataARB_remap_index },
    {  1382, BufferSubDataARB_remap_index },
-   { 27030, DeleteBuffersARB_remap_index },
-   { 29270, GenBuffersARB_remap_index },
-   { 15521, GetBufferParameterivARB_remap_index },
-   { 14743, GetBufferPointervARB_remap_index },
+   { 27066, DeleteBuffersARB_remap_index },
+   { 29306, GenBuffersARB_remap_index },
+   { 15557, GetBufferParameterivARB_remap_index },
+   { 14779, GetBufferPointervARB_remap_index },
    {  1335, GetBufferSubDataARB_remap_index },
-   { 26789, IsBufferARB_remap_index },
-   { 23253, MapBufferARB_remap_index },
-   { 27609, UnmapBufferARB_remap_index },
+   { 26825, IsBufferARB_remap_index },
+   { 23289, MapBufferARB_remap_index },
+   { 27645, UnmapBufferARB_remap_index },
    {   268, BeginQueryARB_remap_index },
-   { 17279, DeleteQueriesARB_remap_index },
+   { 17315, DeleteQueriesARB_remap_index },
    { 10670, EndQueryARB_remap_index },
-   { 25666, GenQueriesARB_remap_index },
+   { 25702, GenQueriesARB_remap_index },
    {  1772, GetQueryObjectivARB_remap_index },
-   { 15223, GetQueryObjectuivARB_remap_index },
+   { 15259, GetQueryObjectuivARB_remap_index },
    {  1624, GetQueryivARB_remap_index },
-   { 17485, IsQueryARB_remap_index },
+   { 17521, IsQueryARB_remap_index },
    {  7247, AttachObjectARB_remap_index },
-   { 16184, CompileShaderARB_remap_index },
+   { 16220, CompileShaderARB_remap_index },
    {  2879, CreateProgramObjectARB_remap_index },
    {  5807, CreateShaderObjectARB_remap_index },
-   { 12685, DeleteObjectARB_remap_index },
-   { 21051, DetachObjectARB_remap_index },
+   { 12721, DeleteObjectARB_remap_index },
+   { 21087, DetachObjectARB_remap_index },
    { 10481, GetActiveUniformARB_remap_index },
    {  8326, GetAttachedObjectsARB_remap_index },
    {  8605, GetHandleARB_remap_index },
-   { 29020, GetInfoLogARB_remap_index },
-   { 28276, GetObjectParameterfvARB_remap_index },
-   { 24091, GetObjectParameterivARB_remap_index },
-   { 25424, GetShaderSourceARB_remap_index },
-   { 24705, GetUniformLocationARB_remap_index },
-   { 20922, GetUniformfvARB_remap_index },
+   { 29056, GetInfoLogARB_remap_index },
+   { 28312, GetObjectParameterfvARB_remap_index },
+   { 24127, GetObjectParameterivARB_remap_index },
+   { 25460, GetShaderSourceARB_remap_index },
+   { 24741, GetUniformLocationARB_remap_index },
+   { 20958, GetUniformfvARB_remap_index },
    { 11067, GetUniformivARB_remap_index },
-   { 17623, LinkProgramARB_remap_index },
-   { 17681, ShaderSourceARB_remap_index },
+   { 17659, LinkProgramARB_remap_index },
+   { 17717, ShaderSourceARB_remap_index },
    {  6529, Uniform1fARB_remap_index },
-   { 26205, Uniform1fvARB_remap_index },
-   { 19407, Uniform1iARB_remap_index },
-   { 18439, Uniform1ivARB_remap_index },
+   { 26241, Uniform1fvARB_remap_index },
+   { 19443, Uniform1iARB_remap_index },
+   { 18475, Uniform1ivARB_remap_index },
    {  2003, Uniform2fARB_remap_index },
-   { 12521, Uniform2fvARB_remap_index },
-   { 23140, Uniform2iARB_remap_index },
+   { 12557, Uniform2fvARB_remap_index },
+   { 23176, Uniform2iARB_remap_index },
    {  2123, Uniform2ivARB_remap_index },
-   { 16294, Uniform3fARB_remap_index },
+   { 16330, Uniform3fARB_remap_index },
    {  8356, Uniform3fvARB_remap_index },
    {  5481, Uniform3iARB_remap_index },
-   { 14849, Uniform3ivARB_remap_index },
-   { 16740, Uniform4fARB_remap_index },
-   { 20786, Uniform4fvARB_remap_index },
-   { 21909, Uniform4iARB_remap_index },
-   { 17991, Uniform4ivARB_remap_index },
+   { 14885, Uniform3ivARB_remap_index },
+   { 16776, Uniform4fARB_remap_index },
+   { 20822, Uniform4fvARB_remap_index },
+   { 21945, Uniform4iARB_remap_index },
+   { 18027, Uniform4ivARB_remap_index },
    {  7299, UniformMatrix2fvARB_remap_index },
    {    17, UniformMatrix3fvARB_remap_index },
    {  2475, UniformMatrix4fvARB_remap_index },
-   { 22342, UseProgramObjectARB_remap_index },
-   { 12956, ValidateProgramARB_remap_index },
-   { 18793, BindAttribLocationARB_remap_index },
+   { 22378, UseProgramObjectARB_remap_index },
+   { 12992, ValidateProgramARB_remap_index },
+   { 18829, BindAttribLocationARB_remap_index },
    {  4315, GetActiveAttribARB_remap_index },
-   { 14530, GetAttribLocationARB_remap_index },
-   { 25944, DrawBuffersARB_remap_index },
+   { 14566, GetAttribLocationARB_remap_index },
+   { 25980, DrawBuffersARB_remap_index },
    { 11550, RenderbufferStorageMultisample_remap_index },
-   { 11918, FramebufferTextureARB_remap_index },
-   { 22685, FramebufferTextureFaceARB_remap_index },
-   { 21200, ProgramParameteriARB_remap_index },
-   { 16788, FlushMappedBufferRange_remap_index },
-   { 24508, MapBufferRange_remap_index },
-   { 14405, BindVertexArray_remap_index },
-   { 12815, GenVertexArrays_remap_index },
-   { 26719, CopyBufferSubData_remap_index },
-   { 27498, ClientWaitSync_remap_index },
+   { 11954, FramebufferTextureARB_remap_index },
+   { 22721, FramebufferTextureFaceARB_remap_index },
+   { 21236, ProgramParameteriARB_remap_index },
+   { 16824, FlushMappedBufferRange_remap_index },
+   { 24544, MapBufferRange_remap_index },
+   { 14441, BindVertexArray_remap_index },
+   { 12851, GenVertexArrays_remap_index },
+   { 26755, CopyBufferSubData_remap_index },
+   { 27534, ClientWaitSync_remap_index },
    {  2394, DeleteSync_remap_index },
    {  6196, FenceSync_remap_index },
-   { 13327, GetInteger64v_remap_index },
-   { 19863, GetSynciv_remap_index },
-   { 25883, IsSync_remap_index },
+   { 13363, GetInteger64v_remap_index },
+   { 19899, GetSynciv_remap_index },
+   { 25919, IsSync_remap_index },
    {  8274, WaitSync_remap_index },
    {  3363, DrawElementsBaseVertex_remap_index },
-   { 26962, DrawRangeElementsBaseVertex_remap_index },
-   { 23284, MultiDrawElementsBaseVertex_remap_index },
+   { 26998, DrawRangeElementsBaseVertex_remap_index },
+   { 23320, MultiDrawElementsBaseVertex_remap_index },
    {  4680, PolygonOffsetEXT_remap_index },
-   { 20422, GetPixelTexGenParameterfvSGIS_remap_index },
+   { 20458, GetPixelTexGenParameterfvSGIS_remap_index },
    {  3895, GetPixelTexGenParameterivSGIS_remap_index },
-   { 20155, PixelTexGenParameterfSGIS_remap_index },
+   { 20191, PixelTexGenParameterfSGIS_remap_index },
    {   580, PixelTexGenParameterfvSGIS_remap_index },
    { 11105, PixelTexGenParameteriSGIS_remap_index },
-   { 12092, PixelTexGenParameterivSGIS_remap_index },
-   { 14493, SampleMaskSGIS_remap_index },
-   { 17425, SamplePatternSGIS_remap_index },
-   { 23213, ColorPointerEXT_remap_index },
-   { 15408, EdgeFlagPointerEXT_remap_index },
+   { 12128, PixelTexGenParameterivSGIS_remap_index },
+   { 14529, SampleMaskSGIS_remap_index },
+   { 17461, SamplePatternSGIS_remap_index },
+   { 23249, ColorPointerEXT_remap_index },
+   { 15444, EdgeFlagPointerEXT_remap_index },
    {  5135, IndexPointerEXT_remap_index },
    {  5215, NormalPointerEXT_remap_index },
-   { 13921, TexCoordPointerEXT_remap_index },
+   { 13957, TexCoordPointerEXT_remap_index },
    {  5985, VertexPointerEXT_remap_index },
    {  3165, PointParameterfEXT_remap_index },
    {  6836, PointParameterfvEXT_remap_index },
-   { 28374, LockArraysEXT_remap_index },
-   { 13020, UnlockArraysEXT_remap_index },
+   { 28410, LockArraysEXT_remap_index },
+   { 13056, UnlockArraysEXT_remap_index },
    {  7843, CullParameterdvEXT_remap_index },
    { 10276, CullParameterfvEXT_remap_index },
    {  1151, SecondaryColor3bEXT_remap_index },
    {  6995, SecondaryColor3bvEXT_remap_index },
    {  9184, SecondaryColor3dEXT_remap_index },
-   { 22488, SecondaryColor3dvEXT_remap_index },
-   { 24754, SecondaryColor3fEXT_remap_index },
-   { 15974, SecondaryColor3fvEXT_remap_index },
+   { 22524, SecondaryColor3dvEXT_remap_index },
+   { 24790, SecondaryColor3fEXT_remap_index },
+   { 16010, SecondaryColor3fvEXT_remap_index },
    {   426, SecondaryColor3iEXT_remap_index },
-   { 14231, SecondaryColor3ivEXT_remap_index },
+   { 14267, SecondaryColor3ivEXT_remap_index },
    {  8842, SecondaryColor3sEXT_remap_index },
-   { 27162, SecondaryColor3svEXT_remap_index },
-   { 23927, SecondaryColor3ubEXT_remap_index },
-   { 18684, SecondaryColor3ubvEXT_remap_index },
+   { 27198, SecondaryColor3svEXT_remap_index },
+   { 23963, SecondaryColor3ubEXT_remap_index },
+   { 18720, SecondaryColor3ubvEXT_remap_index },
    { 11300, SecondaryColor3uiEXT_remap_index },
-   { 20042, SecondaryColor3uivEXT_remap_index },
-   { 22735, SecondaryColor3usEXT_remap_index },
+   { 20078, SecondaryColor3uivEXT_remap_index },
+   { 22771, SecondaryColor3usEXT_remap_index },
    { 11373, SecondaryColor3usvEXT_remap_index },
    { 10352, SecondaryColorPointerEXT_remap_index },
-   { 22549, MultiDrawArraysEXT_remap_index },
-   { 18374, MultiDrawElementsEXT_remap_index },
-   { 18569, FogCoordPointerEXT_remap_index },
+   { 22585, MultiDrawArraysEXT_remap_index },
+   { 18410, MultiDrawElementsEXT_remap_index },
+   { 18605, FogCoordPointerEXT_remap_index },
    {  4044, FogCoorddEXT_remap_index },
-   { 27775, FogCoorddvEXT_remap_index },
+   { 27811, FogCoorddvEXT_remap_index },
    {  4105, FogCoordfEXT_remap_index },
-   { 23850, FogCoordfvEXT_remap_index },
-   { 16692, PixelTexGenSGIX_remap_index },
-   { 24435, BlendFuncSeparateEXT_remap_index },
+   { 23886, FogCoordfvEXT_remap_index },
+   { 16728, PixelTexGenSGIX_remap_index },
+   { 24471, BlendFuncSeparateEXT_remap_index },
    {  5897, FlushVertexArrayRangeNV_remap_index },
    {  4629, VertexArrayRangeNV_remap_index },
-   { 24819, CombinerInputNV_remap_index },
+   { 24855, CombinerInputNV_remap_index },
    {  1946, CombinerOutputNV_remap_index },
-   { 27315, CombinerParameterfNV_remap_index },
+   { 27351, CombinerParameterfNV_remap_index },
    {  4549, CombinerParameterfvNV_remap_index },
-   { 19613, CombinerParameteriNV_remap_index },
-   { 28745, CombinerParameterivNV_remap_index },
+   { 19649, CombinerParameteriNV_remap_index },
+   { 28781, CombinerParameterivNV_remap_index },
    {  6273, FinalCombinerInputNV_remap_index },
    {  8671, GetCombinerInputParameterfvNV_remap_index },
-   { 28582, GetCombinerInputParameterivNV_remap_index },
-   { 22878, GetCombinerOutputParameterfvNV_remap_index },
-   { 12021, GetCombinerOutputParameterivNV_remap_index },
+   { 28618, GetCombinerInputParameterivNV_remap_index },
+   { 22914, GetCombinerOutputParameterfvNV_remap_index },
+   { 12057, GetCombinerOutputParameterivNV_remap_index },
    {  5642, GetFinalCombinerInputParameterfvNV_remap_index },
-   { 21781, GetFinalCombinerInputParameterivNV_remap_index },
+   { 21817, GetFinalCombinerInputParameterivNV_remap_index },
    { 11045, ResizeBuffersMESA_remap_index },
    {  9759, WindowPos2dMESA_remap_index },
    {   944, WindowPos2dvMESA_remap_index },
-   { 29573, WindowPos2fMESA_remap_index },
+   { 29609, WindowPos2fMESA_remap_index },
    {  6940, WindowPos2fvMESA_remap_index },
-   { 15921, WindowPos2iMESA_remap_index },
-   { 17898, WindowPos2ivMESA_remap_index },
-   { 18473, WindowPos2sMESA_remap_index },
+   { 15957, WindowPos2iMESA_remap_index },
+   { 17934, WindowPos2ivMESA_remap_index },
+   { 18509, WindowPos2sMESA_remap_index },
    {  4885, WindowPos2svMESA_remap_index },
    {  6765, WindowPos3dMESA_remap_index },
-   { 12300, WindowPos3dvMESA_remap_index },
+   { 12336, WindowPos3dvMESA_remap_index },
    {   472, WindowPos3fMESA_remap_index },
-   { 13081, WindowPos3fvMESA_remap_index },
-   { 21093, WindowPos3iMESA_remap_index },
-   { 26664, WindowPos3ivMESA_remap_index },
-   { 16438, WindowPos3sMESA_remap_index },
-   { 28031, WindowPos3svMESA_remap_index },
+   { 13117, WindowPos3fvMESA_remap_index },
+   { 21129, WindowPos3iMESA_remap_index },
+   { 26700, WindowPos3ivMESA_remap_index },
+   { 16474, WindowPos3sMESA_remap_index },
+   { 28067, WindowPos3svMESA_remap_index },
    {  9710, WindowPos4dMESA_remap_index },
-   { 14934, WindowPos4dvMESA_remap_index },
-   { 12259, WindowPos4fMESA_remap_index },
-   { 27069, WindowPos4fvMESA_remap_index },
-   { 26817, WindowPos4iMESA_remap_index },
+   { 14970, WindowPos4dvMESA_remap_index },
+   { 12295, WindowPos4fMESA_remap_index },
+   { 27105, WindowPos4fvMESA_remap_index },
+   { 26853, WindowPos4iMESA_remap_index },
    { 10848, WindowPos4ivMESA_remap_index },
-   { 16571, WindowPos4sMESA_remap_index },
+   { 16607, WindowPos4sMESA_remap_index },
    {  2857, WindowPos4svMESA_remap_index },
-   { 12060, MultiModeDrawArraysIBM_remap_index },
-   { 25537, MultiModeDrawElementsIBM_remap_index },
+   { 12096, MultiModeDrawArraysIBM_remap_index },
+   { 25573, MultiModeDrawElementsIBM_remap_index },
    { 10698, DeleteFencesNV_remap_index },
-   { 24666, FinishFenceNV_remap_index },
+   { 24702, FinishFenceNV_remap_index },
    {  3287, GenFencesNV_remap_index },
-   { 14914, GetFenceivNV_remap_index },
+   { 14950, GetFenceivNV_remap_index },
    {  7232, IsFenceNV_remap_index },
-   { 11948, SetFenceNV_remap_index },
+   { 11984, SetFenceNV_remap_index },
    {  3744, TestFenceNV_remap_index },
-   { 28002, AreProgramsResidentNV_remap_index },
-   { 27357, BindProgramNV_remap_index },
-   { 22818, DeleteProgramsNV_remap_index },
-   { 18902, ExecuteProgramNV_remap_index },
-   { 29466, GenProgramsNV_remap_index },
-   { 20501, GetProgramParameterdvNV_remap_index },
+   { 28038, AreProgramsResidentNV_remap_index },
+   { 27393, BindProgramNV_remap_index },
+   { 22854, DeleteProgramsNV_remap_index },
+   { 18938, ExecuteProgramNV_remap_index },
+   { 29502, GenProgramsNV_remap_index },
+   { 20537, GetProgramParameterdvNV_remap_index },
    {  9246, GetProgramParameterfvNV_remap_index },
-   { 23187, GetProgramStringNV_remap_index },
-   { 21470, GetProgramivNV_remap_index },
-   { 20735, GetTrackMatrixivNV_remap_index },
-   { 23007, GetVertexAttribPointervNV_remap_index },
-   { 21714, GetVertexAttribdvNV_remap_index },
-   { 16411, GetVertexAttribfvNV_remap_index },
-   { 16102, GetVertexAttribivNV_remap_index },
-   { 16818, IsProgramNV_remap_index },
+   { 23223, GetProgramStringNV_remap_index },
+   { 21506, GetProgramivNV_remap_index },
+   { 20771, GetTrackMatrixivNV_remap_index },
+   { 23043, GetVertexAttribPointervNV_remap_index },
+   { 21750, GetVertexAttribdvNV_remap_index },
+   { 16447, GetVertexAttribfvNV_remap_index },
+   { 16138, GetVertexAttribivNV_remap_index },
+   { 16854, IsProgramNV_remap_index },
    {  8252, LoadProgramNV_remap_index },
-   { 24531, ProgramParameters4dvNV_remap_index },
-   { 21400, ProgramParameters4fvNV_remap_index },
-   { 18202, RequestResidentProgramsNV_remap_index },
-   { 19591, TrackMatrixNV_remap_index },
-   { 28559, VertexAttrib1dNV_remap_index },
-   { 11859, VertexAttrib1dvNV_remap_index },
-   { 25069, VertexAttrib1fNV_remap_index },
+   { 24567, ProgramParameters4dvNV_remap_index },
+   { 21436, ProgramParameters4fvNV_remap_index },
+   { 18238, RequestResidentProgramsNV_remap_index },
+   { 19627, TrackMatrixNV_remap_index },
+   { 28595, VertexAttrib1dNV_remap_index },
+   { 11895, VertexAttrib1dvNV_remap_index },
+   { 25105, VertexAttrib1fNV_remap_index },
    {  2245, VertexAttrib1fvNV_remap_index },
-   { 27126, VertexAttrib1sNV_remap_index },
-   { 13154, VertexAttrib1svNV_remap_index },
+   { 27162, VertexAttrib1sNV_remap_index },
+   { 13190, VertexAttrib1svNV_remap_index },
    {  4220, VertexAttrib2dNV_remap_index },
-   { 11774, VertexAttrib2dvNV_remap_index },
-   { 17657, VertexAttrib2fNV_remap_index },
+   { 11810, VertexAttrib2dvNV_remap_index },
+   { 17693, VertexAttrib2fNV_remap_index },
    { 11421, VertexAttrib2fvNV_remap_index },
    {  5045, VertexAttrib2sNV_remap_index },
-   { 16492, VertexAttrib2svNV_remap_index },
+   { 16528, VertexAttrib2svNV_remap_index },
    {  9907, VertexAttrib3dNV_remap_index },
-   { 28252, VertexAttrib3dvNV_remap_index },
+   { 28288, VertexAttrib3dvNV_remap_index },
    {  9058, VertexAttrib3fNV_remap_index },
-   { 21741, VertexAttrib3fvNV_remap_index },
-   { 25044, VertexAttrib3sNV_remap_index },
-   { 20762, VertexAttrib3svNV_remap_index },
-   { 25511, VertexAttrib4dNV_remap_index },
-   { 29503, VertexAttrib4dvNV_remap_index },
+   { 21777, VertexAttrib3fvNV_remap_index },
+   { 25080, VertexAttrib3sNV_remap_index },
+   { 20798, VertexAttrib3svNV_remap_index },
+   { 25547, VertexAttrib4dNV_remap_index },
+   { 29539, VertexAttrib4dvNV_remap_index },
    {  3945, VertexAttrib4fNV_remap_index },
    {  8302, VertexAttrib4fvNV_remap_index },
-   { 23509, VertexAttrib4sNV_remap_index },
+   { 23545, VertexAttrib4sNV_remap_index },
    {  1293, VertexAttrib4svNV_remap_index },
    {  4378, VertexAttrib4ubNV_remap_index },
    {   734, VertexAttrib4ubvNV_remap_index },
-   { 19082, VertexAttribPointerNV_remap_index },
+   { 19118, VertexAttribPointerNV_remap_index },
    {  2097, VertexAttribs1dvNV_remap_index },
-   { 16516, VertexAttribs1fvNV_remap_index },
-   { 29303, VertexAttribs1svNV_remap_index },
+   { 16552, VertexAttribs1fvNV_remap_index },
+   { 29339, VertexAttribs1svNV_remap_index },
    {  9083, VertexAttribs2dvNV_remap_index },
-   { 22303, VertexAttribs2fvNV_remap_index },
-   { 15434, VertexAttribs2svNV_remap_index },
+   { 22339, VertexAttribs2fvNV_remap_index },
+   { 15470, VertexAttribs2svNV_remap_index },
    {  4577, VertexAttribs3dvNV_remap_index },
    {  1977, VertexAttribs3fvNV_remap_index },
-   { 26412, VertexAttribs3svNV_remap_index },
-   { 23599, VertexAttribs4dvNV_remap_index },
+   { 26448, VertexAttribs3svNV_remap_index },
+   { 23635, VertexAttribs4dvNV_remap_index },
    {  4603, VertexAttribs4fvNV_remap_index },
-   { 29090, VertexAttribs4svNV_remap_index },
-   { 26160, VertexAttribs4ubvNV_remap_index },
-   { 23669, GetTexBumpParameterfvATI_remap_index },
-   { 29344, GetTexBumpParameterivATI_remap_index },
-   { 16156, TexBumpParameterfvATI_remap_index },
-   { 18073, TexBumpParameterivATI_remap_index },
-   { 13700, AlphaFragmentOp1ATI_remap_index },
+   { 29126, VertexAttribs4svNV_remap_index },
+   { 26196, VertexAttribs4ubvNV_remap_index },
+   { 23705, GetTexBumpParameterfvATI_remap_index },
+   { 29380, GetTexBumpParameterivATI_remap_index },
+   { 16192, TexBumpParameterfvATI_remap_index },
+   { 18109, TexBumpParameterivATI_remap_index },
+   { 13736, AlphaFragmentOp1ATI_remap_index },
    {  9598, AlphaFragmentOp2ATI_remap_index },
-   { 21657, AlphaFragmentOp3ATI_remap_index },
-   { 26339, BeginFragmentShaderATI_remap_index },
-   { 27556, BindFragmentShaderATI_remap_index },
-   { 20891, ColorFragmentOp1ATI_remap_index },
+   { 21693, AlphaFragmentOp3ATI_remap_index },
+   { 26375, BeginFragmentShaderATI_remap_index },
+   { 27592, BindFragmentShaderATI_remap_index },
+   { 20927, ColorFragmentOp1ATI_remap_index },
    {  3823, ColorFragmentOp2ATI_remap_index },
-   { 27897, ColorFragmentOp3ATI_remap_index },
+   { 27933, ColorFragmentOp3ATI_remap_index },
    {  4722, DeleteFragmentShaderATI_remap_index },
-   { 29527, EndFragmentShaderATI_remap_index },
-   { 28773, GenFragmentShadersATI_remap_index },
-   { 22407, PassTexCoordATI_remap_index },
+   { 29563, EndFragmentShaderATI_remap_index },
+   { 28809, GenFragmentShadersATI_remap_index },
+   { 22443, PassTexCoordATI_remap_index },
    {  5965, SampleMapATI_remap_index },
    {  5738, SetFragmentShaderConstantATI_remap_index },
    {   319, PointParameteriNV_remap_index },
-   { 12461, PointParameterivNV_remap_index },
-   { 25350, ActiveStencilFaceEXT_remap_index },
-   { 24191, BindVertexArrayAPPLE_remap_index },
+   { 12497, PointParameterivNV_remap_index },
+   { 25386, ActiveStencilFaceEXT_remap_index },
+   { 24227, BindVertexArrayAPPLE_remap_index },
    {  2522, DeleteVertexArraysAPPLE_remap_index },
-   { 15773, GenVertexArraysAPPLE_remap_index },
-   { 20566, IsVertexArrayAPPLE_remap_index },
+   { 15809, GenVertexArraysAPPLE_remap_index },
+   { 20602, IsVertexArrayAPPLE_remap_index },
    {   775, GetProgramNamedParameterdvNV_remap_index },
    {  3128, GetProgramNamedParameterfvNV_remap_index },
-   { 23700, ProgramNamedParameter4dNV_remap_index },
-   { 12736, ProgramNamedParameter4dvNV_remap_index },
+   { 23736, ProgramNamedParameter4dNV_remap_index },
+   { 12772, ProgramNamedParameter4dvNV_remap_index },
    {  7868, ProgramNamedParameter4fNV_remap_index },
    { 10317, ProgramNamedParameter4fvNV_remap_index },
-   { 21379, DepthBoundsEXT_remap_index },
+   { 21415, DepthBoundsEXT_remap_index },
    {  1043, BlendEquationSeparateEXT_remap_index },
-   { 12855, BindFramebufferEXT_remap_index },
-   { 22594, BindRenderbufferEXT_remap_index },
+   { 12891, BindFramebufferEXT_remap_index },
+   { 22630, BindRenderbufferEXT_remap_index },
    {  8521, CheckFramebufferStatusEXT_remap_index },
-   { 19882, DeleteFramebuffersEXT_remap_index },
-   { 28154, DeleteRenderbuffersEXT_remap_index },
-   { 11798, FramebufferRenderbufferEXT_remap_index },
-   { 11965, FramebufferTexture1DEXT_remap_index },
+   { 19918, DeleteFramebuffersEXT_remap_index },
+   { 28190, DeleteRenderbuffersEXT_remap_index },
+   { 11834, FramebufferRenderbufferEXT_remap_index },
+   { 12001, FramebufferTexture1DEXT_remap_index },
    { 10145, FramebufferTexture2DEXT_remap_index },
    {  9812, FramebufferTexture3DEXT_remap_index },
-   { 20458, GenFramebuffersEXT_remap_index },
-   { 15320, GenRenderbuffersEXT_remap_index },
+   { 20494, GenFramebuffersEXT_remap_index },
+   { 15356, GenRenderbuffersEXT_remap_index },
    {  5684, GenerateMipmapEXT_remap_index },
-   { 19113, GetFramebufferAttachmentParameterivEXT_remap_index },
-   { 28679, GetRenderbufferParameterivEXT_remap_index },
-   { 17953, IsFramebufferEXT_remap_index },
-   { 29426, IsRenderbufferEXT_remap_index },
+   { 19149, GetFramebufferAttachmentParameterivEXT_remap_index },
+   { 28715, GetRenderbufferParameterivEXT_remap_index },
+   { 17989, IsFramebufferEXT_remap_index },
+   { 29462, IsRenderbufferEXT_remap_index },
    {  7179, RenderbufferStorageEXT_remap_index },
    {   651, BlitFramebufferEXT_remap_index },
-   { 12555, BufferParameteriAPPLE_remap_index },
-   { 16850, FlushMappedBufferRangeAPPLE_remap_index },
+   { 12591, BufferParameteriAPPLE_remap_index },
+   { 16886, FlushMappedBufferRangeAPPLE_remap_index },
    {  2701, FramebufferTextureLayerEXT_remap_index },
-   { 26061, ProvokingVertexEXT_remap_index },
+   { 26097, ProvokingVertexEXT_remap_index },
    {  9467, GetTexParameterPointervAPPLE_remap_index },
    {  4405, TextureRangeAPPLE_remap_index },
-   { 25376, StencilFuncSeparateATI_remap_index },
-   { 15840, ProgramEnvParameters4fvEXT_remap_index },
-   { 15058, ProgramLocalParameters4fvEXT_remap_index },
-   { 12389, GetQueryObjecti64vEXT_remap_index },
+   { 25412, StencilFuncSeparateATI_remap_index },
+   { 15876, ProgramEnvParameters4fvEXT_remap_index },
+   { 15094, ProgramLocalParameters4fvEXT_remap_index },
+   { 12425, GetQueryObjecti64vEXT_remap_index },
    {  9109, GetQueryObjectui64vEXT_remap_index },
    {    -1, -1 }
 };
@@ -4748,8 +4749,8 @@ static const struct gl_function_remap MESA_alt_functions[] = {
    /* from GL_EXT_blend_minmax */
    {  9869, _gloffset_BlendEquation },
    /* from GL_EXT_color_subtable */
-   { 14956, _gloffset_ColorSubTable },
-   { 28086, _gloffset_CopyColorSubTable },
+   { 14992, _gloffset_ColorSubTable },
+   { 28122, _gloffset_CopyColorSubTable },
    /* from GL_EXT_convolution */
    {   213, _gloffset_ConvolutionFilter1D },
    {  2284, _gloffset_CopyConvolutionFilter1D },
@@ -4757,62 +4758,62 @@ static const struct gl_function_remap MESA_alt_functions[] = {
    {  7528, _gloffset_ConvolutionFilter2D },
    {  7694, _gloffset_ConvolutionParameteriv },
    {  8154, _gloffset_ConvolutionParameterfv },
-   { 18101, _gloffset_GetSeparableFilter },
-   { 21147, _gloffset_SeparableFilter2D },
-   { 21959, _gloffset_ConvolutionParameteri },
-   { 22082, _gloffset_ConvolutionParameterf },
-   { 23535, _gloffset_GetConvolutionParameterfv },
-   { 24357, _gloffset_GetConvolutionFilter },
-   { 26601, _gloffset_CopyConvolutionFilter2D },
+   { 18137, _gloffset_GetSeparableFilter },
+   { 21183, _gloffset_SeparableFilter2D },
+   { 21995, _gloffset_ConvolutionParameteri },
+   { 22118, _gloffset_ConvolutionParameterf },
+   { 23571, _gloffset_GetConvolutionParameterfv },
+   { 24393, _gloffset_GetConvolutionFilter },
+   { 26637, _gloffset_CopyConvolutionFilter2D },
    /* from GL_EXT_copy_texture */
-   { 13214, _gloffset_CopyTexSubImage3D },
-   { 14696, _gloffset_CopyTexImage2D },
-   { 21567, _gloffset_CopyTexImage1D },
-   { 24038, _gloffset_CopyTexSubImage2D },
-   { 26239, _gloffset_CopyTexSubImage1D },
+   { 13250, _gloffset_CopyTexSubImage3D },
+   { 14732, _gloffset_CopyTexImage2D },
+   { 21603, _gloffset_CopyTexImage1D },
+   { 24074, _gloffset_CopyTexSubImage2D },
+   { 26275, _gloffset_CopyTexSubImage1D },
    /* from GL_EXT_draw_range_elements */
    {  8408, _gloffset_DrawRangeElements },
    /* from GL_EXT_histogram */
    {   812, _gloffset_Histogram },
    {  3088, _gloffset_ResetHistogram },
    {  8780, _gloffset_GetMinmax },
-   { 13548, _gloffset_GetHistogramParameterfv },
-   { 21492, _gloffset_GetMinmaxParameteriv },
-   { 23425, _gloffset_ResetMinmax },
-   { 24254, _gloffset_GetHistogramParameteriv },
-   { 25310, _gloffset_GetHistogram },
-   { 27672, _gloffset_Minmax },
-   { 29173, _gloffset_GetMinmaxParameterfv },
+   { 13584, _gloffset_GetHistogramParameterfv },
+   { 21528, _gloffset_GetMinmaxParameteriv },
+   { 23461, _gloffset_ResetMinmax },
+   { 24290, _gloffset_GetHistogramParameteriv },
+   { 25346, _gloffset_GetHistogram },
+   { 27708, _gloffset_Minmax },
+   { 29209, _gloffset_GetMinmaxParameterfv },
    /* from GL_EXT_paletted_texture */
    {  7390, _gloffset_ColorTable },
-   { 13394, _gloffset_GetColorTable },
-   { 20205, _gloffset_GetColorTableParameterfv },
-   { 22138, _gloffset_GetColorTableParameteriv },
+   { 13430, _gloffset_GetColorTable },
+   { 20241, _gloffset_GetColorTableParameterfv },
+   { 22174, _gloffset_GetColorTableParameteriv },
    /* from GL_EXT_subtexture */
    {  6111, _gloffset_TexSubImage1D },
    {  9394, _gloffset_TexSubImage2D },
    /* from GL_EXT_texture3D */
    {  1658, _gloffset_TexImage3D },
-   { 19974, _gloffset_TexSubImage3D },
+   { 20010, _gloffset_TexSubImage3D },
    /* from GL_EXT_texture_object */
    {  2964, _gloffset_PrioritizeTextures },
    {  6560, _gloffset_AreTexturesResident },
-   { 11883, _gloffset_GenTextures },
-   { 13880, _gloffset_DeleteTextures },
-   { 17131, _gloffset_IsTexture },
-   { 26304, _gloffset_BindTexture },
+   { 11919, _gloffset_GenTextures },
+   { 13916, _gloffset_DeleteTextures },
+   { 17167, _gloffset_IsTexture },
+   { 26340, _gloffset_BindTexture },
    /* from GL_EXT_vertex_array */
-   { 21319, _gloffset_ArrayElement },
-   { 27260, _gloffset_GetPointerv },
-   { 28800, _gloffset_DrawArrays },
+   { 21355, _gloffset_ArrayElement },
+   { 27296, _gloffset_GetPointerv },
+   { 28836, _gloffset_DrawArrays },
    /* from GL_SGI_color_table */
    {  6678, _gloffset_ColorTableParameteriv },
    {  7390, _gloffset_ColorTable },
-   { 13394, _gloffset_GetColorTable },
-   { 13504, _gloffset_CopyColorTable },
-   { 16992, _gloffset_ColorTableParameterfv },
-   { 20205, _gloffset_GetColorTableParameterfv },
-   { 22138, _gloffset_GetColorTableParameteriv },
+   { 13430, _gloffset_GetColorTable },
+   { 13540, _gloffset_CopyColorTable },
+   { 17028, _gloffset_ColorTableParameterfv },
+   { 20241, _gloffset_GetColorTableParameterfv },
+   { 22174, _gloffset_GetColorTableParameteriv },
    /* from GL_VERSION_1_3 */
    {   381, _gloffset_MultiTexCoord3sARB },
    {   613, _gloffset_ActiveTextureARB },
@@ -4827,27 +4828,27 @@ static const struct gl_function_remap MESA_alt_functions[] = {
    { 10613, _gloffset_MultiTexCoord1svARB },
    { 10906, _gloffset_MultiTexCoord3svARB },
    { 10967, _gloffset_MultiTexCoord4iARB },
-   { 11654, _gloffset_MultiTexCoord3iARB },
-   { 12418, _gloffset_MultiTexCoord1dARB },
-   { 12584, _gloffset_MultiTexCoord3dvARB },
-   { 13748, _gloffset_MultiTexCoord3ivARB },
-   { 13793, _gloffset_MultiTexCoord2sARB },
-   { 15013, _gloffset_MultiTexCoord4ivARB },
-   { 16642, _gloffset_ClientActiveTextureARB },
-   { 18858, _gloffset_MultiTexCoord2dARB },
-   { 19233, _gloffset_MultiTexCoord4dvARB },
-   { 19519, _gloffset_MultiTexCoord4fvARB },
-   { 20346, _gloffset_MultiTexCoord3fARB },
-   { 22639, _gloffset_MultiTexCoord4dARB },
-   { 22917, _gloffset_MultiTexCoord1sARB },
-   { 23095, _gloffset_MultiTexCoord1dvARB },
-   { 23882, _gloffset_MultiTexCoord1ivARB },
-   { 23975, _gloffset_MultiTexCoord2ivARB },
-   { 24314, _gloffset_MultiTexCoord1iARB },
-   { 25585, _gloffset_MultiTexCoord4svARB },
-   { 26103, _gloffset_MultiTexCoord1fARB },
-   { 26366, _gloffset_MultiTexCoord4fARB },
-   { 28634, _gloffset_MultiTexCoord2fvARB },
+   { 11690, _gloffset_MultiTexCoord3iARB },
+   { 12454, _gloffset_MultiTexCoord1dARB },
+   { 12620, _gloffset_MultiTexCoord3dvARB },
+   { 13784, _gloffset_MultiTexCoord3ivARB },
+   { 13829, _gloffset_MultiTexCoord2sARB },
+   { 15049, _gloffset_MultiTexCoord4ivARB },
+   { 16678, _gloffset_ClientActiveTextureARB },
+   { 18894, _gloffset_MultiTexCoord2dARB },
+   { 19269, _gloffset_MultiTexCoord4dvARB },
+   { 19555, _gloffset_MultiTexCoord4fvARB },
+   { 20382, _gloffset_MultiTexCoord3fARB },
+   { 22675, _gloffset_MultiTexCoord4dARB },
+   { 22953, _gloffset_MultiTexCoord1sARB },
+   { 23131, _gloffset_MultiTexCoord1dvARB },
+   { 23918, _gloffset_MultiTexCoord1ivARB },
+   { 24011, _gloffset_MultiTexCoord2ivARB },
+   { 24350, _gloffset_MultiTexCoord1iARB },
+   { 25621, _gloffset_MultiTexCoord4svARB },
+   { 26139, _gloffset_MultiTexCoord1fARB },
+   { 26402, _gloffset_MultiTexCoord4fARB },
+   { 28670, _gloffset_MultiTexCoord2fvARB },
    {    -1, -1 }
 };
 
