Merge branch 'origin' into i915tex_privbuffersi915tex_privbuffers

Conflicts: src/mesa/drivers/dri/i915tex/intel_buffers.c src/mesa/drivers/dri/i915tex/intel_context.c src/mesa/drivers/dri/i915tex/intel_pixel_draw.c src/mesa/drivers/dri/nouveau/nouveau_buffers.c src/mesa/main/buffers.c src/mesa/main/buffers.h src/mesa/main/fbobject.c src/mesa/main/framebuffer.c
author: Roland Scheidegger <sroland@tungstengraphics.com> 2007-08-18 22:31:43 +0200
committer: Roland Scheidegger <sroland@tungstengraphics.com> 2007-08-18 22:31:43 +0200
commit: 4b4fc25c3fab92bf292a78cb9c16a5a32a8bb01b (patch)
tree: a7fb45a41c6558a3304210fd72cf211258996dbc
parent: 7aefdd518578687f961bb512a2b408857f0b9824 (diff)
parent: 53cf87be1b93c760228e6a9af8115d2a9ff99337 (diff)
238 files changed, 6828 insertions, 3326 deletions
diff --git a/Makefile b/Makefile
index ea86787ceb..2de764592a 100644
--- a/Makefile
+++ b/Makefile
@@ -66,6 +66,7 @@ aix-64 \
 aix-64-static \
 aix-gcc \
 aix-static \
+bluegene-osmesa \
 beos \
 darwin \
 darwin-static \
@@ -151,8 +152,9 @@ sunos5-v8 \
 sunos5-v8-static \
 sunos5-v9 \
 sunos5-v9-static \
+sunos5-v9-cc-g++ \
 ultrix-gcc:
-	@ if [ -e configs/current ] ; then \
+	@ if test -f configs/current || test -L configs/current ; then \
 		echo "Please run 'make realclean' before changing configs" ; \
 		exit 1 ; \
 	fi
@@ -288,17 +290,6 @@ MAIN_FILES = \
 	$(DIRECTORY)/vms/analyze_map.com				\
 	$(DIRECTORY)/vms/xlib.opt					\
 	$(DIRECTORY)/vms/xlib_share.opt					\
-	$(DIRECTORY)/windows/VC6/mesa/gdi/gdi.dsp			\
-	$(DIRECTORY)/windows/VC6/mesa/glu/*.txt				\
-	$(DIRECTORY)/windows/VC6/mesa/glu/glu.dsp			\
-	$(DIRECTORY)/windows/VC6/mesa/mesa.dsw				\
-	$(DIRECTORY)/windows/VC6/mesa/mesa/mesa.dsp			\
-	$(DIRECTORY)/windows/VC6/mesa/osmesa/osmesa.dsp			\
-	$(DIRECTORY)/windows/VC7/mesa/gdi/gdi.vcproj			\
-	$(DIRECTORY)/windows/VC7/mesa/glu/glu.vcproj			\
-	$(DIRECTORY)/windows/VC7/mesa/mesa.sln				\
-	$(DIRECTORY)/windows/VC7/mesa/mesa/mesa.vcproj			\
-	$(DIRECTORY)/windows/VC7/mesa/osmesa/osmesa.vcproj		\
 	$(DIRECTORY)/windows/VC8/mesa/mesa.sln				\
 	$(DIRECTORY)/windows/VC8/mesa/gdi/gdi.vcproj			\
 	$(DIRECTORY)/windows/VC8/mesa/glu/glu.vcproj			\
@@ -328,6 +319,7 @@ SGI_GLU_FILES = \
 	$(DIRECTORY)/src/glu/Makefile					\
 	$(DIRECTORY)/src/glu/descrip.mms				\
 	$(DIRECTORY)/src/glu/sgi/Makefile				\
+	$(DIRECTORY)/src/glu/sgi/Makefile.mgw				\
 	$(DIRECTORY)/src/glu/sgi/Makefile.win				\
 	$(DIRECTORY)/src/glu/sgi/Makefile.DJ				\
 	$(DIRECTORY)/src/glu/sgi/glu.def				\
@@ -372,6 +364,8 @@ DEMO_FILES = \
 	$(DIRECTORY)/progs/demos/*.cxx			\
 	$(DIRECTORY)/progs/demos/*.dat			\
 	$(DIRECTORY)/progs/demos/README			\
+	$(DIRECTORY)/progs/fbdev/Makefile		\
+	$(DIRECTORY)/progs/fbdev/glfbdevtest.c		\
 	$(DIRECTORY)/progs/osdemos/Makefile		\
 	$(DIRECTORY)/progs/osdemos/*.c			\
 	$(DIRECTORY)/progs/xdemos/Makefile*		\
diff --git a/Makefile.mgw b/Makefile.mgw
index 948860890c..3dc9f62643 100644
--- a/Makefile.mgw
+++ b/Makefile.mgw
@@ -53,11 +53,13 @@
 
 # MinGW core makefile updated for Mesa 7.0
 #
-#  updated : by Heromyth, 2007-6-25
+#  Updated : by Heromyth, on 2007-7-21
 #  Email   : zxpmyth@yahoo.com.cn
-#  Bug     : All the default settings work fine. But the setting X86=1 can't work. 
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
 #            The others havn't been tested yet.
-
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) MAlthough more tests are needed, it can be used individually!
 
 
 .PHONY : all libgl clean realclean
@@ -73,13 +75,14 @@ CFLAGS += -O2 -ffast-math
 
 export CFLAGS
 
+
 ifeq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
 UNLINK = del $(subst /,\,$(1))
 else
 UNLINK = $(RM) $(1)
 endif
 
-all: libgl libglu libglut
+all: libgl libglu libglut example
 
 libgl: lib
 	$(MAKE) -f Makefile.mgw -C src/mesa
diff --git a/configs/bluegene-osmesa b/configs/bluegene-osmesa
new file mode 100644
index 0000000000..02c69e6c67
--- /dev/null
+++ b/configs/bluegene-osmesa
@@ -0,0 +1,29 @@
+# Configuration for building only libOSMesa on BlueGene, no Xlib driver
+# This doesn't really have a lot of dependencies, so it should be usable
+# on other (gcc-based) systems too.
+# It uses static linking and disables multithreading.
+
+include $(TOP)/configs/default
+
+CONFIG_NAME = bluegene-osmesa
+
+# Compiler and flags
+CC = /bgl/BlueLight/ppcfloor/blrts-gnu/bin/powerpc-bgl-blrts-gnu-gcc
+CXX = /bgl/BlueLight/ppcfloor/blrts-gnu/bin/powerpc-bgl-blrts-gnu-g++
+CFLAGS = -O3 -ansi -pedantic -fPIC -ffast-math -D_POSIX_SOURCE -D_POSIX_C_SOURCE=199309L -D_SVID_SOURCE -D_BSD_SOURCE
+CXXFLAGS = -O3 -ansi -pedantic -fPIC -ffast-math -D_POSIX_SOURCE -D_POSIX_C_SOURCE=199309L -D_SVID_SOURCE -D_BSD_SOURC
+
+MKLIB_OPTIONS = -static
+
+OSMESA_LIB_NAME = libOSMesa.a
+
+# Directories
+SRC_DIRS = mesa glu
+DRIVER_DIRS = osmesa
+PROGRAM_DIRS = osdemos
+
+
+# Dependencies
+OSMESA_LIB_DEPS = -lm
+GLU_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(OSMESA_LIB)
+APP_LIB_DEPS = -lOSMesa -lGLU -lm
diff --git a/configs/config.mgw b/configs/config.mgw
new file mode 100644
index 0000000000..b961eb965c
--- /dev/null
+++ b/configs/config.mgw
@@ -0,0 +1,42 @@
+# MinGW config include file updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+# The generated DLLs by MingW with STDCALL are not totally compatible 
+# with the ones linked by Microsoft's compilers.
+#
+# xxx_USING_STDCALL = 1          Compiling MESA with __stdcall. This is default!
+# 
+# xxx_USING_STDCALL = 0          Compiling MESA without __stdcall. I like this:)
+#  
+
+# In fact, GL_USING_STDCALL and GLUT_USING_STDCALL can be
+# different. For example:
+#
+#   GL_USING_STDCALL = 0
+#   GLUT_USING_STDCALL = 1
+# 
+# Suggested setting:
+#
+#     ALL_USING_STDCALL = 1
+#
+# That's default!
+#
+
+
+ALL_USING_STDCALL = 1
+
+
+ifeq ($(ALL_USING_STDCALL),1)
+  GL_USING_STDCALL = 1
+  GLUT_USING_STDCALL = 1
+else
+  GL_USING_STDCALL = 0
+  GLUT_USING_STDCALL = 0
+endif
diff --git a/configs/freebsd b/configs/freebsd
index db7dd3b78a..447faf9f4c 100644
--- a/configs/freebsd
+++ b/configs/freebsd
@@ -13,7 +13,7 @@ OPT_FLAGS  = -O2
 PIC_FLAGS  = -fPIC
 
 DEFINES = -D_POSIX_SOURCE -D_POSIX_C_SOURCE=199309L -D_BSD_SOURCE -DUSE_XSHM \
-	-DHZ=100 -DHAVE_POSIX_MEMALIGN
+	-DHZ=100
 
 X11_INCLUDES = -I/usr/X11R6/include
 
diff --git a/configs/freebsd-dri b/configs/freebsd-dri
index 402883d1de..5a9310a04e 100644
--- a/configs/freebsd-dri
+++ b/configs/freebsd-dri
@@ -15,7 +15,7 @@ EXPAT_INCLUDES = -I/usr/local/include
 X11_INCLUDES = -I/usr/X11R6/include
 DEFINES = -DPTHREADS -DUSE_EXTERNAL_DXTN_LIB=1 -DIN_DRI_DRIVER \
 	-DGLX_DIRECT_RENDERING -DGLX_INDIRECT_RENDERING \
-	-DHAVE_ALIAS -DHAVE_POSIX_MEMALIGN
+	-DHAVE_ALIAS
 
 CFLAGS = $(WARN_FLAGS) $(OPT_FLAGS) $(PIC_FLAGS) -Wmissing-prototypes -std=c99 -Wundef -ffast-math \
 	$(ASM_FLAGS) $(X11_INCLUDES) $(DEFINES)
diff --git a/configs/sunos5-v9-cc-g++ b/configs/sunos5-v9-cc-g++
new file mode 100644
index 0000000000..37b775cc08
--- /dev/null
+++ b/configs/sunos5-v9-cc-g++
@@ -0,0 +1,32 @@
+# Configuration for SunOS 5, SPARC V9 and cc/g++ (for C and C++ sources)
+
+include $(TOP)/configs/default
+
+CONFIG_NAME = sunos5-v9-cc-g++
+
+MKLIB_OPTIONS = -cplusplus
+
+LIB_DIR = lib64
+
+# Compiler and flags
+CC = cc
+CXX = g++
+
+CXX_WARN_FLAGS = -Wall
+CXX_PIC_FLAGS  = -fPIC
+CXX_OPT_FLAGS  = -O3 -m64 -mcpu=ultrasparc -mv8plus -mvis -g -fomit-frame-pointer -pipe
+CXX_ARCH_FLAGS = -m64
+
+
+CXXFLAGS = $(CXX_WARN_FLAGS) $(CXX_OPT_FLAGS) $(CXX_PIC_FLAGS) $(CXX_ARCH_FLAGS) $(DEFINES) \
+        -I/usr/openwin/include
+
+CFLAGS = -xarch=v9 -KPIC -O -I/usr/openwin/include -I/usr/dt/include -DUSE_XSHM -DPTHREADS
+#CXXFLAGS = -xarch=v9 -KPIC -O -I/usr/openwin/include -I/usr/dt/include -DPTHREADS
+GLUT_CFLAGS = -DSOLARIS_2_4_BUG
+
+GL_LIB_DEPS = -L/usr/openwin/lib -L/usr/dt/lib -lX11 -lXext -lXmu -lXi -lm -lpthread
+GLU_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) -lm -lCstd -lCrun
+GLUT_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GLU_LIB) -l$(GL_LIB) -lm
+GLW_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -l$(GL_LIB) -L/usr/openwin/lib -lXt -lX11
+APP_LIB_DEPS = -L$(TOP)/$(LIB_DIR) -lX11 -l$(GLUT_LIB) -l$(GLU_LIB) -l$(GL_LIB) -lm
diff --git a/docs/README.MINGW32 b/docs/README.MINGW32
index 2b39f12090..9477b2bd31 100644
--- a/docs/README.MINGW32
+++ b/docs/README.MINGW32
@@ -88,3 +88,66 @@ Running the Build:
 
 		Paul G. <pgarceau@users.sourceforge.net>
 		Daniel Borca <dborca@users.sourceforge.net>
+
+
+
+******This section is added by Heromyth <zxpmyth@yahoo.com.cn>*************
+
+====================
+Updated on 2007-7-21
+====================
+
+Notice:
+	1) The generated DLLs are *not* compatible with the ones built
+with the other compilers like VC8, especially for GLUT. 
+
+	2) Although more tests are needed, it can be used individually!
+
+	3) You can set the options about whether using STDCALL to build MESA. The 
+config file is <Mesa3D-root>\configs\config.mgw. The default setting is that:
+		ALL_USING_STDCALL = 1
+, which means using STDCALL to build MESA. 
+
+	4) Of course, you can MESA without using STDCALL,I like this:) 
+The setting is :
+		ALL_USING_STDCALL = 0
+To do this, however, you must modify wingdi.h which is in MingW's include dir. 
+For example, run:
+	notepad	C:\MingW\include\wingdi.h
+, and delete all the lines where all the wgl*() functions are. Because they would 
+be conflicted with the ones in <Mesa3D-root>\include\GL\mesa_wgl.h.
+
+>>>>>>>>>> Conflicted Functions List >>>>>>>>>>
+WINGDIAPI BOOL WINAPI wglCopyContext(HGLRC,HGLRC,UINT);
+WINGDIAPI HGLRC WINAPI wglCreateContext(HDC);
+WINGDIAPI HGLRC WINAPI wglCreateLayerContext(HDC,int);
+WINGDIAPI BOOL WINAPI wglDeleteContext(HGLRC);
+WINGDIAPI BOOL WINAPI wglDescribeLayerPlane(HDC,int,int,UINT,LPLAYERPLANEDESCRIPTOR);
+WINGDIAPI HGLRC WINAPI wglGetCurrentContext(void);
+WINGDIAPI HDC WINAPI wglGetCurrentDC(void);
+WINGDIAPI int WINAPI wglGetLayerPaletteEntries(HDC,int,int,int,COLORREF*);
+WINGDIAPI PROC WINAPI wglGetProcAddress(LPCSTR);
+WINGDIAPI BOOL WINAPI wglMakeCurrent(HDC,HGLRC);
+WINGDIAPI BOOL WINAPI wglRealizeLayerPalette(HDC,int,BOOL);
+WINGDIAPI int WINAPI wglSetLayerPaletteEntries(HDC,int,int,int,const COLORREF*);
+WINGDIAPI BOOL WINAPI wglShareLists(HGLRC,HGLRC);
+WINGDIAPI BOOL WINAPI wglSwapLayerBuffers(HDC,UINT);
+WINGDIAPI BOOL WINAPI wglUseFontBitmapsA(HDC,DWORD,DWORD,DWORD);
+WINGDIAPI BOOL WINAPI wglUseFontBitmapsW(HDC,DWORD,DWORD,DWORD);
+WINGDIAPI BOOL WINAPI wglUseFontOutlinesA(HDC,DWORD,DWORD,DWORD,FLOAT,FLOAT,int,LPGLYPHMETRICSFLOAT);
+WINGDIAPI BOOL WINAPI wglUseFontOutlinesW(HDC,DWORD,DWORD,DWORD,FLOAT,FLOAT,int,LPGLYPHMETRICSFLOAT);
+<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+
+====================
+Updated on 2007-7-22
+====================
+	I havn't thought that I would find a better way to solve my problems so soon. 
+I changed the method in which the import-libs and DLLs are made. After this update,
+the DLLs of MESA are more optimized and more compatible. 
+	It seems that there is no need to keep the building way of 'NO-STDCALL'.The 
+way of USING_STDCALL is so nice! The file <Mesa3D-root>\configs\config.mgw is 
+also not needed, and can be deleted safely!
+	
+
+
+*********************************************************************************
+\ No newline at end of file
diff --git a/include/GL/gl.h b/include/GL/gl.h
index a388de36e4..3891a71875 100644
--- a/include/GL/gl.h
+++ b/include/GL/gl.h
@@ -84,7 +84,8 @@
 #include <windows.h>
 #endif
 
-#if defined(_WIN32) && !defined(_WINGDI_) && !defined(_GNU_H_WINDOWS32_DEFINES) && !defined(OPENSTEP) && !defined(__CYGWIN__)
+#if defined(_WIN32) && !defined(_WINGDI_) && !defined(_GNU_H_WINDOWS32_DEFINES) \
+     && !defined(OPENSTEP) && !defined(__CYGWIN__) || defined(__MINGW32__)
 #include <GL/mesa_wgl.h>
 #endif
 
@@ -2161,11 +2162,11 @@ typedef void (APIENTRYP PFNGLMULTITEXCOORD4SVARBPROC) (GLenum target, const GLsh
 #define GL_DEBUG_PRINT_MESA               0x875A
 #define GL_DEBUG_ASSERT_MESA              0x875B
 
-GLAPI GLhandleARB APIENTRY glCreateDebugObjectMESA (void);
-GLAPI void APIENTRY glClearDebugLogMESA (GLhandleARB obj, GLenum logType, GLenum shaderType);
-GLAPI void APIENTRY glGetDebugLogMESA (GLhandleARB obj, GLenum logType, GLenum shaderType, GLsizei maxLength,
+GLAPI GLhandleARB GLAPIENTRY glCreateDebugObjectMESA (void);
+GLAPI void GLAPIENTRY glClearDebugLogMESA (GLhandleARB obj, GLenum logType, GLenum shaderType);
+GLAPI void GLAPIENTRY glGetDebugLogMESA (GLhandleARB obj, GLenum logType, GLenum shaderType, GLsizei maxLength,
                                          GLsizei *length, GLcharARB *debugLog);
-GLAPI GLsizei APIENTRY glGetDebugLogLengthMESA (GLhandleARB obj, GLenum logType, GLenum shaderType);
+GLAPI GLsizei GLAPIENTRY glGetDebugLogLengthMESA (GLhandleARB obj, GLenum logType, GLenum shaderType);
 
 #endif /* GL_MESA_shader_debug */
 
diff --git a/include/GL/glut.h b/include/GL/glut.h
index e0fad6e5cb..137b8efa13 100644
--- a/include/GL/glut.h
+++ b/include/GL/glut.h
@@ -10,6 +10,10 @@
 #include <GL/gl.h>
 #include <GL/glu.h>
 
+#if defined(__MINGW32__)
+#include <GL/mesa_wgl.h>
+#endif
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -130,8 +134,9 @@ extern _CRTIMP void __cdecl exit(int);
 #		pragma message( "----: being multiply defined you should include WINDOWS.H priot to gl/glut.h" )
 #	endif
 #	define CALLBACK __stdcall
-typedef int (GLUTAPIENTRY *PROC)();
+
 #if !defined(__MINGW32__)
+	typedef int (GLUTAPIENTRY *PROC)();
 	typedef void *HGLRC;
 	typedef void *HDC;
 #endif
diff --git a/include/GL/mesa_wgl.h b/include/GL/mesa_wgl.h
index acc7eac2a7..57147232b7 100644
--- a/include/GL/mesa_wgl.h
+++ b/include/GL/mesa_wgl.h
@@ -26,11 +26,12 @@
 /* prototypes for the Mesa WGL functions */
 /* relocated here so that I could make GLUT get them properly */
 
-#define _mesa_wgl_h_
-
 #ifndef _mesa_wgl_h_
 #define _mesa_wgl_h_
 
+#if defined(__MINGW32__)
+#  define __W32API_USE_DLLIMPORT__
+#endif
 
 #include <GL/gl.h>
 
@@ -39,23 +40,16 @@ extern "C" {
 #endif
 
 
-#if !defined(OPENSTEP) && (defined(__WIN32__) || defined(__CYGWIN32__))
-#  if (defined(_MSC_VER) || defined(__MINGW32__)) && defined(BUILD_GL32) /* tag specify we're building mesa as a DLL */
-#    define GLAPI __declspec(dllexport)
-#    define WGLAPI __declspec(dllexport)
-#  elif (defined(_MSC_VER) || defined(__MINGW32__)) && defined(_DLL) /* tag specifying we're building for DLL runtime support */
-#    define GLAPI __declspec(dllimport)
-#    define WGLAPI __declspec(dllimport)
-#  else /* for use with static link lib build of Win32 edition only */
-#    define GLAPI extern
-#    define WGLAPI __declspec(dllimport)
-#  endif /* _STATIC_MESA support */
-#  define GLAPIENTRY __stdcall
-#else
-/* non-Windows compilation */
-#  define GLAPI extern
-#  define GLAPIENTRY
-#endif /* WIN32 / CYGWIN32 bracket */
+#ifndef WGLAPI
+#define WGLAPI GLAPI
+#endif
+
+#if defined(__MINGW32__)
+#  ifndef WIN32_LEAN_AND_MEAN
+#    define WIN32_LEAN_AND_MEAN 1
+#  endif
+#  include <windows.h>
+#endif
 
 
 #if defined(_WIN32) && !defined(_WINGDI_) && !defined(_GNU_H_WINDOWS32_DEFINES) && !defined(OPENSTEP)
@@ -80,23 +74,23 @@ typedef struct tagPIXELFORMATDESCRIPTOR PIXELFORMATDESCRIPTOR, *PPIXELFORMATDESC
 #  pragma warning( disable : 4273 ) /* 'function' : inconsistent DLL linkage. dllexport assumed. */
 #endif
 
-WGLAPI int   GLAPIENTRY wglDeleteContext(HGLRC);
-WGLAPI int   GLAPIENTRY wglMakeCurrent(HDC,HGLRC);
+
 WGLAPI int   GLAPIENTRY wglSetPixelFormat(HDC, int, const PIXELFORMATDESCRIPTOR *);
 WGLAPI int   GLAPIENTRY wglSwapBuffers(HDC hdc);
-WGLAPI HDC   GLAPIENTRY wglGetCurrentDC(void);
-WGLAPI HGLRC GLAPIENTRY wglCreateContext(HDC);
-WGLAPI HGLRC GLAPIENTRY wglCreateLayerContext(HDC,int);
-WGLAPI HGLRC GLAPIENTRY wglGetCurrentContext(void);
-WGLAPI PROC  GLAPIENTRY wglGetProcAddress(const char*);
 WGLAPI int   GLAPIENTRY wglChoosePixelFormat(HDC, const PIXELFORMATDESCRIPTOR *);
+WGLAPI int   GLAPIENTRY wglDescribePixelFormat(HDC,int, unsigned int, LPPIXELFORMATDESCRIPTOR);
+WGLAPI int   GLAPIENTRY wglGetPixelFormat(HDC hdc);
+
 WGLAPI int   GLAPIENTRY wglCopyContext(HGLRC, HGLRC, unsigned int);
+WGLAPI HGLRC GLAPIENTRY wglCreateContext(HDC);
+WGLAPI HGLRC GLAPIENTRY wglCreateLayerContext(HDC,int);
 WGLAPI int   GLAPIENTRY wglDeleteContext(HGLRC);
 WGLAPI int   GLAPIENTRY wglDescribeLayerPlane(HDC, int, int, unsigned int,LPLAYERPLANEDESCRIPTOR);
-WGLAPI int   GLAPIENTRY wglDescribePixelFormat(HDC,int, unsigned int, LPPIXELFORMATDESCRIPTOR);
+WGLAPI HGLRC GLAPIENTRY wglGetCurrentContext(void);
+WGLAPI HDC   GLAPIENTRY wglGetCurrentDC(void);
 WGLAPI int   GLAPIENTRY wglGetLayerPaletteEntries(HDC, int, int, int,COLORREF *);
-WGLAPI int   GLAPIENTRY wglGetPixelFormat(HDC hdc);
-WGLAPI int   GLAPIENTRY wglMakeCurrent(HDC, HGLRC);
+WGLAPI PROC  GLAPIENTRY wglGetProcAddress(const char*);
+WGLAPI int   GLAPIENTRY wglMakeCurrent(HDC,HGLRC);
 WGLAPI int   GLAPIENTRY wglRealizeLayerPalette(HDC, int, int);
 WGLAPI int   GLAPIENTRY wglSetLayerPaletteEntries(HDC, int, int, int,const COLORREF *);
 WGLAPI int   GLAPIENTRY wglShareLists(HGLRC, HGLRC);
@@ -105,12 +99,14 @@ WGLAPI int   GLAPIENTRY wglUseFontBitmapsA(HDC, unsigned long, unsigned long, un
 WGLAPI int   GLAPIENTRY wglUseFontBitmapsW(HDC, unsigned long, unsigned long, unsigned long);
 WGLAPI int   GLAPIENTRY wglUseFontOutlinesA(HDC, unsigned long, unsigned long, unsigned long, float,float, int, LPGLYPHMETRICSFLOAT);
 WGLAPI int   GLAPIENTRY wglUseFontOutlinesW(HDC, unsigned long, unsigned long, unsigned long, float,float, int, LPGLYPHMETRICSFLOAT);
+
+#ifndef __MINGW32__
 WGLAPI int   GLAPIENTRY SwapBuffers(HDC);
 WGLAPI int   GLAPIENTRY ChoosePixelFormat(HDC,const PIXELFORMATDESCRIPTOR *);
 WGLAPI int   GLAPIENTRY DescribePixelFormat(HDC,int,unsigned int,LPPIXELFORMATDESCRIPTOR);
 WGLAPI int   GLAPIENTRY GetPixelFormat(HDC);
 WGLAPI int   GLAPIENTRY SetPixelFormat(HDC,int,const PIXELFORMATDESCRIPTOR *);
-
+#endif
 
 #ifndef WGL_ARB_extensions_string
 #define WGL_ARB_extensions_string 1
diff --git a/progs/demos/geartrain.c b/progs/demos/geartrain.c
index 3feb2a0524..8363f2abc6 100644
--- a/progs/demos/geartrain.c
+++ b/progs/demos/geartrain.c
@@ -1053,6 +1053,7 @@ main (int argc, char *argv[])
     else
        file = argv[1];
 
+    glutInit(&argc, argv); 
     glutInitWindowPosition (0, 0);
     glutInitWindowSize(640,480);
     glutInitDisplayMode (GLUT_RGB | GLUT_DEPTH | GLUT_DOUBLE );
diff --git a/progs/demos/gltestperf.c b/progs/demos/gltestperf.c
index be95390101..2188b02419 100644
--- a/progs/demos/gltestperf.c
+++ b/progs/demos/gltestperf.c
@@ -569,6 +569,7 @@ main(int ac, char **av)
    if (ac == 2)
       frontbuffer = 0;
 
+   glutInit(&ac, av);
    glutInitDisplayMode(GLUT_DOUBLE | GLUT_RGB | GLUT_DEPTH);
    glutInitWindowPosition(0, 0);
    glutInitWindowSize(640, 480);
diff --git a/progs/demos/isosurf.c b/progs/demos/isosurf.c
index 0710bc6047..10f94b6ace 100644
--- a/progs/demos/isosurf.c
+++ b/progs/demos/isosurf.c
@@ -1042,6 +1042,7 @@ int main(int argc, char **argv)
 
    read_surface( "isosurf.dat" );
 
+   glutInit( &argc, argv);
    glutInitWindowPosition(0, 0);
    glutInitWindowSize(400, 400);
 
diff --git a/progs/demos/morph3d.c b/progs/demos/morph3d.c
index 162a6ff847..6aca8270ff 100644
--- a/progs/demos/morph3d.c
+++ b/progs/demos/morph3d.c
@@ -826,7 +826,7 @@ static void pinit(void)
 
 }
 
-static void INIT(void)
+int main(int argc, char **argv)
 {
   printf("Morph 3D - Shows morphing platonic polyhedra\n");
   printf("Author: Marcelo Fernandes Vianna (vianna@cat.cbpf.br)\n\n");
@@ -841,6 +841,7 @@ static void INIT(void)
 
   object=1;
 
+  glutInit(&argc, argv);
   glutInitWindowPosition(0,0);
   glutInitWindowSize(640,480);
 
@@ -888,9 +889,3 @@ static void INIT(void)
   glutMainLoop();
   
 }
-
-int main(int argc, char **argv)
-{
-  INIT();
-  return(0);
-}
diff --git a/progs/demos/winpos.c b/progs/demos/winpos.c
index 3a1a19ecdb..b58e330864 100644
--- a/progs/demos/winpos.c
+++ b/progs/demos/winpos.c
@@ -100,6 +100,7 @@ static void init( void )
 
 int main( int argc, char *argv[] )
 {
+   glutInit(&argc, argv);
    glutInitWindowPosition(0, 0);
    glutInitWindowSize(500, 500);
    glutInitDisplayMode( GLUT_RGB );
diff --git a/progs/fbdev/glfbdevtest.c b/progs/fbdev/glfbdevtest.c
index a461c55e2f..6d9f807030 100644
--- a/progs/fbdev/glfbdevtest.c
+++ b/progs/fbdev/glfbdevtest.c
@@ -23,11 +23,21 @@
 #include <GL/glfbdev.h>
 #include <math.h>
 
-#define DEFAULT_DEPTH 8
+
+/**
+ * Choose one of these modes
+ */
+/*static const int XRes = 1280, YRes = 1024, Hz = 75;*/
+/*static const int XRes = 1280, YRes = 1024, Hz = 70;*/
+/*static const int XRes = 1280, YRes = 1024, Hz = 60;*/
+static const int XRes = 1024, YRes = 768, Hz = 70;
+
+static int DesiredDepth = 32;
+
+static int NumFrames = 100;
 
 static struct fb_fix_screeninfo FixedInfo;
 static struct fb_var_screeninfo VarInfo, OrigVarInfo;
-static int DesiredDepth = 0;
 static int OriginalVT = -1;
 static int ConsoleFD = -1;
 static int FrameBufferFD = -1;
@@ -227,7 +237,6 @@ initialize_fbdev( void )
    VarInfo = OrigVarInfo;
 
    /* set the depth, resolution, etc */
-   DesiredDepth = 32;
    if (DesiredDepth)
       VarInfo.bits_per_pixel = DesiredDepth;
 
@@ -251,16 +260,60 @@ initialize_fbdev( void )
       VarInfo.blue.length = 8;
       VarInfo.transp.length = 8;
    }
-   /* timing values taken from /etc/fb.modes (1280x1024 @ 75Hz) */
-   VarInfo.xres_virtual = VarInfo.xres = 1280;
-   VarInfo.yres_virtual = VarInfo.yres = 1024;
-   VarInfo.pixclock = 7408;
-   VarInfo.left_margin = 248;
-   VarInfo.right_margin = 16;
-   VarInfo.upper_margin = 38;
-   VarInfo.lower_margin = 1;
-   VarInfo.hsync_len = 144;
-   VarInfo.vsync_len = 3;
+
+   /* timing values taken from /etc/fb.modes */
+   if (XRes == 1280 && YRes == 1024) {
+      VarInfo.xres_virtual = VarInfo.xres = XRes;
+      VarInfo.yres_virtual = VarInfo.yres = YRes;
+      if (Hz == 75) {
+         VarInfo.pixclock = 7408;
+         VarInfo.left_margin = 248;
+         VarInfo.right_margin = 16;
+         VarInfo.upper_margin = 38;
+         VarInfo.lower_margin = 1;
+         VarInfo.hsync_len = 144;
+         VarInfo.vsync_len = 3;
+      }
+      else if (Hz == 70) {
+         VarInfo.pixclock = 7937;
+         VarInfo.left_margin = 216;
+         VarInfo.right_margin = 80;
+         VarInfo.upper_margin = 36;
+         VarInfo.lower_margin = 1;
+         VarInfo.hsync_len = 112;
+         VarInfo.vsync_len = 5;
+      }
+      else if (Hz == 60) {
+         VarInfo.pixclock = 9260;
+         VarInfo.left_margin = 248;
+         VarInfo.right_margin = 48;
+         VarInfo.upper_margin = 38;
+         VarInfo.lower_margin = 1;
+         VarInfo.hsync_len = 112;
+         VarInfo.vsync_len = 3;
+      }
+      else {
+         fprintf(stderr, "invalid rate for 1280x1024\n");
+         exit(1);
+      }
+   }
+   else if (XRes == 1024 && YRes == 768 && Hz == 70) {
+      VarInfo.xres_virtual = VarInfo.xres = XRes;
+      VarInfo.yres_virtual = VarInfo.yres = YRes;
+      if (Hz == 70) {
+         VarInfo.pixclock = 13334;
+         VarInfo.left_margin = 144;
+         VarInfo.right_margin = 24;
+         VarInfo.upper_margin = 29;
+         VarInfo.lower_margin = 3;
+         VarInfo.hsync_len = 136;
+         VarInfo.vsync_len = 6;
+      }
+      else {
+         fprintf(stderr, "invalid rate for 1024x768\n");
+         exit(1);
+      }
+   }
 
    VarInfo.xoffset = 0;
    VarInfo.yoffset = 0;
@@ -338,7 +391,7 @@ initialize_fbdev( void )
    printf("MMIOAddress = %p\n", MMIOAddress);
 
    /* try out some simple MMIO register reads */
-   if (1)
+   if (0)
    {
       typedef unsigned int CARD32;
       typedef unsigned char CARD8;
@@ -452,6 +505,7 @@ gltest( void )
    GLFBDevVisualPtr vis;
    int bytes, r, g, b, a;
    float ang;
+   int i;
 
    printf("GLFBDEV_VENDOR = %s\n", glFBDevGetString(GLFBDEV_VENDOR));
    printf("GLFBDEV_VERSION = %s\n", glFBDevGetString(GLFBDEV_VERSION));
@@ -491,13 +545,17 @@ gltest( void )
    glEnable(GL_LIGHT0);
    glEnable(GL_DEPTH_TEST);
 
-   for (ang = 0; ang <= 180; ang += 15) {
+   printf("Drawing %d frames...\n", NumFrames);
+
+   ang = 0.0;
+   for (i = 0; i < NumFrames; i++) {
       glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
       glPushMatrix();
       glRotatef(ang, 1, 0, 0);
       doughnut(1, 3, 40, 20);
       glPopMatrix();
       glFBDevSwapBuffers(buf);
+      ang += 15.0;
    }
 
    /* clean up */
@@ -510,12 +568,29 @@ gltest( void )
 }
 
 
+static void
+parse_args(int argc, char *argv[])
+{
+   int i;
+
+   for (i = 1; i < argc; i++) {
+      if (strcmp(argv[i], "-f") == 0) {
+         NumFrames = atoi(argv[i+1]);
+         i++;
+      }
+   }
+}
+
+
 int
 main( int argc, char *argv[] )
 {
    signal(SIGUSR1, signal_handler);  /* exit if someone tries a vt switch */
    signal(SIGSEGV, signal_handler);  /* catch segfaults */
 
+   parse_args(argc, argv);
+
+   printf("Setting mode to %d x %d @ %d Hz, %d bpp\n", XRes, YRes, Hz, DesiredDepth);
    initialize_fbdev();
    gltest();
    shutdown_fbdev();
diff --git a/progs/glsl/Makefile b/progs/glsl/Makefile
index 37fa312c30..a9800c5414 100644
--- a/progs/glsl/Makefile
+++ b/progs/glsl/Makefile
@@ -8,13 +8,16 @@ INCDIR = $(TOP)/include
 LIB_DEP = $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 PROGS = \
+	bitmap \
 	brick \
 	bump \
 	deriv \
 	mandelbrot \
 	noise \
+	points \
+	texdemo1 \
 	toyball \
-	texdemo1
+	trirast
 
 
 ##### RULES #####
@@ -48,12 +51,16 @@ readtex.h: $(TOP)/progs/util/readtex.h
 readtex.o: readtex.c readtex.h
 	$(CC) -c -I$(INCDIR) $(CFLAGS) readtex.c
 
+bitmap.c: extfuncs.h
+
 brick.c: extfuncs.h
 
 bump.c: extfuncs.h
 
 mandelbrot.c: extfuncs.h
 
+points.c: extfuncs.h
+
 toyball.c: extfuncs.h
 
 texdemo1: texdemo1.o readtex.o
diff --git a/progs/glsl/bitmap.c b/progs/glsl/bitmap.c
new file mode 100644
index 0000000000..4b62686cbf
--- /dev/null
+++ b/progs/glsl/bitmap.c
@@ -0,0 +1,368 @@
+/**
+ * Implement glRasterPos + glBitmap with textures + shaders.
+ * Brian Paul
+ * 14 May 2007
+ */
+
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <math.h>
+#include <GL/gl.h>
+#include <GL/glut.h>
+#include <GL/glext.h>
+#include "extfuncs.h"
+
+
+static GLuint FragShader;
+static GLuint VertShader;
+static GLuint Program;
+
+static GLint Win = 0;
+static GLint WinWidth = 500, WinHeight = 500;
+static GLboolean Anim = GL_TRUE;
+static GLboolean Bitmap = GL_FALSE;
+static GLfloat Xrot = 20.0f, Yrot = 70.0f;
+static GLint uTex, uScale;
+static GLuint Textures[2];
+
+#define TEX_WIDTH 16
+#define TEX_HEIGHT 8
+
+
+static void
+BitmapText(const char *s)
+{
+   while (*s) {
+      glutBitmapCharacter(GLUT_BITMAP_8_BY_13, (int) *s);
+      s++;
+   }
+}
+
+
+static void
+Redisplay(void)
+{
+   static const GLfloat px[3] = { 1.2, 0, 0};
+   static const GLfloat nx[3] = {-1.2, 0, 0};
+
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   glPushMatrix();
+   glRotatef(Xrot, 1.0f, 0.0f, 0.0f);
+   glRotatef(Yrot, 0.0f, 1.0f, 0.0f);
+
+   glEnable(GL_LIGHTING);
+
+   glPushMatrix();
+   glScalef(0.5, 0.5, 0.5);
+   glutSolidDodecahedron();
+   glPopMatrix();
+
+   glDisable(GL_LIGHTING);
+
+   glColor3f(0, 1, 0);
+   glBegin(GL_LINES);
+   glVertex3f(-1, 0, 0);
+   glVertex3f( 1, 0, 0);
+   glEnd();
+
+   glColor3f(1, 1, 0);
+
+   if (Bitmap) {
+      glRasterPos3fv(px);
+      BitmapText("+X");
+      glRasterPos3fv(nx);
+      BitmapText("-X");
+   }
+   else {
+      glUseProgram_func(Program);
+
+      /* vertex positions (deltas) depend on texture size and window size */
+      if (uScale != -1) {
+         glUniform2f_func(uScale,
+                          2.0 * TEX_WIDTH / WinWidth,
+                          2.0 * TEX_HEIGHT / WinHeight);
+      }
+
+      /* draw +X */
+      glBindTexture(GL_TEXTURE_2D, Textures[0]);
+      glBegin(GL_QUADS);
+      glTexCoord2f(0, 0);  glVertex3fv(px);
+      glTexCoord2f(1, 0);  glVertex3fv(px);
+      glTexCoord2f(1, 1);  glVertex3fv(px);
+      glTexCoord2f(0, 1);  glVertex3fv(px);
+      glEnd();
+
+      /* draw -X */
+      glBindTexture(GL_TEXTURE_2D, Textures[1]);
+      glBegin(GL_QUADS);
+      glTexCoord2f(0, 0);  glVertex3fv(nx);
+      glTexCoord2f(1, 0);  glVertex3fv(nx);
+      glTexCoord2f(1, 1);  glVertex3fv(nx);
+      glTexCoord2f(0, 1);  glVertex3fv(nx);
+      glEnd();
+
+      glUseProgram_func(0);
+   }
+
+   glPopMatrix();
+
+   glutSwapBuffers();
+}
+
+
+static void
+Idle(void)
+{
+   Yrot = glutGet(GLUT_ELAPSED_TIME) * 0.01;
+   glutPostRedisplay();
+}
+
+
+static void
+Reshape(int width, int height)
+{
+   WinWidth = width;
+   WinHeight = height;
+   glViewport(0, 0, width, height);
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glFrustum(-1.0, 1.0, -1.0, 1.0, 5.0, 25.0);
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+   glTranslatef(0.0f, 0.0f, -10.0f);
+}
+
+
+static void
+Key(unsigned char key, int x, int y)
+{
+  (void) x;
+  (void) y;
+
+   switch(key) {
+   case ' ':
+   case 'a':
+      Anim = !Anim;
+      if (Anim)
+         glutIdleFunc(Idle);
+      else
+         glutIdleFunc(NULL);
+      break;
+   case 'b':
+      Bitmap = !Bitmap;
+      if (Bitmap)
+         printf("Using glBitmap\n");
+      else
+         printf("Using billboard texture\n");
+      break;
+   case 27:
+      glDeleteShader_func(FragShader);
+      glDeleteShader_func(VertShader);
+      glDeleteProgram_func(Program);
+      glutDestroyWindow(Win);
+      exit(0);
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+SpecialKey(int key, int x, int y)
+{
+   const GLfloat step = 0.125f;
+   switch(key) {
+   case GLUT_KEY_UP:
+      Xrot -= step;
+      break;
+   case GLUT_KEY_DOWN:
+      Xrot += step;
+      break;
+   case GLUT_KEY_LEFT:
+      Yrot -= step;
+      break;
+   case GLUT_KEY_RIGHT:
+      Yrot += step;
+      break;
+   }
+   /*printf("Xrot: %f  Yrot: %f\n", Xrot, Yrot);*/
+   glutPostRedisplay();
+}
+
+
+static void
+MakeTexImage(const char *p, GLuint texobj)
+{
+   GLubyte image[TEX_HEIGHT][TEX_WIDTH];
+   GLuint i, j, k;
+
+   for (i = 0; i < TEX_HEIGHT; i++) {
+      for (j = 0; j < TEX_WIDTH; j++) {
+         k = i * TEX_WIDTH + j;
+         if (p[k] == ' ') {
+            image[i][j] = 0;
+         }
+         else {
+            image[i][j] = 255;
+         }
+      }
+   }
+
+   glBindTexture(GL_TEXTURE_2D, texobj);
+   glTexImage2D(GL_TEXTURE_2D, 0, GL_INTENSITY, TEX_WIDTH, TEX_HEIGHT, 0,
+                GL_RED, GL_UNSIGNED_BYTE, image);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 1);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+}
+
+
+static void
+MakeBitmapTextures(void)
+{
+   const char *px =
+      "        X     X "
+      "   X     X   X  "
+      "   X      X X   "
+      " XXXXX     X    "
+      "   X      X X   "
+      "   X     X   X  "
+      "        X     X "
+      "        X     X ";
+   const char *nx =
+      "        X     X "
+      "         X   X  "
+      "          X X   "
+      " XXXXX     X    "
+      "          X X   "
+      "         X   X  "
+      "        X     X "
+      "        X     X ";
+   glGenTextures(2, Textures);
+   MakeTexImage(px, Textures[0]);
+   MakeTexImage(nx, Textures[1]);
+}
+
+
+static void
+LoadAndCompileShader(GLuint shader, const char *text)
+{
+   GLint stat;
+
+   glShaderSource_func(shader, 1, (const GLchar **) &text, NULL);
+
+   glCompileShader_func(shader);
+
+   glGetShaderiv_func(shader, GL_COMPILE_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetShaderInfoLog_func(shader, 1000, &len, log);
+      fprintf(stderr, "fslight: problem compiling shader:\n%s\n", log);
+      exit(1);
+   }
+}
+
+
+static void
+CheckLink(GLuint prog)
+{
+   GLint stat;
+   glGetProgramiv_func(prog, GL_LINK_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetProgramInfoLog_func(prog, 1000, &len, log);
+      fprintf(stderr, "Linker error:\n%s\n", log);
+   }
+}
+
+
+static void
+Init(void)
+{
+   /* Fragment shader: modulate raster color by texture, discard fragments
+    * with alpha < 1.0
+    */
+   static const char *fragShaderText =
+      "uniform sampler2D tex2d; \n"
+      "void main() {\n"
+      "   vec4 c = texture2D(tex2d, gl_TexCoord[0].xy); \n"
+      "   if (c.w < 1.0) \n"
+      "      discard; \n"
+      "   gl_FragColor = c * gl_Color; \n"
+      "}\n";
+   /* Vertex shader: compute new vertex position based on incoming vertex pos,
+    * texcoords and special scale factor.
+    */
+   static const char *vertShaderText =
+      "uniform vec2 scale; \n"
+      "void main() {\n"
+      "   vec4 p = gl_ModelViewProjectionMatrix * gl_Vertex;\n"
+      "   gl_Position.xy = p.xy + gl_MultiTexCoord0.xy * scale * p.w; \n"
+      "   gl_Position.zw = p.zw; \n"
+      "   gl_TexCoord[0] = gl_MultiTexCoord0; \n"
+      "   gl_FrontColor = gl_Color; \n"
+      "}\n";
+   const char *version;
+
+   version = (const char *) glGetString(GL_VERSION);
+   if (version[0] != '2' || version[1] != '.') {
+      printf("This program requires OpenGL 2.x, found %s\n", version);
+      exit(1);
+   }
+   printf("GL_RENDERER = %s\n",(const char *) glGetString(GL_RENDERER));
+
+   GetExtensionFuncs();
+
+   FragShader = glCreateShader_func(GL_FRAGMENT_SHADER);
+   LoadAndCompileShader(FragShader, fragShaderText);
+
+   VertShader = glCreateShader_func(GL_VERTEX_SHADER);
+   LoadAndCompileShader(VertShader, vertShaderText);
+
+   Program = glCreateProgram_func();
+   glAttachShader_func(Program, FragShader);
+   glAttachShader_func(Program, VertShader);
+   glLinkProgram_func(Program);
+   CheckLink(Program);
+   glUseProgram_func(Program);
+
+   uScale = glGetUniformLocation_func(Program, "scale");
+   uTex = glGetUniformLocation_func(Program, "tex2d");
+   if (uTex != -1) {
+      glUniform1i_func(uTex, 0);  /* tex unit 0 */
+   }
+
+   glUseProgram_func(0);
+
+   glClearColor(0.3f, 0.3f, 0.3f, 0.0f);
+   glEnable(GL_DEPTH_TEST);
+   glEnable(GL_NORMALIZE);
+   glEnable(GL_LIGHT0);
+
+   MakeBitmapTextures();
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   glutInit(&argc, argv);
+   glutInitWindowSize(WinWidth, WinHeight);
+   glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
+   Win = glutCreateWindow(argv[0]);
+   glutReshapeFunc(Reshape);
+   glutKeyboardFunc(Key);
+   glutSpecialFunc(SpecialKey);
+   glutDisplayFunc(Redisplay);
+   if (Anim)
+      glutIdleFunc(Idle);
+   Init();
+   glutMainLoop();
+   return 0;
+}
+
+
diff --git a/progs/glsl/points.c b/progs/glsl/points.c
new file mode 100644
index 0000000000..85115de504
--- /dev/null
+++ b/progs/glsl/points.c
@@ -0,0 +1,305 @@
+/**
+ * Implement smooth (AA) points with shaders.
+ * A simple variation could be used for sprite points.
+ * Brian Paul
+ * 29 July 2007
+ */
+
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <math.h>
+#include <GL/gl.h>
+#include <GL/glut.h>
+#include <GL/glext.h>
+#include "extfuncs.h"
+
+
+static GLuint FragShader;
+static GLuint VertShader;
+static GLuint Program;
+
+static GLint Win = 0;
+static GLint WinWidth = 500, WinHeight = 200;
+static GLfloat Xpos = 0.0f, Ypos = 0.0f;
+static GLint uViewportInv;
+static GLboolean Smooth = GL_TRUE, Blend = GL_TRUE;
+
+
+/**
+ * Issue vertices for a "shader point".
+ * The position is duplicated, only texcoords (or other vertex attrib) change.
+ * The vertex program will compute the "real" quad corners.
+ */
+static void
+PointVertex3f(GLfloat x, GLfloat y, GLfloat z)
+{
+   glTexCoord2f(-1, -1);
+   glVertex3f(x, y, z);
+
+   glTexCoord2f( 1, -1);
+   glVertex3f(x, y, z);
+
+   glTexCoord2f( 1,  1);
+   glVertex3f(x, y, z);
+
+   glTexCoord2f(-1,  1);
+   glVertex3f(x, y, z);
+}
+
+
+static void
+DrawPoints(GLboolean shaderPoints)
+{
+   int i;
+   for (i = 0; i < 9; i++) {
+      GLfloat x = i - 4, y = 0, z = 0;
+      /* note: can't call glPointSize inside Begin/End :( */
+      glPointSize( 2 + i * 5 );
+      if (shaderPoints) {
+         glBegin(GL_QUADS);
+         PointVertex3f(x, y, z);
+         glEnd();
+      }
+      else {
+         glBegin(GL_POINTS);
+         glVertex3f(x, y, z);
+         glEnd();
+      }
+   }
+}
+
+
+/**
+ * Top row of points rendered convetionally,
+ * bottom row rendered with shaders.
+ */
+static void
+Redisplay(void)
+{
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   if (Smooth)
+      glEnable(GL_POINT_SMOOTH);
+   else
+      glDisable(GL_POINT_SMOOTH);
+
+   if (Blend)
+      glEnable(GL_BLEND);
+   else
+      glDisable(GL_BLEND);
+
+   glPushMatrix();
+   glTranslatef(Xpos, Ypos, 0);
+
+   /*
+    * regular points
+    */
+   glPushMatrix();
+   glTranslatef(0, 1.2, 0);
+   glUseProgram_func(0);
+   DrawPoints(GL_FALSE);
+   glPopMatrix();
+
+   /*
+    * shader points
+    */
+   glPushMatrix();
+   glTranslatef(0, -1.2, 0);
+   glUseProgram_func(Program);
+   if (uViewportInv != -1) {
+      glUniform2f_func(uViewportInv, 1.0 / WinWidth, 1.0 / WinHeight);
+   }
+   DrawPoints(GL_TRUE);
+   glPopMatrix();
+
+   glPopMatrix();
+
+   glutSwapBuffers();
+}
+
+
+static void
+Reshape(int width, int height)
+{
+   WinWidth = width;
+   WinHeight = height;
+   glViewport(0, 0, width, height);
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glFrustum(-1.0, 1.0, -1.0, 1.0, 4.0, 30.0);
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+   glTranslatef(0.0f, 0.0f, -20.0f);
+}
+
+
+static void
+Key(unsigned char key, int x, int y)
+{
+  (void) x;
+  (void) y;
+
+   switch(key) {
+   case 'b':
+      Blend = !Blend;
+      break;
+   case 's':
+      Smooth = !Smooth;
+      break;
+   case 27:
+      glDeleteShader_func(FragShader);
+      glDeleteShader_func(VertShader);
+      glDeleteProgram_func(Program);
+      glutDestroyWindow(Win);
+      exit(0);
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+SpecialKey(int key, int x, int y)
+{
+   const GLfloat step = 1/100.0;
+   switch(key) {
+   case GLUT_KEY_UP:
+      Ypos += step;
+      break;
+   case GLUT_KEY_DOWN:
+      Ypos -= step;
+      break;
+   case GLUT_KEY_LEFT:
+      Xpos -= step;
+      break;
+   case GLUT_KEY_RIGHT:
+      Xpos += step;
+      break;
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+LoadAndCompileShader(GLuint shader, const char *text)
+{
+   GLint stat;
+
+   glShaderSource_func(shader, 1, (const GLchar **) &text, NULL);
+
+   glCompileShader_func(shader);
+
+   glGetShaderiv_func(shader, GL_COMPILE_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetShaderInfoLog_func(shader, 1000, &len, log);
+      fprintf(stderr, "fslight: problem compiling shader:\n%s\n", log);
+      exit(1);
+   }
+}
+
+
+static void
+CheckLink(GLuint prog)
+{
+   GLint stat;
+   glGetProgramiv_func(prog, GL_LINK_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetProgramInfoLog_func(prog, 1000, &len, log);
+      fprintf(stderr, "Linker error:\n%s\n", log);
+   }
+}
+
+
+static void
+Init(void)
+{
+   /* Fragment shader: compute distance of fragment from center of point
+    * (we're using texcoords but another varying could be used).
+    *   if dist > 1, discard (coverage==0)
+    *   if dist < k, coverage = 1
+    *   else, coverage = func(dist)
+    * Note: length() uses sqrt() and may be expensive.  The distance could
+    * be squared instead (with adjustments to the threshold (k) test)
+    */
+   static const char *fragShaderText =
+      "void main() {\n"
+      "   float cover; \n"
+      "   float k = 2.0 / gl_Point.size; \n"
+      "   float d = length(gl_TexCoord[0].xy); \n"
+      "   if (d >= 1.0) \n"
+      "      discard; \n"
+      "   if (d < 1.0 - k) \n"
+      "      cover = 1.0; \n"
+      "   else \n"
+      "      cover = (1.0 - d) * 0.5 * gl_Point.size; \n"
+      "   gl_FragColor.rgb = gl_Color.rgb; \n"
+      "   gl_FragColor.a = cover; \n"
+      "}\n";
+   /* Vertex shader: compute new vertex position based on incoming vertex pos,
+    * texcoords, point size, and inverse viewport scale factor.
+    * Note: should compute point size attenuation here too.
+    */
+   static const char *vertShaderText =
+      "uniform vec2 viewportInv; \n"
+      "void main() {\n"
+      "   vec4 pos = gl_ModelViewProjectionMatrix * gl_Vertex;\n"
+      "   gl_Position.xy = pos.xy + gl_MultiTexCoord0.xy * viewportInv \n"
+      "                    * gl_Point.size * pos.w; \n"
+      "   gl_Position.zw = pos.zw; \n"
+      "   gl_TexCoord[0] = gl_MultiTexCoord0; \n"
+      "   gl_FrontColor = gl_Color; \n"
+      "}\n";
+   const char *version;
+
+   version = (const char *) glGetString(GL_VERSION);
+   if (version[0] != '2' || version[1] != '.') {
+      printf("This program requires OpenGL 2.x, found %s\n", version);
+      exit(1);
+   }
+   printf("GL_RENDERER = %s\n",(const char *) glGetString(GL_RENDERER));
+
+   GetExtensionFuncs();
+
+   FragShader = glCreateShader_func(GL_FRAGMENT_SHADER);
+   LoadAndCompileShader(FragShader, fragShaderText);
+
+   VertShader = glCreateShader_func(GL_VERTEX_SHADER);
+   LoadAndCompileShader(VertShader, vertShaderText);
+
+   Program = glCreateProgram_func();
+   glAttachShader_func(Program, FragShader);
+   glAttachShader_func(Program, VertShader);
+   glLinkProgram_func(Program);
+   CheckLink(Program);
+   glUseProgram_func(Program);
+
+   uViewportInv = glGetUniformLocation_func(Program, "viewportInv");
+
+   glUseProgram_func(0);
+
+   glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   glutInit(&argc, argv);
+   glutInitWindowSize(WinWidth, WinHeight);
+   glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
+   Win = glutCreateWindow(argv[0]);
+   glutReshapeFunc(Reshape);
+   glutKeyboardFunc(Key);
+   glutSpecialFunc(SpecialKey);
+   glutDisplayFunc(Redisplay);
+   Init();
+   glutMainLoop();
+   return 0;
+}
+
+
diff --git a/progs/glsl/trirast.c b/progs/glsl/trirast.c
new file mode 100644
index 0000000000..e4325deb1f
--- /dev/null
+++ b/progs/glsl/trirast.c
@@ -0,0 +1,326 @@
+/**
+ * Demonstration of doing triangle rasterization with a fragment program.
+ * Basic idea:
+ *   1. Draw screen-aligned quad / bounding box around the triangle verts.
+ *   2. For each pixel in the quad, determine if pixel is inside/outside
+ *      the triangle edges.
+ *
+ * Brian Paul
+ * 1 Aug 2007
+ */
+
+
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <math.h>
+#include <GL/gl.h>
+#include <GL/glut.h>
+#include <GL/glext.h>
+#include "extfuncs.h"
+
+
+static GLint WinWidth = 300, WinHeight = 300;
+static char *FragProgFile = NULL;
+static char *VertProgFile = NULL;
+static GLuint fragShader;
+static GLuint vertShader;
+static GLuint program;
+static GLint win = 0;
+static GLboolean anim = GL_TRUE;
+static GLfloat Zrot = 0.0f;
+static GLint uv0, uv1, uv2;
+
+
+static const GLfloat TriVerts[3][2] = {
+   { 50,  50 },
+   { 250, 50 },
+   { 150, 250 }
+};
+
+
+static void
+RotateVerts(GLfloat a,
+            GLuint n, const GLfloat vertsIn[][2], GLfloat vertsOut[][2])
+{
+   GLuint i;
+   GLfloat cx = WinWidth / 2, cy = WinHeight / 2;
+   for (i = 0; i < n; i++) {
+      float x = vertsIn[i][0] - cx;
+      float y = vertsIn[i][1] - cy;
+      
+      vertsOut[i][0] =  x * cos(a) + y * sin(a)  + cx;
+      vertsOut[i][1] = -x * sin(a) + y * cos(a)  + cy;
+   }
+}
+
+static void
+ComputeBounds(GLuint n, GLfloat vertsIn[][2],
+              GLfloat *xmin, GLfloat *ymin,
+              GLfloat *xmax, GLfloat *ymax)
+{
+   GLuint i;
+   *xmin = *xmax = vertsIn[0][0];
+   *ymin = *ymax = vertsIn[0][1];
+   for (i = 1; i < n; i++) {
+      if (vertsIn[i][0] < *xmin)
+         *xmin = vertsIn[i][0];
+      else if (vertsIn[i][0] > *xmax)
+         *xmax = vertsIn[i][0];
+      if (vertsIn[i][1] < *ymin)
+         *ymin = vertsIn[i][1];
+      else if (vertsIn[i][1] > *ymax)
+         *ymax = vertsIn[i][1];
+   }
+}
+
+
+static void
+Redisplay(void)
+{
+   GLfloat v[3][2], xmin, ymin, xmax, ymax;
+
+   RotateVerts(Zrot, 3, TriVerts, v);
+   ComputeBounds(3, v, &xmin, &ymin, &xmax, &ymax);
+
+   glUniform2fv_func(uv0, 1, v[0]);
+   glUniform2fv_func(uv1, 1, v[1]);
+   glUniform2fv_func(uv2, 1, v[2]);
+
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   glPushMatrix();
+   glBegin(GL_POLYGON);
+   glVertex2f(xmin, ymin);
+   glVertex2f(xmax, ymin);
+   glVertex2f(xmax, ymax);
+   glVertex2f(xmin, ymax);
+   glEnd();
+   glPopMatrix();
+
+   glutSwapBuffers();
+}
+
+
+static void
+Idle(void)
+{
+   Zrot = glutGet(GLUT_ELAPSED_TIME) * 0.0005;
+   glutPostRedisplay();
+}
+
+
+static void
+Reshape(int width, int height)
+{
+   glViewport(0, 0, width, height);
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glOrtho(0, width, 0, height, -1, 1);
+
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+}
+
+
+static void
+CleanUp(void)
+{
+   glDeleteShader_func(fragShader);
+   glDeleteShader_func(vertShader);
+   glDeleteProgram_func(program);
+   glutDestroyWindow(win);
+}
+
+
+static void
+Key(unsigned char key, int x, int y)
+{
+  (void) x;
+  (void) y;
+
+   switch(key) {
+   case ' ':
+   case 'a':
+      anim = !anim;
+      if (anim)
+         glutIdleFunc(Idle);
+      else
+         glutIdleFunc(NULL);
+      break;
+   case 27:
+      CleanUp();
+      exit(0);
+      break;
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+LoadAndCompileShader(GLuint shader, const char *text)
+{
+   GLint stat;
+
+   glShaderSource_func(shader, 1, (const GLchar **) &text, NULL);
+
+   glCompileShader_func(shader);
+
+   glGetShaderiv_func(shader, GL_COMPILE_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetShaderInfoLog_func(shader, 1000, &len, log);
+      fprintf(stderr, "fslight: problem compiling shader:\n%s\n", log);
+      exit(1);
+   }
+}
+
+
+/**
+ * Read a shader from a file.
+ */
+static void
+ReadShader(GLuint shader, const char *filename)
+{
+   const int max = 100*1000;
+   int n;
+   char *buffer = (char*) malloc(max);
+   FILE *f = fopen(filename, "r");
+   if (!f) {
+      fprintf(stderr, "fslight: Unable to open shader file %s\n", filename);
+      exit(1);
+   }
+
+   n = fread(buffer, 1, max, f);
+   printf("fslight: read %d bytes from shader file %s\n", n, filename);
+   if (n > 0) {
+      buffer[n] = 0;
+      LoadAndCompileShader(shader, buffer);
+   }
+
+   fclose(f);
+   free(buffer);
+}
+
+
+static void
+CheckLink(GLuint prog)
+{
+   GLint stat;
+   glGetProgramiv_func(prog, GL_LINK_STATUS, &stat);
+   if (!stat) {
+      GLchar log[1000];
+      GLsizei len;
+      glGetProgramInfoLog_func(prog, 1000, &len, log);
+      fprintf(stderr, "Linker error:\n%s\n", log);
+   }
+}
+
+
+static void
+Init(void)
+{
+   static const char *fragShaderText =
+      "uniform vec2 v0, v1, v2; \n"
+      "float crs(const vec2 u, const vec2 v) \n"
+      "{ \n"
+      "   return u.x * v.y - u.y * v.x; \n"
+      "} \n"
+      "\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"
+      "      gl_FragColor = vec4(1.0); \n"
+      "   else \n"
+      "      gl_FragColor = vec4(0.5); \n"
+      "}\n";
+   static const char *vertShaderText =
+      "void main() {\n"
+      "   gl_Position = gl_ModelViewProjectionMatrix * gl_Vertex;\n"
+      "}\n";
+   const char *version;
+
+   version = (const char *) glGetString(GL_VERSION);
+   if (version[0] != '2' || version[1] != '.') {
+      printf("This program requires OpenGL 2.x, found %s\n", version);
+      exit(1);
+   }
+
+   GetExtensionFuncs();
+
+   fragShader = glCreateShader_func(GL_FRAGMENT_SHADER);
+   if (FragProgFile)
+      ReadShader(fragShader, FragProgFile);
+   else
+      LoadAndCompileShader(fragShader, fragShaderText);
+
+   vertShader = glCreateShader_func(GL_VERTEX_SHADER);
+   if (VertProgFile)
+      ReadShader(vertShader, VertProgFile);
+   else
+      LoadAndCompileShader(vertShader, vertShaderText);
+
+   program = glCreateProgram_func();
+   glAttachShader_func(program, fragShader);
+   glAttachShader_func(program, vertShader);
+   glLinkProgram_func(program);
+   CheckLink(program);
+   glUseProgram_func(program);
+
+   uv0 = glGetUniformLocation_func(program, "v0");
+   uv1 = glGetUniformLocation_func(program, "v1");
+   uv2 = glGetUniformLocation_func(program, "v2");
+   printf("Uniforms: %d %d %d\n", uv0, uv1, uv2);
+
+   /*assert(glGetError() == 0);*/
+
+   glClearColor(0.3f, 0.3f, 0.3f, 0.0f);
+   glEnable(GL_DEPTH_TEST);
+
+   printf("GL_RENDERER = %s\n",(const char *) glGetString(GL_RENDERER));
+
+   assert(glIsProgram_func(program));
+   assert(glIsShader_func(fragShader));
+   assert(glIsShader_func(vertShader));
+
+   glColor3f(1, 0, 0);
+}
+
+
+static void
+ParseOptions(int argc, char *argv[])
+{
+   int i;
+   for (i = 1; i < argc; i++) {
+      if (strcmp(argv[i], "-fs") == 0) {
+         FragProgFile = argv[i+1];
+      }
+      else if (strcmp(argv[i], "-vs") == 0) {
+         VertProgFile = argv[i+1];
+      }
+   }
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   glutInit(&argc, argv);
+   glutInitWindowPosition( 0, 0);
+   glutInitWindowSize(WinWidth, WinHeight);
+   glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
+   win = glutCreateWindow(argv[0]);
+   glutReshapeFunc(Reshape);
+   glutKeyboardFunc(Key);
+   glutDisplayFunc(Redisplay);
+   if (anim)
+      glutIdleFunc(Idle);
+   ParseOptions(argc, argv);
+   Init();
+   glutMainLoop();
+   return 0;
+}
diff --git a/progs/osdemos/ostest1.c b/progs/osdemos/ostest1.c
index 001e368616..2c7adfc353 100644
--- a/progs/osdemos/ostest1.c
+++ b/progs/osdemos/ostest1.c
@@ -409,6 +409,8 @@ test(GLenum type, GLint bits, const char *filename)
 
    printf("Rendering %d bit/channel image: %s\n", bits, filename);
 
+   OSMesaColorClamp(GL_TRUE);
+
    init_context();
    render_image();
    if (Gradient)
@@ -421,7 +423,7 @@ test(GLenum type, GLint bits, const char *filename)
    if (WriteFiles && filename != NULL) {
       if (type == GL_UNSIGNED_SHORT) {
          GLushort *buffer16 = (GLushort *) buffer;
-         GLubyte *buffer8 = malloc(WIDTH * HEIGHT * 4);
+         GLubyte *buffer8 = (GLubyte *) malloc(WIDTH * HEIGHT * 4);
          int i;
          for (i = 0; i < WIDTH * HEIGHT * 4; i++)
             buffer8[i] = buffer16[i] >> 8;
@@ -430,8 +432,9 @@ test(GLenum type, GLint bits, const char *filename)
       }
       else if (type == GL_FLOAT) {
          GLfloat *buffer32 = (GLfloat *) buffer;
-         GLubyte *buffer8 = malloc(WIDTH * HEIGHT * 4);
+         GLubyte *buffer8 = (GLubyte *) malloc(WIDTH * HEIGHT * 4);
          int i;
+         /* colors may be outside [0,1] so we need to clamp */
          for (i = 0; i < WIDTH * HEIGHT * 4; i++)
             buffer8[i] = (GLubyte) (buffer32[i] * 255.0);
          write_ppm(filename, buffer8, WIDTH, HEIGHT);
diff --git a/progs/samples/Makefile.mgw b/progs/samples/Makefile.mgw
index 1193540578..8bb975be9d 100644
--- a/progs/samples/Makefile.mgw
+++ b/progs/samples/Makefile.mgw
@@ -26,6 +26,15 @@
 #  Email : dborca@users.sourceforge.net
 #  Web   : http://www.geocities.com/dborca
 