@@ -4928,9 +4929,9 @@ static const struct gl_function_remap GL_ARB_map_buffer_range_functions[] = {
 static const struct gl_function_remap GL_ARB_matrix_palette_functions[] = {
    {  3339, -1 }, /* MatrixIndexusvARB */
    { 11511, -1 }, /* MatrixIndexuivARB */
-   { 12706, -1 }, /* MatrixIndexPointerARB */
-   { 17380, -1 }, /* CurrentPaletteMatrixARB */
-   { 20090, -1 }, /* MatrixIndexubvARB */
+   { 12742, -1 }, /* MatrixIndexPointerARB */
+   { 17416, -1 }, /* CurrentPaletteMatrixARB */
+   { 20126, -1 }, /* MatrixIndexubvARB */
    {    -1, -1 }
 };
 #endif
@@ -5003,13 +5004,13 @@ static const struct gl_function_remap GL_ARB_vertex_blend_functions[] = {
    {  2226, -1 }, /* WeightubvARB */
    {  5572, -1 }, /* WeightivARB */
    {  9734, -1 }, /* WeightPointerARB */
-   { 12175, -1 }, /* WeightfvARB */
-   { 15460, -1 }, /* WeightbvARB */
-   { 18526, -1 }, /* WeightusvARB */
-   { 21073, -1 }, /* VertexBlendARB */
-   { 26187, -1 }, /* WeightsvARB */
-   { 28136, -1 }, /* WeightdvARB */
-   { 28834, -1 }, /* WeightuivARB */
+   { 12211, -1 }, /* WeightfvARB */
+   { 15496, -1 }, /* WeightbvARB */
+   { 18562, -1 }, /* WeightusvARB */
+   { 21109, -1 }, /* VertexBlendARB */
+   { 26223, -1 }, /* WeightsvARB */
+   { 28172, -1 }, /* WeightdvARB */
+   { 28870, -1 }, /* WeightuivARB */
    {    -1, -1 }
 };
 #endif
@@ -5107,8 +5108,8 @@ static const struct gl_function_remap GL_EXT_blend_minmax_functions[] = {
 
 #if defined(need_GL_EXT_color_subtable)
 static const struct gl_function_remap GL_EXT_color_subtable_functions[] = {
-   { 14956, _gloffset_ColorSubTable },
-   { 28086, _gloffset_CopyColorSubTable },
+   { 14992, _gloffset_ColorSubTable },
+   { 28122, _gloffset_CopyColorSubTable },
    {    -1, -1 }
 };
 #endif
@@ -5128,13 +5129,13 @@ static const struct gl_function_remap GL_EXT_convolution_functions[] = {
    {  7528, _gloffset_ConvolutionFilter2D },
    {  7694, _gloffset_ConvolutionParameteriv },
    {  8154, _gloffset_ConvolutionParameterfv },
-   { 18101, _gloffset_GetSeparableFilter },
-   { 21147, _gloffset_SeparableFilter2D },
-   { 21959, _gloffset_ConvolutionParameteri },
-   { 22082, _gloffset_ConvolutionParameterf },
-   { 23535, _gloffset_GetConvolutionParameterfv },
-   { 24357, _gloffset_GetConvolutionFilter },
-   { 26601, _gloffset_CopyConvolutionFilter2D },
+   { 18137, _gloffset_GetSeparableFilter },
+   { 21183, _gloffset_SeparableFilter2D },
+   { 21995, _gloffset_ConvolutionParameteri },
+   { 22118, _gloffset_ConvolutionParameterf },
+   { 23571, _gloffset_GetConvolutionParameterfv },
+   { 24393, _gloffset_GetConvolutionFilter },
+   { 26637, _gloffset_CopyConvolutionFilter2D },
    {    -1, -1 }
 };
 #endif
@@ -5143,37 +5144,37 @@ static const struct gl_function_remap GL_EXT_convolution_functions[] = {
 static const struct gl_function_remap GL_EXT_coordinate_frame_functions[] = {
    {  9278, -1 }, /* TangentPointerEXT */
    { 11025, -1 }, /* Binormal3ivEXT */
-   { 11607, -1 }, /* Tangent3sEXT */
-   { 12771, -1 }, /* Tangent3fvEXT */
-   { 16392, -1 }, /* Tangent3dvEXT */
-   { 17078, -1 }, /* Binormal3bvEXT */
-   { 18154, -1 }, /* Binormal3dEXT */
-   { 20022, -1 }, /* Tangent3fEXT */
-   { 22031, -1 }, /* Binormal3sEXT */
-   { 22449, -1 }, /* Tangent3ivEXT */
-   { 22468, -1 }, /* Tangent3dEXT */
-   { 23322, -1 }, /* Binormal3svEXT */
-   { 23780, -1 }, /* Binormal3fEXT */
-   { 24632, -1 }, /* Binormal3dvEXT */
-   { 25807, -1 }, /* Tangent3iEXT */
-   { 26886, -1 }, /* Tangent3bvEXT */
-   { 27295, -1 }, /* Tangent3bEXT */
-   { 27859, -1 }, /* Binormal3fvEXT */
-   { 28533, -1 }, /* BinormalPointerEXT */
-   { 28938, -1 }, /* Tangent3svEXT */
-   { 29375, -1 }, /* Binormal3bEXT */
-   { 29552, -1 }, /* Binormal3iEXT */
+   { 11643, -1 }, /* Tangent3sEXT */
+   { 12807, -1 }, /* Tangent3fvEXT */
+   { 16428, -1 }, /* Tangent3dvEXT */
+   { 17114, -1 }, /* Binormal3bvEXT */
+   { 18190, -1 }, /* Binormal3dEXT */
+   { 20058, -1 }, /* Tangent3fEXT */
+   { 22067, -1 }, /* Binormal3sEXT */
+   { 22485, -1 }, /* Tangent3ivEXT */
+   { 22504, -1 }, /* Tangent3dEXT */
+   { 23358, -1 }, /* Binormal3svEXT */
+   { 23816, -1 }, /* Binormal3fEXT */
+   { 24668, -1 }, /* Binormal3dvEXT */
+   { 25843, -1 }, /* Tangent3iEXT */
+   { 26922, -1 }, /* Tangent3bvEXT */
+   { 27331, -1 }, /* Tangent3bEXT */
+   { 27895, -1 }, /* Binormal3fvEXT */
+   { 28569, -1 }, /* BinormalPointerEXT */
+   { 28974, -1 }, /* Tangent3svEXT */
+   { 29411, -1 }, /* Binormal3bEXT */
+   { 29588, -1 }, /* Binormal3iEXT */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_EXT_copy_texture)
 static const struct gl_function_remap GL_EXT_copy_texture_functions[] = {
-   { 13214, _gloffset_CopyTexSubImage3D },
-   { 14696, _gloffset_CopyTexImage2D },
-   { 21567, _gloffset_CopyTexImage1D },
-   { 24038, _gloffset_CopyTexSubImage2D },
-   { 26239, _gloffset_CopyTexSubImage1D },
+   { 13250, _gloffset_CopyTexSubImage3D },
+   { 14732, _gloffset_CopyTexImage2D },
+   { 21603, _gloffset_CopyTexImage1D },
+   { 24074, _gloffset_CopyTexSubImage2D },
+   { 26275, _gloffset_CopyTexSubImage1D },
    {    -1, -1 }
 };
 #endif
@@ -5213,6 +5214,13 @@ static const struct gl_function_remap GL_EXT_framebuffer_blit_functions[] = {
 };
 #endif
 
+#if defined(need_GL_EXT_framebuffer_multisample)
+/* functions defined in MESA_remap_table_functions are excluded */
+static const struct gl_function_remap GL_EXT_framebuffer_multisample_functions[] = {
+   {    -1, -1 }
+};
+#endif
+
 #if defined(need_GL_EXT_framebuffer_object)
 /* functions defined in MESA_remap_table_functions are excluded */
 static const struct gl_function_remap GL_EXT_framebuffer_object_functions[] = {
@@ -5232,13 +5240,13 @@ static const struct gl_function_remap GL_EXT_histogram_functions[] = {
    {   812, _gloffset_Histogram },
    {  3088, _gloffset_ResetHistogram },
    {  8780, _gloffset_GetMinmax },
-   { 13548, _gloffset_GetHistogramParameterfv },
-   { 21492, _gloffset_GetMinmaxParameteriv },
-   { 23425, _gloffset_ResetMinmax },
-   { 24254, _gloffset_GetHistogramParameteriv },
-   { 25310, _gloffset_GetHistogram },
-   { 27672, _gloffset_Minmax },
-   { 29173, _gloffset_GetMinmaxParameterfv },
+   { 13584, _gloffset_GetHistogramParameterfv },
+   { 21528, _gloffset_GetMinmaxParameteriv },
+   { 23461, _gloffset_ResetMinmax },
+   { 24290, _gloffset_GetHistogramParameteriv },
+   { 25346, _gloffset_GetHistogram },
+   { 27708, _gloffset_Minmax },
+   { 29209, _gloffset_GetMinmaxParameterfv },
    {    -1, -1 }
 };
 #endif
@@ -5252,16 +5260,16 @@ static const struct gl_function_remap GL_EXT_index_func_functions[] = {
 
 #if defined(need_GL_EXT_index_material)
 static const struct gl_function_remap GL_EXT_index_material_functions[] = {
-   { 18613, -1 }, /* IndexMaterialEXT */
+   { 18649, -1 }, /* IndexMaterialEXT */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_EXT_light_texture)
 static const struct gl_function_remap GL_EXT_light_texture_functions[] = {
-   { 23342, -1 }, /* ApplyTextureEXT */
-   { 23379, -1 }, /* TextureMaterialEXT */
-   { 23404, -1 }, /* TextureLightEXT */
+   { 23378, -1 }, /* ApplyTextureEXT */
+   { 23415, -1 }, /* TextureMaterialEXT */
+   { 23440, -1 }, /* TextureLightEXT */
    {    -1, -1 }
 };
 #endif
@@ -5283,9 +5291,9 @@ static const struct gl_function_remap GL_EXT_multisample_functions[] = {
 #if defined(need_GL_EXT_paletted_texture)
 static const struct gl_function_remap GL_EXT_paletted_texture_functions[] = {
    {  7390, _gloffset_ColorTable },
-   { 13394, _gloffset_GetColorTable },
-   { 20205, _gloffset_GetColorTableParameterfv },
-   { 22138, _gloffset_GetColorTableParameteriv },
+   { 13430, _gloffset_GetColorTable },
+   { 20241, _gloffset_GetColorTableParameterfv },
+   { 22174, _gloffset_GetColorTableParameteriv },
    {    -1, -1 }
 };
 #endif
@@ -5293,9 +5301,9 @@ static const struct gl_function_remap GL_EXT_paletted_texture_functions[] = {
 #if defined(need_GL_EXT_pixel_transform)
 static const struct gl_function_remap GL_EXT_pixel_transform_functions[] = {
    {  9519, -1 }, /* PixelTransformParameterfvEXT */
-   { 19198, -1 }, /* PixelTransformParameterfEXT */
-   { 19278, -1 }, /* PixelTransformParameteriEXT */
-   { 28497, -1 }, /* PixelTransformParameterivEXT */
+   { 19234, -1 }, /* PixelTransformParameterfEXT */
+   { 19314, -1 }, /* PixelTransformParameteriEXT */
+   { 28533, -1 }, /* PixelTransformParameterivEXT */
    {    -1, -1 }
 };
 #endif
@@ -5346,7 +5354,7 @@ static const struct gl_function_remap GL_EXT_subtexture_functions[] = {
 #if defined(need_GL_EXT_texture3D)
 static const struct gl_function_remap GL_EXT_texture3D_functions[] = {
    {  1658, _gloffset_TexImage3D },
-   { 19974, _gloffset_TexSubImage3D },
+   { 20010, _gloffset_TexSubImage3D },
    {    -1, -1 }
 };
 #endif
@@ -5362,17 +5370,17 @@ static const struct gl_function_remap GL_EXT_texture_array_functions[] = {
 static const struct gl_function_remap GL_EXT_texture_object_functions[] = {
    {  2964, _gloffset_PrioritizeTextures },
    {  6560, _gloffset_AreTexturesResident },
-   { 11883, _gloffset_GenTextures },
-   { 13880, _gloffset_DeleteTextures },
-   { 17131, _gloffset_IsTexture },
-   { 26304, _gloffset_BindTexture },
+   { 11919, _gloffset_GenTextures },
+   { 13916, _gloffset_DeleteTextures },
+   { 17167, _gloffset_IsTexture },
+   { 26340, _gloffset_BindTexture },
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_EXT_texture_perturb_normal)
 static const struct gl_function_remap GL_EXT_texture_perturb_normal_functions[] = {
-   { 12125, -1 }, /* TextureNormalEXT */
+   { 12161, -1 }, /* TextureNormalEXT */
    {    -1, -1 }
 };
 #endif
@@ -5387,18 +5395,18 @@ static const struct gl_function_remap GL_EXT_timer_query_functions[] = {
 #if defined(need_GL_EXT_vertex_array)
 /* functions defined in MESA_remap_table_functions are excluded */
 static const struct gl_function_remap GL_EXT_vertex_array_functions[] = {
-   { 21319, _gloffset_ArrayElement },
-   { 27260, _gloffset_GetPointerv },
-   { 28800, _gloffset_DrawArrays },
+   { 21355, _gloffset_ArrayElement },
+   { 27296, _gloffset_GetPointerv },
+   { 28836, _gloffset_DrawArrays },
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_EXT_vertex_weighting)
 static const struct gl_function_remap GL_EXT_vertex_weighting_functions[] = {
-   { 17161, -1 }, /* VertexWeightfvEXT */
-   { 23758, -1 }, /* VertexWeightfEXT */
-   { 25279, -1 }, /* VertexWeightPointerEXT */
+   { 17197, -1 }, /* VertexWeightfvEXT */
+   { 23794, -1 }, /* VertexWeightfEXT */
+   { 25315, -1 }, /* VertexWeightPointerEXT */
    {    -1, -1 }
 };
 #endif
@@ -5410,7 +5418,7 @@ static const struct gl_function_remap GL_HP_image_transform_functions[] = {
    {  8972, -1 }, /* ImageTransformParameterfvHP */
    { 10531, -1 }, /* ImageTransformParameteriHP */
    { 10760, -1 }, /* GetImageTransformParameterivHP */
-   { 17225, -1 }, /* ImageTransformParameterivHP */
+   { 17261, -1 }, /* ImageTransformParameterivHP */
    {    -1, -1 }
 };
 #endif
@@ -5429,9 +5437,9 @@ static const struct gl_function_remap GL_IBM_vertex_array_lists_functions[] = {
    {  6734, -1 }, /* FogCoordPointerListIBM */
    {  7041, -1 }, /* VertexPointerListIBM */
    { 10452, -1 }, /* ColorPointerListIBM */
-   { 11714, -1 }, /* TexCoordPointerListIBM */
-   { 12147, -1 }, /* IndexPointerListIBM */
-   { 29116, -1 }, /* EdgeFlagPointerListIBM */
+   { 11750, -1 }, /* TexCoordPointerListIBM */
+   { 12183, -1 }, /* IndexPointerListIBM */
+   { 29152, -1 }, /* EdgeFlagPointerListIBM */
    {    -1, -1 }
 };
 #endif
@@ -5446,9 +5454,9 @@ static const struct gl_function_remap GL_INGR_blend_func_separate_functions[] =
 #if defined(need_GL_INTEL_parallel_arrays)
 static const struct gl_function_remap GL_INTEL_parallel_arrays_functions[] = {
    { 11137, -1 }, /* VertexPointervINTEL */
-   { 13641, -1 }, /* ColorPointervINTEL */
-   { 26575, -1 }, /* NormalPointervINTEL */
-   { 27001, -1 }, /* TexCoordPointervINTEL */
+   { 13677, -1 }, /* ColorPointervINTEL */
+   { 26611, -1 }, /* NormalPointervINTEL */
+   { 27037, -1 }, /* TexCoordPointervINTEL */
    {    -1, -1 }
 };
 #endif
@@ -5465,7 +5473,7 @@ static const struct gl_function_remap GL_MESA_shader_debug_functions[] = {
    {  1522, -1 }, /* GetDebugLogLengthMESA */
    {  3063, -1 }, /* ClearDebugLogMESA */
    {  4018, -1 }, /* GetDebugLogMESA */
-   { 27453, -1 }, /* CreateDebugObjectMESA */
+   { 27489, -1 }, /* CreateDebugObjectMESA */
    {    -1, -1 }
 };
 #endif
@@ -5483,11 +5491,11 @@ static const struct gl_function_remap GL_NV_evaluators_functions[] = {
    {  7496, -1 }, /* MapControlPointsNV */
    {  7595, -1 }, /* MapParameterfvNV */
    {  9377, -1 }, /* EvalMapsNV */
-   { 15130, -1 }, /* GetMapAttribParameterfvNV */
-   { 15296, -1 }, /* MapParameterivNV */
-   { 21882, -1 }, /* GetMapParameterivNV */
-   { 22380, -1 }, /* GetMapParameterfvNV */
-   { 25911, -1 }, /* GetMapControlPointsNV */
+   { 15166, -1 }, /* GetMapAttribParameterfvNV */
+   { 15332, -1 }, /* MapParameterivNV */
+   { 21918, -1 }, /* GetMapParameterivNV */
+   { 22416, -1 }, /* GetMapParameterfvNV */
+   { 25947, -1 }, /* GetMapControlPointsNV */
    {    -1, -1 }
 };
 #endif
@@ -5522,8 +5530,8 @@ static const struct gl_function_remap GL_NV_register_combiners_functions[] = {
 
 #if defined(need_GL_NV_register_combiners2)
 static const struct gl_function_remap GL_NV_register_combiners2_functions[] = {
-   { 14033, -1 }, /* CombinerStageParameterfvNV */
-   { 14348, -1 }, /* GetCombinerStageParameterfvNV */
+   { 14069, -1 }, /* CombinerStageParameterfvNV */
+   { 14384, -1 }, /* GetCombinerStageParameterfvNV */
    {    -1, -1 }
 };
 #endif
@@ -5551,16 +5559,16 @@ static const struct gl_function_remap GL_PGI_misc_hints_functions[] = {
 
 #if defined(need_GL_SGIS_detail_texture)
 static const struct gl_function_remap GL_SGIS_detail_texture_functions[] = {
-   { 14321, -1 }, /* GetDetailTexFuncSGIS */
-   { 14641, -1 }, /* DetailTexFuncSGIS */
+   { 14357, -1 }, /* GetDetailTexFuncSGIS */
+   { 14677, -1 }, /* DetailTexFuncSGIS */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIS_fog_function)
 static const struct gl_function_remap GL_SGIS_fog_function_functions[] = {
-   { 24020, -1 }, /* FogFuncSGIS */
-   { 24685, -1 }, /* GetFogFuncSGIS */
+   { 24056, -1 }, /* FogFuncSGIS */
+   { 24721, -1 }, /* GetFogFuncSGIS */
    {    -1, -1 }
 };
 #endif
@@ -5589,7 +5597,7 @@ static const struct gl_function_remap GL_SGIS_point_parameters_functions[] = {
 #if defined(need_GL_SGIS_sharpen_texture)
 static const struct gl_function_remap GL_SGIS_sharpen_texture_functions[] = {
    {  5834, -1 }, /* GetSharpenTexFuncSGIS */
-   { 19493, -1 }, /* SharpenTexFuncSGIS */
+   { 19529, -1 }, /* SharpenTexFuncSGIS */
    {    -1, -1 }
 };
 #endif
@@ -5597,14 +5605,14 @@ static const struct gl_function_remap GL_SGIS_sharpen_texture_functions[] = {
 #if defined(need_GL_SGIS_texture4D)
 static const struct gl_function_remap GL_SGIS_texture4D_functions[] = {
    {   894, -1 }, /* TexImage4DSGIS */
-   { 13949, -1 }, /* TexSubImage4DSGIS */
+   { 13985, -1 }, /* TexSubImage4DSGIS */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIS_texture_color_mask)
 static const struct gl_function_remap GL_SGIS_texture_color_mask_functions[] = {
-   { 13347, -1 }, /* TextureColorMaskSGIS */
+   { 13383, -1 }, /* TextureColorMaskSGIS */
    {    -1, -1 }
 };
 #endif
@@ -5612,7 +5620,7 @@ static const struct gl_function_remap GL_SGIS_texture_color_mask_functions[] = {
 #if defined(need_GL_SGIS_texture_filter4)
 static const struct gl_function_remap GL_SGIS_texture_filter4_functions[] = {
    {  6011, -1 }, /* GetTexFilterFuncSGIS */
-   { 14467, -1 }, /* TexFilterFuncSGIS */
+   { 14503, -1 }, /* TexFilterFuncSGIS */
    {    -1, -1 }
 };
 #endif
@@ -5622,9 +5630,9 @@ static const struct gl_function_remap GL_SGIX_async_functions[] = {
    {  3014, -1 }, /* AsyncMarkerSGIX */
    {  3997, -1 }, /* FinishAsyncSGIX */
    {  4703, -1 }, /* PollAsyncSGIX */
-   { 19640, -1 }, /* DeleteAsyncMarkersSGIX */
-   { 19669, -1 }, /* IsAsyncMarkerSGIX */
-   { 28913, -1 }, /* GenAsyncMarkersSGIX */
+   { 19676, -1 }, /* DeleteAsyncMarkersSGIX */
+   { 19705, -1 }, /* IsAsyncMarkerSGIX */
+   { 28949, -1 }, /* GenAsyncMarkersSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5647,29 +5655,29 @@ static const struct gl_function_remap GL_SGIX_fragment_lighting_functions[] = {
    {  8106, -1 }, /* FragmentLightModeliSGIX */
    {  9440, -1 }, /* FragmentLightivSGIX */
    {  9677, -1 }, /* GetFragmentMaterialivSGIX */
-   { 17048, -1 }, /* FragmentLightModelfSGIX */
-   { 17348, -1 }, /* FragmentColorMaterialSGIX */
-   { 17720, -1 }, /* FragmentMaterialiSGIX */
-   { 18941, -1 }, /* LightEnviSGIX */
-   { 20297, -1 }, /* FragmentLightModelfvSGIX */
-   { 20606, -1 }, /* FragmentLightfvSGIX */
-   { 25161, -1 }, /* FragmentLightfSGIX */
-   { 27829, -1 }, /* GetFragmentLightfvSGIX */
-   { 29396, -1 }, /* FragmentMaterialivSGIX */
+   { 17084, -1 }, /* FragmentLightModelfSGIX */
+   { 17384, -1 }, /* FragmentColorMaterialSGIX */
+   { 17756, -1 }, /* FragmentMaterialiSGIX */
+   { 18977, -1 }, /* LightEnviSGIX */
+   { 20333, -1 }, /* FragmentLightModelfvSGIX */
+   { 20642, -1 }, /* FragmentLightfvSGIX */
+   { 25197, -1 }, /* FragmentLightfSGIX */
+   { 27865, -1 }, /* GetFragmentLightfvSGIX */
+   { 29432, -1 }, /* FragmentMaterialivSGIX */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIX_framezoom)
 static const struct gl_function_remap GL_SGIX_framezoom_functions[] = {
-   { 19692, -1 }, /* FrameZoomSGIX */
+   { 19728, -1 }, /* FrameZoomSGIX */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIX_igloo_interface)
 static const struct gl_function_remap GL_SGIX_igloo_interface_functions[] = {
-   { 25469, -1 }, /* IglooInterfaceSGIX */
+   { 25505, -1 }, /* IglooInterfaceSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5680,8 +5688,8 @@ static const struct gl_function_remap GL_SGIX_instruments_functions[] = {
    {  5590, -1 }, /* PollInstrumentsSGIX */
    {  9338, -1 }, /* GetInstrumentsSGIX */
    { 11348, -1 }, /* StartInstrumentsSGIX */
-   { 14067, -1 }, /* StopInstrumentsSGIX */
-   { 15673, -1 }, /* InstrumentsBufferSGIX */
+   { 14103, -1 }, /* StopInstrumentsSGIX */
+   { 15709, -1 }, /* InstrumentsBufferSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5690,10 +5698,10 @@ static const struct gl_function_remap GL_SGIX_instruments_functions[] = {
 static const struct gl_function_remap GL_SGIX_list_priority_functions[] = {
    {  1125, -1 }, /* ListParameterfSGIX */
    {  2763, -1 }, /* GetListParameterfvSGIX */
-   { 15588, -1 }, /* ListParameteriSGIX */
-   { 16342, -1 }, /* ListParameterfvSGIX */
-   { 18347, -1 }, /* ListParameterivSGIX */
-   { 28957, -1 }, /* GetListParameterivSGIX */
+   { 15624, -1 }, /* ListParameteriSGIX */
+   { 16378, -1 }, /* ListParameterfvSGIX */
+   { 18383, -1 }, /* ListParameterivSGIX */
+   { 28993, -1 }, /* GetListParameterivSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5708,16 +5716,16 @@ static const struct gl_function_remap GL_SGIX_pixel_texture_functions[] = {
 #if defined(need_GL_SGIX_polynomial_ffd)
 static const struct gl_function_remap GL_SGIX_polynomial_ffd_functions[] = {
    {  3251, -1 }, /* LoadIdentityDeformationMapSGIX */
-   { 14167, -1 }, /* DeformSGIX */
-   { 21431, -1 }, /* DeformationMap3fSGIX */
-   { 27717, -1 }, /* DeformationMap3dSGIX */
+   { 14203, -1 }, /* DeformSGIX */
+   { 21467, -1 }, /* DeformationMap3fSGIX */
+   { 27753, -1 }, /* DeformationMap3dSGIX */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIX_reference_plane)
 static const struct gl_function_remap GL_SGIX_reference_plane_functions[] = {
-   { 12898, -1 }, /* ReferencePlaneSGIX */
+   { 12934, -1 }, /* ReferencePlaneSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5725,16 +5733,16 @@ static const struct gl_function_remap GL_SGIX_reference_plane_functions[] = {
 #if defined(need_GL_SGIX_sprite)
 static const struct gl_function_remap GL_SGIX_sprite_functions[] = {
    {  8493, -1 }, /* SpriteParameterfvSGIX */
-   { 18175, -1 }, /* SpriteParameteriSGIX */
-   { 23459, -1 }, /* SpriteParameterfSGIX */
-   { 26033, -1 }, /* SpriteParameterivSGIX */
+   { 18211, -1 }, /* SpriteParameteriSGIX */
+   { 23495, -1 }, /* SpriteParameterfSGIX */
+   { 26069, -1 }, /* SpriteParameterivSGIX */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SGIX_tag_sample_buffer)
 static const struct gl_function_remap GL_SGIX_tag_sample_buffer_functions[] = {
-   { 18234, -1 }, /* TagSampleBufferSGIX */
+   { 18270, -1 }, /* TagSampleBufferSGIX */
    {    -1, -1 }
 };
 #endif
@@ -5743,18 +5751,18 @@ static const struct gl_function_remap GL_SGIX_tag_sample_buffer_functions[] = {
 static const struct gl_function_remap GL_SGI_color_table_functions[] = {
    {  6678, _gloffset_ColorTableParameteriv },
    {  7390, _gloffset_ColorTable },
-   { 13394, _gloffset_GetColorTable },
-   { 13504, _gloffset_CopyColorTable },
-   { 16992, _gloffset_ColorTableParameterfv },
-   { 20205, _gloffset_GetColorTableParameterfv },
-   { 22138, _gloffset_GetColorTableParameteriv },
+   { 13430, _gloffset_GetColorTable },
+   { 13540, _gloffset_CopyColorTable },
+   { 17028, _gloffset_ColorTableParameterfv },
+   { 20241, _gloffset_GetColorTableParameterfv },
+   { 22174, _gloffset_GetColorTableParameteriv },
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SUNX_constant_data)
 static const struct gl_function_remap GL_SUNX_constant_data_functions[] = {
-   { 27807, -1 }, /* FinishTextureSUNX */
+   { 27843, -1 }, /* FinishTextureSUNX */
    {    -1, -1 }
 };
 #endif
@@ -5766,16 +5774,16 @@ static const struct gl_function_remap GL_SUN_global_alpha_functions[] = {
    {  5615, -1 }, /* GlobalAlphaFactordSUN */
    {  8577, -1 }, /* GlobalAlphaFactoruiSUN */
    {  8929, -1 }, /* GlobalAlphaFactorbSUN */
-   { 11627, -1 }, /* GlobalAlphaFactorfSUN */
-   { 11746, -1 }, /* GlobalAlphaFactorusSUN */
-   { 19931, -1 }, /* GlobalAlphaFactorsSUN */
+   { 11663, -1 }, /* GlobalAlphaFactorfSUN */
+   { 11782, -1 }, /* GlobalAlphaFactorusSUN */
+   { 19967, -1 }, /* GlobalAlphaFactorsSUN */
    {    -1, -1 }
 };
 #endif
 