+# MinGW samples makefile updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
 
 #
 #  Available options:
@@ -45,14 +54,14 @@
 TOP = ../..
 
 CC = mingw32-gcc
-CFLAGS = -Wall -W -pedantic
-CFLAGS += -O2 -ffast-math
+CFLAGS = -Wall -pedantic
+CFLAGS += -O2 -ffast-math -D_DLL
 CFLAGS += -I$(TOP)/include -I../util
 ifeq ($(FX),1)
-CFLAGS += -DFX
+  CFLAGS += -DFX
 endif
-CFLAGS += -DGLUT_DISABLE_ATEXIT_HACK -D_STDCALL_SUPPORTED
-CFLAGS += -D_WINDEF_ -D_WINGDI_
+
+CFLAGS += -DGLUT_DISABLE_ATEXIT_HACK
 
 LD = mingw32-g++
 LDFLAGS = -s -L$(TOP)/lib
@@ -62,7 +71,8 @@ LDLIBS = -lglut32 -lglu32 -lopengl32
 .c.o:
 	$(CC) -o $@ $(CFLAGS) -c $<
 %.exe: ../util/readtex.o ../util/showbuffer.o %.o
-	$(LD) -o $@ $(LDFLAGS) $^ $(LDLIBS)
+	$(LD) -o $@ $(LDFLAGS) $(LDLIBS) $^
+
 
 all:
 	$(error Must specify <filename.exe> to build)
diff --git a/progs/util/extfuncs.h b/progs/util/extfuncs.h
index 5d647cdae7..716bb7df3b 100644
--- a/progs/util/extfuncs.h
+++ b/progs/util/extfuncs.h
@@ -26,6 +26,13 @@ static PFNGLISSHADERPROC glIsShader_func = NULL;
 static PFNGLLINKPROGRAMPROC glLinkProgram_func = NULL;
 static PFNGLSHADERSOURCEPROC glShaderSource_func = NULL;
 static PFNGLUNIFORM1IPROC glUniform1i_func = NULL;
+static PFNGLUNIFORM2IPROC glUniform2i_func = NULL;
+static PFNGLUNIFORM3IPROC glUniform3i_func = NULL;
+static PFNGLUNIFORM4IPROC glUniform4i_func = NULL;
+static PFNGLUNIFORM1FPROC glUniform1f_func = NULL;
+static PFNGLUNIFORM2FPROC glUniform2f_func = NULL;
+static PFNGLUNIFORM3FPROC glUniform3f_func = NULL;
+static PFNGLUNIFORM4FPROC glUniform4f_func = NULL;
 static PFNGLUNIFORM1FVPROC glUniform1fv_func = NULL;
 static PFNGLUNIFORM2FVPROC glUniform2fv_func = NULL;
 static PFNGLUNIFORM3FVPROC glUniform3fv_func = NULL;
@@ -47,6 +54,9 @@ static PFNGLUNIFORMMATRIX4X2FVPROC glUniformMatrix4x2fv_func = NULL;
 static PFNGLUNIFORMMATRIX3X4FVPROC glUniformMatrix3x4fv_func = NULL;
 static PFNGLUNIFORMMATRIX4X3FVPROC glUniformMatrix4x3fv_func = NULL;
 
+/* OpenGL 1.4 */
+static PFNGLPOINTPARAMETERFVPROC glPointParameterfv_func = NULL;
+
 /* GL_ARB_vertex/fragment_program */
 static PFNGLBINDPROGRAMARBPROC glBindProgramARB_func = NULL;
 static PFNGLDELETEPROGRAMSARBPROC glDeleteProgramsARB_func = NULL;
@@ -94,6 +104,13 @@ GetExtensionFuncs(void)
    glLinkProgram_func = (PFNGLLINKPROGRAMPROC) glutGetProcAddress("glLinkProgram");
    glShaderSource_func = (PFNGLSHADERSOURCEPROC) glutGetProcAddress("glShaderSource");
    glUniform1i_func = (PFNGLUNIFORM1IPROC) glutGetProcAddress("glUniform1i");
+   glUniform2i_func = (PFNGLUNIFORM2IPROC) glutGetProcAddress("glUniform2i");
+   glUniform3i_func = (PFNGLUNIFORM3IPROC) glutGetProcAddress("glUniform3i");
+   glUniform4i_func = (PFNGLUNIFORM4IPROC) glutGetProcAddress("glUniform3i");
+   glUniform1f_func = (PFNGLUNIFORM1FPROC) glutGetProcAddress("glUniform1f");
+   glUniform2f_func = (PFNGLUNIFORM2FPROC) glutGetProcAddress("glUniform2f");
+   glUniform3f_func = (PFNGLUNIFORM3FPROC) glutGetProcAddress("glUniform3f");
+   glUniform4f_func = (PFNGLUNIFORM4FPROC) glutGetProcAddress("glUniform3f");
    glUniform1fv_func = (PFNGLUNIFORM1FVPROC) glutGetProcAddress("glUniform1fv");
    glUniform2fv_func = (PFNGLUNIFORM2FVPROC) glutGetProcAddress("glUniform2fv");
    glUniform3fv_func = (PFNGLUNIFORM3FVPROC) glutGetProcAddress("glUniform3fv");
@@ -115,6 +132,8 @@ GetExtensionFuncs(void)
    glUniformMatrix3x4fv_func = (PFNGLUNIFORMMATRIX3X4FVPROC) glutGetProcAddress("glUniformMatrix3x4fv");
    glUniformMatrix4x3fv_func = (PFNGLUNIFORMMATRIX4X3FVPROC) glutGetProcAddress("glUniformMatrix4x3fv");
 
+   glPointParameterfv_func = (PFNGLPOINTPARAMETERFVPROC) glutGetProcAddress("glPointParameterfv");
+
    /* GL_ARB_vertex/fragment_program */
    glBindProgramARB_func = (PFNGLBINDPROGRAMARBPROC) glutGetProcAddress("glBindProgramARB");
    glDeleteProgramsARB_func = (PFNGLDELETEPROGRAMSARBPROC) glutGetProcAddress("glDeleteProgramsARB");
diff --git a/progs/xdemos/glxgears.c b/progs/xdemos/glxgears.c
index 75d63e51a2..ec431c16f0 100644
--- a/progs/xdemos/glxgears.c
+++ b/progs/xdemos/glxgears.c
@@ -433,7 +433,7 @@ make_window( Display *dpy, const char *name,
    attr.override_redirect = fullscreen;
    mask = CWBackPixel | CWBorderPixel | CWColormap | CWEventMask | CWOverrideRedirect;
 
-   win = XCreateWindow( dpy, root, 0, 0, width, height,
+   win = XCreateWindow( dpy, root, x, y, width, height,
 		        0, visinfo->depth, InputOutput,
 		        visinfo->visual, mask, &attr );
 
@@ -548,13 +548,16 @@ usage(void)
    printf("  -stereo                 run in stereo mode\n");
    printf("  -fullscreen             run in fullscreen mode\n");
    printf("  -info                   display OpenGL renderer info\n");
+   printf("  -winwidth <width>       window width (default: 300)\n");
+   printf("  -winheight <height>     window height (default: 300)\n");
 }
  
 
 int
 main(int argc, char *argv[])
 {
-   const int winWidth = 300, winHeight = 300;
+   unsigned int winWidth = 300, winHeight = 300;
+   int x = 0, y = 0;
    Display *dpy;
    Window win;
    GLXContext ctx;
@@ -576,6 +579,10 @@ main(int argc, char *argv[])
       else if (strcmp(argv[i], "-fullscreen") == 0) {
          fullscreen = GL_TRUE;
       }
+      else if (i < argc-1 && strcmp(argv[i], "-geometry") == 0) {
+         XParseGeometry(argv[i+1], &x, &y, &winWidth, &winHeight);
+         i++;
+      }
       else {
          usage();
          return -1;
@@ -589,7 +596,7 @@ main(int argc, char *argv[])
       return -1;
    }
 
-   make_window(dpy, "glxgears", 0, 0, winWidth, winHeight, &win, &ctx);
+   make_window(dpy, "glxgears", x, y, winWidth, winHeight, &win, &ctx);
    XMapWindow(dpy, win);
    glXMakeCurrent(dpy, win, ctx);
 
diff --git a/progs/xdemos/glxswapcontrol.c b/progs/xdemos/glxswapcontrol.c
index d9be984be5..e429d58ecc 100644
--- a/progs/xdemos/glxswapcontrol.c
+++ b/progs/xdemos/glxswapcontrol.c
@@ -814,6 +814,11 @@ main(int argc, char *argv[])
 
    init();
 
+   /* Set initial projection/viewing transformation.
+    * same as glxgears.c
+    */
+   reshape(300, 300);
+
    event_loop(dpy, win);
 
    glXDestroyContext(dpy, ctx);
diff --git a/progs/xdemos/offset.c b/progs/xdemos/offset.c
index 3e92e68daa..0ad9147aea 100644
--- a/progs/xdemos/offset.c
+++ b/progs/xdemos/offset.c
@@ -71,12 +71,12 @@ typedef Vertex Quad[4];
 
 /* data to define the six faces of a unit cube */
 Quad quads[MAXQUAD] = {
-   { {0,0,0}, {1,0,0}, {1,1,0}, {0,1,0} },
-   { {0,0,1}, {1,0,1}, {1,1,1}, {0,1,1} },
-   { {0,0,0}, {1,0,0}, {1,0,1}, {0,0,1} },
-   { {0,1,0}, {1,1,0}, {1,1,1}, {0,1,1} },
-   { {0,0,0}, {0,0,1}, {0,1,1}, {0,1,0} },
-   { {1,0,0}, {1,0,1}, {1,1,1}, {1,1,0} }
+   { {0,0,0}, {0,0,1}, {0,1,1}, {0,1,0} }, /* x = 0 */
+   { {0,0,0}, {1,0,0}, {1,0,1}, {0,0,1} }, /* y = 0 */
+   { {0,0,0}, {1,0,0}, {1,1,0}, {0,1,0} }, /* z = 0 */
+   { {1,0,0}, {1,0,1}, {1,1,1}, {1,1,0} }, /* x = 1 */
+   { {0,1,0}, {1,1,0}, {1,1,1}, {0,1,1} }, /* y = 1 */
+   { {0,0,1}, {1,0,1}, {1,1,1}, {0,1,1} }  /* z = 1 */
 };
 
 #define WIREFRAME	0
@@ -86,7 +86,7 @@ static void error(const char* prog, const char* msg);
 static void cubes(int mx, int my, int mode);
 static void fill(Quad quad);
 static void outline(Quad quad);
-static void draw_hidden(Quad quad, int mode);
+static void draw_hidden(Quad quad, int mode, int face);
 static void process_input(Display *dpy, Window win);
 static int query_extension(char* extName);
 
@@ -101,6 +101,7 @@ int main(int argc, char** argv) {
     XSetWindowAttributes swa;
     Window win;
     GLXContext cx;
+    GLint z;
 
     dpy = XOpenDisplay(0);
     if (!dpy) error(argv[0], "can't open display");
@@ -134,13 +135,16 @@ int main(int argc, char** argv) {
 
     /* set up viewing parameters */
     glMatrixMode(GL_PROJECTION);
-    gluPerspective(20, 1, 0.1, 20);
+    gluPerspective(20, 1, 10, 20);
     glMatrixMode(GL_MODELVIEW);
     glTranslatef(0, 0, -15);
 
     /* set other relevant state information */
     glEnable(GL_DEPTH_TEST);
 
+    glGetIntegerv(GL_DEPTH_BITS, &z);
+    printf("GL_DEPTH_BITS = %d\n", z);
+
 #ifdef GL_EXT_polygon_offset
     printf("using 1.0 offset extension\n");
     glPolygonOffsetEXT( 1.0, 0.00001 );
@@ -160,6 +164,7 @@ int main(int argc, char** argv) {
 
 static void
 draw_scene(int mx, int my) {
+   glClearColor(0.25, 0.25, 0.25, 0);
     glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
 
     glPushMatrix();
@@ -206,7 +211,7 @@ cubes(int mx, int my, int mode) {
 		glTranslatef(x, y, z);
 		glScalef(0.8, 0.8, 0.8);
 		for (i = 0; i < MAXQUAD; i++)
-		    draw_hidden(quads[i], mode);
+                    draw_hidden(quads[i], mode, i);
 		glPopMatrix();
 	    }
 	}
@@ -236,13 +241,18 @@ outline(Quad quad) {
 }
 
 static void
-draw_hidden(Quad quad, int mode) {
+draw_hidden(Quad quad, int mode, int face) {
+    static const GLfloat colors[3][3] = {
+        {0.5, 0.5, 0.0},
+        {0.8, 0.5, 0.0},
+        {0.0, 0.5, 0.8}
+    };
     if (mode == HIDDEN_LINE) {
-	glColor3f(0, 0, 0);
+        glColor3fv(colors[face % 3]);
 	fill(quad);
     }
 
-    /* draw the outline using white, optionally fill the interior with black */
+    /* draw the outline using white */
     glColor3f(1, 1, 1);
     outline(quad);
 }
diff --git a/progs/xdemos/pbdemo.c b/progs/xdemos/pbdemo.c
index efdfdfa452..7db0017b33 100644
--- a/progs/xdemos/pbdemo.c
+++ b/progs/xdemos/pbdemo.c
@@ -93,7 +93,7 @@ MakePbuffer( Display *dpy, int screen, int width, int height )
          None
       },
       {
-         /* Single bufferd, without depth buffer */
+         /* Single buffered, without depth buffer */
          GLX_RENDER_TYPE, GLX_RGBA_BIT,
          GLX_DRAWABLE_TYPE, GLX_PBUFFER_BIT,
          GLX_RED_SIZE, 1,
@@ -105,7 +105,7 @@ MakePbuffer( Display *dpy, int screen, int width, int height )
          None
       },
       {
-         /* Double bufferd, without depth buffer */
+         /* Double buffered, without depth buffer */
          GLX_RENDER_TYPE, GLX_RGBA_BIT,
          GLX_DRAWABLE_TYPE, GLX_PBUFFER_BIT,
          GLX_RED_SIZE, 1,
@@ -130,9 +130,8 @@ MakePbuffer( Display *dpy, int screen, int width, int height )
       /* Get list of possible frame buffer configurations */
       fbConfigs = ChooseFBConfig(dpy, screen, fbAttribs[attempt], &nConfigs);
       if (nConfigs==0 || !fbConfigs) {
-         printf("Error: glXChooseFBConfig failed\n");
-         XCloseDisplay(dpy);
-         return 0;
+         printf("Note: glXChooseFBConfig(%s) failed\n", fbString[attempt]);
+         continue;
       }
 
 #if 0 /*DEBUG*/
diff --git a/src/glu/sgi/Makefile.mgw b/src/glu/sgi/Makefile.mgw
new file mode 100644
index 0000000000..d00d97a3b6
--- /dev/null
+++ b/src/glu/sgi/Makefile.mgw
@@ -0,0 +1,230 @@
+# Mesa 3-D graphics library
+# Version:  5.1
+# 
+# Copyright (C) 1999-2003  Brian Paul   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
+# BRIAN PAUL 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.
+
+# MinGW core makefile v1.4 for Mesa
+#
+#  Copyright (C) 2002 - Daniel Borca
+#  Email : dborca@users.sourceforge.net
+#  Web   : http://www.geocities.com/dborca
+
+# MinGW core-glu makefile updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+#
+#  Available options:
+#
+#    Environment variables:
+#	CFLAGS
+#
+#	GLIDE		path to Glide3 SDK; used with FX.
+#			default = $(TOP)/glide3
+#	FX=1		build for 3dfx Glide3. Note that this disables
+#			compilation of most WMesa code and requires fxMesa.
+#			As a consequence, you'll need the Win32 Glide3
+#			library to build any application.
+#			default = no
+#	ICD=1		build the installable client driver interface
+#			(windows opengl driver interface)
+#			default = no
+#	X86=1		optimize for x86 (if possible, use MMX, SSE, 3DNow).
+#			default = no
+#
+#    Targets:
+#	all:		build GL
+#	clean:		remove object files
+#
+
+
+
+.PHONY: all clean
+.INTERMEDIATE: x86/gen_matypes.exe
+.SUFFIXES: .rc .res
+
+# Set this to the prefix of your build tools, i.e. mingw32-
+TOOLS_PREFIX = mingw32-
+
+TOP = ../../..
+LIBDIR = $(TOP)/lib
+
+LIB_NAME = glu32
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+GLU_DLL = $(LIB_NAME)$(DLL_EXT)
+GLU_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+GLU_DEF = $(LIB_NAME).def
+
+LDLIBS = -L$(LIBDIR) -lopengl32
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GLU32 -D_DLL
+
+
+CC = $(TOOLS_PREFIX)gcc
+CFLAGS += -DNDEBUG -DLIBRARYBUILD -I$(TOP)/include -Iinclude
+CXX = $(TOOLS_PREFIX)g++
+CXXFLAGS = $(CFLAGS) -Ilibnurbs/internals -Ilibnurbs/interface -Ilibnurbs/nurbtess
+
+AR = ar
+ARFLAGS = crus
+
+UNLINK = del $(subst /,\,$(1))
+ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+ifneq ($(wildcard $(addsuffix /rm,$(subst :, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+
+C_SOURCES = \
+	libutil/error.c \
+	libutil/glue.c \
+	libutil/mipmap.c \
+	libutil/project.c \
+	libutil/quad.c \
+	libutil/registry.c \
+	libtess/dict.c \
+	libtess/geom.c \
+	libtess/memalloc.c \
+	libtess/mesh.c \
+	libtess/normal.c \
+	libtess/priorityq.c \
+	libtess/render.c \
+	libtess/sweep.c \
+	libtess/tess.c \
+	libtess/tessmono.c
+
+CC_SOURCES = \
+	libnurbs/interface/bezierEval.cc \
+	libnurbs/interface/bezierPatch.cc \
+	libnurbs/interface/bezierPatchMesh.cc \
+	libnurbs/interface/glcurveval.cc \
+	libnurbs/interface/glinterface.cc \
+	libnurbs/interface/glrenderer.cc \
+	libnurbs/interface/glsurfeval.cc \
+	libnurbs/interface/incurveeval.cc \
+	libnurbs/interface/insurfeval.cc \
+	libnurbs/internals/arc.cc \
+	libnurbs/internals/arcsorter.cc \
+	libnurbs/internals/arctess.cc \
+	libnurbs/internals/backend.cc \
+	libnurbs/internals/basiccrveval.cc \
+	libnurbs/internals/basicsurfeval.cc \
+	libnurbs/internals/bin.cc \
+	libnurbs/internals/bufpool.cc \
+	libnurbs/internals/cachingeval.cc \
+	libnurbs/internals/ccw.cc \
+	libnurbs/internals/coveandtiler.cc \
+	libnurbs/internals/curve.cc \
+	libnurbs/internals/curvelist.cc \
+	libnurbs/internals/curvesub.cc \
+	libnurbs/internals/dataTransform.cc \
+	libnurbs/internals/displaylist.cc \
+	libnurbs/internals/flist.cc \
+	libnurbs/internals/flistsorter.cc \
+	libnurbs/internals/hull.cc \
+	libnurbs/internals/intersect.cc \
+	libnurbs/internals/knotvector.cc \
+	libnurbs/internals/mapdesc.cc \
+	libnurbs/internals/mapdescv.cc \
+	libnurbs/internals/maplist.cc \
+	libnurbs/internals/mesher.cc \
+	libnurbs/internals/monoTriangulationBackend.cc \
+	libnurbs/internals/monotonizer.cc \
+	libnurbs/internals/mycode.cc \
+	libnurbs/internals/nurbsinterfac.cc \
+	libnurbs/internals/nurbstess.cc \
+	libnurbs/internals/patch.cc \
+	libnurbs/internals/patchlist.cc \
+	libnurbs/internals/quilt.cc \
+	libnurbs/internals/reader.cc \
+	libnurbs/internals/renderhints.cc \
+	libnurbs/internals/slicer.cc \
+	libnurbs/internals/sorter.cc \
+	libnurbs/internals/splitarcs.cc \
+	libnurbs/internals/subdivider.cc \
+	libnurbs/internals/tobezier.cc \
+	libnurbs/internals/trimline.cc \
+	libnurbs/internals/trimregion.cc \
+	libnurbs/internals/trimvertpool.cc \
+	libnurbs/internals/uarray.cc \
+	libnurbs/internals/varray.cc \
+	libnurbs/nurbtess/directedLine.cc \
+	libnurbs/nurbtess/gridWrap.cc \
+	libnurbs/nurbtess/monoChain.cc \
+	libnurbs/nurbtess/monoPolyPart.cc \
+	libnurbs/nurbtess/monoTriangulation.cc \
+	libnurbs/nurbtess/partitionX.cc \
+	libnurbs/nurbtess/partitionY.cc \
+	libnurbs/nurbtess/polyDBG.cc \
+	libnurbs/nurbtess/polyUtil.cc \
+	libnurbs/nurbtess/primitiveStream.cc \
+	libnurbs/nurbtess/quicksort.cc \
+	libnurbs/nurbtess/rectBlock.cc \
+	libnurbs/nurbtess/sampleComp.cc \
+	libnurbs/nurbtess/sampleCompBot.cc \
+	libnurbs/nurbtess/sampleCompRight.cc \
+	libnurbs/nurbtess/sampleCompTop.cc \
+	libnurbs/nurbtess/sampleMonoPoly.cc \
+	libnurbs/nurbtess/sampledLine.cc \
+	libnurbs/nurbtess/searchTree.cc
+
+SOURCES = $(C_SOURCES) $(CC_SOURCES)
+
+OBJECTS = $(addsuffix .o,$(basename $(SOURCES)))
+
+.c.o:
+	$(CC) -o $@ $(CFLAGS) -c $<
+.cc.o:
+	$(CXX) -o $@ $(CXXFLAGS) -c $<
+
+
+all: $(LIBDIR) $(LIBDIR)/$(GLU_DLL) $(LIBDIR)/$(GLU_IMP)
+
+$(LIBDIR):
+	mkdir -p $(LIBDIR)
+
+$(LIBDIR)/$(GLU_DLL) $(LIBDIR)/$(GLU_IMP): $(OBJECTS)
+	$(CXX) $(LDFLAGS) -o $(LIBDIR)/$(GLU_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GLU_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GLU_IMP) --def $(LIBDIR)/$(GLU_DEF)
+
+
+clean:
+	-$(call UNLINK,libutil/*.o)
+	-$(call UNLINK,libtess/*.o)
+	-$(call UNLINK,libnurbs/interface/*.o)
+	-$(call UNLINK,libnurbs/internals/*.o)
+	-$(call UNLINK,libnurbs/nurbtess/*.o)
diff --git a/src/glu/sgi/libnurbs/interface/bezierPatch.cc b/src/glu/sgi/libnurbs/interface/bezierPatch.cc
index 836ae94e0a..5b7bda1a00 100644
--- a/src/glu/sgi/libnurbs/interface/bezierPatch.cc
+++ b/src/glu/sgi/libnurbs/interface/bezierPatch.cc
@@ -111,8 +111,11 @@ void bezierPatchDelete(bezierPatch *b)
 void bezierPatchDeleteList(bezierPatch *b)
 {
   bezierPatch *temp;
-  for(temp = b; temp != NULL; temp = temp->next)
-    bezierPatchDelete(temp);
+  while (b != NULL) {
+    temp = b;
+    b = b->next;
+    bezierPatchDelete(temp);    
+  }
 }
 
 bezierPatch* bezierPatchInsert(bezierPatch *list, bezierPatch *b)
diff --git a/src/glu/sgi/libnurbs/interface/glcurveval.h b/src/glu/sgi/libnurbs/interface/glcurveval.h
index 4b44f6e847..a09a74d04c 100644
--- a/src/glu/sgi/libnurbs/interface/glcurveval.h
+++ b/src/glu/sgi/libnurbs/interface/glcurveval.h
@@ -93,7 +93,7 @@ public:
 	output_triangles = flag;
       }
 #ifdef _WIN32
-    void               putCallBack(GLenum which, void (APIENTRY *fn)() );
+    void               putCallBack(GLenum which, void (GLAPIENTRY *fn)() );
 #else
     void               putCallBack(GLenum which, _GLUfuncptr fn );
 #endif
diff --git a/src/glu/sgi/libnurbs/interface/glsurfeval.cc b/src/glu/sgi/libnurbs/interface/glsurfeval.cc
index a36b304508..b5bfab1e28 100644
--- a/src/glu/sgi/libnurbs/interface/glsurfeval.cc
+++ b/src/glu/sgi/libnurbs/interface/glsurfeval.cc
@@ -1184,8 +1184,11 @@ return;
 
 }
 
-void
-OpenGLSurfaceEvaluator::putCallBack(GLenum which, _GLUfuncptr fn )
+#ifdef _WIN32
+void OpenGLSurfaceEvaluator::putCallBack(GLenum which, void (GLAPIENTRY *fn)() )
+#else
+void OpenGLSurfaceEvaluator::putCallBack(GLenum which, _GLUfuncptr fn )
+#endif
 {
   switch(which)
     {
diff --git a/src/glu/sgi/libnurbs/interface/glsurfeval.h b/src/glu/sgi/libnurbs/interface/glsurfeval.h
index c34a58cb9b..b7a88069f5 100644
--- a/src/glu/sgi/libnurbs/interface/glsurfeval.h
+++ b/src/glu/sgi/libnurbs/interface/glsurfeval.h
@@ -145,7 +145,7 @@ public:
     void		newtmeshvert( long, long );
 
 #ifdef _WIN32
-    void 	        putCallBack(GLenum which, void (APIENTRY *fn)() );
+    void 	        putCallBack(GLenum which, void (GLAPIENTRY *fn)() );
 #else
     void 	        putCallBack(GLenum which, _GLUfuncptr fn );
 #endif
diff --git a/src/glu/sgi/libnurbs/interface/insurfeval.cc b/src/glu/sgi/libnurbs/interface/insurfeval.cc
index b314699c7a..e4ee3ef6a2 100644
--- a/src/glu/sgi/libnurbs/interface/insurfeval.cc
+++ b/src/glu/sgi/libnurbs/interface/insurfeval.cc
@@ -1533,8 +1533,8 @@ void OpenGLSurfaceEvaluator::inEvalVStrip(int n_left, REAL u_left, REAL* left_va
     }
   //clean up 
   free(leftXYZ);
-  free(leftXYZ);
-  free(rightNormal);
+  free(rightXYZ);
+  free(leftNormal);
   free(rightNormal);
 }
 
diff --git a/src/glu/sgi/libutil/error.c b/src/glu/sgi/libutil/error.c
index 3d1ce9b210..c200056942 100644
--- a/src/glu/sgi/libutil/error.c
+++ b/src/glu/sgi/libutil/error.c
@@ -82,7 +82,7 @@ gluErrorString(GLenum errorCode)
     if ((errorCode >= GLU_NURBS_ERROR1) && (errorCode <= GLU_NURBS_ERROR37)) {
 	return (const GLubyte *) __gluNURBSErrorString(errorCode - (GLU_NURBS_ERROR1 - 1));
     }
-    if ((errorCode >= GLU_TESS_ERROR1) && (errorCode <= GLU_TESS_ERROR8)) {
+    if ((errorCode >= GLU_TESS_ERROR1) && (errorCode <= GLU_TESS_ERROR6)) {
 	return (const GLubyte *) __gluTessErrorString(errorCode - (GLU_TESS_ERROR1 - 1));
     }
     return (const GLubyte *) 0;
diff --git a/src/glut/glx/Makefile.mgw b/src/glut/glx/Makefile.mgw
new file mode 100644
index 0000000000..9fff2e1503
--- /dev/null
+++ b/src/glut/glx/Makefile.mgw
@@ -0,0 +1,191 @@
+# Mesa 3-D graphics library
+# Version:  5.1
+# 
+# Copyright (C) 1999-2003  Brian Paul   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
+# BRIAN PAUL 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.
+
+# MinGW core makefile v1.4 for Mesa
+#
+#  Copyright (C) 2002 - Daniel Borca
+#  Email : dborca@users.sourceforge.net
+#  Web   : http://www.geocities.com/dborca
+
+# MinGW core-glut makefile updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+
+#
+#  Available options:
+#
+#    Environment variables:
+#	CFLAGS
+#
+#	GLIDE		path to Glide3 SDK; used with FX.
+#			default = $(TOP)/glide3
+#	FX=1		build for 3dfx Glide3. Note that this disables
+#			compilation of most WMesa code and requires fxMesa.
+#			As a consequence, you'll need the Win32 Glide3
+#			library to build any application.
+#			default = no
+#	ICD=1		build the installable client driver interface
+#			(windows opengl driver interface)
+#			default = no
+#	X86=1		optimize for x86 (if possible, use MMX, SSE, 3DNow).
+#			default = no
+#
+#    Targets:
+#	all:		build GL
+#	clean:		remove object files
+#
+
+
+
+.PHONY: all clean
+.INTERMEDIATE: x86/gen_matypes.exe
+.SUFFIXES: .rc .res
+
+# Set this to the prefix of your build tools, i.e. mingw32-
+TOOLS_PREFIX = mingw32-
+
+TOP = ../../..
+
+LIBDIR = $(TOP)/lib
+
+LIB_NAME = glut32
+
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+GLUT_DLL = $(LIB_NAME)$(DLL_EXT)
+GLUT_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+GLUT_DEF = $(LIB_NAME).def
+
+LDLIBS = -L$(LIBDIR) -lwinmm -lgdi32 -luser32 -lopengl32 -lglu32 
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GLUT32 -DGLUT_BUILDING_LIB -DMESA -D_DLL
+CFLAGS += -DNDEBUG -DLIBRARYBUILD -I$(TOP)/include 
+
+CC = $(TOOLS_PREFIX)gcc
+CXX = $(TOOLS_PREFIX)g++
+CXXFLAGS = $(CFLAGS)
+
+AR = ar
+ARFLAGS = crus
+
+UNLINK = del $(subst /,\,$(1))
+ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+ifneq ($(wildcard $(addsuffix /rm,$(subst :, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+
+HDRS = glutint.h glutstroke.h glutbitmap.h glutwin32.h stroke.h win32_glx.h win32_x11.h
+
+SRCS = \
+	glut_bitmap.c \
+	glut_bwidth.c \
+	glut_cindex.c \
+	glut_cmap.c \
+	glut_cursor.c \
+	glut_dials.c \
+	glut_dstr.c \
+	glut_event.c \
+	glut_ext.c \
+	glut_fbc.c \
+	glut_fullscrn.c \
+	glut_gamemode.c \
+	glut_get.c \
+	glut_init.c \
+	glut_input.c \
+	glut_joy.c \
+	glut_key.c \
+	glut_keyctrl.c \
+	glut_keyup.c \
+	glut_mesa.c \
+	glut_modifier.c \
+	glut_overlay.c \
+	glut_shapes.c \
+	glut_space.c \
+	glut_stroke.c \
+	glut_swap.c \
+	glut_swidth.c \
+	glut_tablet.c \
+	glut_teapot.c \
+	glut_util.c \
+	glut_vidresize.c \
+	glut_warp.c \
+	glut_win.c \
+	glut_winmisc.c \
+	win32_glx.c \
+	win32_menu.c \
+	win32_util.c \
+	win32_winproc.c \
+	win32_x11.c
+	
+
+SRCSSEMIGENS = \
+	glut_8x13.c \
+	glut_9x15.c \
+	glut_hel10.c \
+	glut_hel12.c \
+	glut_hel18.c \
+	glut_mroman.c \
+	glut_roman.c \
+	glut_tr10.c \
+	glut_tr24.c
+
+
+
+SOURCES = $(SRCS) $(SRCSSEMIGENS)
+
+OBJECTS = $(addsuffix .o,$(basename $(SOURCES)))
+
+.c.o:
+	$(CC) -o $@ $(CFLAGS) -c $<
+.cc.o:
+	$(CXX) -o $@ $(CXXFLAGS) -c $<
+
+
+all: $(LIBDIR) $(LIBDIR)/$(GLUT_DLL) $(LIBDIR)/$(GLUT_IMP)
+
+$(LIBDIR):
+	mkdir -p $(LIBDIR)
+
+$(LIBDIR)/$(GLUT_DLL) $(LIBDIR)/$(GLUT_IMP): $(OBJECTS)
+	$(CXX) $(LDFLAGS) -o $(LIBDIR)/$(GLUT_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GLUT_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GLUT_IMP) --def $(LIBDIR)/$(GLUT_DEF) 
+
+clean:
+	-$(call UNLINK,*.o)
+\ No newline at end of file
diff --git a/src/glut/glx/glut_dstr.c b/src/glut/glx/glut_dstr.c
index ec9386f3e5..2513af4539 100644
--- a/src/glut/glx/glut_dstr.c
+++ b/src/glut/glx/glut_dstr.c
@@ -291,6 +291,7 @@ loadVisuals(int *nitems_return)
   fbmodes = (FrameBufferMode *) malloc(n * sizeof(FrameBufferMode));
   if (fbmodes == NULL) {
     *nitems_return = -1;
+    free(vlist);
     return NULL;
   }
   for (i = 0; i < n; i++) {
diff --git a/src/glut/glx/glut_event.c b/src/glut/glx/glut_event.c
index 4e67da674e..0a96e8cf71 100644
--- a/src/glut/glx/glut_event.c
+++ b/src/glut/glx/glut_event.c
@@ -24,7 +24,7 @@
 # ifdef __sgi
 #  include <bstring.h>    /* prototype for bzero used by FD_ZERO */
 # endif
-# if (defined(SVR4) || defined(CRAY) || defined(AIXV3)) && !defined(FD_SETSIZE)
+# if (defined(__FreeBSD__) || defined(SVR4) || defined(CRAY) || defined(AIXV3)) && !defined(FD_SETSIZE)
 #  include <sys/select.h> /* select system call interface */
 #  ifdef luna
 #   include <sysent.h>
diff --git a/src/glut/glx/glut_fbc.c b/src/glut/glx/glut_fbc.c
index deb46c3d8d..e93188b862 100644
--- a/src/glut/glx/glut_fbc.c
+++ b/src/glut/glx/glut_fbc.c
@@ -18,7 +18,7 @@
 
 /* Set a Fortran callback function. */
 
-void GLUTAPIENTRY
+void APIENTRY
 __glutSetFCB(int which, void *func)
 {
 #ifdef SUPPORT_FORTRAN
@@ -100,7 +100,7 @@ __glutSetFCB(int which, void *func)
 
 /* Get a Fortran callback function. */
 
-void* GLUTAPIENTRY
+void* APIENTRY
 __glutGetFCB(int which)
 {
 #ifdef SUPPORT_FORTRAN
diff --git a/src/glut/glx/glutint.h b/src/glut/glx/glutint.h
index 6fe09ffe7e..a962c78023 100644
--- a/src/glut/glx/glutint.h
+++ b/src/glut/glx/glutint.h
@@ -26,7 +26,10 @@
 #include <GL/glx.h>
 #endif
 
+#ifndef GLUT_BUILDING_LIB
 #define GLUT_BUILDING_LIB  /* Building the GLUT library itself. */
+#endif
+
 #include <GL/glut.h>
 
 #if defined(MESA) && defined(_WIN32) && !defined(__CYGWIN32__)
diff --git a/src/glut/glx/win32_winproc.c b/src/glut/glx/win32_winproc.c
index a54bac75fa..e1fc785ebb 100644
--- a/src/glut/glx/win32_winproc.c
+++ b/src/glut/glx/win32_winproc.c
@@ -9,6 +9,9 @@
 
 #include "glutint.h"
 #include <sys/timeb.h>
+#ifdef __MINGW32__
+#include <ctype.h>
+#endif
 
 #if defined(_WIN32) && !defined(__CYGWIN32__)
 #include <mmsystem.h>  /* Win32 Multimedia API header. */
diff --git a/src/glx/x11/glxext.c b/src/glx/x11/glxext.c
index af3a5166dc..8fe10338a8 100644
--- a/src/glx/x11/glxext.c
+++ b/src/glx/x11/glxext.c
@@ -672,8 +672,11 @@ filter_modes( __GLcontextModes ** server_modes,
 	if ( do_delete && (m->visualID != 0) ) {
 	    do_delete = GL_FALSE;
 
-	    fprintf(stderr, "libGL warning: 3D driver claims to not support "
-		    "visual 0x%02x\n", m->visualID);
+	    /* don't warn for this visual (Novell #247471 / X.Org #6689) */
+	    if (m->visualRating != GLX_NON_CONFORMANT_CONFIG) {
+		fprintf(stderr, "libGL warning: 3D driver claims to not "
+			"support visual 0x%02x\n", m->visualID);
+	    }
 	}
 
 	if ( do_delete ) {
diff --git a/src/mesa/Makefile.mgw b/src/mesa/Makefile.mgw
index ebec5c055a..6244ded876 100644
--- a/src/mesa/Makefile.mgw
+++ b/src/mesa/Makefile.mgw
@@ -1,5 +1,5 @@
 # Mesa 3-D graphics library
-# Version:  5.1
+# Version:  7.0
 # 
 # Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
 # 
@@ -26,6 +26,16 @@
 #  Email : dborca@users.sourceforge.net
 #  Web   : http://www.geocities.com/dborca
 
+# MinGW core-gl makefile updated for Mesa 7.0
+#
+#  updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
 
 #
 #  Available options:
@@ -52,7 +62,6 @@
 #
 
 
-
 .PHONY: all clean
 .INTERMEDIATE: x86/gen_matypes.exe
 .SUFFIXES: .rc .res
@@ -60,23 +69,37 @@
 # Set this to the prefix of your build tools, i.e. mingw32-
 TOOLS_PREFIX = mingw32-
 
-TOP = ../..
-GLIDE ?= $(TOP)/glide3
-LIBDIR = $(TOP)/lib
 ifeq ($(ICD),1)
-  GL_DLL = mesa32.dll
-  GL_IMP = libmesa32.a
+  LIB_NAME = mesa32
 else
-  GL_DLL = opengl32.dll
-  GL_IMP = libopengl32.a
+  LIB_NAME = opengl32
 endif
 
-LDLIBS = -lgdi32
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+TOP = ../..
+GLIDE ?= $(TOP)/glide3
+LIBDIR = $(TOP)/lib
 
+GL_DLL = $(LIB_NAME)$(DLL_EXT)
+GL_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+
+MESA_LIB = libmesa.a
 CC = $(TOOLS_PREFIX)gcc
-CFLAGS += -DBUILD_GL32 -D_OPENGL32_
-CFLAGS += $(INCLUDE_DIRS)
-CFLAGS += -DUSE_EXTERNAL_DXTN_LIB=1
+
+LDLIBS = -lgdi32 -luser32 -liberty
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GL32 -D_DLL -DMESA_MINWARN
+CFLAGS += -DNDEBUG -DUSE_EXTERNAL_DXTN_LIB=1
+
 ifeq ($(FX),1)
   CFLAGS += -I$(GLIDE)/include -DFX
   LDLIBS += -L$(GLIDE)/lib -lglide3x
@@ -87,12 +110,11 @@ else
     CFLAGS += -DUSE_MGL_NAMESPACE
     GL_DEF = drivers/windows/icd/mesa.def
   else    
-    GL_DEF = drivers/windows/gdi/mesa.def
+    GL_DEF = $(LIB_NAME).def
   endif
 endif
 
-AR = ar
-ARFLAGS = crus
+
 
 UNLINK = del $(subst /,\,$(1))
 ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
@@ -104,6 +126,8 @@ endif
 
 include sources
 
+CFLAGS += $(INCLUDE_DIRS) 
+
 ifeq ($(X86),1)
 CFLAGS += -DUSE_X86_ASM
 CFLAGS += -DUSE_MMX_ASM
@@ -140,22 +164,23 @@ RESOURCE = $(GL_RES:.rc=.res)
 
 .c.o:
 	$(CC) -o $@ $(CFLAGS) -c $<
-.S.o:
-	$(CC) -o $@ $(CFLAGS) -c $<
 .s.o:
 	$(CC) -o $@ $(CFLAGS) -x assembler-with-cpp -c $<
+
 .rc.res:
 	windres -o $@ -Irc -Ocoff $<
 
+
 all: $(LIBDIR) $(LIBDIR)/$(GL_DLL) $(LIBDIR)/$(GL_IMP)
 
 $(LIBDIR):
 	mkdir -p $(LIBDIR)
 
 $(LIBDIR)/$(GL_DLL) $(LIBDIR)/$(GL_IMP): $(OBJECTS) $(RESOURCE)
-	$(TOOLS_PREFIX)dllwrap -o $(LIBDIR)/$(GL_DLL) --output-lib $(LIBDIR)/$(GL_IMP) \
-	                       --target i386-mingw32 --def $(GL_DEF) -Wl,-enable-stdcall-fixup \
-	                       $^ $(LDLIBS)
+	$(CC) $(LDFLAGS) -o $(LIBDIR)/$(GL_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GL_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GL_IMP) --def $(LIBDIR)/$(GL_DEF)
+
 
 $(X86_OBJECTS): x86/matypes.h
 
@@ -187,17 +212,21 @@ tnl/t_vtx_x86_gcc.o: tnl/t_vtx_x86_gcc.S
 	$(CC) -o $@ $(CFLAGS) -DSTDCALL_API -c $<
 
 clean:
-	-$(call UNLINK,array_cache/*.o)
 	-$(call UNLINK,glapi/*.o)
 	-$(call UNLINK,main/*.o)
 	-$(call UNLINK,math/*.o)
+	-$(call UNLINK,vbo/*.o)
 	-$(call UNLINK,shader/*.o)
+	-$(call UNLINK,shader/slang/*.o)
+	-$(call UNLINK,shader/grammar/*.o)
 	-$(call UNLINK,sparc/*.o)
 	-$(call UNLINK,ppc/*.o)
 	-$(call UNLINK,swrast/*.o)
 	-$(call UNLINK,swrast_setup/*.o)
 	-$(call UNLINK,tnl/*.o)
 	-$(call UNLINK,x86/*.o)
+	-$(call UNLINK,x86/rtasm/*.o)
+	-$(call UNLINK,x86-64/*.o)
 	-$(call UNLINK,drivers/common/*.o)
 	-$(call UNLINK,drivers/glide/*.o)
 	-$(call UNLINK,drivers/windows/fx/*.o)
diff --git a/src/mesa/drivers/dri/common/dri_drmpool.c b/src/mesa/drivers/dri/common/dri_drmpool.c
index 878a148b39..592ac10542 100644
--- a/src/mesa/drivers/dri/common/dri_drmpool.c
+++ b/src/mesa/drivers/dri/common/dri_drmpool.c
@@ -51,6 +51,7 @@ pool_create(struct _DriBufferPool *pool,
       return NULL;
 
    if ((alignment > pageSize) && (alignment % pageSize)) {
+      free(buf);
       return NULL;
    }
 
diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c
index d785382f39..6e8a5b5218 100644
--- a/src/mesa/drivers/dri/common/dri_util.c
+++ b/src/mesa/drivers/dri/common/dri_util.c
@@ -434,10 +434,12 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
 
     if (pdp->pClipRects) {
 	_mesa_free(pdp->pClipRects); 
+	pdp->pClipRects = NULL;
     }
 
     if (pdp->pBackClipRects) {
 	_mesa_free(pdp->pBackClipRects); 
+	pdp->pBackClipRects = NULL;
     }
 
     DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h
index 50f3cf5581..53f5f846a0 100644
--- a/src/mesa/drivers/dri/common/spantmp2.h
+++ b/src/mesa/drivers/dri/common/spantmp2.h
@@ -114,7 +114,7 @@
     do {                                                                \
         GLuint p = *(volatile GLuint *) GET_PTR(_x, _y);                \
        __asm__ __volatile__( "bswap	%0; rorl $8, %0"                \
-				: "=r" (p) : "r" (p) );                 \
+				: "=r" (p) : "0" (p) );                 \
        ((GLuint *)rgba)[0] = p;                                         \
     } while (0)
 # elif defined( MESA_BIG_ENDIAN )
diff --git a/src/mesa/drivers/dri/i810/i810screen.c b/src/mesa/drivers/dri/i810/i810screen.c
index f64c10a9ae..f8cf050d7e 100644
--- a/src/mesa/drivers/dri/i810/i810screen.c
+++ b/src/mesa/drivers/dri/i810/i810screen.c
@@ -288,8 +288,8 @@ i810InitDriver(__DRIscreenPrivate *sPriv)
 	      i810Screen->depth.handle,
 	      i810Screen->depth.size,
 	      (drmAddress *)&i810Screen->depth.map) != 0) {
-      FREE(i810Screen);
       drmUnmap(i810Screen->back.map, i810Screen->back.size);
+      FREE(i810Screen);
       sPriv->private = NULL;
       __driUtilMessage("i810InitDriver: drmMap (2) failed");
       return GL_FALSE;
diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c
index 3ad25282d9..e0d5b2b487 100644
--- a/src/mesa/drivers/dri/i810/i810state.c
+++ b/src/mesa/drivers/dri/i810/i810state.c
@@ -380,7 +380,10 @@ static void i810CullFaceFrontFace(GLcontext *ctx, GLenum unused)
 static void i810LineWidth( GLcontext *ctx, GLfloat widthf )
 {
    i810ContextPtr imesa = I810_CONTEXT( ctx );
-   int width = (int)ctx->Line._Width;
+   /* AA, non-AA limits are same */
+   const int width = (int) CLAMP(ctx->Line.Width,
+                                 ctx->Const.MinLineWidth,
+                                 ctx->Const.MaxLineWidth);
 
    imesa->LcsLineWidth = 0;
    if (width & 1) imesa->LcsLineWidth |= LCS_LINEWIDTH_1_0;
@@ -396,7 +399,10 @@ static void i810LineWidth( GLcontext *ctx, GLfloat widthf )
 static void i810PointSize( GLcontext *ctx, GLfloat sz )
 {
    i810ContextPtr imesa = I810_CONTEXT( ctx );
-   int size = (int)ctx->Point._Size;
+   /* AA, non-AA limits are same */
+   const int size = (int) CLAMP(ctx->Point.Size,
+                                ctx->Const.MinPointSize,
+                                ctx->Const.MaxPointSize);
 
    imesa->LcsPointSize = 0;
    if (size & 1) imesa->LcsPointSize |= LCS_LINEWIDTH_1_0;
diff --git a/src/mesa/drivers/dri/i810/i810tris.c b/src/mesa/drivers/dri/i810/i810tris.c
index 3e09427bb9..2c4ee06633 100644
--- a/src/mesa/drivers/dri/i810/i810tris.c
+++ b/src/mesa/drivers/dri/i810/i810tris.c
@@ -112,7 +112,9 @@ static __inline__ void i810_draw_quad( i810ContextPtr imesa,
 static __inline__ void i810_draw_point( i810ContextPtr imesa,
 					i810VertexPtr tmp )
 {
-   GLfloat sz = imesa->glCtx->Point._Size * .5;
+   GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                            imesa->glCtx->Const.MinPointSize,
+                            imesa->glCtx->Const.MaxPointSize);
    int vertsize = imesa->vertex_size;
    GLuint *vb = i810AllocDmaLow( imesa, 2 * 4 * vertsize );
    int j;
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index d40cf705a3..cf1d0b0c1e 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -59,6 +59,11 @@ do {									\
 #define VRTX_TEX_SET_FMT(n, x)          ((x)<<((n)*2))
 #define TEXBIND_SET(n, x) 		((x)<<((n)*4))
 
+ static void
+i830_render_prevalidate(struct intel_context *intel)
+{
+}
+
 static void i830_render_start( intelContextPtr intel )
 {
    GLcontext *ctx = &intel->ctx;
@@ -531,4 +536,5 @@ void i830InitVtbl( i830ContextPtr i830 )
    i830->intel.vtbl.update_texture_state = i830UpdateTextureState;
    i830->intel.vtbl.emit_flush = i830_emit_flush;
    i830->intel.vtbl.render_start = i830_render_start;
+   i830->intel.vtbl.render_prevalidate = i830_render_prevalidate;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 702b878828..abca0bbffe 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -304,7 +304,7 @@ static void upload_program( struct i915_fragment_program *p )
 			 A0_MUL,
 			 tmp, A0_DEST_CHANNEL_X, 0,
 			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
+			 i915_emit_const1f(p, 1.0/(M_PI)),
 			 0);
 
 	 i915_emit_arith( p, 
@@ -319,7 +319,7 @@ static void upload_program( struct i915_fragment_program *p )
 			 A0_MUL,
 			 tmp, A0_DEST_CHANNEL_X, 0,
 			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
+			 i915_emit_const1f(p, (M_PI)),
 			 0);
 
 	 /* 
@@ -645,7 +645,7 @@ static void upload_program( struct i915_fragment_program *p )
 			 A0_MUL,
 			 tmp, A0_DEST_CHANNEL_X, 0,
 			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
+			 i915_emit_const1f(p, 1.0/(M_PI)),
 			 0);
 
 	 i915_emit_arith( p, 
@@ -660,7 +660,7 @@ static void upload_program( struct i915_fragment_program *p )
 			 A0_MUL,
 			 tmp, A0_DEST_CHANNEL_X, 0,
 			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
+			 i915_emit_const1f(p, (M_PI)),
 			 0);
 
 	 /* 
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index a19d4b6584..2b806cc36e 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -60,6 +60,33 @@ static GLint step_offsets[6][2] = { {0,2},
 
 #define I915_TEX_UNIT_ENABLED(unit)		(1<<unit)
 
+static GLuint i915_compressed_alignment(GLint internal_fmt)
+{
+    GLuint alignment = 4;
+
+    switch (internal_fmt) {
+    case GL_COMPRESSED_RGB_FXT1_3DFX:
+    case GL_COMPRESSED_RGBA_FXT1_3DFX:
+        alignment = 8;
+        break;
+
+    default:
+        break;
+    }
+
+    return alignment;
+}
+
+static int align(int value, GLuint alignment)
+{
+    return (value + alignment - 1) & ~(alignment - 1);
+}
+
+static GLuint minify(GLuint d)
+{
+    return MAX2(1, d >> 1);
+}
+
 static void i915LayoutTextureImages( i915ContextPtr i915,
 				     struct gl_texture_object *tObj )
 {
@@ -161,8 +188,15 @@ static void i915LayoutTextureImages( i915ContextPtr i915,
       break;
    }
    default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
+      if (baseImage->IsCompressed) {
+          GLuint alignment = i915_compressed_alignment(baseImage->InternalFormat);
+          
+          pitch = align(tObj->Image[0][firstLevel]->Width, alignment) * t->intel.texelBytes;
+      } else {
+          pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
+          pitch = (pitch + 3) & ~3;
+      }
+
       t->intel.base.dirty_images[0] = ~0;
 
       for ( total_height = i = 0 ; i < numLevels ; i++ ) {
@@ -343,8 +377,23 @@ static void i945LayoutTextureImages( i915ContextPtr i915,
       break;
    }
    default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
+      if (baseImage->IsCompressed) {
+          GLuint alignment = i915_compressed_alignment(baseImage->InternalFormat);
+          
+          pitch = align(tObj->Image[0][firstLevel]->Width, alignment);
+          if (numLevels > 2) {
+              GLint width0 = align(minify(tObj->Image[0][firstLevel]->Width), alignment) +
+                  + align(minify(minify(tObj->Image[0][firstLevel]->Width)), alignment);
+
+              if (width0 > pitch)
+                  pitch = width0;
+          }
+          pitch = pitch * t->intel.texelBytes;
+      } else {
+          pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
+          pitch = (pitch + 3) & ~3;
+      }
+
       t->intel.base.dirty_images[0] = ~0;
       max_offset = 0;
 
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index b0e5f87fc7..4ad95cf4e3 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -41,7 +41,8 @@
 #include "i915_reg.h"
 #include "i915_context.h"
 
-static void i915_render_start( intelContextPtr intel )
+static void
+i915_render_prevalidate(struct intel_context *intel)
 {
    GLcontext *ctx = &intel->ctx;
    i915ContextPtr i915 = I915_CONTEXT(intel);
@@ -54,6 +55,10 @@ static void i915_render_start( intelContextPtr intel )
    }
 }
 
+static void i915_render_start( intelContextPtr intel )
+{
+}
+
 
 static void i915_reduced_primitive_state( intelContextPtr intel,
 					  GLenum rprim )
@@ -64,6 +69,7 @@ static void i915_reduced_primitive_state( intelContextPtr intel,
     st1 &= ~ST1_ENABLE;
 
     switch (rprim) {
+    case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
     case GL_TRIANGLES:
        if (intel->ctx.Polygon.StippleFlag &&
 	   intel->hw_stipple)
@@ -453,6 +459,7 @@ void i915InitVtbl( i915ContextPtr i915 )
    i915->intel.vtbl.lost_hardware = i915_lost_hardware;
    i915->intel.vtbl.reduced_primitive_state = i915_reduced_primitive_state;
    i915->intel.vtbl.render_start = i915_render_start;
+   i915->intel.vtbl.render_prevalidate = i915_render_prevalidate;
    i915->intel.vtbl.set_color_region = i915_set_color_region;
    i915->intel.vtbl.set_z_region = i915_set_z_region;
    i915->intel.vtbl.update_color_z_regions = i915_update_color_z_regions;
diff --git a/src/mesa/drivers/dri/i915/intel_context.h b/src/mesa/drivers/dri/i915/intel_context.h
index 3b50107d73..914533d11b 100644
--- a/src/mesa/drivers/dri/i915/intel_context.h
+++ b/src/mesa/drivers/dri/i915/intel_context.h
@@ -112,6 +112,7 @@ struct intel_context
       void (*update_texture_state)( intelContextPtr intel );
 
       void (*render_start)( intelContextPtr intel );
+      void (*render_prevalidate) (struct intel_context * intel);
       void (*set_color_region)( intelContextPtr intel, const intelRegion *reg );
       void (*set_z_region)( intelContextPtr intel, const intelRegion *reg );
       void (*update_color_z_regions)(intelContextPtr intel,
diff --git a/src/mesa/drivers/dri/i915/intel_pixel.c b/src/mesa/drivers/dri/i915/intel_pixel.c
index d175870a0c..c3030d42b0 100644
--- a/src/mesa/drivers/dri/i915/intel_pixel.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel.c
@@ -450,10 +450,16 @@ intelDrawPixels( GLcontext *ctx,
        * wise happily run the fragment program on each pixel in the image).
        */
       struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
+   /* can't just set current frag prog to 0 here as on buffer resize
+      we'll get new state checks which will segfault. Remains a hack. */
       ctx->FragmentProgram._Current = NULL;
+      ctx->FragmentProgram._UseTexEnvProgram = GL_FALSE;
+      ctx->FragmentProgram._Active = GL_FALSE;
       _swrast_DrawPixels( ctx, x, y, width, height, format, type,
                           unpack, pixels );
       ctx->FragmentProgram._Current = fpSave;
+      ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
+      ctx->FragmentProgram._Active = GL_TRUE;
    }
    else {
       _swrast_DrawPixels( ctx, x, y, width, height, format, type,
diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index d9438ba0fd..7bc02ba807 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -199,6 +199,8 @@ static GLboolean intel_run_render( GLcontext *ctx,
    struct vertex_buffer *VB = &tnl->vb;
    GLuint i;
 
+   intel->vtbl.render_prevalidate( intel );
+
    /* Don't handle clipping or indexed vertices.
     */
    if (intel->RenderIndex != 0 || 
diff --git a/src/mesa/drivers/dri/i915/intel_tex.c b/src/mesa/drivers/dri/i915/intel_tex.c
index 5bd280652a..978945f26b 100644
--- a/src/mesa/drivers/dri/i915/intel_tex.c
+++ b/src/mesa/drivers/dri/i915/intel_tex.c
@@ -643,17 +643,19 @@ static void intelUploadTexImage( intelContextPtr intel,
       switch (image->InternalFormat) {
 	case GL_COMPRESSED_RGB_FXT1_3DFX:
 	case GL_COMPRESSED_RGBA_FXT1_3DFX:
+          row_len = ((image->Width + 7) & ~7) * 2;  
+          break;
 	case GL_RGB_S3TC:
 	case GL_RGB4_S3TC:
 	case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
 	case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-	  row_len = (image->Width * 2 + 7) & ~7;
+	  row_len = ((image->Width + 3) & ~3) * 2;  
 	  break;
 	case GL_RGBA_S3TC:
 	case GL_RGBA4_S3TC:
 	case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
 	case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-	  row_len = (image->Width * 4 + 15) & ~15;
+	  row_len = ((image->Width + 3) & ~3) * 4;  
 	  break;
 	default:
 	  fprintf(stderr,"Internal Compressed format not supported %d\n", image->InternalFormat);
diff --git a/src/mesa/drivers/dri/i915tex/Makefile b/src/mesa/drivers/dri/i915tex/Makefile
index b218929dce..445327f243 100644
--- a/src/mesa/drivers/dri/i915tex/Makefile
+++ b/src/mesa/drivers/dri/i915tex/Makefile
@@ -37,6 +37,7 @@ DRIVER_SOURCES = \
 	i915_texstate.c \
 	i915_context.c \
 	i915_debug.c \
+	i915_debug_fp.c \
 	i915_fragprog.c \
 	i915_metaops.c \
 	i915_program.c \
diff --git a/src/mesa/drivers/dri/i915tex/i830_vtbl.c b/src/mesa/drivers/dri/i915tex/i830_vtbl.c
index 20c6900025..d78d1f2adf 100644
--- a/src/mesa/drivers/dri/i915tex/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915tex/i830_vtbl.c
@@ -524,6 +524,16 @@ i830_emit_state(struct intel_context *intel)
 static void
 i830_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   for (i = 0; i < I830_TEX_UNITS; i++) {
+      if (i830->state.tex_buffer[i] != NULL) {
+	 driBOUnReference(i830->state.tex_buffer[i]);
+	 i830->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
diff --git a/src/mesa/drivers/dri/i915tex/i915_debug.c b/src/mesa/drivers/dri/i915tex/i915_debug.c
index 974527e14c..c0e1242a0e 100644
--- a/src/mesa/drivers/dri/i915tex/i915_debug.c
+++ b/src/mesa/drivers/dri/i915tex/i915_debug.c
@@ -25,310 +25,819 @@
  * 
  **************************************************************************/
 
+#include "imports.h"
+
 #include "i915_reg.h"
 #include "i915_context.h"
-#include <stdio.h>
-
-
-static const char *opcodes[0x20] = {
-   "NOP",
-   "ADD",
-   "MOV",
-   "MUL",
-   "MAD",
-   "DP2ADD",
-   "DP3",
-   "DP4",
-   "FRC",
-   "RCP",
-   "RSQ",
-   "EXP",
-   "LOG",
-   "CMP",
-   "MIN",
-   "MAX",
-   "FLR",
-   "MOD",
-   "TRC",
-   "SGE",
-   "SLT",
-   "TEXLD",
-   "TEXLDP",
-   "TEXLDB",
-   "TEXKILL",
-   "DCL",
-   "0x1a",
-   "0x1b",
-   "0x1c",
-   "0x1d",
-   "0x1e",
-   "0x1f",
-};
-
-
-static const int args[0x20] = {
-   0,                           /* 0 nop */
-   2,                           /* 1 add */
-   1,                           /* 2 mov */
-   2,                           /* 3 m ul */
-   3,                           /* 4 mad */
-   3,                           /* 5 dp2add */
-   2,                           /* 6 dp3 */
-   2,                           /* 7 dp4 */
-   1,                           /* 8 frc */
-   1,                           /* 9 rcp */
-   1,                           /* a rsq */
-   1,                           /* b exp */
-   1,                           /* c log */
-   3,                           /* d cmp */
-   2,                           /* e min */
-   2,                           /* f max */
-   1,                           /* 10 flr */
-   1,                           /* 11 mod */
-   1,                           /* 12 trc */
-   2,                           /* 13 sge */
-   2,                           /* 14 slt */
-   1,
-   1,
-   1,
-   1,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-};
-
-
-static const char *regname[0x8] = {
-   "R",
-   "T",
-   "CONST",
-   "S",
-   "OC",
-   "OD",
-   "U",
-   "UNKNOWN",
-};
-
-static void
-print_reg_type_nr(GLuint type, GLuint nr)
+#include "i915_debug.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
+
+static GLboolean debug( struct debug_stream *stream, const char *name, GLuint len )
 {
-   switch (type) {
-   case REG_TYPE_T:
-      switch (nr) {
-      case T_DIFFUSE:
-         fprintf(stderr, "T_DIFFUSE");
-         return;
-      case T_SPECULAR:
-         fprintf(stderr, "T_SPECULAR");
-         return;
-      case T_FOG_W:
-         fprintf(stderr, "T_FOG_W");
-         return;
-      default:
-         fprintf(stderr, "T_TEX%d", nr);
-         return;
-      }
-   case REG_TYPE_OC:
-      if (nr == 0) {
-         fprintf(stderr, "oC");
-         return;
-      }
-      break;
-   case REG_TYPE_OD:
-      if (nr == 0) {
-         fprintf(stderr, "oD");
-         return;
-      }
-      break;
-   default:
-      break;
+   GLuint i;
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
    }
 
-   fprintf(stderr, "%s[%d]", regname[type], nr);
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);   
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+
+
+static const char *get_prim_name( GLuint val )
+{
+   switch (val & PRIM3D_MASK) {
+   case PRIM3D_TRILIST: return "TRILIST"; break;
+   case PRIM3D_TRISTRIP: return "TRISTRIP"; break;
+   case PRIM3D_TRISTRIP_RVRSE: return "TRISTRIP_RVRSE"; break;
+   case PRIM3D_TRIFAN: return "TRIFAN"; break;
+   case PRIM3D_POLY: return "POLY"; break;
+   case PRIM3D_LINELIST: return "LINELIST"; break;
+   case PRIM3D_LINESTRIP: return "LINESTRIP"; break;
+   case PRIM3D_RECTLIST: return "RECTLIST"; break;
+   case PRIM3D_POINTLIST: return "POINTLIST"; break;
+   case PRIM3D_DIB: return "DIB"; break;
+   case PRIM3D_CLEAR_RECT: return "CLEAR_RECT"; break;
+   case PRIM3D_ZONE_INIT: return "ZONE_INIT"; break;
+   default: return "????"; break;
+   }
 }
 
-#define REG_SWIZZLE_MASK 0x7777
-#define REG_NEGATE_MASK 0x8888
+static GLboolean debug_prim( struct debug_stream *stream, const char *name, 
+			     GLboolean dump_floats,
+			     GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i;
+   
 
-#define REG_SWIZZLE_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
-		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
+
+   PRINTF("%s %s (%d dwords):\n", name, prim, len);
+   PRINTF("\t0x%08x\n",  ptr[0]);   
+   for (i = 1; i < len; i++) {
+      if (dump_floats)
+	 PRINTF("\t0x%08x // %f\n",  ptr[i], *(GLfloat *)&ptr[i]);   
+      else
+	 PRINTF("\t0x%08x\n",  ptr[i]);   
+   }
+
+      
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+   
 
 
-static void
-print_reg_neg_swizzle(GLuint reg)
+
+static GLboolean debug_program( struct debug_stream *stream, const char *name, GLuint len )
 {
-   int i;
-
-   if ((reg & REG_SWIZZLE_MASK) == REG_SWIZZLE_XYZW &&
-       (reg & REG_NEGATE_MASK) == 0)
-      return;
-
-   fprintf(stderr, ".");
-
-   for (i = 3; i >= 0; i--) {
-      if (reg & (1 << ((i * 4) + 3)))
-         fprintf(stderr, "-");
-
-      switch ((reg >> (i * 4)) & 0x7) {
-      case 0:
-         fprintf(stderr, "x");
-         break;
-      case 1:
-         fprintf(stderr, "y");
-         break;
-      case 2:
-         fprintf(stderr, "z");
-         break;
-      case 3:
-         fprintf(stderr, "w");
-         break;
-      case 4:
-         fprintf(stderr, "0");
-         break;
-      case 5:
-         fprintf(stderr, "1");
-         break;
-      default:
-         fprintf(stderr, "?");
-         break;
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
+   }
+
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   i915_disassemble_program( ptr, len );
+
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
+}
+
+
+static GLboolean debug_chain( struct debug_stream *stream, const char *name, GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint old_offset = stream->offset + len * sizeof(GLuint);
+   GLuint i;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+
+   stream->offset = ptr[1] & ~0x3;
+   
+   if (stream->offset < old_offset)
+      PRINTF("\n... skipping backwards from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+   else
+      PRINTF("\n... skipping from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+
+
+   return GL_TRUE;
+}
+
+
+static GLboolean debug_variable_length_prim( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i, len;
+
+   GLushort *idx = (GLushort *)(ptr+1);
+   for (i = 0; idx[i] != 0xffff; i++)
+      ;
+
+   len = 1+(i+2)/2;
+
+   PRINTF("3DPRIM, %s variable length %d indicies (%d dwords):\n", prim, i, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
+}
+
+
+#define BITS( dw, hi, lo, ... )				\
+do {							\
+   unsigned himask = ~0UL >> (31 - (hi));		\
+   PRINTF("\t\t ");				\
+   PRINTF(__VA_ARGS__);			\
+   PRINTF(": 0x%x\n", ((dw) & himask) >> (lo));	\
+} while (0)
+
+#define MBZ( dw, hi, lo) do {							\
+   unsigned x = (dw) >> (lo);				\
+   unsigned lomask = (1 << (lo)) - 1;			\
+   unsigned himask;					\
+   himask = (1UL << (hi)) - 1;				\
+   assert ((x & himask & ~lomask) == 0);	\
+} while (0)
+
+#define FLAG( dw, bit, ... )			\
+do {							\
+   if (((dw) >> (bit)) & 1) {				\
+      PRINTF("\t\t ");				\
+      PRINTF(__VA_ARGS__);			\
+      PRINTF("\n");				\
+   }							\
+} while (0)
+
+static GLboolean debug_load_immediate( struct debug_stream *stream,
+				       const char *name,
+				       GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 4) & 0xff;
+   GLuint j = 0;
+   
+   PRINTF("%s (%d dwords, flags: %x):\n", name, len, bits);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   if (bits & (1<<0)) {
+      PRINTF("\t  LIS0: 0x%08x\n", ptr[j]);
+      PRINTF("\t vb address: 0x%08x\n", (ptr[j] & ~0x3));
+      BITS(ptr[j], 0, 0, "vb invalidate disable");
+      j++;
+   }
+   if (bits & (1<<1)) {
+      PRINTF("\t  LIS1: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 29, 24, "vb dword width");
+      BITS(ptr[j], 21, 16, "vb dword pitch");
+      BITS(ptr[j], 15, 0, "vb max index");
+      j++;
+   }
+   if (bits & (1<<2)) {
+      int i;
+      PRINTF("\t  LIS2: 0x%08x\n", ptr[j]);
+      for (i = 0; i < 8; i++) {
+	 unsigned tc = (ptr[j] >> (i * 4)) & 0xf;
+	 if (tc != 0xf)
+	    BITS(tc, 3, 0, "tex coord %d", i);
+      }
+      j++;
+   }
+   if (bits & (1<<3)) {
+      PRINTF("\t  LIS3: 0x%08x\n", ptr[j]);
+      j++;
+   }
+   if (bits & (1<<4)) {
+      PRINTF("\t  LIS4: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 23, "point width");
+      BITS(ptr[j], 22, 19, "line width");
+      FLAG(ptr[j], 18, "alpha flatshade");
+      FLAG(ptr[j], 17, "fog flatshade");
+      FLAG(ptr[j], 16, "spec flatshade");
+      FLAG(ptr[j], 15, "rgb flatshade");
+      BITS(ptr[j], 14, 13, "cull mode");
+      FLAG(ptr[j], 12, "vfmt: point width");
+      FLAG(ptr[j], 11, "vfmt: specular/fog");
+      FLAG(ptr[j], 10, "vfmt: rgba");
+      FLAG(ptr[j], 9, "vfmt: depth offset");
+      BITS(ptr[j], 8, 6, "vfmt: position (2==xyzw)");
+      FLAG(ptr[j], 5, "force dflt diffuse");
+      FLAG(ptr[j], 4, "force dflt specular");
+      FLAG(ptr[j], 3, "local depth offset enable");
+      FLAG(ptr[j], 2, "vfmt: fp32 fog coord");
+      FLAG(ptr[j], 1, "sprite point");
+      FLAG(ptr[j], 0, "antialiasing");
+      j++;
+   }
+   if (bits & (1<<5)) {
+      PRINTF("\t  LIS5: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 28, "rgba write disables");
+      FLAG(ptr[j], 27,     "force dflt point width");
+      FLAG(ptr[j], 26,     "last pixel enable");
+      FLAG(ptr[j], 25,     "global z offset enable");
+      FLAG(ptr[j], 24,     "fog enable");
+      BITS(ptr[j], 23, 16, "stencil ref");
+      BITS(ptr[j], 15, 13, "stencil test");
+      BITS(ptr[j], 12, 10, "stencil fail op");
+      BITS(ptr[j], 9, 7,   "stencil pass z fail op");
+      BITS(ptr[j], 6, 4,   "stencil pass z pass op");
+      FLAG(ptr[j], 3,      "stencil write enable");
+      FLAG(ptr[j], 2,      "stencil test enable");
+      FLAG(ptr[j], 1,      "color dither enable");
+      FLAG(ptr[j], 0,      "logiop enable");
+      j++;
+   }
+   if (bits & (1<<6)) {
+      PRINTF("\t  LIS6: 0x%08x\n", ptr[j]);
+      FLAG(ptr[j], 31,      "alpha test enable");
+      BITS(ptr[j], 30, 28,  "alpha func");
+      BITS(ptr[j], 27, 20,  "alpha ref");
+      FLAG(ptr[j], 19,      "depth test enable");
+      BITS(ptr[j], 18, 16,  "depth func");
+      FLAG(ptr[j], 15,      "blend enable");
+      BITS(ptr[j], 14, 12,  "blend func");
+      BITS(ptr[j], 11, 8,   "blend src factor");
+      BITS(ptr[j], 7,  4,   "blend dst factor");
+      FLAG(ptr[j], 3,       "depth write enable");
+      FLAG(ptr[j], 2,       "color write enable");
+      BITS(ptr[j], 1,  0,   "provoking vertex"); 
+      j++;
+   }
+
+
+   PRINTF("\n");
+
+   assert(j == len);
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 
+
+
+static GLboolean debug_load_indirect( struct debug_stream *stream,
+				      const char *name,
+				      GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 8) & 0x3f;
+   GLuint i, j = 0;
+   
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   for (i = 0; i < 6; i++) {
+      if (bits & (1<<i)) {
+	 switch (1<<(8+i)) {
+	 case LI0_STATE_STATIC_INDIRECT:
+	    PRINTF("        STATIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_DYNAMIC_INDIRECT:
+	    PRINTF("       DYNAMIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    break;
+	 case LI0_STATE_SAMPLER:
+	    PRINTF("       SAMPLER: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_MAP:
+	    PRINTF("           MAP: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_PROGRAM:
+	    PRINTF("       PROGRAM: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_CONSTANTS:
+	    PRINTF("     CONSTANTS: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 default:
+	    assert(0);
+	    break;
+	 }
       }
    }
+
+   if (bits == 0) {
+      PRINTF("\t  DUMMY: 0x%08x\n", ptr[j++]);
+   }
+
+   PRINTF("\n");
+
+
+   assert(j == len);
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 	
+static void BR13( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   FLAG(val, 30, "clipping enable");
+   BITS(val, 25, 24, "color depth (3==32bpp)");
+   BITS(val, 23, 16, "raster op");
+   BITS(val, 15, 0,  "dest pitch");
 }
 
 
-static void
-print_src_reg(GLuint dword)
+static void BR22( struct debug_stream *stream,
+		  GLuint val )
 {
-   GLuint nr = (dword >> A2_SRC2_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A2_SRC2_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr(type, nr);
-   print_reg_neg_swizzle(dword);
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "dest y1");
+   BITS(val, 15, 0,  "dest x1");
 }
 
-void
-i915_print_ureg(const char *msg, GLuint ureg)
+static void BR23( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "dest y2");
+   BITS(val, 15, 0,  "dest x2");
+}
+
+static void BR09( struct debug_stream *stream,
+		  GLuint val )
 {
-   fprintf(stderr, "%s: ", msg);
-   print_src_reg(ureg >> 8);
-   fprintf(stderr, "\n");
+   PRINTF("\t0x%08x -- dest address\n",  val);
 }
 
-static void
-print_dest_reg(GLuint dword)
+static void BR26( struct debug_stream *stream,
+		  GLuint val )
 {
-   GLuint nr = (dword >> A0_DEST_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A0_DEST_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr(type, nr);
-   if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
-      return;
-   fprintf(stderr, ".");
-   if (dword & A0_DEST_CHANNEL_X)
-      fprintf(stderr, "x");
-   if (dword & A0_DEST_CHANNEL_Y)
-      fprintf(stderr, "y");
-   if (dword & A0_DEST_CHANNEL_Z)
-      fprintf(stderr, "z");
-   if (dword & A0_DEST_CHANNEL_W)
-      fprintf(stderr, "w");
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "src y1");
+   BITS(val, 15, 0,  "src x1");
 }
 
+static void BR11( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 15, 0,  "src pitch");
+}
 
-#define GET_SRC0_REG(r0, r1) ((r0<<14)|(r1>>A1_SRC0_CHANNEL_W_SHIFT))
-#define GET_SRC1_REG(r0, r1) ((r0<<8)|(r1>>A2_SRC1_CHANNEL_W_SHIFT))
-#define GET_SRC2_REG(r)      (r)
+static void BR12( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- src address\n",  val);
+}
 
+static void BR16( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- color\n",  val);
+}
+   
+static GLboolean debug_copy_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   BR13(stream, ptr[j++]);
+   BR22(stream, ptr[j++]);
+   BR23(stream, ptr[j++]);
+   BR09(stream, ptr[j++]);
+   BR26(stream, ptr[j++]);
+   BR11(stream, ptr[j++]);
+   BR12(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
 
-static void
-print_arith_op(GLuint opcode, const GLuint * program)
+static GLboolean debug_color_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
 {
-   if (opcode != A0_NOP) {
-      print_dest_reg(program[0]);
-      if (program[0] & A0_DEST_SATURATE)
-         fprintf(stderr, " = SATURATE ");
-      else
-         fprintf(stderr, " = ");
-   }
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
 
-   print_src_reg(GET_SRC0_REG(program[0], program[1]));
-   if (args[opcode] == 1) {
-      fprintf(stderr, "\n");
-      return;
+   BR13(stream, ptr[j++]);
+   BR22(stream, ptr[j++]);
+   BR23(stream, ptr[j++]);
+   BR09(stream, ptr[j++]);
+   BR16(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_modes4( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j]);
+   BITS(ptr[j], 21, 18, "logicop func");
+   FLAG(ptr[j], 17, "stencil test mask modify-enable");
+   FLAG(ptr[j], 16, "stencil write mask modify-enable");
+   BITS(ptr[j], 15, 8, "stencil test mask");
+   BITS(ptr[j], 7, 0,  "stencil write mask");
+   j++;
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_map_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "map mask");
+      j++;
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC1_REG(program[1], program[2]));
-   if (args[opcode] == 2) {
-      fprintf(stderr, "\n");
-      return;
+   while (j < len) {
+      {
+	 PRINTF("\t  TMn.0: 0x%08x\n", ptr[j]);
+	 PRINTF("\t map address: 0x%08x\n", (ptr[j] & ~0x3));
+	 FLAG(ptr[j], 1, "vertical line stride");
+	 FLAG(ptr[j], 0, "vertical line stride offset");
+	 j++;
+      }
+
+      {
+	 PRINTF("\t  TMn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "height");
+	 BITS(ptr[j], 20, 10, "width");
+	 BITS(ptr[j], 9, 7, "surface format");
+	 BITS(ptr[j], 6, 3, "texel format");
+	 FLAG(ptr[j], 2, "use fence regs");
+	 FLAG(ptr[j], 1, "tiled surface");
+	 FLAG(ptr[j], 0, "tile walk ymajor");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TMn.2: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "dword pitch");
+	 BITS(ptr[j], 20, 15, "cube face enables");
+	 BITS(ptr[j], 14, 9, "max lod");
+	 FLAG(ptr[j], 8,     "mip layout right");
+	 BITS(ptr[j], 7, 0, "depth");
+	 j++;
+      }
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC2_REG(program[2]));
-   fprintf(stderr, "\n");
-   return;
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
 }
 
-
-static void
-print_tex_op(GLuint opcode, const GLuint * program)
+static GLboolean debug_sampler_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
 {
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, " = ");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "sampler mask");
+      j++;
+   }
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   while (j < len) {
+      {
+	 PRINTF("\t  TSn.0: 0x%08x\n", ptr[j]);
+	 FLAG(ptr[j], 31, "reverse gamma");
+	 FLAG(ptr[j], 30, "planar to packed");
+	 FLAG(ptr[j], 29, "yuv->rgb");
+	 BITS(ptr[j], 28, 27, "chromakey index");
+	 BITS(ptr[j], 26, 22, "base mip level");
+	 BITS(ptr[j], 21, 20, "mip mode filter");
+	 BITS(ptr[j], 19, 17, "mag mode filter");
+	 BITS(ptr[j], 16, 14, "min mode filter");
+	 BITS(ptr[j], 13, 5,  "lod bias (s4.4)");
+	 FLAG(ptr[j], 4,      "shadow enable");
+	 FLAG(ptr[j], 3,      "max-aniso-4");
+	 BITS(ptr[j], 2, 0,   "shadow func");
+	 j++;
+      }
 
-   fprintf(stderr, "S[%d],", program[0] & T0_SAMPLER_NR_MASK);
+      {
+	 PRINTF("\t  TSn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 24, "min lod");
+	 MBZ( ptr[j], 23, 18 );
+	 FLAG(ptr[j], 17,     "kill pixel enable");
+	 FLAG(ptr[j], 16,     "keyed tex filter mode");
+	 FLAG(ptr[j], 15,     "chromakey enable");
+	 BITS(ptr[j], 14, 12, "tcx wrap mode");
+	 BITS(ptr[j], 11, 9,  "tcy wrap mode");
+	 BITS(ptr[j], 8,  6,  "tcz wrap mode");
+	 FLAG(ptr[j], 5,      "normalized coords");
+	 BITS(ptr[j], 4,  1,  "map (surface) index");
+	 FLAG(ptr[j], 0,      "EAST deinterlacer enable");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TSn.2: 0x%08x  (default color)\n", ptr[j]);
+	 j++;
+      }
+   }
 
-   print_reg_type_nr((program[1] >> T1_ADDRESS_REG_TYPE_SHIFT) &
-                     REG_TYPE_MASK,
-                     (program[1] >> T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK);
-   fprintf(stderr, "\n");
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
 }
 
-static void
-print_dcl_op(GLuint opcode, const GLuint * program)
+static GLboolean debug_dest_vars( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
 {
-   fprintf(stderr, "%s ", opcodes[opcode]);
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, "\n");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      FLAG(ptr[j], 31,     "early classic ztest");
+      FLAG(ptr[j], 30,     "opengl tex default color");
+      FLAG(ptr[j], 29,     "bypass iz");
+      FLAG(ptr[j], 28,     "lod preclamp");
+      BITS(ptr[j], 27, 26, "dither pattern");
+      FLAG(ptr[j], 25,     "linear gamma blend");
+      FLAG(ptr[j], 24,     "debug dither");
+      BITS(ptr[j], 23, 20, "dstorg x");
+      BITS(ptr[j], 19, 16, "dstorg y");
+      MBZ (ptr[j], 15, 15 );
+      BITS(ptr[j], 14, 12, "422 write select");
+      BITS(ptr[j], 11, 8,  "cbuf format");
+      BITS(ptr[j], 3, 2,   "zbuf format");
+      FLAG(ptr[j], 1,      "vert line stride");
+      FLAG(ptr[j], 1,      "vert line stride offset");
+      j++;
+   }
+   
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
 }
 
+static GLboolean debug_buf_info( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 28, 28, "aux buffer id");
+      BITS(ptr[j], 27, 24, "buffer id (7=depth, 3=back)");
+      FLAG(ptr[j], 23,     "use fence regs");
+      FLAG(ptr[j], 22,     "tiled surface");
+      FLAG(ptr[j], 21,     "tile walk ymajor");
+      MBZ (ptr[j], 20, 14);
+      BITS(ptr[j], 13, 2,  "dword pitch");
+      MBZ (ptr[j], 2,  0);
+      j++;
+   }
+   
+   PRINTF("\t0x%08x -- buffer base address\n",  ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean i915_debug_packet( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint cmd = *ptr;
+   
+   switch (((cmd >> 29) & 0x7)) {
+   case 0x0:
+      switch ((cmd >> 23) & 0x3f) {
+      case 0x0:
+	 return debug(stream, "MI_NOOP", 1);
+      case 0x3:
+	 return debug(stream, "MI_WAIT_FOR_EVENT", 1);
+      case 0x4:
+	 return debug(stream, "MI_FLUSH", 1);
+      case 0xA:
+	 debug(stream, "MI_BATCH_BUFFER_END", 1);
+	 return GL_FALSE;
+      case 0x22:
+	 return debug(stream, "MI_LOAD_REGISTER_IMM", 3);
+      case 0x31:
+	 return debug_chain(stream, "MI_BATCH_BUFFER_START", 2);
+      default:
+	 break;
+      }
+      break;
+   case 0x1:
+      break;
+   case 0x2:
+      switch ((cmd >> 22) & 0xff) {	 
+      case 0x50:
+	 return debug_color_blit(stream, "XY_COLOR_BLT", (cmd & 0xff) + 2);
+      case 0x53:
+	 return debug_copy_blit(stream, "XY_SRC_COPY_BLT", (cmd & 0xff) + 2);
+      default:
+	 return debug(stream, "blit command", (cmd & 0xff) + 2);
+      }
+      break;
+   case 0x3:
+      switch ((cmd >> 24) & 0x1f) {	 
+      case 0x6:
+	 return debug(stream, "3DSTATE_ANTI_ALIASING", 1);
+      case 0x7:
+	 return debug(stream, "3DSTATE_RASTERIZATION_RULES", 1);
+      case 0x8:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_OPS", 2);
+      case 0x9:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_MASKS", 1);
+      case 0xb:
+	 return debug(stream, "3DSTATE_INDEPENDENT_ALPHA_BLEND", 1);
+      case 0xc:
+	 return debug(stream, "3DSTATE_MODES5", 1);	 
+      case 0xd:
+	 return debug_modes4(stream, "3DSTATE_MODES4", 1);
+      case 0x15:
+	 return debug(stream, "3DSTATE_FOG_COLOR", 1);
+      case 0x16:
+	 return debug(stream, "3DSTATE_COORD_SET_BINDINGS", 1);
+      case 0x1c:
+	 /* 3DState16NP */
+	 switch((cmd >> 19) & 0x1f) {
+	 case 0x10:
+	    return debug(stream, "3DSTATE_SCISSOR_ENABLE", 1);
+	 case 0x11:
+	    return debug(stream, "3DSTATE_DEPTH_SUBRECTANGLE_DISABLE", 1);
+	 default:
+	    break;
+	 }
+	 break;
+      case 0x1d:
+	 /* 3DStateMW */
+	 switch ((cmd >> 16) & 0xff) {
+	 case 0x0:
+	    return debug_map_state(stream, "3DSTATE_MAP_STATE", (cmd & 0x1f) + 2);
+	 case 0x1:
+	    return debug_sampler_state(stream, "3DSTATE_SAMPLER_STATE", (cmd & 0x1f) + 2);
+	 case 0x4:
+	    return debug_load_immediate(stream, "3DSTATE_LOAD_STATE_IMMEDIATE", (cmd & 0xf) + 2);
+	 case 0x5:
+	    return debug_program(stream, "3DSTATE_PIXEL_SHADER_PROGRAM", (cmd & 0x1ff) + 2);
+	 case 0x6:
+	    return debug(stream, "3DSTATE_PIXEL_SHADER_CONSTANTS", (cmd & 0xff) + 2);
+	 case 0x7:
+	    return debug_load_indirect(stream, "3DSTATE_LOAD_INDIRECT", (cmd & 0xff) + 2);
+	 case 0x80:
+	    return debug(stream, "3DSTATE_DRAWING_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x81:
+	    return debug(stream, "3DSTATE_SCISSOR_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x83:
+	    return debug(stream, "3DSTATE_SPAN_STIPPLE", (cmd & 0xffff) + 2);
+	 case 0x85:
+	    return debug_dest_vars(stream, "3DSTATE_DEST_BUFFER_VARS", (cmd & 0xffff) + 2);
+	 case 0x88:
+	    return debug(stream, "3DSTATE_CONSTANT_BLEND_COLOR", (cmd & 0xffff) + 2);
+	 case 0x89:
+	    return debug(stream, "3DSTATE_FOG_MODE", (cmd & 0xffff) + 2);
+	 case 0x8e:
+	    return debug_buf_info(stream, "3DSTATE_BUFFER_INFO", (cmd & 0xffff) + 2);
+	 case 0x97:
+	    return debug(stream, "3DSTATE_DEPTH_OFFSET_SCALE", (cmd & 0xffff) + 2);
+	 case 0x98:
+	    return debug(stream, "3DSTATE_DEFAULT_Z", (cmd & 0xffff) + 2);
+	 case 0x99:
+	    return debug(stream, "3DSTATE_DEFAULT_DIFFUSE", (cmd & 0xffff) + 2);
+	 case 0x9a:
+	    return debug(stream, "3DSTATE_DEFAULT_SPECULAR", (cmd & 0xffff) + 2);
+	 case 0x9c:
+	    return debug(stream, "3DSTATE_CLEAR_PARAMETERS", (cmd & 0xffff) + 2);
+	 default:
+	    assert(0);
+	    return 0;
+	 }
+	 break;
+      case 0x1e:
+	 if (cmd & (1 << 23))
+	    return debug(stream, "???", (cmd & 0xffff) + 1);
+	 else
+	    return debug(stream, "", 1);
+	 break;
+      case 0x1f:
+	 if ((cmd & (1 << 23)) == 0)	
+	    return debug_prim(stream, "3DPRIM (inline)", 1, (cmd & 0x1ffff) + 2);
+	 else if (cmd & (1 << 17)) 
+	 {
+	    if ((cmd & 0xffff) == 0)
+	       return debug_variable_length_prim(stream);
+	    else
+	       return debug_prim(stream, "3DPRIM (indexed)", 0, (((cmd & 0xffff) + 1) / 2) + 1);
+	 }
+	 else
+	    return debug_prim(stream, "3DPRIM  (indirect sequential)", 0, 2); 
+	 break;
+      default:
+	 return debug(stream, "", 0);
+      }
+   default:
+      assert(0);
+      return 0;
+   }
+
+   assert(0);
+   return 0;
+}
+
+
 
 void
-i915_disassemble_program(const GLuint * program, GLuint sz)
+i915_dump_batchbuffer( GLuint *start,
+		       GLuint *end )
 {
-   GLuint size = program[0] & 0x1ff;
-   GLint i;
+   struct debug_stream stream;
+   GLuint bytes = (end - start) * 4;
+   GLboolean done = GL_FALSE;
 
-   fprintf(stderr, "BEGIN\n");
+   PRINTF("\n\nBATCH: (%d)\n", bytes / 4);
 
-   if (size + 2 != sz) {
-      fprintf(stderr, "%s: program size mismatch %d/%d\n", __FUNCTION__,
-              size + 2, sz);
-      exit(1);
-   }
+   stream.offset = 0;
+   stream.ptr = (char *)start;
+   stream.print_addresses = 0;
 
-   program++;
-   for (i = 1; i < sz; i += 3, program += 3) {
-      GLuint opcode = program[0] & (0x1f << 24);
+   while (!done &&
+	  stream.offset < bytes &&
+	  stream.offset >= 0)
+   {
+      if (!i915_debug_packet( &stream ))
+	 break;
 
-      if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
-         print_arith_op(opcode >> 24, program);
-      else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
-         print_tex_op(opcode >> 24, program);
-      else if (opcode == D0_DCL)
-         print_dcl_op(opcode >> 24, program);
-      else
-         fprintf(stderr, "Unknown opcode 0x%x\n", opcode);
+      assert(stream.offset <= bytes &&
+	     stream.offset >= 0);
    }
 
-   fprintf(stderr, "END\n\n");
+   PRINTF("END-BATCH\n\n\n");
 }
+
+
diff --git a/src/mesa/drivers/dri/i915tex/i915_debug.h b/src/mesa/drivers/dri/i915tex/i915_debug.h
new file mode 100644
index 0000000000..0643a8c631
--- /dev/null
+++ b/src/mesa/drivers/dri/i915tex/i915_debug.h
@@ -0,0 +1,55 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * 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 TUNGSTEN GRAPHICS 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.
+ * 
+ **************************************************************************/
+
+/* Authors:  Keith Whitwell <keith@tungstengraphics.com>
+ */
+
+#ifndef I915_DEBUG_H
+#define I915_DEBUG_H
+
+struct i915_context;
+
+struct debug_stream 
+{
+   unsigned offset;		/* current gtt offset */
+   char *ptr;		/* pointer to gtt offset zero */
+   char *end;		/* pointer to gtt offset zero */
+   unsigned print_addresses;
+};
+
+
+
+extern void i915_disassemble_program(const unsigned *program, unsigned sz);
+extern void i915_print_ureg(const char *msg, unsigned ureg);
+
+
+void
+i915_dump_batchbuffer( unsigned *start,
+		       unsigned *end );
+
+
+#endif
diff --git a/src/mesa/drivers/dri/i915tex/i915_debug_fp.c b/src/mesa/drivers/dri/i915tex/i915_debug_fp.c
new file mode 100644
index 0000000000..84347a01ef
--- /dev/null
+++ b/src/mesa/drivers/dri/i915tex/i915_debug_fp.c
@@ -0,0 +1,333 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * 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 TUNGSTEN GRAPHICS 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.
+ * 
+ **************************************************************************/
+
+#include <stdio.h>
+
+#include "i915_reg.h"
+#include "i915_debug.h"
+#include "main/imports.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_print.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
+
+static const char *opcodes[0x20] = {
+   "NOP",
+   "ADD",
+   "MOV",
+   "MUL",
+   "MAD",
+   "DP2ADD",
+   "DP3",
+   "DP4",
+   "FRC",
+   "RCP",
+   "RSQ",
+   "EXP",
+   "LOG",
+   "CMP",
+   "MIN",
+   "MAX",
+   "FLR",
+   "MOD",
+   "TRC",
+   "SGE",
+   "SLT",
+   "TEXLD",
+   "TEXLDP",
+   "TEXLDB",
+   "TEXKILL",
+   "DCL",
+   "0x1a",
+   "0x1b",
+   "0x1c",
+   "0x1d",
+   "0x1e",
+   "0x1f",
+};
+
+
+static const int args[0x20] = {
+   0,                           /* 0 nop */
+   2,                           /* 1 add */
+   1,                           /* 2 mov */
+   2,                           /* 3 m ul */
+   3,                           /* 4 mad */
+   3,                           /* 5 dp2add */
+   2,                           /* 6 dp3 */
+   2,                           /* 7 dp4 */
+   1,                           /* 8 frc */
+   1,                           /* 9 rcp */
+   1,                           /* a rsq */
+   1,                           /* b exp */
+   1,                           /* c log */
+   3,                           /* d cmp */
+   2,                           /* e min */
+   2,                           /* f max */
+   1,                           /* 10 flr */
+   1,                           /* 11 mod */
+   1,                           /* 12 trc */
+   2,                           /* 13 sge */
+   2,                           /* 14 slt */
+   1,
+   1,
+   1,
+   1,
+   0,
+   0,
+   0,
+   0,
+   0,
+   0,
+   0,
+};
+
+
+static const char *regname[0x8] = {
+   "R",
+   "T",
+   "CONST",
+   "S",
+   "OC",
+   "OD",
+   "U",
+   "UNKNOWN",
+};
+
+static void
+print_reg_type_nr(GLuint type, GLuint nr)
+{
+   switch (type) {
+   case REG_TYPE_T:
+      switch (nr) {
+      case T_DIFFUSE:
+         PRINTF("T_DIFFUSE");
+         return;
+      case T_SPECULAR:
+         PRINTF("T_SPECULAR");
+         return;
+      case T_FOG_W:
+         PRINTF("T_FOG_W");
+         return;
+      default:
+         PRINTF("T_TEX%d", nr);
+         return;
+      }
+   case REG_TYPE_OC:
+      if (nr == 0) {
+         PRINTF("oC");
+         return;
+      }
+      break;
+   case REG_TYPE_OD:
+      if (nr == 0) {
+         PRINTF("oD");
+         return;
+      }
+      break;
+   default:
+      break;
+   }
+
+   PRINTF("%s[%d]", regname[type], nr);
+}
+
+#define REG_SWIZZLE_MASK 0x7777
+#define REG_NEGATE_MASK 0x8888
+
+#define REG_SWIZZLE_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
+		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
+		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
+		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
+
+
+static void
+print_reg_neg_swizzle(GLuint reg)
+{
+   int i;
+
+   if ((reg & REG_SWIZZLE_MASK) == REG_SWIZZLE_XYZW &&
+       (reg & REG_NEGATE_MASK) == 0)
+      return;
+
+   PRINTF(".");
+
+   for (i = 3; i >= 0; i--) {
+      if (reg & (1 << ((i * 4) + 3)))
+         PRINTF("-");
+
+      switch ((reg >> (i * 4)) & 0x7) {
+      case 0:
+         PRINTF("x");
+         break;
+      case 1:
+         PRINTF("y");
+         break;
+      case 2:
+         PRINTF("z");
+         break;
+      case 3:
+         PRINTF("w");
+         break;
+      case 4:
+         PRINTF("0");
+         break;
+      case 5:
+         PRINTF("1");
+         break;
+      default:
+         PRINTF("?");
+         break;
+      }
+   }
+}
+
+
+static void
+print_src_reg(GLuint dword)
+{
+   GLuint nr = (dword >> A2_SRC2_NR_SHIFT) & REG_NR_MASK;
+   GLuint type = (dword >> A2_SRC2_TYPE_SHIFT) & REG_TYPE_MASK;
+   print_reg_type_nr(type, nr);
+   print_reg_neg_swizzle(dword);
+}
+
+
+static void
+print_dest_reg(GLuint dword)
+{
+   GLuint nr = (dword >> A0_DEST_NR_SHIFT) & REG_NR_MASK;
+   GLuint type = (dword >> A0_DEST_TYPE_SHIFT) & REG_TYPE_MASK;
+   print_reg_type_nr(type, nr);
+   if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
+      return;
+   PRINTF(".");
+   if (dword & A0_DEST_CHANNEL_X)
+      PRINTF("x");
+   if (dword & A0_DEST_CHANNEL_Y)
+      PRINTF("y");
+   if (dword & A0_DEST_CHANNEL_Z)
+      PRINTF("z");
+   if (dword & A0_DEST_CHANNEL_W)
+      PRINTF("w");
+}
+
+
+#define GET_SRC0_REG(r0, r1) ((r0<<14)|(r1>>A1_SRC0_CHANNEL_W_SHIFT))
+#define GET_SRC1_REG(r0, r1) ((r0<<8)|(r1>>A2_SRC1_CHANNEL_W_SHIFT))
+#define GET_SRC2_REG(r)      (r)
+
+
+static void
+print_arith_op(GLuint opcode, const GLuint * program)
+{
+   if (opcode != A0_NOP) {
+      print_dest_reg(program[0]);
+      if (program[0] & A0_DEST_SATURATE)
+         PRINTF(" = SATURATE ");
+      else
+         PRINTF(" = ");
+   }
+
+   PRINTF("%s ", opcodes[opcode]);
+
+   print_src_reg(GET_SRC0_REG(program[0], program[1]));
+   if (args[opcode] == 1) {
+      PRINTF("\n");
+      return;
+   }
+
+   PRINTF(", ");
+   print_src_reg(GET_SRC1_REG(program[1], program[2]));
+   if (args[opcode] == 2) {
+      PRINTF("\n");
+      return;
+   }
+
+   PRINTF(", ");
+   print_src_reg(GET_SRC2_REG(program[2]));
+   PRINTF("\n");
+   return;
+}
+
+
+static void
+print_tex_op(GLuint opcode, const GLuint * program)
+{
+   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
+   PRINTF(" = ");
+
+   PRINTF("%s ", opcodes[opcode]);
+
+   PRINTF("S[%d],", program[0] & T0_SAMPLER_NR_MASK);
+
+   print_reg_type_nr((program[1] >> T1_ADDRESS_REG_TYPE_SHIFT) &
+                     REG_TYPE_MASK,
+                     (program[1] >> T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK);
+   PRINTF("\n");
+}
+
+static void
+print_dcl_op(GLuint opcode, const GLuint * program)
+{
+   PRINTF("%s ", opcodes[opcode]);
+   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
+   PRINTF("\n");
+}
+
+
+void
+i915_disassemble_program(const GLuint * program, GLuint sz)
+{
+   GLuint size = program[0] & 0x1ff;
+   GLint i;
+
+   PRINTF("\t\tBEGIN\n");
+
+   assert(size + 2 == sz);
+
+   program++;
+   for (i = 1; i < sz; i += 3, program += 3) {
+      GLuint opcode = program[0] & (0x1f << 24);
+
+      PRINTF("\t\t");
+
+      if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
+         print_arith_op(opcode >> 24, program);
+      else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
+         print_tex_op(opcode >> 24, program);
+      else if (opcode == D0_DCL)
+         print_dcl_op(opcode >> 24, program);
+      else
+         PRINTF("Unknown opcode 0x%x\n", opcode);
+   }
+
+   PRINTF("\t\tEND\n\n");
+}
+
+
diff --git a/src/mesa/drivers/dri/i915tex/i915_vtbl.c b/src/mesa/drivers/dri/i915tex/i915_vtbl.c
index 13ca9aee6b..16fb92ce4b 100644
--- a/src/mesa/drivers/dri/i915tex/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915tex/i915_vtbl.c
@@ -66,6 +66,7 @@ i915_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
    st1 &= ~ST1_ENABLE;
 
    switch (rprim) {
+   case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
    case GL_TRIANGLES:
       if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
          st1 |= ST1_ENABLE;
@@ -202,7 +203,7 @@ i915_emit_invarient_state(struct intel_context *intel)
 
    /* Need to initialize this to zero.
     */
-   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (1));
+   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (0));
    OUT_BATCH(0);
 
    /* XXX: Use this */
@@ -220,6 +221,7 @@ i915_emit_invarient_state(struct intel_context *intel)
 
    /* Don't support twosided stencil yet */
    OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS | BFO_ENABLE_STENCIL_TWO_SIDE | 0);
+   OUT_BATCH(0);
 
    ADVANCE_BATCH();
 }
@@ -251,6 +253,9 @@ get_state_size(struct i915_hw_state *state)
    GLuint i;
    GLuint sz = 0;
 
+   if (dirty & I915_UPLOAD_INVARIENT)
+      sz += 30 * 4;
+
    if (dirty & I915_UPLOAD_CTX)
       sz += sizeof(state->Ctx);
 
@@ -306,6 +311,7 @@ i915_emit_state(struct intel_context *intel)
     * causing more state to be dirty!
     */
    dirty = get_dirty(state);
+   state->emitted |= dirty;
 
    if (INTEL_DEBUG & DEBUG_STATE)
       fprintf(stderr, "%s dirty: %x\n", __FUNCTION__, dirty);
@@ -428,12 +434,22 @@ i915_emit_state(struct intel_context *intel)
          i915_disassemble_program(state->Program, state->ProgramSize);
    }
 
-   state->emitted |= dirty;
+   assert(get_dirty(state) == 0);
 }
 
 static void
 i915_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
+   for (i = 0; i < I915_TEX_UNITS; i++) {
+      if (i915->state.tex_buffer[i] != NULL) {
+	 driBOUnReference(i915->state.tex_buffer[i]);
+	 i915->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
index c92b83bcb3..6665629d03 100644
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
@@ -27,6 +27,7 @@
 
 #include "intel_batchbuffer.h"
 #include "intel_ioctl.h"
+#include "i915_debug.h"
 
 /* Relocations in kernel space:
  *    - pass dma buffer seperately
@@ -71,9 +72,8 @@ intel_dump_batchbuffer(GLuint offset, GLuint * ptr, GLuint count)
 {
    int i;
    fprintf(stderr, "\n\n\nSTART BATCH (%d dwords):\n", count / 4);
-   for (i = 0; i < count / 4; i += 4)
-      fprintf(stderr, "0x%x:\t0x%08x 0x%08x 0x%08x 0x%08x\n",
-              offset + i * 4, ptr[i], ptr[i + 1], ptr[i + 2], ptr[i + 3]);
+   for (i = 0; i < count / 4; i += 1)
+      fprintf(stderr, "\t0x%08x\n", ptr[i]);
    fprintf(stderr, "END BATCH\n\n\n");
 }
 
@@ -186,8 +186,11 @@ do_flush_locked(struct intel_batchbuffer *batch,
       ptr[r->offset / 4] = driBOOffset(r->buf) + r->delta;
    }
 
-   if (INTEL_DEBUG & DEBUG_BATCH)
+   if (INTEL_DEBUG & DEBUG_BATCH) {
+      if (0) i915_dump_batchbuffer(ptr, ptr + used/4);
       intel_dump_batchbuffer(0, ptr, used);
+   }
+
 
    driBOUnmap(batch->buffer);
    batch->map = NULL;
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffers.c b/src/mesa/drivers/dri/i915tex/intel_buffers.c
index 4b2a4acaf9..13f26137aa 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffers.c
+++ b/src/mesa/drivers/dri/i915tex/intel_buffers.c
@@ -250,7 +250,8 @@ intelWindowMoved(struct intel_context *intel)
 	 flags = intel_fb->vblank_flags & ~VBLANK_FLAG_SECONDARY;
       }
 
-      if (flags != intel_fb->vblank_flags) {
+      if (flags != intel_fb->vblank_flags && intel_fb->vblank_flags &&
+	  !(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ)) {
 	 drmVBlank vbl;
 	 int i;
 
@@ -261,7 +262,9 @@ intelWindowMoved(struct intel_context *intel)
 	 }
 
 	 for (i = 0; i < intel_fb->pf_num_pages; i++) {
-	    if (!intel_fb->color_rb[i])
+	    if (!intel_fb->color_rb[i] ||
+		(intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <=
+		(1<<23))
 	       continue;
 
 	    vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending;
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.c b/src/mesa/drivers/dri/i915tex/intel_context.c
index f032c97dfb..8a5b4d1f83 100644
--- a/src/mesa/drivers/dri/i915tex/intel_context.c
+++ b/src/mesa/drivers/dri/i915tex/intel_context.c
@@ -605,12 +605,23 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
       if (intel->ctx.DrawBuffer == &intel_fb->Base) {
 
 	 if (intel->driDrawable != driDrawPriv) {
-	    intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0)
-	       ? driGetDefaultVBlankFlags(&intel->optionCache)
-	       : VBLANK_FLAG_NO_IRQ;
-	    (*dri_interface->getUST) (&intel_fb->swap_ust);
-	    driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags,
-				  &intel_fb->vbl_seq);
+	    if (driDrawPriv->pdraw->swap_interval == (unsigned)-1) {
+	       int i;
+
+	       intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0)
+		  ? driGetDefaultVBlankFlags(&intel->optionCache)
+		 : VBLANK_FLAG_NO_IRQ;
+
+	       (*dri_interface->getUST) (&intel_fb->swap_ust);
+	       driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags,
+				     &intel_fb->vbl_seq);
+	       intel_fb->vbl_waited = intel_fb->vbl_seq;
+
+	       for (i = 0; i < (2); i++) {
+		  if (intel_fb->color_rb[i])
+		     intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_seq;
+	       }
+	    }
 	 }
       }
 
@@ -660,37 +671,27 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
    if (sarea->width != intel->width ||
        sarea->height != intel->height ||
        sarea->rotation != intel->current_rotation) {
-      
-      void *batchMap = intel->batch->map;
-      
+      int numClipRects = intel->numClipRects;
+
       /*
        * FIXME: Really only need to do this when drawing to a
        * common back- or front buffer.
        */
 
       /*
-       * This will drop the outstanding batchbuffer on the floor
+       * This will essentially drop the outstanding batchbuffer on the floor.
        */
+      intel->numClipRects = 0;
 
-      if (batchMap != NULL) {
-	 driBOUnmap(intel->batch->buffer);
-	 intel->batch->map = NULL;
-      }
-
-      intel_batchbuffer_reset(intel->batch);
+      if (intel->Fallback)
+	 _swrast_flush(&intel->ctx);
 
-      if (batchMap == NULL) {
-	 driBOUnmap(intel->batch->buffer);
-	 intel->batch->map = NULL;
-      }
+      INTEL_FIREVERTICES(intel);
 
-      /* lose all primitives */
-      intel->prim.primitive = ~0;
-      intel->prim.start_ptr = 0;
-      intel->prim.flush = 0;
+      if (intel->batch->map != intel->batch->ptr)
+	 intel_batchbuffer_flush(intel->batch);
 
-      /* re-emit all state */
-      intel->vtbl.lost_hardware(intel);
+      intel->numClipRects = numClipRects;
 
       /* force window update */
       intel->lastStamp = 0;
@@ -724,7 +725,9 @@ void LOCK_HARDWARE( struct intel_context *intel )
 				    BUFFER_BACK_LEFT);
     }
 
-    if (intel_rb && (intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
+    if (intel_rb && intel_fb->vblank_flags &&
+	!(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) &&
+	(intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
 	drmVBlank vbl;
 
 	vbl.request.type = DRM_VBLANK_ABSOLUTE;
diff --git a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
index 843a78eb82..74f6b2d851 100644
--- a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
@@ -97,9 +97,19 @@ intel_miptree_create(struct intel_context *intel,
       break;
    }
 
-   if (ok)
+   if (ok) {
+      if (!mt->compressed) {
+	 /* 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.
+	  */
+	 mt->pitch = ((mt->pitch * cpp + 63) & ~63) / cpp;
+      }
+
       mt->region = intel_region_alloc(intel->intelScreen,
                                       mt->cpp, mt->pitch, mt->total_height);
+   }
 
    if (!mt->region) {
       free(mt);
@@ -325,6 +335,7 @@ intel_miptree_image_data(struct intel_context *intel,
    }
 }
 
+extern GLuint intel_compressed_alignment(GLenum);
 /* Copy mipmap image between trees
  */
 void
@@ -342,8 +353,12 @@ intel_miptree_image_copy(struct intel_context *intel,
    const GLuint *src_depth_offset = intel_miptree_depth_offsets(src, level);
    GLuint i;
 
-   if (dst->compressed)
-      height /= 4;
+   if (dst->compressed) {
+       GLuint alignment = intel_compressed_alignment(dst->internal_format);
+       height = (height + 3) / 4;
+       width = ((width + alignment - 1) & ~(alignment - 1));
+   }
+
    for (i = 0; i < depth; i++) {
       intel_region_copy(intel->intelScreen,
                         dst->region, dst_offset + dst_depth_offset[i],
diff --git a/src/mesa/drivers/dri/i915tex/intel_span.c b/src/mesa/drivers/dri/i915tex/intel_span.c
index 5a978d9ce2..cc0dec1843 100644
--- a/src/mesa/drivers/dri/i915tex/intel_span.c
+++ b/src/mesa/drivers/dri/i915tex/intel_span.c
@@ -190,9 +190,9 @@ intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
                   intel_region_map(intel->intelScreen, irb->region);
                else
                   intel_region_unmap(intel->intelScreen, irb->region);
+               irb->pfMap = irb->region->map;
+               irb->pfPitch = irb->region->pitch;
             }
-            irb->pfMap = irb->region->map;
-            irb->pfPitch = irb->region->pitch;
          }
       }
    }
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index e8f878a701..9bb7d2f703 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -240,6 +240,8 @@
 #define BRW_FRONTWINDING_CW      0
 #define BRW_FRONTWINDING_CCW     1
 
+#define BRW_SPRITE_POINT_ENABLE  16
+
 #define BRW_INDEX_BYTE     0
 #define BRW_INDEX_WORD     1
 #define BRW_INDEX_DWORD    2
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index d5175399d6..73232b3c7c 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -74,6 +74,11 @@ static void compile_sf_prog( struct brw_context *brw,
       if (c.key.attrs & (1<<i)) {
 	 c.attr_to_idx[i] = idx;
 	 c.idx_to_attr[idx] = i;
+	 if (i >= VERT_RESULT_TEX0 && i <= VERT_RESULT_TEX7) {
+		 c.point_attrs[i].CoordReplace = 
+			brw->attribs.Point->CoordReplace[i - VERT_RESULT_TEX0];
+	 } else
+		 c.point_attrs[i].CoordReplace = GL_FALSE;
 	 idx++;
       }
    
@@ -90,7 +95,10 @@ static void compile_sf_prog( struct brw_context *brw,
       break;
    case SF_POINTS:
       c.nr_verts = 1;
-      brw_emit_point_setup( &c );
+      if (key->do_point_sprite)
+	  brw_emit_point_sprite_setup( &c );
+      else
+	  brw_emit_point_setup( &c );
       break;
    case SF_UNFILLED_TRIS:
       c.nr_verts = 3;
@@ -162,7 +170,8 @@ static void upload_sf_prog( struct brw_context *brw )
       break;
    }
 
-
+   key.do_point_sprite = brw->attribs.Point->PointSprite;
+   key.SpriteOrigin = brw->attribs.Point->SpriteOrigin;
    /* _NEW_LIGHT */
    key.do_flat_shading = (brw->attribs.Light->ShadeModel == GL_FLAT);
    key.do_twoside_color = (brw->attribs.Light->Enabled && brw->attribs.Light->Model.TwoSide);
@@ -179,7 +188,7 @@ static void upload_sf_prog( struct brw_context *brw )
 
 const struct brw_tracked_state brw_sf_prog = {
    .dirty = {
-      .mesa  = (_NEW_LIGHT|_NEW_POLYGON),
+      .mesa  = (_NEW_LIGHT|_NEW_POLYGON|_NEW_POINT),
       .brw   = (BRW_NEW_REDUCED_PRIMITIVE),
       .cache = CACHE_NEW_VS_PROG
    },
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index fb72b84ba8..e374e372bb 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -50,9 +50,14 @@ struct brw_sf_prog_key {
    GLuint do_flat_shading:1;
    GLuint attrs:16;
    GLuint frontface_ccw:1;
-   GLuint pad:11;
+   GLuint do_point_sprite:1;
+   GLuint pad:10;
+   GLenum SpriteOrigin;
 };
 
+struct brw_sf_point_tex {
+	GLboolean CoordReplace;	
+};
 
 struct brw_sf_compile {
    struct brw_compile func;
@@ -94,12 +99,14 @@ struct brw_sf_compile {
 
    GLubyte attr_to_idx[VERT_RESULT_MAX];   
    GLubyte idx_to_attr[VERT_RESULT_MAX];   
+   struct brw_sf_point_tex point_attrs[VERT_RESULT_MAX];
 };
 
  
 void brw_emit_tri_setup( struct brw_sf_compile *c );
 void brw_emit_line_setup( struct brw_sf_compile *c );
 void brw_emit_point_setup( struct brw_sf_compile *c );
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c );
 void brw_emit_anyprim_setup( struct brw_sf_compile *c );
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_sf_emit.c b/src/mesa/drivers/dri/i965/brw_sf_emit.c
index cbaf018c44..22911a4649 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_emit.c
@@ -497,6 +497,86 @@ void brw_emit_line_setup( struct brw_sf_compile *c )
    } 
 }
 
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c )
+{
+   struct brw_compile *p = &c->func;
+   GLuint i;
+
+   c->nr_verts = 1;
+   alloc_regs(c);
+   copy_z_inv_w(c);
+   for (i = 0; i < c->nr_setup_regs; i++)
+   {
+      struct brw_sf_point_tex *tex = &c->point_attrs[c->idx_to_attr[2*i]];
+      struct brw_reg a0 = offset(c->vert[0], i);
+      GLushort pc, pc_persp, pc_linear;
+      GLboolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
+            
+      if (pc_persp)
+      {				
+	  if (!tex->CoordReplace) {
+	      brw_set_predicate_control_flag_value(p, pc_persp);
+	      brw_MUL(p, a0, a0, c->inv_w[0]);
+	  }
+      }
+
+      if (tex->CoordReplace) {
+	  /* Caculate 1.0/PointWidth */
+	  brw_math(&c->func,
+		  c->tmp,
+		  BRW_MATH_FUNCTION_INV,
+		  BRW_MATH_SATURATE_NONE,
+		  0,
+		  c->dx0,
+		  BRW_MATH_DATA_SCALAR,
+		  BRW_MATH_PRECISION_FULL);
+
+	  if (c->key.SpriteOrigin == GL_UPPER_LEFT) {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, negate(c->inv_w[0]));
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  } else {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  }
+      } else {
+	  brw_MOV(p, c->m1Cx, brw_imm_ud(0));
+	  brw_MOV(p, c->m2Cy, brw_imm_ud(0));
+      }
+
+      {
+	 brw_set_predicate_control_flag_value(p, pc); 
+	 if (tex->CoordReplace) {
+	     if (c->key.SpriteOrigin == GL_UPPER_LEFT) {
+		 brw_MUL(p, c->m3C0, c->inv_w[0], brw_imm_f(1.0));
+		 brw_MOV(p, vec1(suboffset(c->m3C0, 0)), brw_imm_f(0.0));
+	     }
+	     else
+		 brw_MOV(p, c->m3C0, brw_imm_f(0.0));
+	 } else {
+	 	brw_MOV(p, c->m3C0, a0); /* constant value */
+	 }
+
+	 /* Copy m0..m3 to URB. 
+	  */
+	 brw_urb_WRITE(p, 
+		       brw_null_reg(),
+		       0,
+		       brw_vec8_grf(0, 0),
+		       0, 	/* allocate */
+		       1,	/* used */
+		       4, 	/* msg len */
+		       0,	/* response len */
+		       last, 	/* eot */
+		       last, 	/* writes complete */
+		       i*4,	/* urb destination offset */
+		       BRW_URB_SWIZZLE_TRANSPOSE);
+      }
+   }
+}
 
 /* Points setup - several simplifications as all attributes are
  * constant across the face of the point (point sprites excluded!)
@@ -561,6 +641,7 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
    struct brw_compile *p = &c->func;
    struct brw_reg ip = brw_ip_reg();
    struct brw_reg payload_prim = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0);
+   struct brw_reg payload_attr = get_element_ud(brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0), 0); 
    struct brw_reg primmask;
    struct brw_instruction *jmp;
    struct brw_reg v1_null_ud = vec1(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD));
@@ -601,6 +682,14 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
    }
    brw_land_fwd_jump(p, jmp); 
 
+   brw_set_conditionalmod(p, BRW_CONDITIONAL_Z);
+   brw_AND(p, v1_null_ud, payload_attr, brw_imm_ud(1<<BRW_SPRITE_POINT_ENABLE));
+   jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
+   {
+      brw_emit_point_sprite_setup( c );
+   }
+   brw_land_fwd_jump(p, jmp); 
+
    brw_emit_point_setup( c );
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index bfac52d765..236c6fd42a 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -173,7 +173,8 @@ static void upload_sf_unit( struct brw_context *brw )
       
 
    /* _NEW_LINE */
-   sf.sf6.line_width = brw->attribs.Line->_Width * (1<<1);
+   /* XXX use ctx->Const.Min/MaxLineWidth here */
+   sf.sf6.line_width = CLAMP(brw->attribs.Line->Width, 1.0, 5.0) * (1<<1);
 
    sf.sf6.line_endcap_aa_region_width = 1;
    if (brw->attribs.Line->SmoothFlag)
@@ -183,7 +184,10 @@ static void upload_sf_unit( struct brw_context *brw )
 
    /* _NEW_POINT */
    sf.sf6.point_rast_rule = 1;	/* opengl conventions */
-   sf.sf7.point_size = brw->attribs.Point->_Size * (1<<3);
+   /* XXX clamp max depends on AA vs. non-AA */
+
+   sf.sf7.sprite_point = brw->attribs.Point->PointSprite;
+   sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 255.0) * (1<<3);
    sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated;
       
    /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c
index 9d4b9867d2..ad29316a42 100644
--- a/src/mesa/drivers/dri/i965/brw_tex.c
+++ b/src/mesa/drivers/dri/i965/brw_tex.c
@@ -154,13 +154,19 @@ brwChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
+   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
+       return &_mesa_texformat_rgb_dxt1;
+
+   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
+       return &_mesa_texformat_rgba_dxt1;
+
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
+       return &_mesa_texformat_rgba_dxt3;
+
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgb_dxt1; /* there is no rgba support? */
+       return &_mesa_texformat_rgba_dxt5;
 
    case GL_DEPTH_COMPONENT:
    case GL_DEPTH_COMPONENT16:
@@ -168,6 +174,25 @@ brwChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_DEPTH_COMPONENT32:
       return &_mesa_texformat_z16;
 
+   case GL_SRGB_EXT:
+   case GL_SRGB8_EXT:
+   case GL_SRGB_ALPHA_EXT:
+   case GL_SRGB8_ALPHA8_EXT:
+   case GL_SLUMINANCE_EXT:
+   case GL_SLUMINANCE8_EXT:
+   case GL_SLUMINANCE_ALPHA_EXT:
+   case GL_SLUMINANCE8_ALPHA8_EXT:
+   case GL_COMPRESSED_SRGB_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_EXT:
+   case GL_COMPRESSED_SLUMINANCE_EXT:
+   case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
+	return &_mesa_texformat_srgba8;
+   case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
+     return &_mesa_texformat_srgb_dxt1;
+
    default:
       fprintf(stderr, "unexpected texture format %s in %s\n", 
 	      _mesa_lookup_enum_by_nr(internalFormat),
diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
index af1ad0f1ef..2094a1c8ad 100644
--- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
+++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
@@ -37,6 +37,7 @@
 #include "intel_tex_layout.h"
 #include "macros.h"
 
+#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
 
 GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 {
@@ -53,11 +54,20 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
       GLuint pack_x_pitch, pack_x_nr;
       GLuint pack_y_pitch;
       GLuint level;
+      GLuint align_h = 2;
+      GLuint align_w = 4;
 
-      mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
       mt->total_height = 0;
+      
+      if (mt->compressed) {
+          align_w = intel_compressed_alignment(mt->internal_format);
+          mt->pitch = ALIGN(width, align_w);
+          pack_y_pitch = (height + 3) / 4;
+      } else {
+          mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
+          pack_y_pitch = ALIGN(mt->height0, align_h);
+      }
 
-      pack_y_pitch = MAX2(mt->height0, 2);
       pack_x_pitch = mt->pitch;
       pack_x_nr = 1;
 
@@ -83,20 +93,30 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 
 
 	 mt->total_height += y;
-
-	 if (pack_x_pitch > 4) {
-	    pack_x_pitch >>= 1;
-	    pack_x_nr <<= 1;
-	    assert(pack_x_pitch * pack_x_nr <= mt->pitch);
-	 }
-
-	 if (pack_y_pitch > 2) {
-	    pack_y_pitch >>= 1;
-	 }
-
 	 width  = minify(width);
 	 height = minify(height);
 	 depth  = minify(depth);
+
+    if (mt->compressed) {
+        pack_y_pitch = (height + 3) / 4;
+        
+        if (pack_x_pitch > ALIGN(width, align_w)) {
+            pack_x_pitch = ALIGN(width, align_w);
+            pack_x_nr <<= 1;
+        }
+    } else {
+        if (pack_x_pitch > 4) {
+            pack_x_pitch >>= 1;
+            pack_x_nr <<= 1;
+            assert(pack_x_pitch * pack_x_nr <= mt->pitch);
+        }
+
+        if (pack_y_pitch > 2) {
+            pack_y_pitch >>= 1;
+            pack_y_pitch = ALIGN(pack_y_pitch, align_h);
+        }
+    }
+
       }
       break;
    }
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 6eb11b19ad..d00f0c71a8 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -201,7 +201,7 @@ static void unalias2( struct brw_vs_compile *c,
 				    struct brw_reg,
 				    struct brw_reg ))
 {
-   if ((dst.file == arg0.file && dst.nr == arg0.nr) &&
+   if ((dst.file == arg0.file && dst.nr == arg0.nr) ||
        (dst.file == arg1.file && dst.nr == arg1.nr)) {
       struct brw_compile *p = &c->func;
       struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 1497dc7968..904c00bef8 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -285,7 +285,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
 
 
    /* BRW_NEW_WM_INPUT_DIMENSIONS */
-   key->projtex_mask = brw->wm.input_size_masks[4-1]; 
+   key->projtex_mask = brw->wm.input_size_masks[4-1] >> (FRAG_ATTRIB_TEX0 - FRAG_ATTRIB_WPOS); 
 
    /* _NEW_LIGHT */
    key->flat_shade = (brw->attribs.Light->ShadeModel == GL_FLAT);
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index ff97d87dc4..62bb081206 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -660,7 +660,7 @@ static GLboolean projtex( struct brw_wm_compile *c,
       return 0;  /* ut2004 gun rendering !?! */
    else if (src.File == PROGRAM_INPUT && 
 	    GET_SWZ(src.Swizzle, W) == W &&
-	    (c->key.projtex_mask & (1<<src.Index)) == 0)
+           (c->key.projtex_mask & (1<<(src.Index + FRAG_ATTRIB_WPOS - FRAG_ATTRIB_TEX0))) == 0)
       return 0;
    else
       return 1;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
index 93d4cfc3a5..83a4b02e48 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
@@ -54,7 +54,7 @@ static GLuint translate_wrap_mode( GLenum wrap )
    case GL_REPEAT: 
       return BRW_TEXCOORDMODE_WRAP;
    case GL_CLAMP:  
-      return BRW_TEXCOORDMODE_CLAMP_BORDER; /* conform likes it this way */
+      return BRW_TEXCOORDMODE_CLAMP;
    case GL_CLAMP_TO_EDGE: 
       return BRW_TEXCOORDMODE_CLAMP; /* conform likes it this way */
    case GL_CLAMP_TO_BORDER: 
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 d24c618a66..0a45164a0f 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -114,11 +114,24 @@ static GLuint translate_tex_format( GLuint mesa_format )
       return BRW_SURFACEFORMAT_FXT1;
 
    case MESA_FORMAT_Z16:
-      return BRW_SURFACEFORMAT_L16_UNORM;
+      return BRW_SURFACEFORMAT_I16_UNORM;
 
-   case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-      return BRW_SURFACEFORMAT_DXT1_RGB;
+       return BRW_SURFACEFORMAT_DXT1_RGB;
+
+   case MESA_FORMAT_RGBA_DXT1:
+       return BRW_SURFACEFORMAT_BC1_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT3:
+       return BRW_SURFACEFORMAT_BC2_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT5:
+       return BRW_SURFACEFORMAT_BC3_UNORM;
+
+   case MESA_FORMAT_SRGBA8:
+      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB;
+   case MESA_FORMAT_SRGB_DXT1:
+      return BRW_SURFACEFORMAT_BC1_UNORM_SRGB;
 
    default:
       assert(0);
diff --git a/src/mesa/drivers/dri/i965/intel_blit.c b/src/mesa/drivers/dri/i965/intel_blit.c
index 173d1d5b6c..f88cbb2328 100644
--- a/src/mesa/drivers/dri/i965/intel_blit.c
+++ b/src/mesa/drivers/dri/i965/intel_blit.c
@@ -532,12 +532,15 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  GLuint dst_offset,
 				  GLboolean dst_tiled,
 				  GLshort x, GLshort y, 
-				  GLshort w, GLshort h)
+				  GLshort w, GLshort h,
+				  GLenum logic_op)
 {
    struct xy_setup_blit setup;
    struct xy_text_immediate_blit text;
    int dwords = ((src_size + 7) & ~7) / 4;
 
+   assert( logic_op - GL_CLEAR >= 0 );
+   assert( logic_op - GL_CLEAR < 0x10 );
 
    if (w < 0 || h < 0) 
       return;
@@ -561,7 +564,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
    setup.br0.length = (sizeof(setup) / sizeof(int)) - 2;
       
    setup.br13.dest_pitch = dst_pitch;
-   setup.br13.rop = 0xcc;
+   setup.br13.rop = translate_raster_op(logic_op);
    setup.br13.color_depth = (cpp == 4) ? BR13_8888 : BR13_565;
    setup.br13.clipping_enable = 0;
    setup.br13.mono_source_transparency = 1;
diff --git a/src/mesa/drivers/dri/i965/intel_blit.h b/src/mesa/drivers/dri/i965/intel_blit.h
index 8b0cc65243..e361545c8f 100644
--- a/src/mesa/drivers/dri/i965/intel_blit.h
+++ b/src/mesa/drivers/dri/i965/intel_blit.h
@@ -72,6 +72,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  GLuint dst_offset,
 				  GLboolean dst_tiled,
 				  GLshort dst_x, GLshort dst_y, 
-				  GLshort w, GLshort h);
+				  GLshort w, GLshort h,
+				  GLenum logic_op );
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/intel_context.c b/src/mesa/drivers/dri/i965/intel_context.c
index 4f51fefe0f..eafe809d3a 100644
--- a/src/mesa/drivers/dri/i965/intel_context.c
+++ b/src/mesa/drivers/dri/i965/intel_context.c
@@ -66,6 +66,7 @@
 int INTEL_DEBUG = (0);
 #endif
 
+#define need_GL_NV_point_sprite
 #define need_GL_ARB_multisample
 #define need_GL_ARB_point_parameters
 #define need_GL_ARB_texture_compression
@@ -81,6 +82,7 @@ int INTEL_DEBUG = (0);
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
+#define need_GL_EXT_point_parameters
 #include "extension_helper.h"
 
 #ifndef VERBOSE
@@ -146,6 +148,7 @@ const struct dri_extension card_extensions[] =
     { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
+    { "GL_NV_point_sprite",                GL_NV_point_sprite_functions },
     { "GL_ARB_texture_border_clamp",       NULL },
     { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_cube_map",           NULL },
@@ -158,6 +161,8 @@ const struct dri_extension card_extensions[] =
     { "GL_NV_texture_rectangle",           NULL },
     { "GL_EXT_texture_rectangle",          NULL },
     { "GL_ARB_texture_rectangle",          NULL },
+    { "GL_ARB_point_sprite",               NULL},
+    { "GL_ARB_point_parameters",	   NULL }, 
     { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
     { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
@@ -177,6 +182,7 @@ const struct dri_extension card_extensions[] =
     { "GL_EXT_texture_env_dot3",           NULL },
     { "GL_EXT_texture_filter_anisotropic", NULL },
     { "GL_EXT_texture_lod_bias",           NULL },
+    { "GL_EXT_texture_sRGB",               NULL },
     { "GL_3DFX_texture_compression_FXT1",  NULL },
     { "GL_APPLE_client_storage",           NULL },
     { "GL_MESA_pack_invert",               NULL },
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
index 8486086b27..0fb33e27f4 100644
--- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
@@ -75,7 +75,7 @@ struct intel_mipmap_tree *intel_miptree_create( struct intel_context *intel,
    mt->width0 = width0;
    mt->height0 = height0;
    mt->depth0 = depth0;
-   mt->cpp = compressed ? 2 : cpp;
+   mt->cpp = cpp;
    mt->compressed = compressed;
 
    switch (intel->intelScreen->deviceID) {
@@ -211,7 +211,7 @@ GLuint intel_miptree_image_offset(struct intel_mipmap_tree *mt,
 
 
 
-
+extern GLuint intel_compressed_alignment(GLenum);
 /* Upload data for a particular image.
  */
 GLboolean intel_miptree_image_data(struct intel_context *intel, 
@@ -226,6 +226,17 @@ GLboolean intel_miptree_image_data(struct intel_context *intel,
    GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
    const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
    GLuint i;
+   GLuint width, height, alignment;
+
+   width = dst->level[level].width;
+   height = dst->level[level].height;
+
+   if (dst->compressed) {
+       alignment = intel_compressed_alignment(dst->internal_format);
+       src_row_pitch = ((src_row_pitch + alignment - 1) & ~(alignment - 1));
+       width = ((width + alignment - 1) & ~(alignment - 1));
+       height = (height + 3) / 4;
+   }
 
    DBG("%s\n", __FUNCTION__);
    for (i = 0; i < depth; i++) {
@@ -237,8 +248,8 @@ GLboolean intel_miptree_image_data(struct intel_context *intel,
 			     src,
 			     src_row_pitch,
 			     0, 0,	/* source x,y */
-			     dst->level[level].width,
-			     dst->level[level].height))
+			     width,
+			     height))
 	 return GL_FALSE;
       src += src_image_pitch;
    }
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
index 5841afaa3e..f3f062e651 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
@@ -226,10 +226,10 @@ do_blit_bitmap( GLcontext *ctx,
       dsty = dPriv->y + (dPriv->h - dsty - height);  
       dstx = dPriv->x + dstx;
 
-      dest_rect.x1 = dstx;
-      dest_rect.y1 = dsty;
-      dest_rect.x2 = dstx + width;
-      dest_rect.y2 = dsty + height;
+      dest_rect.x1 = dstx < 0 ? 0 : dstx;
+      dest_rect.y1 = dsty < 0 ? 0 : dsty;
+      dest_rect.x2 = dstx + width < 0 ? 0 : dstx + width;
+      dest_rect.y2 = dsty + height < 0 ? 0 : dsty + height;
 
       for (i = 0; i < nbox; i++) {
          drm_clip_rect_t rect;
@@ -260,7 +260,9 @@ do_blit_bitmap( GLcontext *ctx,
 	       int h = MIN2(DY, box_h - py);
 	       int w = MIN2(DX, box_w - px); 
 	       GLuint sz = align(align(w,8) * h, 64)/8;
-	    
+	       GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
+		  ctx->Color.LogicOp : GL_COPY;
+
 	       assert(sz <= sizeof(stipple));
 	       memset(stipple, 0, sz);
 
@@ -288,7 +290,8 @@ do_blit_bitmap( GLcontext *ctx,
 						  dst->tiled,
 						  rect.x1 + px,
 						  rect.y2 - (py + h),
-						  w, h);
+						  w, h,
+						  logic_op);
 	    } 
 	 } 
       }
diff --git a/src/mesa/drivers/dri/i965/intel_tex_validate.c b/src/mesa/drivers/dri/i965/intel_tex_validate.c
index 44ee94614d..8c05e7cdab 100644
--- a/src/mesa/drivers/dri/i965/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i965/intel_tex_validate.c
@@ -122,6 +122,29 @@ static void intel_texture_invalidate_cb( struct intel_context *intel,
    intel_texture_invalidate( (struct intel_texture_object *) ptr );
 }
 
+#include "texformat.h"
+static GLuint intel_compressed_num_bytes(GLenum mesaFormat)
+{
+    GLuint 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;
+}
 
 /*  
  */
@@ -132,7 +155,8 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
    GLuint face, i;
    GLuint nr_faces = 0;
    struct gl_texture_image *firstImage;
-
+   GLuint cpp = 0;
+   
    if( tObj == intel->frame_buffer_texobj )
       return GL_FALSE;
    
@@ -165,6 +189,12 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 
 
 
+   if (firstImage->IsCompressed) {
+       cpp = intel_compressed_num_bytes(firstImage->TexFormat->MesaFormat);
+   } else {
+       cpp = firstImage->TexFormat->TexelBytes;
+   }
+       
    /* Check tree can hold all active levels.  Check tree matches
     * target, imageFormat, etc.
     */
@@ -176,7 +206,7 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 	intelObj->mt->width0 != firstImage->Width ||
 	intelObj->mt->height0 != firstImage->Height ||
 	intelObj->mt->depth0 != firstImage->Depth ||
-	intelObj->mt->cpp != firstImage->TexFormat->TexelBytes ||
+	intelObj->mt->cpp != cpp ||
 	intelObj->mt->compressed != firstImage->IsCompressed)) 
    {
       intel_miptree_destroy(intel, intelObj->mt);
@@ -199,7 +229,7 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 					  firstImage->Width,
 					  firstImage->Height,
 					  firstImage->Depth,
-					  firstImage->TexFormat->TexelBytes,
+					  cpp,
 					  firstImage->IsCompressed);
 