 #if defined(need_GL_SUN_mesh_array)
 static const struct gl_function_remap GL_SUN_mesh_array_functions[] = {
-   { 25845, -1 }, /* DrawMeshArraysSUN */
+   { 25881, -1 }, /* DrawMeshArraysSUN */
    {    -1, -1 }
 };
 #endif
@@ -5784,11 +5792,11 @@ static const struct gl_function_remap GL_SUN_mesh_array_functions[] = {
 static const struct gl_function_remap GL_SUN_triangle_list_functions[] = {
    {  3971, -1 }, /* ReplacementCodeubSUN */
    {  5454, -1 }, /* ReplacementCodeubvSUN */
-   { 16713, -1 }, /* ReplacementCodeusvSUN */
-   { 16901, -1 }, /* ReplacementCodePointerSUN */
-   { 18258, -1 }, /* ReplacementCodeusSUN */
-   { 19005, -1 }, /* ReplacementCodeuiSUN */
-   { 26490, -1 }, /* ReplacementCodeuivSUN */
+   { 16749, -1 }, /* ReplacementCodeusvSUN */
+   { 16937, -1 }, /* ReplacementCodePointerSUN */
+   { 18294, -1 }, /* ReplacementCodeusSUN */
+   { 19041, -1 }, /* ReplacementCodeuiSUN */
+   { 26526, -1 }, /* ReplacementCodeuivSUN */
    {    -1, -1 }
 };
 #endif
@@ -5815,26 +5823,26 @@ static const struct gl_function_remap GL_SUN_vertex_functions[] = {
    {  9303, -1 }, /* Color4fNormal3fVertex3fvSUN */
    {  9975, -1 }, /* ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN */
    { 11211, -1 }, /* ReplacementCodeuiColor4fNormal3fVertex3fvSUN */
-   { 12629, -1 }, /* ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN */
-   { 13040, -1 }, /* TexCoord2fColor3fVertex3fSUN */
-   { 14092, -1 }, /* TexCoord4fColor4fNormal3fVertex4fSUN */
-   { 14426, -1 }, /* Color4ubVertex2fvSUN */
-   { 14666, -1 }, /* Normal3fVertex3fSUN */
-   { 15614, -1 }, /* ReplacementCodeuiColor4fNormal3fVertex3fSUN */
-   { 15875, -1 }, /* TexCoord2fColor4fNormal3fVertex3fvSUN */
-   { 16542, -1 }, /* TexCoord2fVertex3fvSUN */
-   { 17318, -1 }, /* Color4ubVertex2fSUN */
-   { 17511, -1 }, /* ReplacementCodeuiColor4ubVertex3fSUN */
-   { 19364, -1 }, /* TexCoord2fColor4ubVertex3fSUN */
-   { 19711, -1 }, /* Normal3fVertex3fvSUN */
-   { 20114, -1 }, /* Color4fNormal3fVertex3fSUN */
-   { 20980, -1 }, /* ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN */
-   { 22960, -1 }, /* ReplacementCodeuiColor3fVertex3fSUN */
-   { 24136, -1 }, /* TexCoord4fVertex4fSUN */
-   { 24562, -1 }, /* TexCoord2fColor3fVertex3fvSUN */
-   { 24888, -1 }, /* ReplacementCodeuiNormal3fVertex3fvSUN */
-   { 25015, -1 }, /* TexCoord4fVertex4fvSUN */
-   { 25717, -1 }, /* ReplacementCodeuiVertex3fSUN */
+   { 12665, -1 }, /* ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN */
+   { 13076, -1 }, /* TexCoord2fColor3fVertex3fSUN */
+   { 14128, -1 }, /* TexCoord4fColor4fNormal3fVertex4fSUN */
+   { 14462, -1 }, /* Color4ubVertex2fvSUN */
+   { 14702, -1 }, /* Normal3fVertex3fSUN */
+   { 15650, -1 }, /* ReplacementCodeuiColor4fNormal3fVertex3fSUN */
+   { 15911, -1 }, /* TexCoord2fColor4fNormal3fVertex3fvSUN */
+   { 16578, -1 }, /* TexCoord2fVertex3fvSUN */
+   { 17354, -1 }, /* Color4ubVertex2fSUN */
+   { 17547, -1 }, /* ReplacementCodeuiColor4ubVertex3fSUN */
+   { 19400, -1 }, /* TexCoord2fColor4ubVertex3fSUN */
+   { 19747, -1 }, /* Normal3fVertex3fvSUN */
+   { 20150, -1 }, /* Color4fNormal3fVertex3fSUN */
+   { 21016, -1 }, /* ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN */
+   { 22996, -1 }, /* ReplacementCodeuiColor3fVertex3fSUN */
+   { 24172, -1 }, /* TexCoord4fVertex4fSUN */
+   { 24598, -1 }, /* TexCoord2fColor3fVertex3fvSUN */
+   { 24924, -1 }, /* ReplacementCodeuiNormal3fVertex3fvSUN */
+   { 25051, -1 }, /* TexCoord4fVertex4fvSUN */
+   { 25753, -1 }, /* ReplacementCodeuiVertex3fSUN */
    {    -1, -1 }
 };
 #endif
@@ -5855,27 +5863,27 @@ static const struct gl_function_remap GL_VERSION_1_3_functions[] = {
    { 10613, _gloffset_MultiTexCoord1svARB },
    { 10906, _gloffset_MultiTexCoord3svARB },
    { 10967, _gloffset_MultiTexCoord4iARB },
-   { 11654, _gloffset_MultiTexCoord3iARB },
-   { 12418, _gloffset_MultiTexCoord1dARB },
-   { 12584, _gloffset_MultiTexCoord3dvARB },
-   { 13748, _gloffset_MultiTexCoord3ivARB },
-   { 13793, _gloffset_MultiTexCoord2sARB },
-   { 15013, _gloffset_MultiTexCoord4ivARB },
-   { 16642, _gloffset_ClientActiveTextureARB },
-   { 18858, _gloffset_MultiTexCoord2dARB },
-   { 19233, _gloffset_MultiTexCoord4dvARB },
-   { 19519, _gloffset_MultiTexCoord4fvARB },
-   { 20346, _gloffset_MultiTexCoord3fARB },
-   { 22639, _gloffset_MultiTexCoord4dARB },
-   { 22917, _gloffset_MultiTexCoord1sARB },
-   { 23095, _gloffset_MultiTexCoord1dvARB },
-   { 23882, _gloffset_MultiTexCoord1ivARB },
-   { 23975, _gloffset_MultiTexCoord2ivARB },
-   { 24314, _gloffset_MultiTexCoord1iARB },
-   { 25585, _gloffset_MultiTexCoord4svARB },
-   { 26103, _gloffset_MultiTexCoord1fARB },
-   { 26366, _gloffset_MultiTexCoord4fARB },
-   { 28634, _gloffset_MultiTexCoord2fvARB },
+   { 11690, _gloffset_MultiTexCoord3iARB },
+   { 12454, _gloffset_MultiTexCoord1dARB },
+   { 12620, _gloffset_MultiTexCoord3dvARB },
+   { 13784, _gloffset_MultiTexCoord3ivARB },
+   { 13829, _gloffset_MultiTexCoord2sARB },
+   { 15049, _gloffset_MultiTexCoord4ivARB },
+   { 16678, _gloffset_ClientActiveTextureARB },
+   { 18894, _gloffset_MultiTexCoord2dARB },
+   { 19269, _gloffset_MultiTexCoord4dvARB },
+   { 19555, _gloffset_MultiTexCoord4fvARB },
+   { 20382, _gloffset_MultiTexCoord3fARB },
+   { 22675, _gloffset_MultiTexCoord4dARB },
+   { 22953, _gloffset_MultiTexCoord1sARB },
+   { 23131, _gloffset_MultiTexCoord1dvARB },
+   { 23918, _gloffset_MultiTexCoord1ivARB },
+   { 24011, _gloffset_MultiTexCoord2ivARB },
+   { 24350, _gloffset_MultiTexCoord1iARB },
+   { 25621, _gloffset_MultiTexCoord4svARB },
+   { 26139, _gloffset_MultiTexCoord1fARB },
+   { 26402, _gloffset_MultiTexCoord4fARB },
+   { 28670, _gloffset_MultiTexCoord2fvARB },
    {    -1, -1 }
 };
 #endif
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index d7e77e759e..f439d4addb 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -367,7 +367,7 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx )
    else {
       /* calculate from vp->outputs */
       struct gl_vertex_program *vprog;
-      GLbitfield vp_outputs;
+      GLbitfield64 vp_outputs;
 
       /* Choose GLSL vertex shader over ARB vertex program.  Need this
        * since vertex shader state validation comes after fragment state
diff --git a/src/mesa/main/texfetch.c b/src/mesa/main/texfetch.c
index 314ccb7b65..b37039429f 100644
--- a/src/mesa/main/texfetch.c
+++ b/src/mesa/main/texfetch.c
@@ -376,6 +376,13 @@ texfetch_funcs[MESA_FORMAT_COUNT] =
       store_texel_xrgb8888
    },
    {
+      MESA_FORMAT_XRGB8888_REV,
+      fetch_texel_1d_f_xrgb8888_rev,
+      fetch_texel_2d_f_xrgb8888_rev,
+      fetch_texel_3d_f_xrgb8888_rev,
+      store_texel_xrgb8888_rev,
+   },
+   {
       MESA_FORMAT_RGB888,
       fetch_texel_1d_f_rgb888,
       fetch_texel_2d_f_rgb888,
@@ -453,6 +460,20 @@ texfetch_funcs[MESA_FORMAT_COUNT] =
       store_texel_al88_rev
    },
    {
+      MESA_FORMAT_AL1616,
+      fetch_texel_1d_f_al1616,
+      fetch_texel_2d_f_al1616,
+      fetch_texel_3d_f_al1616,
+      store_texel_al1616
+   },
+   {
+      MESA_FORMAT_AL1616_REV,
+      fetch_texel_1d_f_al1616_rev,
+      fetch_texel_2d_f_al1616_rev,
+      fetch_texel_3d_f_al1616_rev,
+      store_texel_al1616_rev
+   },
+   {
       MESA_FORMAT_RGB332,
       fetch_texel_1d_f_rgb332,
       fetch_texel_2d_f_rgb332,
@@ -549,7 +570,7 @@ texfetch_funcs[MESA_FORMAT_COUNT] =
 static FetchTexelFuncF
 _mesa_get_texel_fetch_func(gl_format format, GLuint dims)
 {
-   FetchTexelFuncF f;
+   FetchTexelFuncF f = NULL;
    GLuint i;
    /* XXX replace loop with direct table lookup */
    for (i = 0; i < MESA_FORMAT_COUNT; i++) {
diff --git a/src/mesa/main/texfetch_tmp.h b/src/mesa/main/texfetch_tmp.h
index 27434946ec..1f0d436236 100644
--- a/src/mesa/main/texfetch_tmp.h
+++ b/src/mesa/main/texfetch_tmp.h
@@ -559,6 +559,30 @@ static void store_texel_xrgb8888(struct gl_texture_image *texImage,
 #endif
 
 
+/* MESA_FORMAT_XRGB8888_REV **************************************************/
+
+/* Fetch texel from 1D, 2D or 3D xrgb8888_rev texture, return 4 GLfloats */
+static void FETCH(f_xrgb8888_rev)( const struct gl_texture_image *texImage,
+                                   GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   const GLuint s = *TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
+   texel[RCOMP] = UBYTE_TO_FLOAT( (s >>  8) & 0xff );
+   texel[GCOMP] = UBYTE_TO_FLOAT( (s >> 16) & 0xff );
+   texel[BCOMP] = UBYTE_TO_FLOAT( (s >> 24)        );
+   texel[ACOMP] = 1.0f;
+}
+
+#if DIM == 3
+static void store_texel_xrgb8888_rev(struct gl_texture_image *texImage,
+                                     GLint i, GLint j, GLint k, const void *texel)
+{
+   const GLubyte *rgba = (const GLubyte *) texel;
+   GLuint *dst = TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
+   *dst = PACK_COLOR_8888(rgba[BCOMP], rgba[GCOMP], rgba[RCOMP], 0xff);
+}
+#endif
+
+
 /* MESA_FORMAT_RGB888 ********************************************************/
 
 /* Fetch texel from 1D, 2D or 3D rgb888 texture, return 4 GLchans */
@@ -834,6 +858,54 @@ static void store_texel_al88_rev(struct gl_texture_image *texImage,
 #endif
 
 
+/* MESA_FORMAT_AL1616 ********************************************************/
+
+/* Fetch texel from 1D, 2D or 3D al1616 texture, return 4 GLchans */
+static void FETCH(f_al1616)( const struct gl_texture_image *texImage,
+			     GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   const GLuint s = *TEXEL_ADDR(GLushort, texImage, i, j, k, 1);
+   texel[RCOMP] =
+   texel[GCOMP] =
+   texel[BCOMP] = USHORT_TO_FLOAT( s & 0xffff );
+   texel[ACOMP] = USHORT_TO_FLOAT( s >> 16 );
+}
+
+#if DIM == 3
+static void store_texel_al1616(struct gl_texture_image *texImage,
+                             GLint i, GLint j, GLint k, const void *texel)
+{
+   const GLushort *rgba = (const GLushort *) texel;
+   GLuint *dst = TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
+   *dst = PACK_COLOR_1616(rgba[ACOMP], rgba[RCOMP]);
+}
+#endif
+
+
+/* MESA_FORMAT_AL1616_REV ****************************************************/
+
+/* Fetch texel from 1D, 2D or 3D al1616_rev texture, return 4 GLchans */
+static void FETCH(f_al1616_rev)( const struct gl_texture_image *texImage,
+				 GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   const GLuint s = *TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
+   texel[RCOMP] =
+   texel[GCOMP] =
+   texel[BCOMP] = USHORT_TO_FLOAT( s >> 16 );
+   texel[ACOMP] = USHORT_TO_FLOAT( s & 0xffff );
+}
+
+#if DIM == 3
+static void store_texel_al1616_rev(struct gl_texture_image *texImage,
+				   GLint i, GLint j, GLint k, const void *texel)
+{
+   const GLushort *rgba = (const GLushort *) texel;
+   GLuint *dst = TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
+   *dst = PACK_COLOR_1616(rgba[RCOMP], rgba[ACOMP]);
+}
+#endif
+
+
 /* MESA_FORMAT_RGB332 ********************************************************/
 
 /* Fetch texel from 1D, 2D or 3D rgb332 texture, return 4 GLchans */
diff --git a/src/mesa/main/texformat.c b/src/mesa/main/texformat.c
index 39c561e278..1a374e7bee 100644
--- a/src/mesa/main/texformat.c
+++ b/src/mesa/main/texformat.c
@@ -115,11 +115,13 @@ _mesa_choose_tex_format( GLcontext *ctx, GLint internalFormat,
       case GL_LUMINANCE_ALPHA:
       case GL_LUMINANCE4_ALPHA4:
       case GL_LUMINANCE6_ALPHA2:
+      case GL_LUMINANCE8_ALPHA8:
+         return MESA_FORMAT_AL88;
+
       case GL_LUMINANCE12_ALPHA4:
       case GL_LUMINANCE12_ALPHA12:
       case GL_LUMINANCE16_ALPHA16:
-      case GL_LUMINANCE8_ALPHA8:
-         return MESA_FORMAT_AL88;
+         return MESA_FORMAT_AL1616;
 
       case GL_INTENSITY:
       case GL_INTENSITY4:
diff --git a/src/mesa/main/texgetimage.c b/src/mesa/main/texgetimage.c
index 2f88718933..bd7cc8d278 100644
--- a/src/mesa/main/texgetimage.c
+++ b/src/mesa/main/texgetimage.c
@@ -521,10 +521,11 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
                               struct gl_texture_object *texObj,
                               struct gl_texture_image *texImage)
 {
-   const GLuint size = _mesa_format_image_size(texImage->TexFormat,
-                                               texImage->Width,
-                                               texImage->Height,
-                                               texImage->Depth);
+   const GLuint row_stride = _mesa_format_row_stride(texImage->TexFormat,
+                                                     texImage->Width);
+   const GLuint row_stride_stored = _mesa_format_row_stride(texImage->TexFormat,
+                                                            texImage->RowStride);
+   GLuint i;
 
    if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) {
       /* pack texture image into a PBO */
@@ -540,8 +541,24 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
       img = ADD_POINTERS(buf, img);
    }
 
-   /* just memcpy, no pixelstore or pixel transfer */
-   _mesa_memcpy(img, texImage->Data, size);
+   /* no pixelstore or pixel transfer, but respect stride */
+
+   if (row_stride == row_stride_stored) {
+      const GLuint size = _mesa_format_image_size(texImage->TexFormat,
+                                                  texImage->Width,
+                                                  texImage->Height,
+                                                  texImage->Depth);
+      _mesa_memcpy(img, texImage->Data, size);
+   }
+   else {
+      GLuint bw, bh;
+      _mesa_get_format_block_size(texImage->TexFormat, &bw, &bh);
+      for (i = 0; i < (texImage->Height + bh - 1) / bh; i++) {
+         memcpy((GLubyte *)img + i * row_stride,
+                (GLubyte *)texImage->Data + i * row_stride_stored,
+                row_stride);
+      }
+   }
 
    if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) {
       ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT,
diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c
index e48d472061..5387eb1283 100644
--- a/src/mesa/main/texstore.c
+++ b/src/mesa/main/texstore.c
@@ -1410,7 +1410,8 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
 
    ASSERT(dstFormat == MESA_FORMAT_ARGB8888 ||
           dstFormat == MESA_FORMAT_ARGB8888_REV ||
-          dstFormat == MESA_FORMAT_XRGB8888);
+          dstFormat == MESA_FORMAT_XRGB8888 ||
+          dstFormat == MESA_FORMAT_XRGB8888_REV );
    ASSERT(texelBytes == 4);
 
    if (!ctx->_ImageTransferState &&
@@ -1431,7 +1432,8 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
    }
    else if (!ctx->_ImageTransferState &&
        !srcPacking->SwapBytes &&
-       dstFormat == MESA_FORMAT_ARGB8888_REV &&
+       (dstFormat == MESA_FORMAT_ARGB8888_REV ||
+        dstFormat == MESA_FORMAT_XRGB8888_REV) &&
        baseInternalFormat == GL_RGBA &&
        srcFormat == GL_BGRA &&
        ((srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
@@ -1524,7 +1526,8 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
        */
       if ((littleEndian && dstFormat == MESA_FORMAT_ARGB8888) ||
           (littleEndian && dstFormat == MESA_FORMAT_XRGB8888) ||
-	  (!littleEndian && dstFormat == MESA_FORMAT_ARGB8888_REV)) {
+	  (!littleEndian && dstFormat == MESA_FORMAT_ARGB8888_REV) ||
+	  (!littleEndian && dstFormat == MESA_FORMAT_XRGB8888_REV)) {
 	 dstmap[3] = 3;		/* alpha */
 	 dstmap[2] = 0;		/* red */
 	 dstmap[1] = 1;		/* green */
@@ -1533,6 +1536,7 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
       else {
 	 assert((littleEndian && dstFormat == MESA_FORMAT_ARGB8888_REV) ||
 		(!littleEndian && dstFormat == MESA_FORMAT_ARGB8888) ||
+		(littleEndian && dstFormat == MESA_FORMAT_XRGB8888_REV) ||
 		(!littleEndian && dstFormat == MESA_FORMAT_XRGB8888));
 	 dstmap[3] = 2;
 	 dstmap[2] = 1;
@@ -2146,6 +2150,77 @@ _mesa_texstore_al88(TEXSTORE_PARAMS)
 
 
 static GLboolean
+_mesa_texstore_al1616(TEXSTORE_PARAMS)
+{
+   const GLboolean littleEndian = _mesa_little_endian();
+   const GLuint texelBytes = _mesa_get_format_bytes(dstFormat);
+   const GLenum baseFormat = _mesa_get_format_base_format(dstFormat);
+
+   ASSERT(dstFormat == MESA_FORMAT_AL1616 ||
+          dstFormat == MESA_FORMAT_AL1616_REV);
+   ASSERT(texelBytes == 4);
+
+   if (!ctx->_ImageTransferState &&
+       !srcPacking->SwapBytes &&
+       dstFormat == MESA_FORMAT_AL1616 &&
+       baseInternalFormat == GL_LUMINANCE_ALPHA &&
+       srcFormat == GL_LUMINANCE_ALPHA &&
+       srcType == GL_UNSIGNED_SHORT &&
+       littleEndian) {
+      /* simple memcpy path */
+      memcpy_texture(ctx, dims,
+                     dstFormat, dstAddr, dstXoffset, dstYoffset, dstZoffset,
+                     dstRowStride,
+                     dstImageOffsets,
+                     srcWidth, srcHeight, srcDepth, srcFormat, srcType,
+                     srcAddr, srcPacking);
+   }
+   else {
+      /* general path */
+      const GLfloat *tempImage = make_temp_float_image(ctx, dims,
+                                                 baseInternalFormat,
+                                                 baseFormat,
+                                                 srcWidth, srcHeight, srcDepth,
+                                                 srcFormat, srcType, srcAddr,
+                                                 srcPacking);
+      const GLfloat *src = tempImage;
+      GLint img, row, col;
+      if (!tempImage)
+         return GL_FALSE;
+      _mesa_adjust_image_for_convolution(ctx, dims, &srcWidth, &srcHeight);
+      for (img = 0; img < srcDepth; img++) {
+         GLubyte *dstRow = (GLubyte *) dstAddr
+            + dstImageOffsets[dstZoffset + img] * texelBytes
+            + dstYoffset * dstRowStride
+            + dstXoffset * texelBytes;
+         for (row = 0; row < srcHeight; row++) {
+            GLuint *dstUI = (GLuint *) dstRow;
+            if (dstFormat == MESA_FORMAT_AL1616) {
+               for (col = 0; col < srcWidth; col++) {
+                  /* src[0] is luminance, src[1] is alpha */
+                 dstUI[col] = PACK_COLOR_1616( FLOAT_TO_USHORT(src[1]),
+					       FLOAT_TO_USHORT(src[0]) );
+                 src += 2;
+               }
+            }
+            else {
+               for (col = 0; col < srcWidth; col++) {
+                  /* src[0] is luminance, src[1] is alpha */
+                 dstUI[col] = PACK_COLOR_1616_REV( FLOAT_TO_UBYTE(src[1]),
+						   FLOAT_TO_UBYTE(src[0]) );
+                 src += 2;
+               }
+            }
+            dstRow += dstRowStride;
+         }
+      }
+      _mesa_free((void *) tempImage);
+   }
+   return GL_TRUE;
+}
+
+
+static GLboolean
 _mesa_texstore_rgb332(TEXSTORE_PARAMS)
 {
    const GLuint texelBytes = _mesa_get_format_bytes(dstFormat);
@@ -3050,7 +3125,7 @@ _mesa_texstore_sla8(TEXSTORE_PARAMS)
  * Table mapping MESA_FORMAT_8 to _mesa_texstore_*()
  * XXX this is somewhat temporary.
  */
-const static struct {
+static const struct {
    gl_format Name;
    StoreTexImageFunc Store;
 }
@@ -3062,6 +3137,7 @@ texstore_funcs[MESA_FORMAT_COUNT] =
    { MESA_FORMAT_ARGB8888, _mesa_texstore_argb8888 },
    { MESA_FORMAT_ARGB8888_REV, _mesa_texstore_argb8888 },
    { MESA_FORMAT_XRGB8888, _mesa_texstore_argb8888 },
+   { MESA_FORMAT_XRGB8888_REV, _mesa_texstore_argb8888 },
    { MESA_FORMAT_RGB888, _mesa_texstore_rgb888 },
    { MESA_FORMAT_BGR888, _mesa_texstore_bgr888 },
    { MESA_FORMAT_RGB565, _mesa_texstore_rgb565 },
@@ -3073,6 +3149,8 @@ texstore_funcs[MESA_FORMAT_COUNT] =
    { MESA_FORMAT_ARGB1555_REV, _mesa_texstore_argb1555 },
    { MESA_FORMAT_AL88, _mesa_texstore_al88 },
    { MESA_FORMAT_AL88_REV, _mesa_texstore_al88 },
+   { MESA_FORMAT_AL1616, _mesa_texstore_al1616 },
+   { MESA_FORMAT_AL1616_REV, _mesa_texstore_al1616 },
    { MESA_FORMAT_RGB332, _mesa_texstore_rgb332 },
    { MESA_FORMAT_A8, _mesa_texstore_a8 },
    { MESA_FORMAT_L8, _mesa_texstore_a8 },
diff --git a/src/mesa/main/version.h b/src/mesa/main/version.h
index 0cae1860a3..dc55cb7ccc 100644
--- a/src/mesa/main/version.h
+++ b/src/mesa/main/version.h
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.7
+ * Version:  7.8
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
@@ -30,9 +30,9 @@
 
 /* Mesa version */
 #define MESA_MAJOR 7
-#define MESA_MINOR 7
+#define MESA_MINOR 8
 #define MESA_PATCH 0
-#define MESA_VERSION_STRING "7.7-devel"
+#define MESA_VERSION_STRING "7.8-devel"
 
 /* To make version comparison easy */
 #define MESA_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index e96675ae05..f2fa2bb896 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -314,7 +314,6 @@ struct prog_dst_register
     */
    GLuint CondSrc:1;
    /*@}*/
-   GLuint pad:28;
 };
 
 
diff --git a/src/mesa/shader/prog_optimize.c b/src/mesa/shader/prog_optimize.c
index 3d28d885a4..4fe351251e 100644
--- a/src/mesa/shader/prog_optimize.c
+++ b/src/mesa/shader/prog_optimize.c
@@ -443,7 +443,7 @@ _mesa_remove_extra_move_use(struct gl_program *prog)
     *    FOO tmpY, arg0, arg1;
     */
 
-   for (i = 0; i < prog->NumInstructions - 1; i++) {
+   for (i = 0; i + 1 < prog->NumInstructions; i++) {
       const struct prog_instruction *mov = prog->Instructions + i;
 
       if (mov->Opcode != OPCODE_MOV ||
diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c
index b6529ed02d..81cfe212ab 100644
--- a/src/mesa/shader/prog_print.c
+++ b/src/mesa/shader/prog_print.c
@@ -834,11 +834,11 @@ _mesa_print_program(const struct gl_program *prog)
  * XXX move to imports.[ch] if useful elsewhere.
  */
 static const char *
-binary(GLbitfield val)
+binary(GLbitfield64 val)
 {
-   static char buf[50];
+   static char buf[80];
    GLint i, len = 0;
-   for (i = 31; i >= 0; --i) {
+   for (i = 63; i >= 0; --i) {
       if (val & (1 << i))
          buf[len++] = '1';
       else if (len > 0 || i == 0)
diff --git a/src/mesa/shader/program_parse.tab.c b/src/mesa/shader/program_parse.tab.c
index b9ef88b64b..d4f8429488 100644
--- a/src/mesa/shader/program_parse.tab.c
+++ b/src/mesa/shader/program_parse.tab.c
@@ -2622,7 +2622,7 @@ yyreduce:
 		 YYERROR;
 	      }
 
-	      state->prog->OutputsWritten |= (1U << (yyval.dst_reg).Index);
+	      state->prog->OutputsWritten |= BITFIELD64_BIT((yyval.dst_reg).Index);
 	   }
 	;}
     break;
diff --git a/src/mesa/shader/program_parse.y b/src/mesa/shader/program_parse.y
index d07bf85b36..8ca6f9805b 100644
--- a/src/mesa/shader/program_parse.y
+++ b/src/mesa/shader/program_parse.y
@@ -643,7 +643,7 @@ maskedDstReg: dstReg optionalMask optionalCcMask
 		 YYERROR;
 	      }
 