       /* Tell the buffer manager that we will manage the backing
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.c b/src/mesa/drivers/dri/intel/intel_tex_layout.c
index fcb5cc3906..fdecd3e186 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.c
@@ -40,6 +40,23 @@ static int align(int value, int alignment)
    return (value + alignment - 1) & ~(alignment - 1);
 }
 
+GLuint intel_compressed_alignment(GLenum internalFormat)
+{
+    GLuint alignment = 4;
+
+    switch (internalFormat) {
+    case GL_COMPRESSED_RGB_FXT1_3DFX:
+    case GL_COMPRESSED_RGBA_FXT1_3DFX:
+        alignment = 8;
+        break;
+
+    default:
+        break;
+    }
+
+    return alignment;
+}
+
 void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 {
    GLint align_h = 2, align_w = 4;
@@ -51,17 +68,30 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 
    mt->pitch = mt->width0;
 
+   if (mt->compressed) {
+       align_w = intel_compressed_alignment(mt->internal_format);
+       mt->pitch = align(mt->width0, align_w);
+   }
+
    /* May need to adjust pitch to accomodate the placement of
     * the 2nd mipmap.  This occurs when the alignment
     * constraints of mipmap placement push the right edge of the
     * 2nd mipmap out past the width of its parent.
     */
    if (mt->first_level != mt->last_level) {
-      GLuint mip1_width = align(minify(mt->width0), align_w)
-			+ minify(minify(mt->width0));
+       GLuint mip1_width;
+
+       if (mt->compressed) {
+           mip1_width = align(minify(mt->width0), align_w)
+               + align(minify(minify(mt->width0)), align_w);
+       } else {
+           mip1_width = align(minify(mt->width0), align_w)
+               + minify(minify(mt->width0));
+       }
 
-      if (mip1_width > mt->width0)
-	 mt->pitch = mip1_width;
+       if (mip1_width > mt->pitch) {
+           mt->pitch = mip1_width;
+       }
    }
 
    /* Pitch must be a whole number of dwords, even though we
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.h b/src/mesa/drivers/dri/intel/intel_tex_layout.h
index 1e37f8f525..99d41c3629 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.h
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.h
@@ -39,3 +39,4 @@ static GLuint minify( GLuint d )
 }
 
 extern void i945_miptree_layout_2d( struct intel_mipmap_tree *mt );
+extern GLuint intel_compressed_alignment(GLenum);
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vb.c b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
index 75cf0e2ed2..248fa2a9a2 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vb.c
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
@@ -103,7 +103,7 @@ void TAG(translate_vertex)(GLcontext *ctx,
 
    assert( p + 1 - (CARD32 *)src == 10 );
 	 
-   dst->pointSize = ctx->Point._Size;
+   dst->pointSize = ctx->Point.Size;
 }
 
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.c b/src/mesa/drivers/dri/mach64/mach64_tris.c
index 08cc1849a1..369f610442 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tris.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tris.c
@@ -673,7 +673,10 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa,
 #if MACH64_NATIVE_VTXFMT
    GLcontext *ctx = mmesa->glCtx;
    const GLuint vertsize = mmesa->vertex_size;
-   GLint width = (GLint)(mmesa->glCtx->Line._Width * 2.0); /* 2 fractional bits for hardware */
+   /* 2 fractional bits for hardware: */
+   const int width = (int) (2.0 * CLAMP(mmesa->glCtx->Line.Width,
+                                        mmesa->glCtx->Const.MinLineWidth,
+                                        mmesa->glCtx->Const.MaxLineWidth));
    GLfloat ooa;
    GLuint *pxy0, *pxy1;
    GLuint xy0old, xy0, xy1old, xy1;
@@ -691,9 +694,6 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa,
       mach64_print_vertex( ctx, v1 );
    }
   
-   if( !width )
-      width = 1;	/* round to the nearest supported width */
-      
    pxy0 = &v0->ui[xyoffset];
    xy0old = *pxy0;
    xy0 = LE32_IN( &xy0old );
@@ -961,7 +961,10 @@ static __inline void mach64_draw_point( mach64ContextPtr mmesa,
 #if MACH64_NATIVE_VTXFMT
    GLcontext *ctx = mmesa->glCtx;
    const GLuint vertsize = mmesa->vertex_size;
-   GLint sz = (GLint)(mmesa->glCtx->Point._Size * 2.0); /* 2 fractional bits for hardware */
+   /* 2 fractional bits for hardware: */
+   GLint sz = (GLint) (2.0 * CLAMP(mmesa->glCtx->Point.Size,
+                                   ctx->Const.MinPointSize,
+                                   ctx->Const.MaxPointSize));
    GLfloat ooa;
    GLuint *pxy;
    GLuint xyold, xy;
diff --git a/src/mesa/drivers/dri/mga/mgatris.c b/src/mesa/drivers/dri/mga/mgatris.c
index 2b7ea05b14..91b413ae76 100644
--- a/src/mesa/drivers/dri/mga/mgatris.c
+++ b/src/mesa/drivers/dri/mga/mgatris.c
@@ -104,8 +104,10 @@ static void __inline__ mga_draw_quad( mgaContextPtr mmesa,
 static __inline__ void mga_draw_point( mgaContextPtr mmesa,
 					mgaVertexPtr tmp )
 {
-   GLfloat sz = mmesa->glCtx->Point._Size * .5;
-   int vertex_size = mmesa->vertex_size;
+   const GLfloat sz = 0.5 * CLAMP(mmesa->glCtx->Point.Size,
+                                  mmesa->glCtx->Const.MinPointSize,
+                                  mmesa->glCtx->Const.MaxPointSize);
+   const int vertex_size = mmesa->vertex_size;
    GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size );
    int j;
    
@@ -165,7 +167,9 @@ static __inline__ void mga_draw_line( mgaContextPtr mmesa,
    GLuint vertex_size = mmesa->vertex_size;
    GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size );
    GLfloat dx, dy, ix, iy;
-   GLfloat width = mmesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(mmesa->glCtx->Line.Width,
+                               mmesa->glCtx->Const.MinLineWidth,
+                               mmesa->glCtx->Const.MaxLineWidth);
    GLint j;
 
 #if 0
diff --git a/src/mesa/drivers/dri/nouveau/Makefile b/src/mesa/drivers/dri/nouveau/Makefile
index 20d2de5eef..6ea4594f1e 100644
--- a/src/mesa/drivers/dri/nouveau/Makefile
+++ b/src/mesa/drivers/dri/nouveau/Makefile
@@ -9,12 +9,13 @@ MINIGLX_SOURCES =
 
 DRIVER_SOURCES = \
 	nouveau_bufferobj.c      \
-	nouveau_buffers.c        \
 	nouveau_card.c           \
 	nouveau_context.c        \
 	nouveau_driver.c         \
+	nouveau_fbo.c            \
 	nouveau_fifo.c           \
 	nouveau_lock.c           \
+	nouveau_mem.c            \
 	nouveau_object.c         \
 	nouveau_screen.c         \
 	nouveau_span.c           \
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
index fc14060c04..be6455a01e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
@@ -2,11 +2,11 @@
 #include "enums.h"
 
 #include "nouveau_bufferobj.h"
-#include "nouveau_buffers.h"
 #include "nouveau_context.h"
 #include "nouveau_drm.h"
-#include "nouveau_object.h"
+#include "nouveau_mem.h"
 #include "nouveau_msg.h"
+#include "nouveau_object.h"
 
 #define NOUVEAU_MEM_FREE(mem) do {      \
 	nouveau_mem_free(ctx, (mem));   \
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
index 3439a35e7c..cbc89a151d 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
@@ -2,7 +2,7 @@
 #define __NOUVEAU_BUFFEROBJ_H__
 
 #include "mtypes.h"
-#include "nouveau_buffers.h"
+#include "nouveau_mem.h"
 
 #define NOUVEAU_BO_VRAM_OK (NOUVEAU_MEM_FB | NOUVEAU_MEM_FB_ACCEPTABLE)
 #define NOUVEAU_BO_GART_OK  (NOUVEAU_MEM_AGP | NOUVEAU_MEM_AGP_ACCEPTABLE)
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
deleted file mode 100644
index 6a2c2013e3..0000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
+++ /dev/null
@@ -1,441 +0,0 @@
-#include "utils.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "fbobject.h"
-
-#include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_msg.h"
-
-#define MAX_MEMFMT_LENGTH 32768
-
-/* Unstrided blit using NV_MEMORY_TO_MEMORY_FORMAT */
-GLboolean
-nouveau_memformat_flat_emit(GLcontext *ctx,
-      			    nouveau_mem *dst, nouveau_mem *src,
-			    GLuint dst_offset, GLuint src_offset,
-			    GLuint size)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   uint32_t src_handle, dst_handle;
-   GLuint count;
-
-   if (src_offset + size > src->size) {
-      MESSAGE("src out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-   if (dst_offset + size > dst->size) {
-      MESSAGE("dst out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-
-   src_handle = (src->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   dst_handle = (dst->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   src_offset += nouveau_mem_gpu_offset_get(ctx, src);
-   dst_offset += nouveau_mem_gpu_offset_get(ctx, dst);
-
-   BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN, 2);
-   OUT_RING       (src_handle);
-   OUT_RING       (dst_handle);
-
-   count = (size / MAX_MEMFMT_LENGTH) + ((size % MAX_MEMFMT_LENGTH) ? 1 : 0);
-
-   while (count--) {
-      GLuint length = (size > MAX_MEMFMT_LENGTH) ? MAX_MEMFMT_LENGTH : size;
-
-      BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
-      OUT_RING       (src_offset);
-      OUT_RING       (dst_offset);
-      OUT_RING       (0); /* pitch in */
-      OUT_RING       (0); /* pitch out */
-      OUT_RING       (length); /* line length */
-      OUT_RING       (1); /* number of lines */
-      OUT_RING       ((1 << 8) /* dst_inc */ | (1 << 0) /* src_inc */);
-      OUT_RING       (0); /* buffer notify? */
-      FIRE_RING();
-
-      src_offset += length;
-      dst_offset += length;
-      size       -= length;
-   }
-
-   return GL_TRUE;
-}
-
-void
-nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   drm_nouveau_mem_free_t memf;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (mem->map)
-      drmUnmap(mem->map, mem->size);
-   memf.flags         = mem->type;
-   memf.region_offset = mem->offset;
-   drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf, sizeof(memf));
-   FREE(mem);
-}
-
-nouveau_mem *
-nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   drm_nouveau_mem_alloc_t mema;
-   nouveau_mem *mem;
-   int ret;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: requested: type=0x%x, size=0x%x, align=0x%x\n",
-	    __func__, type, (GLuint)size, align);
-   }
-
-   mem = CALLOC(sizeof(nouveau_mem));
-   if (!mem)
-      return NULL;
-
-   mema.flags     = type;
-   mema.size      = mem->size = size;
-   mema.alignment = align;
-   mem->map       = NULL;
-   ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_MEM_ALLOC,
-	 		     &mema, sizeof(mema));
-   if (ret) {
-      FREE(mem);
-      return NULL;
-   }
-   mem->offset = mema.region_offset;
-   mem->type   = mema.flags;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: actual: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (type & NOUVEAU_MEM_MAPPED)
-      ret = drmMap(nmesa->driFd, mem->offset, mem->size, &mem->map);
-   if (ret) {
-      mem->map = NULL;
-      nouveau_mem_free(ctx, mem);
-      mem = NULL;
-   }
-
-   return mem;
-}
-
-uint32_t
-nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   if (mem->type & NOUVEAU_MEM_FB)
-      return (uint32_t)mem->offset - nmesa->vram_phys;
-   else if (mem->type & NOUVEAU_MEM_AGP)
-      return (uint32_t)mem->offset - nmesa->gart_phys;
-   else
-      return 0xDEADF00D;
-}
-
-static GLboolean
-nouveau_renderbuffer_pixelformat(nouveau_renderbuffer *nrb,
-      				 GLenum internalFormat)
-{
-      nrb->mesa.InternalFormat	= internalFormat;
-
-   /*TODO: We probably want to extend this a bit, and maybe make
-    *      card-specific? 
-    */
-      switch (internalFormat) {
-      case GL_RGBA:
-      case GL_RGBA8:
-	 nrb->mesa._BaseFormat	= GL_RGBA;
-	 nrb->mesa._ActualFormat= GL_RGBA8;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 8;
-	 nrb->mesa.GreenBits	= 8;
-	 nrb->mesa.BlueBits	= 8;
-	 nrb->mesa.AlphaBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_RGB:
-      case GL_RGB5:
-	 nrb->mesa._BaseFormat	= GL_RGB;
-	 nrb->mesa._ActualFormat= GL_RGB5;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 5;
-	 nrb->mesa.GreenBits	= 6;
-	 nrb->mesa.BlueBits	= 5;
-	 nrb->mesa.AlphaBits	= 0;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT16:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH_COMPONENT16;
-	 nrb->mesa.DataType	= GL_UNSIGNED_SHORT;
-	 nrb->mesa.DepthBits	= 16;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT24:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_STENCIL_INDEX8_EXT:
-	 nrb->mesa._BaseFormat	= GL_STENCIL_INDEX;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.StencilBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_DEPTH24_STENCIL8_EXT:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_STENCIL_EXT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->mesa.StencilBits  = 8;
-	 nrb->cpp		= 4;
-	 break;
-      default:
-	 return GL_FALSE;
-	 break;
-      }
-
-      return GL_TRUE;
-}
-
-static GLboolean
-nouveau_renderbuffer_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
-      			     GLenum internalFormat,
-			     GLuint width,
-			     GLuint height)
-{
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
-      fprintf(stderr, "%s: unknown internalFormat\n", __func__);
-      return GL_FALSE;
-   }
-
-   /* If this buffer isn't statically alloc'd, we may need to ask the
-    * drm for more memory */
-   if (!nrb->dPriv && (rb->Width != width || rb->Height != height)) {
-      GLuint pitch;
-
-      /* align pitches to 64 bytes */
-      pitch = ((width * nrb->cpp) + 63) & ~63;
-
-      if (nrb->mem)
-	 nouveau_mem_free(ctx, nrb->mem);
-      nrb->mem = nouveau_mem_alloc(ctx,
-	    			   NOUVEAU_MEM_FB | NOUVEAU_MEM_MAPPED,
-				   pitch*height,
-				   0);
-      if (!nrb->mem)
-	 return GL_FALSE;
-
-      /* update nouveau_renderbuffer info */
-      nrb->offset = nouveau_mem_gpu_offset_get(ctx, nrb->mem);
-      nrb->pitch  = pitch;
-   }
-
-   rb->Width = width;
-   rb->Height = height;
-   rb->InternalFormat = internalFormat;
-   return GL_TRUE;
-}
-
-static void
-nouveau_renderbuffer_delete(struct gl_renderbuffer *rb)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (nrb->mem)
-      nouveau_mem_free(ctx, nrb->mem);
-   FREE(nrb);
-}
-
-nouveau_renderbuffer *
-nouveau_renderbuffer_new(GLenum internalFormat, GLvoid *map,
-      			 GLuint offset,  GLuint pitch,
-			 __DRIdrawablePrivate *dPriv)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, 0);
-
-      nouveau_renderbuffer_pixelformat(nrb, internalFormat);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-
-      nrb->dPriv  = dPriv;
-      nrb->offset = offset;
-      nrb->pitch  = pitch;
-      nrb->map    = map;
-   }
-
-   return nrb;
-}
-
-static void
-nouveau_cliprects_drawable_set(nouveauContextPtr nmesa,
-      			       nouveau_renderbuffer *nrb)
-{
-   __DRIdrawablePrivate *dPriv = nrb->dPriv;
-
-   nmesa->numClipRects	= dPriv->numClipRects;
-   nmesa->pClipRects	= dPriv->pClipRects;
-   nmesa->drawX		= dPriv->x;
-   nmesa->drawY		= dPriv->y;
-   nmesa->drawW		= dPriv->w;
-   nmesa->drawH		= dPriv->h;
-}
-
-static void
-nouveau_cliprects_renderbuffer_set(nouveauContextPtr nmesa,
-      				   nouveau_renderbuffer *nrb)
-{
-   nmesa->numClipRects	= 1;
-   nmesa->pClipRects	= &nmesa->osClipRect;
-   nmesa->osClipRect.x1	= 0;
-   nmesa->osClipRect.y1	= 0;
-   nmesa->osClipRect.x2	= nrb->mesa.Width;
-   nmesa->osClipRect.y2	= nrb->mesa.Height;
-   nmesa->drawX		= 0;
-   nmesa->drawY		= 0;
-   nmesa->drawW		= nrb->mesa.Width;
-   nmesa->drawH		= nrb->mesa.Height;
-}
-
-void
-nouveau_window_moved(GLcontext *ctx)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb;
-
-   nrb = (nouveau_renderbuffer *)ctx->DrawBuffer->_ColorDrawBuffers[0][0];
-   if (!nrb)
-      return;
-
-   if (!nrb->dPriv)
-      nouveau_cliprects_renderbuffer_set(nmesa, nrb);
-   else
-      nouveau_cliprects_drawable_set(nmesa, nrb);
-
-   /* Viewport depends on window size/position, nouveauCalcViewport
-    * will take care of calling the hw-specific WindowMoved
-    */
-   ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
-	 		     ctx->Viewport.Width, ctx->Viewport.Height);
-   /* Scissor depends on window position */
-   ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	 		    ctx->Scissor.Width, ctx->Scissor.Height);
-}
-
-GLboolean
-nouveau_build_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *color[MAX_DRAW_BUFFERS];
-   nouveau_renderbuffer *depth;
-
-   _mesa_update_framebuffer(ctx);
-   _mesa_update_draw_buffer_bounds(ctx);
-
-   color[0] = (nouveau_renderbuffer *)fb->_ColorDrawBuffers[0][0];
-   if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped)
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer->Wrapped;
-   else
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer;
-
-   if (!nmesa->hw_func.BindBuffers(nmesa, 1, color, depth))
-      return GL_FALSE;
-   nouveau_window_moved(ctx);
-
-   return GL_TRUE;
-}
-
-static void
-nouveauDrawBuffer(GLcontext *ctx, GLenum buffer)
-{
-   nouveau_build_framebuffer(ctx, ctx->DrawBuffer);
-}
-
-static struct gl_framebuffer *
-nouveauNewFramebuffer(GLcontext *ctx, GLuint name)
-{
-   return _mesa_new_framebuffer(ctx, name);
-}
-
-static struct gl_renderbuffer *
-nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, name);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-   }
-   return &nrb->mesa;
-}
-
-static void
-nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
-                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
-{
-   if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
-      nouveau_build_framebuffer(ctx, fb);
-   }
-}
-
-static void
-nouveauFramebufferRenderbuffer(GLcontext *ctx,
-      			       struct gl_framebuffer *fb,
-      			       GLenum attachment,
-			       struct gl_renderbuffer *rb)
-{
-   _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
-   nouveau_build_framebuffer(ctx, fb);
-}
-
-static void
-nouveauRenderTexture(GLcontext *ctx,
-      		     struct gl_framebuffer *fb,
-      		     struct gl_renderbuffer_attachment *att)
-{
-}
-
-static void
-nouveauFinishRenderTexture(GLcontext *ctx,
-      			   struct gl_renderbuffer_attachment *att)
-{
-}
-
-void
-nouveauInitBufferFuncs(struct dd_function_table *func)
-{
-   func->DrawBuffer		 = nouveauDrawBuffer;
-
-   func->NewFramebuffer		 = nouveauNewFramebuffer;
-   func->NewRenderbuffer	 = nouveauNewRenderbuffer;
-   func->BindFramebuffer	 = nouveauBindFramebuffer;
-   func->FramebufferRenderbuffer = nouveauFramebufferRenderbuffer;
-   func->RenderTexture		 = nouveauRenderTexture;
-   func->FinishRenderTexture	 = nouveauFinishRenderTexture;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h b/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
deleted file mode 100644
index d86455184c..0000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
+++ /dev/null
@@ -1,48 +0,0 @@
-#ifndef __NOUVEAU_BUFFERS_H__
-#define __NOUVEAU_BUFFERS_H__
-
-#include <stdint.h>
-#include "mtypes.h"
-#include "utils.h"
-#include "renderbuffer.h"
-
-typedef struct nouveau_mem_t {
-   int type;
-   uint64_t offset;
-   uint64_t size;
-   void*    map;
-} nouveau_mem;
-
-extern nouveau_mem *nouveau_mem_alloc(GLcontext *ctx, int type,
-      				      GLuint size, GLuint align);
-extern void nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem);
-extern uint32_t nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem);
-
-extern GLboolean nouveau_memformat_flat_emit(GLcontext *ctx,
-      					     nouveau_mem *dst,
-					     nouveau_mem *src,
-					     GLuint dst_offset,
-					     GLuint src_offset,
-					     GLuint size);
-
-typedef struct nouveau_renderbuffer_t {
-   struct gl_renderbuffer mesa; /* must be first! */
-   __DRIdrawablePrivate  *dPriv;
-
-   nouveau_mem *mem;
-   void *	map;
-
-   int		cpp;
-   uint32_t	offset;
-   uint32_t	pitch;
-} nouveau_renderbuffer;
-
-extern nouveau_renderbuffer *nouveau_renderbuffer_new(GLenum internalFormat,
-      GLvoid *map, GLuint offset, GLuint pitch, __DRIdrawablePrivate *dPriv);
-extern void nouveau_window_moved(GLcontext *ctx);
-extern GLboolean nouveau_build_framebuffer(GLcontext *, struct gl_framebuffer *);
-extern nouveau_renderbuffer *nouveau_current_draw_buffer(GLcontext *ctx);
-
-extern void nouveauInitBufferFuncs(struct dd_function_table *func);
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c
index d96b00242c..1e13324b98 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c
@@ -180,7 +180,7 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 	driParseConfigFiles (&nmesa->optionCache, &screen->optionCache,
 			screen->driScreen->myNum, "nouveau");
 
-	nmesa->sarea = (drm_nouveau_sarea_t *)((char *)sPriv->pSAREA +
+	nmesa->sarea = (struct drm_nouveau_sarea *)((char *)sPriv->pSAREA +
 			screen->sarea_priv_offset);
 
 	/* Enable any supported extensions */
@@ -224,6 +224,8 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 			nv04TriInitFunctions( ctx );
 			break;
 		case NV_10:
+		case NV_11:
+		case NV_17:
 		case NV_20:
 		case NV_30:
 		case NV_40:
@@ -313,21 +315,24 @@ GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv )
 	return GL_TRUE;
 }
 
-static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
-				 __DRIdrawablePrivate *dPriv)
+void
+nouveauDoSwapBuffers(nouveauContextPtr nmesa, __DRIdrawablePrivate *dPriv)
 {
 	struct gl_framebuffer *fb;
-	nouveau_renderbuffer *src, *dst;
+	nouveauScreenPtr screen = dPriv->driScreenPriv->private;
+	nouveau_renderbuffer_t *src;
 	drm_clip_rect_t *box;
 	int nbox, i;
 
 	fb = (struct gl_framebuffer *)dPriv->driverPrivate;
-	dst = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
-	src = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
+	if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
+		src = (nouveau_renderbuffer_t *)
+			fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
+	} else {
+		src = (nouveau_renderbuffer_t *)
+			fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
+	}
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	LOCK_HARDWARE(nmesa);
 	nbox = dPriv->numClipRects;
 	box  = dPriv->pClipRects;
@@ -339,9 +344,10 @@ static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
 			OUT_RING       (6); /* X8R8G8B8 */
 		else
 			OUT_RING       (4); /* R5G6B5 */
-		OUT_RING       ((dst->pitch << 16) | src->pitch);
-		OUT_RING       (src->offset);
-		OUT_RING       (dst->offset);
+		OUT_RING(((screen->frontPitch * screen->fbFormat) << 16) |
+			  src->pitch);
+		OUT_RING(src->offset);
+		OUT_RING(screen->frontOffset);
 	}
 
 	for (i=0; i<nbox; i++, box++) {
@@ -355,7 +361,6 @@ static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
 	FIRE_RING();
 
 	UNLOCK_HARDWARE(nmesa);
-#endif
 }
 
 void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv)
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h
index 10d2ed6e17..65ecf21cab 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h
@@ -36,25 +36,24 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "mtypes.h"
 #include "tnl/t_vertex.h"
 
+#include "nouveau_fbo.h"
 #include "nouveau_screen.h"
-#include "nouveau_state_cache.h"
-#include "nouveau_buffers.h"
 #include "nouveau_shader.h"
+#include "nouveau_state_cache.h"
 #include "nouveau_sync.h"
 
 #include "xmlconfig.h"
 
-typedef struct nouveau_fifo_t{
-	int channel;
-	u_int32_t* buffer;
-	u_int32_t* mmio;
-	u_int32_t put_base;
-	u_int32_t current;
-	u_int32_t put;
-	u_int32_t free;
-	u_int32_t max;
-}
-nouveau_fifo;
+typedef struct nouveau_fifo {
+	struct drm_nouveau_channel_alloc drm;
+	uint32_t *pushbuf;
+	uint32_t *mmio;
+	uint32_t *notifier_block;
+	uint32_t  current;
+	uint32_t  put;
+	uint32_t  free;
+	uint32_t  max;
+} nouveau_fifo_t;
 
 #define TAG(x) nouveau##x
 #include "tnl_dd/t_dd_vertex.h"
@@ -83,8 +82,8 @@ typedef struct nouveau_hw_func_t {
 	GLboolean (*InitCard)(struct nouveau_context *);
 	/* Update buffer offset/pitch/format */
 	GLboolean (*BindBuffers)(struct nouveau_context *, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth);
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth);
 	/* Update anything that depends on the window position/size */
 	void      (*WindowMoved)(struct nouveau_context *);
 } nouveau_hw_func;
@@ -94,13 +93,7 @@ typedef struct nouveau_context {
 	GLcontext *glCtx;
 
 	/* The per-context fifo */
-	nouveau_fifo fifo;
-
-	/* The read-only regs */
-	volatile unsigned char* mmio;
-
-	/* The per-channel notifier block */
-	volatile void *notifier_block;
+	nouveau_fifo_t fifo;
 
 	/* Physical addresses of AGP/VRAM apertures */
 	uint64_t vram_phys;
@@ -109,12 +102,12 @@ typedef struct nouveau_context {
 	uint64_t gart_size;
 
 	/* Channel synchronisation */
-	drm_nouveau_notifier_alloc_t *syncNotifier;
+	struct drm_nouveau_notifierobj_alloc *syncNotifier;
 
 	/* ARB_occlusion_query / EXT_timer_query */
 	GLuint		  query_object_max;
 	GLboolean *	  query_alloc;
-	drm_nouveau_notifier_alloc_t *queryNotifier;
+	struct drm_nouveau_notifierobj_alloc *queryNotifier;
 
 	/* Additional hw-specific functions */
 	nouveau_hw_func hw_func;
@@ -130,6 +123,10 @@ typedef struct nouveau_context {
 	struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
 	GLuint vertex_attr_count;
 
+	/* Color and depth renderbuffers */
+	nouveau_renderbuffer_t *color_buffer[2];
+	nouveau_renderbuffer_t *depth_buffer;
+
 	/* Color buffer clear value */
 	uint32_t clear_color_value;
 
@@ -168,7 +165,7 @@ typedef struct nouveau_context {
 	nouveauShader *passthrough_fp;
 
 	nouveauScreenRec *screen;
-	drm_nouveau_sarea_t *sarea;
+	struct drm_nouveau_sarea *sarea;
 
 	__DRIcontextPrivate  *driContext;    /* DRI context */
 	__DRIscreenPrivate   *driScreen;     /* DRI screen */
@@ -223,6 +220,9 @@ extern GLboolean nouveauMakeCurrent( __DRIcontextPrivate *driContextPriv,
 
 extern GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv );
 
+extern void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
+				 __DRIdrawablePrivate *dPriv);
+
 extern void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv);
 
 extern void nouveauCopySubBuffer(__DRIdrawablePrivate *dPriv,
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.c b/src/mesa/drivers/dri/nouveau/nouveau_driver.c
index 00956aa8f8..4851c66835 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_driver.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_driver.c
@@ -41,7 +41,7 @@ GLboolean nouveauDRMGetParam(nouveauContextPtr nmesa,
 			     unsigned int      param,
 			     uint64_t*         value)
 {
-	drm_nouveau_getparam_t getp;
+	struct drm_nouveau_getparam getp;
 
 	getp.param = param;
 	if (!value || drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_GETPARAM,
@@ -56,7 +56,7 @@ GLboolean nouveauDRMSetParam(nouveauContextPtr nmesa,
 			     unsigned int      param,
 			     uint64_t          value)
 {
-	drm_nouveau_setparam_t setp;
+	struct drm_nouveau_setparam setp;
 
 	setp.param = param;
 	setp.value = value;
@@ -117,6 +117,9 @@ static const GLubyte *nouveauGetString( GLcontext *ctx, GLenum name )
 static void nouveauFlush( GLcontext *ctx )
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (ctx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT)
+		nouveauDoSwapBuffers(nmesa, nmesa->driDrawable);
 	FIRE_RING();
 }
 
@@ -124,6 +127,7 @@ static void nouveauFlush( GLcontext *ctx )
 static void nouveauFinish( GLcontext *ctx )
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
 	nouveauFlush( ctx );
 	nouveauWaitForIdle( nmesa );
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fbo.c b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
new file mode 100644
index 0000000000..cc3da8b9bd
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
@@ -0,0 +1,290 @@
+#include "utils.h"
+#include "framebuffer.h"
+#include "renderbuffer.h"
+#include "fbobject.h"
+
+#include "nouveau_context.h"
+#include "nouveau_fbo.h"
+#include "nouveau_fifo.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
+
+static GLboolean
+nouveau_renderbuffer_pixelformat(nouveau_renderbuffer_t * nrb,
+				 GLenum internalFormat)
+{
+	nrb->mesa.InternalFormat = internalFormat;
+
+	/*TODO: We probably want to extend this a bit, and maybe make
+	 *      card-specific? 
+	 */
+	switch (internalFormat) {
+	case GL_RGBA:
+	case GL_RGBA8:
+		nrb->mesa._BaseFormat = GL_RGBA;
+		nrb->mesa._ActualFormat = GL_RGBA8;
+		nrb->mesa.DataType = GL_UNSIGNED_BYTE;
+		nrb->mesa.RedBits = 8;
+		nrb->mesa.GreenBits = 8;
+		nrb->mesa.BlueBits = 8;
+		nrb->mesa.AlphaBits = 8;
+		nrb->cpp = 4;
+		break;
+	case GL_RGB:
+	case GL_RGB5:
+		nrb->mesa._BaseFormat = GL_RGB;
+		nrb->mesa._ActualFormat = GL_RGB5;
+		nrb->mesa.DataType = GL_UNSIGNED_BYTE;
+		nrb->mesa.RedBits = 5;
+		nrb->mesa.GreenBits = 6;
+		nrb->mesa.BlueBits = 5;
+		nrb->mesa.AlphaBits = 0;
+		nrb->cpp = 2;
+		break;
+	case GL_DEPTH_COMPONENT16:
+		nrb->mesa._BaseFormat = GL_DEPTH_COMPONENT;
+		nrb->mesa._ActualFormat = GL_DEPTH_COMPONENT16;
+		nrb->mesa.DataType = GL_UNSIGNED_SHORT;
+		nrb->mesa.DepthBits = 16;
+		nrb->cpp = 2;
+		break;
+	case GL_DEPTH_COMPONENT24:
+		nrb->mesa._BaseFormat = GL_DEPTH_COMPONENT;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.DepthBits = 24;
+		nrb->cpp = 4;
+		break;
+	case GL_STENCIL_INDEX8_EXT:
+		nrb->mesa._BaseFormat = GL_STENCIL_INDEX;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.StencilBits = 8;
+		nrb->cpp = 4;
+		break;
+	case GL_DEPTH24_STENCIL8_EXT:
+		nrb->mesa._BaseFormat = GL_DEPTH_STENCIL_EXT;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.DepthBits = 24;
+		nrb->mesa.StencilBits = 8;
+		nrb->cpp = 4;
+		break;
+	default:
+		return GL_FALSE;
+		break;
+	}
+
+	return GL_TRUE;
+}
+
+static GLboolean
+nouveau_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
+			     GLenum internalFormat,
+			     GLuint width, GLuint height)
+{
+	nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *) rb;
+
+	if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
+		fprintf(stderr, "%s: unknown internalFormat\n", __func__);
+		return GL_FALSE;
+	}
+
+	/* If this buffer isn't statically alloc'd, we may need to ask the
+	 * drm for more memory */
+	if (rb->Width != width || rb->Height != height) {
+		GLuint pitch;
+
+		/* align pitches to 64 bytes */
+		pitch = ((width * nrb->cpp) + 63) & ~63;
+
+		if (nrb->mem)
+			nouveau_mem_free(ctx, nrb->mem);
+		nrb->mem = nouveau_mem_alloc(ctx,
+					     NOUVEAU_MEM_FB |
+					     NOUVEAU_MEM_MAPPED,
+					     pitch * height, 0);
+		if (!nrb->mem)
+			return GL_FALSE;
+
+		/* update nouveau_renderbuffer info */
+		nrb->offset = nouveau_mem_gpu_offset_get(ctx, nrb->mem);
+		nrb->pitch = pitch;
+	}
+
+	rb->Width = width;
+	rb->Height = height;
+	rb->InternalFormat = internalFormat;
+
+	nouveauSpanSetFunctions(nrb);
+
+	return GL_TRUE;
+}
+
+static void
+nouveau_renderbuffer_delete(struct gl_renderbuffer *rb)
+{
+	GET_CURRENT_CONTEXT(ctx);
+	nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *) rb;
+
+	if (nrb->mem)
+		nouveau_mem_free(ctx, nrb->mem);
+	FREE(nrb);
+}
+
+nouveau_renderbuffer_t *
+nouveau_renderbuffer_new(GLenum internalFormat)
+{
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = CALLOC_STRUCT(nouveau_renderbuffer);
+	if (!nrb)
+		return NULL;
+
+	_mesa_init_renderbuffer(&nrb->mesa, 0);
+
+	if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
+		fprintf(stderr, "%s: unknown internalFormat\n", __func__);
+		return GL_FALSE;
+	}
+
+	nrb->mesa.AllocStorage = nouveau_renderbuffer_storage;
+	nrb->mesa.Delete = nouveau_renderbuffer_delete;
+
+	return nrb;
+}
+
+static void
+nouveau_cliprects_renderbuffer_set(nouveauContextPtr nmesa,
+				   nouveau_renderbuffer_t * nrb)
+{
+	nmesa->numClipRects = 1;
+	nmesa->pClipRects = &nmesa->osClipRect;
+	nmesa->osClipRect.x1 = 0;
+	nmesa->osClipRect.y1 = 0;
+	nmesa->osClipRect.x2 = nrb->mesa.Width;
+	nmesa->osClipRect.y2 = nrb->mesa.Height;
+	nmesa->drawX = 0;
+	nmesa->drawY = 0;
+	nmesa->drawW = nrb->mesa.Width;
+	nmesa->drawH = nrb->mesa.Height;
+}
+
+void
+nouveau_window_moved(GLcontext * ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = (nouveau_renderbuffer_t *)
+		ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+	if (!nrb)
+		return;
+
+	nouveau_cliprects_renderbuffer_set(nmesa, nrb);
+
+	/* Viewport depends on window size/position, nouveauCalcViewport
+	 * will take care of calling the hw-specific WindowMoved
+	 */
+	ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
+			     ctx->Viewport.Width, ctx->Viewport.Height);
+	/* Scissor depends on window position */
+	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			    ctx->Scissor.Width, ctx->Scissor.Height);
+}
+
+GLboolean
+nouveau_build_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_renderbuffer_t *color[MAX_DRAW_BUFFERS];
+	nouveau_renderbuffer_t *depth;
+
+	_mesa_update_framebuffer(ctx);
+	_mesa_update_draw_buffer_bounds(ctx);
+
+	color[0] = (nouveau_renderbuffer_t *) fb->_ColorDrawBuffers[0][0];
+	if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped)
+		depth = (nouveau_renderbuffer_t *) fb->_DepthBuffer->Wrapped;
+	else
+		depth = (nouveau_renderbuffer_t *) fb->_DepthBuffer;
+
+	if (!nmesa->hw_func.BindBuffers(nmesa, 1, color, depth))
+		return GL_FALSE;
+	nouveau_window_moved(ctx);
+
+	return GL_TRUE;
+}
+
+static void
+nouveauDrawBuffer(GLcontext *ctx, GLenum buffer)
+{
+	nouveau_build_framebuffer(ctx, ctx->DrawBuffer);
+}
+
+static struct gl_framebuffer *
+nouveauNewFramebuffer(GLcontext *ctx, GLuint name)
+{
+	return _mesa_new_framebuffer(ctx, name);
+}
+
+static struct gl_renderbuffer *
+nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
+{
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = CALLOC_STRUCT(nouveau_renderbuffer);
+	if (!nrb)
+		return NULL;
+
+	_mesa_init_renderbuffer(&nrb->mesa, name);
+
+	nrb->mesa.AllocStorage = nouveau_renderbuffer_storage;
+	nrb->mesa.Delete = nouveau_renderbuffer_delete;
+	return &nrb->mesa;
+}
+
+static void
+nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
+		       struct gl_framebuffer *fb,
+		       struct gl_framebuffer *fbread)
+{
+	if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
+		nouveau_build_framebuffer(ctx, fb);
+	}
+}
+
+static void
+nouveauFramebufferRenderbuffer(GLcontext *ctx, struct gl_framebuffer *fb,
+			       GLenum attachment, struct gl_renderbuffer *rb)
+{
+	_mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
+	nouveau_build_framebuffer(ctx, fb);
+}
+
+static void
+nouveauRenderTexture(GLcontext * ctx, struct gl_framebuffer *fb,
+		     struct gl_renderbuffer_attachment *att)
+{
+}
+
+static void
+nouveauFinishRenderTexture(GLcontext * ctx,
+			   struct gl_renderbuffer_attachment *att)
+{
+}
+
+void
+nouveauInitBufferFuncs(struct dd_function_table *func)
+{
+	func->DrawBuffer = nouveauDrawBuffer;
+
+	func->NewFramebuffer = nouveauNewFramebuffer;
+	func->NewRenderbuffer = nouveauNewRenderbuffer;
+	func->BindFramebuffer = nouveauBindFramebuffer;
+	func->FramebufferRenderbuffer = nouveauFramebufferRenderbuffer;
+	func->RenderTexture = nouveauRenderTexture;
+	func->FinishRenderTexture = nouveauFinishRenderTexture;
+}
+
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fbo.h b/src/mesa/drivers/dri/nouveau/nouveau_fbo.h
new file mode 100644
index 0000000000..787af4e9e4
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fbo.h
@@ -0,0 +1,30 @@
+#ifndef __NOUVEAU_BUFFERS_H__
+#define __NOUVEAU_BUFFERS_H__
+
+#include <stdint.h>
+#include "mtypes.h"
+#include "utils.h"
+#include "renderbuffer.h"
+
+#include "nouveau_mem.h"
+
+typedef struct nouveau_renderbuffer {
+	struct gl_renderbuffer mesa;	/* must be first! */
+
+	nouveau_mem *mem;
+	void *map;
+
+	int cpp;
+	uint32_t offset;
+	uint32_t pitch;
+} nouveau_renderbuffer_t;
+
+extern nouveau_renderbuffer_t *nouveau_renderbuffer_new(GLenum internalFormat);
+extern void nouveau_window_moved(GLcontext *);
+extern GLboolean nouveau_build_framebuffer(GLcontext *,
+					   struct gl_framebuffer *);
+extern nouveau_renderbuffer_t *nouveau_current_draw_buffer(GLcontext *);
+
+extern void nouveauInitBufferFuncs(struct dd_function_table *);
+
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
index e9320918f9..5dc94e0520 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
@@ -57,7 +57,8 @@ void WAIT_RING(nouveauContextPtr nmesa,u_int32_t size)
 		if(nmesa->fifo.put >= fifo_get) {
 			nmesa->fifo.free = nmesa->fifo.max - nmesa->fifo.current;
 			if(nmesa->fifo.free < size+1) {
-				OUT_RING(NV03_FIFO_CMD_JUMP | nmesa->fifo.put_base);
+				OUT_RING(NV03_FIFO_CMD_JUMP |
+					 nmesa->fifo.drm.put_base);
 				if(fifo_get <= RING_SKIPS) {
 					if(nmesa->fifo.put <= RING_SKIPS) /* corner case - will be idle */
 						NV_FIFO_WRITE_PUT(RING_SKIPS + 1);
@@ -98,54 +99,54 @@ void nouveauWaitForIdle(nouveauContextPtr nmesa)
 // here we call the fifo initialization ioctl and fill in stuff accordingly
 GLboolean nouveauFifoInit(nouveauContextPtr nmesa)
 {
-	drm_nouveau_fifo_alloc_t fifo_init;
 	int i, ret;
 
 #ifdef NOUVEAU_RING_DEBUG
 	return GL_TRUE;
 #endif
 
-	fifo_init.fb_ctxdma_handle = NvDmaFB;
-	fifo_init.tt_ctxdma_handle = NvDmaTT;
-	ret=drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_FIFO_ALLOC, &fifo_init, sizeof(fifo_init));
+	nmesa->fifo.drm.fb_ctxdma_handle = NvDmaFB;
+	nmesa->fifo.drm.tt_ctxdma_handle = NvDmaTT;
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_CHANNEL_ALLOC,
+				  &nmesa->fifo.drm, sizeof(nmesa->fifo.drm));
 	if (ret) {
-		FATAL("Fifo initialization ioctl failed (returned %d)\n",ret);
+		FATAL("Fifo initialization ioctl failed (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.cmdbuf, fifo_init.cmdbuf_size, &nmesa->fifo.buffer);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.cmdbuf,
+		     nmesa->fifo.drm.cmdbuf_size, &nmesa->fifo.pushbuf);
 	if (ret) {
-		FATAL("Unable to map the fifo (returned %d)\n",ret);
+		FATAL("Unable to map the fifo (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.ctrl, fifo_init.ctrl_size, &nmesa->fifo.mmio);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.ctrl,
+		     nmesa->fifo.drm.ctrl_size, &nmesa->fifo.mmio);
 	if (ret) {
-		FATAL("Unable to map the control regs (returned %d)\n",ret);
+		FATAL("Unable to map the control regs (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.notifier,
-				   fifo_init.notifier_size,
-				   &nmesa->notifier_block);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.notifier,
+				   nmesa->fifo.drm.notifier_size,
+				   &nmesa->fifo.notifier_block);
 	if (ret) {
-		FATAL("Unable to map the notifier block (returned %d)\n",ret);
+		FATAL("Unable to map the notifier block (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
 	/* Setup our initial FIFO tracking params */
-	nmesa->fifo.channel  = fifo_init.channel;
-	nmesa->fifo.put_base = fifo_init.put_base;
 	nmesa->fifo.current  = 0;
 	nmesa->fifo.put      = 0;
-	nmesa->fifo.max      = (fifo_init.cmdbuf_size >> 2) - 1;
+	nmesa->fifo.max      = (nmesa->fifo.drm.cmdbuf_size >> 2) - 1;
 	nmesa->fifo.free     = nmesa->fifo.max - nmesa->fifo.current;
 
 	for (i=0; i<RING_SKIPS; i++)
 	   OUT_RING(0);
 	nmesa->fifo.free -= RING_SKIPS;
 
-	MESSAGE("Fifo init ok. Using context %d\n", fifo_init.channel);
+	MESSAGE("Fifo init ok. Using context %d\n", nmesa->fifo.drm.channel);
 	return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
index 23325dcea5..956dd549dc 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
@@ -48,14 +48,14 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define NV_FIFO_READ(reg) *(volatile u_int32_t *)(nmesa->fifo.mmio + (reg/4))
 #define NV_FIFO_WRITE(reg,value) *(volatile u_int32_t *)(nmesa->fifo.mmio + (reg/4)) = value;
-#define NV_FIFO_READ_GET() ((NV_FIFO_READ(NV03_FIFO_REGS_DMAGET) - nmesa->fifo.put_base) >> 2)
+#define NV_FIFO_READ_GET() ((NV_FIFO_READ(NV03_FIFO_REGS_DMAGET) - nmesa->fifo.drm.put_base) >> 2)
 #define NV_FIFO_WRITE_PUT(val) do { \
 	if (NOUVEAU_RING_TRACE) {\
 		printf("FIRE_RING : 0x%08x\n", nmesa->fifo.current << 2); \
 		fflush(stdout); \
 		sleep(1); \
 	} \
-	NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.put_base); \
+	NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.drm.put_base); \
 } while(0)
 
 /* 
@@ -110,20 +110,20 @@ nouveau_fifo_remaining-=sz;							\
 		uint32_t* p=(uint32_t*)(ptr);							\
 		int i; printf("OUT_RINGp: (size 0x%x dwords) (%s)\n",sz, __func__); for(i=0;i<sz;i++) printf(" [0x%08x] 0x%08x   %f\n", (nmesa->fifo.current+i) << 2, *(p+i), *((float*)(p+i))); 	\
 	} \
-	memcpy(nmesa->fifo.buffer+nmesa->fifo.current,ptr,(sz)*4);		\
+	memcpy(nmesa->fifo.pushbuf+nmesa->fifo.current,ptr,(sz)*4);		\
 	nmesa->fifo.current+=(sz);						\
 }while(0)
 
 #define OUT_RING(n) do {							\
 if (NOUVEAU_RING_TRACE) \
     printf("OUT_RINGn: [0x%08x] 0x%08x (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-nmesa->fifo.buffer[nmesa->fifo.current++]=(n);					\
+nmesa->fifo.pushbuf[nmesa->fifo.current++]=(n);					\
 }while(0)
 
 #define OUT_RINGf(n) do {							\
 if (NOUVEAU_RING_TRACE) \
     printf("OUT_RINGf: [0x%08x] %.04f (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-*((float*)(nmesa->fifo.buffer+nmesa->fifo.current++))=(n);			\
+*((float*)(nmesa->fifo.pushbuf+nmesa->fifo.current++))=(n);			\
 }while(0)
 
 #define BEGIN_RING_SIZE(subchannel,tag,size) do {					\
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_lock.c b/src/mesa/drivers/dri/nouveau/nouveau_lock.c
index c119d14dd7..aa86c9e783 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_lock.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_lock.c
@@ -44,7 +44,7 @@ void nouveauGetLock( nouveauContextPtr nmesa, GLuint flags )
 {
    __DRIdrawablePrivate *dPriv = nmesa->driDrawable;
    __DRIscreenPrivate *sPriv = nmesa->driScreen;
-   drm_nouveau_sarea_t *sarea = nmesa->sarea;
+   struct drm_nouveau_sarea *sarea = nmesa->sarea;
 
    drmGetLock( nmesa->driFd, nmesa->hHWContext, flags );
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_mem.c b/src/mesa/drivers/dri/nouveau/nouveau_mem.c
new file mode 100644
index 0000000000..35c53268b0
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_mem.c
@@ -0,0 +1,144 @@
+#include "mtypes.h"
+
+#include "nouveau_context.h"
+#include "nouveau_fifo.h"
+#include "nouveau_mem.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
+
+#define MAX_MEMFMT_LENGTH 32768
+
+/* Unstrided blit using NV_MEMORY_TO_MEMORY_FORMAT */
+GLboolean
+nouveau_memformat_flat_emit(GLcontext * ctx,
+			    nouveau_mem * dst, nouveau_mem * src,
+			    GLuint dst_offset, GLuint src_offset,
+			    GLuint size)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	uint32_t src_handle, dst_handle;
+	GLuint count;
+
+	if (src_offset + size > src->size) {
+		MESSAGE("src out of nouveau_mem bounds\n");
+		return GL_FALSE;
+	}
+	if (dst_offset + size > dst->size) {
+		MESSAGE("dst out of nouveau_mem bounds\n");
+		return GL_FALSE;
+	}
+
+	src_handle = (src->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
+	dst_handle = (dst->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
+	src_offset += nouveau_mem_gpu_offset_get(ctx, src);
+	dst_offset += nouveau_mem_gpu_offset_get(ctx, dst);
+
+	BEGIN_RING_SIZE(NvSubMemFormat,
+			NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN, 2);
+	OUT_RING(src_handle);
+	OUT_RING(dst_handle);
+
+	count = (size / MAX_MEMFMT_LENGTH) + 
+		((size % MAX_MEMFMT_LENGTH) ? 1 : 0);
+
+	while (count--) {
+		GLuint length =
+		    (size > MAX_MEMFMT_LENGTH) ? MAX_MEMFMT_LENGTH : size;
+
+		BEGIN_RING_SIZE(NvSubMemFormat,
+				NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
+		OUT_RING(src_offset);
+		OUT_RING(dst_offset);
+		OUT_RING(0);	/* pitch in */
+		OUT_RING(0);	/* pitch out */
+		OUT_RING(length);	/* line length */
+		OUT_RING(1);	/* number of lines */
+		OUT_RING((1 << 8) /* dst_inc */ |(1 << 0) /* src_inc */ );
+		OUT_RING(0);	/* buffer notify? */
+		FIRE_RING();
+
+		src_offset += length;
+		dst_offset += length;
+		size -= length;
+	}
+
+	return GL_TRUE;
+}
+
+void nouveau_mem_free(GLcontext * ctx, nouveau_mem * mem)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	struct drm_nouveau_mem_free memf;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n",
+			__func__, mem->type, (GLuint) mem->offset,
+			(GLuint) mem->size);
+	}
+
+	if (mem->map)
+		drmUnmap(mem->map, mem->size);
+	memf.flags = mem->type;
+	memf.offset = mem->offset;
+	drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf,
+			sizeof(memf));
+	FREE(mem);
+}
+
+nouveau_mem *nouveau_mem_alloc(GLcontext *ctx, uint32_t flags, GLuint size,
+			       GLuint align)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	struct drm_nouveau_mem_alloc mema;
+	nouveau_mem *mem;
+	int ret;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr,
+			"%s: requested: flags=0x%x, size=0x%x, align=0x%x\n",
+			__func__, flags, (GLuint) size, align);
+	}
+
+	mem = CALLOC(sizeof(nouveau_mem));
+	if (!mem)
+		return NULL;
+
+	mema.flags = flags;
+	mema.size = mem->size = size;
+	mema.alignment = align;
+	mem->map = NULL;
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_MEM_ALLOC,
+				  &mema, sizeof(mema));
+	if (ret) {
+		FREE(mem);
+		return NULL;
+	}
+	mem->offset = mema.offset;
+	mem->type = mema.flags;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr,
+			"%s: actual: type=0x%x, offset=0x%x, size=0x%x\n",
+			__func__, mem->type, (GLuint) mem->offset,
+			(GLuint) mem->size);
+	}
+
+	if (flags & NOUVEAU_MEM_MAPPED)
+		ret = drmMap(nmesa->driFd, mema.map_handle, mem->size,
+			     &mem->map);
+	if (ret) {
+		mem->map = NULL;
+		nouveau_mem_free(ctx, mem);
+		mem = NULL;
+	}
+
+	return mem;
+}
+
+uint32_t nouveau_mem_gpu_offset_get(GLcontext * ctx, nouveau_mem * mem)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	return mem->offset;
+}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_mem.h b/src/mesa/drivers/dri/nouveau/nouveau_mem.h
new file mode 100644
index 0000000000..6db73f4708
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_mem.h
@@ -0,0 +1,23 @@
+#ifndef __NOUVEAU_MEM_H__
+#define __NOUVEAU_MEM_H__
+
+typedef struct nouveau_mem_t {
+	int type;
+	uint64_t offset;
+	uint64_t size;
+	void *map;
+} nouveau_mem;
+
+extern nouveau_mem *nouveau_mem_alloc(GLcontext *, uint32_t flags,
+				      GLuint size, GLuint align);
+extern void nouveau_mem_free(GLcontext *, nouveau_mem *);
+extern uint32_t nouveau_mem_gpu_offset_get(GLcontext *, nouveau_mem *);
+
+extern GLboolean nouveau_memformat_flat_emit(GLcontext *,
+					     nouveau_mem *dst,
+					     nouveau_mem *src,
+					     GLuint dst_offset,
+					     GLuint src_offset,
+					     GLuint size);
+
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.c b/src/mesa/drivers/dri/nouveau/nouveau_object.c
index 69f8dbf794..4169530be6 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.c
@@ -7,10 +7,10 @@
 GLboolean nouveauCreateContextObject(nouveauContextPtr nmesa,
 				     uint32_t handle, int class)
 {
-	drm_nouveau_grobj_alloc_t cto;
+	struct drm_nouveau_grobj_alloc cto;
 	int ret;
 
-	cto.channel = nmesa->fifo.channel;
+	cto.channel = nmesa->fifo.drm.channel;
 	cto.handle  = handle;
 	cto.class   = class;
 	ret = drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_GROBJ_ALLOC,
@@ -34,15 +34,22 @@ void nouveauObjectInit(nouveauContextPtr nmesa)
 	nouveauCreateContextObject(nmesa, Nv3D, nmesa->screen->card->class_3d);
 	if (nmesa->screen->card->type>=NV_10) {
 		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV10_CONTEXT_SURFACES_2D);
-		nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT);
 	} else {
 		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D);
 		nouveauCreateContextObject(nmesa, NvCtxSurf3D, NV04_CONTEXT_SURFACES_3D);
+	}
+	if (nmesa->screen->card->type>=NV_11) {
+		nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT);
+	} else {
 		nouveauCreateContextObject(nmesa, NvImageBlit, NV_IMAGE_BLIT);
 	}
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<NV_20)) {
+		nouveauCreateContextObject(nmesa, NvGdiRectText, NV04_GDI_RECTANGLE_TEXT);
+		nouveauCreateContextObject(nmesa, NvRasterOp, NV03_PRIMITIVE_RASTER_OP);
+		nouveauCreateContextObject(nmesa, NvPattern, NV04_IMAGE_PATTERN);
+	}
 	nouveauCreateContextObject(nmesa, NvMemFormat, NV_MEMORY_TO_MEMORY_FORMAT);
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf2D, NvCtxSurf2D);
 	BEGIN_RING_SIZE(NvSubCtxSurf2D, NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY0, 2);
 	OUT_RING(NvDmaFB);
@@ -54,8 +61,35 @@ void nouveauObjectInit(nouveauContextPtr nmesa)
 	BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_OPERATION, 1);
 	OUT_RING(3); /* SRCCOPY */
 
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<NV_20)) {
+		nouveauObjectOnSubchannel(nmesa, NvSubPattern, NvPattern);
+
+		BEGIN_RING_SIZE(NvSubPattern, NV04_IMAGE_PATTERN_COLOR_FORMAT, 4);
+		OUT_RING(1);	/* A16R5G6B5 */
+		OUT_RING(1);	/* little endian */
+		OUT_RING(0);	/* 8x8 */
+		OUT_RING(1);	/* monochrome */
+
+		nouveauObjectOnSubchannel(nmesa, NvSubRasterOp, NvRasterOp);
+
+		BEGIN_RING_SIZE(NvSubRasterOp, NV03_PRIMITIVE_RASTER_OP_DMA_NOTIFY, 1);
+		OUT_RING(NvDmaFB);
+
+		nouveauObjectOnSubchannel(nmesa, NvSubGdiRectText, NvGdiRectText);
+
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_SET_DMA_NOTIFY, 1);
+		OUT_RING(NvDmaFB);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_PATTERN, 2);
+		OUT_RING(NvPattern);
+		OUT_RING(NvRasterOp);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1);
+		OUT_RING(NvCtxSurf2D);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_FORMAT, 2);
+		OUT_RING(1);	/* X1R5G5B5 */
+		OUT_RING(1);	/* little endian */
+	}
+
 	nouveauObjectOnSubchannel(nmesa, NvSubMemFormat, NvMemFormat);
-#endif
 
 	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.h b/src/mesa/drivers/dri/nouveau/nouveau_object.h
index 8c72d014da..10859de119 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.h
@@ -3,8 +3,6 @@
 
 #include "nouveau_context.h"
 
-#define ALLOW_MULTI_SUBCHANNEL
-
 void nouveauObjectInit(nouveauContextPtr nmesa);
 
 enum DMAObjects {
@@ -13,6 +11,9 @@ enum DMAObjects {
 	NvImageBlit		= 0x80000021,
 	NvMemFormat		= 0x80000022,
 	NvCtxSurf3D		= 0x80000023,
+	NvGdiRectText		= 0x80000024,
+	NvPattern		= 0x80000025,
+	NvRasterOp		= 0x80000026,
 	NvDmaFB			= 0xD0FB0001,
 	NvDmaTT			= 0xD0AA0001,
 	NvSyncNotify		= 0xD0000001,
@@ -24,6 +25,9 @@ enum DMASubchannel {
 	NvSubImageBlit	= 1,
 	NvSubMemFormat	= 2,
 	NvSubCtxSurf3D	= 3,
+	NvSubGdiRectText= 4,
+	NvSubPattern	= 5,
+	NvSubRasterOp	= 6,
 	NvSub3D		= 7,
 };
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.c b/src/mesa/drivers/dri/nouveau/nouveau_query.c
index 0154140069..e5c1750a8e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_query.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_query.c
@@ -167,7 +167,8 @@ nouveauQueryInitFuncs(GLcontext *ctx)
 	if (nmesa->screen->card->type < NV_20)
 		return;
 
-	nmesa->query_object_max = (0x4000 / 32);
+	/* Reserve half the notifier block for use as query objects */
+	nmesa->query_object_max = (nmesa->fifo.drm.notifier_size / 2) / 32;
 	nmesa->queryNotifier =
 		nouveau_notifier_new(ctx, NvQueryNotify,
 					  nmesa->query_object_max);
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.c b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
index bc7f39b042..065aa81746 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_screen.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
@@ -132,10 +132,11 @@ nouveauCreateBuffer(__DRIscreenPrivate *driScrnPriv,
                     GLboolean isPixmap)
 {
 	nouveauScreenPtr screen = (nouveauScreenPtr) driScrnPriv->private;
-	nouveau_renderbuffer  *nrb;
+	nouveau_renderbuffer_t *nrb;
 	struct gl_framebuffer *fb;
 	const GLboolean swAccum = mesaVis->accumRedBits > 0;
-	const GLboolean swStencil = mesaVis->stencilBits > 0 && mesaVis->depthBits != 24;
+	const GLboolean swStencil = (mesaVis->stencilBits > 0 &&
+				     mesaVis->depthBits != 24);
 	GLenum color_format = screen->fbFormat == 4 ? GL_RGBA8 : GL_RGB5;
 
 	if (isPixmap)
@@ -146,44 +147,26 @@ nouveauCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 		return GL_FALSE;
 
 	/* Front buffer */
-	nrb = nouveau_renderbuffer_new(color_format,
-				       driScrnPriv->pFB + screen->frontOffset,
-				       screen->frontOffset,
-				       screen->frontPitch * screen->fbFormat,
-				       driDrawPriv);
-	nouveauSpanSetFunctions(nrb, mesaVis);
+	nrb = nouveau_renderbuffer_new(color_format);
 	_mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &nrb->mesa);
 
-	if (0 /* unified buffers if we choose to support them.. */) {
-	} else {
-		if (mesaVis->doubleBufferMode) {
-			nrb = nouveau_renderbuffer_new(color_format, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &nrb->mesa);
-		}
+	if (mesaVis->doubleBufferMode) {
+		nrb = nouveau_renderbuffer_new(color_format);
+		_mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &nrb->mesa);
+	}
 
-		if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH24_STENCIL8_EXT, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-			_mesa_add_renderbuffer(fb, BUFFER_STENCIL, &nrb->mesa);
-		} else if (mesaVis->depthBits == 24) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT24, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		} else if (mesaVis->depthBits == 16) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT16, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		}
+	if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH24_STENCIL8_EXT);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
+		_mesa_add_renderbuffer(fb, BUFFER_STENCIL, &nrb->mesa);
+	} else
+	if (mesaVis->depthBits == 24) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT24);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
+	} else
+	if (mesaVis->depthBits == 16) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT16);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
 	}
 
 	_mesa_add_soft_renderbuffers(fb,
@@ -328,7 +311,7 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc
 	static const __DRIversion ddx_expected = { 1, 2, 0 };
 	static const __DRIversion dri_expected = { 4, 0, 0 };
 	static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL };
-#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 7
+#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10
 #error nouveau_drm.h version doesn't match expected version
 #endif
 	dri_interface = interface;
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.c b/src/mesa/drivers/dri/nouveau/nouveau_span.c
index 6e3f9fadf4..d62830ff53 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_span.c
@@ -37,8 +37,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define HAVE_HW_STENCIL_SPANS	0
 #define HAVE_HW_STENCIL_PIXELS	0
 
-static char *fake_span[1280*1024*4];
-
 #define HW_CLIPLOOP()							\
    do {									\
       int _nc = nmesa->numClipRects;					\
@@ -50,11 +48,10 @@ static char *fake_span[1280*1024*4];
 
 #define LOCAL_VARS							\
    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);			\
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer *)rb;		\
+   nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *)rb;		\
    GLuint height = nrb->mesa.Height;					\
    GLubyte *map = (GLubyte *)(nrb->map ? nrb->map : nrb->mem->map) +    \
 	 (nmesa->drawY * nrb->pitch) + (nmesa->drawX * nrb->cpp);       \
-   map = fake_span; \
    GLuint p;								\
    (void) p;
 
@@ -70,7 +67,7 @@ static char *fake_span[1280*1024*4];
  * Color buffers
  */
 
-/* RGB565 */ 
+/* RGB565 */
 #define SPANTMP_PIXEL_FMT GL_RGB
 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
 
@@ -89,40 +86,38 @@ static char *fake_span[1280*1024*4];
 #define GET_PTR(X,Y) (map + (Y)*nrb->pitch + (X)*nrb->cpp)
 #include "spantmp2.h"
 
-static void
-nouveauSpanRenderStart( GLcontext *ctx )
+static void nouveauSpanRenderStart(GLcontext * ctx)
 {
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   FIRE_RING();
-   LOCK_HARDWARE(nmesa);
-   nouveauWaitForIdleLocked( nmesa );
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	FIRE_RING();
+	LOCK_HARDWARE(nmesa);
+	nouveauWaitForIdleLocked(nmesa);
 }
 
-static void
-nouveauSpanRenderFinish( GLcontext *ctx )
+static void nouveauSpanRenderFinish(GLcontext * ctx)
 {
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   _swrast_flush( ctx );
-   nouveauWaitForIdleLocked( nmesa );
-   UNLOCK_HARDWARE( nmesa );
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	_swrast_flush(ctx);
+	nouveauWaitForIdleLocked(nmesa);
+	UNLOCK_HARDWARE(nmesa);
 }
 
-void nouveauSpanInitFunctions( GLcontext *ctx )
+void nouveauSpanInitFunctions(GLcontext * ctx)
 {
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart	= nouveauSpanRenderStart;
-   swdd->SpanRenderFinish	= nouveauSpanRenderFinish;
+	struct swrast_device_driver *swdd =
+	    _swrast_GetDeviceDriverReference(ctx);
+	swdd->SpanRenderStart = nouveauSpanRenderStart;
+	swdd->SpanRenderFinish = nouveauSpanRenderFinish;
 }
 
 
 /**
  * Plug in the Get/Put routines for the given driRenderbuffer.
  */
-void
-nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis)
+void nouveauSpanSetFunctions(nouveau_renderbuffer_t * nrb)
 {
-   if (nrb->mesa._ActualFormat == GL_RGBA8)
-      nouveauInitPointers_ARGB8888(&nrb->mesa);
-   else // if (nrb->mesa._ActualFormat == GL_RGB5)
-      nouveauInitPointers_RGB565(&nrb->mesa);
+	if (nrb->mesa._ActualFormat == GL_RGBA8)
+		nouveauInitPointers_ARGB8888(&nrb->mesa);
+	else			// if (nrb->mesa._ActualFormat == GL_RGB5)
+		nouveauInitPointers_RGB565(&nrb->mesa);
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.h b/src/mesa/drivers/dri/nouveau/nouveau_span.h
index bc39ecd17b..d3f31c9cb2 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_span.h
@@ -30,10 +30,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define __NOUVEAU_SPAN_H__
 
 #include "drirenderbuffer.h"
-#include "nouveau_buffers.h"
+#include "nouveau_fbo.h"
 
-extern void nouveauSpanInitFunctions( GLcontext *ctx );
-extern void nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis);
+extern void nouveauSpanInitFunctions(GLcontext *ctx);
+extern void nouveauSpanSetFunctions(nouveau_renderbuffer_t *nrb);
 
 #endif /* __NOUVEAU_SPAN_H__ */
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c
index 7cb805902a..f618dcfc99 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c
@@ -162,6 +162,8 @@ void nouveauDDInitState(nouveauContextPtr nmesa)
             nv04InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
             break;
         case NV_10:
+        case NV_11:
+        case NV_17:
             nv10InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
             break;
         case NV_20:
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.c b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
index 1d1eeede18..2ca038f4f8 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
@@ -28,36 +28,36 @@
 #include "vblank.h" /* for DO_USLEEP */
 
 #include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
 #include "nouveau_fifo.h"
-#include "nouveau_reg.h"
+#include "nouveau_mem.h"
 #include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
 #include "nouveau_sync.h"
 
 #define NOTIFIER(__v) \
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); \
-	volatile uint32_t *__v = (void*)nmesa->notifier_block + notifier->offset
+	volatile uint32_t *__v = (void*)nmesa->fifo.notifier_block + \
+				 notifier->offset
 
-drm_nouveau_notifier_alloc_t *
+struct drm_nouveau_notifierobj_alloc *
 nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	drm_nouveau_notifier_alloc_t *notifier;
+	struct drm_nouveau_notifierobj_alloc *notifier;
 	int ret;
 
 #ifdef NOUVEAU_RING_DEBUG
 	return NULL;
 #endif
-
-	notifier = CALLOC_STRUCT(drm_nouveau_notifier_alloc);
+	notifier = CALLOC_STRUCT(drm_nouveau_notifierobj_alloc);
 	if (!notifier)
 		return NULL;
 
-	notifier->channel = nmesa->fifo.channel;
+	notifier->channel = nmesa->fifo.drm.channel;
 	notifier->handle  = handle;
 	notifier->count   = count;
-	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_NOTIFIER_ALLOC,
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_NOTIFIEROBJ_ALLOC,
 				  notifier, sizeof(*notifier));
 	if (ret) {
 		MESSAGE("Failed to create notifier 0x%08x: %d\n", handle, ret);
@@ -69,14 +69,16 @@ nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
 }
 
 void
-nouveau_notifier_destroy(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier)
+nouveau_notifier_destroy(GLcontext *ctx,
+			 struct drm_nouveau_notifierobj_alloc *notifier)
 {
 	/*XXX: free notifier object.. */
 	FREE(notifier);
 }
 
 void
-nouveau_notifier_reset(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier,
+nouveau_notifier_reset(GLcontext *ctx,
+		       struct drm_nouveau_notifierobj_alloc *notifier,
 		       GLuint id)
 {
 	NOTIFIER(n);
@@ -93,7 +95,8 @@ nouveau_notifier_reset(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier,
 }
 
 GLuint
-nouveau_notifier_status(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier,
+nouveau_notifier_status(GLcontext *ctx,
+			struct drm_nouveau_notifierobj_alloc *notifier,
 			GLuint id)
 {
 	NOTIFIER(n);
@@ -103,7 +106,8 @@ nouveau_notifier_status(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier,
 
 GLuint
 nouveau_notifier_return_val(GLcontext *ctx,
-			    drm_nouveau_notifier_alloc_t *notifier, GLuint id)
+			    struct drm_nouveau_notifierobj_alloc *notifier,
+			    GLuint id)
 {
 	NOTIFIER(n);
 
@@ -112,8 +116,8 @@ nouveau_notifier_return_val(GLcontext *ctx,
 
 GLboolean
 nouveau_notifier_wait_status(GLcontext *ctx,
-			     drm_nouveau_notifier_alloc_t *notifier, GLuint id,
-			     GLuint status, GLuint timeout)
+			     struct drm_nouveau_notifierobj_alloc *notifier,
+			     GLuint id, GLuint status, GLuint timeout)
 {
 	NOTIFIER(n);
 	unsigned int time = 0;
@@ -146,7 +150,8 @@ nouveau_notifier_wait_status(GLcontext *ctx,
 
 void
 nouveau_notifier_wait_nop(GLcontext *ctx,
-			  drm_nouveau_notifier_alloc_t *notifier, GLuint subc)
+			  struct drm_nouveau_notifierobj_alloc *notifier,
+			  GLuint subc)
 {
 	NOTIFIER(n);
 	GLboolean ret;
@@ -184,11 +189,9 @@ GLboolean nouveauSyncInitFuncs(GLcontext *ctx)
 	 */
 	BEGIN_RING_CACHE(NvSub3D, 0x180, 1);
 	OUT_RING_CACHE  (NvSyncNotify);
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	BEGIN_RING_SIZE(NvSubMemFormat,
 	      		NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1);
 	OUT_RING       (NvSyncNotify);
-#endif
 
 	return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.h b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
index b56cc5fb54..fc37efbe6b 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
@@ -28,8 +28,6 @@
 #ifndef __NOUVEAU_SYNC_H__
 #define __NOUVEAU_SYNC_H__
 
-#include "nouveau_buffers.h"
-
 #define NV_NOTIFIER_SIZE                                                      32
 #define NV_NOTIFY_TIME_0                                              0x00000000
 #define NV_NOTIFY_TIME_1                                              0x00000004
@@ -47,22 +45,24 @@
 #define NV_NOTIFY                                                     0x00000104
 #define NV_NOTIFY_STYLE_WRITE_ONLY                                             0
 
-extern drm_nouveau_notifier_alloc_t *
+extern struct drm_nouveau_notifierobj_alloc *
 nouveau_notifier_new(GLcontext *, GLuint handle, GLuint count);
 extern void 
-nouveau_notifier_destroy(GLcontext *, drm_nouveau_notifier_alloc_t *);
+nouveau_notifier_destroy(GLcontext *, struct drm_nouveau_notifierobj_alloc *);
 extern void
-nouveau_notifier_reset(GLcontext *, drm_nouveau_notifier_alloc_t *, GLuint id);
+nouveau_notifier_reset(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
+		       GLuint id);
 extern GLuint
-nouveau_notifier_status(GLcontext *, drm_nouveau_notifier_alloc_t *, GLuint id);
+nouveau_notifier_status(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
+			GLuint id);
 extern GLuint
-nouveau_notifier_return_val(GLcontext *, drm_nouveau_notifier_alloc_t *,
+nouveau_notifier_return_val(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 			    GLuint id);
 extern GLboolean
-nouveau_notifier_wait_status(GLcontext *, drm_nouveau_notifier_alloc_t *,
+nouveau_notifier_wait_status(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 			     GLuint id, GLuint status, GLuint timeout);
 extern void
-nouveau_notifier_wait_nop(GLcontext *ctx, drm_nouveau_notifier_alloc_t *,
+nouveau_notifier_wait_nop(GLcontext *ctx, struct drm_nouveau_notifierobj_alloc *,
 			  GLuint subc);
 
 extern GLboolean nouveauSyncInitFuncs(GLcontext *ctx);
diff --git a/src/mesa/drivers/dri/nouveau/nv04_state.c b/src/mesa/drivers/dri/nouveau/nv04_state.c
index 25df3d2a62..d3031aa5b1 100644
--- a/src/mesa/drivers/dri/nouveau/nv04_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv04_state.c
@@ -451,8 +451,8 @@ static GLboolean nv04InitCard(nouveauContextPtr nmesa)
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv04BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
+		nouveau_renderbuffer_t **color,
+		nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	uint32_t depth_pitch=(depth?depth->pitch:0+15)&~15+16;
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state.c b/src/mesa/drivers/dri/nouveau/nv10_state.c
index 5f304ccab9..4a3646ea4f 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state.c
@@ -111,9 +111,62 @@ static void nv10BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfac
 	OUT_RING_CACHE(dfactorRGB);
 }
 
+static void nv10ClearBuffer(GLcontext *ctx, nouveau_renderbuffer_t *buffer, int fill, int mask)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	int dimensions;
+
+	if (!buffer) {
+		return;
+	}
+
+	/* Surface that we will work on */
+	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf2D, NvCtxSurf2D);
+
+	BEGIN_RING_SIZE(NvSubCtxSurf2D, NV10_CONTEXT_SURFACES_2D_FORMAT, 4);
+	OUT_RING(0x0b); /* Y32 color format */
+	OUT_RING((buffer->pitch<<16)|buffer->pitch);
+	OUT_RING(buffer->offset);
+	OUT_RING(buffer->offset);
+
+	/* Now clear a rectangle */
+	dimensions = ((buffer->mesa.Height)<<16) | (buffer->mesa.Width);
+
+	nouveauObjectOnSubchannel(nmesa, NvSubGdiRectText, NvGdiRectText);
+
+	BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1);
+	OUT_RING(3);	/* SRCCOPY */
+
+	BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL, 5);
+	OUT_RING(0);	/* top left */
+	OUT_RING(dimensions);	/* bottom right */
+	OUT_RING(fill);
+	OUT_RING(0);	/* top left */
+	OUT_RING(dimensions);	/* bottom right */
+}
+
 static void nv10Clear(GLcontext *ctx, GLbitfield mask)
 {
-	/* TODO */
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (mask & (BUFFER_BIT_FRONT_LEFT)) {
+		nv10ClearBuffer(ctx, nmesa->color_buffer[0],
+			nmesa->clear_color_value, 0xffffffff);
+	}
+	if (mask & (BUFFER_BIT_BACK_LEFT)) {
+		nv10ClearBuffer(ctx, nmesa->color_buffer[1],
+			nmesa->clear_color_value, 0xffffffff);
+	}
+	/* FIXME: check depth bits */
+	if (mask & (BUFFER_BIT_DEPTH)) {
+		nv10ClearBuffer(ctx, nmesa->depth_buffer,
+			nmesa->clear_value, 0xffffff00);
+	}
+	/* FIXME: check about stencil? */
+	if (mask & (BUFFER_BIT_STENCIL)) {
+		nv10ClearBuffer(ctx, nmesa->depth_buffer,
+			nmesa->clear_value, 0x000000ff);
+	}
 }
 
 static void nv10ClearColor(GLcontext *ctx, const GLfloat color[4])
@@ -697,8 +750,7 @@ static GLboolean nv10InitCard(nouveauContextPtr nmesa)
 	BEGIN_RING_SIZE(NvSub3D, 0x03f4, 1);
 	OUT_RING(0);
 
-	/* not for nv10, only for >= nv11 */
-	if ((nmesa->screen->card->id>>4) >= 0x11) {
+	if (nmesa->screen->card->type >= NV_11) {
 	        BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
         	OUT_RING(0);
 	        OUT_RING(1);
@@ -710,12 +762,17 @@ static GLboolean nv10InitCard(nouveauContextPtr nmesa)
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	GLuint pitch, format, depth_pitch;
 
+	/* Store buffer pointers in context */
+	nmesa->color_buffer[0] = color[0];
+	nmesa->color_buffer[1] = color[1];
+	nmesa->depth_buffer = depth;
+
 	w = color[0]->mesa.Width;
 	h = color[0]->mesa.Height;
 	x = nmesa->drawX;
@@ -739,11 +796,11 @@ static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
 	OUT_RING_CACHE(depth ? depth->offset : color[0]->offset);
 
 	/* Always set to bottom left of buffer */
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
+	/*BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
 	OUT_RING_CACHEf (0.0);
 	OUT_RING_CACHEf ((GLfloat) h);
 	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (0.0);
+	OUT_RING_CACHEf (0.0);*/
 
 	return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
index 4576c1ede4..611469b6e4 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
@@ -58,7 +58,7 @@ static void nv10ResetLineStipple( GLcontext *ctx );
 
 static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t primitive,uint32_t size)
 {
-	if (nmesa->screen->card->type==NV_10)
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
 		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
 	else if (nmesa->screen->card->type==NV_20)
 		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1);
@@ -66,7 +66,7 @@ static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t pri
 		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_BEGIN_END,1);
 	OUT_RING(primitive);
 
-	if (nmesa->screen->card->type==NV_10)
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
 		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA|NONINC_METHOD,size);
 	else if (nmesa->screen->card->type==NV_20)
 		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_VERTEX_DATA|NONINC_METHOD,size);
@@ -76,7 +76,7 @@ static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t pri
 
 inline void nv10FinishPrimitive(struct nouveau_context *nmesa)
 {
-	if (nmesa->screen->card->type==NV_10)
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
 		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
 	else if (nmesa->screen->card->type==NV_20)
 		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1);
@@ -454,7 +454,7 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	/* 
 	 * Tell the hardware about the vertex format
 	 */
-	if (nmesa->screen->card->type==NV_10) {
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) {
 		int size;
 
 #define NV_VERTEX_ATTRIBUTE_TYPE_FLOAT 2
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state.c b/src/mesa/drivers/dri/nouveau/nv20_state.c
index 3d8d83a865..6b583980a4 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state.c
@@ -568,10 +568,10 @@ static void nv20Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 	   y += nmesa->drawY;
 	}
 
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 1);
+        /*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 1);
         OUT_RING_CACHE((w << 16) | x );
         BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1, 1);
-        OUT_RING_CACHE((h << 16) | y );
+        OUT_RING_CACHE((h << 16) | y );*/
 
 }
 
@@ -728,8 +728,8 @@ static GLboolean nv20InitCard(nouveauContextPtr nmesa)
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	GLuint pitch, format, depth_pitch;
@@ -764,11 +764,11 @@ static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color,
 	}
 
 	/* Always set to bottom left of buffer */
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
+	/*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
 	OUT_RING_CACHEf (0.0);
 	OUT_RING_CACHEf ((GLfloat) h);
 	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (0.0);
+	OUT_RING_CACHEf (0.0);*/
 
 	return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nv30_state.c b/src/mesa/drivers/dri/nouveau/nv30_state.c
index 9b010954b3..cd3ee98688 100644
--- a/src/mesa/drivers/dri/nouveau/nv30_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv30_state.c
@@ -905,9 +905,9 @@ static GLboolean nv40InitCard(nouveauContextPtr nmesa)
 	return GL_TRUE;
 }
 
-static GLboolean nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
+static GLboolean
+nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
+		nouveau_renderbuffer_t **color, nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 
diff --git a/src/mesa/drivers/dri/nouveau/nv50_state.c b/src/mesa/drivers/dri/nouveau/nv50_state.c
index 818e268615..a9236f093c 100644
--- a/src/mesa/drivers/dri/nouveau/nv50_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv50_state.c
@@ -584,9 +584,9 @@ static GLboolean nv50InitCard(nouveauContextPtr nmesa)
 	return GL_FALSE;
 }
 
-static GLboolean nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
-      				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+static GLboolean
+nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
+		nouveau_renderbuffer_t **color, nouveau_renderbuffer_t *depth)
 {
 	return GL_FALSE;
 }
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index 2115799b9b..1d975ecd57 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -772,9 +772,11 @@ static void r200LineWidth( GLcontext *ctx, GLfloat widthf )
    R200_STATECHANGE( rmesa, set );
 
    /* Line width is stored in U6.4 format.
+    * Same min/max limits for AA, non-AA lines.
     */
    rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] &= ~0xffff;
-   rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)(ctx->Line._Width * 16.0);
+   rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)
+      (CLAMP(widthf, ctx->Const.MinLineWidth, ctx->Const.MaxLineWidth) * 16.0);
 
    if ( widthf > 1.0 ) {
       rmesa->hw.set.cmd[SET_SE_CNTL] |=  R200_WIDELINE_ENABLE;
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 6615bc79fb..be6909724a 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -568,19 +568,10 @@ struct r300_vertex_shader_fragment {
 	union {
 		GLuint d[VSF_MAX_FRAGMENT_LENGTH];
 		float f[VSF_MAX_FRAGMENT_LENGTH];
-		VERTEX_SHADER_INSTRUCTION i[VSF_MAX_FRAGMENT_LENGTH / 4];
+		GLuint i[VSF_MAX_FRAGMENT_LENGTH];
 	} body;
 };
 
-#define VSF_DEST_PROGRAM	0x0
-#define VSF_DEST_MATRIX0	0x200
-#define VSF_DEST_MATRIX1	0x204
-#define VSF_DEST_MATRIX2	0x208
-#define VSF_DEST_VECTOR0	0x20c
-#define VSF_DEST_VECTOR1	0x20d
-#define VSF_DEST_UNKNOWN1	0x400
-#define VSF_DEST_UNKNOWN2	0x406
-
 struct r300_vertex_shader_state {
 	struct r300_vertex_shader_fragment program;
 };
diff --git a/src/mesa/drivers/dri/r300/r300_emit.c b/src/mesa/drivers/dri/r300/r300_emit.c
index 229439dfa8..424bf44e59 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.c
+++ b/src/mesa/drivers/dri/r300/r300_emit.c
@@ -299,13 +299,14 @@ GLuint r300VAPOutputCntl0(GLcontext * ctx, GLuint OutputsWritten)
 	if (OutputsWritten & (1 << VERT_RESULT_COL1))
 		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT;
 
-#if 0
-	if (OutputsWritten & (1 << VERT_RESULT_BFC0))
-		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT;
-
-	if (OutputsWritten & (1 << VERT_RESULT_BFC1))
-		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT;
+	if (OutputsWritten & (1 << VERT_RESULT_BFC0)
+	    || OutputsWritten & (1 << VERT_RESULT_BFC1))
+		ret |=
+		    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT |
+		    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT |
+		    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT;
 
+#if 0
 	if (OutputsWritten & (1 << VERT_RESULT_FOGC)) ;
 #endif
 
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index e59919be49..1baa74c526 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -282,9 +282,32 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 #define R300_VAP_PVS_UPLOAD_ADDRESS         0x2200
 #       define R300_PVS_UPLOAD_PROGRAM           0x00000000
+/* gap */
 #       define R300_PVS_UPLOAD_PARAMETERS        0x00000200
+/* gap */
+#       define R300_PVS_UPLOAD_CLIP_PLANE0       0x00000400
+#       define R300_PVS_UPLOAD_CLIP_PLANE1       0x00000401
+#       define R300_PVS_UPLOAD_CLIP_PLANE2       0x00000402
+#       define R300_PVS_UPLOAD_CLIP_PLANE3       0x00000403
+#       define R300_PVS_UPLOAD_CLIP_PLANE4       0x00000404
+#       define R300_PVS_UPLOAD_CLIP_PLANE5       0x00000405
 #       define R300_PVS_UPLOAD_POINTSIZE         0x00000406
 
+/*
+ * These are obsolete defines form r300_context.h, but they might give some
+ * clues when investigating the addresses further...
+ */
+#if 0
+#define VSF_DEST_PROGRAM        0x0
+#define VSF_DEST_MATRIX0        0x200
+#define VSF_DEST_MATRIX1        0x204
+#define VSF_DEST_MATRIX2        0x208
+#define VSF_DEST_VECTOR0        0x20c
+#define VSF_DEST_VECTOR1        0x20d
+#define VSF_DEST_UNKNOWN1       0x400
+#define VSF_DEST_UNKNOWN2       0x406
+#endif
+
 /* gap */
 
 #define R300_VAP_PVS_UPLOAD_DATA            0x2208
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index b5cf21d644..088216c76e 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -82,6 +82,8 @@ static void r300BlendColor(GLcontext * ctx, const GLfloat cf[4])
 
 	rmesa->hw.blend_color.cmd[1] = PACK_COLOR_8888(color[3], color[0],
 						       color[1], color[2]);
+	rmesa->hw.blend_color.cmd[2] = 0;
+	rmesa->hw.blend_color.cmd[3] = 0;
 }
 
 /**
@@ -317,20 +319,34 @@ static void r300UpdateCulling(GLcontext * ctx)
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 	uint32_t val = 0;
 
-	R300_STATECHANGE(r300, cul);
 	if (ctx->Polygon.CullFlag) {
-		if (ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK)
-			val = R300_CULL_FRONT | R300_CULL_BACK;
-		else if (ctx->Polygon.CullFaceMode == GL_FRONT)
+		switch (ctx->Polygon.CullFaceMode) {
+		case GL_FRONT:
 			val = R300_CULL_FRONT;
-		else
+			break;
+		case GL_BACK:
 			val = R300_CULL_BACK;
+			break;
+		case GL_FRONT_AND_BACK:
+			val = R300_CULL_FRONT | R300_CULL_BACK;
+			break;
+		default:
+			break;
+		}
+	}
 
-		if (ctx->Polygon.FrontFace == GL_CW)
-			val |= R300_FRONT_FACE_CW;
-		else
-			val |= R300_FRONT_FACE_CCW;
+	switch (ctx->Polygon.FrontFace) {
+	case GL_CW:
+		val |= R300_FRONT_FACE_CW;
+		break;
+	case GL_CCW:
+		val |= R300_FRONT_FACE_CCW;
+		break;
+	default:
+		break;
 	}
+
+	R300_STATECHANGE(r300, cul);
 	r300->hw.cul.cmd[R300_CUL_CULL] = val;
 }
 
@@ -344,6 +360,20 @@ static void r300SetEarlyZState(GLcontext * ctx)
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
 	R300_STATECHANGE(r300, zstencil_format);
+	switch (ctx->Visual.depthBits) {
+	case 16:
+		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z;
+		break;
+	case 24:
+		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z;
+		break;
+	default:
+		fprintf(stderr, "Error: Unsupported depth %d... exiting\n", ctx->Visual.depthBits);
+		_mesa_exit(-1);
+	}
+
+	// r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32;
+
 	if (ctx->Color.AlphaEnabled && ctx->Color.AlphaFunc != GL_ALWAYS)
 		/* disable early Z */
 		r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
@@ -355,6 +385,9 @@ static void r300SetEarlyZState(GLcontext * ctx)
 			/* disable early Z */
 			r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
 	}
+
+	r300->hw.zstencil_format.cmd[3] = 0x00000003;
+	r300->hw.zstencil_format.cmd[4] = 0x00000000;
 }
 
 static void r300SetAlphaState(GLcontext * ctx)
@@ -403,6 +436,7 @@ static void r300SetAlphaState(GLcontext * ctx)
 
 	R300_STATECHANGE(r300, at);
 	r300->hw.at.cmd[R300_AT_ALPHA_TEST] = pp_misc;
+	r300->hw.at.cmd[R300_AT_UNKNOWN] = 0;
 
 	r300SetEarlyZState(ctx);
 }
@@ -513,6 +547,9 @@ static void r300UpdatePolygonMode(GLcontext * ctx)
 		R300_STATECHANGE(r300, polygon_mode);
 		r300->hw.polygon_mode.cmd[1] = hw_mode;
 	}
+
+	r300->hw.polygon_mode.cmd[2] = 0x00000001;
+	r300->hw.polygon_mode.cmd[3] = 0x00000000;
 }
 
 /**
@@ -696,8 +733,8 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 static void r300PointSize(GLcontext * ctx, GLfloat size)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-
-	size = ctx->Point._Size;
+        /* same size limits for AA, non-AA points */
+	size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
 
 	R300_STATECHANGE(r300, ps);
 	r300->hw.ps.cmd[R300_PS_POINTSIZE] =
@@ -712,8 +749,9 @@ static void r300LineWidth(GLcontext * ctx, GLfloat widthf)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
-	widthf = ctx->Line._Width;
-
+	widthf = CLAMP(widthf,
+                       ctx->Const.MinPointSize,
+                       ctx->Const.MaxPointSize);
 	R300_STATECHANGE(r300, lcntl);
 	r300->hw.lcntl.cmd[1] =
 	    R300_LINE_CNT_HO | R300_LINE_CNT_VE | (int)(widthf * 6.0);
@@ -762,6 +800,7 @@ static void r300ShadeModel(GLcontext * ctx, GLenum mode)
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 
 	R300_STATECHANGE(rmesa, shade);
+	rmesa->hw.shade.cmd[1] = 0x00000002;
 	switch (mode) {
 	case GL_FLAT:
 		rmesa->hw.shade.cmd[2] = R300_RE_SHADE_MODEL_FLAT;
@@ -772,6 +811,8 @@ static void r300ShadeModel(GLcontext * ctx, GLenum mode)
 	default:
 		return;
 	}
+	rmesa->hw.shade.cmd[3] = 0x00000000;
+	rmesa->hw.shade.cmd[4] = 0x00000000;
 }
 
 static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
@@ -1526,30 +1567,31 @@ static void r300SetupDefaultVertexProgram(r300ContextPtr rmesa)
 
 	for (i = VERT_ATTRIB_POS; i < VERT_ATTRIB_MAX; i++) {
 		if (rmesa->state.sw_tcl_inputs[i] != -1) {
-			prog->program.body.i[program_end].op = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT);
-			prog->program.body.i[program_end].src[0] = VSF_REG(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end].src[1] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end].src[2] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]);
-			program_end++;
+			prog->program.body.i[program_end + 0] = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT);
+			prog->program.body.i[program_end + 1] = VSF_REG(rmesa->state.sw_tcl_inputs[i]);
+			prog->program.body.i[program_end + 2] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]);
+			prog->program.body.i[program_end + 3] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]);
+			program_end += 4;
 		}
 	}
 
-	prog->program.length = program_end * 4;
+	prog->program.length = program_end;
 
-	r300SetupVertexProgramFragment(rmesa, VSF_DEST_PROGRAM, &(prog->program));
+	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM,
+				       &(prog->program));
 	inst_count = (prog->program.length / 4) - 1;
 
 	R300_STATECHANGE(rmesa, pvs);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_1] =
-	  (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
+	    (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_2] =
-	  (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
-	  (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
+	    (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
+	    (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_3] =
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
+	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
+	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
 }
 
 static void r300SetupRealVertexProgram(r300ContextPtr rmesa)
@@ -1570,7 +1612,7 @@ static void r300SetupRealVertexProgram(r300ContextPtr rmesa)
 	bump_vpu_count(rmesa->hw.vpp.cmd, param_count);
 	param_count /= 4;
 
-	r300SetupVertexProgramFragment(rmesa, VSF_DEST_PROGRAM, &(prog->program));
+	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM, &(prog->program));
 	inst_count = (prog->program.length / 4) - 1;
 
 	R300_STATECHANGE(rmesa, pvs);
@@ -1848,15 +1890,10 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300->hw.unk4260.cmd[2] = r300PackFloat32(0.0);
 	r300->hw.unk4260.cmd[3] = r300PackFloat32(1.0);
 
-	r300->hw.shade.cmd[1] = 0x00000002;
 	r300ShadeModel(ctx, ctx->Light.ShadeModel);
-	r300->hw.shade.cmd[3] = 0x00000000;
-	r300->hw.shade.cmd[4] = 0x00000000;
 
 	r300PolygonMode(ctx, GL_FRONT, ctx->Polygon.FrontMode);
 	r300PolygonMode(ctx, GL_BACK, ctx->Polygon.BackMode);
-	r300->hw.polygon_mode.cmd[2] = 0x00000001;
-	r300->hw.polygon_mode.cmd[3] = 0x00000000;
 	r300->hw.zbias_cntl.cmd[1] = 0x00000000;
 
 	r300PolygonOffset(ctx, ctx->Polygon.OffsetFactor,
@@ -1887,14 +1924,11 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300Fogfv(ctx, GL_FOG_COLOR, ctx->Fog.Color);
 	r300Fogfv(ctx, GL_FOG_COORDINATE_SOURCE_EXT, NULL);
 
-	r300->hw.at.cmd[R300_AT_UNKNOWN] = 0;
 	r300->hw.unk4BD8.cmd[1] = 0;
 
 	r300->hw.unk4E00.cmd[1] = 0;
 
 	r300BlendColor(ctx, ctx->Color.BlendColor);
-	r300->hw.blend_color.cmd[2] = 0;
-	r300->hw.blend_color.cmd[3] = 0;
 
 	/* Again, r300ClearBuffer uses this */
 	r300->hw.cb.cmd[R300_CB_OFFSET] =
@@ -1925,25 +1959,6 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300->hw.unk4EA0.cmd[1] = 0x00000000;
 	r300->hw.unk4EA0.cmd[2] = 0xffffffff;
 
-	switch (ctx->Visual.depthBits) {
-	case 16:
-		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z;
-		break;
-	case 24:
-		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z;
-		break;
-	default:
-		fprintf(stderr, "Error: Unsupported depth %d... exiting\n",
-			ctx->Visual.depthBits);
-		_mesa_exit(-1);
-
-	}
-	/* z compress? */
-	//r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32;
-
-	r300->hw.zstencil_format.cmd[3] = 0x00000003;
-	r300->hw.zstencil_format.cmd[4] = 0x00000000;
-
 	r300->hw.zb.cmd[R300_ZB_OFFSET] =
 	    r300->radeon.radeonScreen->depthOffset +
 	    r300->radeon.radeonScreen->fbLocation;
diff --git a/src/mesa/drivers/dri/r300/r300_texmem.c b/src/mesa/drivers/dri/r300/r300_texmem.c
index 38f0da8b7c..723601ac4a 100644
--- a/src/mesa/drivers/dri/r300/r300_texmem.c
+++ b/src/mesa/drivers/dri/r300/r300_texmem.c
@@ -505,7 +505,7 @@ int r300UploadTexImages(r300ContextPtr rmesa, r300TexObjPtr t, GLuint face)
 			t->base.lastLevel);
 	}
 
-	if (!t || t->base.totalSize == 0)
+	if (t->base.totalSize == 0)
 		return 0;
 
 	if (RADEON_DEBUG & DEBUG_SYNC) {
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 16dddf6557..7d4e8c9511 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -29,6 +29,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \file
  *
  * \author Aapo Tahkola <aet@rasterburn.org>
+ *
+ * \author Oliver McFadden <z3ro.geek@gmail.com>
+ *
+ * For a description of the vertex program instruction set see r300_reg.h.
  */
 
 #include "glheader.h"
@@ -55,54 +59,58 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #error Cannot change these!
 #endif
 
-#define SCALAR_FLAG (1<<31)
-#define FLAG_MASK (1<<31)
-#define OP_MASK	(0xf)		/* we are unlikely to have more than 15 */
-#define OPN(operator, ip) {#operator, OPCODE_##operator, ip}
-
-static struct {
-	char *name;
-	int opcode;
-	unsigned long ip;	/* number of input operands and flags */
-} op_names[] = {
-	/* *INDENT-OFF* */
-	OPN(ABS, 1),
-	OPN(ADD, 2),
-	OPN(ARL, 1 | SCALAR_FLAG),
-	OPN(DP3, 2),
-	OPN(DP4, 2),
-	OPN(DPH, 2),
-	OPN(DST, 2),
-	OPN(EX2, 1 | SCALAR_FLAG),
-	OPN(EXP, 1 | SCALAR_FLAG),
-	OPN(FLR, 1),
-	OPN(FRC, 1),
-	OPN(LG2, 1 | SCALAR_FLAG),
-	OPN(LIT, 1),
-	OPN(LOG, 1 | SCALAR_FLAG),
-	OPN(MAD, 3),
-	OPN(MAX, 2),
-	OPN(MIN, 2),
-	OPN(MOV, 1),
-	OPN(MUL, 2),
-	OPN(POW, 2 | SCALAR_FLAG),
-	OPN(RCP, 1 | SCALAR_FLAG),
-	OPN(RSQ, 1 | SCALAR_FLAG),
-	OPN(SGE, 2),
-	OPN(SLT, 2),
-	OPN(SUB, 2),
-	OPN(SWZ, 1),
-	OPN(XPD, 2),
-	OPN(RCC, 0),	//extra
-	OPN(PRINT, 0),
-	OPN(END, 0)
-	/* *INDENT-ON* */
-};
-
-#undef OPN
+/* TODO: Get rid of t_src_class call */
+#define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \
+		       ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \
+			 t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \
+			(t_src_class(a.File) == VSF_IN_CLASS_ATTR && \
+			 t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \
+
+#define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+
+#define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+
+#define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
+				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+
+#define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+
+#define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+
+#define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    SWIZZLE_ONE, SWIZZLE_ONE, \
+				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+
+/* DP4 version seems to trigger some hw peculiarity */
+//#define PREFER_DP4
+
+#define FREE_TEMPS() \
+	do { \
+		if(u_temp_i < vp->num_temporaries) { \
+			WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \
+			vp->native = GL_FALSE; \
+		} \
+		u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \
+	} while (0)
 
 int r300VertexProgUpdateParams(GLcontext * ctx,
-			       struct r300_vertex_program_cont *vp, float *dst)
+			       struct r300_vertex_program_cont *vp,
+			       float *dst)
 {
 	int pi;
 	struct gl_vertex_program *mesa_vp = &vp->mesa_program;
@@ -234,8 +242,8 @@ static void vp_dump_inputs(struct r300_vertex_program *vp, char *caller)
 	int i;
 
 	if (vp == NULL) {
-		fprintf(stderr, "vp null in call to %s from %s\n", __FUNCTION__,
-			caller);
+		fprintf(stderr, "vp null in call to %s from %s\n",
+			__FUNCTION__, caller);
 		return;
 	}
 
@@ -276,6 +284,8 @@ static unsigned long t_src_index(struct r300_vertex_program *vp,
 	}
 }
 
+/* these two functions should probably be merged... */
+
 static unsigned long t_src(struct r300_vertex_program *vp,
 			   struct prog_src_register *src)
 {
@@ -294,7 +304,9 @@ static unsigned long t_src(struct r300_vertex_program *vp,
 static unsigned long t_src_scalar(struct r300_vertex_program *vp,
 				  struct prog_src_register *src)
 {
-
+	/* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
+	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
+	 */
 	return MAKE_VSF_SOURCE(t_src_index(vp, src),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
@@ -306,128 +318,741 @@ static unsigned long t_src_scalar(struct r300_vertex_program *vp,
 	    (src->RelAddr << 4);
 }
 
-static unsigned long t_opcode(enum prog_opcode opcode)
+static GLboolean valid_dst(struct r300_vertex_program *vp,
+			   struct prog_dst_register *dst)
 {
+	if (dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1) {
+		return GL_FALSE;
+	} else if (dst->File == PROGRAM_ADDRESS) {
+		assert(dst->Index == 0);
+	}
 
-	switch (opcode) {
-	/* *INDENT-OFF* */
-	case OPCODE_ARL: return R300_VPI_OUT_OP_ARL;
-	case OPCODE_DST: return R300_VPI_OUT_OP_DST;
-	case OPCODE_EX2: return R300_VPI_OUT_OP_EX2;
-	case OPCODE_EXP: return R300_VPI_OUT_OP_EXP;
-	case OPCODE_FRC: return R300_VPI_OUT_OP_FRC;
-	case OPCODE_LG2: return R300_VPI_OUT_OP_LG2;
-	case OPCODE_LOG: return R300_VPI_OUT_OP_LOG;
-	case OPCODE_MAX: return R300_VPI_OUT_OP_MAX;
-	case OPCODE_MIN: return R300_VPI_OUT_OP_MIN;
-	case OPCODE_MUL: return R300_VPI_OUT_OP_MUL;
-	case OPCODE_RCP: return R300_VPI_OUT_OP_RCP;
-	case OPCODE_RSQ: return R300_VPI_OUT_OP_RSQ;
-	case OPCODE_SGE: return R300_VPI_OUT_OP_SGE;
-	case OPCODE_SLT: return R300_VPI_OUT_OP_SLT;
-	case OPCODE_DP4: return R300_VPI_OUT_OP_DOT;
-	/* *INDENT-ON* */
+	return GL_TRUE;
+}
 
-	default:
-		fprintf(stderr, "%s: Should not be called with opcode %d!",
-			__FUNCTION__, opcode);
-	}
-	_mesa_exit(-1);
-	return 0;
+/*
+ * Instruction    Inputs  Output   Description
+ * -----------    ------  ------   --------------------------------
+ * ABS            v       v        absolute value
+ * ADD            v,v     v        add
+ * ARL            s       a        address register load
+ * DP3            v,v     ssss     3-component dot product
+ * DP4            v,v     ssss     4-component dot product
+ * DPH            v,v     ssss     homogeneous dot product
+ * DST            v,v     v        distance vector
+ * EX2            s       ssss     exponential base 2
+ * EXP            s       v        exponential base 2 (approximate)
+ * FLR            v       v        floor
+ * FRC            v       v        fraction
+ * LG2            s       ssss     logarithm base 2
+ * LIT            v       v        compute light coefficients
+ * LOG            s       v        logarithm base 2 (approximate)
+ * MAD            v,v,v   v        multiply and add
+ * MAX            v,v     v        maximum
+ * MIN            v,v     v        minimum
+ * MOV            v       v        move
+ * MUL            v,v     v        multiply
+ * POW            s,s     ssss     exponentiate
+ * RCP            s       ssss     reciprocal
+ * RSQ            s       ssss     reciprocal square root
+ * SGE            v,v     v        set on greater than or equal
+ * SLT            v,v     v        set on less than
+ * SUB            v,v     v        subtract
+ * SWZ            v       v        extended swizzle
+ * XPD            v,v     v        cross product
+ *
+ * Table X.5:  Summary of vertex program instructions.  "v" indicates a
+ * floating-point vector input or output, "s" indicates a floating-point
+ * scalar input, "ssss" indicates a scalar output replicated across a
+ * 4-component result vector, and "a" indicates a single address register
+ * component.
+ */
+
+static GLuint *t_opcode_abs(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
+			    t_src_class(src[0].File),
+			    (!src[0].
+			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[3] = 0;
+
+	return inst;
 }
 
-static unsigned long op_operands(enum prog_opcode opcode)
+static GLuint *t_opcode_add(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	int i;
+	unsigned long hw_op;
 
-	/* Can we trust mesas opcodes to be in order ? */
-	for (i = 0; i < sizeof(op_names) / sizeof(*op_names); i++)
-		if (op_names[i].opcode == opcode)
-			return op_names[i].ip;
+#if 1
+	hw_op = (src[0].File == PROGRAM_TEMPORARY
+		 && src[1].File ==
+		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = ONE_SRC_0;
+	inst[2] = t_src(vp, &src[0]);
+	inst[3] = t_src(vp, &src[1]);
+#else
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
 
-	fprintf(stderr, "op %d not found in op_names\n", opcode);
-	_mesa_exit(-1);
-	return 0;
+#endif
+
+	return inst;
 }
 
-static GLboolean valid_dst(struct r300_vertex_program *vp,
-			   struct prog_dst_register *dst)
+static GLuint *t_opcode_arl(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	if (dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1) {
-		return GL_FALSE;
-	} else if (dst->File == PROGRAM_ADDRESS) {
-		assert(dst->Index == 0);
-	}
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ARL, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
 
-	return GL_TRUE;
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
 }
 
-/* TODO: Get rid of t_src_class call */
-#define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \
-		       ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \
-			 t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \
-			(t_src_class(a.File) == VSF_IN_CLASS_ATTR && \
-			 t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \
+static GLuint *t_opcode_dp3(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO}
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+			    SWIZZLE_ZERO, t_src_class(src[0].File),
+			    src[0].
+			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+
+	inst[2] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+			    SWIZZLE_ZERO, t_src_class(src[1].File),
+			    src[1].
+			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+	    (src[1].RelAddr << 4);
+
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
 
-#define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+static GLuint *t_opcode_dp4(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
 
-#define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
 
-#define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+	return inst;
+}
 
-#define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+static GLuint *t_opcode_dph(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W}
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+			    VSF_IN_COMPONENT_ONE, t_src_class(src[0].File),
+			    src[0].
+			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
 
-#define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+static GLuint *t_opcode_dst(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_DST, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
 
-#define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
 
-/* DP4 version seems to trigger some hw peculiarity */
-//#define PREFER_DP4
+	return inst;
+}
 
-#define FREE_TEMPS() \
-	do { \
-		if(u_temp_i < vp->num_temporaries) { \
-			WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \
-			vp->native = GL_FALSE; \
-		} \
-		u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \
-	} while (0)
+static GLuint *t_opcode_ex2(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_EX2, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
 
-static void r300TranslateVertexShader(struct r300_vertex_program *vp,
-				      struct prog_instruction *vpi)
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_exp(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_EXP, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_flr(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3], int *u_temp_i)
+{
+	/* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W}
+	   ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, *u_temp_i,
+			t_dst_mask(vpi->DstReg.WriteMask),
+			VSF_OUT_CLASS_TMP);
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+	inst += 4;
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
+	    MAKE_VSF_SOURCE(*u_temp_i, VSF_IN_COMPONENT_X,
+			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
+			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
+			    /* Not 100% sure about this */
+			    (!src[0].
+			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE
+			    /*VSF_FLAG_ALL */ );
+
+	inst[3] = ZERO_SRC_0;
+	(*u_temp_i)--;
+
+	return inst;
+}
+
+static GLuint *t_opcode_frc(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_lg2(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	// LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X}
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+			    t_src_class(src[0].File),
+			    src[0].
+			    NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_lit(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	/* NOTE: Users swizzling might not work. */
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+	inst[3] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  VSF_IN_COMPONENT_ZERO,	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	return inst;
+}
+
+static GLuint *t_opcode_log(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_LOG, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_mad(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
 {
-	int i, cur_reg = 0;
-	VERTEX_SHADER_INSTRUCTION *o_inst;
-	unsigned long operands;
-	int are_srcs_scalar;
 	unsigned long hw_op;
-	/* Initial value should be last tmp reg that hw supports.
-	   Strangely enough r300 doesnt mind even though these would be out of range.
-	   Smart enough to realize that it doesnt need it? */
-	int u_temp_i = VSF_MAX_FRAGMENT_TEMPS - 1;
-	struct prog_src_register src[3];
 
-	vp->pos_end = 0;	/* Not supported yet */
-	vp->program.length = 0;
-	/*vp->num_temporaries=mesa_vp->Base.NumTemporaries; */
+	hw_op = (src[0].File == PROGRAM_TEMPORARY
+		 && src[1].File == PROGRAM_TEMPORARY
+		 && src[2].File ==
+		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = t_src(vp, &src[2]);
+
+	return inst;
+}
+
+static GLuint *t_opcode_max(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
+
+static GLuint *t_opcode_min(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_MIN, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
+
+static GLuint *t_opcode_mov(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
+
+#if 1
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+#else
+	hw_op =
+	    (src[0].File ==
+	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] = ZERO_SRC_0;
+#endif
+
+	return inst;
+}
+
+static GLuint *t_opcode_mul(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	unsigned long hw_op;
+
+	// HW mul can take third arg but appears to have some other limitations.
+
+	hw_op = (src[0].File == PROGRAM_TEMPORARY
+		 && src[1].File ==
+		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
+
+static GLuint *t_opcode_pow(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_POW, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = t_src_scalar(vp, &src[1]);
+
+	return inst;
+}
+
+static GLuint *t_opcode_rcp(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_RCP, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_rsq(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_RSQ, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src_scalar(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+
+	return inst;
+}
+
+static GLuint *t_opcode_sge(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_SGE, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
+
+static GLuint *t_opcode_slt(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_SLT, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = t_src(vp, &src[1]);
+	inst[3] = ZERO_SRC_1;
+
+	return inst;
+}
+
+static GLuint *t_opcode_sub(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	unsigned long hw_op;
+
+	//ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
+
+#if 1
+	hw_op = (src[0].File == PROGRAM_TEMPORARY
+		 && src[1].File ==
+		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+			    t_src_class(src[1].File),
+			    (!src[1].
+			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[1].RelAddr << 4);
+#else
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] =
+	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+			    t_src_class(src[1].File),
+			    (!src[1].
+			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[1].RelAddr << 4);
+	inst[3] = 0;
+#endif
+
+	return inst;
+}
+
+static GLuint *t_opcode_swz(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3])
+{
+	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
+
+#if 1
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ZERO_SRC_0;
+	inst[3] = ZERO_SRC_0;
+#else
+	hw_op =
+	    (src[0].File ==
+	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
+	    R300_VPI_OUT_OP_MAD;
+
+	inst[0] =
+	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+	inst[1] = t_src(vp, &src[0]);
+	inst[2] = ONE_SRC_0;
+	inst[3] = ZERO_SRC_0;
+#endif
+
+	return inst;
+}
+
+static GLuint *t_opcode_xpd(struct r300_vertex_program *vp,
+			    struct prog_instruction *vpi, GLuint * inst,
+			    struct prog_src_register src[3], int *u_temp_i)
+{
+	/* mul r0, r1.yzxw, r2.zxyw
+	   mad r0, -r2.yzxw, r1.zxyw, r0
+	   NOTE: might need MAD_2
+	 */
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, *u_temp_i,
+			t_dst_mask(vpi->DstReg.WriteMask),
+			VSF_OUT_CLASS_TMP);
+
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+				  t_src_class(src[1].File),
+				  src[1].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+	inst[3] = ZERO_SRC_1;
+	inst += 4;
+	(*u_temp_i)--;
+
+	inst[0] =
+	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, t_dst_index(vp, &vpi->DstReg),
+			t_dst_mask(vpi->DstReg.WriteMask),
+			t_dst_class(vpi->DstReg.File));
+
+	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+				  t_src_class(src[1].File),
+				  (!src[1].
+				   NegateBase) ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL :
+				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+	inst[3] =
+	    MAKE_VSF_SOURCE(*u_temp_i + 1, VSF_IN_COMPONENT_X,
+			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
+			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
+			    VSF_FLAG_NONE);
+
+	return inst;
+}
+
+static void t_inputs_outputs(struct r300_vertex_program *vp)
+{
+	int i;
+	int cur_reg = 0;
 
 	for (i = 0; i < VERT_ATTRIB_MAX; i++)
 		vp->inputs[i] = -1;
@@ -437,39 +1062,71 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 	assert(vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS));
 
-	/* Assign outputs */
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS))
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS)) {
 		vp->outputs[VERT_RESULT_HPOS] = cur_reg++;
+	}
 
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_PSIZ))
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_PSIZ)) {
 		vp->outputs[VERT_RESULT_PSIZ] = cur_reg++;
+	}
 
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL0))
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL0)) {
 		vp->outputs[VERT_RESULT_COL0] = cur_reg++;
+	}
 
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL1))
-		vp->outputs[VERT_RESULT_COL1] = cur_reg++;
-
-#if 0				/* Not supported yet */
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0))
-		vp->outputs[VERT_RESULT_BFC0] = cur_reg++;
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL1)) {
+		vp->outputs[VERT_RESULT_COL1] =
+		    vp->outputs[VERT_RESULT_COL0] + 1;
+		cur_reg = vp->outputs[VERT_RESULT_COL1] + 1;
+	}
 
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1))
-		vp->outputs[VERT_RESULT_BFC1] = cur_reg++;
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0)) {
+		vp->outputs[VERT_RESULT_BFC0] =
+		    vp->outputs[VERT_RESULT_COL0] + 2;
+		cur_reg = vp->outputs[VERT_RESULT_BFC0] + 2;
+	}
 
-	if (vp->key.OutputsWritten & (1 << VERT_RESULT_FOGC))
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1)) {
+		vp->outputs[VERT_RESULT_BFC1] =
+		    vp->outputs[VERT_RESULT_COL0] + 3;
+		cur_reg = vp->outputs[VERT_RESULT_BFC1] + 1;
+	}
+#if 0
+	if (vp->key.OutputsWritten & (1 << VERT_RESULT_FOGC)) {
 		vp->outputs[VERT_RESULT_FOGC] = cur_reg++;
+	}
 #endif
 
-	for (i = VERT_RESULT_TEX0; i <= VERT_RESULT_TEX7; i++)
-		if (vp->key.OutputsWritten & (1 << i))
+	for (i = VERT_RESULT_TEX0; i <= VERT_RESULT_TEX7; i++) {
+		if (vp->key.OutputsWritten & (1 << i)) {
 			vp->outputs[i] = cur_reg++;
+		}
+	}
+}
+
+static void r300TranslateVertexShader(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi)
+{
+	int i;
+	GLuint *inst;
+	unsigned long num_operands;
+	/* Initial value should be last tmp reg that hw supports.
+	   Strangely enough r300 doesnt mind even though these would be out of range.
+	   Smart enough to realize that it doesnt need it? */
+	int u_temp_i = VSF_MAX_FRAGMENT_TEMPS - 1;
+	struct prog_src_register src[3];
 
+	vp->pos_end = 0;	/* Not supported yet */
+	vp->program.length = 0;
+	/*vp->num_temporaries=mesa_vp->Base.NumTemporaries; */
 	vp->translated = GL_TRUE;
 	vp->native = GL_TRUE;
 
-	o_inst = vp->program.body.i;
-	for (; vpi->Opcode != OPCODE_END; vpi++, o_inst++) {
+	t_inputs_outputs(vp);
+
+	for (inst = vp->program.body.i; vpi->Opcode != OPCODE_END;
+	     vpi++, inst += 4) {
+
 		FREE_TEMPS();
 
 		if (!valid_dst(vp, &vpi->DstReg)) {
@@ -478,61 +1135,62 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 			vpi->DstReg.Index = u_temp_i;
 		}
 
-		operands = op_operands(vpi->Opcode);
-		are_srcs_scalar = operands & SCALAR_FLAG;
-		operands &= OP_MASK;
+		num_operands = _mesa_num_inst_src_regs(vpi->Opcode);
 
-		for (i = 0; i < operands; i++)
+		/* copy the sources (src) from mesa into a local variable... is this needed? */
+		for (i = 0; i < num_operands; i++) {
 			src[i] = vpi->SrcReg[i];
+		}
 
-		if (operands == 3) {	/* TODO: scalars */
+		if (num_operands == 3) {	/* TODO: scalars */
 			if (CMP_SRCS(src[1], src[2])
 			    || CMP_SRCS(src[0], src[2])) {
-				o_inst->op =
-				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i,
-						VSF_FLAG_ALL,
+				inst[0] =
+				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
+						u_temp_i, VSF_FLAG_ALL,
 						VSF_OUT_CLASS_TMP);
 
-				o_inst->src[0] =
-				    MAKE_VSF_SOURCE(t_src_index(vp, &src[2]),
+				inst[1] =
+				    MAKE_VSF_SOURCE(t_src_index
+						    (vp, &src[2]),
 						    SWIZZLE_X, SWIZZLE_Y,
 						    SWIZZLE_Z, SWIZZLE_W,
-						    t_src_class(src[2].File),
-						    VSF_FLAG_NONE) | (src[2].
-								      RelAddr <<
-								      4);
+						    t_src_class(src[2].
+								File),
+						    VSF_FLAG_NONE) |
+				    (src[2].RelAddr << 4);
 
-				o_inst->src[1] = ZERO_SRC_2;
-				o_inst->src[2] = ZERO_SRC_2;
-				o_inst++;
+				inst[2] = ZERO_SRC_2;
+				inst[3] = ZERO_SRC_2;
+				inst += 4;
 
 				src[2].File = PROGRAM_TEMPORARY;
 				src[2].Index = u_temp_i;
 				src[2].RelAddr = 0;
 				u_temp_i--;
 			}
-
 		}
 
-		if (operands >= 2) {
+		if (num_operands >= 2) {
 			if (CMP_SRCS(src[1], src[0])) {
-				o_inst->op =
-				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i,
-						VSF_FLAG_ALL,
+				inst[0] =
+				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
+						u_temp_i, VSF_FLAG_ALL,
 						VSF_OUT_CLASS_TMP);
 
-				o_inst->src[0] =
-				    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+				inst[1] =
+				    MAKE_VSF_SOURCE(t_src_index
+						    (vp, &src[0]),
 						    SWIZZLE_X, SWIZZLE_Y,
 						    SWIZZLE_Z, SWIZZLE_W,
-						    t_src_class(src[0].File),
-						    VSF_FLAG_NONE) | (src[0].
-								      RelAddr <<
-								      4);
+						    t_src_class(src[0].
+								File),
+						    VSF_FLAG_NONE) |
+				    (src[0].RelAddr << 4);
 
-				o_inst->src[1] = ZERO_SRC_0;
-				o_inst->src[2] = ZERO_SRC_0;
-				o_inst++;
+				inst[2] = ZERO_SRC_0;
+				inst[3] = ZERO_SRC_0;
+				inst += 4;
 
 				src[0].File = PROGRAM_TEMPORARY;
 				src[0].Index = u_temp_i;
@@ -541,517 +1199,103 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 			}
 		}
 
-		/* These ops need special handling. */
 		switch (vpi->Opcode) {
-		case OPCODE_POW:
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_POW,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src_scalar(vp, &src[0]);
-			o_inst->src[1] = ZERO_SRC_0;
-			o_inst->src[2] = t_src_scalar(vp, &src[1]);
-			goto next;
-
-		case OPCODE_MOV:	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
-		case OPCODE_SWZ:
-#if 1
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = ZERO_SRC_0;
-			o_inst->src[2] = ZERO_SRC_0;
-#else
-			hw_op =
-			    (src[0].File ==
-			     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-			    R300_VPI_OUT_OP_MAD;
-
-			o_inst->op =
-			    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = ONE_SRC_0;
-			o_inst->src[2] = ZERO_SRC_0;
-#endif
-
-			goto next;
-
+		case OPCODE_ABS:
+			inst = t_opcode_abs(vp, vpi, inst, src);
+			break;
 		case OPCODE_ADD:
-#if 1
-			hw_op = (src[0].File == PROGRAM_TEMPORARY &&
-				 src[1].File ==
-				 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-			    R300_VPI_OUT_OP_MAD;
-
-			o_inst->op =
-			    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = ONE_SRC_0;
-			o_inst->src[1] = t_src(vp, &src[0]);
-			o_inst->src[2] = t_src(vp, &src[1]);
-#else
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = t_src(vp, &src[1]);
-			o_inst->src[2] = ZERO_SRC_1;
-
-#endif
-			goto next;
-
-		case OPCODE_MAD:
-			hw_op = (src[0].File == PROGRAM_TEMPORARY &&
-				 src[1].File == PROGRAM_TEMPORARY &&
-				 src[2].File ==
-				 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-			    R300_VPI_OUT_OP_MAD;
-
-			o_inst->op =
-			    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = t_src(vp, &src[1]);
-			o_inst->src[2] = t_src(vp, &src[2]);
-			goto next;
-
-		case OPCODE_MUL:	/* HW mul can take third arg but appears to have some other limitations. */
-			hw_op = (src[0].File == PROGRAM_TEMPORARY &&
-				 src[1].File ==
-				 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-			    R300_VPI_OUT_OP_MAD;
-
-			o_inst->op =
-			    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = t_src(vp, &src[1]);
-
-			o_inst->src[2] = ZERO_SRC_1;
-			goto next;
-
-		case OPCODE_DP3:	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO}
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 2)),
-					    SWIZZLE_ZERO,
-					    t_src_class(src[0].File),
-					    src[0].
-					    NegateBase ? VSF_FLAG_XYZ :
-					    VSF_FLAG_NONE) | (src[0].
-							      RelAddr << 4);
-
-			o_inst->src[1] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 2)),
-					    SWIZZLE_ZERO,
-					    t_src_class(src[1].File),
-					    src[1].
-					    NegateBase ? VSF_FLAG_XYZ :
-					    VSF_FLAG_NONE) | (src[1].
-							      RelAddr << 4);
-
-			o_inst->src[2] = ZERO_SRC_1;
-			goto next;
-
-		case OPCODE_SUB:	//ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
-#if 1
-			hw_op = (src[0].File == PROGRAM_TEMPORARY &&
-				 src[1].File ==
-				 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-			    R300_VPI_OUT_OP_MAD;
-
-			o_inst->op =
-			    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = ONE_SRC_0;
-			o_inst->src[2] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 2)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 3)),
-					    t_src_class(src[1].File),
-					    (!src[1].
-					     NegateBase) ? VSF_FLAG_ALL :
-					    VSF_FLAG_NONE) | (src[1].
-							      RelAddr << 4);
-#else
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 2)),
-					    t_swizzle(GET_SWZ
-						      (src[1].Swizzle, 3)),
-					    t_src_class(src[1].File),
-					    (!src[1].
-					     NegateBase) ? VSF_FLAG_ALL :
-					    VSF_FLAG_NONE) | (src[1].
-							      RelAddr << 4);
-			o_inst->src[2] = 0;
-#endif
-			goto next;
-
-		case OPCODE_ABS:	//MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 2)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 3)),
-					    t_src_class(src[0].File),
-					    (!src[0].
-					     NegateBase) ? VSF_FLAG_ALL :
-					    VSF_FLAG_NONE) | (src[0].
-							      RelAddr << 4);
-			o_inst->src[2] = 0;
-			goto next;
-
+			inst = t_opcode_add(vp, vpi, inst, src);
+			break;
+		case OPCODE_ARL:
+			inst = t_opcode_arl(vp, vpi, inst, src);
+			break;
+		case OPCODE_DP3:
+			inst = t_opcode_dp3(vp, vpi, inst, src);
+			break;
+		case OPCODE_DP4:
+			inst = t_opcode_dp4(vp, vpi, inst, src);
+			break;
+		case OPCODE_DPH:
+			inst = t_opcode_dph(vp, vpi, inst, src);
+			break;
+		case OPCODE_DST:
+			inst = t_opcode_dst(vp, vpi, inst, src);
+			break;
+		case OPCODE_EX2:
+			inst = t_opcode_ex2(vp, vpi, inst, src);
+			break;
+		case OPCODE_EXP:
+			inst = t_opcode_exp(vp, vpi, inst, src);
+			break;
 		case OPCODE_FLR:
-			/* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W}
-			   ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */
-
-			o_inst->op = MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, u_temp_i,
-						 t_dst_mask(vpi->DstReg.
-							    WriteMask),
-						 VSF_OUT_CLASS_TMP);
-
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = ZERO_SRC_0;
-			o_inst->src[2] = ZERO_SRC_0;
-			o_inst++;
-
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] = t_src(vp, &src[0]);
-			o_inst->src[1] = MAKE_VSF_SOURCE(u_temp_i,
-							 VSF_IN_COMPONENT_X,
-							 VSF_IN_COMPONENT_Y,
-							 VSF_IN_COMPONENT_Z,
-							 VSF_IN_COMPONENT_W,
-							 VSF_IN_CLASS_TMP,
-							 /* Not 100% sure about this */
-							 (!src[0].
-							  NegateBase) ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE
-							 /*VSF_FLAG_ALL */ );
-
-			o_inst->src[2] = ZERO_SRC_0;
-			u_temp_i--;
-			goto next;
-
-		case OPCODE_LG2:	// LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X}
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_LG2,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_src_class(src[0].File),
-					    src[0].
-					    NegateBase ? VSF_FLAG_ALL :
-					    VSF_FLAG_NONE) | (src[0].
-							      RelAddr << 4);
-			o_inst->src[1] = ZERO_SRC_0;
-			o_inst->src[2] = ZERO_SRC_0;
-			goto next;
-
-		case OPCODE_LIT:	//LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_LIT,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-			/* NOTE: Users swizzling might not work. */
-			o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-							 VSF_IN_COMPONENT_ZERO,	// z
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-							 t_src_class(src[0].
-								     File),
-							 src[0].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[0].RelAddr << 4);
-			o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-							 VSF_IN_COMPONENT_ZERO,	// z
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-							 t_src_class(src[0].
-								     File),
-							 src[0].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[0].RelAddr << 4);
-			o_inst->src[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-							 VSF_IN_COMPONENT_ZERO,	// z
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-							 t_src_class(src[0].
-								     File),
-							 src[0].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[0].RelAddr << 4);
-			goto next;
-
-		case OPCODE_DPH:	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W}
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] =
-			    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 0)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 1)),
-					    t_swizzle(GET_SWZ
-						      (src[0].Swizzle, 2)),
-					    VSF_IN_COMPONENT_ONE,
-					    t_src_class(src[0].File),
-					    src[0].
-					    NegateBase ? VSF_FLAG_XYZ :
-					    VSF_FLAG_NONE) | (src[0].
-							      RelAddr << 4);
-			o_inst->src[1] = t_src(vp, &src[1]);
-			o_inst->src[2] = ZERO_SRC_1;
-			goto next;
-
-		case OPCODE_XPD:
-			/* mul r0, r1.yzxw, r2.zxyw
-			   mad r0, -r2.yzxw, r1.zxyw, r0
-			   NOTE: might need MAD_2
-			 */
-
-			o_inst->op = MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, u_temp_i,
-						 t_dst_mask(vpi->DstReg.
-							    WriteMask),
-						 VSF_OUT_CLASS_TMP);
-
-			o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-							 t_src_class(src[0].
-								     File),
-							 src[0].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[0].RelAddr << 4);
-
-			o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
-							 t_src_class(src[1].
-								     File),
-							 src[1].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[1].RelAddr << 4);
-
-			o_inst->src[2] = ZERO_SRC_1;
-			o_inst++;
-			u_temp_i--;
-
-			o_inst->op =
-			    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD,
-					t_dst_index(vp, &vpi->DstReg),
-					t_dst_mask(vpi->DstReg.WriteMask),
-					t_dst_class(vpi->DstReg.File));
-
-			o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-							 t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
-							 t_src_class(src[1].
-								     File),
-							 (!src[1].
-							  NegateBase) ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[1].RelAddr << 4);
-
-			o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-							 t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-							 t_src_class(src[0].
-								     File),
-							 src[0].
-							 NegateBase ?
-							 VSF_FLAG_ALL :
-							 VSF_FLAG_NONE) |
-			    (src[0].RelAddr << 4);
-
-			o_inst->src[2] = MAKE_VSF_SOURCE(u_temp_i + 1,
-							 VSF_IN_COMPONENT_X,
-							 VSF_IN_COMPONENT_Y,
-							 VSF_IN_COMPONENT_Z,
-							 VSF_IN_COMPONENT_W,
-							 VSF_IN_CLASS_TMP,
-							 VSF_FLAG_NONE);
-
-			goto next;
-
-		case OPCODE_RCC:
-			fprintf(stderr, "Dont know how to handle op %d yet\n",
-				vpi->Opcode);
-			_mesa_exit(-1);
+			inst =
+			    t_opcode_flr(vp, vpi, inst, src, /* FIXME */
+					 &u_temp_i);
+			break;
+		case OPCODE_FRC:
+			inst = t_opcode_frc(vp, vpi, inst, src);
+			break;
+		case OPCODE_LG2:
+			inst = t_opcode_lg2(vp, vpi, inst, src);
+			break;
+		case OPCODE_LIT:
+			inst = t_opcode_lit(vp, vpi, inst, src);
+			break;
+		case OPCODE_LOG:
+			inst = t_opcode_log(vp, vpi, inst, src);
+			break;
+		case OPCODE_MAD:
+			inst = t_opcode_mad(vp, vpi, inst, src);
+			break;
+		case OPCODE_MAX:
+			inst = t_opcode_max(vp, vpi, inst, src);
+			break;
+		case OPCODE_MIN:
+			inst = t_opcode_min(vp, vpi, inst, src);
+			break;
+		case OPCODE_MOV:
+			inst = t_opcode_mov(vp, vpi, inst, src);
 			break;
-		case OPCODE_END:
+		case OPCODE_MUL:
+			inst = t_opcode_mul(vp, vpi, inst, src);
+			break;
+		case OPCODE_POW:
+			inst = t_opcode_pow(vp, vpi, inst, src);
+			break;
+		case OPCODE_RCP:
+			inst = t_opcode_rcp(vp, vpi, inst, src);
+			break;
+		case OPCODE_RSQ:
+			inst = t_opcode_rsq(vp, vpi, inst, src);
+			break;
+		case OPCODE_SGE:
+			inst = t_opcode_sge(vp, vpi, inst, src);
+			break;
+		case OPCODE_SLT:
+			inst = t_opcode_slt(vp, vpi, inst, src);
+			break;
+		case OPCODE_SUB:
+			inst = t_opcode_sub(vp, vpi, inst, src);
+			break;
+		case OPCODE_SWZ:
+			inst = t_opcode_swz(vp, vpi, inst, src);
+			break;
+		case OPCODE_XPD:
+			inst =
+			    t_opcode_xpd(vp, vpi, inst, src, /* FIXME */
+					 &u_temp_i);
 			break;
 		default:
+			assert(0);
 			break;
 		}
-
-		o_inst->op =
-		    MAKE_VSF_OP(t_opcode(vpi->Opcode),
-				t_dst_index(vp, &vpi->DstReg),
-				t_dst_mask(vpi->DstReg.WriteMask),
-				t_dst_class(vpi->DstReg.File));
-
-		if (are_srcs_scalar) {
-			switch (operands) {
-			case 1:
-				o_inst->src[0] = t_src_scalar(vp, &src[0]);
-				o_inst->src[1] = ZERO_SRC_0;
-				o_inst->src[2] = ZERO_SRC_0;
-				break;
-
-			case 2:
-				o_inst->src[0] = t_src_scalar(vp, &src[0]);
-				o_inst->src[1] = t_src_scalar(vp, &src[1]);
-				o_inst->src[2] = ZERO_SRC_1;
-				break;
-
-			case 3:
-				o_inst->src[0] = t_src_scalar(vp, &src[0]);
-				o_inst->src[1] = t_src_scalar(vp, &src[1]);
-				o_inst->src[2] = t_src_scalar(vp, &src[2]);
-				break;
-
-			default:
-				fprintf(stderr,
-					"scalars and op RCC not handled yet");
-				_mesa_exit(-1);
-				break;
-			}
-		} else {
-			switch (operands) {
-			case 1:
-				o_inst->src[0] = t_src(vp, &src[0]);
-				o_inst->src[1] = ZERO_SRC_0;
-				o_inst->src[2] = ZERO_SRC_0;
-				break;
-
-			case 2:
-				o_inst->src[0] = t_src(vp, &src[0]);
-				o_inst->src[1] = t_src(vp, &src[1]);
-				o_inst->src[2] = ZERO_SRC_1;
-				break;
-
-			case 3:
-				o_inst->src[0] = t_src(vp, &src[0]);
-				o_inst->src[1] = t_src(vp, &src[1]);
-				o_inst->src[2] = t_src(vp, &src[2]);
-				break;
-
-			default:
-				fprintf(stderr,
-					"scalars and op RCC not handled yet");
-				_mesa_exit(-1);
-				break;
-			}
-		}
-	      next:;
 	}
 
-	/* Will most likely segfault before we get here... fix later. */
-	if (o_inst - vp->program.body.i >= VSF_MAX_FRAGMENT_LENGTH / 4) {
+	vp->program.length = (inst - vp->program.body.i);
+	if (vp->program.length >= VSF_MAX_FRAGMENT_LENGTH) {
 		vp->program.length = 0;
 		vp->native = GL_FALSE;
-		return;
 	}
-	vp->program.length = (o_inst - vp->program.body.i) * 4;
 #if 0
 	fprintf(stderr, "hw program:\n");
 	for (i = 0; i < vp->program.length; i++)
@@ -1065,7 +1309,8 @@ static void position_invariant(struct gl_program *prog)
 	struct gl_program_parameter_list *paramList;
 	int i;
 
-	gl_state_index tokens[STATE_LENGTH] = { STATE_MVP_MATRIX, 0, 0, 0, 0 };
+	gl_state_index tokens[STATE_LENGTH] =
+	    { STATE_MVP_MATRIX, 0, 0, 0, 0 };
 
 	/* tokens[4] = matrix modifier */
 #ifdef PREFER_DP4
@@ -1159,8 +1404,8 @@ static void insert_wpos(struct r300_vertex_program *vp,
 				prog->NumInstructions - 1);
 	/* END */
 	_mesa_copy_instructions(&vpi[prog->NumInstructions + 1],
-				&prog->Instructions[prog->NumInstructions - 1],
-				1);
+				&prog->Instructions[prog->NumInstructions -
+						    1], 1);
 	vpi_insert = &vpi[prog->NumInstructions - 1];
 
 	vpi_insert[i].Opcode = OPCODE_MOV;
@@ -1206,8 +1451,8 @@ static void pos_as_texcoord(struct r300_vertex_program *vp,
 	prog->NumTemporaries++;
 
 	for (vpi = prog->Instructions; vpi->Opcode != OPCODE_END; vpi++) {
-		if (vpi->DstReg.File == PROGRAM_OUTPUT &&
-		    vpi->DstReg.Index == VERT_RESULT_HPOS) {
+		if (vpi->DstReg.File == PROGRAM_OUTPUT
+		    && vpi->DstReg.Index == VERT_RESULT_HPOS) {
 			vpi->DstReg.File = PROGRAM_TEMPORARY;
 			vpi->DstReg.Index = tempregi;
 		}
@@ -1223,20 +1468,18 @@ static struct r300_vertex_program *build_program(struct r300_vertex_program_key
 
 	vp = _mesa_calloc(sizeof(*vp));
 	_mesa_memcpy(&vp->key, wanted_key, sizeof(vp->key));
-
 	vp->wpos_idx = wpos_idx;
 
 	if (mesa_vp->IsPositionInvariant) {
 		position_invariant(&mesa_vp->Base);
 	}
 
-	if (wpos_idx > -1)
+	if (wpos_idx > -1) {
 		pos_as_texcoord(vp, &mesa_vp->Base);
+	}
 
 	assert(mesa_vp->Base.NumInstructions);
-
 	vp->num_temporaries = mesa_vp->Base.NumTemporaries;
-
 	r300TranslateVertexShader(vp, mesa_vp->Base.Instructions);
 
 	return vp;
@@ -1252,11 +1495,10 @@ void r300SelectVertexShader(r300ContextPtr r300)
 	struct r300_vertex_program *vp;
 	GLint wpos_idx;
 