-	      state->prog->OutputsWritten |= (1U << $$.Index);
+	      state->prog->OutputsWritten |= BITFIELD64_BIT($$.Index);
 	   }
 	}
 	;
diff --git a/src/mesa/shader/program_parser.h b/src/mesa/shader/program_parser.h
index bce6041381..c170948f73 100644
--- a/src/mesa/shader/program_parser.h
+++ b/src/mesa/shader/program_parser.h
@@ -35,7 +35,7 @@ enum asm_type {
    at_attrib,
    at_param,
    at_temp,
-   at_output,
+   at_output
 };
 
 struct asm_symbol {
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index 3b8529592d..a0daac1b80 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -109,7 +109,7 @@ _mesa_insert_mvp_dp4_code(GLcontext *ctx, struct gl_vertex_program *vprog)
    vprog->Base.Instructions = newInst;
    vprog->Base.NumInstructions = newLen;
    vprog->Base.InputsRead |= VERT_BIT_POS;
-   vprog->Base.OutputsWritten |= (1 << VERT_RESULT_HPOS);
+   vprog->Base.OutputsWritten |= BITFIELD64_BIT(VERT_RESULT_HPOS);
 }
 
 
@@ -211,7 +211,7 @@ _mesa_insert_mvp_mad_code(GLcontext *ctx, struct gl_vertex_program *vprog)
    vprog->Base.Instructions = newInst;
    vprog->Base.NumInstructions = newLen;
    vprog->Base.InputsRead |= VERT_BIT_POS;
-   vprog->Base.OutputsWritten |= (1 << VERT_RESULT_HPOS);
+   vprog->Base.OutputsWritten |= BITFIELD64_BIT(VERT_RESULT_HPOS);
 }
 
 
@@ -613,7 +613,7 @@ _mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog)
    prog->Base.Instructions = inst;
    prog->Base.NumInstructions = 2;
    prog->Base.InputsRead = 1 << inputAttr;
-   prog->Base.OutputsWritten = 1 << FRAG_RESULT_COLOR;
+   prog->Base.OutputsWritten = BITFIELD64_BIT(FRAG_RESULT_COLOR);
 }
 
 
@@ -657,7 +657,7 @@ _mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog)
    prog->Base.Instructions = inst;
    prog->Base.NumInstructions = 2;
    prog->Base.InputsRead = 1 << inputAttr;
-   prog->Base.OutputsWritten = 1 << VERT_RESULT_COL0;
+   prog->Base.OutputsWritten = BITFIELD64_BIT(VERT_RESULT_COL0);
 
    /*
     * Now insert code to do standard modelview/projection transformation.
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index 1942d65cea..54e0d17dc5 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -927,7 +927,7 @@ gen_return_with_expression(slang_assemble_ctx *A, slang_operation *oper)
       slang_operation_copy(rhs, &oper->children[0]);
    }
 
-   ///blockOper->locals->outer_scope = oper->locals->outer_scope;
+   /*blockOper->locals->outer_scope = oper->locals->outer_scope;*/
 
    /*slang_print_tree(blockOper, 0);*/
 
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 7670223a39..4b5bb56edf 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -81,8 +81,8 @@ new_subroutine(slang_emit_info *emitInfo, GLuint *id)
 
    emitInfo->Subroutines = (struct gl_program **)
       _mesa_realloc(emitInfo->Subroutines,
-                    n * sizeof(struct gl_program),
-                    (n + 1) * sizeof(struct gl_program));
+                    n * sizeof(struct gl_program *),
+                    (n + 1) * sizeof(struct gl_program *));
    emitInfo->Subroutines[n] = ctx->Driver.NewProgram(ctx, emitInfo->prog->Target, 0);
    emitInfo->Subroutines[n]->Parameters = emitInfo->prog->Parameters;
    emitInfo->NumSubroutines++;
@@ -551,6 +551,9 @@ emit_instruction(slang_emit_info *emitInfo,
                                        &srcRelAddr,
                                        NULL,
                                        NULL);
+               if (!inst) {
+                  return NULL;
+               }
 
                src[i] = &newSrc[i];
             }
@@ -948,6 +951,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                               n->Children[0]->Store,
                               n->Children[1]->Store,
                               NULL);
+      if (!inst) {
+         return NULL;
+      }
       inst_comment(inst, "Compare values");
 
       /* Compute val = DOT(temp, temp)  (reduction) */
@@ -957,6 +963,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                               &tempStore,
                               &tempStore,
                               NULL);
+      if (!inst) {
+         return NULL;
+      }
       inst->SrcReg[0].Swizzle = inst->SrcReg[1].Swizzle = swizzle; /*override*/
       inst_comment(inst, "Reduce vec to bool");
 
@@ -972,6 +981,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                                  n->Store,
                                  &zero,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
          inst_comment(inst, "Invert true/false");
       }
    }
@@ -1001,6 +1013,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                                     &srcStore0,
                                     &srcStore1,
                                     NULL);
+            if (!inst) {
+               return NULL;
+            }
             inst_comment(inst, "Begin struct/array comparison");
          }
          else {
@@ -1010,12 +1025,18 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                                     &srcStore0,
                                     &srcStore1,
                                     NULL);
+            if (!inst) {
+               return NULL;
+            }
             /* ADD accTemp, accTemp, sneTemp; # like logical-OR */
             inst = emit_instruction(emitInfo, OPCODE_ADD,
                                     &accTemp, /* dest */
                                     &accTemp,
                                     &sneTemp,
                                     NULL);
+            if (!inst) {
+               return NULL;
+            }
          }
       }
 
@@ -1025,6 +1046,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                               &accTemp,
                               &accTemp,
                               NULL);
+      if (!inst) {
+         return NULL;
+      }
       inst_comment(inst, "End struct/array comparison");
 
       if (n->Opcode == IR_EQUAL) {
@@ -1036,6 +1060,9 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
                                  n->Store,
                                  &zero,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
          inst_comment(inst, "Invert true/false");
       }
 
@@ -1119,6 +1146,9 @@ emit_clamp(slang_emit_info *emitInfo, slang_ir_node *n)
                            n->Children[0]->Store,
                            n->Children[1]->Store,
                            NULL);
+   if (!inst) {
+      return NULL;
+   }
 
    /* n->dest = min(tmp, ch[2]) */
    inst = emit_instruction(emitInfo, OPCODE_MIN,
@@ -1153,7 +1183,9 @@ emit_negation(slang_emit_info *emitInfo, slang_ir_node *n)
                            n->Children[0]->Store,
                            NULL,
                            NULL);
-   inst->SrcReg[0].Negate = NEGATE_XYZW;
+   if (inst) {
+      inst->SrcReg[0].Negate = NEGATE_XYZW;
+   }
    return inst;
 }
 
@@ -1356,6 +1388,9 @@ emit_tex(slang_emit_info *emitInfo, slang_ir_node *n)
                            n->Children[1]->Store,
                            NULL,
                            NULL);
+   if (!inst) {
+      return NULL;
+   }
 
    inst->TexShadow = shadow;
 
@@ -1458,6 +1493,9 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
                                     &srcStore,
                                     NULL,
                                     NULL);
+            if (!inst) {
+               return NULL;
+            }
             inst_comment(inst, "IR_COPY block");
             srcStore.Index++;
             dstStore.Index++;
@@ -1473,6 +1511,9 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
                                  n->Children[1]->Store,
                                  NULL,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
          dstAnnot = storage_annotation(n->Children[0], emitInfo->prog);
          srcAnnot = storage_annotation(n->Children[1], emitInfo->prog);
          inst->Comment = instruction_annotation(inst->Opcode, dstAnnot,
@@ -1534,6 +1575,9 @@ emit_cond(slang_emit_info *emitInfo, slang_ir_node *n)
                                  n->Children[0]->Store,
                                  NULL,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
          inst->CondUpdate = GL_TRUE;
          inst_comment(inst, "COND expr");
          _slang_free_temp(emitInfo->vt, n->Store);
@@ -1596,6 +1640,9 @@ emit_not(slang_emit_info *emitInfo, slang_ir_node *n)
                            n->Children[0]->Store,
                            &zero,
                            NULL);
+   if (!inst) {
+      return NULL;
+   }
    inst_comment(inst, "NOT");
 
    free_node_storage(emitInfo->vt, n->Children[0]);
@@ -1646,12 +1693,17 @@ emit_if(slang_emit_info *emitInfo, slang_ir_node *n)
          ifInst->DstReg.CondSwizzle = writemask_to_swizzle(condWritemask);
       }
       else {
+         struct prog_instruction *inst;
+
          /* IF src[0] THEN ... */
-         emit_instruction(emitInfo, OPCODE_IF,
-                          NULL, /* dst */
-                          n->Children[0]->Store, /* op0 */
-                          NULL,
-                          NULL);
+         inst = emit_instruction(emitInfo, OPCODE_IF,
+                                 NULL, /* dst */
+                                 n->Children[0]->Store, /* op0 */
+                                 NULL,
+                                 NULL);
+         if (!inst) {
+            return NULL;
+         }
       }
    }
    else {
@@ -1875,6 +1927,9 @@ emit_cont_break_if_true(slang_emit_info *emitInfo, slang_ir_node *n)
                                  n->Children[0]->Store,
                                  NULL,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
          n->InstLocation = emitInfo->prog->NumInstructions;
 
          inst = new_instruction(emitInfo, opcode);
@@ -2045,6 +2100,9 @@ emit_array_element(slang_emit_info *emitInfo, slang_ir_node *n)
                                  indexStore, /* the index */
                                  &elemSizeStore,
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
 
          indexStore = indexTemp;
       }
@@ -2071,6 +2129,9 @@ emit_array_element(slang_emit_info *emitInfo, slang_ir_node *n)
                                  indexStore,     /* the index */
                                  &indirectArray, /* indirect array base */
                                  NULL);
+         if (!inst) {
+            return NULL;
+         }
 
          indexStore = indexTemp;
       }
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index 646db43fda..4636fa4998 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -530,7 +530,7 @@ _slang_update_inputs_outputs(struct gl_program *prog)
       }
 
       if (inst->DstReg.File == PROGRAM_OUTPUT) {
-         prog->OutputsWritten |= 1 << inst->DstReg.Index;
+         prog->OutputsWritten |= BITFIELD64_BIT(inst->DstReg.Index);
          if (inst->DstReg.RelAddr) {
             /* If the output attribute is indexed with relative addressing
              * we know that it must be a varying or texcoord such as
@@ -543,14 +543,17 @@ _slang_update_inputs_outputs(struct gl_program *prog)
             if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
                if (inst->DstReg.Index == VERT_RESULT_TEX0) {
                   /* mark all texcoord outputs as written */
-                  const GLbitfield mask =
-                     ((1 << MAX_TEXTURE_COORD_UNITS) - 1) << VERT_RESULT_TEX0;
+                  const GLbitfield64 mask =
+		     BITFIELD64_RANGE(VERT_RESULT_TEX0,
+				      (VERT_RESULT_TEX0
+				       + MAX_TEXTURE_COORD_UNITS - 1));
                   prog->OutputsWritten |= mask;
                }
                else if (inst->DstReg.Index == VERT_RESULT_VAR0) {
                   /* mark all generic varying outputs as written */
-                  const GLbitfield mask =
-                     ((1 << MAX_VARYING) - 1) << VERT_RESULT_VAR0;
+                  const GLbitfield64 mask =
+		     BITFIELD64_RANGE(VERT_RESULT_VAR0,
+				      (VERT_RESULT_VAR0 + MAX_VARYING - 1));
                   prog->OutputsWritten |= mask;
                }
             }
@@ -602,11 +605,16 @@ concat_shaders(struct gl_shader_program *shProg, GLenum shaderType)
 {
    struct gl_shader *newShader;
    const struct gl_shader *firstShader = NULL;
-   GLuint shaderLengths[100];
+   GLuint *shaderLengths;
    GLchar *source;
    GLuint totalLen = 0, len = 0;
    GLuint i;
 
+   shaderLengths = (GLuint *)_mesa_malloc(shProg->NumShaders * sizeof(GLuint));
+   if (!shaderLengths) {
+      return NULL;
+   }
+
    /* compute total size of new shader source code */
    for (i = 0; i < shProg->NumShaders; i++) {
       const struct gl_shader *shader = shProg->Shaders[i];
@@ -618,12 +626,16 @@ concat_shaders(struct gl_shader_program *shProg, GLenum shaderType)
       }
    }
 
-   if (totalLen == 0)
+   if (totalLen == 0) {
+      _mesa_free(shaderLengths);
       return NULL;
+   }
 
    source = (GLchar *) _mesa_malloc(totalLen + 1);
-   if (!source)
+   if (!source) {
+      _mesa_free(shaderLengths);
       return NULL;
+   }
 
    /* concatenate shaders */
    for (i = 0; i < shProg->NumShaders; i++) {
@@ -638,9 +650,16 @@ concat_shaders(struct gl_shader_program *shProg, GLenum shaderType)
    _mesa_printf("---NEW CONCATENATED SHADER---:\n%s\n------------\n", source);
    */
 
+   _mesa_free(shaderLengths);
+
    remove_extra_version_directives(source);
 
    newShader = CALLOC_STRUCT(gl_shader);
+   if (!newShader) {
+      _mesa_free(source);
+      return NULL;
+   }
+
    newShader->Type = shaderType;
    newShader->Source = source;
    newShader->Pragmas = firstShader->Pragmas;
@@ -845,7 +864,8 @@ _slang_link(GLcontext *ctx,
    if (shProg->VertexProgram) {
       _slang_update_inputs_outputs(&shProg->VertexProgram->Base);
       _slang_count_temporaries(&shProg->VertexProgram->Base);
-      if (!(shProg->VertexProgram->Base.OutputsWritten & (1 << VERT_RESULT_HPOS))) {
+      if (!(shProg->VertexProgram->Base.OutputsWritten
+	    & BITFIELD64_BIT(VERT_RESULT_HPOS))) {
          /* the vertex program did not compute a vertex position */
          link_error(shProg,
                     "gl_Position was not written by vertex shader\n");
@@ -878,7 +898,7 @@ _slang_link(GLcontext *ctx,
    if (shProg->FragmentProgram) {
       const GLbitfield varyingRead
          = shProg->FragmentProgram->Base.InputsRead >> FRAG_ATTRIB_VAR0;
-      const GLbitfield varyingWritten = shProg->VertexProgram ?
+      const GLbitfield64 varyingWritten = shProg->VertexProgram ?
          shProg->VertexProgram->Base.OutputsWritten >> VERT_RESULT_VAR0 : 0x0;
       if ((varyingRead & varyingWritten) != varyingRead) {
          link_error(shProg,
@@ -889,9 +909,10 @@ _slang_link(GLcontext *ctx,
 
    /* check that gl_FragColor and gl_FragData are not both written to */
    if (shProg->FragmentProgram) {
-      GLbitfield outputsWritten = shProg->FragmentProgram->Base.OutputsWritten;
-      if ((outputsWritten & ((1 << FRAG_RESULT_COLOR))) &&
-          (outputsWritten >= (1 << FRAG_RESULT_DATA0))) {
+      const GLbitfield64 outputsWritten =
+	 shProg->FragmentProgram->Base.OutputsWritten;
+      if ((outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) &&
+          (outputsWritten >= BITFIELD64_BIT(FRAG_RESULT_DATA0))) {
          link_error(shProg, "Fragment program cannot write both gl_FragColor"
                     " and gl_FragData[].\n");
          return;
diff --git a/src/mesa/sparc/glapi_sparc.S b/src/mesa/sparc/glapi_sparc.S
index 91ea2afa43..299d5712a7 100644
--- a/src/mesa/sparc/glapi_sparc.S
+++ b/src/mesa/sparc/glapi_sparc.S
@@ -1275,6 +1275,7 @@ gl_dispatch_functions_start:
 	GL_STUB_ALIAS(glGetAttribLocation, glGetAttribLocationARB)
 	GL_STUB_ALIAS(glDrawBuffers, glDrawBuffersARB)
 	GL_STUB_ALIAS(glDrawBuffersATI, glDrawBuffersARB)
+	GL_STUB_ALIAS(glRenderbufferStorageMultisampleEXT, glRenderbufferStorageMultisample)
 	GL_STUB_ALIAS(glPointParameterf, glPointParameterfEXT)
 	GL_STUB_ALIAS(glPointParameterfARB, glPointParameterfEXT)
 	GL_STUB_ALIAS(glPointParameterfSGIS, glPointParameterfEXT)
diff --git a/src/mesa/state_tracker/st_atom.c b/src/mesa/state_tracker/st_atom.c
index bbbfa9478f..62a0c93203 100644
--- a/src/mesa/state_tracker/st_atom.c
+++ b/src/mesa/state_tracker/st_atom.c
@@ -46,7 +46,9 @@ static const struct st_tracked_state *atoms[] =
    &st_update_clip,
 
    &st_finalize_textures,
-   &st_update_shader,
+   &st_update_fp,
+   &st_update_gp,
+   &st_update_vp,
 
    &st_update_rasterizer,
    &st_update_polygon_stipple,
@@ -140,7 +142,7 @@ void st_validate_state( struct st_context *st )
    if (state->st == 0)
       return;
 
-//   _mesa_printf("%s %x/%x\n", __FUNCTION__, state->mesa, state->st);
+   /*_mesa_printf("%s %x/%x\n", __FUNCTION__, state->mesa, state->st);*/
 
    if (1) {
       /* Debug version which enforces various sanity checks on the
@@ -155,7 +157,7 @@ void st_validate_state( struct st_context *st )
 	 const struct st_tracked_state *atom = atoms[i];
 	 struct st_state_flags generated;
 	 
-//	 _mesa_printf("atom %s %x/%x\n", atom->name, atom->dirty.mesa, atom->dirty.st);
+	 /*_mesa_printf("atom %s %x/%x\n", atom->name, atom->dirty.mesa, atom->dirty.st);*/
 
 	 if (!(atom->dirty.mesa || atom->dirty.st) ||
 	     !atom->update) {
@@ -165,7 +167,7 @@ void st_validate_state( struct st_context *st )
 
 	 if (check_state(state, &atom->dirty)) {
 	    atoms[i]->update( st );
-//	    _mesa_printf("after: %x\n", atom->dirty.mesa);
+	    /*_mesa_printf("after: %x\n", atom->dirty.mesa);*/
 	 }
 
 	 accumulate_state(&examined, &atom->dirty);
@@ -178,7 +180,7 @@ void st_validate_state( struct st_context *st )
 	 assert(!check_state(&examined, &generated));
 	 prev = *state;
       }
-//      _mesa_printf("\n");
+      /*_mesa_printf("\n");*/
 
    }
    else {
diff --git a/src/mesa/state_tracker/st_atom.h b/src/mesa/state_tracker/st_atom.h
index 58688e9f21..c8c9e42619 100644
--- a/src/mesa/state_tracker/st_atom.h
+++ b/src/mesa/state_tracker/st_atom.h
@@ -47,7 +47,9 @@ void st_validate_state( struct st_context *st );
 extern const struct st_tracked_state st_update_framebuffer;
 extern const struct st_tracked_state st_update_clip;
 extern const struct st_tracked_state st_update_depth_stencil_alpha;
-extern const struct st_tracked_state st_update_shader;
+extern const struct st_tracked_state st_update_fp;
+extern const struct st_tracked_state st_update_gp;
+extern const struct st_tracked_state st_update_vp;
 extern const struct st_tracked_state st_update_rasterizer;
 extern const struct st_tracked_state st_update_polygon_stipple;
 extern const struct st_tracked_state st_update_viewport;
diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c
index e18c0f6e0a..8ca4335e33 100644
--- a/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -40,6 +40,7 @@
 #include "pipe/p_inlines.h"
 #include "cso_cache/cso_context.h"
 #include "util/u_rect.h"
+#include "util/u_math.h"
 
 
 
@@ -64,8 +65,8 @@ update_renderbuffer_surface(struct st_context *st,
       GLuint level;
       /* find matching mipmap level size */
       for (level = 0; level <= texture->last_level; level++) {
-         if (texture->width[level] == rtt_width &&
-             texture->height[level] == rtt_height) {
+         if (u_minify(texture->width0, level) == rtt_width &&
+             u_minify(texture->height0, level) == rtt_height) {
 
             pipe_surface_reference(&strb->surface, NULL);
 
diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c
index babfcc87b4..4b35f59cc2 100644
--- a/src/mesa/state_tracker/st_atom_pixeltransfer.c
+++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c
@@ -145,7 +145,7 @@ load_color_map_texture(GLcontext *ctx, struct pipe_texture *pt)
    const GLuint gSize = ctx->PixelMaps.GtoG.Size;
    const GLuint bSize = ctx->PixelMaps.BtoB.Size;
    const GLuint aSize = ctx->PixelMaps.AtoA.Size;
-   const uint texSize = pt->width[0];
+   const uint texSize = pt->width0;
    uint *dest;
    uint i, j;
 
diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c
index 705e3ae284..a1b0d93f9a 100644
--- a/src/mesa/state_tracker/st_atom_shader.c
+++ b/src/mesa/state_tracker/st_atom_shader.c
@@ -56,208 +56,18 @@
 #include "st_mesa_to_tgsi.h"
 
 
-/**
- * This represents a vertex program, especially translated to match
- * the inputs of a particular fragment shader.
- */
-struct translated_vertex_program
-{
-   struct st_vertex_program *master;
-
-   /** The fragment/geometry shader "signature" this vertex
-    * shader is meant for: */
-   GLbitfield linked_inputs;
-
-   /** Compared against master vertex program's serialNo: */
-   GLuint serialNo;
-
-   /** Maps VERT_RESULT_x to slot */
-   GLuint output_to_slot[VERT_RESULT_MAX];
-   ubyte output_to_semantic_name[VERT_RESULT_MAX];
-   ubyte output_to_semantic_index[VERT_RESULT_MAX];
-
-   /** Pointer to the translated vertex program */
-   struct st_vertex_program *vp;
-
-   struct translated_vertex_program *next;  /**< next in linked list */
-};
-
-struct translated_geometry_program
-{
-   struct st_geometry_program *master;
-
-   /** The fragment shader "signature" this vertex
-    * shader is meant for: */
-   GLbitfield frag_inputs;
-
-   /** Compared against master vertex program's serialNo: */
-   GLuint serialNo;
-
-   /** Maps VERT_RESULT_x to slot */
-   GLuint output_to_slot[VERT_RESULT_MAX];
-   ubyte output_to_semantic_name[VERT_RESULT_MAX];
-   ubyte output_to_semantic_index[VERT_RESULT_MAX];
-
-   /** Pointer to the translated vertex program */
-   struct st_geometry_program *gp;
 
-   struct translated_geometry_program *next;  /**< next in linked list */
-};
 
 
-/**
- * Given a vertex program output attribute, return the corresponding
- * fragment program input attribute.
- * \return -1 for vertex outputs that have no corresponding fragment input
- */
-static GLint
-vp_out_to_fp_in(GLuint vertResult)
-{
-   if (vertResult >= VERT_RESULT_TEX0 &&
-       vertResult < VERT_RESULT_TEX0 + MAX_TEXTURE_COORD_UNITS)
-      return FRAG_ATTRIB_TEX0 + (vertResult - VERT_RESULT_TEX0);
-
-   if (vertResult >= VERT_RESULT_VAR0 &&
-       vertResult < VERT_RESULT_VAR0 + MAX_VARYING)
-      return FRAG_ATTRIB_VAR0 + (vertResult - VERT_RESULT_VAR0);
-
-   switch (vertResult) {
-   case VERT_RESULT_HPOS:
-      return FRAG_ATTRIB_WPOS;
-   case VERT_RESULT_COL0:
-      return FRAG_ATTRIB_COL0;
-   case VERT_RESULT_COL1:
-      return FRAG_ATTRIB_COL1;
-   case VERT_RESULT_FOGC:
-      return FRAG_ATTRIB_FOGC;
-   default:
-      /* Back-face colors, edge flags, etc */
-      return -1;
-   }
-}
-
-/**
- * Given a geometry program output attribute, return the corresponding
- * fragment program input attribute.
- * \return -1 for geometry outputs that have no corresponding fragment input
- */
-static GLint
-gp_out_to_fp_in(GLuint geomResult)
-{
-   if (geomResult >= GEOM_RESULT_TEX0 &&
-       geomResult < GEOM_RESULT_TEX0 + MAX_TEXTURE_COORD_UNITS)
-      return FRAG_ATTRIB_TEX0 + (geomResult - GEOM_RESULT_TEX0);
-
-   if (geomResult >= GEOM_RESULT_VAR0 &&
-       geomResult < GEOM_RESULT_VAR0 + MAX_VARYING - 2)
-      return FRAG_ATTRIB_VAR0 + (geomResult - GEOM_RESULT_VAR0);
-
-   switch (geomResult) {
-   case GEOM_RESULT_POS:
-      return FRAG_ATTRIB_WPOS;
-   case GEOM_RESULT_COL0:
-      return FRAG_ATTRIB_COL0;
-   case GEOM_RESULT_COL1:
-      return FRAG_ATTRIB_COL1;
-   case GEOM_RESULT_FOGC:
-      return FRAG_ATTRIB_FOGC;
-   default:
-      /* Back-face colors, primitive id, etc */
-      return -1;
-   }
-}
-
-/**
- * Given a vertex program output attribute, return the corresponding
- * geometry program input attribute.
- * \return -1 for vertex outputs that have no corresponding geometry input
- */
-static GLint
-vp_out_to_gp_in(GLuint vertResult)
-{
-   if (vertResult >= VERT_RESULT_TEX0 &&
-       vertResult < VERT_RESULT_TEX0 + MAX_TEXTURE_COORD_UNITS)
-      return GEOM_ATTRIB_TEX_COORD;
-
-   if (vertResult >= VERT_RESULT_VAR0 &&
-       vertResult < VERT_RESULT_VAR0 + MAX_VARYING)
-      return GEOM_ATTRIB_VAR0 + (vertResult - VERT_RESULT_VAR0);
-
-   switch (vertResult) {
-   case VERT_RESULT_HPOS:
-      return GEOM_ATTRIB_POSITION;
-   case VERT_RESULT_COL0:
-      return GEOM_ATTRIB_COLOR0;
-   case VERT_RESULT_COL1:
-      return GEOM_ATTRIB_COLOR1;
-   case VERT_RESULT_FOGC:
-      return GEOM_ATTRIB_FOG_FRAG_COORD;
-   case VERT_RESULT_PSIZ:
-      return GEOM_ATTRIB_POINT_SIZE;
-   case VERT_RESULT_BFC0:
-      return GEOM_ATTRIB_SECONDARY_COLOR0;
-   case VERT_RESULT_BFC1:
-      return GEOM_ATTRIB_SECONDARY_COLOR1;
-   default:
-      /* edge flags, etc */
-      return -1;
-   }
-}
-
-static INLINE int num_vertices_for_prim(int prim)
-{
-   switch(prim) {
-   case PIPE_PRIM_POINTS:
-      return 1;
-   case PIPE_PRIM_LINES:
-      return 2;
-   case PIPE_PRIM_LINE_LOOP:
-      return 2;
-   case PIPE_PRIM_LINE_STRIP:
-      return 2;
-   case PIPE_PRIM_TRIANGLES:
-      return 3;
-   case PIPE_PRIM_TRIANGLE_STRIP:
-      return 3;
-   case PIPE_PRIM_TRIANGLE_FAN:
-      return 3;
-   case PIPE_PRIM_LINES_ADJACENCY:
-   case PIPE_PRIM_LINE_STRIP_ADJACENCY:
-      return 4;
-   case PIPE_PRIM_TRIANGLES_ADJACENCY:
-   case PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY:
-      return 6;
-   default:
-      assert(!"Bad geometry shader input");
-      return 0;
-   }
-}
-
-/**
- * Find a translated vertex program that corresponds to stvp and
- * has outputs matched to stfp's inputs.
- * This performs vertex and fragment translation (to TGSI) when needed.
+/*
+ * Translate fragment program if needed.
  */
-static struct translated_vertex_program *
-find_translated_vp(struct st_context *st,
-                   struct st_vertex_program *stvp,
-                   struct st_fragment_program *stfp,
-                   struct st_geometry_program *stgp)
+static void
+translate_fp(struct st_context *st,
+             struct st_fragment_program *stfp)
 {
-   static const GLuint UNUSED = ~0;
-   struct translated_vertex_program *xvp;
-   struct translated_geometry_program *xgp;
    const GLbitfield fragInputsRead = stfp->Base.Base.InputsRead;
-   GLbitfield geomInputsRead = 0, linkedInputs = fragInputsRead;
 
-   if (stgp) {
-      geomInputsRead = stgp->Base.Base.InputsRead;
-      linkedInputs = geomInputsRead;
-   }
-
-   /*
-    * Translate fragment program if needed.
-    */
    if (!stfp->state.tokens) {
       GLuint inAttr, numIn = 0;
 
@@ -267,7 +77,7 @@ find_translated_vp(struct st_context *st,
             numIn++;
          }
          else {
-            stfp->input_to_slot[inAttr] = UNUSED;
+            stfp->input_to_slot[inAttr] = -1;
          }
       }
 