-	vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
+	vpc =
+	    (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
 	InputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
 
-	wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS;
-
 	wpos_idx = -1;
 	if (InputsRead & FRAG_BIT_WPOS) {
 		for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
@@ -1271,28 +1513,35 @@ void r300SelectVertexShader(r300ContextPtr r300)
 		InputsRead |= (FRAG_BIT_TEX0 << i);
 		wpos_idx = i;
 	}
+	wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
+	wanted_key.OutputsWritten = vpc->mesa_program.Base.OutputsWritten;
+
+	wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS;
 
-	if (InputsRead & FRAG_BIT_COL0)
+	if (InputsRead & FRAG_BIT_COL0) {
 		wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL0;
+	}
 
-	if ((InputsRead & FRAG_BIT_COL1)	/*||
-						   (InputsRead & FRAG_BIT_FOGC) */ )
+	if ((InputsRead & FRAG_BIT_COL1)) {
 		wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL1;
+	}
 
-	for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
-		if (InputsRead & (FRAG_BIT_TEX0 << i))
+	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
 			wanted_key.OutputsWritten |=
 			    1 << (VERT_RESULT_TEX0 + i);
+		}
+	}
 
-	wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
 	if (vpc->mesa_program.IsPositionInvariant) {
 		/* we wan't position don't we ? */
 		wanted_key.InputsRead |= (1 << VERT_ATTRIB_POS);
+		wanted_key.OutputsWritten |= (1 << VERT_RESULT_HPOS);
 	}
 
 	for (vp = vpc->progs; vp; vp = vp->next)
-		if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key)) ==
-		    0) {
+		if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key))
+		    == 0) {
 			r300->selected_vp = vp;
 			return;
 		}
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.h b/src/mesa/drivers/dri/r300/r300_vertprog.h
index 252d5a901f..3df0eee799 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.h
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.h
@@ -3,11 +3,6 @@
 
 #include "r300_reg.h"
 
-typedef struct {
-	GLuint op;
-	GLuint src[3];
-} VERTEX_SHADER_INSTRUCTION;
-
 #define VSF_FLAG_X	1
 #define VSF_FLAG_Y	2
 #define VSF_FLAG_Z	4
diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c
index 3dd821a4d3..4ce2f60b4f 100644
--- a/src/mesa/drivers/dri/savage/savagetris.c
+++ b/src/mesa/drivers/dri/savage/savagetris.c
@@ -131,7 +131,9 @@ static __inline__ void savage_draw_point (savageContextPtr imesa,
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
    const GLfloat x = tmp->v.x;
    const GLfloat y = tmp->v.y;
-   const GLfloat sz = imesa->glCtx->Point._Size * .5;
+   const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                                  imesa->glCtx->Const.MinPointSize,
+                                  imesa->glCtx->Const.MaxPointSize);
    GLuint j;
 
    *(float *)&vb[0] = x - sz;
@@ -164,7 +166,9 @@ static __inline__ void savage_draw_line (savageContextPtr imesa,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
-   GLfloat width = imesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(imesa->glCtx->Line.Width,
+                               imesa->glCtx->Const.MinLineWidth,
+                               imesa->glCtx->Const.MaxLineWidth);
    GLfloat dx, dy, ix, iy;
    GLuint j;
 
@@ -234,7 +238,9 @@ static __inline__ void savage_ptex_line (savageContextPtr imesa,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
-   GLfloat width = imesa->glCtx->Line._Width;
+   const GLfloat width = CLAMP(imesa->glCtx->Line.Width,
+                               imesa->glCtx->Const.MinLineWidth,
+                               imesa->glCtx->Const.MaxLineWidth);
    GLfloat dx, dy, ix, iy;
    savageVertex tmp0, tmp1;
    GLuint j;
@@ -281,7 +287,9 @@ static __inline__ void savage_ptex_point (savageContextPtr imesa,
    u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
    const GLfloat x = v0->v.x;
    const GLfloat y = v0->v.y;
-   const GLfloat sz = imesa->glCtx->Point._Size * .5;
+   const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
+                                  imesa->glCtx->Const.MinPointSize,
+                                  imesa->glCtx->Const.MaxPointSize);
    savageVertex tmp;
    GLuint j;
 
diff --git a/src/mesa/drivers/dri/sis/sis_tex.c b/src/mesa/drivers/dri/sis/sis_tex.c
index be87f16e29..8fd7b26ba5 100644
--- a/src/mesa/drivers/dri/sis/sis_tex.c
+++ b/src/mesa/drivers/dri/sis/sis_tex.c
@@ -130,6 +130,8 @@ sisAllocTexImage( sisContextPtr smesa, sisTexObjPtr t, int level,
 static void
 sisFreeTexImage( sisContextPtr smesa, sisTexObjPtr t, int level )
 {
+   assert(level >= 0);
+   assert(level < SIS_MAX_TEXTURE_LEVELS);
    if (t->image[level].Data == NULL)
       return;
 
@@ -213,7 +215,7 @@ sisDeleteTexture( GLcontext * ctx, struct gl_texture_object *texObj )
        */
       return;
    }
-   for (i = 0; i < MAX_TEXTURE_LEVELS; i++) {
+   for (i = 0; i < SIS_MAX_TEXTURE_LEVELS; i++) {
       sisFreeTexImage( smesa, t, i );
    }
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
index 89865d9637..64946385b6 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
@@ -1822,7 +1822,7 @@ tdfxTestProxyTexImage(GLcontext *ctx, GLenum target,
             tdfxTexInfo *ti;
             int memNeeded;
 
-            tObj = ctx->Texture.Proxy2D;
+            tObj = ctx->Texture.ProxyTex[TEXTURE_2D_INDEX];
             if (!tObj->DriverData)
                 tObj->DriverData = fxAllocTexObjData(fxMesa);
             ti = TDFX_TEXTURE_DATA(tObj);
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tris.c b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
index 96f9ae27fc..7252a7e7dc 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tris.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
@@ -184,7 +184,7 @@ tdfx_translate_vertex( GLcontext *ctx, const tdfxVertex *src, SWvertex *dst)
       }
    }
 
-   dst->pointSize = ctx->Point._Size;
+   dst->pointSize = ctx->Point.Size;
 }
 
 
diff --git a/src/mesa/drivers/dri/unichrome/via_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index 4d25d328e3..7c73877720 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -733,14 +733,15 @@ void viaXMesaWindowMoved(struct via_context *vmesa)
 {
    __DRIdrawablePrivate *const drawable = vmesa->driDrawable;
    __DRIdrawablePrivate *const readable = vmesa->driReadable;
-   struct via_renderbuffer *const draw_buffer = 
-     (struct via_renderbuffer *) drawable->driverPrivate;
-   struct via_renderbuffer *const read_buffer =
-     (struct via_renderbuffer *) readable->driverPrivate;
+   struct via_renderbuffer * draw_buffer;
+   struct via_renderbuffer * read_buffer;
    GLuint bytePerPixel = vmesa->viaScreen->bitsPerPixel >> 3;
 
    if (!drawable)
       return;
+
+   draw_buffer =  (struct via_renderbuffer *) drawable->driverPrivate;
+   read_buffer =  (struct via_renderbuffer *) readable->driverPrivate;
    
    switch (vmesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0]) {
    case BUFFER_BIT_BACK_LEFT: 
diff --git a/src/mesa/drivers/fbdev/glfbdev.c b/src/mesa/drivers/fbdev/glfbdev.c
index 6c6511b7e5..e95a424698 100644
--- a/src/mesa/drivers/fbdev/glfbdev.c
+++ b/src/mesa/drivers/fbdev/glfbdev.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -63,6 +63,9 @@
 #include "drivers/common/driverfuncs.h"
 
 
+/**
+ * Pixel formats we support:
+ */
 #define PF_B8G8R8     1
 #define PF_B8G8R8A8   2
 #define PF_B5G6R5     3
@@ -70,7 +73,7 @@
 #define PF_CI8        5
 
 
-/*
+/**
  * Derived from Mesa's GLvisual class.
  */
 struct GLFBDevVisualRec {
@@ -80,7 +83,7 @@ struct GLFBDevVisualRec {
    int pixelFormat;
 };
 
-/*
+/**
  * Derived from Mesa's GLframebuffer class.
  */
 struct GLFBDevBufferRec {
@@ -92,7 +95,7 @@ struct GLFBDevBufferRec {
    GLuint bytesPerPixel;
 };
 
-/*
+/**
  * Derived from Mesa's GLcontext class.
  */
 struct GLFBDevContextRec {
@@ -103,7 +106,7 @@ struct GLFBDevContextRec {
    GLFBDevBufferPtr curBuffer;
 };
 
-/*
+/**
  * Derived from Mesa's gl_renderbuffer class.
  */
 struct GLFBDevRenderbufferRec {
@@ -114,11 +117,6 @@ struct GLFBDevRenderbufferRec {
 };
 
 
-
-#define GLFBDEV_CONTEXT(CTX)  ((GLFBDevContextPtr) (CTX))
-#define GLFBDEV_BUFFER(BUF)  ((GLFBDevBufferPtr) (BUF))
-
-
 /**********************************************************************/
 /* Internal device driver functions                                   */
 /**********************************************************************/
@@ -151,7 +149,7 @@ update_state( GLcontext *ctx, GLuint new_state )
 static void
 get_buffer_size( GLframebuffer *buffer, GLuint *width, GLuint *height )
 {
-   const GLFBDevBufferPtr fbdevbuffer = GLFBDEV_BUFFER(buffer);
+   const GLFBDevBufferPtr fbdevbuffer = (GLFBDevBufferPtr) buffer;
    *width = fbdevbuffer->var.xres;
    *height = fbdevbuffer->var.yres;
 }
@@ -389,8 +387,8 @@ glFBDevCreateVisual( const struct fb_fix_screeninfo *fixInfo,
          /* ignored for now */
          break;
       case GLFBDEV_MULTISAMPLE:
-	 numSamples = attrib[1];
-	 attrib++;
+         numSamples = attrib[1];
+         attrib++;
          break;
       default:
          /* unexpected token */
@@ -406,36 +404,36 @@ glFBDevCreateVisual( const struct fb_fix_screeninfo *fixInfo,
       alphaBits = varInfo->transp.length;
 
       if (fixInfo->visual == FB_VISUAL_TRUECOLOR ||
-	  fixInfo->visual == FB_VISUAL_DIRECTCOLOR) {
-	 if(varInfo->bits_per_pixel == 24
-	    && varInfo->red.offset == 16
-	    && varInfo->green.offset == 8
-	    && varInfo->blue.offset == 0)
-	    vis->pixelFormat = PF_B8G8R8;
-
-	 else if(varInfo->bits_per_pixel == 32
-		 && varInfo->red.offset == 16
-		 && varInfo->green.offset == 8
-		 && varInfo->blue.offset == 0)
-	    vis->pixelFormat = PF_B8G8R8A8;
-
-	 else if(varInfo->bits_per_pixel == 16
-		 && varInfo->red.offset == 11
-		 && varInfo->green.offset == 5
-		 && varInfo->blue.offset == 0)
-	    vis->pixelFormat = PF_B5G6R5;
-
-	 else if(varInfo->bits_per_pixel == 16
-		 && varInfo->red.offset == 10
-		 && varInfo->green.offset == 5
-		 && varInfo->blue.offset == 0)
-	    vis->pixelFormat = PF_B5G5R5;
-
-	 else {
-	    _mesa_problem(NULL, "Unsupported fbdev RGB visual/bitdepth!\n");
-	    _mesa_free(vis);
-	    return NULL;
-	 }
+          fixInfo->visual == FB_VISUAL_DIRECTCOLOR) {
+         if (varInfo->bits_per_pixel == 24
+             && varInfo->red.offset == 16
+             && varInfo->green.offset == 8
+             && varInfo->blue.offset == 0) {
+            vis->pixelFormat = PF_B8G8R8;
+         }
+         else if (varInfo->bits_per_pixel == 32
+                  && varInfo->red.offset == 16
+                  && varInfo->green.offset == 8
+                  && varInfo->blue.offset == 0) {
+            vis->pixelFormat = PF_B8G8R8A8;
+         }
+         else if (varInfo->bits_per_pixel == 16
+                  && varInfo->red.offset == 11
+                  && varInfo->green.offset == 5
+                  && varInfo->blue.offset == 0) {
+            vis->pixelFormat = PF_B5G6R5;
+         }
+         else if (varInfo->bits_per_pixel == 16
+                  && varInfo->red.offset == 10
+                  && varInfo->green.offset == 5
+                  && varInfo->blue.offset == 0) {
+            vis->pixelFormat = PF_B5G5R5;
+         }
+         else {
+            _mesa_problem(NULL, "Unsupported fbdev RGB visual/bitdepth!\n");
+            _mesa_free(vis);
+            return NULL;
+         }
       }
    }
    else {
@@ -578,7 +576,7 @@ new_glfbdev_renderbuffer(void *bufferStart, const GLFBDevVisualPtr visual)
 
       rb->rowStride = visual->var.xres_virtual * visual->var.bits_per_pixel / 8;
       rb->bottom = (GLubyte *) bufferStart
-	  + (visual->var.yres - 1) * rb->rowStride;
+                 + (visual->var.yres - 1) * rb->rowStride;
 
       rb->Base.Width = visual->var.xres;
       rb->Base.Height = visual->var.yres;
@@ -635,7 +633,7 @@ glFBDevCreateBuffer( const struct fb_fix_screeninfo *fixInfo,
                           &frontrb->Base);
    /* add back renderbuffer */
    if (visual->glvisual.doubleBufferMode) {
-      int malloced = !backBuffer;
+      const int malloced = !backBuffer;
       if (malloced) {
          /* malloc a back buffer */
          backBuffer = _mesa_malloc(size);
@@ -647,8 +645,11 @@ glFBDevCreateBuffer( const struct fb_fix_screeninfo *fixInfo,
       }
 
       backrb = new_glfbdev_renderbuffer(backBuffer, visual);
-      if(malloced)
-	 backrb->mallocedBuffer = GL_TRUE;
+      if (!backrb) {
+         /* out of mem */
+         return NULL;
+      }
+      backrb->mallocedBuffer = malloced;
 
       _mesa_add_renderbuffer(&buf->glframebuffer, BUFFER_BACK_LEFT,
                              &backrb->Base);
@@ -682,16 +683,10 @@ glFBDevDestroyBuffer( GLFBDevBufferPtr buffer )
       if (buffer == curDraw || buffer == curRead) {
          glFBDevMakeCurrent( NULL, NULL, NULL);
       }
-#if 0
-      /* free the software depth, stencil, accum buffers */
-      _mesa_free_framebuffer_data(&buffer->glframebuffer);
-      _mesa_free(buffer);
-#else
       {
          struct gl_framebuffer *fb = &buffer->glframebuffer;
          _mesa_unreference_framebuffer(&fb);
       }
-#endif
    }
 }
 
diff --git a/src/mesa/drivers/windows/gdi/wgl.c b/src/mesa/drivers/windows/gdi/wgl.c
index fb23d210db..dad3dc1160 100644
--- a/src/mesa/drivers/windows/gdi/wgl.c
+++ b/src/mesa/drivers/windows/gdi/wgl.c
@@ -33,15 +33,30 @@
 /* We're essentially building part of GDI here, so define this so that
  * we get the right export linkage. */
 #ifdef __MINGW32__
-#include <GL/gl.h>
+
+#include <stdarg.h>
+#include <windef.h>
+#include <wincon.h>
+#include <winbase.h>
+
+#  if defined(BUILD_GL32)
+#    define WINGDIAPI __declspec(dllexport)	
+#  else
+#    define __W32API_USE_DLLIMPORT__
+#  endif
+
+#include <wingdi.h>
+#include "GL/mesa_wgl.h"
 #include <stdlib.h>
+
 #else
+
 #define _GDI32_
-#endif
 #include <windows.h>
 
-#include "glapi.h"
+#endif
 
+#include "glapi.h"
 #include "GL/wmesa.h"   /* protos for wmesa* functions */
 
 /*
@@ -339,7 +354,7 @@ WINGDIAPI int GLAPIENTRY wglGetPixelFormat(HDC hdc)
 }
 
 WINGDIAPI BOOL GLAPIENTRY wglSetPixelFormat(HDC hdc,int iPixelFormat,
-					    const PIXELFORMATDESCRIPTOR *ppfd)
+					const PIXELFORMATDESCRIPTOR *ppfd)
 {
     (void) hdc;
     
@@ -392,12 +407,12 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
     
     bitDevice = CreateCompatibleDC(fontDevice);
     
-    // Swap fore and back colors so the bitmap has the right polarity
+    /* Swap fore and back colors so the bitmap has the right polarity */
     tempColor = GetBkColor(bitDevice);
     SetBkColor(bitDevice, GetTextColor(bitDevice));
     SetTextColor(bitDevice, tempColor);
     
-    // Place chars based on base line
+    /* Place chars based on base line */
     VERIFY(SetTextAlign(bitDevice, TA_BASELINE) != GDI_ERROR ? 1 : 0);
     
     for(i = 0; i < (int)numChars; i++) {
@@ -410,36 +425,36 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
 	
 	curChar = (char)(i + firstChar);
 	
-	// Find how high/wide this character is
+	/* Find how high/wide this character is */
 	VERIFY(GetTextExtentPoint32(bitDevice, &curChar, 1, &size));
 	
-	// Create the output bitmap
+	/* Create the output bitmap */
 	charWidth = size.cx;
 	charHeight = size.cy;
-	// Round up to the next multiple of 32 bits
+	/* Round up to the next multiple of 32 bits */
 	bmapWidth = ((charWidth + 31) / 32) * 32;   
 	bmapHeight = charHeight;
 	bitObject = CreateCompatibleBitmap(bitDevice,
 					   bmapWidth,
 					   bmapHeight);
-	//VERIFY(bitObject);
+	/* VERIFY(bitObject); */
 	
-	// Assign the output bitmap to the device
+	/* Assign the output bitmap to the device */
 	origBmap = SelectObject(bitDevice, bitObject);
 	(void) VERIFY(origBmap);
 	
 	VERIFY( PatBlt( bitDevice, 0, 0, bmapWidth, bmapHeight,BLACKNESS ) );
 	
-	// Use our source font on the device
+	/* Use our source font on the device */
 	VERIFY(SelectObject(bitDevice, GetCurrentObject(fontDevice,OBJ_FONT)));
 	
-	// Draw the character
+	/* Draw the character */
 	VERIFY(TextOut(bitDevice, 0, metric.tmAscent, &curChar, 1));
 	
-	// Unselect our bmap object
+	/* Unselect our bmap object */
 	VERIFY(SelectObject(bitDevice, origBmap));
 	
-	// Convert the display dependant representation to a 1 bit deep DIB
+	/* Convert the display dependant representation to a 1 bit deep DIB */
 	numBytes = (bmapWidth * bmapHeight) / 8;
 	bmap = malloc(numBytes);
 	dibInfo->bmiHeader.biWidth = bmapWidth;
@@ -447,24 +462,24 @@ static BOOL wglUseFontBitmaps_FX(HDC fontDevice, DWORD firstChar,
 	res = GetDIBits(bitDevice, bitObject, 0, bmapHeight, bmap,
 			dibInfo,
 			DIB_RGB_COLORS);
-	//VERIFY(res);
+	/* VERIFY(res); */
 	
-	// Create the GL object
+	/* Create the GL object */
 	glNewList(i + listBase, GL_COMPILE);
 	glBitmap(bmapWidth, bmapHeight, 0.0, (GLfloat)metric.tmDescent,
 		 (GLfloat)charWidth, 0.0,
 		 bmap);
 	glEndList();
-	// CheckGL();
+	/* CheckGL(); */
 	
-	// Destroy the bmap object
+	/* Destroy the bmap object */
 	DeleteObject(bitObject);
 	
-	// Deallocate the bitmap data
+	/* Deallocate the bitmap data */
 	free(bmap);
     }
     
-    // Destroy the DC
+    /* Destroy the DC */
     VERIFY(DeleteDC(bitDevice));
     
     free(dibInfo);
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 2eec188912..5b67439f0f 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -6,6 +6,7 @@
 #include "wmesadef.h"
 #include "colors.h"
 #include <GL/wmesa.h>
+#include <winuser.h>
 #include "context.h"
 #include "extensions.h"
 #include "framebuffer.h"
@@ -114,7 +115,7 @@ static void wmSetPixelFormat(WMesaFramebuffer pwfb, HDC hDC)
 {
     pwfb->cColorBits = GetDeviceCaps(hDC, BITSPIXEL);
 
-    // Only 16 and 32 bit targets are supported now
+    /* Only 16 and 32 bit targets are supported now */
     assert(pwfb->cColorBits == 0 ||
 	   pwfb->cColorBits == 16 || 
 	   pwfb->cColorBits == 32);
@@ -1171,7 +1172,7 @@ WMesaContext WMesaCreateContext(HDC hDC,
     /* I do not understand this contributed code */
     /* Support memory and device contexts */
     if(WindowFromDC(hDC) != NULL) {
-	c->hDC = GetDC(WindowFromDC(hDC)); // huh ????
+	c->hDC = GetDC(WindowFromDC(hDC)); /* huh ???? */
     }
     else {
 	c->hDC = hDC;
@@ -1404,6 +1405,7 @@ void WMesaSwapBuffers( HDC hdc )
  * table entries.  Hopefully, I'll find a better solution.  The
  * dispatch table generation scripts ought to be making these dummy
  * stubs as well. */
+#if !defined(__MINGW32__) || !defined(GL_NO_STDCALL)
 void gl_dispatch_stub_543(void){}
 void gl_dispatch_stub_544(void){}
 void gl_dispatch_stub_545(void){}
@@ -1471,3 +1473,4 @@ void gl_dispatch_stub_769(void){}
 void gl_dispatch_stub_770(void){}
 void gl_dispatch_stub_771(void){}
 
+#endif
diff --git a/src/mesa/drivers/windows/gdi/wmesadef.h b/src/mesa/drivers/windows/gdi/wmesadef.h
index 97b063a8ba..83a42e6082 100644
--- a/src/mesa/drivers/windows/gdi/wmesadef.h
+++ b/src/mesa/drivers/windows/gdi/wmesadef.h
@@ -1,6 +1,8 @@
 #ifndef WMESADEF_H
 #define WMESADEF_H
-
+#ifdef __MINGW32__
+#include <windows.h>
+#endif
 #include "context.h"
 
 
diff --git a/src/mesa/drivers/x11/fakeglx.c b/src/mesa/drivers/x11/fakeglx.c
index 1587df66bc..7a170b4d3d 100644
--- a/src/mesa/drivers/x11/fakeglx.c
+++ b/src/mesa/drivers/x11/fakeglx.c
@@ -296,6 +296,11 @@ save_glx_visual( Display *dpy, XVisualInfo *vinfo,
       }
    }
 
+   if (stereoFlag) {
+      /* stereo not supported */
+      return NULL;
+   }
+
    /* Comparing IDs uses less memory but sometimes fails. */
    /* XXX revisit this after 3.0 is finished. */
    if (_mesa_getenv("MESA_GLX_VISUAL_HACK"))
@@ -1079,7 +1084,7 @@ choose_visual( Display *dpy, int screen, const int *list, GLboolean fbConfig )
             else {
                stereo_flag = GL_TRUE;
             }
-            return NULL; /* stereo not supported */
+            break;
 	 case GLX_AUX_BUFFERS:
 	    parselist++;
             numAux = *parselist++;
diff --git a/src/mesa/main/api_arrayelt.c b/src/mesa/main/api_arrayelt.c
index 1899975213..72091b0789 100644
--- a/src/mesa/main/api_arrayelt.c
+++ b/src/mesa/main/api_arrayelt.c
@@ -31,8 +31,8 @@
 #include "context.h"
 #include "imports.h"
 #include "macros.h"
-#include "glapioffsets.h"
-#include "dispatch.h"
+#include "glapi/glapioffsets.h"
+#include "glapi/dispatch.h"
 
 typedef void (GLAPIENTRY *array_func)( const void * );
 
diff --git a/src/mesa/main/api_loopback.c b/src/mesa/main/api_loopback.c
index efe5a77d58..924d7134a2 100644
--- a/src/mesa/main/api_loopback.c
+++ b/src/mesa/main/api_loopback.c
@@ -30,14 +30,14 @@
 
 
 #include "glheader.h"
-#include "glapi.h"
-#include "glapitable.h"
 #include "macros.h"
 #include "colormac.h"
 #include "api_loopback.h"
-#include "glthread.h"
 #include "mtypes.h"
-#include "dispatch.h"
+#include "glapi/glapi.h"
+#include "glapi/glapitable.h"
+#include "glapi/glthread.h"
+#include "glapi/dispatch.h"
 
 /* KW: A set of functions to convert unusual Color/Normal/Vertex/etc
  * calls to a smaller set of driver-provided formats.  Currently just
diff --git a/src/mesa/main/api_noop.c b/src/mesa/main/api_noop.c
index 0c1a35361f..3df64362ea 100644
--- a/src/mesa/main/api_noop.c
+++ b/src/mesa/main/api_noop.c
@@ -31,7 +31,7 @@
 #include "light.h"
 #include "macros.h"
 #include "dlist.h"
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 
 /**
diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c
index 3d20ba7d14..18dead61c3 100644
--- a/src/mesa/main/api_validate.c
+++ b/src/mesa/main/api_validate.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.1
+ * Version:  7.0.1
  *
- * Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -35,6 +35,7 @@ _mesa_validate_DrawElements(GLcontext *ctx,
 			    GLenum mode, GLsizei count, GLenum type,
 			    const GLvoid *indices)
 {
+   GLboolean mapped = GL_FALSE;
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx,  GL_FALSE);
 
    if (count <= 0) {
@@ -67,13 +68,19 @@ _mesa_validate_DrawElements(GLcontext *ctx,
    /* Vertex buffer object tests */
    if (ctx->Array.ElementArrayBufferObj->Name) {
       GLuint indexBytes;
+      const GLvoid *map = ctx->Driver.MapBuffer(ctx,
+                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                                GL_READ_ONLY,
+                                                ctx->Array.ElementArrayBufferObj);
 
       /* use indices in the buffer object */
-      if (!ctx->Array.ElementArrayBufferObj->Data) {
+      if (!map) {
          _mesa_warning(ctx, "DrawElements with empty vertex elements buffer!");
          return GL_FALSE;
       }
 
+      mapped = GL_TRUE;
+
       /* make sure count doesn't go outside buffer bounds */
       if (type == GL_UNSIGNED_INT) {
          indexBytes = count * sizeof(GLuint);
@@ -86,20 +93,25 @@ _mesa_validate_DrawElements(GLcontext *ctx,
          indexBytes = count * sizeof(GLushort);
       }
 
-      if ((GLubyte *) indices + indexBytes >
-          ctx->Array.ElementArrayBufferObj->Data +
-          ctx->Array.ElementArrayBufferObj->Size) {
+      if (ADD_POINTERS(map, indices) + indexBytes > 
+          (GLubyte *)map + ctx->Array.ElementArrayBufferObj->Size) {
          _mesa_warning(ctx, "glDrawElements index out of buffer bounds");
+         ctx->Driver.UnmapBuffer(ctx,
+                                 GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                 ctx->Array.ElementArrayBufferObj);
          return GL_FALSE;
       }
 
       /* Actual address is the sum of pointers.  Indices may be used below. */
       if (ctx->Const.CheckArrayBounds) {
-         indices = (const GLvoid *)
-            ADD_POINTERS(ctx->Array.ElementArrayBufferObj->Data,
-                         (const GLubyte *) indices);
+         indices = ADD_POINTERS(map, indices);
       }
    }
+   else {
+      /* not using a VBO */
+      if (!indices)
+         return GL_FALSE;
+   }
 
    if (ctx->Const.CheckArrayBounds) {
       /* find max array index */
@@ -121,12 +133,25 @@ _mesa_validate_DrawElements(GLcontext *ctx,
             if (((GLubyte *) indices)[i] > max)
                max = ((GLubyte *) indices)[i];
       }
+
       if (max >= ctx->Array._MaxElement) {
          /* the max element is out of bounds of one or more enabled arrays */
+         if (mapped) {
+            ctx->Driver.UnmapBuffer(ctx,
+                                    GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                    ctx->Array.ElementArrayBufferObj);
+         }
+
          return GL_FALSE;
       }
    }
 
+   if (mapped) {
+      ctx->Driver.UnmapBuffer(ctx,
+                              GL_ELEMENT_ARRAY_BUFFER_ARB,
+                              ctx->Array.ElementArrayBufferObj);
+   }
+
    return GL_TRUE;
 }
 
@@ -170,6 +195,16 @@ _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode,
        && !(ctx->VertexProgram._Enabled && ctx->Array.ArrayObj->VertexAttrib[0].Enabled))
       return GL_FALSE;
 
+   /* Vertex buffer object tests */
+   if (ctx->Array.ElementArrayBufferObj->Name) {
+      /* XXX re-use code from above? */
+   }
+   else {
+      /* not using VBO */
+      if (!indices)
+         return GL_FALSE;
+   }
+
    if (ctx->Const.CheckArrayBounds) {
       /* Find max array index.
        * We don't trust the user's start and end values.
diff --git a/src/mesa/main/arrayobj.c b/src/mesa/main/arrayobj.c
index d601ee461e..f08f99d8e1 100644
--- a/src/mesa/main/arrayobj.c
+++ b/src/mesa/main/arrayobj.c
@@ -46,7 +46,7 @@
 #include "bufferobj.h"
 #endif
 #include "arrayobj.h"
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 
 /**
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index b422198f92..3199c9b426 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -52,6 +52,24 @@
 
 
 /**
+ * Special struct for saving/restoring texture state (GL_TEXTURE_BIT)
+ */
+struct texture_state
+{
+   struct gl_texture_attrib Texture;  /**< The usual context state */
+
+   /** to save per texture object state (wrap modes, filters, etc): */
+   struct gl_texture_object SavedObj[MAX_TEXTURE_UNITS][NUM_TEXTURE_TARGETS];
+
+   /**
+    * To save references to texture objects (so they don't get accidentally
+    * deleted while saved in the attribute stack).
+    */
+   struct gl_texture_object *SavedTexRef[MAX_TEXTURE_UNITS][NUM_TEXTURE_TARGETS];
+};
+
+
+/**
  * Allocate a new attribute state node.  These nodes have a
  * "kind" value and a pointer to a struct of state data.
  */
@@ -339,46 +357,61 @@ _mesa_PushAttrib(GLbitfield mask)
    }
 
    if (mask & GL_TEXTURE_BIT) {
-      struct gl_texture_attrib *attr;
+      struct texture_state *texstate = CALLOC_STRUCT(texture_state);
       GLuint u;
 
+      if (!texstate) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib(GL_TEXTURE_BIT)");
+         goto end;
+      }
+
       _mesa_lock_context_textures(ctx);
-      /* Bump the texture object reference counts so that they don't
-       * inadvertantly get deleted.
+
+      /* copy/save the bulk of texture state here */
+      _mesa_memcpy(&texstate->Texture, &ctx->Texture, sizeof(ctx->Texture));
+
+      /* Save references to the currently bound texture objects so they don't
+       * accidentally get deleted while referenced in the attribute stack.
        */
       for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-	 ctx->Texture.Unit[u].Current1D->RefCount++;
-	 ctx->Texture.Unit[u].Current2D->RefCount++;
-	 ctx->Texture.Unit[u].Current3D->RefCount++;
-	 ctx->Texture.Unit[u].CurrentCubeMap->RefCount++;
-	 ctx->Texture.Unit[u].CurrentRect->RefCount++;
-	 ctx->Texture.Unit[u].Current1DArray->RefCount++;
-	 ctx->Texture.Unit[u].Current2DArray->RefCount++;
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_1D_INDEX],
+                                ctx->Texture.Unit[u].Current1D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_2D_INDEX],
+                                ctx->Texture.Unit[u].Current2D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_3D_INDEX],
+                                ctx->Texture.Unit[u].Current3D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_CUBE_INDEX],
+                                ctx->Texture.Unit[u].CurrentCubeMap);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_RECT_INDEX],
+                                ctx->Texture.Unit[u].CurrentRect);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_1D_ARRAY_INDEX],
+                                ctx->Texture.Unit[u].Current1DArray);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_2D_ARRAY_INDEX],
+                                ctx->Texture.Unit[u].Current2DArray);
       }
-      attr = MALLOC_STRUCT( gl_texture_attrib );
-      MEMCPY( attr, &ctx->Texture, sizeof(struct gl_texture_attrib) );
-      /* copy state of the currently bound texture objects */
+
+      /* copy state/contents of the currently bound texture objects */
       for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-         _mesa_copy_texture_object(&attr->Unit[u].Saved1D,
-                                   attr->Unit[u].Current1D);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved2D,
-                                   attr->Unit[u].Current2D);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved3D,
-                                   attr->Unit[u].Current3D);
-         _mesa_copy_texture_object(&attr->Unit[u].SavedCubeMap,
-                                   attr->Unit[u].CurrentCubeMap);
-         _mesa_copy_texture_object(&attr->Unit[u].SavedRect,
-                                   attr->Unit[u].CurrentRect);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved1DArray,
-                                   attr->Unit[u].Current1DArray);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved2DArray,
-                                   attr->Unit[u].Current2DArray);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_1D_INDEX],
+                                   ctx->Texture.Unit[u].Current1D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_2D_INDEX],
+                                   ctx->Texture.Unit[u].Current2D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_3D_INDEX],
+                                   ctx->Texture.Unit[u].Current3D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_CUBE_INDEX],
+                                   ctx->Texture.Unit[u].CurrentCubeMap);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_RECT_INDEX],
+                                   ctx->Texture.Unit[u].CurrentRect);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_1D_ARRAY_INDEX],
+                                   ctx->Texture.Unit[u].Current1DArray);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_2D_ARRAY_INDEX],
+                                   ctx->Texture.Unit[u].Current2DArray);
       }
 
       _mesa_unlock_context_textures(ctx);
 
       newnode = new_attrib_node( GL_TEXTURE_BIT );
-      newnode->data = attr;
+      newnode->data = texstate;
       newnode->next = head;
       head = newnode;
    }
@@ -414,6 +447,7 @@ _mesa_PushAttrib(GLbitfield mask)
       head = newnode;
    }
 
+end:
    ctx->AttribStack[ctx->AttribStackDepth] = head;
    ctx->AttribStackDepth++;
 }
@@ -623,14 +657,19 @@ pop_enable_group(GLcontext *ctx, const struct gl_enable_attrib *enable)
 }
 
 
+/**
+ * Pop/restore texture attribute/group state.
+ */
 static void
-pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
+pop_texture_group(GLcontext *ctx, struct texture_state *texstate)
 {
    GLuint u;
 
+   _mesa_lock_context_textures(ctx);
+
    for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-      const struct gl_texture_unit *unit = &texAttrib->Unit[u];
-      GLuint i;
+      const struct gl_texture_unit *unit = &texstate->Texture.Unit[u];
+      GLuint tgt;
 
       _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + u);
       _mesa_set_enable(ctx, GL_TEXTURE_1D,
@@ -723,53 +762,33 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
                        1 << unit->Combine.ScaleShiftA);
       }
 
-      /* Restore texture object state */
-      for (i = 0; i < NUM_TEXTURE_TARGETS; i++) {
-         GLenum target = 0;
+      /* Restore texture object state for each target */
+      for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
          const struct gl_texture_object *obj = NULL;
          GLfloat bordColor[4];
+         GLenum target;
 
-         switch (i) {
-         case 0:
-            target = GL_TEXTURE_1D;
-            obj = &unit->Saved1D;
-            break;
-         case 1:
-            target = GL_TEXTURE_2D;
-            obj = &unit->Saved2D;
-            break;
-         case 2:
-            target = GL_TEXTURE_3D;
-            obj = &unit->Saved3D;
-            break;
-         case 3:
-            if (!ctx->Extensions.ARB_texture_cube_map)
-               continue;
-            target = GL_TEXTURE_CUBE_MAP_ARB;
-            obj = &unit->SavedCubeMap;
-            break;
-         case 4:
-            if (!ctx->Extensions.NV_texture_rectangle)
-               continue;
-            target = GL_TEXTURE_RECTANGLE_NV;
-            obj = &unit->SavedRect;
-            break;
-         case 5:
-            if (!ctx->Extensions.MESA_texture_array)
-               continue;
-            target = GL_TEXTURE_1D_ARRAY_EXT;
-            obj = &unit->Saved1DArray;
-            break;
-         case 6:
-            if (!ctx->Extensions.MESA_texture_array)
-               continue;
-            target = GL_TEXTURE_2D_ARRAY_EXT;
-            obj = &unit->Saved2DArray;
-            break;
-         default:
-            ; /* silence warnings */
+         obj = &texstate->SavedObj[u][tgt];
+
+         /* don't restore state for unsupported targets to prevent
+          * raising GL errors.
+          */
+         if (obj->Target == GL_TEXTURE_CUBE_MAP_ARB &&
+             !ctx->Extensions.ARB_texture_cube_map) {
+            continue;
+         }
+         else if (obj->Target == GL_TEXTURE_RECTANGLE_NV &&
+                  !ctx->Extensions.NV_texture_rectangle) {
+            continue;
+         }
+         else if ((obj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
+                   obj->Target == GL_TEXTURE_2D_ARRAY_EXT) &&
+                  !ctx->Extensions.MESA_texture_array) {
+            continue;
          }
 
+         target = obj->Target;
+
          _mesa_BindTexture(target, obj->Name);
 
          bordColor[0] = CHAN_TO_FLOAT(obj->BorderColor[0]);
@@ -777,8 +796,8 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
          bordColor[2] = CHAN_TO_FLOAT(obj->BorderColor[2]);
          bordColor[3] = CHAN_TO_FLOAT(obj->BorderColor[3]);
 
-         _mesa_TexParameterf(target, GL_TEXTURE_PRIORITY, obj->Priority);
          _mesa_TexParameterfv(target, GL_TEXTURE_BORDER_COLOR, bordColor);
+         _mesa_TexParameterf(target, GL_TEXTURE_PRIORITY, obj->Priority);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, obj->WrapS);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, obj->WrapT);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, obj->WrapR);
@@ -804,25 +823,17 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
             _mesa_TexParameterf(target, GL_SHADOW_AMBIENT_SGIX,
                                 obj->ShadowAmbient);
          }
+      }
 
+      /* remove saved references to the texture objects */
+      for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][tgt], NULL);
       }
    }
-   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB
-                          + texAttrib->CurrentUnit);
 
-   /* "un-bump" the texture object reference counts.  We did that so they
-    * wouldn't inadvertantly get deleted while they were still referenced
-    * inside the attribute state stack.
-    */
-   for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-      ctx->Texture.Unit[u].Current1D->RefCount--;
-      ctx->Texture.Unit[u].Current2D->RefCount--;
-      ctx->Texture.Unit[u].Current3D->RefCount--;
-      ctx->Texture.Unit[u].CurrentCubeMap->RefCount--;
-      ctx->Texture.Unit[u].CurrentRect->RefCount--;
-      ctx->Texture.Unit[u].Current1DArray->RefCount--;
-      ctx->Texture.Unit[u].Current2DArray->RefCount--;
-   }
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + texstate->Texture.CurrentUnit);
+
+   _mesa_unlock_context_textures(ctx);
 }
 
 
@@ -1201,9 +1212,9 @@ _mesa_PopAttrib(void)
          case GL_TEXTURE_BIT:
             /* Take care of texture object reference counters */
             {
-               const struct gl_texture_attrib *texture;
-               texture = (const struct gl_texture_attrib *) attr->data;
-               pop_texture_group(ctx, texture);
+               struct texture_state *texstate
+                  = (struct texture_state *) attr->data;
+               pop_texture_group(ctx, texstate);
 	       ctx->NewState |= _NEW_TEXTURE;
             }
             break;
@@ -1425,6 +1436,42 @@ _mesa_PopClientAttrib(void)
 }
 
 
+/**
+ * Free any attribute state data that might be attached to the context.
+ */
+void
+_mesa_free_attrib_data(GLcontext *ctx)
+{
+   while (ctx->AttribStackDepth > 0) {
+      struct gl_attrib_node *attr, *next;
+
+      ctx->AttribStackDepth--;
+      attr = ctx->AttribStack[ctx->AttribStackDepth];
+
+      while (attr) {
+         if (attr->kind == GL_TEXTURE_BIT) {
+            struct texture_state *texstate = (struct texture_state*)attr->data;
+            GLuint u, tgt;
+            /* clear references to the saved texture objects */
+            for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
+               for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
+                  _mesa_reference_texobj(&texstate->SavedTexRef[u][tgt], NULL);
+               }
+            }
+         }
+         else {
+            /* any other chunks of state that requires special handling? */
+         }
+
+         next = attr->next;
+         _mesa_free(attr->data);
+         _mesa_free(attr);
+         attr = next;
+      }
+   }
+}
+
+
 void _mesa_init_attrib( GLcontext *ctx )
 {
    /* Renderer and client attribute stacks */
diff --git a/src/mesa/main/attrib.h b/src/mesa/main/attrib.h
index 09d75196b2..2cf8fe6934 100644
--- a/src/mesa/main/attrib.h
+++ b/src/mesa/main/attrib.h
@@ -1,18 +1,8 @@
-/**
- * \file attrib.h
- * Attribute stacks.
- * 
- * \if subset
- * (No-op)
- *
- * \endif
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  5.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -32,8 +22,6 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
-
-
 #ifndef ATTRIB_H
 #define ATTRIB_H
 
@@ -58,10 +46,14 @@ _mesa_PopClientAttrib( void );
 extern void 
 _mesa_init_attrib( GLcontext *ctx );
 
+extern void 
+_mesa_free_attrib_data( GLcontext *ctx );
+
 #else
 
 /** No-op */
 #define _mesa_init_attrib( c ) ((void)0)
+#define _mesa_free_attrib_data( c ) ((void)0)
 
 #endif
 
diff --git a/src/mesa/main/buffers.c b/src/mesa/main/buffers.c
index 0e6ca8ea1c..cbbe3e8698 100644
--- a/src/mesa/main/buffers.c
+++ b/src/mesa/main/buffers.c
@@ -336,6 +336,20 @@ read_buffer_enum_to_index(GLenum buffer)
  * \sa _mesa_DrawBuffersARB
  *
  * \param buffer  buffer token such as GL_LEFT or GL_FRONT_AND_BACK, etc.
+ *
+ * Note that the behaviour of this function depends on whether the
+ * current ctx->DrawBuffer is a window-system framebuffer (Name=0) or
+ * a user-created framebuffer object (Name!=0).
+ *   In the former case, we update the per-context ctx->Color.DrawBuffer
+ *   state var _and_ the FB's ColorDrawBuffer state.
+ *   In the later case, we update the FB's ColorDrawBuffer state only.
+ *
+ * Furthermore, upon a MakeCurrent() or BindFramebuffer() call, if the
+ * new FB is a window system FB, we need to re-update the FB's
+ * ColorDrawBuffer state to match the context.  This is handled in
+ * _mesa_update_framebuffer().
+ *
+ * See the GL_EXT_framebuffer_object spec for more info.
  */
 void GLAPIENTRY
 _mesa_DrawBuffer(GLenum buffer)
@@ -479,17 +493,22 @@ set_color_output(GLcontext *ctx, GLuint output, GLenum buffer,
    /* not really needed, will be set later */
    fb->_NumColorDrawBuffers[output] = 0;
 
-   if (fb->Name == 0)
-   /* Set traditional state var */
+   if (fb->Name == 0) {
+      /* Only set the per-context DrawBuffer state if we're currently
+       * drawing to a window system framebuffer.
+       */
       ctx->Color.DrawBuffer[output] = buffer;
+   }
 }
 
 
 /**
- * Helper routine used by _mesa_DrawBuffer, _mesa_DrawBuffersARB and
- * other places (window fbo fixup) to set fbo (and the old ctx) fields.
+ * Helper function to set the GL_DRAW_BUFFER state in the context and
+ * current FBO.
+ *
  * All error checking will have been done prior to calling this function
  * so nothing should go wrong at this point.
+ *
  * \param ctx  current context
  * \param n    number of color outputs to set
  * \param buffers  array[n] of colorbuffer names, like GL_LEFT.
@@ -529,43 +548,29 @@ _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
 }
 
 
-GLboolean
-_mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer)
+/**
+ * Like \sa _mesa_drawbuffers(), this is a helper function for setting
+ * GL_READ_BUFFER state in the context and current FBO.
+ * \param ctx  the rendering context
+ * \param buffer  GL_FRONT, GL_BACK, GL_COLOR_ATTACHMENT0, etc.
+ * \param bufferIndex  the numerical index corresponding to 'buffer'
+ */
+void
+_mesa_readbuffer(GLcontext *ctx, GLenum buffer, GLint bufferIndex)
 {
-   struct gl_framebuffer *fb;
-   GLbitfield supportedMask;
-   GLint srcBuffer;
-
-   fb = ctx->ReadBuffer;
-
-   if (MESA_VERBOSE & VERBOSE_API)
-      _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
-
-   if (fb->Name > 0 && buffer == GL_NONE) {
-      /* This is legal for user-created framebuffer objects */
-      srcBuffer = -1;
-   }
-   else {
-      /* general case / window-system framebuffer */
-      srcBuffer = read_buffer_enum_to_index(buffer);
-      if (srcBuffer == -1) {
-         _mesa_error(ctx, GL_INVALID_ENUM, "glReadBuffer(buffer=0x%x)", buffer);
-         return GL_FALSE;
-      }
-      supportedMask = supported_buffer_bitmask(ctx, fb);
-      if (((1 << srcBuffer) & supportedMask) == 0) {
-         _mesa_error(ctx, GL_INVALID_OPERATION, "glReadBuffer(buffer=0x%x)", buffer);
-         return GL_FALSE;
-      }
-   }
+   struct gl_framebuffer *fb = ctx->ReadBuffer;
 
    if (fb->Name == 0) {
+      /* Only update the per-context READ_BUFFER state if we're bound to
+       * a window-system framebuffer.
+       */
       ctx->Pixel.ReadBuffer = buffer;
    }
+
    fb->ColorReadBuffer = buffer;
-   fb->_ColorReadBufferIndex = srcBuffer;
+   fb->_ColorReadBufferIndex = bufferIndex;
 
-   return GL_TRUE;
+   ctx->NewState |= _NEW_PIXEL;
 }
 
 
@@ -577,16 +582,43 @@ _mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer)
 void GLAPIENTRY
 _mesa_ReadBuffer(GLenum buffer)
 {
+   struct gl_framebuffer *fb;
+   GLbitfield supportedMask;
+   GLint srcBuffer;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
 
-   if (!_mesa_readbuffer_update_fields(ctx, buffer))
-      return;
+   fb = ctx->ReadBuffer;
 
-   ctx->NewState |= _NEW_PIXEL;
+   if (MESA_VERBOSE & VERBOSE_API)
+      _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
+
+   if (fb->Name > 0 && buffer == GL_NONE) {
+      /* This is legal for user-created framebuffer objects */
+      srcBuffer = -1;
+   }
+   else {
+      /* general case / window-system framebuffer */
+      srcBuffer = read_buffer_enum_to_index(buffer);
+      if (srcBuffer == -1) {
+         _mesa_error(ctx, GL_INVALID_ENUM,
+                     "glReadBuffer(buffer=0x%x)", buffer);
+         return;
+      }
+      supportedMask = supported_buffer_bitmask(ctx, fb);
+      if (((1 << srcBuffer) & supportedMask) == 0) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glReadBuffer(buffer=0x%x)", buffer);
+         return;
+      }
+   }
+
+   /* OK, all error checking has been completed now */
+
+   _mesa_readbuffer(ctx, buffer, srcBuffer);
 
    /*
     * Call device driver function.
diff --git a/src/mesa/main/buffers.h b/src/mesa/main/buffers.h
index 208e7af2b9..9c0070e00f 100644
--- a/src/mesa/main/buffers.h
+++ b/src/mesa/main/buffers.h
@@ -5,9 +5,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -56,8 +56,8 @@ extern void
 _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
                   const GLbitfield *destMask);
 
-extern GLboolean
-_mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer);
+extern void
+_mesa_readbuffer(GLcontext *ctx, GLenum buffer, GLint bufferIndex);
 
 extern void GLAPIENTRY
 _mesa_ReadBuffer( GLenum mode );
diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c
index 610acba306..97120398f9 100644
--- a/src/mesa/main/colortab.c
+++ b/src/mesa/main/colortab.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -30,6 +30,7 @@
 #include "image.h"
 #include "macros.h"
 #include "state.h"
+#include "teximage.h"
 
 
 /**
@@ -305,49 +306,6 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat,
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); /* too complex */
 
    switch (target) {
-      case GL_TEXTURE_1D:
-         texObj = texUnit->Current1D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_2D:
-         texObj = texUnit->Current2D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_3D:
-         texObj = texUnit->Current3D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTable(target)");
-            return;
-         }
-         texObj = texUnit->CurrentCubeMap;
-         table = &texObj->Palette;
-         break;
-      case GL_PROXY_TEXTURE_1D:
-         texObj = ctx->Texture.Proxy1D;
-         table = &texObj->Palette;
-         proxy = GL_TRUE;
-         break;
-      case GL_PROXY_TEXTURE_2D:
-         texObj = ctx->Texture.Proxy2D;
-         table = &texObj->Palette;
-         proxy = GL_TRUE;
-         break;
-      case GL_PROXY_TEXTURE_3D:
-         texObj = ctx->Texture.Proxy3D;
-         table = &texObj->Palette;
-         proxy = GL_TRUE;
-         break;
-      case GL_PROXY_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTable(target)");
-            return;
-         }
-         texObj = ctx->Texture.ProxyCubeMap;
-         table = &texObj->Palette;
-         break;
       case GL_SHARED_TEXTURE_PALETTE_EXT:
          table = &ctx->Texture.Palette;
          break;
@@ -396,8 +354,19 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat,
          proxy = GL_TRUE;
          break;
       default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glColorTable(target)");
-         return;
+         /* try texture targets */
+         {
+            struct gl_texture_object *texobj
+               = _mesa_select_tex_object(ctx, texUnit, target);
+            if (texobj) {
+               table = &texobj->Palette;
+               proxy = _mesa_is_proxy_texture(target);
+            }
+            else {
+               _mesa_error(ctx, GL_INVALID_ENUM, "glColorTable(target)");
+               return;
+            }
+         }
    }
 
    assert(table);
@@ -499,26 +468,6 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    switch (target) {
-      case GL_TEXTURE_1D:
-         texObj = texUnit->Current1D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_2D:
-         texObj = texUnit->Current2D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_3D:
-         texObj = texUnit->Current3D;
-         table = &texObj->Palette;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorSubTable(target)");
-            return;
-         }
-         texObj = texUnit->CurrentCubeMap;
-         table = &texObj->Palette;
-         break;
       case GL_SHARED_TEXTURE_PALETTE_EXT:
          table = &ctx->Texture.Palette;
          break;
@@ -547,8 +496,15 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
          bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX];
          break;
       default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glColorSubTable(target)");
-         return;
+         /* try texture targets */
+         texObj = _mesa_select_tex_object(ctx, texUnit, target);
+         if (texObj && !_mesa_is_proxy_texture(target)) {
+            table = &texObj->Palette;
+         }
+         else {
+            _mesa_error(ctx, GL_INVALID_ENUM, "glColorSubTable(target)");
+            return;
+         }
    }
 
    assert(table);
@@ -636,22 +592,6 @@ _mesa_GetColorTable( GLenum target, GLenum format,
    }
 
    switch (target) {
-      case GL_TEXTURE_1D:
-         table = &texUnit->Current1D->Palette;
-         break;
-      case GL_TEXTURE_2D:
-         table = &texUnit->Current2D->Palette;
-         break;
-      case GL_TEXTURE_3D:
-         table = &texUnit->Current3D->Palette;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTable(target)");
-            return;
-         }
-         table = &texUnit->CurrentCubeMap->Palette;
-         break;
       case GL_SHARED_TEXTURE_PALETTE_EXT:
          table = &ctx->Texture.Palette;
          break;
@@ -672,8 +612,18 @@ _mesa_GetColorTable( GLenum target, GLenum format,
          table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
          break;
       default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTable(target)");
-         return;
+         /* try texture targets */
+         {
+            struct gl_texture_object *texobj
+               = _mesa_select_tex_object(ctx, texUnit, target);
+            if (texobj && !_mesa_is_proxy_texture(target)) {
+               table = &texobj->Palette;
+            }
+            else {
+               _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTable(target)");
+               return;
+            }
+         }
    }
 
    ASSERT(table);
@@ -781,65 +731,41 @@ _mesa_GetColorTable( GLenum target, GLenum format,
 void GLAPIENTRY
 _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 {
+   GLfloat *scale, *bias;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    switch (target) {
-      case GL_COLOR_TABLE_SGI:
-         if (pname == GL_COLOR_TABLE_SCALE_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION], params);
-         }
-         else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION], params);
-         }
-         else {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
-            return;
-         }
-         break;
-      case GL_TEXTURE_COLOR_TABLE_SGI:
-         if (!ctx->Extensions.SGI_texture_color_table) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameter(target)");
-            return;
-         }
-         if (pname == GL_COLOR_TABLE_SCALE_SGI) {
-            COPY_4V(ctx->Pixel.TextureColorTableScale, params);
-         }
-         else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
-            COPY_4V(ctx->Pixel.TextureColorTableBias, params);
-         }
-         else {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
-            return;
-         }
-         break;
-      case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
-         if (pname == GL_COLOR_TABLE_SCALE_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION], params);
-         }
-         else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION], params);
-         }
-         else {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
-            return;
-         }
-         break;
-      case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
-         if (pname == GL_COLOR_TABLE_SCALE_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX], params);
-         }
-         else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
-            COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX], params);
-         }
-         else {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
-            return;
-         }
-         break;
-      default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameter(target)");
-         return;
+   case GL_COLOR_TABLE_SGI:
+      scale = ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION];
+      bias  = ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION];
+      break;
+   case GL_TEXTURE_COLOR_TABLE_SGI:
+      scale = ctx->Pixel.TextureColorTableScale;
+      bias  = ctx->Pixel.TextureColorTableBias;
+      break;
+   case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
+      scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION];
+      bias  = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION];
+      break;
+   case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
+      scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX];
+      bias  = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX];
+      break;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameter(target)");
+      return;
+   }
+
+   if (pname == GL_COLOR_TABLE_SCALE_SGI) {
+      COPY_4V(scale, params);
+   }
+   else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
+      COPY_4V(bias, params);
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
+      return;
    }
 
    ctx->NewState |= _NEW_PIXEL;
@@ -879,40 +805,6 @@ _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params )
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    switch (target) {
-      case GL_TEXTURE_1D:
-         table = &texUnit->Current1D->Palette;
-         break;
-      case GL_TEXTURE_2D:
-         table = &texUnit->Current2D->Palette;
-         break;
-      case GL_TEXTURE_3D:
-         table = &texUnit->Current3D->Palette;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetColorTableParameterfv(target)");
-            return;
-         }
-         table = &texUnit->CurrentCubeMap->Palette;
-         break;
-      case GL_PROXY_TEXTURE_1D:
-         table = &ctx->Texture.Proxy1D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_2D:
-         table = &ctx->Texture.Proxy2D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_3D:
-         table = &ctx->Texture.Proxy3D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetColorTableParameterfv(target)");
-            return;
-         }
-         table = &ctx->Texture.ProxyCubeMap->Palette;
-         break;
       case GL_SHARED_TEXTURE_PALETTE_EXT:
          table = &ctx->Texture.Palette;
          break;
@@ -981,8 +873,19 @@ _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params )
          table = &ctx->ProxyColorTable[COLORTABLE_POSTCOLORMATRIX];
          break;
       default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameterfv(target)");
-         return;
+         /* try texture targets */
+         {
+            struct gl_texture_object *texobj
+               = _mesa_select_tex_object(ctx, texUnit, target);
+            if (texobj) {
+               table = &texobj->Palette;
+            }
+            else {
+               _mesa_error(ctx, GL_INVALID_ENUM,
+                           "glGetColorTableParameterfv(target)");
+               return;
+            }
+         }
    }
 
    assert(table);
@@ -1029,40 +932,6 @@ _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params )
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    switch (target) {
-      case GL_TEXTURE_1D:
-         table = &texUnit->Current1D->Palette;
-         break;
-      case GL_TEXTURE_2D:
-         table = &texUnit->Current2D->Palette;
-         break;
-      case GL_TEXTURE_3D:
-         table = &texUnit->Current3D->Palette;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetColorTableParameteriv(target)");
-            return;
-         }
-         table = &texUnit->CurrentCubeMap->Palette;
-         break;
-      case GL_PROXY_TEXTURE_1D:
-         table = &ctx->Texture.Proxy1D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_2D:
-         table = &ctx->Texture.Proxy2D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_3D:
-         table = &ctx->Texture.Proxy3D->Palette;
-         break;
-      case GL_PROXY_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetColorTableParameteriv(target)");
-            return;
-         }
-         table = &ctx->Texture.ProxyCubeMap->Palette;
-         break;
       case GL_SHARED_TEXTURE_PALETTE_EXT:
          table = &ctx->Texture.Palette;
          break;
@@ -1161,8 +1030,19 @@ _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params )
          table = &ctx->ProxyColorTable[COLORTABLE_POSTCOLORMATRIX];
          break;
       default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameteriv(target)");
-         return;
+         /* Try texture targets */
+         {
+            struct gl_texture_object *texobj
+               = _mesa_select_tex_object(ctx, texUnit, target);
+            if (texobj) {
+               table = &texobj->Palette;
+            }
+            else {
+               _mesa_error(ctx, GL_INVALID_ENUM,
+                           "glGetColorTableParameteriv(target)");
+               return;
+            }
+         }
    }
 
    assert(table);
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index 92d8ceda29..bb48109317 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -6,9 +6,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -97,12 +97,9 @@
 #include "fog.h"
 #include "framebuffer.h"
 #include "get.h"
-#include "glthread.h"
-#include "glapioffsets.h"
 #include "histogram.h"
 #include "hint.h"
 #include "hash.h"
-#include "atifragshader.h"
 #include "light.h"
 #include "lines.h"
 #include "macros.h"
@@ -110,9 +107,6 @@
 #include "pixel.h"
 #include "points.h"
 #include "polygon.h"
-#if FEATURE_NV_vertex_program || FEATURE_NV_fragment_program
-#include "program.h"
-#endif
 #include "queryobj.h"
 #include "rastpos.h"
 #include "simple_list.h"
@@ -126,13 +120,19 @@
 #include "varray.h"
 #include "version.h"
 #include "vtxfmt.h"
+#include "glapi/glthread.h"
+#include "glapi/glapioffsets.h"
+#if FEATURE_NV_vertex_program || FEATURE_NV_fragment_program
+#include "shader/program.h"
+#endif
+#include "shader/shader_api.h"
+#include "shader/atifragshader.h"
 #if _HAVE_FULL_GL
 #include "math/m_translate.h"
 #include "math/m_matrix.h"
 #include "math/m_xform.h"
 #include "math/mathmod.h"
 #endif
-#include "shader_api.h"
 
 #ifdef USE_SPARC_ASM
 #include "sparc/sparc.h"
@@ -475,19 +475,12 @@ alloc_shared_state( GLcontext *ctx )
    if (!ss->Default2DArray)
       goto cleanup;
 
-   /* Effectively bind the default textures to all texture units */
-   ss->Default1D->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->Default2D->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->Default3D->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->DefaultCubeMap->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->DefaultRect->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->Default1DArray->RefCount += MAX_TEXTURE_IMAGE_UNITS;
-   ss->Default2DArray->RefCount += MAX_TEXTURE_IMAGE_UNITS;
+   /* sanity check */
+   assert(ss->Default1D->RefCount == 1);
 
    _glthread_INIT_MUTEX(ss->TexMutex);
    ss->TextureStateStamp = 0;
 
-
 #if FEATURE_EXT_framebuffer_object
    ss->FrameBuffers = _mesa_NewHashTable();
    if (!ss->FrameBuffers)
@@ -497,10 +490,9 @@ alloc_shared_state( GLcontext *ctx )
       goto cleanup;
 #endif
 
-
    return GL_TRUE;
 
- cleanup:
+cleanup:
    /* Ran out of memory at some point.  Free everything and return NULL */
    if (ss->DisplayList)
       _mesa_DeleteHashTable(ss->DisplayList);
@@ -552,6 +544,10 @@ alloc_shared_state( GLcontext *ctx )
       (*ctx->Driver.DeleteTexture)(ctx, ss->DefaultCubeMap);
    if (ss->DefaultRect)
       (*ctx->Driver.DeleteTexture)(ctx, ss->DefaultRect);
+   if (ss->Default1DArray)
+      (*ctx->Driver.DeleteTexture)(ctx, ss->Default1DArray);
+   if (ss->Default2DArray)
+      (*ctx->Driver.DeleteTexture)(ctx, ss->Default2DArray);
    if (ss)
       _mesa_free(ss);
    return GL_FALSE;
@@ -644,6 +640,33 @@ delete_shader_cb(GLuint id, void *data, void *userData)
    }
 }
 
+/**
+ * Callback for deleting a framebuffer object.  Called by _mesa_HashDeleteAll()
+ */
+static void
+delete_framebuffer_cb(GLuint id, void *data, void *userData)
+{
+   struct gl_framebuffer *fb = (struct gl_framebuffer *) data;
+   /* The fact that the framebuffer is in the hashtable means its refcount
+    * is one, but we're removing from the hashtable now.  So clear refcount.
+    */
+   /*assert(fb->RefCount == 1);*/
+   fb->RefCount = 0;
+   fb->Delete(fb);
+}
+
+/**
+ * Callback for deleting a renderbuffer object. Called by _mesa_HashDeleteAll()
+ */
+static void
+delete_renderbuffer_cb(GLuint id, void *data, void *userData)
+{
+   struct gl_renderbuffer *rb = (struct gl_renderbuffer *) data;
+   rb->RefCount = 0;  /* see comment for FBOs above */
+   rb->Delete(rb);
+}
+
+
 
 /**
  * Deallocate a shared state object and all children structures.
@@ -666,20 +689,6 @@ free_shared_state( GLcontext *ctx, struct gl_shared_state *ss )
    _mesa_HashDeleteAll(ss->DisplayList, delete_displaylist_cb, ctx);
    _mesa_DeleteHashTable(ss->DisplayList);
 
-   /*
-    * Free texture objects
-    */
-   ASSERT(ctx->Driver.DeleteTexture);
-   /* the default textures */
-   ctx->Driver.DeleteTexture(ctx, ss->Default1D);
-   ctx->Driver.DeleteTexture(ctx, ss->Default2D);
-   ctx->Driver.DeleteTexture(ctx, ss->Default3D);
-   ctx->Driver.DeleteTexture(ctx, ss->DefaultCubeMap);
-   ctx->Driver.DeleteTexture(ctx, ss->DefaultRect);
-   /* all other textures */
-   _mesa_HashDeleteAll(ss->TexObjects, delete_texture_cb, ctx);
-   _mesa_DeleteHashTable(ss->TexObjects);
-
 #if defined(FEATURE_NV_vertex_program) || defined(FEATURE_NV_fragment_program)
    _mesa_HashDeleteAll(ss->Programs, delete_program_cb, ctx);
    _mesa_DeleteHashTable(ss->Programs);
@@ -711,10 +720,29 @@ free_shared_state( GLcontext *ctx, struct gl_shared_state *ss )
 #endif
 
 #if FEATURE_EXT_framebuffer_object
+   _mesa_HashDeleteAll(ss->FrameBuffers, delete_framebuffer_cb, ctx);
    _mesa_DeleteHashTable(ss->FrameBuffers);
+   _mesa_HashDeleteAll(ss->RenderBuffers, delete_renderbuffer_cb, ctx);
    _mesa_DeleteHashTable(ss->RenderBuffers);
 #endif
 
+   /*
+    * Free texture objects (after FBOs since some textures might have
+    * been bound to FBOs).
+    */
+   ASSERT(ctx->Driver.DeleteTexture);
+   /* the default textures */
+   ctx->Driver.DeleteTexture(ctx, ss->Default1D);
+   ctx->Driver.DeleteTexture(ctx, ss->Default2D);
+   ctx->Driver.DeleteTexture(ctx, ss->Default3D);
+   ctx->Driver.DeleteTexture(ctx, ss->DefaultCubeMap);
+   ctx->Driver.DeleteTexture(ctx, ss->DefaultRect);
+   ctx->Driver.DeleteTexture(ctx, ss->Default1DArray);
+   ctx->Driver.DeleteTexture(ctx, ss->Default2DArray);
+   /* all other textures */
+   _mesa_HashDeleteAll(ss->TexObjects, delete_texture_cb, ctx);
+   _mesa_DeleteHashTable(ss->TexObjects);
+
    _glthread_DESTROY_MUTEX(ss->Mutex);
 
    _mesa_free(ss);
@@ -1168,18 +1196,20 @@ _mesa_create_context(const GLvisual *visual,
 void
 _mesa_free_context_data( GLcontext *ctx )
 {
-   /* if we're destroying the current context, unbind it first */
-   if (ctx == _mesa_get_current_context()) {
-      _mesa_make_current(NULL, NULL, NULL);
-   }
-   else {
-      /* unreference WinSysDraw/Read buffers */
-      _mesa_unreference_framebuffer(&ctx->WinSysDrawBuffer);
-      _mesa_unreference_framebuffer(&ctx->WinSysReadBuffer);
-      _mesa_unreference_framebuffer(&ctx->DrawBuffer);
-      _mesa_unreference_framebuffer(&ctx->ReadBuffer);
+   if (!_mesa_get_current_context()){
+      /* No current context, but we may need one in order to delete
+       * texture objs, etc.  So temporarily bind the context now.
+       */
+      _mesa_make_current(ctx, NULL, NULL);
    }
 
+   /* unreference WinSysDraw/Read buffers */
+   _mesa_unreference_framebuffer(&ctx->WinSysDrawBuffer);
+   _mesa_unreference_framebuffer(&ctx->WinSysReadBuffer);
+   _mesa_unreference_framebuffer(&ctx->DrawBuffer);
+   _mesa_unreference_framebuffer(&ctx->ReadBuffer);
+
+   _mesa_free_attrib_data(ctx);
    _mesa_free_lighting_data( ctx );
    _mesa_free_eval_data( ctx );
    _mesa_free_texture_data( ctx );
@@ -1211,6 +1241,11 @@ _mesa_free_context_data( GLcontext *ctx )
 
    if (ctx->Extensions.String)
       _mesa_free((void *) ctx->Extensions.String);
+
+   /* unbind the context if it's currently bound */
+   if (ctx == _mesa_get_current_context()) {
+      _mesa_make_current(NULL, NULL, NULL);
+   }
 }
 
 
@@ -1495,21 +1530,14 @@ _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer,
           */
          if (!newCtx->DrawBuffer || newCtx->DrawBuffer->Name == 0) {
             _mesa_reference_framebuffer(&newCtx->DrawBuffer, drawBuffer);
-         /* fix up the fb fields - these will end up wrong otherwise
-            if the DRIdrawable changes, and everything relies on them.
-            This is a bit messy (same as needed in _mesa_BindFramebufferEXT) */
-            int i;
-            GLenum buffers[MAX_DRAW_BUFFERS];
-            for(i = 0; i < newCtx->Const.MaxDrawBuffers; i++) {
-               buffers[i] = newCtx->Color.DrawBuffer[i];
-            }
-            _mesa_drawbuffers(newCtx, newCtx->Const.MaxDrawBuffers, buffers, NULL);
          }
          if (!newCtx->ReadBuffer || newCtx->ReadBuffer->Name == 0) {
             _mesa_reference_framebuffer(&newCtx->ReadBuffer, readBuffer);
-            _mesa_readbuffer_update_fields(newCtx, newCtx->Pixel.ReadBuffer);
          }
 
+         /* XXX only set this flag if we're really changing the draw/read
+          * framebuffer bindings.
+          */
 	 newCtx->NewState |= _NEW_BUFFERS;
 
 #if 1
diff --git a/src/mesa/main/context.h b/src/mesa/main/context.h
index b6a9d13149..099912aa15 100644
--- a/src/mesa/main/context.h
+++ b/src/mesa/main/context.h
@@ -49,7 +49,7 @@
 #define CONTEXT_H
 
 
-#include "glapi.h"
+#include "glapi/glapi.h"
 #include "imports.h"
 #include "mtypes.h"
 
diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c
index 91ddcbf9ed..be2f438a94 100644
--- a/src/mesa/main/dlist.c
+++ b/src/mesa/main/dlist.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -33,10 +33,6 @@
 #include "api_arrayelt.h"
 #include "api_loopback.h"
 #include "config.h"
-#if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
-#include "arbprogram.h"
-#include "program.h"
-#endif
 #include "attrib.h"
 #include "blend.h"
 #include "buffers.h"
@@ -57,7 +53,7 @@
 #include "extensions.h"
 #include "feedback.h"
 #include "get.h"
-#include "glapi.h"
+#include "glapi/glapi.h"
 #include "hash.h"
 #include "histogram.h"
 #include "image.h"
@@ -76,18 +72,22 @@
 #include "texstate.h"
 #include "mtypes.h"
 #include "varray.h"
+#if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
+#include "shader/arbprogram.h"
+#include "shader/program.h"
+#endif
 #if FEATURE_NV_vertex_program || FEATURE_NV_fragment_program
-#include "nvprogram.h"
-#include "program.h"
+#include "shader/nvprogram.h"
+#include "shader/program.h"
 #endif
 #if FEATURE_ATI_fragment_shader
-#include "atifragshader.h"
+#include "shader/atifragshader.h"
 #endif
 
 #include "math/m_matrix.h"
 #include "math/m_xform.h"
 
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 
 /**
@@ -2716,21 +2716,20 @@ save_PolygonMode(GLenum face, GLenum mode)
 }
 
 
-/*
- * Polygon stipple must have been upacked already!
- */
 static void GLAPIENTRY
 save_PolygonStipple(const GLubyte * pattern)
 {
    GET_CURRENT_CONTEXT(ctx);
+   GLvoid *image = unpack_image(2, 32, 32, 1, GL_COLOR_INDEX, GL_BITMAP,
+                                pattern, &ctx->Unpack);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
    n = ALLOC_INSTRUCTION(ctx, OPCODE_POLYGON_STIPPLE, 1);
    if (n) {
-      void *data;
-      n[1].data = _mesa_malloc(32 * 4);
-      data = n[1].data;         /* This needed for Acorn compiler */
-      MEMCPY(data, pattern, 32 * 4);
+      n[1].data = image; 
+   }
+   else if (image) {
+      _mesa_free(image);
    }
    if (ctx->ExecuteFlag) {
       CALL_PolygonStipple(ctx->Exec, ((GLubyte *) pattern));
@@ -6169,7 +6168,12 @@ execute_list(GLcontext *ctx, GLuint list)
             CALL_PolygonMode(ctx->Exec, (n[1].e, n[2].e));
             break;
          case OPCODE_POLYGON_STIPPLE:
-            CALL_PolygonStipple(ctx->Exec, ((GLubyte *) n[1].data));
+            {
+               const struct gl_pixelstore_attrib save = ctx->Unpack;
+               ctx->Unpack = ctx->DefaultPacking;
+               CALL_PolygonStipple(ctx->Exec, ((GLubyte *) n[1].data));
+               ctx->Unpack = save;      /* restore */
+            }
             break;
          case OPCODE_POLYGON_OFFSET:
             CALL_PolygonOffset(ctx->Exec, (n[1].f, n[2].f));
diff --git a/src/mesa/main/execmem.c b/src/mesa/main/execmem.c
index df3095232d..40f66d7da2 100644
--- a/src/mesa/main/execmem.c
+++ b/src/mesa/main/execmem.c
@@ -32,7 +32,7 @@
 
 
 #include "imports.h"
-#include "glthread.h"
+#include "glapi/glthread.h"
 
 
 
diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c
index 6f7effcce7..db04c785de 100644
--- a/src/mesa/main/fbobject.c
+++ b/src/mesa/main/fbobject.c
@@ -151,22 +151,18 @@ _mesa_remove_attachment(GLcontext *ctx, struct gl_renderbuffer_attachment *att)
 {
    if (att->Type == GL_TEXTURE) {
       ASSERT(att->Texture);
-      att->Texture->RefCount--;
-      if (att->Texture->RefCount == 0) {
-	 ctx->Driver.DeleteTexture(ctx, att->Texture);
+      if (ctx->Driver.FinishRenderTexture) {
+         /* tell driver we're done rendering to this texobj */
+         ctx->Driver.FinishRenderTexture(ctx, att);
       }
-      else {
-         /* tell driver that we're done rendering to this texture. */
-         if (ctx->Driver.FinishRenderTexture) {
-            ctx->Driver.FinishRenderTexture(ctx, att);
-         }
-      }
-      att->Texture = NULL;
+      _mesa_reference_texobj(&att->Texture, NULL); /* unbind */
+      ASSERT(!att->Texture);
    }
    if (att->Type == GL_TEXTURE || att->Type == GL_RENDERBUFFER_EXT) {
       ASSERT(att->Renderbuffer);
       ASSERT(!att->Texture);
-      _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
+      _mesa_reference_renderbuffer(&att->Renderbuffer, NULL); /* unbind */
+      ASSERT(!att->Renderbuffer);
    }
    att->Type = GL_NONE;
    att->Complete = GL_TRUE;
@@ -192,8 +188,8 @@ _mesa_set_texture_attachment(GLcontext *ctx,
       /* new attachment */
       _mesa_remove_attachment(ctx, att);
       att->Type = GL_TEXTURE;
-      att->Texture = texObj;
-      texObj->RefCount++;
+      assert(!att->Texture);
+      _mesa_reference_texobj(&att->Texture, texObj);
    }
 
    /* always update these fields */
@@ -966,9 +962,11 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
    }
 
    FLUSH_VERTICES(ctx, _NEW_BUFFERS);
+
    if (ctx->Driver.Flush) {  
       ctx->Driver.Flush(ctx);
    }
+
    if (framebuffer) {
       /* Binding a user-created framebuffer object */
       newFb = _mesa_lookup_framebuffer(ctx, framebuffer);
@@ -1002,32 +1000,18 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
     * XXX check if re-binding same buffer and skip some of this code.
     */
 
-   /* for window-framebuffers, re-initialize the fbo values, as they
-      could be wrong (makecurrent with a new drawable while still a fbo
-      was bound will lead to default init fbo values).
-      note that therefore the context ReadBuffer/DrawBuffer values are not
-      valid while fbo's are bound!!! */
    if (bindReadBuf) {
       _mesa_reference_framebuffer(&ctx->ReadBuffer, newFbread);
-      if (!newFbread->Name) {
-         _mesa_readbuffer_update_fields(ctx, ctx->Pixel.ReadBuffer);
-      }
    }
 
    if (bindDrawBuf) {
       /* check if old FB had any texture attachments */
       check_end_texture_render(ctx, ctx->DrawBuffer);
+
       /* check if time to delete this framebuffer */
       _mesa_reference_framebuffer(&ctx->DrawBuffer, newFb);
-      if (!newFb->Name) {
-         GLuint i;
-         GLenum buffers[MAX_DRAW_BUFFERS];
-         for(i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-            buffers[i] = ctx->Color.DrawBuffer[i];
-         }
-         _mesa_drawbuffers(ctx, ctx->Const.MaxDrawBuffers, buffers, NULL);
-      }
-      else {
+
+      if (newFb->Name != 0) {
          /* check if newly bound framebuffer has any texture attachments */
          check_begin_texture_render(ctx, newFb);
       }
diff --git a/src/mesa/main/framebuffer.c b/src/mesa/main/framebuffer.c
index f737c74f35..3e36197d88 100644
--- a/src/mesa/main/framebuffer.c
+++ b/src/mesa/main/framebuffer.c
@@ -32,12 +32,14 @@
 
 #include "glheader.h"
 #include "imports.h"
+#include "buffers.h"
 #include "context.h"
 #include "depthstencil.h"
 #include "mtypes.h"
 #include "fbobject.h"
 #include "framebuffer.h"
 #include "renderbuffer.h"
+#include "texobj.h"
 
 
 
@@ -65,7 +67,9 @@ compute_depth_max(struct gl_framebuffer *fb)
       fb->_DepthMax = 0xffffffff;
    }
    fb->_DepthMaxF = (GLfloat) fb->_DepthMax;
-   fb->_MRD = 1.0;  /* Minimum resolvable depth value, for polygon offset */
+
+   /* Minimum resolvable depth value, for polygon offset */
+   fb->_MRD = 1.0 / fb->_DepthMaxF;
 }
 
 
@@ -135,7 +139,7 @@ _mesa_initialize_framebuffer(struct gl_framebuffer *fb, const GLvisual *visual)
    /* save the visual */
    fb->Visual = *visual;
 
-   /* Init glRead/DrawBuffer state */
+   /* Init read/draw renderbuffer state */
    if (visual->doubleBufferMode) {
       fb->ColorDrawBuffer[0] = GL_BACK;
       fb->_ColorDrawBufferMask[0] = BUFFER_BIT_BACK_LEFT;
@@ -190,17 +194,11 @@ _mesa_free_framebuffer_data(struct gl_framebuffer *fb)
          _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
       }
       if (att->Texture) {
-         /* render to texture */
-         att->Texture->RefCount--;
-         if (att->Texture->RefCount == 0) {
-            GET_CURRENT_CONTEXT(ctx);
-            if (ctx) {
-               ctx->Driver.DeleteTexture(ctx, att->Texture);
-            }
-         }
+         _mesa_reference_texobj(&att->Texture, NULL);
       }
+      ASSERT(!att->Renderbuffer);
+      ASSERT(!att->Texture);
       att->Type = GL_NONE;
-      att->Texture = NULL;
    }
 
    /* unbind _Depth/_StencilBuffer to decr ref counts */
@@ -586,7 +584,7 @@ _mesa_update_stencil_buffer(GLcontext *ctx,
 
 
 /**
- * Update the list of color drawing renderbuffer pointers.
+ * Update the (derived) list of color drawing renderbuffer pointers.
  * Later, when we're rendering we'll loop from 0 to _NumColorDrawBuffers
  * writing colors.
  */
@@ -630,7 +628,7 @@ update_color_draw_buffers(GLcontext *ctx, struct gl_framebuffer *fb)
 
 
 /**
- * Update the color read renderbuffer pointer.
+ * Update the (derived) color read renderbuffer pointer.
  * Unlike the DrawBuffer, we can only read from one (or zero) color buffers.
  */
 static void
@@ -652,19 +650,51 @@ update_color_read_buffer(GLcontext *ctx, struct gl_framebuffer *fb)
 }
 
 
+/**
+ * Update a gl_framebuffer's derived state.
+ *
+ * Specifically, update these framebuffer fields:
+ *    _ColorDrawBuffers
+ *    _NumColorDrawBuffers
+ *    _ColorReadBuffer
+ *    _DepthBuffer
+ *    _StencilBuffer
+ *
+ * If the framebuffer is user-created, make sure it's complete.
+ *
+ * The following functions (at least) can effect framebuffer state:
+ * glReadBuffer, glDrawBuffer, glDrawBuffersARB, glFramebufferRenderbufferEXT,
+ * glRenderbufferStorageEXT.
+ */
 static void
 update_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
 {
-   /* Completeness only matters for user-created framebuffers */
-   if (fb->Name != 0) {
-      /* XXX: EXT_framebuffer_blit:
-         framebuffer must still be complete wrt read/draw? */
+   if (fb->Name == 0) {
+      /* This is a window-system framebuffer */
+      /* Need to update the FB's GL_DRAW_BUFFER state to match the
+       * context state (GL_READ_BUFFER too).
+       */
+      if (fb->ColorDrawBuffer[0] != ctx->Color.DrawBuffer[0]) {
+         _mesa_drawbuffers(ctx, ctx->Const.MaxDrawBuffers,
+                           ctx->Color.DrawBuffer, NULL);
+      }
+      if (fb->ColorReadBuffer != ctx->Pixel.ReadBuffer) {
+         
+      }
+   }
+   else {
+      /* This is a user-created framebuffer.
+       * Completeness only matters for user-created framebuffers.
+       */
       _mesa_test_framebuffer_completeness(ctx, fb);
       _mesa_update_framebuffer_visual(fb);
    }
 
-   /* update_color_draw/read_buffers not needed for
-      read/draw only fb, but shouldn't hurt ??? */
+   /* Strictly speaking, we don't need to update the draw-state
+    * if this FB is bound as ctx->ReadBuffer (and conversely, the
+    * read-state if this FB is bound as ctx->DrawBuffer), but no
+    * harm.
+    */
    update_color_draw_buffers(ctx, fb);
    update_color_read_buffer(ctx, fb);
    _mesa_update_depth_buffer(ctx, fb, BUFFER_DEPTH);
@@ -673,28 +703,19 @@ update_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
    compute_depth_max(fb);
 }
 
+
 /**
  * Update state related to the current draw/read framebuffers.
- * Specifically, update these framebuffer fields:
- *    _ColorDrawBuffers
- *    _NumColorDrawBuffers
- *    _ColorReadBuffer
- *    _DepthBuffer
- *    _StencilBuffer
- * If the current framebuffer is user-created, make sure it's complete.
- * The following functions can effect this state:  glReadBuffer,
- * glDrawBuffer, glDrawBuffersARB, glFramebufferRenderbufferEXT,
- * glRenderbufferStorageEXT.
  */
 void
 _mesa_update_framebuffer(GLcontext *ctx)
 {
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_framebuffer *fbread = ctx->ReadBuffer;
+   struct gl_framebuffer *drawFb = ctx->DrawBuffer;
+   struct gl_framebuffer *readFb = ctx->ReadBuffer;
 
-   update_framebuffer(ctx, fb);
-   if (fbread != fb)
-      update_framebuffer(ctx, fbread);
+   update_framebuffer(ctx, drawFb);
+   if (readFb != drawFb)
+      update_framebuffer(ctx, readFb);
 }
 
 
diff --git a/src/mesa/main/glheader.h b/src/mesa/main/glheader.h
index 63dd002a41..fd4127558a 100644
--- a/src/mesa/main/glheader.h
+++ b/src/mesa/main/glheader.h
@@ -92,7 +92,7 @@
 #endif
 
 #ifdef WGLAPI
-#undef WGLAPI
+#	undef WGLAPI
 #endif
 
 #if !defined(OPENSTEP) && (defined(__WIN32__) && !defined(__CYGWIN__)) && !defined(BUILD_FOR_SNAP)
diff --git a/src/mesa/main/hash.c b/src/mesa/main/hash.c
index 2d5bcc3e01..ffb2c4d946 100644
--- a/src/mesa/main/hash.c
+++ b/src/mesa/main/hash.c
@@ -37,7 +37,7 @@
 
 #include "glheader.h"
 #include "imports.h"
-#include "glthread.h"
+#include "glapi/glthread.h"
 #include "hash.h"
 
 
diff --git a/src/mesa/main/image.c b/src/mesa/main/image.c
index e2e7f806ab..347cec66ef 100644
--- a/src/mesa/main/image.c
+++ b/src/mesa/main/image.c
@@ -850,7 +850,7 @@ _mesa_unpack_bitmap( GLint width, GLint height, const GLubyte *pixels,
          return NULL;
       }
 
-      if (packing->SkipPixels == 0) {
+      if ((packing->SkipPixels & 7) == 0) {
          _mesa_memcpy( dst, src, width_in_bytes );
          if (packing->LsbFirst) {
             flip_bytes( dst, width_in_bytes );
@@ -942,7 +942,7 @@ _mesa_pack_bitmap( GLint width, GLint height, const GLubyte *source,
       if (!dst)
          return;
 
-      if (packing->SkipPixels == 0) {
+      if ((packing->SkipPixels & 7) == 0) {
          _mesa_memcpy( dst, src, width_in_bytes );
          if (packing->LsbFirst) {
             flip_bytes( dst, width_in_bytes );
@@ -961,20 +961,20 @@ _mesa_pack_bitmap( GLint width, GLint height, const GLubyte *source,
                if (*s & srcMask) {
                   *d |= dstMask;
                }
-               if (srcMask == 128) {
-                  srcMask = 1;
+               if (srcMask == 1) {
+                  srcMask = 128;
                   s++;
                }
                else {
-                  srcMask = srcMask << 1;
+                  srcMask = srcMask >> 1;
                }
-               if (dstMask == 1) {
-                  dstMask = 128;
+               if (dstMask == 128) {
+                  dstMask = 1;
                   d++;
                   *d = 0;
                }
                else {
-                  dstMask = dstMask >> 1;
+                  dstMask = dstMask << 1;
                }
             }
          }
@@ -3648,11 +3648,13 @@ _mesa_unpack_stencil_span( const GLcontext *ctx, GLuint n,
     * Try simple cases first
     */
    if (transferOps == 0 &&
+       !ctx->Pixel.MapStencilFlag &&
        srcType == GL_UNSIGNED_BYTE &&
        dstType == GL_UNSIGNED_BYTE) {
       _mesa_memcpy(dest, source, n * sizeof(GLubyte));
    }
    else if (transferOps == 0 &&
+            !ctx->Pixel.MapStencilFlag &&
             srcType == GL_UNSIGNED_INT &&
             dstType == GL_UNSIGNED_INT &&
             !srcPacking->SwapBytes) {
@@ -3668,19 +3670,17 @@ _mesa_unpack_stencil_span( const GLcontext *ctx, GLuint n,
       extract_uint_indexes(n, indexes, GL_STENCIL_INDEX, srcType, source,
                            srcPacking);
 
-      if (transferOps) {
-         if (transferOps & IMAGE_SHIFT_OFFSET_BIT) {
-            /* shift and offset indexes */
-            shift_and_offset_ci(ctx, n, indexes);
-         }
+      if (transferOps & IMAGE_SHIFT_OFFSET_BIT) {
+         /* shift and offset indexes */
+         shift_and_offset_ci(ctx, n, indexes);
+      }
 
-         if (ctx->Pixel.MapStencilFlag) {
-            /* Apply stencil lookup table */
-            GLuint mask = ctx->PixelMaps.StoS.Size - 1;
-            GLuint i;
-            for (i=0;i<n;i++) {
-               indexes[i] = ctx->PixelMaps.StoS.Map[ indexes[i] & mask ];
-            }
+      if (ctx->Pixel.MapStencilFlag) {
+         /* Apply stencil lookup table */
+         const GLuint mask = ctx->PixelMaps.StoS.Size - 1;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            indexes[i] = ctx->PixelMaps.StoS.Map[ indexes[i] & mask ];
          }
       }
 
@@ -3733,7 +3733,7 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
 
    switch (dstType) {
    case GL_UNSIGNED_BYTE:
-      if (sizeof(GLstencil) == 8) {
+      if (sizeof(GLstencil) == 1) {
          _mesa_memcpy( dest, source, n );
       }
       else {
@@ -3745,14 +3745,11 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
       }
       break;
    case GL_BYTE:
-      if (sizeof(GLstencil) == 8) {
-         _mesa_memcpy( dest, source, n );
-      }
-      else {
+      {
          GLbyte *dst = (GLbyte *) dest;
          GLuint i;
          for (i=0;i<n;i++) {
-            dst[i] = (GLbyte) source[i];
+            dst[i] = (GLbyte) (source[i] & 0x7f);
          }
       }
       break;
@@ -3797,7 +3794,7 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
          GLint *dst = (GLint *) dest;
          GLuint i;
          for (i=0;i<n;i++) {
-            *dst++ = (GLint) source[i];
+            dst[i] = (GLint) source[i];
          }
          if (dstPacking->SwapBytes) {
             _mesa_swap4( (GLuint *) dst, n );
@@ -3882,9 +3879,19 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
         }                                                               \
     } while (0)
 
+
+/**
+ * Unpack a row of depth/z values from memory, returning GLushort, GLuint
+ * or GLfloat values.
+ * The glPixelTransfer (scale/bias) params will be applied.
+ *
+ * \param dstType  one of GL_UNSIGNED_SHORT, GL_UNSIGNED_INT, GL_FLOAT
+ * \param depthMax  max value for returned GLushort or GLuint values
+ *                  (ignored for GLfloat).
+ */
 void
 _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
-                         GLenum dstType, GLvoid *dest, GLfloat depthScale,
+                         GLenum dstType, GLvoid *dest, GLuint depthMax,
                          GLenum srcType, const GLvoid *source,
                          const struct gl_pixelstore_attrib *srcPacking )
 {
@@ -3907,7 +3914,9 @@ _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
          }
          return;
       }
-      if (srcType == GL_UNSIGNED_SHORT && dstType == GL_UNSIGNED_INT) {
+      if (srcType == GL_UNSIGNED_SHORT
+          && dstType == GL_UNSIGNED_INT
+          && depthMax == 0xffffffff) {
          const GLushort *src = (const GLushort *) source;
          GLuint *dst = (GLuint *) dest;
          GLuint i;
@@ -3955,7 +3964,7 @@ _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
          break;
       case GL_UNSIGNED_INT_24_8_EXT: /* GL_EXT_packed_depth_stencil */
          if (dstType == GL_UNSIGNED_INT &&
-             depthScale == (GLfloat) 0xffffff &&
+             depthMax == 0xffffff &&
              ctx->Pixel.DepthScale == 1.0 &&
              ctx->Pixel.DepthBias == 0.0) {
             const GLuint *src = (const GLuint *) source;
@@ -4033,16 +4042,16 @@ _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
    if (dstType == GL_UNSIGNED_INT) {
       GLuint *zValues = (GLuint *) dest;
       GLuint i;
-      if (depthScale <= (GLfloat) 0xffffff) {
+      if (depthMax <= 0xffffff) {
          /* no overflow worries */
          for (i = 0; i < n; i++) {
-            zValues[i] = (GLuint) (depthValues[i] * depthScale);
+            zValues[i] = (GLuint) (depthValues[i] * (GLfloat) depthMax);
          }
       }
       else {
          /* need to use double precision to prevent overflow problems */
          for (i = 0; i < n; i++) {
-            GLdouble z = depthValues[i] * depthScale;
+            GLdouble z = depthValues[i] * (GLfloat) depthMax;
             if (z >= (GLdouble) 0xffffffff)
                zValues[i] = 0xffffffff;
             else
@@ -4053,14 +4062,14 @@ _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
    else if (dstType == GL_UNSIGNED_SHORT) {
       GLushort *zValues = (GLushort *) dest;
       GLuint i;
-      ASSERT(depthScale <= 65535.0);
+      ASSERT(depthMax <= 0xffff);
       for (i = 0; i < n; i++) {
-         zValues[i] = (GLushort) (depthValues[i] * depthScale);
+         zValues[i] = (GLushort) (depthValues[i] * (GLfloat) depthMax);
       }
    }
    else {
       ASSERT(dstType == GL_FLOAT);
-      ASSERT(depthScale == 1.0F);
+      /*ASSERT(depthMax == 1.0F);*/
    }
 }
 
diff --git a/src/mesa/main/image.h b/src/mesa/main/image.h
index 990398a7c4..2a16989fa7 100644
--- a/src/mesa/main/image.h
+++ b/src/mesa/main/image.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -181,7 +181,7 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
 
 extern void
 _mesa_unpack_depth_span( const GLcontext *ctx, GLuint n,
-                         GLenum dstType, GLvoid *dest, GLfloat depthScale,
+                         GLenum dstType, GLvoid *dest, GLuint depthMax,
                          GLenum srcType, const GLvoid *source,
                          const struct gl_pixelstore_attrib *srcPacking );
 
diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c
index 8a5dfdb4b8..3ae56c8b0b 100644
--- a/src/mesa/main/imports.c
+++ b/src/mesa/main/imports.c
@@ -575,7 +575,11 @@ _mesa_ffs(int i)
  *          if no bits set.
  */
 int
+#ifdef __MINGW32__
+_mesa_ffsll(long val)
+#else
 _mesa_ffsll(long long val)
+#endif
 {
 #ifdef ffsll
    return ffsll(val);
diff --git a/src/mesa/main/imports.h b/src/mesa/main/imports.h
index 9be8014a13..ebdfc452a7 100644
--- a/src/mesa/main/imports.h
+++ b/src/mesa/main/imports.h
@@ -700,7 +700,11 @@ extern int
 _mesa_ffs(int i);
 
 extern int
+#ifdef __MINGW32__
+_mesa_ffsll(long i);
+#else
 _mesa_ffsll(long long i);
+#endif
 
 extern unsigned int
 _mesa_bitcount(unsigned int n);
diff --git a/src/mesa/main/lines.c b/src/mesa/main/lines.c
index dc7195d4eb..0c2dbf915a 100644
--- a/src/mesa/main/lines.c
+++ b/src/mesa/main/lines.c
@@ -55,9 +55,6 @@ _mesa_LineWidth( GLfloat width )
 
    FLUSH_VERTICES(ctx, _NEW_LINE);
    ctx->Line.Width = width;
-   ctx->Line._Width = CLAMP(width,
-			    ctx->Const.MinLineWidth,
-			    ctx->Const.MaxLineWidth);
 
    if (ctx->Driver.LineWidth)
       ctx->Driver.LineWidth(ctx, width);
@@ -105,13 +102,12 @@ _mesa_LineStipple( GLint factor, GLushort pattern )
  * Initializes __GLcontextRec::Line and line related constants in
  * __GLcontextRec::Const.
  */
-void GLAPIENTRY _mesa_init_line( GLcontext * ctx )
+void GLAPIENTRY
+_mesa_init_line( GLcontext * ctx )
 {
-   /* Line group */
    ctx->Line.SmoothFlag = GL_FALSE;
    ctx->Line.StippleFlag = GL_FALSE;
    ctx->Line.Width = 1.0;
-   ctx->Line._Width = 1.0;
    ctx->Line.StipplePattern = 0xffff;
    ctx->Line.StippleFactor = 1;
 }
diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c
index 9f3db22b75..a9260c847b 100644
--- a/src/mesa/main/mipmap.c
+++ b/src/mesa/main/mipmap.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -957,7 +957,8 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target,
       GLint  components, size;
       GLchan *dst;
 
-      assert(texObj->Target == GL_TEXTURE_2D);
+      assert(texObj->Target == GL_TEXTURE_2D ||
+             texObj->Target == GL_TEXTURE_CUBE_MAP_ARB);
 
       if (srcImage->_BaseFormat == GL_RGB) {
          convertFormat = &_mesa_texformat_rgb;
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 05c08c19fe..1145f361b1 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -38,8 +38,8 @@
 #include "glheader.h"
 #include <GL/internal/glcore.h>	/* __GLcontextModes (GLvisual) */
 #include "config.h"		/* Hardwired parameters */
-#include "glapitable.h"
-#include "glthread.h"
+#include "glapi/glapitable.h"
+#include "glapi/glthread.h"
 #include "math/m_matrix.h"	/* GLmatrix */
 #include "bitset.h"
 
@@ -917,7 +917,6 @@ struct gl_line_attrib
    GLushort StipplePattern;	/**< Stipple pattern */
    GLint StippleFactor;		/**< Stipple repeat factor */
    GLfloat Width;		/**< Line width */
-   GLfloat _Width;		/**< Clamped Line width */
 };
 
 
@@ -1063,7 +1062,6 @@ struct gl_point_attrib
 {
    GLboolean SmoothFlag;	/**< True if GL_POINT_SMOOTH is enabled */
    GLfloat Size;		/**< User-specified point size */
-   GLfloat _Size;		/**< Size clamped to Const.Min/MaxPointSize */
    GLfloat Params[3];		/**< GL_EXT_point_parameters */
    GLfloat MinSize, MaxSize;	/**< GL_EXT_point_parameters */
    GLfloat Threshold;		/**< GL_EXT_point_parameters */
@@ -1133,13 +1131,13 @@ struct gl_stencil_attrib
  * An index for each type of texture object
  */
 /*@{*/
-#define TEXTURE_1D_INDEX    0
-#define TEXTURE_2D_INDEX    1
-#define TEXTURE_3D_INDEX    2
-#define TEXTURE_CUBE_INDEX  3
-#define TEXTURE_RECT_INDEX  4
-#define TEXTURE_1D_ARRAY_INDEX    5
-#define TEXTURE_2D_ARRAY_INDEX    6
+#define TEXTURE_1D_INDEX       0
+#define TEXTURE_2D_INDEX       1
+#define TEXTURE_3D_INDEX       2
+#define TEXTURE_CUBE_INDEX     3
+#define TEXTURE_RECT_INDEX     4
+#define TEXTURE_1D_ARRAY_INDEX 5
+#define TEXTURE_2D_ARRAY_INDEX 6
 /*@}*/
 
 /**
@@ -1147,13 +1145,13 @@ struct gl_stencil_attrib
  * Used for Texture.Unit[]._ReallyEnabled flags.
  */
 /*@{*/
-#define TEXTURE_1D_BIT   (1 << TEXTURE_1D_INDEX)
-#define TEXTURE_2D_BIT   (1 << TEXTURE_2D_INDEX)
-#define TEXTURE_3D_BIT   (1 << TEXTURE_3D_INDEX)
-#define TEXTURE_CUBE_BIT (1 << TEXTURE_CUBE_INDEX)
-#define TEXTURE_RECT_BIT (1 << TEXTURE_RECT_INDEX)
-#define TEXTURE_1D_ARRAY_BIT   (1 << TEXTURE_1D_ARRAY_INDEX)
-#define TEXTURE_2D_ARRAY_BIT   (1 << TEXTURE_2D_ARRAY_INDEX)
+#define TEXTURE_1D_BIT       (1 << TEXTURE_1D_INDEX)
+#define TEXTURE_2D_BIT       (1 << TEXTURE_2D_INDEX)
+#define TEXTURE_3D_BIT       (1 << TEXTURE_3D_INDEX)
+#define TEXTURE_CUBE_BIT     (1 << TEXTURE_CUBE_INDEX)
+#define TEXTURE_RECT_BIT     (1 << TEXTURE_RECT_INDEX)
+#define TEXTURE_1D_ARRAY_BIT (1 << TEXTURE_1D_ARRAY_INDEX)
+#define TEXTURE_2D_ARRAY_BIT (1 << TEXTURE_2D_ARRAY_INDEX)
 /*@}*/
 
 
@@ -1327,8 +1325,6 @@ struct gl_texture_format
 };
 
 
-#define MAX_3D_TEXTURE_SIZE (1 << (MAX_3D_TEXTURE_LEVELS - 1))
-
 /**
  * Texture image state.  Describes the dimensions of a texture image,
  * the texel format and pointers to Texel Fetch functions.
@@ -1393,7 +1389,7 @@ struct gl_texture_image
 #define FACE_NEG_Y   3
 #define FACE_POS_Z   4
 #define FACE_NEG_Z   5
-#define MAX_FACES  6
+#define MAX_FACES    6
 /*@}*/
 
 
@@ -1404,6 +1400,7 @@ struct gl_texture_image
  */
 struct gl_texture_object
 {
+   _glthread_Mutex Mutex;	/**< for thread safety */
    GLint RefCount;		/**< reference count */
    GLuint Name;			/**< the user-visible texture object ID */
    GLenum Target;               /**< GL_TEXTURE_1D, GL_TEXTURE_2D, etc. */
@@ -1539,17 +1536,6 @@ struct gl_texture_unit
 
    struct gl_texture_object *_Current; /**< Points to really enabled tex obj */
 
-   /** These are used for glPush/PopAttrib */
-   /*@{*/
-   struct gl_texture_object Saved1D;
-   struct gl_texture_object Saved2D;
-   struct gl_texture_object Saved3D;
-   struct gl_texture_object SavedCubeMap;
-   struct gl_texture_object SavedRect;
-   struct gl_texture_object Saved1DArray;
-   struct gl_texture_object Saved2DArray;
-   /*@}*/
-
    /** GL_SGI_texture_color_table */
    /*@{*/
    struct gl_color_table ColorTable;
@@ -1590,13 +1576,8 @@ struct gl_texture_attrib
 
    struct gl_texture_unit Unit[MAX_TEXTURE_UNITS];
 
-   struct gl_texture_object *Proxy1D;
-   struct gl_texture_object *Proxy2D;
-   struct gl_texture_object *Proxy3D;
-   struct gl_texture_object *ProxyCubeMap;
-   struct gl_texture_object *ProxyRect;
-   struct gl_texture_object *Proxy1DArray;
-   struct gl_texture_object *Proxy2DArray;
+   /** Proxy texture objects */
+   struct gl_texture_object *ProxyTex[NUM_TEXTURE_TARGETS];
 
    /** GL_EXT_shared_texture_palette */
    GLboolean SharedPalette;
@@ -2193,12 +2174,11 @@ struct gl_shared_state
     * \todo Improve the granularity of locking.
     */
    /*@{*/
-   _glthread_Mutex TexMutex;		   /**< texobj thread safety */
-   GLuint TextureStateStamp;	           /**< state notification for shared tex  */
+   _glthread_Mutex TexMutex;		/**< texobj thread safety */
+   GLuint TextureStateStamp;	        /**< state notification for shared tex */
    /*@}*/
 
 
-
    /**
     * \name Vertex/fragment programs
     */
diff --git a/src/mesa/main/points.c b/src/mesa/main/points.c
index 0f562420b0..88e4989405 100644
--- a/src/mesa/main/points.c
+++ b/src/mesa/main/points.c
@@ -5,7 +5,7 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  7.0
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -57,10 +57,6 @@ _mesa_PointSize( GLfloat size )
 
    FLUSH_VERTICES(ctx, _NEW_POINT);
    ctx->Point.Size = size;
-   /* XXX correct clamp limits? */
-   ctx->Point._Size = CLAMP(ctx->Point.Size,
-			    ctx->Point.MinSize,
-			    ctx->Point.MaxSize);
 
    if (ctx->Driver.PointSize)
       ctx->Driver.PointSize(ctx, size);
@@ -86,8 +82,13 @@ _mesa_PointParameteriNV( GLenum pname, GLint param )
 void GLAPIENTRY
 _mesa_PointParameterivNV( GLenum pname, const GLint *params )
 {
-   const GLfloat value = (GLfloat) params[0];
-   _mesa_PointParameterfvEXT(pname, &value);
+   GLfloat p[3];
+   p[0] = (GLfloat) params[0];
+   if (pname == GL_DISTANCE_ATTENUATION_EXT) {
+      p[1] = (GLfloat) params[1];
+      p[2] = (GLfloat) params[2];
+   }
+   _mesa_PointParameterfvEXT(pname, p);
 }
 
 
@@ -253,7 +254,6 @@ _mesa_init_point(GLcontext *ctx)
 
    ctx->Point.SmoothFlag = GL_FALSE;
    ctx->Point.Size = 1.0;
-   ctx->Point._Size = 1.0;
    ctx->Point.Params[0] = 1.0;
    ctx->Point.Params[1] = 0.0;
    ctx->Point.Params[2] = 0.0;
diff --git a/src/mesa/main/shaders.c b/src/mesa/main/shaders.c
index 58be1f46e5..7bf8808767 100644
--- a/src/mesa/main/shaders.c
+++ b/src/mesa/main/shaders.c
@@ -83,7 +83,7 @@ _mesa_CreateShader(GLenum type)
 }
 
 
-GLhandleARB APIENTRY
+GLhandleARB GLAPIENTRY
 _mesa_CreateShaderObjectARB(GLenum type)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -99,7 +99,7 @@ _mesa_CreateProgram(void)
 }
 
 
-GLhandleARB APIENTRY
+GLhandleARB GLAPIENTRY
 _mesa_CreateProgramObjectARB(void)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -319,7 +319,7 @@ _mesa_GetUniformivARB(GLhandleARB program, GLint location, GLint * params)
 
 
 #if 0
-GLint APIENTRY
+GLint GLAPIENTRY
 _mesa_GetUniformLocation(GLuint program, const GLcharARB *name)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -336,7 +336,7 @@ _mesa_GetHandleARB(GLenum pname)
 }
 
 
-GLint APIENTRY
+GLint GLAPIENTRY
 _mesa_GetUniformLocationARB(GLhandleARB programObj, const GLcharARB *name)
 {
    GET_CURRENT_CONTEXT(ctx);
diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c
index 0a83abc7dd..444f227760 100644
--- a/src/mesa/main/state.c
+++ b/src/mesa/main/state.c
@@ -35,10 +35,10 @@
 #include "accum.h"
 #include "api_loopback.h"
 #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
-#include "arbprogram.h"
+#include "shader/arbprogram.h"
 #endif
 #if FEATURE_ATI_fragment_shader
-#include "atifragshader.h"
+#include "shader/atifragshader.h"
 #endif
 #include "attrib.h"
 #include "blend.h"
@@ -85,18 +85,18 @@
 #include "mtypes.h"
 #include "varray.h"
 #if FEATURE_NV_vertex_program
-#include "nvprogram.h"
+#include "shader/nvprogram.h"
 #endif
 #if FEATURE_NV_fragment_program
-#include "nvprogram.h"
-#include "program.h"
+#include "shader/nvprogram.h"
+#include "shader/program.h"
 #include "texenvprogram.h"
 #endif
 #if FEATURE_ARB_shader_objects
 #include "shaders.h"
 #endif
 #include "debug.h"
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 
 
@@ -1068,7 +1068,7 @@ update_tricaps(GLcontext *ctx, GLbitfield new_state)
    if (1/*new_state & _NEW_POINT*/) {
       if (ctx->Point.SmoothFlag)
          ctx->_TriangleCaps |= DD_POINT_SMOOTH;
-      if (ctx->Point._Size != 1.0F)
+      if (ctx->Point.Size != 1.0F)
          ctx->_TriangleCaps |= DD_POINT_SIZE;
       if (ctx->Point._Attenuated)
          ctx->_TriangleCaps |= DD_POINT_ATTEN;
@@ -1082,7 +1082,7 @@ update_tricaps(GLcontext *ctx, GLbitfield new_state)
          ctx->_TriangleCaps |= DD_LINE_SMOOTH;
       if (ctx->Line.StippleFlag)
          ctx->_TriangleCaps |= DD_LINE_STIPPLE;
-      if (ctx->Line._Width != 1.0)
+      if (ctx->Line.Width != 1.0)
          ctx->_TriangleCaps |= DD_LINE_WIDTH;
    }
 
diff --git a/src/mesa/main/texcompress_fxt1.c b/src/mesa/main/texcompress_fxt1.c
index 411d51cfcc..b6991f45ed 100644
--- a/src/mesa/main/texcompress_fxt1.c
+++ b/src/mesa/main/texcompress_fxt1.c
@@ -302,7 +302,12 @@ const struct gl_texture_format _mesa_texformat_rgba_fxt1 = {
 
 #define FX64_NATIVE 1
 
+#ifdef __MINGW32__
+typedef unsigned long Fx64;
+#else
 typedef unsigned long long Fx64;
+#endif
+
 
 #define FX64_MOV32(a, b) a = b
 #define FX64_OR32(a, b)  a |= b
diff --git a/src/mesa/main/texcompress_s3tc.c b/src/mesa/main/texcompress_s3tc.c
index c48063d919..4f329cdf59 100644
--- a/src/mesa/main/texcompress_s3tc.c
+++ b/src/mesa/main/texcompress_s3tc.c
@@ -577,6 +577,32 @@ const struct gl_texture_format _mesa_texformat_rgb_dxt1 = {
    NULL					/* StoreTexel */
 };
 
+#if FEATURE_EXT_texture_sRGB
+const struct gl_texture_format _mesa_texformat_srgb_dxt1 = {
+   MESA_FORMAT_SRGB_DXT1,		/* MesaFormat */
+   GL_RGB,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   4, /*approx*/			/* RedBits */
+   4, /*approx*/			/* GreenBits */
+   4, /*approx*/			/* BlueBits */
+   0,					/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   0,					/* TexelBytes */
+   texstore_rgb_dxt1,			/* StoreTexImageFunc */
+   NULL, /*impossible*/ 		/* FetchTexel1D */
+   fetch_texel_2d_rgb_dxt1, 		/* FetchTexel2D */
+   NULL, /*impossible*/ 		/* FetchTexel3D */
+   NULL, /*impossible*/ 		/* FetchTexel1Df */
+   fetch_texel_2d_f_rgb_dxt1, 		/* FetchTexel2Df */
+   NULL, /*impossible*/ 		/* FetchTexel3Df */
+   NULL					/* StoreTexel */
+};
+#endif
+
 const struct gl_texture_format _mesa_texformat_rgba_dxt1 = {
    MESA_FORMAT_RGBA_DXT1,		/* MesaFormat */
    GL_RGBA,				/* BaseFormat */
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index 1a46c10ffa..9c84da985e 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -28,17 +28,17 @@
 #include "glheader.h"
 #include "macros.h"
 #include "enums.h"
-#include "prog_parameter.h"
-#include "prog_instruction.h"
-#include "prog_print.h"
-#include "prog_statevars.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_print.h"
+#include "shader/prog_statevars.h"
 #include "texenvprogram.h"
 
 /**
  * According to Glean's texCombine test, no more than 21 instructions
  * are needed.  Allow a few extra just in case.
  */
-#define MAX_INSTRUCTIONS 24
+#define MAX_INSTRUCTIONS ((MAX_TEXTURE_UNITS * 6) + 10) /* see bug 9829 */
 
 #define DISASSEM (MESA_VERBOSE & VERBOSE_DISASSEM)
 
diff --git a/src/mesa/main/texformat.h b/src/mesa/main/texformat.h
index 55851db701..82023b946d 100644
--- a/src/mesa/main/texformat.h
+++ b/src/mesa/main/texformat.h
@@ -97,6 +97,7 @@ enum _format {
    MESA_FORMAT_SRGBA8,
    MESA_FORMAT_SL8,
    MESA_FORMAT_SLA8,
+   MESA_FORMAT_SRGB_DXT1,
    /*@}*/
 #endif
 
@@ -168,6 +169,7 @@ extern const struct gl_texture_format _mesa_texformat_srgb8;
 extern const struct gl_texture_format _mesa_texformat_srgba8;
 extern const struct gl_texture_format _mesa_texformat_sl8;
 extern const struct gl_texture_format _mesa_texformat_sla8;
+extern const struct gl_texture_format _mesa_texformat_srgb_dxt1;
 /*@}*/
 #endif
 
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index f902365b9b..1bc52d220a 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -760,15 +760,15 @@ _mesa_select_tex_object(GLcontext *ctx, const struct gl_texture_unit *texUnit,
       case GL_TEXTURE_1D:
          return texUnit->Current1D;
       case GL_PROXY_TEXTURE_1D:
-         return ctx->Texture.Proxy1D;
+         return ctx->Texture.ProxyTex[TEXTURE_1D_INDEX];
       case GL_TEXTURE_2D:
          return texUnit->Current2D;
       case GL_PROXY_TEXTURE_2D:
-         return ctx->Texture.Proxy2D;
+         return ctx->Texture.ProxyTex[TEXTURE_2D_INDEX];
       case GL_TEXTURE_3D:
          return texUnit->Current3D;
       case GL_PROXY_TEXTURE_3D:
-         return ctx->Texture.Proxy3D;
+         return ctx->Texture.ProxyTex[TEXTURE_3D_INDEX];
       case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
       case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
       case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
@@ -780,25 +780,25 @@ _mesa_select_tex_object(GLcontext *ctx, const struct gl_texture_unit *texUnit,
                 ? texUnit->CurrentCubeMap : NULL;
       case GL_PROXY_TEXTURE_CUBE_MAP_ARB:
          return ctx->Extensions.ARB_texture_cube_map
-                ? ctx->Texture.ProxyCubeMap : NULL;
+                ? ctx->Texture.ProxyTex[TEXTURE_CUBE_INDEX] : NULL;
       case GL_TEXTURE_RECTANGLE_NV:
          return ctx->Extensions.NV_texture_rectangle
                 ? texUnit->CurrentRect : NULL;
       case GL_PROXY_TEXTURE_RECTANGLE_NV:
          return ctx->Extensions.NV_texture_rectangle
-                ? ctx->Texture.ProxyRect : NULL;
+                ? ctx->Texture.ProxyTex[TEXTURE_RECT_INDEX] : NULL;
       case GL_TEXTURE_1D_ARRAY_EXT:
          return ctx->Extensions.MESA_texture_array
                 ? texUnit->Current1DArray : NULL;
       case GL_PROXY_TEXTURE_1D_ARRAY_EXT:
          return ctx->Extensions.MESA_texture_array
-                ? ctx->Texture.Proxy1DArray : NULL;
+                ? ctx->Texture.ProxyTex[TEXTURE_1D_ARRAY_INDEX] : NULL;
       case GL_TEXTURE_2D_ARRAY_EXT:
          return ctx->Extensions.MESA_texture_array
                 ? texUnit->Current2DArray : NULL;
       case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
          return ctx->Extensions.MESA_texture_array
-                ? ctx->Texture.Proxy2DArray : NULL;
+                ? ctx->Texture.ProxyTex[TEXTURE_2D_ARRAY_INDEX] : NULL;
       default:
          _mesa_problem(NULL, "bad target in _mesa_select_tex_object()");
          return NULL;
@@ -924,106 +924,106 @@ _mesa_get_proxy_tex_image(GLcontext *ctx, GLenum target, GLint level)
    case GL_PROXY_TEXTURE_1D:
       if (level >= ctx->Const.MaxTextureLevels)
          return NULL;
-      texImage = ctx->Texture.Proxy1D->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_1D_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.Proxy1D->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_1D_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.Proxy1D;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_1D_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_2D:
       if (level >= ctx->Const.MaxTextureLevels)
          return NULL;
-      texImage = ctx->Texture.Proxy2D->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_2D_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.Proxy2D->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_2D_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.Proxy2D;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_2D_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_3D:
       if (level >= ctx->Const.Max3DTextureLevels)
          return NULL;
-      texImage = ctx->Texture.Proxy3D->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_3D_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.Proxy3D->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_3D_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.Proxy3D;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_3D_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_CUBE_MAP:
       if (level >= ctx->Const.MaxCubeTextureLevels)
          return NULL;
-      texImage = ctx->Texture.ProxyCubeMap->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_CUBE_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.ProxyCubeMap->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_CUBE_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.ProxyCubeMap;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_CUBE_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_RECTANGLE_NV:
       if (level > 0)
          return NULL;
-      texImage = ctx->Texture.ProxyRect->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_RECT_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.ProxyRect->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_RECT_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.ProxyRect;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_RECT_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_1D_ARRAY_EXT:
       if (level >= ctx->Const.MaxTextureLevels)
          return NULL;
-      texImage = ctx->Texture.Proxy1DArray->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_1D_ARRAY_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.Proxy1DArray->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_1D_ARRAY_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.Proxy1DArray;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_1D_ARRAY_INDEX];
       }
       return texImage;
    case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
       if (level >= ctx->Const.MaxTextureLevels)
          return NULL;
-      texImage = ctx->Texture.Proxy2DArray->Image[0][level];
+      texImage = ctx->Texture.ProxyTex[TEXTURE_2D_ARRAY_INDEX]->Image[0][level];
       if (!texImage) {
          texImage = ctx->Driver.NewTextureImage(ctx);
          if (!texImage) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "proxy texture allocation");
             return NULL;
          }
-         ctx->Texture.Proxy2DArray->Image[0][level] = texImage;
+         ctx->Texture.ProxyTex[TEXTURE_2D_ARRAY_INDEX]->Image[0][level] = texImage;
          /* Set the 'back' pointer */
-         texImage->TexObject = ctx->Texture.Proxy2DArray;
+         texImage->TexObject = ctx->Texture.ProxyTex[TEXTURE_2D_ARRAY_INDEX];
       }
       return texImage;
    default:
@@ -1264,6 +1264,10 @@ _mesa_init_teximage_fields(GLcontext *ctx, GLenum target,
  * A hardware driver might override this function if, for example, the
  * max 3D texture size is 512x512x64 (i.e. not a cube).
  *
+ * Note that width, height, depth == 0 is not an error.  However, a
+ * texture with zero width/height/depth will be considered "incomplete"
+ * and texturing will effectively be disabled.
+ *
  * \param target  one of GL_PROXY_TEXTURE_1D, GL_PROXY_TEXTURE_2D,
  *                GL_PROXY_TEXTURE_3D, GL_PROXY_TEXTURE_RECTANGLE_NV,
  *                GL_PROXY_TEXTURE_CUBE_MAP_ARB.
@@ -1293,7 +1297,7 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width >0 && _mesa_bitcount(width - 2 * border) != 1) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or level */
          return GL_FALSE;
@@ -1303,10 +1307,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or height or level */
          return GL_FALSE;
@@ -1316,21 +1320,21 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.Max3DTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
           depth < 2 * border || depth > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(depth - 2 * border) != 1) ||
+           depth > 0 && _mesa_bitcount(depth - 2 * border) != 1) ||
           level >= ctx->Const.Max3DTextureLevels) {
          /* bad width or height or depth or level */
          return GL_FALSE;
       }
       return GL_TRUE;
    case GL_PROXY_TEXTURE_RECTANGLE_NV:
-      if (width < 1 || width > ctx->Const.MaxTextureRectSize ||
-          height < 1 || height > ctx->Const.MaxTextureRectSize ||
+      if (width < 0 || width > ctx->Const.MaxTextureRectSize ||
+          height < 0 || height > ctx->Const.MaxTextureRectSize ||
           level != 0) {
          /* bad width or height or level */
          return GL_FALSE;
@@ -1340,10 +1344,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxCubeTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
           level >= ctx->Const.MaxCubeTextureLevels) {
          /* bad width or height */
          return GL_FALSE;
@@ -1353,7 +1357,7 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or level */
          return GL_FALSE;
@@ -1367,10 +1371,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or height or level */
          return GL_FALSE;
@@ -1472,7 +1476,7 @@ texture_error_check( GLcontext *ctx, GLenum target,
       if (target == GL_PROXY_TEXTURE_1D || target == GL_TEXTURE_1D) {
          proxy_target = GL_PROXY_TEXTURE_1D;
          height = 1;
-         width = 1;
+         depth = 1;
       }
       else {
          _mesa_error( ctx, GL_INVALID_ENUM, "glTexImage1D(target)" );
@@ -1994,7 +1998,7 @@ copytexture_error_check( GLcontext *ctx, GLuint dimensions,
    }
 
    if (is_compressed_format(ctx, internalFormat)) {
-      if (target != GL_TEXTURE_2D) {
+      if (!target_can_be_compressed(ctx, target)) {
          _mesa_error(ctx, GL_INVALID_ENUM,
                      "glCopyTexImage%d(target)", dimensions);
          return GL_TRUE;
@@ -2177,7 +2181,7 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
    }
 
    if (teximage->IsCompressed) {
-      if (target != GL_TEXTURE_2D) {
+      if (!target_can_be_compressed(ctx, target)) {
          _mesa_error(ctx, GL_INVALID_ENUM,
                      "glCopyTexSubImage%d(target)", dimensions);
          return GL_TRUE;
@@ -2587,7 +2591,7 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat,
                               1, border)) {
          /* when error, clear all proxy texture image parameters */
          if (texImage)
-            clear_teximage_fields(ctx->Texture.Proxy2D->Image[0][level]);
+            clear_teximage_fields(ctx->Texture.ProxyTex[TEXTURE_2D_INDEX]->Image[0][level]);
       }
       else {
          /* no error, set the tex image parameters */
diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c
index df64002f99..3e017c1eda 100644
--- a/src/mesa/main/texobj.c
+++ b/src/mesa/main/texobj.c
@@ -5,9 +5,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -106,6 +106,7 @@ _mesa_initialize_texture_object( struct gl_texture_object *obj,
 
    _mesa_bzero(obj, sizeof(*obj));
    /* init the non-zero fields */
+   _glthread_INIT_MUTEX(obj->Mutex);
    obj->RefCount = 1;
    obj->Name = name;
    obj->Target = target;
@@ -153,6 +154,11 @@ _mesa_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj )
 
    (void) ctx;
 
+   /* Set Target to an invalid value.  With some assertions elsewhere
+    * we can try to detect possible use of deleted textures.
+    */
+   texObj->Target = 0x99;
+
    _mesa_free_colortable_data(&texObj->Palette);
 
    /* free the texture images */
@@ -164,6 +170,9 @@ _mesa_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj )
       }
    }
 
+   /* destroy the mutex -- it may have allocated memory (eg on bsd) */
+   _glthread_DESTROY_MUTEX(texObj->Mutex);
+
    /* free this object */
    _mesa_free(texObj);
 }
@@ -182,6 +191,7 @@ void
 _mesa_copy_texture_object( struct gl_texture_object *dest,
                            const struct gl_texture_object *src )
 {
+   dest->Target = src->Target;
    dest->Name = src->Name;
    dest->Priority = src->Priority;
    dest->BorderColor[0] = src->BorderColor[0];
@@ -214,6 +224,94 @@ _mesa_copy_texture_object( struct gl_texture_object *dest,
 
 
 /**
+ * Check if the given texture object is valid by examining its Target field.
+ * For debugging only.
+ */
+static GLboolean
+valid_texture_object(const struct gl_texture_object *tex)
+{
+   switch (tex->Target) {
+   case 0:
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_3D:
+   case GL_TEXTURE_CUBE_MAP_ARB:
+   case GL_TEXTURE_RECTANGLE_NV:
+   case GL_TEXTURE_1D_ARRAY_EXT:
+   case GL_TEXTURE_2D_ARRAY_EXT:
+      return GL_TRUE;
+   case 0x99:
+      _mesa_problem(NULL, "invalid reference to a deleted texture object");
+      return GL_FALSE;
+   default:
+      _mesa_problem(NULL, "invalid texture object Target value");
+      return GL_FALSE;
+   }
+}
+
+
+/**
+ * Reference (or unreference) a texture object.
+ * If '*ptr', decrement *ptr's refcount (and delete if it becomes zero).
+ * If 'tex' is non-null, increment its refcount.
+ */
+void
+_mesa_reference_texobj(struct gl_texture_object **ptr,
+                       struct gl_texture_object *tex)
+{
+   assert(ptr);
+   if (*ptr == tex) {
+      /* no change */
+      return;
+   }
+
+   if (*ptr) {
+      /* Unreference the old texture */
+      GLboolean deleteFlag = GL_FALSE;
+      struct gl_texture_object *oldTex = *ptr;
+
+      assert(valid_texture_object(oldTex));
+
+      _glthread_LOCK_MUTEX(oldTex->Mutex);
+      ASSERT(oldTex->RefCount > 0);
+      oldTex->RefCount--;
+
+      deleteFlag = (oldTex->RefCount == 0);
+      _glthread_UNLOCK_MUTEX(oldTex->Mutex);
+
+      if (deleteFlag) {
+         GET_CURRENT_CONTEXT(ctx);
+         if (ctx)
+            ctx->Driver.DeleteTexture(ctx, oldTex);
+         else
+            _mesa_problem(NULL, "Unable to delete texture, no context");
+      }
+
+      *ptr = NULL;
+   }
+   assert(!*ptr);
+
+   if (tex) {
+      /* reference new texture */
+      assert(valid_texture_object(tex));
+      _glthread_LOCK_MUTEX(tex->Mutex);
+      if (tex->RefCount == 0) {
+         /* this texture's being deleted (look just above) */
+         /* Not sure this can every really happen.  Warn if it does. */
+         _mesa_problem(NULL, "referencing deleted texture object");
+         *ptr = NULL;
+      }
+      else {
+         tex->RefCount++;
+         *ptr = tex;
+      }
+      _glthread_UNLOCK_MUTEX(tex->Mutex);
+   }
+}
+
+
+
+/**
  * Report why a texture object is incomplete.  
  *
  * \param t texture object.
@@ -609,8 +707,7 @@ unbind_texobj_from_fbo(GLcontext *ctx, struct gl_texture_object *texObj)
 
 /**
  * Check if the given texture object is bound to any texture image units and
- * unbind it if so.
- * XXX all RefCount accesses should be protected by a mutex.
+ * unbind it if so (revert to default textures).
  */
 static void
 unbind_texobj_from_texunits(GLcontext *ctx, struct gl_texture_object *texObj)
@@ -619,42 +716,26 @@ unbind_texobj_from_texunits(GLcontext *ctx, struct gl_texture_object *texObj)
 
    for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++) {
       struct gl_texture_unit *unit = &ctx->Texture.Unit[u];
-      struct gl_texture_object **curr = NULL;
-
       if (texObj == unit->Current1D) {
-         curr = &unit->Current1D;
-         unit->Current1D = ctx->Shared->Default1D;
+         _mesa_reference_texobj(&unit->Current1D, ctx->Shared->Default1D);
       }
       else if (texObj == unit->Current2D) {
-         curr = &unit->Current2D;
-         unit->Current2D = ctx->Shared->Default2D;
+         _mesa_reference_texobj(&unit->Current2D, ctx->Shared->Default2D);
       }
       else if (texObj == unit->Current3D) {
-         curr = &unit->Current3D;
-         unit->Current3D = ctx->Shared->Default3D;
+         _mesa_reference_texobj(&unit->Current3D, ctx->Shared->Default3D);
       }
       else if (texObj == unit->CurrentCubeMap) {
-         curr = &unit->CurrentCubeMap;
-         unit->CurrentCubeMap = ctx->Shared->DefaultCubeMap;
+         _mesa_reference_texobj(&unit->CurrentCubeMap, ctx->Shared->DefaultCubeMap);
       }
       else if (texObj == unit->CurrentRect) {
-         curr = &unit->CurrentRect;
-         unit->CurrentRect = ctx->Shared->DefaultRect;
+         _mesa_reference_texobj(&unit->CurrentRect, ctx->Shared->DefaultRect);
       }
       else if (texObj == unit->Current1DArray) {
-         curr = &unit->Current1DArray;
-         unit->CurrentRect = ctx->Shared->Default1DArray;
+         _mesa_reference_texobj(&unit->Current1DArray, ctx->Shared->Default1DArray);
       }
       else if (texObj == unit->Current2DArray) {
-         curr = &unit->Current1DArray;
-         unit->CurrentRect = ctx->Shared->Default2DArray;
-      }
-
-      if (curr) {
-         (*curr)->RefCount++;
-         texObj->RefCount--;
-         if (texObj == unit->_Current)
-            unit->_Current = *curr;
+         _mesa_reference_texobj(&unit->Current2DArray, ctx->Shared->Default2DArray);
       }
    }
 }
@@ -690,8 +771,6 @@ _mesa_DeleteTextures( GLsizei n, const GLuint *textures)
             = _mesa_lookup_texture(ctx, textures[i]);
 
          if (delObj) {
-	    GLboolean deleted;
-
 	    _mesa_lock_texture(ctx, delObj);
 
             /* Check if texture is bound to any framebuffer objects.
@@ -701,10 +780,12 @@ _mesa_DeleteTextures( GLsizei n, const GLuint *textures)
             unbind_texobj_from_fbo(ctx, delObj);
 
             /* Check if this texture is currently bound to any texture units.
-             * If so, unbind it and decrement the reference count.
+             * If so, unbind it.
              */
             unbind_texobj_from_texunits(ctx, delObj);
 
+	    _mesa_unlock_texture(ctx, delObj);
+
             ctx->NewState |= _NEW_TEXTURE;
 
             /* The texture _name_ is now free for re-use.
@@ -714,23 +795,10 @@ _mesa_DeleteTextures( GLsizei n, const GLuint *textures)
             _mesa_HashRemove(ctx->Shared->TexObjects, delObj->Name);
             _glthread_UNLOCK_MUTEX(ctx->Shared->Mutex);
 
-            /* The actual texture object will not be freed until it's no
-             * longer bound in any context.
-             * XXX all RefCount accesses should be protected by a mutex.
+            /* Unreference the texobj.  If refcount hits zero, the texture
+             * will be deleted.
              */
-            delObj->RefCount--;
-	    deleted = (delObj->RefCount == 0);
-	    _mesa_unlock_texture(ctx, delObj);
-
-	    /* We know that refcount went to zero above, so this is
-	     * the only pointer left to delObj, so we don't have to
-	     * worry about locking any more:
-	     */
-            if (deleted) {
-               ASSERT(delObj->Name != 0); /* Never delete default tex objs */
-               ASSERT(ctx->Driver.DeleteTexture);
-               (*ctx->Driver.DeleteTexture)(ctx, delObj);
-            }
+            _mesa_reference_texobj(&delObj, NULL);
          }
       }
    }
@@ -758,7 +826,6 @@ _mesa_BindTexture( GLenum target, GLuint texName )
    GET_CURRENT_CONTEXT(ctx);
    const GLuint unit = ctx->Texture.CurrentUnit;
    struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *oldTexObj;
    struct gl_texture_object *newTexObj = NULL;
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -767,62 +834,6 @@ _mesa_BindTexture( GLenum target, GLuint texName )
                   _mesa_lookup_enum_by_nr(target), (GLint) texName);
 
    /*
-    * Get pointer to currently bound texture object (oldTexObj)
-    */
-   switch (target) {
-      case GL_TEXTURE_1D:
-         oldTexObj = texUnit->Current1D;
-         break;
-      case GL_TEXTURE_2D:
-         oldTexObj = texUnit->Current2D;
-         break;
-      case GL_TEXTURE_3D:
-         oldTexObj = texUnit->Current3D;
-         break;
-      case GL_TEXTURE_CUBE_MAP_ARB:
-         if (!ctx->Extensions.ARB_texture_cube_map) {
-            _mesa_error( ctx, GL_INVALID_ENUM, "glBindTexture(target)" );
-            return;
-         }
-         oldTexObj = texUnit->CurrentCubeMap;
-         break;
-      case GL_TEXTURE_RECTANGLE_NV:
-         if (!ctx->Extensions.NV_texture_rectangle) {
-            _mesa_error( ctx, GL_INVALID_ENUM, "glBindTexture(target)" );
-            return;
-         }
-         oldTexObj = texUnit->CurrentRect;
-         break;
-      case GL_TEXTURE_1D_ARRAY_EXT:
-         if (!ctx->Extensions.MESA_texture_array) {
-            _mesa_error( ctx, GL_INVALID_ENUM, "glBindTexture(target)" );
-            return;
-         }
-         oldTexObj = texUnit->Current1DArray;
-         break;
-      case GL_TEXTURE_2D_ARRAY_EXT:
-         if (!ctx->Extensions.MESA_texture_array) {
-            _mesa_error( ctx, GL_INVALID_ENUM, "glBindTexture(target)" );
-            return;
-         }
-         oldTexObj = texUnit->Current2DArray;
-         break;
-      default:
-         _mesa_error( ctx, GL_INVALID_ENUM, "glBindTexture(target)" );
-         return;
-   }
-
-   if (oldTexObj->Name == texName) {
-      /* XXX this might be wrong.  If the texobj is in use by another
-       * context and a texobj parameter was changed, this might be our
-       * only chance to update this context's hardware state.
-       * Note that some applications re-bind the same texture a lot so we
-       * want to handle that case quickly.
-       */
-      return;   /* rebinding the same texture- no change */
-   }
-
-   /*
     * Get pointer to new texture object (newTexObj)
     */
    if (texName == 0) {
@@ -896,28 +907,30 @@ _mesa_BindTexture( GLenum target, GLuint texName )
       newTexObj->Target = target;
    }
 
-   /* XXX all RefCount accesses should be protected by a mutex. */
-   newTexObj->RefCount++;
+   assert(valid_texture_object(newTexObj));
 
-   /* do the actual binding, but first flush outstanding vertices:
-    */
+   /* flush before changing binding */
    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
 