@@ -277,344 +87,93 @@ find_translated_vp(struct st_context *st,
 
       st_translate_fragment_program(st, stfp, stfp->input_to_slot);
    }
+}
 
-   /* See if we've got a translated geometry program whose outputs match
-    * the fragment  program's inputs.
-    * XXX This could be a hash lookup, using InputsRead as the key.
-    */
-   if (stgp) {
-      for (xgp = stfp->geometry_programs; xgp; xgp = xgp->next) {
-         if (xgp->master == stgp && xgp->frag_inputs == fragInputsRead)
-            break;
-      }
-      /* No?  Allocate translated gp object now */
-      if (!xgp) {
-         xgp = ST_CALLOC_STRUCT(translated_geometry_program);
-         xgp->frag_inputs = fragInputsRead;
-         xgp->master = stgp;
-
-         xgp->next = stfp->geometry_programs;
-         stfp->geometry_programs = xgp;
-      }
-
-      if (!stgp->state.shader.tokens) {
-         GLuint inAttr, numIn = 0;
-
-         for (inAttr = 0; inAttr < GEOM_ATTRIB_MAX; inAttr++) {
-            if (geomInputsRead & (1 << inAttr)) {
-               stgp->input_to_slot[inAttr] = numIn;
-               switch(inAttr) {
-               case GEOM_ATTRIB_VERTICES:
-               case GEOM_ATTRIB_PRIMITIVE_ID:
-                  ++numIn;
-                  break;
-               default:
-                  numIn += num_vertices_for_prim(stgp->Base.InputType);
-                  break;
-               }
-#if 1
-               debug_printf("%d attr = slot %d\n",
-                            inAttr, stgp->input_to_slot[inAttr]);
-#endif
-            }
-            else {
-               stgp->input_to_slot[inAttr] = UNUSED;
-            }
-         }
-
-         stgp->num_input_slots = numIn;
-         assert(stgp->Base.Base.NumInstructions > 1);
-
-         if (xgp->serialNo != stgp->serialNo) {
-            GLuint outAttr;
-            const GLbitfield outputsWritten = stgp->Base.Base.OutputsWritten;
-            GLuint numGpOuts = 0;
-            GLboolean emitPntSize = GL_FALSE, emitBFC0 = GL_FALSE, emitBFC1 = GL_FALSE;
-            GLbitfield usedGenerics = 0x0;
-            GLbitfield usedOutputSlots = 0x0;
-
-            /* Compute mapping of geometry program outputs to slots, which depends
-             * on the fragment program's input->slot mapping.
-             */
-            for (outAttr = 0; outAttr < GEOM_RESULT_MAX; outAttr++) {
-               /* set defaults: */
-               xgp->output_to_slot[outAttr] = UNUSED;
-               xgp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_COUNT;
-               xgp->output_to_semantic_index[outAttr] = 99;
-
-               if (outAttr == GEOM_RESULT_POS) {
-                  /* always put xformed position into slot zero */
-                  GLuint slot = 0;
-                  xgp->output_to_slot[GEOM_RESULT_POS] = slot;
-                  xgp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_POSITION;
-                  xgp->output_to_semantic_index[outAttr] = 0;
-                  numGpOuts++;
-                  usedOutputSlots |= (1 << slot);
-               }
-               else if (outputsWritten & (1 << outAttr)) {
-                  /* see if the frag prog wants this geom output */
-                  GLint fpInAttrib = gp_out_to_fp_in(outAttr);
-                  if (fpInAttrib >= 0) {
-                     GLuint fpInSlot = stfp->input_to_slot[fpInAttrib];
-                     if (fpInSlot != ~0) {
-                        /* match this gp output to the fp input */
-                        GLuint gpOutSlot = stfp->input_map[fpInSlot];
-                        xgp->output_to_slot[outAttr] = gpOutSlot;
-                        xgp->output_to_semantic_name[outAttr] = stfp->input_semantic_name[fpInSlot];
-                        xgp->output_to_semantic_index[outAttr] = stfp->input_semantic_index[fpInSlot];
-                        numGpOuts++;
-                        usedOutputSlots |= (1 << gpOutSlot);
-                     }
-                     else {
-#if 0 /*debug*/
-                        printf("GP output %d not used by FP\n", outAttr);
-#endif
-                     }
-                  }
-                  else if (outAttr == GEOM_RESULT_PSIZ)
-                     emitPntSize = GL_TRUE;
-                  else if (outAttr == GEOM_RESULT_SCOL0)
-                     emitBFC0 = GL_TRUE;
-                  else if (outAttr == GEOM_RESULT_SCOL1)
-                     emitBFC1 = GL_TRUE;
-               }
-#if 0 /*debug*/
-               printf("assign gp output_to_slot[%d] = %d\n", outAttr,
-                      xgp->output_to_slot[outAttr]);
-#endif
-            }
-
-            /* must do these last */
-            if (emitPntSize) {
-               GLuint slot = numGpOuts++;
-               xgp->output_to_slot[GEOM_RESULT_PSIZ] = slot;
-               xgp->output_to_semantic_name[GEOM_RESULT_PSIZ] = TGSI_SEMANTIC_PSIZE;
-               xgp->output_to_semantic_index[GEOM_RESULT_PSIZ] = 0;
-               usedOutputSlots |= (1 << slot);
-            }
-            if (emitBFC0) {
-               GLuint slot = numGpOuts++;
-               xgp->output_to_slot[GEOM_RESULT_SCOL0] = slot;
-               xgp->output_to_semantic_name[GEOM_RESULT_SCOL0] = TGSI_SEMANTIC_COLOR;
-               xgp->output_to_semantic_index[GEOM_RESULT_SCOL0] = 0;
-               usedOutputSlots |= (1 << slot);
-            }
-            if (emitBFC1) {
-               GLuint slot = numGpOuts++;
-               xgp->output_to_slot[GEOM_RESULT_SCOL1] = slot;
-               xgp->output_to_semantic_name[GEOM_RESULT_SCOL1] = TGSI_SEMANTIC_COLOR;
-               xgp->output_to_semantic_index[GEOM_RESULT_SCOL1] = 1;
-               usedOutputSlots |= (1 << slot);
-            }
-
-            /* build usedGenerics mask */
-            usedGenerics = 0x0;
-            for (outAttr = 0; outAttr < GEOM_RESULT_MAX; outAttr++) {
-               if (xgp->output_to_semantic_name[outAttr] == TGSI_SEMANTIC_GENERIC) {
-                  usedGenerics |= (1 << xgp->output_to_semantic_index[outAttr]);
-               }
-            }
-
-            /* For each vertex geometry output that doesn't match up to a fragment
-             * program input, map the geometry program output to a free slot and
-             * free generic attribute.
-             */
-            for (outAttr = 0; outAttr < GEOM_RESULT_MAX; outAttr++) {
-               if (outputsWritten & (1 << outAttr)) {
-                  if (xgp->output_to_slot[outAttr] == UNUSED) {
-                     GLint freeGeneric = _mesa_ffs(~usedGenerics) - 1;
-                     GLint freeSlot = _mesa_ffs(~usedOutputSlots) - 1;
-                     usedGenerics |= (1 << freeGeneric);
-                     usedOutputSlots |= (1 << freeSlot);
-                     xgp->output_to_slot[outAttr] = freeSlot;
-                     xgp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_GENERIC;
-                     xgp->output_to_semantic_index[outAttr] = freeGeneric;
-                  }
-               }
-
-#if 0 /*debug*/
-               printf("gp output_to_slot[%d] = %d\n", outAttr,
-                      xgp->output_to_slot[outAttr]);
-#endif
-            }
-         }
 
-         st_translate_geometry_program(st, stgp, stgp->input_to_slot,
-                                       xgp->output_to_slot,
-                                       xgp->output_to_semantic_name,
-                                       xgp->output_to_semantic_index);
+/*
+ * Translate geometry program if needed.
+ */
+static void
+translate_gp(struct st_context *st,
+             struct st_geometry_program *stgp)
+{
+   const GLbitfield geomInputsRead = stgp->Base.Base.InputsRead;
 
-         xgp->gp = stgp;
+   if (!stgp->state.shader.tokens) {
+      GLuint inAttr, numIn = 0;
 
-         /* translated GP is up to date now */
-         xgp->serialNo = stgp->serialNo;
+      for (inAttr = 0; inAttr < GEOM_ATTRIB_MAX; inAttr++) {
+         if (geomInputsRead & (1 << inAttr)) {
+            stgp->input_to_slot[inAttr] = numIn;
+            numIn++;
+         }
+         else {
+            stgp->input_to_slot[inAttr] = -1;
+         }
       }
-   }
-   /* See if we've got a translated vertex program whose outputs match
-    * the fragment or geometry program's inputs.
-    * XXX This could be a hash lookup, using InputsRead as the key.
-    */
-#define LP_FUNC(FP, GP) ((stgp)?GP:FP)
-#define LP(X) ((stgp)?stgp->X:stfp->X)
-#define LP_SET(X, Y) do { if (stgp) stgp->X = Y; else stfp->X = Y; } while(0)
 
-   for (xvp = LP(vertex_programs); xvp; xvp = xvp->next) {
-      if (xvp->master == stvp && xvp->linked_inputs == linkedInputs)
-         break;
-   }
+      stgp->num_input_slots = numIn;
 
-   /* No?  Allocate translated vp object now */
-   if (!xvp) {
-      xvp = ST_CALLOC_STRUCT(translated_vertex_program);
-      xvp->linked_inputs = linkedInputs;
-      xvp->master = stvp;
+      assert(stgp->Base.Base.NumInstructions > 1);
 
-      xvp->next = LP(vertex_programs);
-      LP_SET(vertex_programs, xvp);
+      st_translate_geometry_program(st, stgp, stgp->input_to_slot);
    }
+}
 
-   /* See if we need to translate vertex program to TGSI form */
-   if (xvp->serialNo != stvp->serialNo) {
-      GLuint outAttr;
-      const GLbitfield outputsWritten = stvp->Base.Base.OutputsWritten;
-      GLuint numVpOuts = 0;
-      GLboolean emitPntSize = GL_FALSE, emitBFC0 = GL_FALSE, emitBFC1 = GL_FALSE;
-      GLbitfield usedGenerics = 0x0;
-      GLbitfield usedOutputSlots = 0x0;
-
-      /* Compute mapping of vertex program outputs to slots, which depends
-       * on the fragment program's input->slot mapping.
-       */
-      for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
-         /* set defaults: */
-         xvp->output_to_slot[outAttr] = UNUSED;
-         xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_COUNT;
-         xvp->output_to_semantic_index[outAttr] = 99;
-
-         if (outAttr == VERT_RESULT_HPOS) {
-            /* always put xformed position into slot zero */
-            GLuint slot = 0;
-            xvp->output_to_slot[VERT_RESULT_HPOS] = slot;
-            xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_POSITION;
-            xvp->output_to_semantic_index[outAttr] = 0;
-            numVpOuts++;
-            usedOutputSlots |= (1 << slot);
-         }
-         else if (outputsWritten & (1 << outAttr)) {
-            /* see if the frag prog wants this vert output */
-            GLint lpInAttrib = LP_FUNC(vp_out_to_fp_in(outAttr), vp_out_to_gp_in(outAttr));
-            if (lpInAttrib >= 0) {
-               GLuint lpInSlot = LP(input_to_slot[lpInAttrib]);
-               if (lpInSlot != ~0) {
-                  /* match this vp output to the lp input */
-                  GLuint vpOutSlot = LP(input_map[lpInSlot]);
-                  xvp->output_to_slot[outAttr] = vpOutSlot;
-                  xvp->output_to_semantic_name[outAttr] = LP(input_semantic_name[lpInSlot]);
-                  xvp->output_to_semantic_index[outAttr] = LP(input_semantic_index[lpInSlot]);
-                  numVpOuts++;
-                  usedOutputSlots |= (1 << vpOutSlot);
-               }
-               else {
-#if 0 /*debug*/
-                  printf("VP output %d not used by LP\n", outAttr);
-#endif
-               }
-            }
-            else if (outAttr == VERT_RESULT_PSIZ)
-               emitPntSize = GL_TRUE;
-            else if (outAttr == VERT_RESULT_BFC0)
-               emitBFC0 = GL_TRUE;
-            else if (outAttr == VERT_RESULT_BFC1)
-               emitBFC1 = GL_TRUE;
-         }
-#if 0 /*debug*/
-         printf("assign vp output_to_slot[%d] = %d\n", outAttr, 
-                xvp->output_to_slot[outAttr]);
-#endif
-      }
 
-      /* must do these last */
-      if (emitPntSize) {
-         GLuint slot = numVpOuts++;
-         xvp->output_to_slot[VERT_RESULT_PSIZ] = slot;
-         xvp->output_to_semantic_name[VERT_RESULT_PSIZ] = TGSI_SEMANTIC_PSIZE;
-         xvp->output_to_semantic_index[VERT_RESULT_PSIZ] = 0;
-         usedOutputSlots |= (1 << slot);
-      }
-      if (emitBFC0) {
-         GLuint slot = numVpOuts++;
-         xvp->output_to_slot[VERT_RESULT_BFC0] = slot;
-         xvp->output_to_semantic_name[VERT_RESULT_BFC0] = TGSI_SEMANTIC_COLOR;
-         xvp->output_to_semantic_index[VERT_RESULT_BFC0] = 0;
-         usedOutputSlots |= (1 << slot);
-      }
-      if (emitBFC1) {
-         GLuint slot = numVpOuts++;
-         xvp->output_to_slot[VERT_RESULT_BFC1] = slot;
-         xvp->output_to_semantic_name[VERT_RESULT_BFC1] = TGSI_SEMANTIC_COLOR;
-         xvp->output_to_semantic_index[VERT_RESULT_BFC1] = 1;
-         usedOutputSlots |= (1 << slot);
-      }
+/**
+ * Find a translated vertex program that corresponds to stvp and
+ * has outputs matched to stfp's inputs.
+ * This performs vertex and fragment translation (to TGSI) when needed.
+ */
+static struct st_vp_varient *
+find_translated_vp(struct st_context *st,
+                   struct st_vertex_program *stvp )
+{
+   struct st_vp_varient *vpv;
+   struct st_vp_varient_key key;
 
-      /* build usedGenerics mask */
-      usedGenerics = 0x0;
-      for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
-         if (xvp->output_to_semantic_name[outAttr] == TGSI_SEMANTIC_GENERIC) {
-            usedGenerics |= (1 << xvp->output_to_semantic_index[outAttr]);
-         }
-      }
+   /* Nothing in our key yet.  This will change:
+    */
+   memset(&key, 0, sizeof key);
+   key.dummy = 0;
 
-      /* For each vertex program output that doesn't match up to a fragment
-       * program input, map the vertex program output to a free slot and
-       * free generic attribute.
+   /* Do we need to throw away old translations after a change in the
+    * GL program string?
+    */
+   if (stvp->serialNo != stvp->lastSerialNo) {
+      /* These may have changed if the program string changed.
        */
-      for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
-         if (outputsWritten & (1 << outAttr)) {
-            if (xvp->output_to_slot[outAttr] == UNUSED) {
-               GLint freeGeneric = _mesa_ffs(~usedGenerics) - 1;
-               GLint freeSlot = _mesa_ffs(~usedOutputSlots) - 1;
-               usedGenerics |= (1 << freeGeneric);
-               usedOutputSlots |= (1 << freeSlot);
-               xvp->output_to_slot[outAttr] = freeSlot;
-               xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_GENERIC;
-               xvp->output_to_semantic_index[outAttr] = freeGeneric;
-            }
-         }
+      st_prepare_vertex_program( st, stvp );
 
-#if 0 /*debug*/
-         printf("vp output_to_slot[%d] = %d\n", outAttr, 
-                xvp->output_to_slot[outAttr]);
-#endif
+      /* We are now up-to-date:
+       */
+      stvp->lastSerialNo = stvp->serialNo;
+   }
+   
+   /* See if we've got a translated vertex program whose outputs match
+    * the fragment program's inputs.
+    */
+   for (vpv = stvp->varients; vpv; vpv = vpv->next) {
+      if (memcmp(&vpv->key, &key, sizeof key) == 0) {
+         break;
       }
+   }
 
-      assert(stvp->Base.Base.NumInstructions > 1);
-
-      st_translate_vertex_program(st, stvp, xvp->output_to_slot,
-                                  xvp->output_to_semantic_name,
-                                  xvp->output_to_semantic_index);
-
-      xvp->vp = stvp;
-
-      /* translated VP is up to date now */
-      xvp->serialNo = stvp->serialNo;
+   /* No?  Perform new translation here. */
+   if (!vpv) {
+      vpv = st_translate_vertex_program(st, stvp, &key);
+      if (!vpv)
+         return NULL;
+      
+      vpv->next = stvp->varients;
+      stvp->varients = vpv;
    }
 
-   return xvp;
+   return vpv;
 }
 
 
-void
-st_free_translated_vertex_programs(struct st_context *st,
-                                   struct translated_vertex_program *xvp)
-{
-   struct translated_vertex_program *next;
-
-   while (xvp) {
-      next = xvp->next;
-      _mesa_free(xvp);
-      xvp = next;
-   }
-}
 
 
 static void *
@@ -628,38 +187,18 @@ get_passthrough_fs(struct st_context *st)
    return st->passthrough_fs;
 }
 
-
 static void
-update_linkage( struct st_context *st )
+update_fp( struct st_context *st )
 {
-   struct st_vertex_program *stvp;
    struct st_fragment_program *stfp;
-   struct st_geometry_program *stgp = NULL;
-   struct translated_vertex_program *xvp;
-
-   /* find active shader and params -- Should be covered by
-    * ST_NEW_VERTEX_PROGRAM
-    */
-   assert(st->ctx->VertexProgram._Current);
-   stvp = st_vertex_program(st->ctx->VertexProgram._Current);
-   assert(stvp->Base.Base.Target == GL_VERTEX_PROGRAM_ARB);
 
    assert(st->ctx->FragmentProgram._Current);
    stfp = st_fragment_program(st->ctx->FragmentProgram._Current);
    assert(stfp->Base.Base.Target == GL_FRAGMENT_PROGRAM_ARB);
 
-   if (st->ctx->GeometryProgram._Current) {
-      stgp = st_geometry_program(st->ctx->GeometryProgram._Current);
-      assert(stgp->Base.Base.Target == MESA_GEOMETRY_PROGRAM);
-   }
+   translate_fp(st, stfp);
 
-   xvp = find_translated_vp(st, stvp, stfp, stgp);
-
-   st_reference_vertprog(st, &st->vp, stvp);
    st_reference_fragprog(st, &st->fp, stfp);
-   st_reference_geomprog(st, &st->gp, stgp);
-
-   cso_set_vertex_shader_handle(st->cso_context, stvp->driver_shader);
 
    if (st->missing_textures) {
       /* use a pass-through frag shader that uses no textures */
@@ -669,18 +208,75 @@ update_linkage( struct st_context *st )
    else {
       cso_set_fragment_shader_handle(st->cso_context, stfp->driver_shader);
    }
-   if (stgp && stgp->driver_shader)
-      cso_set_geometry_shader_handle(st->cso_context, stgp->driver_shader);
+}
 
-   st->vertex_result_to_slot = xvp->output_to_slot;
+const struct st_tracked_state st_update_fp = {
+   "st_update_fp",					/* name */
+   {							/* dirty */
+      0,						/* mesa */
+      ST_NEW_FRAGMENT_PROGRAM                           /* st */
+   },
+   update_fp  					/* update */
+};
+
+
+
+
+static void
+update_vp( struct st_context *st )
+{
+   struct st_vertex_program *stvp;
+
+   /* find active shader and params -- Should be covered by
+    * ST_NEW_VERTEX_PROGRAM
+    */
+   assert(st->ctx->VertexProgram._Current);
+   stvp = st_vertex_program(st->ctx->VertexProgram._Current);
+   assert(stvp->Base.Base.Target == GL_VERTEX_PROGRAM_ARB);
+
+   st->vp_varient = find_translated_vp(st, stvp);
+
+   st_reference_vertprog(st, &st->vp, stvp);
+
+   cso_set_vertex_shader_handle(st->cso_context, 
+                                st->vp_varient->driver_shader);
+
+   st->vertex_result_to_slot = stvp->result_to_output;
 }
 
 
-const struct st_tracked_state st_update_shader = {
-   "st_update_shader",					/* name */
+const struct st_tracked_state st_update_vp = {
+   "st_update_vp",					/* name */
    {							/* dirty */
       0,						/* mesa */
-      ST_NEW_VERTEX_PROGRAM | ST_NEW_FRAGMENT_PROGRAM | ST_NEW_GEOMETRY_PROGRAM	/* st */
+      ST_NEW_VERTEX_PROGRAM                             /* st */
    },
-   update_linkage					/* update */
+   update_vp					/* update */
 };
+
+
+static void
+update_gp( struct st_context *st )
+{
+   struct st_geometry_program *stgp;
+
+   assert(st->ctx->GeometryProgram._Current);
+   stgp = st_geometry_program(st->ctx->GeometryProgram._Current);
+   assert(stgp->Base.Base.Target == MESA_GEOMETRY_PROGRAM);
+
+   translate_gp(st, stgp);
+
+   st_reference_geomprog(st, &st->gp, stgp);
+
+   cso_set_geometry_shader_handle(st->cso_context, stgp->driver_shader);
+}
+
+const struct st_tracked_state st_update_gp = {
+   "st_update_gp",					/* name */
+   {							/* dirty */
+      0,						/* mesa */
+      ST_NEW_GEOMETRY_PROGRAM                           /* st */
+   },
+   update_gp  					/* update */
+};
+
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index a22fa68299..1bdeaccda3 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -169,11 +169,6 @@ make_bitmap_fragment_program(GLcontext *ctx, GLuint samplerIndex)
    stfp = (struct st_fragment_program *) p;
    stfp->Base.UsesKill = GL_TRUE;
 
-   /* No need to send this incomplete program down to hardware:
-    *
-    * st_translate_fragment_program(ctx->st, stfp, NULL);
-    */
-
    return stfp;
 }
 
@@ -526,7 +521,7 @@ reset_cache(struct st_context *st)
    struct pipe_screen *screen = pipe->screen;
    struct bitmap_cache *cache = st->bitmap.cache;
 
-   //memset(cache->buffer, 0xff, sizeof(cache->buffer));
+   /*memset(cache->buffer, 0xff, sizeof(cache->buffer));*/
    cache->empty = GL_TRUE;
 
    cache->xmin = 1000000;
diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index 1d33e81c2c..a68a29e126 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -60,6 +60,7 @@
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "tgsi/tgsi_ureg.h"
 #include "util/u_tile.h"
 #include "util/u_draw_quad.h"
 #include "util/u_math.h"
@@ -236,78 +237,41 @@ make_fragment_shader_z(struct st_context *st)
  * Create a simple vertex shader that just passes through the
  * vertex position and texcoord (and optionally, color).
  */
-static struct st_vertex_program *
-st_make_passthrough_vertex_shader(struct st_context *st, GLboolean passColor)
+static void *
+st_make_passthrough_vertex_shader(struct st_context *st, 
+                                  GLboolean passColor)
 {
-   GLcontext *ctx = st->ctx;
-   struct st_vertex_program *stvp;
-   struct gl_program *p;
-   GLuint ic = 0;
-
-   if (st->drawpix.vert_shaders[passColor])
-      return st->drawpix.vert_shaders[passColor];
-
-   /*
-    * Create shader now
-    */
-   p = ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0);
-   if (!p)
-      return NULL;
-
-   if (passColor)
-      p->NumInstructions = 4;
-   else
-      p->NumInstructions = 3;
-
-   p->Instructions = _mesa_alloc_instructions(p->NumInstructions);
-   if (!p->Instructions) {
-      ctx->Driver.DeleteProgram(ctx, p);
-      return NULL;
-   }
-   _mesa_init_instructions(p->Instructions, p->NumInstructions);
-   /* MOV result.pos, vertex.pos; */
-   p->Instructions[0].Opcode = OPCODE_MOV;
-   p->Instructions[0].DstReg.File = PROGRAM_OUTPUT;
-   p->Instructions[0].DstReg.Index = VERT_RESULT_HPOS;
-   p->Instructions[0].SrcReg[0].File = PROGRAM_INPUT;
-   p->Instructions[0].SrcReg[0].Index = VERT_ATTRIB_POS;
-   /* MOV result.texcoord0, vertex.texcoord0; */
-   p->Instructions[1].Opcode = OPCODE_MOV;
-   p->Instructions[1].DstReg.File = PROGRAM_OUTPUT;
-   p->Instructions[1].DstReg.Index = VERT_RESULT_TEX0;
-   p->Instructions[1].SrcReg[0].File = PROGRAM_INPUT;
-   p->Instructions[1].SrcReg[0].Index = VERT_ATTRIB_TEX0;
-   ic = 2;
-   if (passColor) {
-      /* MOV result.color0, vertex.color0; */
-      p->Instructions[ic].Opcode = OPCODE_MOV;
-      p->Instructions[ic].DstReg.File = PROGRAM_OUTPUT;
-      p->Instructions[ic].DstReg.Index = VERT_RESULT_COL0;
-      p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT;
-      p->Instructions[ic].SrcReg[0].Index = VERT_ATTRIB_COLOR0;
-      ic++;
-   }
-
-   /* END; */
-   p->Instructions[ic].Opcode = OPCODE_END;
-   ic++;
-
-   assert(ic == p->NumInstructions);
+   if (!st->drawpix.vert_shaders[passColor]) {
+      struct ureg_program *ureg = 
+         ureg_create( TGSI_PROCESSOR_VERTEX );
+
+      if (ureg == NULL)
+         return NULL;
+
+      /* MOV result.pos, vertex.pos; */
+      ureg_MOV(ureg, 
+               ureg_DECL_output( ureg, TGSI_SEMANTIC_POSITION, 0 ),
+               ureg_DECL_vs_input( ureg, 0 ));
+      
+      /* MOV result.texcoord0, vertex.texcoord0; */
+      ureg_MOV(ureg, 
+               ureg_DECL_output( ureg, TGSI_SEMANTIC_GENERIC, 0 ),
+               ureg_DECL_vs_input( ureg, 1 ));
+      
+      if (passColor) {
+         /* MOV result.color0, vertex.color0; */
+         ureg_MOV(ureg, 
+                  ureg_DECL_output( ureg, TGSI_SEMANTIC_COLOR, 0 ),
+                  ureg_DECL_vs_input( ureg, 2 ));
+      }
 
-   p->InputsRead = VERT_BIT_POS | VERT_BIT_TEX0;
-   p->OutputsWritten = ((1 << VERT_RESULT_TEX0) |
-                        (1 << VERT_RESULT_HPOS));
-   if (passColor) {
-      p->InputsRead |= VERT_BIT_COLOR0;
-      p->OutputsWritten |= (1 << VERT_RESULT_COL0);
+      ureg_END( ureg );
+      
+      st->drawpix.vert_shaders[passColor] = 
+         ureg_create_shader_and_destroy( ureg, st->pipe );
    }
 
-   stvp = (struct st_vertex_program *) p;
-   st_translate_vertex_program(st, stvp, NULL, NULL, NULL);
-
-   st->drawpix.vert_shaders[passColor] = stvp;
-
-   return stvp;
+   return st->drawpix.vert_shaders[passColor];
 }
 
 
@@ -539,8 +503,8 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
                    GLsizei width, GLsizei height,
                    GLfloat zoomX, GLfloat zoomY,
                    struct pipe_texture *pt,
-                   struct st_vertex_program *stvp,
-                   struct st_fragment_program *stfp,
+                   void *driver_vp,
+                   void *driver_fp,
                    const GLfloat *color,
                    GLboolean invertTex)
 {
@@ -575,10 +539,10 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
    }
 
    /* fragment shader state: TEX lookup program */
-   cso_set_fragment_shader_handle(cso, stfp->driver_shader);
+   cso_set_fragment_shader_handle(cso, driver_fp);
 
    /* vertex shader state: position + texcoord pass-through */
-   cso_set_vertex_shader_handle(cso, stvp->driver_shader);
+   cso_set_vertex_shader_handle(cso, driver_vp);
 
 
    /* texture sampling state: */
@@ -637,8 +601,8 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
    y1 = y + height * ctx->Pixel.ZoomY;
 
    draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex,
-	     (GLfloat) width / pt->width[0],
-	     (GLfloat) height / pt->height[0]);
+	     (GLfloat) width / pt->width0,
+	     (GLfloat) height / pt->height0);
 
    /* restore state */
    cso_restore_rasterizer(cso);
@@ -806,7 +770,7 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
               const struct gl_pixelstore_attrib *unpack, const GLvoid *pixels)
 {
    struct st_fragment_program *stfp;
-   struct st_vertex_program *stvp;
+   void *driver_vp;
    struct st_context *st = st_context(ctx);
    struct pipe_surface *ps;
    const GLfloat *color;
@@ -826,13 +790,13 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
    if (format == GL_DEPTH_COMPONENT) {
       ps = st->state.framebuffer.zsbuf;
       stfp = make_fragment_shader_z(st);
-      stvp = st_make_passthrough_vertex_shader(st, GL_TRUE);
+      driver_vp = st_make_passthrough_vertex_shader(st, GL_TRUE);
       color = ctx->Current.RasterColor;
    }
    else {
       ps = st->state.framebuffer.cbufs[0];
       stfp = combined_drawpix_fragment_program(ctx);
-      stvp = st_make_passthrough_vertex_shader(st, GL_FALSE);
+      driver_vp = st_make_passthrough_vertex_shader(st, GL_FALSE);
       color = NULL;
    }
 
@@ -843,7 +807,10 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height,
       if (pt) {
          draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2],
                             width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY,
-                            pt, stvp, stfp, color, GL_FALSE);
+                            pt, 
+                            driver_vp, 
+                            stfp->driver_shader,
+                            color, GL_FALSE);
          pipe_texture_reference(&pt, NULL);
       }
    }
@@ -960,7 +927,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
    struct pipe_context *pipe = st->pipe;
    struct pipe_screen *screen = pipe->screen;
    struct st_renderbuffer *rbRead;
-   struct st_vertex_program *stvp;
+   void *driver_vp;
    struct st_fragment_program *stfp;
    struct pipe_texture *pt;
    GLfloat *color;
@@ -1009,14 +976,14 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
       rbRead = st_get_color_read_renderbuffer(ctx);
       color = NULL;
       stfp = combined_drawpix_fragment_program(ctx);
-      stvp = st_make_passthrough_vertex_shader(st, GL_FALSE);
+      driver_vp = st_make_passthrough_vertex_shader(st, GL_FALSE);
    }
    else {
       assert(type == GL_DEPTH);
       rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer);
       color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
       stfp = make_fragment_shader_z(st);
-      stvp = st_make_passthrough_vertex_shader(st, GL_TRUE);
+      driver_vp = st_make_passthrough_vertex_shader(st, GL_TRUE);
    }
 
    srcFormat = rbRead->texture->format;
@@ -1148,7 +1115,10 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy,
    /* draw textured quad */
    draw_textured_quad(ctx, dstx, dsty, ctx->Current.RasterPos[2],
                       width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY,
-                      pt, stvp, stfp, color, GL_TRUE);
+                      pt, 
+                      driver_vp, 
+                      stfp->driver_shader,
+                      color, GL_TRUE);
 
    pipe_texture_reference(&pt, NULL);
 }
diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c
index 65ce12ccd4..659a6c9193 100644
--- a/src/mesa/state_tracker/st_cb_fbo.c
+++ b/src/mesa/state_tracker/st_cb_fbo.c
@@ -128,9 +128,9 @@ st_renderbuffer_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
       template.target = PIPE_TEXTURE_2D;
       template.format = format;
       pf_get_block(format, &template.block);
-      template.width[0] = width;
-      template.height[0] = height;
-      template.depth[0] = 1;
+      template.width0 = width;
+      template.height0 = height;
+      template.depth0 = 1;
       template.last_level = 0;
       template.nr_samples = rb->NumSamples;
       if (pf_is_depth_stencil(format)) {
@@ -376,7 +376,7 @@ st_render_texture(GLcontext *ctx,
    rb->_BaseFormat = texImage->_BaseFormat;
    /*printf("***** render to texture level %d: %d x %d\n", att->TextureLevel, rb->Width, rb->Height);*/
 
-   /*printf("***** pipe texture %d x %d\n", pt->width[0], pt->height[0]);*/
+   /*printf("***** pipe texture %d x %d\n", pt->width0, pt->height0);*/
 
    pipe_texture_reference( &strb->texture, pt );
 
diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c
index 96eece82ee..3940ec0606 100644
--- a/src/mesa/state_tracker/st_cb_program.c
+++ b/src/mesa/state_tracker/st_cb_program.c
@@ -150,24 +150,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog)
    case GL_VERTEX_PROGRAM_ARB:
       {
          struct st_vertex_program *stvp = (struct st_vertex_program *) prog;
-
-         if (stvp->driver_shader) {
-            cso_delete_vertex_shader(st->cso_context, stvp->driver_shader);
-            stvp->driver_shader = NULL;
-         }
-
-         if (stvp->draw_shader) {
-#if FEATURE_feedback || FEATURE_drawpix
-            /* this would only have been allocated for the RasterPos path */
-            draw_delete_vertex_shader(st->draw, stvp->draw_shader);
-            stvp->draw_shader = NULL;
-#endif
-         }
-
-         if (stvp->state.tokens) {
-            st_free_tokens(stvp->state.tokens);
-            stvp->state.tokens = NULL;
-         }
+         st_vp_release_varients( st, stvp );
       }
       break;
    case GL_FRAGMENT_PROGRAM_ARB:
@@ -189,8 +172,6 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog)
             _mesa_reference_program(ctx, &prg, NULL);
             stfp->bitmap_program = NULL;
          }
-
-         st_free_translated_vertex_programs(st, stfp->vertex_programs);
       }
       break;
    case MESA_GEOMETRY_PROGRAM:
@@ -246,8 +227,6 @@ static void st_program_string_notify( GLcontext *ctx,
          stfp->state.tokens = NULL;
       }
 
-      stfp->param_state = stfp->Base.Base.Parameters->StateFlags;
-
       if (st->fp == stfp)
 	 st->dirty.st |= ST_NEW_FRAGMENT_PROGRAM;
    }
@@ -256,25 +235,7 @@ static void st_program_string_notify( GLcontext *ctx,
 
       stvp->serialNo++;
 
-      if (stvp->driver_shader) {
-         cso_delete_vertex_shader(st->cso_context, stvp->driver_shader);
-         stvp->driver_shader = NULL;
-      }
-
-      if (stvp->draw_shader) {
-#if FEATURE_feedback || FEATURE_drawpix
-         /* this would only have been allocated for the RasterPos path */
-         draw_delete_vertex_shader(st->draw, stvp->draw_shader);
-         stvp->draw_shader = NULL;
-#endif
-      }
-
-      if (stvp->state.tokens) {
-         st_free_tokens(stvp->state.tokens);
-         stvp->state.tokens = NULL;
-      }
-
-      stvp->param_state = stvp->Base.Base.Parameters->StateFlags;
+      st_vp_release_varients( st, stvp );
 
       if (st->vp == stvp)
 	 st->dirty.st |= ST_NEW_VERTEX_PROGRAM;
diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c
index 772bb3bb69..103861d6f9 100644
--- a/src/mesa/state_tracker/st_cb_readpixels.c
+++ b/src/mesa/state_tracker/st_cb_readpixels.c
@@ -243,7 +243,7 @@ st_fast_readpixels(GLcontext *ctx, struct st_renderbuffer *strb,
       GLint row, col, dy, dstStride;
 
       if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
-         y = strb->texture->height[0] - y - height;
+         y = strb->texture->height0 - y - height;
       }
 
       trans = st_cond_flush_get_tex_transfer(st_context(ctx), strb->texture,
diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index 9186db76e1..3a2337802f 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -93,51 +93,6 @@ gl_target_to_pipe(GLenum target)
 }
 
 
-/**
- * Return nominal bytes per texel for a compressed format, 0 for non-compressed
- * format.
- */
-static GLuint
-compressed_num_bytes(gl_format format)
-{
-   switch (format) {
-#if FEATURE_texture_fxt1
-   case MESA_FORMAT_RGB_FXT1:
-   case MESA_FORMAT_RGBA_FXT1:
-#endif
-#if FEATURE_texture_s3tc
-   case MESA_FORMAT_RGB_DXT1:
-   case MESA_FORMAT_RGBA_DXT1:
-      return 2;
-   case MESA_FORMAT_RGBA_DXT3:
-   case MESA_FORMAT_RGBA_DXT5:
-      return 4;
-#endif
-   default:
-      return 0;
-   }
-}
-
-
-static GLboolean
-is_compressed_mesa_format(gl_format format)
-{
-   switch (format) {
-   case MESA_FORMAT_RGB_DXT1:
-   case MESA_FORMAT_RGBA_DXT1:
-   case MESA_FORMAT_RGBA_DXT3:
-   case MESA_FORMAT_RGBA_DXT5:
-   case MESA_FORMAT_SRGB_DXT1:
-   case MESA_FORMAT_SRGBA_DXT1:
-   case MESA_FORMAT_SRGBA_DXT3:
-   case MESA_FORMAT_SRGBA_DXT5:
-      return GL_TRUE;
-   default:
-      return GL_FALSE;
-   }
-}
-
-
 /** called via ctx->Driver.NewTextureImage() */
 static struct gl_texture_image *
 st_NewTextureImage(GLcontext * ctx)
@@ -451,9 +406,9 @@ compress_with_blit(GLcontext * ctx,
    templ.target = PIPE_TEXTURE_2D;
    templ.format = st_mesa_format_to_pipe_format(mesa_format);
    pf_get_block(templ.format, &templ.block);
-   templ.width[0] = width;
-   templ.height[0] = height;
-   templ.depth[0] = 1;
+   templ.width0 = width;
+   templ.height0 = height;
+   templ.depth0 = 1;
    templ.last_level = 0;
    templ.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
    src_tex = screen->texture_create(screen, &templ);
@@ -663,7 +618,7 @@ st_TexImage(GLcontext * ctx,
     */
    if (!compressed_src &&
        !ctx->Mesa_DXTn &&
-       is_compressed_mesa_format(texImage->TexFormat) &&
+       _mesa_is_format_compressed(texImage->TexFormat) &&
        screen->is_format_supported(screen,
                                    stImage->pt->format,
                                    stImage->pt->target,
@@ -1066,7 +1021,7 @@ st_TexSubimage(GLcontext *ctx, GLint dims, GLenum target, GLint level,
    /* See if we can do texture compression with a blit/render.
     */
    if (!ctx->Mesa_DXTn &&
-       is_compressed_mesa_format(texImage->TexFormat) &&
+       _mesa_is_format_compressed(texImage->TexFormat) &&
        screen->is_format_supported(screen,
                                    stImage->pt->format,
                                    stImage->pt->target,
@@ -1724,8 +1679,6 @@ copy_image_data_to_texture(struct st_context *st,
       pipe_texture_reference(&stImage->pt, NULL);
    }
    else if (stImage->base.Data) {
-      assert(stImage->base.Data != NULL);
-
       /* More straightforward upload.  
        */
 
@@ -1764,7 +1717,7 @@ st_finalize_texture(GLcontext *ctx,
 {
    struct st_texture_object *stObj = st_texture_object(tObj);
    const GLuint nr_faces = (stObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-   GLuint cpp, face;
+   GLuint blockSize, face;
    struct st_texture_image *firstImage;
 
    *needFlush = GL_FALSE;
@@ -1796,13 +1749,8 @@ st_finalize_texture(GLcontext *ctx,
       pipe_texture_reference(&stObj->pt, firstImage->pt);
    }
 
-   /* FIXME: determine format block instead of cpp */
-   if (_mesa_is_format_compressed(firstImage->base.TexFormat)) {
-      cpp = compressed_num_bytes(firstImage->base.TexFormat);
-   }
-   else {
-      cpp = _mesa_get_format_bytes(firstImage->base.TexFormat);
-   }
+   /* bytes per pixel block (blocks are usually 1x1) */
+   blockSize = _mesa_get_format_bytes(firstImage->base.TexFormat);
 
    /* If we already have a gallium texture, check that it matches the texture
     * object's format, target, size, num_levels, etc.
@@ -1813,11 +1761,10 @@ st_finalize_texture(GLcontext *ctx,
       if (stObj->pt->target != gl_target_to_pipe(stObj->base.Target) ||
           stObj->pt->format != fmt ||
           stObj->pt->last_level < stObj->lastLevel ||
-          stObj->pt->width[0] != firstImage->base.Width2 ||
-          stObj->pt->height[0] != firstImage->base.Height2 ||
-          stObj->pt->depth[0] != firstImage->base.Depth2 ||
-          /* Nominal bytes per pixel: */
-          stObj->pt->block.size / stObj->pt->block.width != cpp)
+          stObj->pt->width0 != firstImage->base.Width2 ||
+          stObj->pt->height0 != firstImage->base.Height2 ||
+          stObj->pt->depth0 != firstImage->base.Depth2 ||
+          stObj->pt->block.size != blockSize)
       {
          pipe_texture_reference(&stObj->pt, NULL);
          ctx->st->dirty.st |= ST_NEW_FRAMEBUFFER;
diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h
index 7efaed9e05..a2845adb8b 100644
--- a/src/mesa/state_tracker/st_context.h
+++ b/src/mesa/state_tracker/st_context.h
@@ -129,6 +129,8 @@ struct st_context
    struct st_fragment_program *fp;  /**< Currently bound fragment program */
    struct st_geometry_program *gp;  /**< Currently bound geometry program */
 
+   struct st_vp_varient *vp_varient;
+
    struct gl_texture_object *default_texture;
 
    struct {
diff --git a/src/mesa/state_tracker/st_debug.c b/src/mesa/state_tracker/st_debug.c
index 3009cde9d5..6e699ca552 100644
--- a/src/mesa/state_tracker/st_debug.c
+++ b/src/mesa/state_tracker/st_debug.c
@@ -86,7 +86,8 @@ st_print_current(void)
    }
 #endif
 
-   tgsi_dump( st->vp->state.tokens, 0 );
+   if (st->vp->varients)
+      tgsi_dump( st->vp->varients[0].state.tokens, 0 );
    if (st->vp->Base.Base.Parameters)
       _mesa_print_parameter_list(st->vp->Base.Base.Parameters);
 
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 68bc76b572..e13ae57a0e 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -573,7 +573,7 @@ st_draw_vbo(GLcontext *ctx,
 
    /* must get these after state validation! */
    vp = ctx->st->vp;
-   vs = &ctx->st->vp->state;
+   vs = &ctx->st->vp_varient->state;
 
 #if 0
    if (MESA_VERBOSE & VERBOSE_GLSL) {
diff --git a/src/mesa/state_tracker/st_draw_feedback.c b/src/mesa/state_tracker/st_draw_feedback.c
index dd0b4ef9a4..e9afc0e529 100644
--- a/src/mesa/state_tracker/st_draw_feedback.c
+++ b/src/mesa/state_tracker/st_draw_feedback.c
@@ -120,10 +120,10 @@ st_feedback_draw_vbo(GLcontext *ctx,
 
    /* must get these after state validation! */
    vp = ctx->st->vp;
-   vs = &st->vp->state;
+   vs = &st->vp_varient->state;
 
-   if (!st->vp->draw_shader) {
-      st->vp->draw_shader = draw_create_vertex_shader(draw, vs);
+   if (!st->vp_varient->draw_shader) {
+      st->vp_varient->draw_shader = draw_create_vertex_shader(draw, vs);
    }
 
    /*
@@ -136,7 +136,7 @@ st_feedback_draw_vbo(GLcontext *ctx,
    draw_set_viewport_state(draw, &st->state.viewport);
    draw_set_clip_state(draw, &st->state.clip);
    draw_set_rasterizer_state(draw, &st->state.rasterizer);
-   draw_bind_vertex_shader(draw, st->vp->draw_shader);
+   draw_bind_vertex_shader(draw, st->vp_varient->draw_shader);
    set_feedback_vertex_format(ctx);
 
    /* loop over TGSI shader inputs to determine vertex buffer
diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c
index 16ca2771b0..f8068fa12b 100644
--- a/src/mesa/state_tracker/st_gen_mipmap.c
+++ b/src/mesa/state_tracker/st_gen_mipmap.c
@@ -38,6 +38,7 @@
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
 #include "util/u_gen_mipmap.h"
+#include "util/u_math.h"
 
 #include "cso_cache/cso_cache.h"
 #include "cso_cache/cso_context.h"
@@ -133,14 +134,14 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target,
       srcTrans = st_cond_flush_get_tex_transfer(st_context(ctx), pt, face,
 						srcLevel, zslice,
 						PIPE_TRANSFER_READ, 0, 0,
-						pt->width[srcLevel],
-						pt->height[srcLevel]);
+						u_minify(pt->width0, srcLevel),
+						u_minify(pt->height0, srcLevel));
 
       dstTrans = st_cond_flush_get_tex_transfer(st_context(ctx), pt, face,
 						dstLevel, zslice,
 						PIPE_TRANSFER_WRITE, 0, 0,
-						pt->width[dstLevel],
-						pt->height[dstLevel]);
+						u_minify(pt->width0, dstLevel),
+						u_minify(pt->height0, dstLevel));
 
       srcData = (ubyte *) screen->transfer_map(screen, srcTrans);
       dstData = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -149,13 +150,17 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target,
       dstStride = dstTrans->stride / dstTrans->block.size;
 
       _mesa_generate_mipmap_level(target, datatype, comps,
-                   0 /*border*/,
-                   pt->width[srcLevel], pt->height[srcLevel], pt->depth[srcLevel],
-                   srcData,
-                   srcStride, /* stride in texels */
-                   pt->width[dstLevel], pt->height[dstLevel], pt->depth[dstLevel],
-                   dstData,
-                   dstStride); /* stride in texels */
+                                  0 /*border*/,
+                                  u_minify(pt->width0, srcLevel),
+                                  u_minify(pt->height0, srcLevel),
+                                  u_minify(pt->depth0, srcLevel),
+                                  srcData,
+                                  srcStride, /* stride in texels */
+                                  u_minify(pt->width0, dstLevel),
+                                  u_minify(pt->height0, dstLevel),
+                                  u_minify(pt->depth0, dstLevel),
+                                  dstData,
+                                  dstStride); /* stride in texels */
 
       screen->transfer_unmap(screen, srcTrans);
       screen->transfer_unmap(screen, dstTrans);
@@ -232,9 +237,9 @@ st_generate_mipmap(GLcontext *ctx, GLenum target,
                                     oldTex->target,
                                     oldTex->format,
                                     lastLevel,
-                                    oldTex->width[0],
-                                    oldTex->height[0],
-                                    oldTex->depth[0],
+                                    oldTex->width0,
+                                    oldTex->height0,
+                                    oldTex->depth0,
                                     oldTex->tex_usage);
 
       /* The texture isn't in a "complete" state yet so set the expected
@@ -269,9 +274,9 @@ st_generate_mipmap(GLcontext *ctx, GLenum target,
          = _mesa_get_tex_image(ctx, texObj, target, srcLevel);
       struct gl_texture_image *dstImage;
       struct st_texture_image *stImage;
-      uint dstWidth = pt->width[dstLevel];
-      uint dstHeight = pt->height[dstLevel];
-      uint dstDepth = pt->depth[dstLevel];
+      uint dstWidth = u_minify(pt->width0, dstLevel);
+      uint dstHeight = u_minify(pt->height0, dstLevel);
+      uint dstDepth = u_minify(pt->depth0, dstLevel); 
       uint border = srcImage->Border;
 
       dstImage = _mesa_get_tex_image(ctx, texObj, target, dstLevel);
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index c2121f4585..46f64a7d50 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -708,12 +708,10 @@ st_translate_mesa_program(
    const ubyte inputSemanticName[],
    const ubyte inputSemanticIndex[],
    const GLuint interpMode[],
-   const GLbitfield inputFlags[],
    GLuint numOutputs,
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
-   const ubyte outputSemanticIndex[],
-   const GLbitfield outputFlags[] )
+   const ubyte outputSemanticIndex[] )
 {
    struct st_translate translate, *t;
    struct ureg_program *ureg;
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h
index c0d1ff59e1..dc0362fe79 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.h
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h
@@ -49,12 +49,10 @@ st_translate_mesa_program(
    const ubyte inputSemanticName[],
    const ubyte inputSemanticIndex[],
    const GLuint interpMode[],
-   const GLbitfield inputFlags[],
    GLuint numOutputs,
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
-   const ubyte outputSemanticIndex[],
-   const GLbitfield outputFlags[] );
+   const ubyte outputSemanticIndex[] );
 
 void
 st_free_tokens(const struct tgsi_token *tokens);
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index d928269f03..603497c796 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -49,6 +49,37 @@
 #include "st_mesa_to_tgsi.h"
 #include "cso_cache/cso_context.h"
 
+   /* Clean out any old compilations:
+    */
+void
+st_vp_release_varients( struct st_context *st,
+                        struct st_vertex_program *stvp )
+{
+   struct st_vp_varient *vpv;
+
+   for (vpv = stvp->varients; vpv; ) {
+      struct st_vp_varient *next = vpv->next;
+
+      if (vpv->driver_shader) 
+         cso_delete_vertex_shader(st->cso_context, vpv->driver_shader);
+      
+      if (vpv->draw_shader)
+         draw_delete_vertex_shader( st->draw, vpv->draw_shader );
+      
+      if (vpv->state.tokens)
+         st_free_tokens(vpv->state.tokens);
+      
+      FREE( vpv );
+
+      vpv = next;
+   }
+
+   stvp->varients = NULL;
+}
+
+
+
+
 /**
  * Translate a Mesa vertex shader into a TGSI shader.
  * \param outputMapping  to map vertex program output registers (VERT_RESULT_x)
@@ -57,31 +88,13 @@
  * \return  pointer to cached pipe_shader object.
  */
 void
-st_translate_vertex_program(struct st_context *st,
-                            struct st_vertex_program *stvp,
-                            const GLuint outputMapping[],
-                            const ubyte *outputSemanticName,
-                            const ubyte *outputSemanticIndex)
+st_prepare_vertex_program(struct st_context *st,
+                            struct st_vertex_program *stvp)
 {
-   struct pipe_context *pipe = st->pipe;
-   GLuint defaultOutputMapping[VERT_RESULT_MAX];
-   GLuint attr, i;
-   GLuint num_generic = 0;
-
-   ubyte vs_input_semantic_name[PIPE_MAX_SHADER_INPUTS];
-   ubyte vs_input_semantic_index[PIPE_MAX_SHADER_INPUTS];
-   uint vs_num_inputs = 0;
-
-   ubyte vs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS];
-   ubyte vs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
-   uint vs_num_outputs = 0;
-
-   GLbitfield input_flags[MAX_PROGRAM_INPUTS];
-   GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
+   GLuint attr;
 
-//   memset(&vs, 0, sizeof(vs));
-   memset(input_flags, 0, sizeof(input_flags));
-   memset(output_flags, 0, sizeof(output_flags));
+   stvp->num_inputs = 0;
+   stvp->num_outputs = 0;
 
    if (stvp->Base.IsPositionInvariant)
       _mesa_insert_mvp_code(st->ctx, &stvp->Base);
@@ -92,161 +105,56 @@ st_translate_vertex_program(struct st_context *st,
     */
    for (attr = 0; attr < VERT_ATTRIB_MAX; attr++) {
       if (stvp->Base.Base.InputsRead & (1 << attr)) {
-         const GLuint slot = vs_num_inputs;
-
-         vs_num_inputs++;
-
-         stvp->input_to_index[attr] = slot;
-         stvp->index_to_input[slot] = attr;
-
-         switch (attr) {
-         case VERT_ATTRIB_POS:
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-            vs_input_semantic_index[slot] = 0;
-            break;
-         case VERT_ATTRIB_WEIGHT:
-            /* fall-through */
-         case VERT_ATTRIB_NORMAL:
-            /* just label as a generic */
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            vs_input_semantic_index[slot] = 0;
-            break;
-         case VERT_ATTRIB_COLOR0:
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            vs_input_semantic_index[slot] = 0;
-            break;
-         case VERT_ATTRIB_COLOR1:
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            vs_input_semantic_index[slot] = 1;
-            break;
-         case VERT_ATTRIB_FOG:
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-            vs_input_semantic_index[slot] = 0;
-            break;
-         case VERT_ATTRIB_POINT_SIZE:
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
-            vs_input_semantic_index[slot] = 0;
-            break;
-         case VERT_ATTRIB_TEX0:
-         case VERT_ATTRIB_TEX1:
-         case VERT_ATTRIB_TEX2:
-         case VERT_ATTRIB_TEX3:
-         case VERT_ATTRIB_TEX4:
-         case VERT_ATTRIB_TEX5:
-         case VERT_ATTRIB_TEX6:
-         case VERT_ATTRIB_TEX7:
-            assert(slot < Elements(vs_input_semantic_name));
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            vs_input_semantic_index[slot] = num_generic++;
-            break;
-         case VERT_ATTRIB_GENERIC0:
-         case VERT_ATTRIB_GENERIC1:
-         case VERT_ATTRIB_GENERIC2:
-         case VERT_ATTRIB_GENERIC3:
-         case VERT_ATTRIB_GENERIC4:
-         case VERT_ATTRIB_GENERIC5:
-         case VERT_ATTRIB_GENERIC6:
-         case VERT_ATTRIB_GENERIC7:
-         case VERT_ATTRIB_GENERIC8:
-         case VERT_ATTRIB_GENERIC9:
-         case VERT_ATTRIB_GENERIC10:
-         case VERT_ATTRIB_GENERIC11:
-         case VERT_ATTRIB_GENERIC12:
-         case VERT_ATTRIB_GENERIC13:
-         case VERT_ATTRIB_GENERIC14:
-         case VERT_ATTRIB_GENERIC15:
-            assert(attr < VERT_ATTRIB_MAX);
-            assert(slot < Elements(vs_input_semantic_name));
-            vs_input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            vs_input_semantic_index[slot] = num_generic++;
-            break;
-         default:
-            assert(0);
-         }
-
-         input_flags[slot] = stvp->Base.Base.InputFlags[attr];
+         stvp->input_to_index[attr] = stvp->num_inputs;
+         stvp->index_to_input[stvp->num_inputs] = attr;
+         stvp->num_inputs++;
       }
    }
 
-#if 0
-   if (outputMapping && outputSemanticName) {
-      printf("VERT_RESULT  written  out_slot  semantic_name  semantic_index\n");
-      for (attr = 0; attr < VERT_RESULT_MAX; attr++) {
-         printf("    %-2d          %c       %3d          %2d              %2d\n",
-                attr, 
-                ((stvp->Base.Base.OutputsWritten & (1 << attr)) ? 'Y' : ' '),
-                outputMapping[attr],
-                outputSemanticName[attr],
-                outputSemanticIndex[attr]);
-      }
-   }
-#endif
-
-   /* initialize output semantics to defaults */
-   for (i = 0; i < PIPE_MAX_SHADER_OUTPUTS; i++) {
-      assert(i < Elements(vs_output_semantic_name));
-      vs_output_semantic_name[i] = TGSI_SEMANTIC_GENERIC;
-      vs_output_semantic_index[i] = 0;
-      output_flags[i] = 0x0;
-   }
-
-   num_generic = 0;
-   /*
-    * Determine number of outputs, the (default) output register
-    * mapping and the semantic information for each output.
+   /* Compute mapping of vertex program outputs to slots.
     */
    for (attr = 0; attr < VERT_RESULT_MAX; attr++) {
-      if (stvp->Base.Base.OutputsWritten & (1 << attr)) {
-         GLuint slot;
+      if ((stvp->Base.Base.OutputsWritten & (1 << attr)) == 0) {
+         stvp->result_to_output[attr] = ~0;
+      }
+      else {
+         unsigned slot = stvp->num_outputs++;
 
-         /* XXX
-          * Pass in the fragment program's input's semantic info.
-          * Use the generic semantic indexes from there, instead of
-          * guessing below.
-          */
-         if (outputMapping) {
-            slot = outputMapping[attr];
-            assert(slot != ~0);
-         }
-         else {
-            slot = vs_num_outputs;
-            vs_num_outputs++;
-            defaultOutputMapping[attr] = slot;
-         }
+         stvp->result_to_output[attr] = slot;
 
          switch (attr) {
          case VERT_RESULT_HPOS:
-            assert(slot == 0);
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-            vs_output_semantic_index[slot] = 0;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+            stvp->output_semantic_index[slot] = 0;
             break;
          case VERT_RESULT_COL0:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            vs_output_semantic_index[slot] = 0;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stvp->output_semantic_index[slot] = 0;
             break;
          case VERT_RESULT_COL1:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            vs_output_semantic_index[slot] = 1;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stvp->output_semantic_index[slot] = 1;
             break;
          case VERT_RESULT_BFC0:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            vs_output_semantic_index[slot] = 0;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stvp->output_semantic_index[slot] = 0;
             break;
          case VERT_RESULT_BFC1:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            vs_output_semantic_index[slot] = 1;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stvp->output_semantic_index[slot] = 1;
             break;
          case VERT_RESULT_FOGC:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-            vs_output_semantic_index[slot] = 0;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            stvp->output_semantic_index[slot] = 0;
             break;
          case VERT_RESULT_PSIZ:
-            vs_output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
-            vs_output_semantic_index[slot] = 0;
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
+            stvp->output_semantic_index[slot] = 0;
             break;
          case VERT_RESULT_EDGE:
             assert(0);
             break;
+
          case VERT_RESULT_TEX0:
          case VERT_RESULT_TEX1:
          case VERT_RESULT_TEX2:
@@ -255,92 +163,50 @@ st_translate_vertex_program(struct st_context *st,
          case VERT_RESULT_TEX5:
          case VERT_RESULT_TEX6:
          case VERT_RESULT_TEX7:
-            /* fall-through */
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stvp->output_semantic_index[slot] = attr - VERT_RESULT_TEX0;
+            break;
+
          case VERT_RESULT_VAR0:
-            /* fall-through */
          default:
-            assert(slot < Elements(vs_output_semantic_name));
-            if (outputSemanticName) {
-               /* use provided semantic into */
-               assert(outputSemanticName[attr] != TGSI_SEMANTIC_COUNT);
-               vs_output_semantic_name[slot] = outputSemanticName[attr];
-               vs_output_semantic_index[slot] = outputSemanticIndex[attr];
-            }
-            else {
-               /* use default semantic info */
-               vs_output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-               vs_output_semantic_index[slot] = num_generic++;
-            }
+            assert(attr < VERT_RESULT_MAX);
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stvp->output_semantic_index[slot] = (FRAG_ATTRIB_VAR0 - 
+                                                FRAG_ATTRIB_TEX0 +
+                                                attr - 
+                                                VERT_RESULT_VAR0);
+            break;
          }
-
-         assert(slot < Elements(output_flags));
-         output_flags[slot] = stvp->Base.Base.OutputFlags[attr];
       }
    }
+}
 
-   if (outputMapping) {
-      /* find max output slot referenced to compute vs_num_outputs */
-      GLuint maxSlot = 0;
-      for (attr = 0; attr < VERT_RESULT_MAX; attr++) {
-         if (outputMapping[attr] != ~0 && outputMapping[attr] > maxSlot)
-            maxSlot = outputMapping[attr];
-      }
-      vs_num_outputs = maxSlot + 1;
-   }
-   else {
-      outputMapping = defaultOutputMapping;
-   }
 
-#if 0 /* debug */
-   {
-      GLuint i;
-      printf("outputMapping? %d\n", outputMapping ? 1 : 0);
-      if (outputMapping) {
-         printf("attr -> slot\n");
-         for (i = 0; i < 16;  i++) {
-            printf(" %2d       %3d\n", i, outputMapping[i]);
-         }
-      }
-      printf("slot    sem_name  sem_index\n");
-      for (i = 0; i < vs_num_outputs; i++) {
-         printf(" %2d         %d         %d\n",
-                i,
-                vs_output_semantic_name[i],
-                vs_output_semantic_index[i]);
-      }
-   }
-#endif
-
-   /* free old shader state, if any */
-   if (stvp->state.tokens) {
-      st_free_tokens(stvp->state.tokens);
-      stvp->state.tokens = NULL;
-   }
-   if (stvp->driver_shader) {
-      cso_delete_vertex_shader(st->cso_context, stvp->driver_shader);
-      stvp->driver_shader = NULL;
-   }
+struct st_vp_varient *
+st_translate_vertex_program(struct st_context *st,
+                            struct st_vertex_program *stvp,
+                            const struct st_vp_varient_key *key)
+{
+   struct st_vp_varient *vpv = CALLOC_STRUCT(st_vp_varient);
+   struct pipe_context *pipe = st->pipe;
 
-   stvp->state.tokens = 
+   vpv->state.tokens = 
       st_translate_mesa_program(st->ctx,
                                 TGSI_PROCESSOR_VERTEX,
                                 &stvp->Base.Base,
                                 /* inputs */
-                                vs_num_inputs,
+                                stvp->num_inputs,
                                 stvp->input_to_index,
-                                vs_input_semantic_name,
-                                vs_input_semantic_index,
+                                NULL, /* input semantic name */
+                                NULL, /* input semantic index */
                                 NULL,
-                                input_flags,
                                 /* outputs */
-                                vs_num_outputs,
-                                outputMapping,
-                                vs_output_semantic_name,
-                                vs_output_semantic_index,
-                                output_flags );
+                                stvp->num_outputs,
+                                stvp->result_to_output,
+                                stvp->output_semantic_name,
+                                stvp->output_semantic_index );
 
-   stvp->num_inputs = vs_num_inputs;
-   stvp->driver_shader = pipe->create_vs_state(pipe, &stvp->state);
+   vpv->driver_shader = pipe->create_vs_state(pipe, &vpv->state);
 
    if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) {
       _mesa_print_program(&stvp->Base.Base);
@@ -348,9 +214,11 @@ st_translate_vertex_program(struct st_context *st,
    }
 
    if (ST_DEBUG & DEBUG_TGSI) {
-      tgsi_dump( stvp->state.tokens, 0 );
+      tgsi_dump( vpv->state.tokens, 0 );
       debug_printf("\n");
    }
+
+   return vpv;
 }
 
 
@@ -373,7 +241,6 @@ st_translate_fragment_program(struct st_context *st,
    GLuint attr;
    const GLbitfield inputsRead = stfp->Base.Base.InputsRead;
    GLuint vslot = 0;
-   GLuint num_generic = 0;
 
    uint fs_num_inputs = 0;
 
@@ -381,13 +248,6 @@ st_translate_fragment_program(struct st_context *st,
    ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
    uint fs_num_outputs = 0;
 
-   GLbitfield input_flags[MAX_PROGRAM_INPUTS];
-   GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
-
-//   memset(&fs, 0, sizeof(fs));
-   memset(input_flags, 0, sizeof(input_flags));
-   memset(output_flags, 0, sizeof(output_flags));
-
    /* which vertex output goes to the first fragment input: */
    if (inputsRead & FRAG_BIT_WPOS)
       vslot = 0;
@@ -430,14 +290,25 @@ st_translate_fragment_program(struct st_context *st,
             break;
          case FRAG_ATTRIB_FACE:
             stfp->input_semantic_name[slot] = TGSI_SEMANTIC_FACE;
-            stfp->input_semantic_index[slot] = num_generic++;
+            stfp->input_semantic_index[slot] = 0;
             interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
             break;
-         case FRAG_ATTRIB_PNTC:
-            stfp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stfp->input_semantic_index[slot] = num_generic++;
-            interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
-            break;
+
+            /* In most cases, there is nothing special about these
+             * inputs, so adopt a convention to use the generic
+             * semantic name and the mesa FRAG_ATTRIB_ number as the
+             * index. 
+             * 
+             * All that is required is that the vertex shader labels
+             * its own outputs similarly, and that the vertex shader
+             * generates at least every output required by the
+             * fragment shader plus fixed-function hardware (such as
+             * BFC).
+             * 
+             * There is no requirement that semantic indexes start at
+             * zero or be restricted to a particular range -- nobody
+             * should be building tables based on semantic index.
+             */
          case FRAG_ATTRIB_TEX0:
          case FRAG_ATTRIB_TEX1:
          case FRAG_ATTRIB_TEX2:
@@ -446,19 +317,18 @@ st_translate_fragment_program(struct st_context *st,
          case FRAG_ATTRIB_TEX5:
          case FRAG_ATTRIB_TEX6:
          case FRAG_ATTRIB_TEX7:
-            stfp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stfp->input_semantic_index[slot] = num_generic++;
-            interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
-            break;
+         case FRAG_ATTRIB_PNTC:
          case FRAG_ATTRIB_VAR0:
-            /* fall-through */
          default:
+            /* Actually, let's try and zero-base this just for
+             * readability of the generated TGSI.
+             */
+            assert(attr >= FRAG_ATTRIB_TEX0);
+            stfp->input_semantic_index[slot] = (attr - FRAG_ATTRIB_TEX0);
             stfp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stfp->input_semantic_index[slot] = num_generic++;
             interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
+            break;
          }
-
-         input_flags[slot] = stfp->Base.Base.InputFlags[attr];
       }
    }
 
@@ -467,7 +337,7 @@ st_translate_fragment_program(struct st_context *st,
     */
    {
       uint numColors = 0;
-      GLbitfield outputsWritten = stfp->Base.Base.OutputsWritten;
+      GLbitfield64 outputsWritten = stfp->Base.Base.OutputsWritten;
 
       /* if z is written, emit that first */
       if (outputsWritten & (1 << FRAG_RESULT_DEPTH)) {
@@ -496,8 +366,6 @@ st_translate_fragment_program(struct st_context *st,
                break;
             }
 
-            output_flags[fs_num_outputs] = stfp->Base.Base.OutputFlags[attr];
-
             fs_num_outputs++;
          }
       }
@@ -516,13 +384,11 @@ st_translate_fragment_program(struct st_context *st,
                                 stfp->input_semantic_name,
                                 stfp->input_semantic_index,
                                 interpMode,
-                                input_flags,
                                 /* outputs */
                                 fs_num_outputs,
                                 outputMapping,
                                 fs_output_semantic_name,
-                                fs_output_semantic_index,
-                                output_flags );
+                                fs_output_semantic_index );
 
    stfp->driver_shader = pipe->create_fs_state(pipe, &stfp->state);
 
@@ -538,6 +404,97 @@ st_translate_fragment_program(struct st_context *st,
 }
 
 void
+st_prepare_geometry_program(struct st_context *st,
+                            struct st_geometry_program *stgp)
+{
+   GLuint attr;
+
+   stgp->num_inputs = 0;
+   stgp->num_outputs = 0;
+
+/*
+ * Determine number of inputs, the mappings between VERT_ATTRIB_x
+ * and TGSI generic input indexes, plus input attrib semantic info.
+ */
+   for (attr = 0; attr < GEOM_ATTRIB_MAX; attr++) {
+      if (stgp->Base.Base.InputsRead & (1 << attr)) {
+         stgp->input_to_index[attr] = stgp->num_inputs;
+         stgp->index_to_input[stgp->num_inputs] = attr;
+         stgp->num_inputs++;
+      }
+   }
+
+   /* Compute mapping of vertex program outputs to slots.
+    */
+   for (attr = 0; attr < GEOM_RESULT_MAX; attr++) {
+      if ((stgp->Base.Base.OutputsWritten & (1 << attr)) == 0) {
+         stgp->result_to_output[attr] = ~0;
+      }
+      else {
+         unsigned slot = stgp->num_outputs++;
+
+         stgp->result_to_output[attr] = slot;
+
+         switch (attr) {
+         case GEOM_RESULT_HPOS:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+            stgp->output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_COL0:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stgp->output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_COL1:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stgp->output_semantic_index[slot] = 1;
+            break;
+         case GEOM_RESULT_BFC0:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stgp->output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_BFC1:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stgp->output_semantic_index[slot] = 1;
+            break;
+         case GEOM_RESULT_FOGC:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            stgp->output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_PSIZ:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
+            stgp->output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_EDGE:
+            assert(0);
+            break;
+
+         case GEOM_RESULT_TEX0:
+         case GEOM_RESULT_TEX1:
+         case GEOM_RESULT_TEX2:
+         case GEOM_RESULT_TEX3:
+         case GEOM_RESULT_TEX4:
+         case GEOM_RESULT_TEX5:
+         case GEOM_RESULT_TEX6:
+         case GEOM_RESULT_TEX7:
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stgp->output_semantic_index[slot] = attr - GEOM_RESULT_TEX0;
+            break;
+
+         case GEOM_RESULT_VAR0:
+         default:
+            assert(attr < GEOM_RESULT_MAX);
+            stgp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stgp->output_semantic_index[slot] = (FRAG_ATTRIB_VAR0 -
+                                                 FRAG_ATTRIB_TEX0 +
+                                                 attr -
+                                                 GEOM_RESULT_VAR0);
+            break;
+         }
+      }
+   }
+}
+
+void
 st_translate_geometry_program(struct st_context *st,
                               struct st_geometry_program *stgp,
                               const GLuint inputMapping[],
diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h
index 595733d60e..c75a8dd4e7 100644
--- a/src/mesa/state_tracker/st_program.h
+++ b/src/mesa/state_tracker/st_program.h
@@ -65,42 +65,70 @@ struct st_fragment_program
    struct pipe_shader_state state;
    void *driver_shader;
 
-   GLuint param_state;
-
-   /** List of vertex programs which have been translated such that their
-    * outputs match this fragment program's inputs.
-    */
-   struct translated_vertex_program *vertex_programs;
-   struct translated_geometry_program *geometry_programs;
-
    /** Program prefixed with glBitmap prologue */
    struct st_fragment_program *bitmap_program;
    uint bitmap_sampler;
 };
 
 
+
+struct st_vp_varient_key
+{
+   char dummy;                  /* currently unused */
+};
+
+
+/**
+ * This represents a vertex program, especially translated to match
+ * the inputs of a particular fragment shader.
+ */
+struct st_vp_varient
+{
+   /* Parameters which generated this translated version of a vertex
+    * shader:
+    */
+   struct st_vp_varient_key key;
+
+   /** TGSI tokens -- why?
+    */
+   struct pipe_shader_state state;
+
+   /** Driver's compiled shader */
+   void *driver_shader;
+
+   /** For using our private draw module (glRasterPos) */
+   struct draw_vertex_shader *draw_shader;
+
+   /** Next in linked list */
+   struct st_vp_varient *next;  
+};
+
+
+
+
 /**
  * Derived from Mesa gl_vertex_program:
  */
 struct st_vertex_program
 {
    struct gl_vertex_program Base;  /**< The Mesa vertex program */
-   GLuint serialNo;
+   GLuint serialNo, lastSerialNo;
 
    /** maps a Mesa VERT_ATTRIB_x to a packed TGSI input index */
    GLuint input_to_index[VERT_ATTRIB_MAX];
    /** maps a TGSI input index back to a Mesa VERT_ATTRIB_x */
    GLuint index_to_input[PIPE_MAX_SHADER_INPUTS];
-
    GLuint num_inputs;
 
-   struct pipe_shader_state state;
-   void *driver_shader;
-
-   /** For using our private draw module (glRasterPos) */
-   struct draw_vertex_shader *draw_shader;
+   /** Maps VERT_RESULT_x to slot */
+   GLuint result_to_output[VERT_RESULT_MAX];
+   ubyte output_semantic_name[VERT_RESULT_MAX];
+   ubyte output_semantic_index[VERT_RESULT_MAX];
+   GLuint num_outputs;
 
-   GLuint param_state;
+   /** List of translated varients of this vertex program.
+    */
+   struct st_vp_varient *varients;
 };
 
 
@@ -204,13 +232,21 @@ st_translate_geometry_program(struct st_context *st,
                               const ubyte *outputSemanticName,
                               const ubyte *outputSemanticIndex);
 
+/* Called after program string change, discard all previous
+ * compilation results.
+ */
 extern void
+st_prepare_vertex_program(struct st_context *st,
+                          struct st_vertex_program *stvp);
+
+extern struct st_vp_varient *
 st_translate_vertex_program(struct st_context *st,
-                            struct st_vertex_program *vp,
-                            const GLuint vert_output_to_slot[],
-                            const ubyte *fs_input_semantic_name,
-                            const ubyte *fs_input_semantic_index);
+                            struct st_vertex_program *stvp,
+                            const struct st_vp_varient_key *key);
 
+void
+st_vp_release_varients( struct st_context *st,
+                        struct st_vertex_program *stvp );
 
 extern void
 st_print_shaders(GLcontext *ctx);
diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c
index 3945822f66..dbccee86c1 100644
--- a/src/mesa/state_tracker/st_texture.c
+++ b/src/mesa/state_tracker/st_texture.c
@@ -44,6 +44,7 @@
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
 #include "util/u_rect.h"
+#include "util/u_math.h"
 
 
 #define DBG if(0) printf
@@ -100,9 +101,9 @@ st_texture_create(struct st_context *st,
    pt.target = target;
    pt.format = format;
    pt.last_level = last_level;
-   pt.width[0] = width0;
-   pt.height[0] = height0;
-   pt.depth[0] = depth0;
+   pt.width0 = width0;
+   pt.height0 = height0;
+   pt.depth0 = depth0;
    pf_get_block(format, &pt.block);
    pt.tex_usage = usage;
 
@@ -135,9 +136,9 @@ st_texture_match_image(const struct pipe_texture *pt,
    /* Test if this image's size matches what's expected in the
     * established texture.
     */
-   if (image->Width != pt->width[level] ||
-       image->Height != pt->height[level] ||
-       image->Depth != pt->depth[level])
+   if (image->Width != u_minify(pt->width0, level) ||
+       image->Height != u_minify(pt->height0, level) ||
+       image->Depth != u_minify(pt->depth0, level))
       return GL_FALSE;
 
    return GL_TRUE;
@@ -265,7 +266,7 @@ st_texture_image_data(struct st_context *st,
 {
    struct pipe_context *pipe = st->pipe;
    struct pipe_screen *screen = pipe->screen;
-   GLuint depth = dst->depth[level];
+   GLuint depth = u_minify(dst->depth0, level);
    GLuint i;
    const GLubyte *srcUB = src;
    struct pipe_transfer *dst_transfer;
@@ -275,15 +276,16 @@ st_texture_image_data(struct st_context *st,
    for (i = 0; i < depth; i++) {
       dst_transfer = st_no_flush_get_tex_transfer(st, dst, face, level, i,
 						  PIPE_TRANSFER_WRITE, 0, 0,
-						  dst->width[level],
-						  dst->height[level]);
+						  u_minify(dst->width0, level),
+                                                  u_minify(dst->height0, level));
 
       st_surface_data(pipe, dst_transfer,
 		      0, 0,                             /* dstx, dsty */
 		      srcUB,
 		      src_row_stride,
 		      0, 0,                             /* source x, y */
-		      dst->width[level], dst->height[level]);       /* width, height */
+		      u_minify(dst->width0, level),
+                      u_minify(dst->height0, level));      /* width, height */
 
       screen->tex_transfer_destroy(dst_transfer);
 
@@ -301,9 +303,9 @@ st_texture_image_copy(struct pipe_context *pipe,
                       GLuint face)
 {
    struct pipe_screen *screen = pipe->screen;
-   GLuint width = dst->width[dstLevel];
-   GLuint height = dst->height[dstLevel];
-   GLuint depth = dst->depth[dstLevel];
+   GLuint width = u_minify(dst->width0, dstLevel); 
+   GLuint height = u_minify(dst->height0, dstLevel); 
+   GLuint depth = u_minify(dst->depth0, dstLevel); 
    struct pipe_surface *src_surface;
    struct pipe_surface *dst_surface;
    GLuint i;
@@ -313,13 +315,13 @@ st_texture_image_copy(struct pipe_context *pipe,
 
       /* find src texture level of needed size */
       for (srcLevel = 0; srcLevel <= src->last_level; srcLevel++) {
-         if (src->width[srcLevel] == width &&
-             src->height[srcLevel] == height) {
+         if (u_minify(src->width0, srcLevel) == width &&
+             u_minify(src->height0, srcLevel) == height) {
             break;
          }
       }
-      assert(src->width[srcLevel] == width);
-      assert(src->height[srcLevel] == height);
+      assert(u_minify(src->width0, srcLevel) == width);
+      assert(u_minify(src->height0, srcLevel) == height);
 
 #if 0
       {
@@ -588,5 +590,5 @@ st_teximage_flush_before_map(struct st_context *st,
 
    if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) ||
 		      (usage & PIPE_TRANSFER_WRITE)))
-      st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL);
+      st->pipe->flush(st->pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
 }
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 77a77f0bcb..a22d34415d 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -190,7 +190,7 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
-   const GLbitfield outputsWritten = program->Base.OutputsWritten;
+   const GLbitfield64 outputsWritten = program->Base.OutputsWritten;
    struct gl_program_machine *machine = &swrast->FragProgMachine;
    GLuint i;
 
@@ -201,7 +201,7 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
          if (_mesa_execute_program(ctx, &program->Base, machine)) {
 
             /* Store result color */
-            if (outputsWritten & (1 << FRAG_RESULT_COLOR)) {
+	    if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
                COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
                        machine->Outputs[FRAG_RESULT_COLOR]);
             }
@@ -212,7 +212,7 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
                 */
                GLuint buf;
                for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
-                  if (outputsWritten & (1 << (FRAG_RESULT_DATA0 + buf))) {
+                  if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {
                      COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
                              machine->Outputs[FRAG_RESULT_DATA0 + buf]);
                   }
@@ -220,7 +220,7 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
             }
 
             /* Store result depth/z */
-            if (outputsWritten & (1 << FRAG_RESULT_DEPTH)) {
+            if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
                const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
                if (depth <= 0.0)
                   span->array->z[i] = 0;
@@ -256,12 +256,12 @@ _swrast_exec_fragment_program( GLcontext *ctx, SWspan *span )
 
    run_program(ctx, span, 0, span->end);
 
-   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_COLOR)) {
+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
       span->interpMask &= ~SPAN_RGBA;
       span->arrayMask |= SPAN_RGBA;
    }
 
-   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) {
+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
       span->interpMask &= ~SPAN_Z;
       span->arrayMask |= SPAN_Z;
    }
diff --git a/src/mesa/swrast_setup/ss_tritmp.h b/src/mesa/swrast_setup/ss_tritmp.h
index 724b5e94fa..17f3863956 100644
--- a/src/mesa/swrast_setup/ss_tritmp.h
+++ b/src/mesa/swrast_setup/ss_tritmp.h
@@ -67,8 +67,8 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	 if (facing == 1) {
 	    if (IND & SS_TWOSIDE_BIT) {
 	       if (IND & SS_RGBA_BIT) {
-                  if (VB->ColorPtr[1]) {
-                     GLfloat (*vbcolor)[4] = VB->ColorPtr[1]->data;
+                  if (VB->BackfaceColorPtr) {
+                     GLfloat (*vbcolor)[4] = VB->BackfaceColorPtr->data;
 
                      if (swsetup->intColors) {
                         COPY_CHAN4(saved_color[0], v[0]->color);
@@ -81,7 +81,7 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
                         COPY_4V(saved_col0[2], v[2]->attrib[FRAG_ATTRIB_COL0]);
                      }
 
-                     if (VB->ColorPtr[1]->stride) {
+                     if (VB->BackfaceColorPtr->stride) {
                         if (swsetup->intColors) {
                            SS_COLOR(v[0]->color, vbcolor[e0]);
                            SS_COLOR(v[1]->color, vbcolor[e1]);
@@ -108,14 +108,14 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
                      }
                   }
 
-		  if (VB->SecondaryColorPtr[1]) {
-		     GLfloat (*vbspec)[4] = VB->SecondaryColorPtr[1]->data;
+		  if (VB->BackfaceSecondaryColorPtr) {
+		     GLfloat (*vbspec)[4] = VB->BackfaceSecondaryColorPtr->data;
 
 		     COPY_4V(saved_spec[0], v[0]->attrib[FRAG_ATTRIB_COL1]);
 		     COPY_4V(saved_spec[1], v[1]->attrib[FRAG_ATTRIB_COL1]);
 		     COPY_4V(saved_spec[2], v[2]->attrib[FRAG_ATTRIB_COL1]);
 
-		     if (VB->SecondaryColorPtr[1]->stride) {
+		     if (VB->BackfaceSecondaryColorPtr->stride) {
 			SS_SPEC(v[0]->attrib[FRAG_ATTRIB_COL1], vbspec[e0]);
 			SS_SPEC(v[1]->attrib[FRAG_ATTRIB_COL1], vbspec[e1]);
 			SS_SPEC(v[2]->attrib[FRAG_ATTRIB_COL1], vbspec[e2]);
@@ -127,7 +127,7 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 		     }
 		  }
 	       } else {
-		  GLfloat *vbindex = (GLfloat *)VB->IndexPtr[1]->data;
+		  GLfloat *vbindex = (GLfloat *)VB->BackfaceIndexPtr->data;
 		  saved_index[0] = v[0]->attrib[FRAG_ATTRIB_CI][0];
 		  saved_index[1] = v[1]->attrib[FRAG_ATTRIB_CI][0];
 		  saved_index[2] = v[2]->attrib[FRAG_ATTRIB_CI][0];
@@ -200,7 +200,7 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
    if (IND & SS_TWOSIDE_BIT) {
       if (facing == 1) {
 	 if (IND & SS_RGBA_BIT) {
-            if (VB->ColorPtr[1]) {
+            if (VB->BackfaceColorPtr) {
                if (swsetup->intColors) {
                   COPY_CHAN4(v[0]->color, saved_color[0]);
                   COPY_CHAN4(v[1]->color, saved_color[1]);
@@ -213,7 +213,7 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
                }
             }
 
-	    if (VB->SecondaryColorPtr[1]) {
+	    if (VB->BackfaceSecondaryColorPtr) {
 	       COPY_4V(v[0]->attrib[FRAG_ATTRIB_COL1], saved_spec[0]);
 	       COPY_4V(v[1]->attrib[FRAG_ATTRIB_COL1], saved_spec[1]);
 	       COPY_4V(v[2]->attrib[FRAG_ATTRIB_COL1], saved_spec[2]);
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index f2771cde09..db21b4589d 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -171,7 +171,7 @@ _tnl_InvalidateState( GLcontext *ctx, GLuint new_state )
    if (vp) {
       GLuint i;
       for (i = 0; i < MAX_VARYING; i++) {
-         if (vp->Base.OutputsWritten & (1 << (VERT_RESULT_VAR0 + i))) {
+	 if (vp->Base.OutputsWritten & BITFIELD64_BIT(VERT_RESULT_VAR0 + i)) {
             RENDERINPUTS_SET(tnl->render_inputs_bitset,
                              _TNL_ATTRIB_GENERIC(i));
          }
diff --git a/src/mesa/tnl/t_context.h b/src/mesa/tnl/t_context.h
index 6137c2d2fe..ebaae6335b 100644
--- a/src/mesa/tnl/t_context.h
+++ b/src/mesa/tnl/t_context.h
@@ -198,26 +198,23 @@ struct vertex_buffer
     */
    GLuint Count;  /**< Number of vertices currently in buffer */
 
-   /* Pointers to current data.
-    * XXX some of these fields alias AttribPtr below and should be removed
-    * such as NormalPtr, TexCoordPtr, FogCoordPtr, etc.
+   /* Pointers to current data.  Most of the data is in AttribPtr -- all of
+    * it that is one of VERT_ATTRIB_X.  For things only produced by TNL,
+    * such as backface color or eye-space coordinates, they are stored
+    * here.
     */
    GLuint      *Elts;		                
-   GLvector4f  *ObjPtr;		                /* _TNL_BIT_POS */
    GLvector4f  *EyePtr;		                /* _TNL_BIT_POS */
    GLvector4f  *ClipPtr;	                /* _TNL_BIT_POS */
    GLvector4f  *NdcPtr;                         /* _TNL_BIT_POS */
    GLubyte     ClipOrMask;	                /* _TNL_BIT_POS */
    GLubyte     ClipAndMask;	                /* _TNL_BIT_POS */
    GLubyte     *ClipMask;		        /* _TNL_BIT_POS */
-   GLvector4f  *NormalPtr;	                /* _TNL_BIT_NORMAL */
    GLfloat     *NormalLengthPtr;	        /* _TNL_BIT_NORMAL */
    GLboolean   *EdgeFlag;	                /* _TNL_BIT_EDGEFLAG */
-   GLvector4f  *TexCoordPtr[MAX_TEXTURE_COORD_UNITS]; /* VERT_TEX_0..n */
-   GLvector4f  *IndexPtr[2];	                /* _TNL_BIT_INDEX */
-   GLvector4f  *ColorPtr[2];	                /* _TNL_BIT_COLOR0 */
-   GLvector4f  *SecondaryColorPtr[2];           /* _TNL_BIT_COLOR1 */
-   GLvector4f  *FogCoordPtr;	                /* _TNL_BIT_FOG */
+   GLvector4f  *BackfaceIndexPtr;
+   GLvector4f  *BackfaceColorPtr;
+   GLvector4f  *BackfaceSecondaryColorPtr;
 
    const struct _mesa_prim  *Primitive;	              
    GLuint      PrimitiveCount;	      
@@ -402,11 +399,6 @@ struct tnl_device_driver
    /* Alert tnl-aware drivers of changes to material.
     */
 
-   void (*NotifyInputChanges)(GLcontext *ctx, GLuint bitmask);
-   /* Alert tnl-aware drivers of changes to size and stride of input
-    * arrays.
-    */
-
    /***
     *** Rendering -- These functions called only from t_vb_render.c
     ***/
diff --git a/src/mesa/tnl/t_draw.c b/src/mesa/tnl/t_draw.c
index 04fa106300..1c7c733883 100644
--- a/src/mesa/tnl/t_draw.c
+++ b/src/mesa/tnl/t_draw.c
@@ -251,22 +251,10 @@ static void bind_inputs( GLcontext *ctx,
     */
    VB->Count = count;
 
-
-   /* Legacy pointers -- remove one day.
-    */
-   VB->ObjPtr = VB->AttribPtr[_TNL_ATTRIB_POS];
-   VB->NormalPtr = VB->AttribPtr[_TNL_ATTRIB_NORMAL];
-   VB->ColorPtr[0] = VB->AttribPtr[_TNL_ATTRIB_COLOR0];
-   VB->ColorPtr[1] = NULL;
-   VB->IndexPtr[0] = VB->AttribPtr[_TNL_ATTRIB_COLOR_INDEX];
-   VB->IndexPtr[1] = NULL;
-   VB->SecondaryColorPtr[0] = VB->AttribPtr[_TNL_ATTRIB_COLOR1];
-   VB->SecondaryColorPtr[1] = NULL;
-   VB->FogCoordPtr = VB->AttribPtr[_TNL_ATTRIB_FOG];
-
-   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
-      VB->TexCoordPtr[i] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i];
-   }
+   /* These should perhaps be part of _TNL_ATTRIB_* */
+   VB->BackfaceColorPtr = NULL;
+   VB->BackfaceIndexPtr = NULL;
+   VB->BackfaceSecondaryColorPtr = NULL;
 
    /* Clipping and drawing code still requires this to be a packed
     * array of ubytes which can be written into.  TODO: Fix and
diff --git a/src/mesa/tnl/t_pipeline.c b/src/mesa/tnl/t_pipeline.c
index 357ef1e24b..01b30babb4 100644
--- a/src/mesa/tnl/t_pipeline.c
+++ b/src/mesa/tnl/t_pipeline.c
@@ -86,10 +86,6 @@ static GLuint check_input_changes( GLcontext *ctx )
       }
    }
 
-   if (tnl->pipeline.input_changes &&
-      tnl->Driver.NotifyInputChanges) 
-      tnl->Driver.NotifyInputChanges( ctx, tnl->pipeline.input_changes );
-
    return tnl->pipeline.input_changes;
 }
 
diff --git a/src/mesa/tnl/t_vb_fog.c b/src/mesa/tnl/t_vb_fog.c
index f3a7bd49f4..4a0e6ad4f9 100644
--- a/src/mesa/tnl/t_vb_fog.c
+++ b/src/mesa/tnl/t_vb_fog.c
@@ -156,7 +156,7 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
       GLuint i;
       GLfloat *coord;
       /* Fog is computed from vertex or fragment Z values */
-      /* source = VB->ObjPtr or VB->EyePtr coords */
+      /* source = VB->AttribPtr[_TNL_ATTRIB_POS] or VB->EyePtr coords */
       /* dest = VB->AttribPtr[_TNL_ATTRIB_FOG] = fog stage private storage */
       VB->AttribPtr[_TNL_ATTRIB_FOG] = &store->fogcoord;
 
@@ -176,11 +176,12 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
 	 /* Full eye coords weren't required, just calculate the
 	  * eye Z values.
 	  */
-	 _mesa_dotprod_tab[VB->ObjPtr->size]( (GLfloat *) input->data,
-					      4 * sizeof(GLfloat),
-					      VB->ObjPtr, plane );
+	 _mesa_dotprod_tab[VB->AttribPtr[_TNL_ATTRIB_POS]->size]
+	    ( (GLfloat *) input->data,
+	      4 * sizeof(GLfloat),
+	      VB->AttribPtr[_TNL_ATTRIB_POS], plane );
 
-	 input->count = VB->ObjPtr->count;
+	 input->count = VB->AttribPtr[_TNL_ATTRIB_POS]->count;
 
 	 /* make sure coords are really positive
 	    NOTE should avoid going through array twice */
@@ -213,7 +214,7 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
       /* input->count may be one if glFogCoord was only called once
        * before glBegin.  But we need to compute fog for all vertices.
        */
-      input->count = VB->ObjPtr->count;
+      input->count = VB->AttribPtr[_TNL_ATTRIB_POS]->count;
 
       VB->AttribPtr[_TNL_ATTRIB_FOG] = &store->fogcoord;  /* dest data */
    }
@@ -227,7 +228,6 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
       VB->AttribPtr[_TNL_ATTRIB_FOG] = input;
    }
 
-   VB->FogCoordPtr = VB->AttribPtr[_TNL_ATTRIB_FOG];
    return GL_TRUE;
 }
 
diff --git a/src/mesa/tnl/t_vb_light.c b/src/mesa/tnl/t_vb_light.c
index f47f99397c..8a0fe63fd8 100644
--- a/src/mesa/tnl/t_vb_light.c
+++ b/src/mesa/tnl/t_vb_light.c
@@ -127,7 +127,7 @@ prepare_materials(GLcontext *ctx,
       const GLuint bitmask = ctx->Light.ColorMaterialBitmask;
       for (i = 0 ; i < MAT_ATTRIB_MAX ; i++)
 	 if (bitmask & (1<<i))
-	    VB->AttribPtr[_TNL_ATTRIB_MAT_FRONT_AMBIENT + i] = VB->ColorPtr[0];
+	    VB->AttribPtr[_TNL_ATTRIB_MAT_FRONT_AMBIENT + i] = VB->AttribPtr[_TNL_ATTRIB_COLOR0];
    }
 
    /* Now, for each material attribute that's tracking vertex color, save
@@ -200,7 +200,7 @@ static GLboolean run_lighting( GLcontext *ctx,
    struct light_stage_data *store = LIGHT_STAGE_DATA(stage);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
-   GLvector4f *input = ctx->_NeedEyeCoords ? VB->EyePtr : VB->ObjPtr;
+   GLvector4f *input = ctx->_NeedEyeCoords ? VB->EyePtr : VB->AttribPtr[_TNL_ATTRIB_POS];
    GLuint idx;
 
    if (!ctx->Light.Enabled || ctx->VertexProgram._Current)
@@ -208,13 +208,13 @@ static GLboolean run_lighting( GLcontext *ctx,
 
    /* Make sure we can talk about position x,y and z:
     */
-   if (input->size <= 2 && input == VB->ObjPtr) {
+   if (input->size <= 2 && input == VB->AttribPtr[_TNL_ATTRIB_POS]) {
 
       _math_trans_4f( store->Input.data,
-		      VB->ObjPtr->data,
-		      VB->ObjPtr->stride,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
 		      GL_FLOAT,
-		      VB->ObjPtr->size,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->size,
 		      0,
 		      VB->Count );
 
@@ -246,10 +246,6 @@ static GLboolean run_lighting( GLcontext *ctx,
     */
    store->light_func_tab[idx]( ctx, VB, stage, input );
 
-   VB->AttribPtr[_TNL_ATTRIB_COLOR0] = VB->ColorPtr[0];
-   VB->AttribPtr[_TNL_ATTRIB_COLOR1] = VB->SecondaryColorPtr[0];
-   VB->AttribPtr[_TNL_ATTRIB_COLOR_INDEX] = VB->IndexPtr[0];
-
    return GL_TRUE;
 }
 
diff --git a/src/mesa/tnl/t_vb_lighttmp.h b/src/mesa/tnl/t_vb_lighttmp.h
index 124ca3c74f..4ebef2356f 100644
--- a/src/mesa/tnl/t_vb_lighttmp.h
+++ b/src/mesa/tnl/t_vb_lighttmp.h
@@ -72,13 +72,13 @@ static void TAG(light_rgba_spec)( GLcontext *ctx,
    fprintf(stderr, "%s\n", __FUNCTION__ );
 #endif
 
-   VB->ColorPtr[0] = &store->LitColor[0];
-   VB->SecondaryColorPtr[0] = &store->LitSecondary[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR0] = &store->LitColor[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR1] = &store->LitSecondary[0];
    sumA[0] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE][3];
 
 #if IDX & LIGHT_TWOSIDE
-   VB->ColorPtr[1] = &store->LitColor[1];
-   VB->SecondaryColorPtr[1] = &store->LitSecondary[1];
+   VB->BackfaceColorPtr = &store->LitColor[1];
+   VB->BackfaceSecondaryColorPtr = &store->LitSecondary[1];
    sumA[1] = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_DIFFUSE][3];
 #endif
 
@@ -259,11 +259,11 @@ static void TAG(light_rgba)( GLcontext *ctx,
    fprintf(stderr, "%s\n", __FUNCTION__ );
 #endif
 
-   VB->ColorPtr[0] = &store->LitColor[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR0] = &store->LitColor[0];
    sumA[0] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE][3];
 
 #if IDX & LIGHT_TWOSIDE
-   VB->ColorPtr[1] = &store->LitColor[1];
+   VB->BackfaceColorPtr = &store->LitColor[1];
    sumA[1] = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_DIFFUSE][3];
 #endif
 
@@ -449,9 +449,9 @@ static void TAG(light_fast_rgba_single)( GLcontext *ctx,
 
    (void) input;		/* doesn't refer to Eye or Obj */
 
-   VB->ColorPtr[0] = &store->LitColor[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR0] = &store->LitColor[0];
 #if IDX & LIGHT_TWOSIDE
-   VB->ColorPtr[1] = &store->LitColor[1];
+   VB->BackfaceColorPtr = &store->LitColor[1];
 #endif
 
    if (nr > 1) {
@@ -559,9 +559,9 @@ static void TAG(light_fast_rgba)( GLcontext *ctx,
    sumA[0] = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE][3];
    sumA[1] = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_DIFFUSE][3];
 
-   VB->ColorPtr[0] = &store->LitColor[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR0] = &store->LitColor[0];
 #if IDX & LIGHT_TWOSIDE
-   VB->ColorPtr[1] = &store->LitColor[1];
+   VB->BackfaceColorPtr = &store->LitColor[1];
 #endif
 
    if (nr > 1) {
@@ -665,14 +665,14 @@ static void TAG(light_ci)( GLcontext *ctx,
    fprintf(stderr, "%s\n", __FUNCTION__ );
 #endif
 
-   VB->IndexPtr[0] = &store->LitIndex[0];
+   VB->AttribPtr[_TNL_ATTRIB_COLOR_INDEX] = &store->LitIndex[0];
 #if IDX & LIGHT_TWOSIDE
-   VB->IndexPtr[1] = &store->LitIndex[1];
+   VB->BackfaceIndexPtr = &store->LitIndex[1];
 #endif
 
-   indexResult[0] = (GLfloat *)VB->IndexPtr[0]->data;
+   indexResult[0] = (GLfloat *)VB->AttribPtr[_TNL_ATTRIB_COLOR_INDEX]->data;
 #if IDX & LIGHT_TWOSIDE
-   indexResult[1] = (GLfloat *)VB->IndexPtr[1]->data;
+   indexResult[1] = (GLfloat *)VB->BackfaceIndexPtr->data;
 #endif
 
    /* loop over vertices */
diff --git a/src/mesa/tnl/t_vb_normals.c b/src/mesa/tnl/t_vb_normals.c
index a4821cc1cc..693d3dc118 100644
--- a/src/mesa/tnl/t_vb_normals.c
+++ b/src/mesa/tnl/t_vb_normals.c
@@ -79,7 +79,6 @@ run_normal_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
    }
 
    VB->AttribPtr[_TNL_ATTRIB_NORMAL] = &store->normal;
-   VB->NormalPtr = &store->normal;
 
    VB->NormalLengthPtr = NULL;	/* no longer valid */
    return GL_TRUE;
diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c
index c10a27614f..c289cdfbaa 100644
--- a/src/mesa/tnl/t_vb_program.c
+++ b/src/mesa/tnl/t_vb_program.c
@@ -329,7 +329,7 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
    /* make list of outputs to save some time below */
    numOutputs = 0;
    for (i = 0; i < VERT_RESULT_MAX; i++) {
-      if (program->Base.OutputsWritten & (1 << i)) {
+      if (program->Base.OutputsWritten & BITFIELD64_BIT(i)) {
          outputs[numOutputs++] = i;
       }
    }
@@ -407,14 +407,14 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
    /* Fixup fog and point size results if needed */
    if (program->IsNVProgram) {
       if (ctx->Fog.Enabled &&
-          (program->Base.OutputsWritten & (1 << VERT_RESULT_FOGC)) == 0) {
+          (program->Base.OutputsWritten & BITFIELD64_BIT(VERT_RESULT_FOGC)) == 0) {
          for (i = 0; i < VB->Count; i++) {
             store->results[VERT_RESULT_FOGC].data[i][0] = 1.0;
          }
       }
 
       if (ctx->VertexProgram.PointSizeEnabled &&
-          (program->Base.OutputsWritten & (1 << VERT_RESULT_PSIZ)) == 0) {
+          (program->Base.OutputsWritten & BITFIELD64_BIT(VERT_RESULT_PSIZ)) == 0) {
          for (i = 0; i < VB->Count; i++) {
             store->results[VERT_RESULT_PSIZ].data[i][0] = ctx->Point.Size;
          }
@@ -454,25 +454,20 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
       VB->ClipPtr->count = VB->Count;
    }
 
-   VB->ColorPtr[0] = &store->results[VERT_RESULT_COL0];
-   VB->ColorPtr[1] = &store->results[VERT_RESULT_BFC0];
-   VB->SecondaryColorPtr[0] = &store->results[VERT_RESULT_COL1];
-   VB->SecondaryColorPtr[1] = &store->results[VERT_RESULT_BFC1];
-   VB->FogCoordPtr = &store->results[VERT_RESULT_FOGC];
-
    VB->AttribPtr[VERT_ATTRIB_COLOR0] = &store->results[VERT_RESULT_COL0];
    VB->AttribPtr[VERT_ATTRIB_COLOR1] = &store->results[VERT_RESULT_COL1];
    VB->AttribPtr[VERT_ATTRIB_FOG] = &store->results[VERT_RESULT_FOGC];
    VB->AttribPtr[_TNL_ATTRIB_POINTSIZE] = &store->results[VERT_RESULT_PSIZ];
+   VB->BackfaceColorPtr = &store->results[VERT_RESULT_BFC0];
+   VB->BackfaceSecondaryColorPtr = &store->results[VERT_RESULT_BFC1];
 
    for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
-      VB->TexCoordPtr[i] = 
       VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]
          = &store->results[VERT_RESULT_TEX0 + i];
    }
 
    for (i = 0; i < ctx->Const.MaxVarying; i++) {
-      if (program->Base.OutputsWritten & (1 << (VERT_RESULT_VAR0 + i))) {
+      if (program->Base.OutputsWritten & BITFIELD64_BIT(VERT_RESULT_VAR0 + i)) {
          /* Note: varying results get put into the generic attributes */
 	 VB->AttribPtr[VERT_ATTRIB_GENERIC0+i]
             = &store->results[VERT_RESULT_VAR0 + i];
diff --git a/src/mesa/tnl/t_vb_texgen.c b/src/mesa/tnl/t_vb_texgen.c
index 7c1819b223..9ef13bc96d 100644
--- a/src/mesa/tnl/t_vb_texgen.c
+++ b/src/mesa/tnl/t_vb_texgen.c
@@ -341,7 +341,7 @@ static void texgen( GLcontext *ctx,
    GLvector4f *in = VB->AttribPtr[VERT_ATTRIB_TEX0 + unit];
    GLvector4f *out = &store->texcoord[unit];
    const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   const GLvector4f *obj = VB->ObjPtr;
+   const GLvector4f *obj = VB->AttribPtr[_TNL_ATTRIB_POS];
    const GLvector4f *eye = VB->EyePtr;
    const GLvector4f *normal = VB->AttribPtr[_TNL_ATTRIB_NORMAL];
    const GLfloat *m = store->tmp_m;
@@ -498,7 +498,6 @@ static GLboolean run_texgen_stage( GLcontext *ctx,
 
 	 store->TexgenFunc[i]( ctx, store, i );
 
-         VB->TexCoordPtr[i] =
          VB->AttribPtr[VERT_ATTRIB_TEX0 + i] = &store->texcoord[i];
       }
    }
diff --git a/src/mesa/tnl/t_vb_texmat.c b/src/mesa/tnl/t_vb_texmat.c
index 0abe8cc35d..83688290e5 100644
--- a/src/mesa/tnl/t_vb_texmat.c
+++ b/src/mesa/tnl/t_vb_texmat.c
@@ -73,7 +73,6 @@ static GLboolean run_texmat_stage( GLcontext *ctx,
 			      ctx->TextureMatrixStack[i].Top,
 			      VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]);
 
-         VB->TexCoordPtr[i] = 
 	 VB->AttribPtr[VERT_ATTRIB_TEX0+i] = &store->texcoord[i];
       }
    }
diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c
index 4734754ea4..bc7e0951ec 100644
--- a/src/mesa/tnl/t_vb_vertex.c
+++ b/src/mesa/tnl/t_vb_vertex.c
@@ -152,16 +152,16 @@ static GLboolean run_vertex_stage( GLcontext *ctx,
        * Use combined ModelProject to avoid some depth artifacts
        */
       if (ctx->ModelviewMatrixStack.Top->type == MATRIX_IDENTITY)
-	 VB->EyePtr = VB->ObjPtr;
+	 VB->EyePtr = VB->AttribPtr[_TNL_ATTRIB_POS];
       else
 	 VB->EyePtr = TransformRaw( &store->eye,
 				    ctx->ModelviewMatrixStack.Top,
-				    VB->ObjPtr);
+				    VB->AttribPtr[_TNL_ATTRIB_POS]);
    }
 
    VB->ClipPtr = TransformRaw( &store->clip,
 			       &ctx->_ModelProjectMatrix,
-			       VB->ObjPtr );
+			       VB->AttribPtr[_TNL_ATTRIB_POS] );
 
    /* Drivers expect this to be clean to element 4...
     */
diff --git a/src/mesa/tnl/t_vertex_generic.c b/src/mesa/tnl/t_vertex_generic.c
index 9812f8c808..fa34d11d7b 100644
--- a/src/mesa/tnl/t_vertex_generic.c
+++ b/src/mesa/tnl/t_vertex_generic.c
@@ -1092,33 +1092,33 @@ void _tnl_generic_interp_extras( GLcontext *ctx,
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   /* If stride is zero, ColorPtr[1] is constant across the VB, so
+   /* If stride is zero, BackfaceColorPtr is constant across the VB, so
     * there is no point interpolating between two values as they will
     * be identical.  In all other cases, this value is generated by
     * t_vb_lighttmp.h and has a stride of 4 dwords.
     */
-   if (VB->ColorPtr[1] && VB->ColorPtr[1]->stride) {
-      assert(VB->ColorPtr[1]->stride == 4 * sizeof(GLfloat));
+   if (VB->BackfaceColorPtr && VB->BackfaceColorPtr->stride) {
+      assert(VB->BackfaceColorPtr->stride == 4 * sizeof(GLfloat));
 
       INTERP_4F( t,
-		 VB->ColorPtr[1]->data[dst],
-		 VB->ColorPtr[1]->data[out],
-		 VB->ColorPtr[1]->data[in] );
+		 VB->BackfaceColorPtr->data[dst],
+		 VB->BackfaceColorPtr->data[out],
+		 VB->BackfaceColorPtr->data[in] );
    }
 
-   if (VB->SecondaryColorPtr[1]) {
-      assert(VB->SecondaryColorPtr[1]->stride == 4 * sizeof(GLfloat));
+   if (VB->BackfaceSecondaryColorPtr) {
+      assert(VB->BackfaceSecondaryColorPtr->stride == 4 * sizeof(GLfloat));
       
       INTERP_3F( t,
-		 VB->SecondaryColorPtr[1]->data[dst],
-		 VB->SecondaryColorPtr[1]->data[out],
-		 VB->SecondaryColorPtr[1]->data[in] );
+		 VB->BackfaceSecondaryColorPtr->data[dst],
+		 VB->BackfaceSecondaryColorPtr->data[out],
+		 VB->BackfaceSecondaryColorPtr->data[in] );
    }
    
-   if (VB->IndexPtr[1]) {
-      VB->IndexPtr[1]->data[dst][0] = LINTERP( t,
-					       VB->IndexPtr[1]->data[out][0],
-					       VB->IndexPtr[1]->data[in][0] );
+   if (VB->BackfaceIndexPtr) {
+      VB->BackfaceIndexPtr->data[dst][0] = LINTERP( t,
+					       VB->BackfaceIndexPtr->data[out][0],
+					       VB->BackfaceIndexPtr->data[in][0] );
    }
 
    if (VB->EdgeFlag) {
@@ -1135,18 +1135,18 @@ void _tnl_generic_copy_pv_extras( GLcontext *ctx,
 
    /* See above comment:
     */
-   if (VB->ColorPtr[1] && VB->ColorPtr[1]->stride) {
-      COPY_4FV( VB->ColorPtr[1]->data[dst], 
-		VB->ColorPtr[1]->data[src] );
+   if (VB->BackfaceColorPtr && VB->BackfaceColorPtr->stride) {
+      COPY_4FV( VB->BackfaceColorPtr->data[dst],
+		VB->BackfaceColorPtr->data[src] );
    }
 
-   if (VB->SecondaryColorPtr[1]) {
-      COPY_4FV( VB->SecondaryColorPtr[1]->data[dst], 
-		VB->SecondaryColorPtr[1]->data[src] );
+   if (VB->BackfaceSecondaryColorPtr) {
+      COPY_4FV( VB->BackfaceSecondaryColorPtr->data[dst],
+		VB->BackfaceSecondaryColorPtr->data[src] );
    }
 
-   if (VB->IndexPtr[1]) {
-      VB->IndexPtr[1]->data[dst][0] = VB->IndexPtr[1]->data[src][0];
+   if (VB->BackfaceIndexPtr) {
+      VB->BackfaceIndexPtr->data[dst][0] = VB->BackfaceIndexPtr->data[src][0];
    }
 
    _tnl_generic_copy_pv(ctx, dst, src);
diff --git a/src/mesa/tnl_dd/t_dd_dmatmp.h b/src/mesa/tnl_dd/t_dd_dmatmp.h
index e4b535fb68..e5885782c7 100644
--- a/src/mesa/tnl_dd/t_dd_dmatmp.h
+++ b/src/mesa/tnl_dd/t_dd_dmatmp.h
@@ -443,7 +443,7 @@ static void TAG(render_quad_strip_verts)( GLcontext *ctx,
 
    } else if (HAVE_TRI_STRIPS && 
 	      ctx->Light.ShadeModel == GL_FLAT &&
-	      TNL_CONTEXT(ctx)->vb.ColorPtr[0]->stride) {
+	      TNL_CONTEXT(ctx)->vb.AttribPtr[_TNL_ATTRIB_COLOR0]->stride) {
       if (HAVE_ELTS) {
 	 LOCAL_VARS;
 	 int dmasz = GET_SUBSEQUENT_VB_MAX_ELTS();
@@ -1221,7 +1221,7 @@ static GLboolean TAG(validate_render)( GLcontext *ctx,
 	    ok = GL_TRUE;
 	 } else if (HAVE_TRI_STRIPS && 
 		    ctx->Light.ShadeModel == GL_FLAT &&
-		    VB->ColorPtr[0]->stride != 0) {
+		    VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride != 0) {
 	    if (HAVE_ELTS) {
 	       ok = (GLint) count < GET_SUBSEQUENT_VB_MAX_ELTS();
 	    }
diff --git a/src/mesa/tnl_dd/t_dd_tritmp.h b/src/mesa/tnl_dd/t_dd_tritmp.h
index 1ae70f4059..8574fe618b 100644
--- a/src/mesa/tnl_dd/t_dd_tritmp.h
+++ b/src/mesa/tnl_dd/t_dd_tritmp.h
@@ -195,7 +195,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 		  }
 	       }
 	       else {
-		  GLfloat (*vbcolor)[4] = VB->ColorPtr[1]->data;
+		  GLfloat (*vbcolor)[4] = VB->BackfaceColorPtr->data;
 		  (void) vbcolor;
 
 		  if (!DO_FLAT) {
@@ -204,8 +204,8 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 		  }
 		  VERT_SAVE_RGBA( 2 );
 
-		  if (VB->ColorPtr[1]->stride) {
-		     ASSERT(VB->ColorPtr[1]->stride == 4*sizeof(GLfloat));
+		  if (VB->BackfaceColorPtr->stride) {
+		     ASSERT(VB->BackfaceColorPtr->stride == 4*sizeof(GLfloat));
 
 		     if (!DO_FLAT) {		  
 			VERT_SET_RGBA( v[0], vbcolor[e0] );
@@ -221,9 +221,9 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 		     VERT_SET_RGBA( v[2], vbcolor[0] );
 		  }
 
-		  if (HAVE_SPEC && VB->SecondaryColorPtr[1]) {
-		     GLfloat (*vbspec)[4] = VB->SecondaryColorPtr[1]->data;
-		     ASSERT(VB->SecondaryColorPtr[1]->stride == 4*sizeof(GLfloat));
+		  if (HAVE_SPEC && VB->BackfaceSecondaryColorPtr) {
+		     GLfloat (*vbspec)[4] = VB->BackfaceSecondaryColorPtr->data;
+		     ASSERT(VB->BackfaceSecondaryColorPtr->stride == 4*sizeof(GLfloat));
 
 		     if (!DO_FLAT) {
 			VERT_SAVE_SPEC( 0 );
@@ -237,7 +237,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	       }
 	    }
 	    else {
-	       GLfloat (*vbindex) = (GLfloat *)VB->IndexPtr[1]->data;
+	       GLfloat (*vbindex) = (GLfloat *)VB->BackfaceIndexPtr->data;
 	       if (!DO_FLAT) {
 		  VERT_SAVE_IND( 0 );
 		  VERT_SAVE_IND( 1 );
@@ -279,7 +279,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	 VERT_SAVE_RGBA( 1 );
 	 VERT_COPY_RGBA( v[0], v[2] );
 	 VERT_COPY_RGBA( v[1], v[2] );
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_SAVE_SPEC( 0 );
 	    VERT_SAVE_SPEC( 1 );
 	    VERT_COPY_SPEC( v[0], v[2] );
@@ -374,7 +374,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
       if (HAVE_RGBA) {
 	 VERT_RESTORE_RGBA( 0 );
 	 VERT_RESTORE_RGBA( 1 );
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_RESTORE_SPEC( 0 );
 	    VERT_RESTORE_SPEC( 1 );
 	 }
@@ -436,7 +436,7 @@ static void TAG(quadr)( GLcontext *ctx,
 	 if (DO_TWOSIDE && facing == 1)
 	 {
 	    if (HAVE_RGBA) {
-	       GLfloat (*vbcolor)[4] = VB->ColorPtr[1]->data;
+	       GLfloat (*vbcolor)[4] = VB->BackfaceColorPtr->data;
 	       (void)vbcolor;
 
 	       if (HAVE_BACK_COLORS) {
@@ -471,7 +471,7 @@ static void TAG(quadr)( GLcontext *ctx,
 		  }
 	          VERT_SAVE_RGBA( 3 );
 
-		  if (VB->ColorPtr[1]->stride) {
+		  if (VB->BackfaceColorPtr->stride) {
 		     if (!DO_FLAT) {
 			VERT_SET_RGBA( v[0], vbcolor[e0] );
 			VERT_SET_RGBA( v[1], vbcolor[e1] );
@@ -488,9 +488,9 @@ static void TAG(quadr)( GLcontext *ctx,
 		     VERT_SET_RGBA( v[3], vbcolor[0] );
 		  }
 
-	          if (HAVE_SPEC && VB->SecondaryColorPtr[1]) {
-		     GLfloat (*vbspec)[4] = VB->SecondaryColorPtr[1]->data;
-		     ASSERT(VB->SecondaryColorPtr[1]->stride==4*sizeof(GLfloat));
+	          if (HAVE_SPEC && VB->BackfaceSecondaryColorPtr) {
+		     GLfloat (*vbspec)[4] = VB->BackfaceSecondaryColorPtr->data;
+		     ASSERT(VB->BackfaceSecondaryColorPtr->stride==4*sizeof(GLfloat));
 
 		     if (!DO_FLAT) {
 		        VERT_SAVE_SPEC( 0 );
@@ -506,7 +506,7 @@ static void TAG(quadr)( GLcontext *ctx,
 	       }
 	    }
 	    else {
-	       GLfloat *vbindex = (GLfloat *)VB->IndexPtr[1]->data;
+	       GLfloat *vbindex = (GLfloat *)VB->BackfaceIndexPtr->data;
 	       if (!DO_FLAT) {
 		  VERT_SAVE_IND( 0 );
 		  VERT_SAVE_IND( 1 );
@@ -553,7 +553,7 @@ static void TAG(quadr)( GLcontext *ctx,
 	 VERT_COPY_RGBA( v[0], v[3] );
 	 VERT_COPY_RGBA( v[1], v[3] );
 	 VERT_COPY_RGBA( v[2], v[3] );
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_SAVE_SPEC( 0 );
 	    VERT_SAVE_SPEC( 1 );
 	    VERT_SAVE_SPEC( 2 );
@@ -659,7 +659,7 @@ static void TAG(quadr)( GLcontext *ctx,
 	 VERT_RESTORE_RGBA( 0 );
 	 VERT_RESTORE_RGBA( 1 );
 	 VERT_RESTORE_RGBA( 2 );
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_RESTORE_SPEC( 0 );
 	    VERT_RESTORE_SPEC( 1 );
 	    VERT_RESTORE_SPEC( 2 );
@@ -708,7 +708,7 @@ static void TAG(line)( GLcontext *ctx, GLuint e0, GLuint e1 )
       if (HAVE_RGBA) {
 	 VERT_SAVE_RGBA( 0 );
 	 VERT_COPY_RGBA( v[0], v[1] );
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_SAVE_SPEC( 0 );
 	    VERT_COPY_SPEC( v[0], v[1] );
 	 }
@@ -725,7 +725,7 @@ static void TAG(line)( GLcontext *ctx, GLuint e0, GLuint e1 )
       if (HAVE_RGBA) {
 	 VERT_RESTORE_RGBA( 0 );
 
-	 if (HAVE_SPEC && VB->SecondaryColorPtr[0]) {
+	 if (HAVE_SPEC && VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
 	    VERT_RESTORE_SPEC( 0 );
 	 }
       }
diff --git a/src/mesa/tnl_dd/t_dd_vb.c b/src/mesa/tnl_dd/t_dd_vb.c
index b3937c29a0..a8a0a69768 100644
--- a/src/mesa/tnl_dd/t_dd_vb.c
+++ b/src/mesa/tnl_dd/t_dd_vb.c
@@ -297,19 +297,19 @@ INTERP_QUALIFIER void TAG(interp_extras)( GLcontext *ctx,
    LOCALVARS
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      assert(VB->ColorPtr[1]->stride == 4 * sizeof(GLfloat));
+   if (VB->BackfaceColorPtr) {
+      assert(VB->BackfaceColorPtr->stride == 4 * sizeof(GLfloat));
 
       INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		    GET_COLOR(VB->BackfaceColorPtr, dst),
+		    GET_COLOR(VB->BackfaceColorPtr, out),
+		    GET_COLOR(VB->BackfaceColorPtr, in) );
 
-      if (VB->SecondaryColorPtr[1]) {
+      if (VB->BackfaceSecondaryColorPtr) {
 	 INTERP_3F( t,
-		       GET_COLOR(VB->SecondaryColorPtr[1], dst),
-		       GET_COLOR(VB->SecondaryColorPtr[1], out),
-		       GET_COLOR(VB->SecondaryColorPtr[1], in) );
+		       GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		       GET_COLOR(VB->BackfaceSecondaryColorPtr, out),
+		       GET_COLOR(VB->BackfaceSecondaryColorPtr, in) );
       }
    }
 
@@ -326,13 +326,13 @@ INTERP_QUALIFIER void TAG(copy_pv_extras)( GLcontext *ctx,
    LOCALVARS
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst), 
-		GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+      COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		GET_COLOR(VB->BackfaceColorPtr, src) );
 
-      if (VB->SecondaryColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->SecondaryColorPtr[1], dst), 
-		   GET_COLOR(VB->SecondaryColorPtr[1], src) );
+      if (VB->BackfaceSecondaryColorPtr) {
+	 COPY_4FV( GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		   GET_COLOR(VB->BackfaceSecondaryColorPtr, src) );
       }
    }
 
diff --git a/src/mesa/tnl_dd/t_dd_vbtmp.h b/src/mesa/tnl_dd/t_dd_vbtmp.h
index 92dd8931c3..85101b9ceb 100644
--- a/src/mesa/tnl_dd/t_dd_vbtmp.h
+++ b/src/mesa/tnl_dd/t_dd_vbtmp.h
@@ -153,46 +153,46 @@ static void TAG(emit)( GLcontext *ctx,
 
    if (DO_TEX3) {
       const GLuint t3 = GET_TEXSOURCE(3);
-      tc3 = VB->TexCoordPtr[t3]->data;
-      tc3_stride = VB->TexCoordPtr[t3]->stride;
+      tc3 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->data;
+      tc3_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->stride;
       if (DO_PTEX)
-	 tc3_size = VB->TexCoordPtr[t3]->size;
+	 tc3_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->size;
    }
 
    if (DO_TEX2) {
       const GLuint t2 = GET_TEXSOURCE(2);
-      tc2 = VB->TexCoordPtr[t2]->data;
-      tc2_stride = VB->TexCoordPtr[t2]->stride;
+      tc2 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->data;
+      tc2_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->stride;
       if (DO_PTEX)
-	 tc2_size = VB->TexCoordPtr[t2]->size;
+	 tc2_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size;
    }
 
    if (DO_TEX1) {
       const GLuint t1 = GET_TEXSOURCE(1);
-      tc1 = VB->TexCoordPtr[t1]->data;
-      tc1_stride = VB->TexCoordPtr[t1]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
       if (DO_PTEX)
-	 tc1_size = VB->TexCoordPtr[t1]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size;
    }
 
    if (DO_TEX0) {
       const GLuint t0 = GET_TEXSOURCE(0);
-      tc0_stride = VB->TexCoordPtr[t0]->stride;
-      tc0 = VB->TexCoordPtr[t0]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
       if (DO_PTEX) 
-	 tc0_size = VB->TexCoordPtr[t0]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size;
    }
 
    if (DO_RGBA) {
-      col_stride = VB->ColorPtr[0]->stride;
-      col = VB->ColorPtr[0]->data;
-      col_size = VB->ColorPtr[0]->size;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_size = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
    }
 
    if (DO_SPEC) {
-      if (VB->SecondaryColorPtr[0]) {
-	 spec_stride = VB->SecondaryColorPtr[0]->stride;
-	 spec = VB->SecondaryColorPtr[0]->data;
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
+	 spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
+	 spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
       } else {
 	 spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
 	 spec_stride = 0;
@@ -200,9 +200,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_FOG) {
-      if (VB->FogCoordPtr) {
-	 fog = VB->FogCoordPtr->data;
-	 fog_stride = VB->FogCoordPtr->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+	 fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+	 fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
       }
       else {
 	 static GLfloat tmp[4] = {0, 0, 0, 0};
@@ -356,9 +356,9 @@ static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
 
    ASSERT(stride == 4);
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
-   col_size = VB->ColorPtr[0]->size;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
+   col_size = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
 
 /*     fprintf(stderr, "%s(small) importable %x\n",  */
 /*  	   __FUNCTION__, VB->importable_data); */
@@ -410,22 +410,22 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 0] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1];
 
    if (DO_PTEX)
       return GL_TRUE;
    
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4) ||
-       (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4))
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4) ||
+       (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4))
       return GL_FALSE;
 
    return GL_TRUE;
@@ -438,14 +438,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + 0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0 + 0] = VB->AttribPtr[_TNL_ATTRIB_TEX0 + 1];
 
    if (DO_PTEX)
       return GL_TRUE;
@@ -453,14 +453,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
    /* No hardware support for projective texture.  Can fake it for
     * TEX0 only.
     */
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4)) {
       PTEX_FALLBACK();
       return GL_FALSE;
    }
 
-   if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
+   if (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4) {
       if (DO_TEX1 || DO_TEX2 || DO_TEX3) {
 	 PTEX_FALLBACK();
       }
diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c
index f72d2d84f3..c90565eae8 100644
--- a/src/mesa/vbo/vbo_exec_api.c
+++ b/src/mesa/vbo/vbo_exec_api.c
@@ -867,15 +867,27 @@ void vbo_exec_FlushVertices_internal( GLcontext *ctx, GLboolean unmap )
 }
 
 
-
+/**
+ * \param flags  bitmask of FLUSH_STORED_VERTICES, FLUSH_UPDATE_CURRENT
+ */
 void vbo_exec_FlushVertices( GLcontext *ctx, GLuint flags )
 {
    struct vbo_exec_context *exec = &vbo_context(ctx)->exec;
 
+#ifdef DEBUG
+   /* debug check: make sure we don't get called recursively */
+   static GLuint callDepth = 0;
+   callDepth++;
+   assert(callDepth == 1);
+#endif
+
    if (0) _mesa_printf("%s\n", __FUNCTION__);
 
    if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END) {
       if (0) _mesa_printf("%s - inside begin/end\n", __FUNCTION__);
+#ifdef DEBUG
+      callDepth--;
+#endif
       return;
    }
 
@@ -889,6 +901,10 @@ void vbo_exec_FlushVertices( GLcontext *ctx, GLuint flags )
    }
 
    exec->ctx->Driver.NeedFlush &= ~flags;
+
+#ifdef DEBUG
+   callDepth--;
+#endif
 }
 
 
diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c
index fd70b57b72..6de8f059b7 100644
--- a/src/mesa/vbo/vbo_exec_array.c
+++ b/src/mesa/vbo/vbo_exec_array.c
@@ -146,7 +146,7 @@ check_array_data(GLcontext *ctx, struct gl_client_array *array,
                                array->Ptr, array->BufferObj->Name);
                   f[k] = 1.0; /* XXX replace the bad value! */
                }
-               //assert(!IS_INF_OR_NAN(f[k]));
+               /*assert(!IS_INF_OR_NAN(f[k]));*/
             }
          }
          break;
diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S
index 48142d8d77..1bc2a4c115 100644
--- a/src/mesa/x86-64/glapi_x86-64.S
+++ b/src/mesa/x86-64/glapi_x86-64.S
@@ -30429,6 +30429,7 @@ GL_PREFIX(_dispatch_stub_797):
 	.globl GL_PREFIX(GetAttribLocation) ; .set GL_PREFIX(GetAttribLocation), GL_PREFIX(GetAttribLocationARB)
 	.globl GL_PREFIX(DrawBuffers) ; .set GL_PREFIX(DrawBuffers), GL_PREFIX(DrawBuffersARB)
 	.globl GL_PREFIX(DrawBuffersATI) ; .set GL_PREFIX(DrawBuffersATI), GL_PREFIX(DrawBuffersARB)
+	.globl GL_PREFIX(RenderbufferStorageMultisampleEXT) ; .set GL_PREFIX(RenderbufferStorageMultisampleEXT), GL_PREFIX(RenderbufferStorageMultisample)
 	.globl GL_PREFIX(PointParameterf) ; .set GL_PREFIX(PointParameterf), GL_PREFIX(PointParameterfEXT)
 	.globl GL_PREFIX(PointParameterfARB) ; .set GL_PREFIX(PointParameterfARB), GL_PREFIX(PointParameterfEXT)
 	.globl GL_PREFIX(PointParameterfv) ; .set GL_PREFIX(PointParameterfv), GL_PREFIX(PointParameterfvEXT)
diff --git a/src/mesa/x86/gen_matypes.c b/src/mesa/x86/gen_matypes.c
index d56b701aa8..0d7e0f1f98 100644
--- a/src/mesa/x86/gen_matypes.c
+++ b/src/mesa/x86/gen_matypes.c
@@ -120,22 +120,22 @@ int main( int argc, char **argv )
    OFFSET( "VB_COUNT               ", struct vertex_buffer, Count );
    printf( "\n" );
    OFFSET( "VB_ELTS                ", struct vertex_buffer, Elts );
-   OFFSET( "VB_OBJ_PTR             ", struct vertex_buffer, ObjPtr );
+   OFFSET( "VB_OBJ_PTR             ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_POS] );
    OFFSET( "VB_EYE_PTR             ", struct vertex_buffer, EyePtr );
    OFFSET( "VB_CLIP_PTR            ", struct vertex_buffer, ClipPtr );
    OFFSET( "VB_PROJ_CLIP_PTR       ", struct vertex_buffer, NdcPtr );
    OFFSET( "VB_CLIP_OR_MASK        ", struct vertex_buffer, ClipOrMask );
    OFFSET( "VB_CLIP_MASK           ", struct vertex_buffer, ClipMask );
-   OFFSET( "VB_NORMAL_PTR          ", struct vertex_buffer, NormalPtr );
+   OFFSET( "VB_NORMAL_PTR          ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_NORMAL] );
    OFFSET( "VB_EDGE_FLAG           ", struct vertex_buffer, EdgeFlag );
-   OFFSET( "VB_TEX0_COORD_PTR      ", struct vertex_buffer, TexCoordPtr[0] );
-   OFFSET( "VB_TEX1_COORD_PTR      ", struct vertex_buffer, TexCoordPtr[1] );
-   OFFSET( "VB_TEX2_COORD_PTR      ", struct vertex_buffer, TexCoordPtr[2] );
-   OFFSET( "VB_TEX3_COORD_PTR      ", struct vertex_buffer, TexCoordPtr[3] );
-   OFFSET( "VB_INDEX_PTR           ", struct vertex_buffer, IndexPtr );
-   OFFSET( "VB_COLOR_PTR           ", struct vertex_buffer, ColorPtr );
-   OFFSET( "VB_SECONDARY_COLOR_PTR ", struct vertex_buffer, SecondaryColorPtr );
-   OFFSET( "VB_FOG_COORD_PTR       ", struct vertex_buffer, FogCoordPtr );
+   OFFSET( "VB_TEX0_COORD_PTR      ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_TEX0] );
+   OFFSET( "VB_TEX1_COORD_PTR      ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_TEX1] );
+   OFFSET( "VB_TEX2_COORD_PTR      ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_TEX2] );
+   OFFSET( "VB_TEX3_COORD_PTR      ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_TEX3] );
+   OFFSET( "VB_INDEX_PTR           ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_COLOR_INDEX] );
+   OFFSET( "VB_COLOR_PTR           ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_COLOR0] );
+   OFFSET( "VB_SECONDARY_COLOR_PTR ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_COLOR1] );
+   OFFSET( "VB_FOG_COORD_PTR       ", struct vertex_buffer, AttribPtr[_TNL_ATTRIB_FOG] );
    OFFSET( "VB_PRIMITIVE           ", struct vertex_buffer, Primitive );
    printf( "\n" );
 
diff --git a/src/mesa/x86/glapi_x86.S b/src/mesa/x86/glapi_x86.S
index 38c988efea..64acf9646b 100644
--- a/src/mesa/x86/glapi_x86.S
+++ b/src/mesa/x86/glapi_x86.S
@@ -1229,6 +1229,7 @@ GLNAME(gl_dispatch_functions_start):
 	GL_STUB_ALIAS(GetAttribLocation, _gloffset_GetAttribLocationARB, GetAttribLocation@8, GetAttribLocationARB, GetAttribLocationARB@8)
 	GL_STUB_ALIAS(DrawBuffers, _gloffset_DrawBuffersARB, DrawBuffers@8, DrawBuffersARB, DrawBuffersARB@8)
 	GL_STUB_ALIAS(DrawBuffersATI, _gloffset_DrawBuffersARB, DrawBuffersATI@8, DrawBuffersARB, DrawBuffersARB@8)
+	GL_STUB_ALIAS(RenderbufferStorageMultisampleEXT, _gloffset_RenderbufferStorageMultisample, RenderbufferStorageMultisampleEXT@20, RenderbufferStorageMultisample, RenderbufferStorageMultisample@20)
 	GL_STUB_ALIAS(PointParameterf, _gloffset_PointParameterfEXT, PointParameterf@8, PointParameterfEXT, PointParameterfEXT@8)
 	GL_STUB_ALIAS(PointParameterfARB, _gloffset_PointParameterfEXT, PointParameterfARB@8, PointParameterfEXT, PointParameterfEXT@8)
 	GL_STUB_ALIAS(PointParameterfSGIS, _gloffset_PointParameterfEXT, PointParameterfSGIS@8, PointParameterfEXT, PointParameterfEXT@8)