+   /* Do the actual binding.  The refcount on the previously bound
+    * texture object will be decremented.  It'll be deleted if the
+    * count hits zero.
+    */
    switch (target) {
       case GL_TEXTURE_1D:
-         texUnit->Current1D = newTexObj;
+         _mesa_reference_texobj(&texUnit->Current1D, newTexObj);
          break;
       case GL_TEXTURE_2D:
-         texUnit->Current2D = newTexObj;
+         _mesa_reference_texobj(&texUnit->Current2D, newTexObj);
          break;
       case GL_TEXTURE_3D:
-         texUnit->Current3D = newTexObj;
+         _mesa_reference_texobj(&texUnit->Current3D, newTexObj);
          break;
       case GL_TEXTURE_CUBE_MAP_ARB:
-         texUnit->CurrentCubeMap = newTexObj;
+         _mesa_reference_texobj(&texUnit->CurrentCubeMap, newTexObj);
          break;
       case GL_TEXTURE_RECTANGLE_NV:
-         texUnit->CurrentRect = newTexObj;
+         _mesa_reference_texobj(&texUnit->CurrentRect, newTexObj);
          break;
       case GL_TEXTURE_1D_ARRAY_EXT:
          texUnit->Current1DArray = newTexObj;
@@ -933,18 +946,6 @@ _mesa_BindTexture( GLenum target, GLuint texName )
    /* Pass BindTexture call to device driver */
    if (ctx->Driver.BindTexture)
       (*ctx->Driver.BindTexture)( ctx, target, newTexObj );
-
-   /* Decrement the reference count on the old texture and check if it's
-    * time to delete it.
-    */
-   /* XXX all RefCount accesses should be protected by a mutex. */
-   oldTexObj->RefCount--;
-   ASSERT(oldTexObj->RefCount >= 0);
-   if (oldTexObj->RefCount == 0) {
-      ASSERT(oldTexObj->Name != 0);
-      ASSERT(ctx->Driver.DeleteTexture);
-      (*ctx->Driver.DeleteTexture)( ctx, oldTexObj );
-   }
 }
 
 
diff --git a/src/mesa/main/texobj.h b/src/mesa/main/texobj.h
index 2a2bde3601..d5374c5d6c 100644
--- a/src/mesa/main/texobj.h
+++ b/src/mesa/main/texobj.h
@@ -58,6 +58,10 @@ _mesa_copy_texture_object( struct gl_texture_object *dest,
                            const struct gl_texture_object *src );
 
 extern void
+_mesa_reference_texobj(struct gl_texture_object **ptr,
+                       struct gl_texture_object *tex);
+
+extern void
 _mesa_test_texobj_completeness( const GLcontext *ctx,
                                 struct gl_texture_object *obj );
 
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index c9f8a0656e..020e3eef4c 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -63,31 +63,6 @@ static const struct gl_tex_env_combine_state default_combine_state = {
 };
 
 
-/**
- * Copy a texture binding.  Helper used by _mesa_copy_texture_state().
- */
-static void
-copy_texture_binding(const GLcontext *ctx,
-                     struct gl_texture_object **dst,
-                     struct gl_texture_object *src)
-{
-   /* only copy if names differ (per OpenGL SI) */
-   if ((*dst)->Name != src->Name) {
-      /* unbind/delete dest binding which we're changing */
-      (*dst)->RefCount--;
-      if ((*dst)->RefCount == 0) {
-         /* time to delete this texture object */
-         ASSERT((*dst)->Name != 0);
-         ASSERT(ctx->Driver.DeleteTexture);
-         /* XXX cast-away const, unfortunately */
-         (*ctx->Driver.DeleteTexture)((GLcontext *) ctx, *dst);
-      }
-      /* make new binding, incrementing ref count */
-      *dst = src;
-      src->RefCount++;
-   }
-}
-
 
 /**
  * Used by glXCopyContext to copy texture state from one context to another.
@@ -144,20 +119,20 @@ _mesa_copy_texture_state( const GLcontext *src, GLcontext *dst )
       /* copy texture object bindings, not contents of texture objects */
       _mesa_lock_context_textures(dst);
 
-      copy_texture_binding(src, &dst->Texture.Unit[i].Current1D,
-                           src->Texture.Unit[i].Current1D);
-      copy_texture_binding(src, &dst->Texture.Unit[i].Current2D,
-                           src->Texture.Unit[i].Current2D);
-      copy_texture_binding(src, &dst->Texture.Unit[i].Current3D,
-                           src->Texture.Unit[i].Current3D);
-      copy_texture_binding(src, &dst->Texture.Unit[i].CurrentCubeMap,
-                           src->Texture.Unit[i].CurrentCubeMap);
-      copy_texture_binding(src, &dst->Texture.Unit[i].CurrentRect,
-                           src->Texture.Unit[i].CurrentRect);
-      copy_texture_binding(src, &dst->Texture.Unit[i].Current1DArray,
-                           src->Texture.Unit[i].Current1DArray);
-      copy_texture_binding(src, &dst->Texture.Unit[i].Current2DArray,
-                           src->Texture.Unit[i].Current2DArray);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current1D,
+                             src->Texture.Unit[i].Current1D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current2D,
+                             src->Texture.Unit[i].Current2D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current3D,
+                             src->Texture.Unit[i].Current3D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentCubeMap,
+                             src->Texture.Unit[i].CurrentCubeMap);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentRect,
+                             src->Texture.Unit[i].CurrentRect);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current1DArray,
+                             src->Texture.Unit[i].Current1DArray);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current2DArray,
+                             src->Texture.Unit[i].Current2DArray);
 
       _mesa_unlock_context_textures(dst);
    }
@@ -3085,52 +3060,32 @@ _mesa_update_texture( GLcontext *ctx, GLuint new_state )
 static GLboolean
 alloc_proxy_textures( GLcontext *ctx )
 {
-   ctx->Texture.Proxy1D = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_1D);
-   if (!ctx->Texture.Proxy1D)
-      goto cleanup;
-
-   ctx->Texture.Proxy2D = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_2D);
-   if (!ctx->Texture.Proxy2D)
-      goto cleanup;
-
-   ctx->Texture.Proxy3D = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_3D);
-   if (!ctx->Texture.Proxy3D)
-      goto cleanup;
-
-   ctx->Texture.ProxyCubeMap = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_CUBE_MAP_ARB);
-   if (!ctx->Texture.ProxyCubeMap)
-      goto cleanup;
-
-   ctx->Texture.ProxyRect = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_RECTANGLE_NV);
-   if (!ctx->Texture.ProxyRect)
-      goto cleanup;
-
-   ctx->Texture.Proxy1DArray = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_1D_ARRAY_EXT);
-   if (!ctx->Texture.Proxy1DArray)
-      goto cleanup;
-
-   ctx->Texture.Proxy2DArray = (*ctx->Driver.NewTextureObject)(ctx, 0, GL_TEXTURE_2D_ARRAY_EXT);
-   if (!ctx->Texture.Proxy2DArray)
-      goto cleanup;
+   static const GLenum targets[] = {
+      GL_TEXTURE_1D,
+      GL_TEXTURE_2D,
+      GL_TEXTURE_3D,
+      GL_TEXTURE_CUBE_MAP_ARB,
+      GL_TEXTURE_RECTANGLE_NV,
+      GL_TEXTURE_1D_ARRAY_EXT,
+      GL_TEXTURE_2D_ARRAY_EXT
+   };
+   GLint tgt;
+
+   ASSERT(Elements(targets) == NUM_TEXTURE_TARGETS);
+
+   for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
+      if (!(ctx->Texture.ProxyTex[tgt]
+            = ctx->Driver.NewTextureObject(ctx, 0, targets[tgt]))) {
+         /* out of memory, free what we did allocate */
+         while (--tgt >= 0) {
+            ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]);
+         }
+         return GL_FALSE;
+      }
+   }
 
+   assert(ctx->Texture.ProxyTex[0]->RefCount == 1); /* sanity check */
    return GL_TRUE;
-
- cleanup:
-   if (ctx->Texture.Proxy1D)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.Proxy1D);
-   if (ctx->Texture.Proxy2D)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.Proxy2D);
-   if (ctx->Texture.Proxy3D)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.Proxy3D);
-   if (ctx->Texture.ProxyCubeMap)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.ProxyCubeMap);
-   if (ctx->Texture.ProxyRect)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.ProxyRect);
-   if (ctx->Texture.Proxy1DArray)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.Proxy1DArray);
-   if (ctx->Texture.Proxy2DArray)
-      (ctx->Driver.DeleteTexture)(ctx, ctx->Texture.Proxy2DArray);
-   return GL_FALSE;
 }
 
 
@@ -3172,13 +3127,14 @@ init_texture_unit( GLcontext *ctx, GLuint unit )
    ASSIGN_4V( texUnit->EyePlaneR, 0.0, 0.0, 0.0, 0.0 );
    ASSIGN_4V( texUnit->EyePlaneQ, 0.0, 0.0, 0.0, 0.0 );
 
-   texUnit->Current1D = ctx->Shared->Default1D;
-   texUnit->Current2D = ctx->Shared->Default2D;
-   texUnit->Current3D = ctx->Shared->Default3D;
-   texUnit->CurrentCubeMap = ctx->Shared->DefaultCubeMap;
-   texUnit->CurrentRect = ctx->Shared->DefaultRect;
-   texUnit->Current1DArray = ctx->Shared->Default1DArray;
-   texUnit->Current2DArray = ctx->Shared->Default2DArray;
+   /* initialize current texture object ptrs to the shared default objects */
+   _mesa_reference_texobj(&texUnit->Current1D, ctx->Shared->Default1D);
+   _mesa_reference_texobj(&texUnit->Current2D, ctx->Shared->Default2D);
+   _mesa_reference_texobj(&texUnit->Current3D, ctx->Shared->Default3D);
+   _mesa_reference_texobj(&texUnit->CurrentCubeMap, ctx->Shared->DefaultCubeMap);
+   _mesa_reference_texobj(&texUnit->CurrentRect, ctx->Shared->DefaultRect);
+   _mesa_reference_texobj(&texUnit->Current1DArray, ctx->Shared->Default1DArray);
+   _mesa_reference_texobj(&texUnit->Current2DArray, ctx->Shared->Default2DArray);
 }
 
 
@@ -3193,23 +3149,20 @@ _mesa_init_texture(GLcontext *ctx)
    assert(MAX_TEXTURE_LEVELS >= MAX_3D_TEXTURE_LEVELS);
    assert(MAX_TEXTURE_LEVELS >= MAX_CUBE_TEXTURE_LEVELS);
 
-   /* Effectively bind the default textures to all texture units */
-   ctx->Shared->Default1D->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->Default2D->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->Default3D->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->DefaultCubeMap->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->DefaultRect->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->Default1DArray->RefCount += MAX_TEXTURE_UNITS;
-   ctx->Shared->Default2DArray->RefCount += MAX_TEXTURE_UNITS;
-
    /* Texture group */
    ctx->Texture.CurrentUnit = 0;      /* multitexture */
    ctx->Texture._EnabledUnits = 0;
-   for (i=0; i<MAX_TEXTURE_UNITS; i++)
-      init_texture_unit( ctx, i );
    ctx->Texture.SharedPalette = GL_FALSE;
    _mesa_init_colortable(&ctx->Texture.Palette);
 
+   for (i = 0; i < MAX_TEXTURE_UNITS; i++)
+      init_texture_unit( ctx, i );
+
+   /* After we're done initializing the context's texture state the default
+    * texture objects' refcounts should be at least MAX_TEXTURE_UNITS + 1.
+    */
+   assert(ctx->Shared->Default1D->RefCount >= MAX_TEXTURE_UNITS + 1);
+
    _mesa_TexEnvProgramCacheInit( ctx );
 
    /* Allocate proxy textures */
@@ -3226,19 +3179,26 @@ _mesa_init_texture(GLcontext *ctx)
 void
 _mesa_free_texture_data(GLcontext *ctx)
 {
-   GLuint i;
+   GLuint u, tgt;
+
+   /* unreference current textures */
+   for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++) {
+      struct gl_texture_unit *unit = ctx->Texture.Unit + u;
+      _mesa_reference_texobj(&unit->Current1D, NULL);
+      _mesa_reference_texobj(&unit->Current2D, NULL);
+      _mesa_reference_texobj(&unit->Current3D, NULL);
+      _mesa_reference_texobj(&unit->CurrentCubeMap, NULL);
+      _mesa_reference_texobj(&unit->CurrentRect, NULL);
+      _mesa_reference_texobj(&unit->Current1DArray, NULL);
+      _mesa_reference_texobj(&unit->Current2DArray, NULL);
+   }
 
    /* Free proxy texture objects */
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.Proxy1D );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.Proxy2D );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.Proxy3D );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.ProxyCubeMap );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.ProxyRect );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.Proxy1DArray );
-   (ctx->Driver.DeleteTexture)(ctx,  ctx->Texture.Proxy2DArray );
-
-   for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++)
-      _mesa_free_colortable_data( &ctx->Texture.Unit[i].ColorTable );
+   for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++)
+      ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]);
+
+   for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++)
+      _mesa_free_colortable_data( &ctx->Texture.Unit[u].ColorTable );
 
    _mesa_TexEnvProgramCacheDestroy( ctx );
 }
diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c
index 9b8a06df14..3b5151ed17 100644
--- a/src/mesa/main/texstore.c
+++ b/src/mesa/main/texstore.c
@@ -1077,7 +1077,7 @@ _mesa_texstore_rgba(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_z32(TEXSTORE_PARAMS)
 {
-   const GLfloat depthScale = (GLfloat) 0xffffffff;
+   const GLuint depthScale = 0xffffffff;
    (void) dims;
    ASSERT(dstFormat == &_mesa_texformat_z32);
    ASSERT(dstFormat->TexelBytes == sizeof(GLuint));
@@ -1124,7 +1124,7 @@ _mesa_texstore_z32(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_z16(TEXSTORE_PARAMS)
 {
-   const GLfloat depthScale = 65535.0f;
+   const GLuint depthScale = 0xffff;
    (void) dims;
    ASSERT(dstFormat == &_mesa_texformat_z16);
    ASSERT(dstFormat->TexelBytes == sizeof(GLushort));
@@ -2319,6 +2319,8 @@ _mesa_texstore_ycbcr(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
 {
+   const GLuint depthScale = 0xffffff;
+
    ASSERT(dstFormat == &_mesa_texformat_z24_s8);
    ASSERT(srcFormat == GL_DEPTH_STENCIL_EXT);
    ASSERT(srcType == GL_UNSIGNED_INT_24_8_EXT);
@@ -2357,7 +2359,7 @@ _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
             _mesa_unpack_depth_span(ctx, srcWidth,
                                     GL_UNSIGNED_INT, /* dst type */
                                     dstRow, /* dst addr */
-                                    (GLfloat) 0xffffff, /* depthScale */
+                                    depthScale,
                                     srcType, src, srcPacking);
             /* get the 8-bit stencil values */
             _mesa_unpack_stencil_span(ctx, srcWidth,
diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index bf1ad0165e..fe4a7c684f 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -32,7 +32,7 @@
 #include "mtypes.h"
 #include "varray.h"
 #include "arrayobj.h"
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 
 /**
diff --git a/src/mesa/main/vtxfmt_tmp.h b/src/mesa/main/vtxfmt_tmp.h
index 783b06558d..6f5d01e40f 100644
--- a/src/mesa/main/vtxfmt_tmp.h
+++ b/src/mesa/main/vtxfmt_tmp.h
@@ -29,8 +29,8 @@
 #define PRE_LOOPBACK( FUNC )
 #endif
 
-#include "dispatch.h"
-#include "glapioffsets.h"
+#include "glapi/dispatch.h"
+#include "glapi/glapioffsets.h"
 
 static void GLAPIENTRY TAG(ArrayElement)( GLint i )
 {
diff --git a/src/mesa/math/m_eval.c b/src/mesa/math/m_eval.c
index 42ffd4133d..d324673c5d 100644
--- a/src/mesa/math/m_eval.c
+++ b/src/mesa/math/m_eval.c
@@ -37,8 +37,8 @@
  */
 
 
-#include "glheader.h"
-#include "config.h"
+#include "main/glheader.h"
+#include "main/config.h"
 #include "m_eval.h"
 
 static GLfloat inv_tab[MAX_EVAL_ORDER];
diff --git a/src/mesa/math/m_eval.h b/src/mesa/math/m_eval.h
index a23cbd402e..d73ecaafb2 100644
--- a/src/mesa/math/m_eval.h
+++ b/src/mesa/math/m_eval.h
@@ -26,7 +26,7 @@
 #ifndef _M_EVAL_H
 #define _M_EVAL_H
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 void _math_init_eval( void );
 
diff --git a/src/mesa/math/m_translate.h b/src/mesa/math/m_translate.h
index 0bcf96005c..c677682d50 100644
--- a/src/mesa/math/m_translate.h
+++ b/src/mesa/math/m_translate.h
@@ -26,8 +26,8 @@
 #ifndef _M_TRANSLATE_H_
 #define _M_TRANSLATE_H_
 
-#include "config.h"
-#include "mtypes.h"		/* hack for GLchan */
+#include "main/config.h"
+#include "main/mtypes.h"		/* hack for GLchan */
 
 
 /**
diff --git a/src/mesa/math/m_xform.c b/src/mesa/math/m_xform.c
index fa3f57a8e5..901ae5b416 100644
--- a/src/mesa/math/m_xform.c
+++ b/src/mesa/math/m_xform.c
@@ -33,8 +33,8 @@
  * 3. Transformation of a point p by a matrix M is: p' = M * p
  */
 
-#include "glheader.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/macros.h"
 
 #include "m_eval.h"
 #include "m_matrix.h"
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index 5d8f763741..9a5290d920 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -30,10 +30,10 @@
  * \author Karl Rasche
  */
 
-#include "glheader.h"
-#include "imports.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "shader/grammar/grammar_mesa.h"
 #include "arbprogparse.h"
-#include "grammar_mesa.h"
 #include "program.h"
 #include "prog_parameter.h"
 #include "prog_statevars.h"
@@ -3573,7 +3573,7 @@ parse_instructions(GLcontext * ctx, const GLubyte * inst,
 
 /* XXX temporary */
 LONGSTRING static char core_grammar_text[] =
-#include "grammar_syn.h"
+#include "shader/grammar/grammar_syn.h"
 ;
 
 
diff --git a/src/mesa/shader/atifragshader.c b/src/mesa/shader/atifragshader.c
index 4727c1a436..854c911874 100644
--- a/src/mesa/shader/atifragshader.c
+++ b/src/mesa/shader/atifragshader.c
@@ -440,7 +440,7 @@ _mesa_PassTexCoordATI(GLuint dst, GLuint coord, GLenum swizzle)
       _mesa_error(ctx, GL_INVALID_OPERATION, "glPassTexCoordATI(coord)");
       return;
    }
-   if ((swizzle < GL_SWIZZLE_STR_ATI) && (swizzle > GL_SWIZZLE_STQ_DQ_ATI)) {
+   if (!(swizzle >= GL_SWIZZLE_STR_ATI) && (swizzle <= GL_SWIZZLE_STQ_DQ_ATI)) {
       _mesa_error(ctx, GL_INVALID_ENUM, "glPassTexCoordATI(swizzle)");
       return;
    }
@@ -513,7 +513,7 @@ _mesa_SampleMapATI(GLuint dst, GLuint interp, GLenum swizzle)
       _mesa_error(ctx, GL_INVALID_OPERATION, "glSampleMapATI(interp)");
       return;
    }
-   if ((swizzle < GL_SWIZZLE_STR_ATI) && (swizzle > GL_SWIZZLE_STQ_DQ_ATI)) {
+   if (!(swizzle >= GL_SWIZZLE_STR_ATI) && (swizzle <= GL_SWIZZLE_STQ_DQ_ATI)) {
       _mesa_error(ctx, GL_INVALID_ENUM, "glSampleMapATI(swizzle)");
       return;
    }
diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c
index 9faf9d8613..28d195d0ee 100644
--- a/src/mesa/shader/prog_execute.c
+++ b/src/mesa/shader/prog_execute.c
@@ -43,7 +43,7 @@
 #include "prog_instruction.h"
 #include "prog_parameter.h"
 #include "prog_print.h"
-#include "slang_library_noise.h"
+#include "shader/slang/slang_library_noise.h"
 
 
 /* See comments below for info about this */
diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c
index 4205919828..1f227390af 100644
--- a/src/mesa/shader/program.c
+++ b/src/mesa/shader/program.c
@@ -333,7 +333,7 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog)
 {
    struct gl_program *clone;
 
-   clone = _mesa_new_program(ctx, prog->Target, prog->Id);
+   clone = ctx->Driver.NewProgram(ctx, prog->Target, prog->Id);
    if (!clone)
       return NULL;
 
diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c
index 3a54e68d0d..a1e73ef125 100644
--- a/src/mesa/shader/shader_api.c
+++ b/src/mesa/shader/shader_api.c
@@ -43,10 +43,9 @@
 #include "prog_parameter.h"
 #include "prog_print.h"
 #include "prog_statevars.h"
-#include "shader_api.h"
-
-#include "slang_compile.h"
-#include "slang_link.h"
+#include "shader/shader_api.h"
+#include "shader/slang/slang_compile.h"
+#include "shader/slang/slang_link.h"
 
 
 
@@ -378,7 +377,7 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader)
    struct gl_shader_program *shProg
       = _mesa_lookup_shader_program(ctx, program);
    struct gl_shader *sh = _mesa_lookup_shader(ctx, shader);
-   const GLuint n = shProg->NumShaders;
+   GLuint n;
    GLuint i;
 
    if (!shProg || !sh) {
@@ -387,6 +386,8 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader)
       return;
    }
 
+   n = shProg->NumShaders;
+
    for (i = 0; i < n; i++) {
       if (shProg->Shaders[i] == sh) {
          /* already attached */
@@ -548,7 +549,7 @@ _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader)
 {
    struct gl_shader_program *shProg
       = _mesa_lookup_shader_program(ctx, program);
-   const GLuint n = shProg->NumShaders;
+   GLuint n;
    GLuint i, j;
 
    if (!shProg) {
@@ -557,6 +558,8 @@ _mesa_detach_shader(GLcontext *ctx, GLuint program, GLuint shader)
       return;
    }
 
+   n = shProg->NumShaders;
+
    for (i = 0; i < n; i++) {
       if (shProg->Shaders[i]->Name == shader) {
          /* found it */
@@ -780,7 +783,7 @@ _mesa_get_programiv(GLcontext *ctx, GLuint program,
       *params = shProg->Validated;
       break;
    case GL_INFO_LOG_LENGTH:
-      *params = shProg->InfoLog ? strlen(shProg->InfoLog) : 0;
+      *params = shProg->InfoLog ? strlen(shProg->InfoLog) + 1 : 0;
       break;
    case GL_ATTACHED_SHADERS:
       *params = shProg->NumShaders;
@@ -832,10 +835,10 @@ _mesa_get_shaderiv(GLcontext *ctx, GLuint name, GLenum pname, GLint *params)
       *params = shader->CompileStatus;
       break;
    case GL_INFO_LOG_LENGTH:
-      *params = shader->InfoLog ? strlen(shader->InfoLog) : 0;
+      *params = shader->InfoLog ? strlen(shader->InfoLog) + 1 : 0;
       break;
    case GL_SHADER_SOURCE_LENGTH:
-      *params = shader->Source ? strlen((char *) shader->Source) : 0;
+      *params = shader->Source ? strlen((char *) shader->Source) + 1 : 0;
       break;
    default:
       _mesa_error(ctx, GL_INVALID_ENUM, "glGetShaderiv(pname)");
diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c
index 6ee0fd33b6..1081d8ff8d 100644
--- a/src/mesa/shader/slang/slang_builtin.c
+++ b/src/mesa/shader/slang/slang_builtin.c
@@ -28,14 +28,14 @@
  * \author Brian Paul
  */
 
-#include "imports.h"
-#include "slang_builtin.h"
-#include "slang_ir.h"
-#include "mtypes.h"
-#include "program.h"
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "prog_statevars.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_statevars.h"
+#include "shader/slang/slang_ir.h"
+#include "shader/slang/slang_builtin.h"
 
 
 /**
diff --git a/src/mesa/shader/slang/slang_builtin.h b/src/mesa/shader/slang/slang_builtin.h
index ae20c844d5..58629f4f7f 100644
--- a/src/mesa/shader/slang/slang_builtin.h
+++ b/src/mesa/shader/slang/slang_builtin.h
@@ -26,7 +26,7 @@
 #ifndef SLANG_BUILTIN_H
 #define SLANG_BUILTIN_H
 
-#include "prog_parameter.h"
+#include "shader/prog_parameter.h"
 #include "slang_utility.h"
 #include "slang_ir.h"
 
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index f3a6d04428..d368009ca5 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.1
  *
  * Copyright (C) 2005-2007  Brian Paul   All Rights Reserved.
  *
@@ -37,13 +37,13 @@
 
 
 
-#include "imports.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "program.h"
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "prog_statevars.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_statevars.h"
 #include "slang_typeinfo.h"
 #include "slang_codegen.h"
 #include "slang_compile.h"
@@ -491,6 +491,9 @@ new_node0(slang_ir_opcode op)
 }
 
 
+/**
+ * Create sequence of two nodes.
+ */
 static slang_ir_node *
 new_seq(slang_ir_node *left, slang_ir_node *right)
 {
@@ -531,12 +534,12 @@ new_not(slang_ir_node *n)
 
 
 /**
- * Inlined subroutine.
+ * Non-inlined function call.
  */
 static slang_ir_node *
-new_inlined_function_call(slang_ir_node *code, slang_label *name)
+new_function_call(slang_ir_node *code, slang_label *name)
 {
-   slang_ir_node *n = new_node1(IR_FUNC, code);
+   slang_ir_node *n = new_node1(IR_CALL, code);
    assert(name);
    if (n)
       n->Label = name;
@@ -1202,17 +1205,29 @@ _slang_gen_function_call(slang_assemble_ctx *A, slang_function *fun,
       /* non-assembly function */
       inlined = slang_inline_function_call(A, fun, oper, dest);
       if (inlined && _slang_find_node_type(inlined, SLANG_OPER_RETURN)) {
-         /* This inlined function has one or more 'return' statements.
+         slang_operation *callOper;
+         /* The function we're calling has one or more 'return' statements.
           * So, we can't truly inline this function because we need to
           * implement 'return' with RET (and CAL).
+          * Nevertheless, we performed "inlining" to make a new instance
+          * of the function body to deal with static register allocation.
+          *
           * XXX check if there's one 'return' and if it's the very last
           * statement in the function - we can optimize that case.
           */
          assert(inlined->type == SLANG_OPER_BLOCK_NEW_SCOPE ||
                 inlined->type == SLANG_OPER_SEQUENCE);
-         inlined->type = SLANG_OPER_INLINED_CALL;
-         inlined->fun = fun;
-         inlined->label = _slang_label_new_unique((char*) fun->header.a_name);
+         if (_slang_function_has_return_value(fun) && !dest) {
+            assert(inlined->children[0].type == SLANG_OPER_VARIABLE_DECL);
+            assert(inlined->children[2].type == SLANG_OPER_IDENTIFIER);
+            callOper = &inlined->children[1];
+         }
+         else {
+            callOper = inlined;
+         }
+         callOper->type = SLANG_OPER_NON_INLINED_CALL;
+         callOper->fun = fun;
+         callOper->label = _slang_label_new_unique((char*) fun->header.a_name);
       }
    }
 
@@ -1949,8 +1964,7 @@ static slang_ir_node *
 _slang_gen_return(slang_assemble_ctx * A, slang_operation *oper)
 {
    const GLboolean haveReturnValue
-      = (oper->num_children == 1 &&
-         oper->children[0].type != SLANG_OPER_VOID);
+      = (oper->num_children == 1 && oper->children[0].type != SLANG_OPER_VOID);
 
    /* error checking */
    assert(A->CurFunction);
@@ -1960,7 +1974,7 @@ _slang_gen_return(slang_assemble_ctx * A, slang_operation *oper)
       return NULL;
    }
    else if (!haveReturnValue &&
-       A->CurFunction->header.type.specifier.type != SLANG_SPEC_VOID) {
+            A->CurFunction->header.type.specifier.type != SLANG_SPEC_VOID) {
       slang_info_log_error(A->log, "return statement requires an expression");
       return NULL;
    }
@@ -2045,7 +2059,8 @@ _slang_gen_declaration(slang_assemble_ctx *A, slang_operation *oper)
       }
       /* XXX make copy of this initializer? */
       rhs = _slang_gen_operation(A, &oper->children[0]);
-      assert(rhs);
+      if (!rhs)
+         return NULL;  /* must have found an error */
       init = new_node2(IR_MOVE, var, rhs);
       /*assert(rhs->Opcode != IR_SEQ);*/
       n = new_seq(varDecl, init);
@@ -2236,7 +2251,9 @@ _slang_gen_assignment(slang_assemble_ctx * A, slang_operation *oper)
       }
       if (var->type.qualifier == SLANG_QUAL_CONST ||
           var->type.qualifier == SLANG_QUAL_ATTRIBUTE ||
-          var->type.qualifier == SLANG_QUAL_UNIFORM) {
+          var->type.qualifier == SLANG_QUAL_UNIFORM ||
+          (var->type.qualifier == SLANG_QUAL_VARYING &&
+           A->program->Target == GL_FRAGMENT_PROGRAM_ARB)) {
          slang_info_log_error(A->log,
                               "illegal assignment to read-only variable '%s'",
                               (char *) oper->children[0].a_id);
@@ -2264,10 +2281,11 @@ _slang_gen_assignment(slang_assemble_ctx * A, slang_operation *oper)
       lhs = _slang_gen_operation(A, &oper->children[0]);
 
       if (lhs) {
-         if (lhs->Store->File != PROGRAM_OUTPUT &&
-             lhs->Store->File != PROGRAM_TEMPORARY &&
-             lhs->Store->File != PROGRAM_VARYING &&
-             lhs->Store->File != PROGRAM_UNDEFINED) {
+         if (!(lhs->Store->File == PROGRAM_OUTPUT ||
+               lhs->Store->File == PROGRAM_TEMPORARY ||
+               (lhs->Store->File == PROGRAM_VARYING &&
+                A->program->Target == GL_VERTEX_PROGRAM_ARB) ||
+               lhs->Store->File == PROGRAM_UNDEFINED)) {
             slang_info_log_error(A->log,
                                  "illegal assignment to read-only l-value");
             return NULL;
@@ -2328,7 +2346,8 @@ _slang_gen_field(slang_assemble_ctx * A, slang_operation *oper)
          n = _slang_gen_swizzle(n, swizzle);
       return n;
    }
-   else if (ti.spec.type == SLANG_SPEC_FLOAT) {
+   else if (   ti.spec.type == SLANG_SPEC_FLOAT
+            || ti.spec.type == SLANG_SPEC_INT) {
       const GLuint rows = 1;
       slang_swizzle swz;
       slang_ir_node *n;
@@ -2570,7 +2589,7 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper)
                _slang_free_ir_tree(tree);
                return NULL; /* error must have occured */
             }
-            tree = tree ? new_seq(tree, n) : n;
+            tree = new_seq(tree, n);
          }
 
 #if 00
@@ -2798,17 +2817,17 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper)
 	 return n;
       }
 
-   case SLANG_OPER_INLINED_CALL:
+   case SLANG_OPER_NON_INLINED_CALL:
    case SLANG_OPER_SEQUENCE:
       {
          slang_ir_node *tree = NULL;
          GLuint i;
          for (i = 0; i < oper->num_children; i++) {
             slang_ir_node *n = _slang_gen_operation(A, &oper->children[i]);
-            tree = tree ? new_seq(tree, n) : n;
+            tree = new_seq(tree, n);
          }
-         if (oper->type == SLANG_OPER_INLINED_CALL) {
-            tree = new_inlined_function_call(tree, oper->label);
+         if (oper->type == SLANG_OPER_NON_INLINED_CALL) {
+            tree = new_function_call(tree, oper->label);
          }
          return tree;
       }
diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c
index a4dd5b8b4a..4e29e8dcc9 100644
--- a/src/mesa/shader/slang/slang_compile.c
+++ b/src/mesa/shader/slang/slang_compile.c
@@ -28,11 +28,11 @@
  * \author Michal Krol
  */
 
-#include "imports.h"
-#include "context.h"
-#include "program.h"
-#include "prog_parameter.h"
-#include "grammar_mesa.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "shader/program.h"
+#include "shader/prog_parameter.h"
+#include "shader/grammar/grammar_mesa.h"
 #include "slang_codegen.h"
 #include "slang_compile.h"
 #include "slang_preprocess.h"
@@ -2135,7 +2135,7 @@ _slang_compile(GLcontext *ctx, struct gl_shader *shader)
          progTarget = GL_FRAGMENT_PROGRAM_ARB;
       shader->Programs
          = (struct gl_program **) malloc(sizeof(struct gl_program*));
-      shader->Programs[0] = _mesa_new_program(ctx, progTarget, 1);
+      shader->Programs[0] = ctx->Driver.NewProgram(ctx, progTarget, 1);
       shader->NumPrograms = 1;
 
       shader->Programs[0]->Parameters = _mesa_new_parameter_list();
diff --git a/src/mesa/shader/slang/slang_compile_operation.h b/src/mesa/shader/slang/slang_compile_operation.h
index d497b6f66f..d5cbe779a6 100644
--- a/src/mesa/shader/slang/slang_compile_operation.h
+++ b/src/mesa/shader/slang/slang_compile_operation.h
@@ -93,7 +93,7 @@ typedef enum slang_operation_type_
    SLANG_OPER_NOT,              /* "!" [expr] */
    SLANG_OPER_SUBSCRIPT,        /* [expr] "[" [expr] "]" */
    SLANG_OPER_CALL,             /* [func name] [param] [param] [...] */
-   SLANG_OPER_INLINED_CALL,     /* inlined function call */
+   SLANG_OPER_NON_INLINED_CALL, /* a real function call */
    SLANG_OPER_FIELD,            /* i.e.: ".next" or ".xzy" or ".xxx" etc */
    SLANG_OPER_POSTINCREMENT,    /* [var] "++" */
    SLANG_OPER_POSTDECREMENT     /* [var] "--" */
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 7804e19236..fe13f2865c 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -36,13 +36,13 @@
  ***/
 
 
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "program.h"
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "prog_print.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
 #include "slang_builtin.h"
 #include "slang_emit.h"
 #include "slang_mem.h"
@@ -780,16 +780,18 @@ emit_label(slang_emit_info *emitInfo, const slang_ir_node *n)
 
 
 /**
- * Emit code for an inlined function call (subroutine).
+ * Emit code for a function call.
+ * Note that for each time a function is called, we emit the function's
+ * body code again because the set of available registers may be different.
  */
 static struct prog_instruction *
-emit_func(slang_emit_info *emitInfo, slang_ir_node *n)
+emit_fcall(slang_emit_info *emitInfo, slang_ir_node *n)
 {
    struct gl_program *progSave;
    struct prog_instruction *inst;
    GLuint subroutineId;
 
-   assert(n->Opcode == IR_FUNC);
+   assert(n->Opcode == IR_CALL);
    assert(n->Label);
 
    /* save/push cur program */
@@ -1687,10 +1689,10 @@ emit(slang_emit_info *emitInfo, slang_ir_node *n)
    case IR_KILL:
       return emit_kill(emitInfo);
 
-   case IR_FUNC:
-      /* new variable scope for subroutines/function calls*/
+   case IR_CALL:
+      /* new variable scope for subroutines/function calls */
       _slang_push_var_table(emitInfo->vt);
-      inst = emit_func(emitInfo, n);
+      inst = emit_fcall(emitInfo, n);
       _slang_pop_var_table(emitInfo->vt);
       return inst;
 
@@ -1782,7 +1784,7 @@ _slang_resolve_subroutines(slang_emit_info *emitInfo)
    emitInfo->NumSubroutines = 0;
 
    /* Examine CAL instructions.
-    * At this point, the BranchTarget field of the CAL instructions is
+    * At this point, the BranchTarget field of the CAL instruction is
     * the number/id of the subroutine to call (an index into the
     * emitInfo->Subroutines list).
     * Translate that into an actual instruction location now.
diff --git a/src/mesa/shader/slang/slang_ir.c b/src/mesa/shader/slang/slang_ir.c
index a6903cc8b6..92e8d0345e 100644
--- a/src/mesa/shader/slang/slang_ir.c
+++ b/src/mesa/shader/slang/slang_ir.c
@@ -27,7 +27,7 @@
 #include "context.h"
 #include "slang_ir.h"
 #include "slang_mem.h"
-#include "prog_print.h"
+#include "shader/prog_print.h"
 
 
 static const slang_ir_info IrInfo[] = {
@@ -311,7 +311,7 @@ _slang_print_ir_tree(const slang_ir_node *n, int indent)
       printf("RETURN\n");
       break;
    case IR_CALL:
-      printf("CALL\n");
+      printf("CALL %s\n", n->Label->Name);
       break;
 
    case IR_LOOP:
diff --git a/src/mesa/shader/slang/slang_ir.h b/src/mesa/shader/slang/slang_ir.h
index 69db4b5451..c7c0ddbf9a 100644
--- a/src/mesa/shader/slang/slang_ir.h
+++ b/src/mesa/shader/slang/slang_ir.h
@@ -62,8 +62,6 @@ typedef enum
    IR_RETURN,    /* return from subroutine */
    IR_CALL,      /* call subroutine */
 
-   IR_FUNC,      /* inlined function code */
-
    IR_LOOP,      /* high-level loop-begin / loop-end */
                  /* Children[0] = loop body */
                  /* Children[1] = loop tail code, or NULL */
diff --git a/src/mesa/shader/slang/slang_label.h b/src/mesa/shader/slang/slang_label.h
index 0f1a45b30f..87068ae7a7 100644
--- a/src/mesa/shader/slang/slang_label.h
+++ b/src/mesa/shader/slang/slang_label.h
@@ -1,9 +1,9 @@
 #ifndef SLANG_LABEL_H
 #define SLANG_LABEL_H 1
 
-#include "imports.h"
-#include "mtypes.h"
-#include "prog_instruction.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "shader/prog_instruction.h"
 
 
 struct slang_label_
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index d6d1c7523e..c8457fc483 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -28,16 +28,16 @@
  * \author Brian Paul
  */
 
-#include "imports.h"
-#include "context.h"
-#include "hash.h"
-#include "macros.h"
-#include "program.h"
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "prog_print.h"
-#include "prog_statevars.h"
-#include "shader_api.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/hash.h"
+#include "main/macros.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
+#include "shader/prog_statevars.h"
+#include "shader/shader_api.h"
 #include "slang_link.h"
 
 
@@ -329,8 +329,10 @@ _slang_resolve_attributes(struct gl_shader_program *shProg,
                 * glVertex/position.
                 */
                for (attr = 1; attr < MAX_VERTEX_ATTRIBS; attr++) {
-                  if (((1 << attr) & usedAttributes) == 0)
+                  if (((1 << attr) & usedAttributes) == 0) {
+                     usedAttributes |= (1 << attr);
                      break;
+                  }
                }
                if (attr == MAX_VERTEX_ATTRIBS) {
                   /* too many!  XXX record error log */
diff --git a/src/mesa/shader/slang/slang_preprocess.c b/src/mesa/shader/slang/slang_preprocess.c
index 72281eda57..076e982f8f 100644
--- a/src/mesa/shader/slang/slang_preprocess.c
+++ b/src/mesa/shader/slang/slang_preprocess.c
@@ -29,7 +29,7 @@
  */
 
 #include "imports.h"
-#include "grammar_mesa.h"
+#include "shader/grammar/grammar_mesa.h"
 #include "slang_preprocess.h"
 
 LONGSTRING static const char *slang_pp_directives_syn =
diff --git a/src/mesa/shader/slang/slang_typeinfo.c b/src/mesa/shader/slang/slang_typeinfo.c
index da0b32bc44..8a1c3abf48 100644
--- a/src/mesa/shader/slang/slang_typeinfo.c
+++ b/src/mesa/shader/slang/slang_typeinfo.c
@@ -28,12 +28,12 @@
  * \author Michal Krol
  */
 
-#include "imports.h"
+#include "main/imports.h"
+#include "shader/prog_instruction.h"
 #include "slang_typeinfo.h"
 #include "slang_compile.h"
 #include "slang_log.h"
 #include "slang_mem.h"
-#include "prog_instruction.h"
 
 
 /**
diff --git a/src/mesa/shader/slang/slang_vartable.c b/src/mesa/shader/slang/slang_vartable.c
index 8a3c299d19..1d817000c6 100644
--- a/src/mesa/shader/slang/slang_vartable.c
+++ b/src/mesa/shader/slang/slang_vartable.c
@@ -1,11 +1,11 @@
 
-#include "imports.h"
+#include "main/imports.h"
+#include "shader/prog_instruction.h"
 #include "slang_compile.h"
 #include "slang_compile_variable.h"
 #include "slang_mem.h"
 #include "slang_vartable.h"
 #include "slang_ir.h"
-#include "prog_instruction.h"
 
 
 static int dbg = 0;
diff --git a/src/mesa/swrast/s_aalinetemp.h b/src/mesa/swrast/s_aalinetemp.h
index 69a1f0cd39..e7911fec3b 100644
--- a/src/mesa/swrast/s_aalinetemp.h
+++ b/src/mesa/swrast/s_aalinetemp.h
@@ -132,7 +132,9 @@ NAME(line)(GLcontext *ctx, const SWvertex *v0, const SWvertex *v1)
    line.dx = line.x1 - line.x0;
    line.dy = line.y1 - line.y0;
    line.len = SQRTF(line.dx * line.dx + line.dy * line.dy);
-   line.halfWidth = 0.5F * ctx->Line._Width;
+   line.halfWidth = 0.5F * CLAMP(ctx->Line.Width,
+                                 ctx->Const.MinLineWidthAA,
+                                 ctx->Const.MaxLineWidthAA);
 
    if (line.len == 0.0 || IS_INF_OR_NAN(line.len))
       return;
diff --git a/src/mesa/swrast/s_atifragshader.c b/src/mesa/swrast/s_atifragshader.c
index 947054faa3..55ec757ee0 100644
--- a/src/mesa/swrast/s_atifragshader.c
+++ b/src/mesa/swrast/s_atifragshader.c
@@ -23,11 +23,10 @@
 #include "glheader.h"
 #include "colormac.h"
 #include "context.h"
-#include "atifragshader.h"
 #include "macros.h"
-#include "program.h"
-
-#include "s_atifragshader.h"
+#include "shader/program.h"
+#include "shader/atifragshader.h"
+#include "swrast/s_atifragshader.h"
 
 
 /**
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index 791850cb50..3956925651 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.c
@@ -31,9 +31,9 @@
 #include "context.h"
 #include "colormac.h"
 #include "mtypes.h"
-#include "prog_statevars.h"
 #include "teximage.h"
 #include "swrast.h"
+#include "shader/prog_statevars.h"
 #include "s_blend.h"
 #include "s_context.h"
 #include "s_lines.h"
diff --git a/src/mesa/swrast/s_context.h b/src/mesa/swrast/s_context.h
index f118eb92ca..daa07e1578 100644
--- a/src/mesa/swrast/s_context.h
+++ b/src/mesa/swrast/s_context.h
@@ -43,10 +43,10 @@
 #ifndef S_CONTEXT_H
 #define S_CONTEXT_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
+#include "shader/prog_execute.h"
 #include "swrast.h"
 #include "s_span.h"
-#include "prog_execute.h"
 
 
 typedef void (*texture_sample_func)(GLcontext *ctx,
diff --git a/src/mesa/swrast/s_drawpix.c b/src/mesa/swrast/s_drawpix.c
index 1c9f64b275..0cf425e1c6 100644
--- a/src/mesa/swrast/s_drawpix.c
+++ b/src/mesa/swrast/s_drawpix.c
@@ -484,7 +484,7 @@ draw_depth_pixels( GLcontext *ctx, GLint x, GLint y,
    }
    else {
       /* General case */
-      const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;
+      const GLuint depthMax = ctx->DrawBuffer->_DepthMax;
       GLint skipPixels = 0;
 
       /* in case width > MAX_WIDTH do the copy in chunks */
@@ -695,7 +695,7 @@ draw_depth_stencil_pixels(GLcontext *ctx, GLint x, GLint y,
    const GLint imgX = x, imgY = y;
    const GLboolean scaleOrBias
       = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;
-   const GLfloat depthScale = ctx->DrawBuffer->_DepthMaxF;
+   const GLuint depthMax = ctx->DrawBuffer->_DepthMax;
    const GLuint stencilMask = ctx->Stencil.WriteMask[0];
    const GLuint stencilType = (STENCIL_BITS == 8) ? 
       GL_UNSIGNED_BYTE : GL_UNSIGNED_SHORT;
@@ -783,7 +783,7 @@ draw_depth_stencil_pixels(GLcontext *ctx, GLint x, GLint y,
                /* general case */
                GLuint zValues[MAX_WIDTH];  /* 16 or 32-bit Z value storage */
                _mesa_unpack_depth_span(ctx, width,
-                                       depthRb->DataType, zValues, depthScale,
+                                       depthRb->DataType, zValues, depthMax,
                                        type, depthStencilSrc, &clippedUnpack);
                if (zoom) {
                   _swrast_write_zoomed_z_span(ctx, imgX, imgY, width, x,
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 923b67e78e..14c9868c18 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -22,11 +22,11 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include "glheader.h"
-#include "colormac.h"
-#include "context.h"
-#include "prog_instruction.h"
-#include "texstate.h"
+#include "main/glheader.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/texstate.h"
+#include "shader/prog_instruction.h"
 
 #include "s_fragprog.h"
 #include "s_span.h"
diff --git a/src/mesa/swrast/s_lines.c b/src/mesa/swrast/s_lines.c
index 781146e67f..3de438760b 100644
--- a/src/mesa/swrast/s_lines.c
+++ b/src/mesa/swrast/s_lines.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -63,12 +63,13 @@ compute_stipple_mask( GLcontext *ctx, GLuint len, GLubyte mask[] )
 static void
 draw_wide_line( GLcontext *ctx, SWspan *span, GLboolean xMajor )
 {
-   GLint width, start;
+   const GLint width = (GLint) CLAMP(ctx->Line.Width,
+                                     ctx->Const.MinLineWidth,
+                                     ctx->Const.MaxLineWidth);
+   GLint start;
 
    ASSERT(span->end < MAX_WIDTH);
 
-   width = (GLint) CLAMP( ctx->Line._Width, MIN_LINE_WIDTH, MAX_LINE_WIDTH );
-
    if (width & 1)
       start = width / 2;
    else
@@ -143,7 +144,7 @@ draw_wide_line( GLcontext *ctx, SWspan *span, GLboolean xMajor )
       span.arrayMask |= SPAN_MASK;				\
       compute_stipple_mask(ctx, span.end, span.array->mask);    \
    }								\
-   if (ctx->Line._Width > 1.0) {				\
+   if (ctx->Line.Width > 1.0) {					\
       draw_wide_line(ctx, &span, (GLboolean)(dx > dy));		\
    }								\
    else {							\
@@ -161,7 +162,7 @@ draw_wide_line( GLcontext *ctx, SWspan *span, GLboolean xMajor )
       span.arrayMask |= SPAN_MASK;				\
       compute_stipple_mask(ctx, span.end, span.array->mask);	\
    }								\
-   if (ctx->Line._Width > 1.0) {				\
+   if (ctx->Line.Width > 1.0) {					\
       draw_wide_line(ctx, &span, (GLboolean)(dx > dy));		\
    }								\
    else {							\
@@ -180,7 +181,7 @@ draw_wide_line( GLcontext *ctx, SWspan *span, GLboolean xMajor )
       span.arrayMask |= SPAN_MASK;				\
       compute_stipple_mask(ctx, span.end, span.array->mask);	\
    }								\
-   if (ctx->Line._Width > 1.0) {				\
+   if (ctx->Line.Width > 1.0) {					\
       draw_wide_line(ctx, &span, (GLboolean)(dx > dy));		\
    }								\
    else {							\
@@ -274,16 +275,21 @@ _swrast_choose_line( GLcontext *ctx )
          USE(general_line);
       }
       else if (ctx->Depth.Test
-               || ctx->Line._Width != 1.0
+               || ctx->Line.Width != 1.0
                || ctx->Line.StippleFlag) {
          /* no texture, but Z, fog, width>1, stipple, etc. */
          if (rgbmode)
+#if CHAN_BITS == 32
+            USE(general_line);
+#else
             USE(rgba_line);
+#endif
          else
             USE(ci_line);
       }
       else {
          ASSERT(!ctx->Depth.Test);
+         ASSERT(ctx->Line.Width == 1.0);
          /* simple lines */
          if (rgbmode)
             USE(simple_no_z_rgba_line);
diff --git a/src/mesa/swrast/s_points.c b/src/mesa/swrast/s_points.c
index 8eba53c807..21316372e8 100644
--- a/src/mesa/swrast/s_points.c
+++ b/src/mesa/swrast/s_points.c
@@ -76,7 +76,7 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
    }
    else {
       /* use constant point size */
-      size = ctx->Point._Size; /* already clamped to user range */
+      size = ctx->Point.Size;
    }
    /* clamp to non-AA implementation limits */
    size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
@@ -172,9 +172,9 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       }
       else {
          /* even size */
-         xmin = (GLint) x - iRadius + 1;
+         xmin = (GLint) x - iRadius;
          xmax = xmin + iSize - 1;
-         ymin = (GLint) y - iRadius + 1;
+         ymin = (GLint) y - iRadius;
          ymax = ymin + iSize - 1;
       }
 
@@ -227,7 +227,7 @@ smooth_point(GLcontext *ctx, const SWvertex *vert)
    }
    else {
       /* use constant point size */
-      size = ctx->Point._Size; /* this is already clamped */
+      size = ctx->Point.Size;
    }
    /* clamp to AA implementation limits */
    size = CLAMP(size, ctx->Const.MinPointSizeAA, ctx->Const.MaxPointSizeAA);
@@ -361,7 +361,7 @@ large_point(GLcontext *ctx, const SWvertex *vert)
    }
    else {
       /* use constant point size */
-      size = ctx->Point._Size; /* already clamped to user range */
+      size = ctx->Point.Size;
    }
    /* clamp to non-AA implementation limits */
    size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
@@ -418,9 +418,9 @@ large_point(GLcontext *ctx, const SWvertex *vert)
       }
       else {
          /* even size */
-         xmin = (GLint) x - iRadius + 1;
+         xmin = (GLint) x - iRadius;
          xmax = xmin + iSize - 1;
-         ymin = (GLint) y - iRadius + 1;
+         ymin = (GLint) y - iRadius;
          ymax = ymin + iSize - 1;
       }
 
@@ -550,7 +550,7 @@ _swrast_choose_point(GLcontext *ctx)
       else if (ctx->Point.SmoothFlag) {
          swrast->Point = smooth_point;
       }
-      else if (ctx->Point._Size > 1.0 ||
+      else if (ctx->Point.Size > 1.0 ||
                ctx->Point._Attenuated ||
                ctx->VertexProgram.PointSizeEnabled) {
          swrast->Point = large_point;
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index f23272c2be..bbde0025f4 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -873,7 +873,7 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
 #endif
 
    /* we have to wait until after occlusion to do this test */
-   if (ctx->Color.DrawBuffer == GL_NONE || ctx->Color.IndexMask == 0) {
+   if (ctx->Color.IndexMask == 0) {
       /* write no pixels */
       span->arrayMask = origArrayMask;
       return;
@@ -1358,7 +1358,7 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
 
 #if CHAN_BITS == 32
    if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-      interpolate_int_colors(ctx, span);
+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
    }
 #else
    if ((span->arrayMask & SPAN_RGBA) == 0) {
diff --git a/src/mesa/swrast/s_stencil.c b/src/mesa/swrast/s_stencil.c
index 89991fad02..d0cbdd6917 100644
--- a/src/mesa/swrast/s_stencil.c
+++ b/src/mesa/swrast/s_stencil.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   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"),
@@ -392,6 +392,23 @@ do_stencil_test( GLcontext *ctx, GLuint face, GLuint n, GLstencil stencil[],
 }
 
 
+/**
+ * Compute the zpass/zfail masks by comparing the pre- and post-depth test
+ * masks.
+ */
+static INLINE void
+compute_pass_fail_masks(GLuint n, const GLubyte origMask[],
+                        const GLubyte newMask[],
+                        GLubyte passMask[], GLubyte failMask[])
+{
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      ASSERT(newMask[i] == 0 || newMask[i] == 1);
+      passMask[i] = origMask[i] & newMask[i];
+      failMask[i] = origMask[i] & (newMask[i] ^ 1);
+   }
+}
+
 
 /**
  * Apply stencil and depth testing to the span of pixels.
@@ -460,39 +477,24 @@ stencil_and_ztest_span(GLcontext *ctx, SWspan *span, GLuint face)
       /*
        * Perform depth buffering, then apply zpass or zfail stencil function.
        */
-      GLubyte passmask[MAX_WIDTH], failmask[MAX_WIDTH], oldmask[MAX_WIDTH];
-      GLuint i;
+      GLubyte passMask[MAX_WIDTH], failMask[MAX_WIDTH], origMask[MAX_WIDTH];
 
       /* save the current mask bits */
-      _mesa_memcpy(oldmask, mask, n * sizeof(GLubyte));
+      _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));
 
       /* apply the depth test */
       _swrast_depth_test_span(ctx, span);
 
-      /* Set the stencil pass/fail flags according to result of depth testing.
-       * if oldmask[i] == 0 then
-       *    Don't touch the stencil value
-       * else if oldmask[i] and newmask[i] then
-       *    Depth test passed
-       * else
-       *    assert(oldmask[i] && !newmask[i])
-       *    Depth test failed
-       * endif
-       */
-      for (i=0;i<n;i++) {
-         ASSERT(mask[i] == 0 || mask[i] == 1);
-         passmask[i] = oldmask[i] & mask[i];
-         failmask[i] = oldmask[i] & (mask[i] ^ 1);
-      }
+      compute_pass_fail_masks(n, origMask, mask, passMask, failMask);
 
       /* apply the pass and fail operations */
       if (ctx->Stencil.ZFailFunc[face] != GL_KEEP) {
          apply_stencil_op( ctx, ctx->Stencil.ZFailFunc[face], face,
-                           n, stencil, failmask );
+                           n, stencil, failMask );
       }
       if (ctx->Stencil.ZPassFunc[face] != GL_KEEP) {
          apply_stencil_op( ctx, ctx->Stencil.ZPassFunc[face], face,
-                           n, stencil, passmask );
+                           n, stencil, passMask );
       }
    }
 
@@ -902,6 +904,7 @@ stencil_test_pixels( GLcontext *ctx, GLuint face, GLuint n,
 static GLboolean
 stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
 {
+   GLubyte passMask[MAX_WIDTH], failMask[MAX_WIDTH], origMask[MAX_WIDTH];
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    struct gl_renderbuffer *rb = fb->_StencilBuffer;
    const GLuint n = span->end;
@@ -916,13 +919,10 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
    if (!rb->GetPointer(ctx, rb, 0, 0)) {
       /* No direct access */
       GLstencil stencil[MAX_WIDTH];
-      GLubyte origMask[MAX_WIDTH];
 
       ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
       _swrast_get_values(ctx, rb, n, x, y, stencil, sizeof(GLubyte));
 
-      _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));
-
       (void) do_stencil_test(ctx, face, n, stencil, mask);
 
       if (ctx->Depth.Test == GL_FALSE) {
@@ -930,27 +930,19 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
                           n, stencil, mask);
       }
       else {
+         _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));
+
          _swrast_depth_test_span(ctx, span);
 
+         compute_pass_fail_masks(n, origMask, mask, passMask, failMask);
+
          if (ctx->Stencil.ZFailFunc[face] != GL_KEEP) {
-            GLubyte failmask[MAX_WIDTH];
-            GLuint i;
-            for (i = 0; i < n; i++) {
-               ASSERT(mask[i] == 0 || mask[i] == 1);
-               failmask[i] = origMask[i] & (mask[i] ^ 1);
-            }
             apply_stencil_op(ctx, ctx->Stencil.ZFailFunc[face], face,
-                             n, stencil, failmask);
+                             n, stencil, failMask);
          }
          if (ctx->Stencil.ZPassFunc[face] != GL_KEEP) {
-            GLubyte passmask[MAX_WIDTH];
-            GLuint i;
-            for (i = 0; i < n; i++) {
-               ASSERT(mask[i] == 0 || mask[i] == 1);
-               passmask[i] = origMask[i] & mask[i];
-            }
             apply_stencil_op(ctx, ctx->Stencil.ZPassFunc[face], face,
-                             n, stencil, passmask);
+                             n, stencil, passMask);
          }
       }
 
@@ -972,28 +964,21 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
                                     ctx->Stencil.ZPassFunc[face], face, mask);
       }
       else {
-         GLubyte passmask[MAX_WIDTH], failmask[MAX_WIDTH], oldmask[MAX_WIDTH];
-         GLuint i;
-
-         _mesa_memcpy(oldmask, mask, n * sizeof(GLubyte));
+         _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));
 
          _swrast_depth_test_span(ctx, span);
 
-         for (i=0;i<n;i++) {
-            ASSERT(mask[i] == 0 || mask[i] == 1);
-            passmask[i] = oldmask[i] & mask[i];
-            failmask[i] = oldmask[i] & (mask[i] ^ 1);
-         }
+         compute_pass_fail_masks(n, origMask, mask, passMask, failMask);
 
          if (ctx->Stencil.ZFailFunc[face] != GL_KEEP) {
             apply_stencil_op_to_pixels(ctx, n, x, y,
                                        ctx->Stencil.ZFailFunc[face],
-                                       face, failmask);
+                                       face, failMask);
          }
          if (ctx->Stencil.ZPassFunc[face] != GL_KEEP) {
             apply_stencil_op_to_pixels(ctx, n, x, y,
                                        ctx->Stencil.ZPassFunc[face],
-                                       face, passmask);
+                                       face, passMask);
          }
       }
 
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index d101a9e2ae..85a27fd55b 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -32,7 +32,7 @@
 #ifndef SWRAST_H
 #define SWRAST_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 /**
  * \struct SWvertex
diff --git a/src/mesa/swrast_setup/ss_context.c b/src/mesa/swrast_setup/ss_context.c
index f8a1cadfa5..a9c7d941e5 100644
--- a/src/mesa/swrast_setup/ss_context.c
+++ b/src/mesa/swrast_setup/ss_context.c
@@ -280,26 +280,29 @@ _swsetup_Translate( GLcontext *ctx, const void *vertex, SWvertex *dest )
 
    /** XXX try to limit these loops someday */
    for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++)
-      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_TEX0+i,
+      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_TEX0 + i,
                      dest->attrib[FRAG_ATTRIB_TEX0 + i] );
 
    for (i = 0 ; i < ctx->Const.MaxVarying ; i++)
-      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_GENERIC0+i,
+      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_GENERIC0 + i,
                      dest->attrib[FRAG_ATTRIB_VAR0 + i] );
 
-   _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR0, tmp );
-   /* XXX need float color FRAG_ATTRIB_COL0?? */
-   UNCLAMPED_FLOAT_TO_RGBA_CHAN( dest->color, tmp );
+   if (ctx->Visual.rgbMode) {
+      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR0,
+                     dest->attrib[FRAG_ATTRIB_COL0] );
+      UNCLAMPED_FLOAT_TO_RGBA_CHAN( dest->color, tmp );
 
-   _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR1, tmp );
-   COPY_4V(dest->attrib[FRAG_ATTRIB_COL1], tmp);
+      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR1,
+                     dest->attrib[FRAG_ATTRIB_COL1]);
+   }
+   else {
+      _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR_INDEX, tmp );
+      dest->attrib[FRAG_ATTRIB_CI][0] = tmp[0];
+   }
 
    _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_FOG, tmp );
    dest->attrib[FRAG_ATTRIB_FOGC][0] = tmp[0];
 
-   _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR_INDEX, tmp );
-   dest->attrib[FRAG_ATTRIB_CI][0] = tmp[0];
-
    /* XXX See _tnl_get_attr about pointsize ... */
    _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_POINTSIZE, tmp );
    dest->pointSize = tmp[0];
diff --git a/src/mesa/swrast_setup/ss_tritmp.h b/src/mesa/swrast_setup/ss_tritmp.h
index c14468e951..f6b738d60d 100644
--- a/src/mesa/swrast_setup/ss_tritmp.h
+++ b/src/mesa/swrast_setup/ss_tritmp.h
@@ -37,7 +37,7 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
    SWvertex *verts = SWSETUP_CONTEXT(ctx)->verts;
    SWvertex *v[3];
    GLfloat z[3];
-   GLfloat offset;
+   GLfloat offset, oz0, oz1, oz2;
    GLenum mode = GL_FILL;
    GLuint facing = 0;
    GLchan saved_color[3][4];
@@ -142,12 +142,16 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	 }
       }
 
-      if (IND & SS_OFFSET_BIT)
-      {
-	 offset = ctx->Polygon.OffsetUnits * ctx->DrawBuffer->_MRD;
+      if (IND & SS_OFFSET_BIT) {
+         const GLfloat max = ctx->DrawBuffer->_DepthMaxF;
+         /* save original Z values (restored later) */
 	 z[0] = v[0]->attrib[FRAG_ATTRIB_WPOS][2];
 	 z[1] = v[1]->attrib[FRAG_ATTRIB_WPOS][2];
 	 z[2] = v[2]->attrib[FRAG_ATTRIB_WPOS][2];
+         /* Note that Z values are already scaled to [0,65535] (for example)
+          * so no MRD value is used here.
+          */
+	 offset = ctx->Polygon.OffsetUnits;
 	 if (cc * cc > 1e-16) {
 	    const GLfloat ez = z[0] - z[2];
 	    const GLfloat fz = z[1] - z[2];
@@ -155,35 +159,33 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	    const GLfloat dzdx = FABSF((ey * fz - ez * fy) * oneOverArea);
 	    const GLfloat dzdy = FABSF((ez * fx - ex * fz) * oneOverArea);
 	    offset += MAX2(dzdx, dzdy) * ctx->Polygon.OffsetFactor;
-            /* Unfortunately, we need to clamp to prevent negative Zs below.
-             * Technically, we should do the clamping per-fragment.
-             */
-            offset = MAX2(offset, -v[0]->attrib[FRAG_ATTRIB_WPOS][2]);
-            offset = MAX2(offset, -v[1]->attrib[FRAG_ATTRIB_WPOS][2]);
-            offset = MAX2(offset, -v[2]->attrib[FRAG_ATTRIB_WPOS][2]);
 	 }
+         /* new Z values */
+         oz0 = CLAMP(v[0]->attrib[FRAG_ATTRIB_WPOS][2] + offset, 0.0, max);
+         oz1 = CLAMP(v[1]->attrib[FRAG_ATTRIB_WPOS][2] + offset, 0.0, max);
+         oz2 = CLAMP(v[2]->attrib[FRAG_ATTRIB_WPOS][2] + offset, 0.0, max);
       }
    }
 
    if (mode == GL_POINT) {
       if ((IND & SS_OFFSET_BIT) && ctx->Polygon.OffsetPoint) {
-	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
+	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] = oz0;
+	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] = oz1;
+	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] = oz2;
       }
       _swsetup_render_point_tri( ctx, e0, e1, e2, facing );
    } else if (mode == GL_LINE) {
       if ((IND & SS_OFFSET_BIT) && ctx->Polygon.OffsetLine) {
-	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
+	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] = oz0;
+	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] = oz1;
+	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] = oz2;
       }
       _swsetup_render_line_tri( ctx, e0, e1, e2, facing );
    } else {
       if ((IND & SS_OFFSET_BIT) && ctx->Polygon.OffsetFill) {
-	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
-	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] += offset;
+	 v[0]->attrib[FRAG_ATTRIB_WPOS][2] = oz0;
+	 v[1]->attrib[FRAG_ATTRIB_WPOS][2] = oz1;
+	 v[2]->attrib[FRAG_ATTRIB_WPOS][2] = oz2;
       }
       _swrast_Triangle( ctx, v[0], v[1], v[2] );
    }
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index 3017c73cf1..3b8dd18bbb 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -26,12 +26,12 @@
  */
 
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "light.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/light.h"
 
 #include "tnl.h"
 #include "t_context.h"
diff --git a/src/mesa/tnl/t_context.h b/src/mesa/tnl/t_context.h
index 31b89aca41..baf283ef0f 100644
--- a/src/mesa/tnl/t_context.h
+++ b/src/mesa/tnl/t_context.h
@@ -49,8 +49,8 @@
 #ifndef _T_CONTEXT_H
 #define _T_CONTEXT_H
 
-#include "glheader.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
 
 #include "math/m_matrix.h"
 #include "math/m_vector.h"
diff --git a/src/mesa/tnl/t_pipeline.c b/src/mesa/tnl/t_pipeline.c
index c7188da34a..2a0ed8852a 100644
--- a/src/mesa/tnl/t_pipeline.c
+++ b/src/mesa/tnl/t_pipeline.c
@@ -25,11 +25,11 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "imports.h"
-#include "state.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/state.h"
+#include "main/mtypes.h"
 
 #include "t_context.h"
 #include "t_pipeline.h"
diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c
index 9961af70ce..f8e561ac57 100644
--- a/src/mesa/tnl/t_vb_program.c
+++ b/src/mesa/tnl/t_vb_program.c
@@ -35,16 +35,16 @@
 #include "context.h"
 #include "macros.h"
 #include "imports.h"
-#include "prog_instruction.h"
-#include "prog_statevars.h"
-#include "prog_execute.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_statevars.h"
+#include "shader/prog_execute.h"
+#include "swrast/s_context.h"
+#include "swrast/s_texfilter.h"
 
 #include "tnl.h"
 #include "t_context.h"
 #include "t_pipeline.h"
 
-#include "swrast/s_context.h"
-#include "swrast/s_texfilter.h"
 
 /**
  * XXX the texture sampling code in this module is a bit of a hack.
diff --git a/src/mesa/tnl/t_vertex.c b/src/mesa/tnl/t_vertex.c
index 6aae602037..a6728c318f 100644
--- a/src/mesa/tnl/t_vertex.c
+++ b/src/mesa/tnl/t_vertex.c
@@ -233,7 +233,7 @@ void _tnl_get_attr( GLcontext *ctx, const void *vin,
       /* If the hardware vertex doesn't have point size then use size from
        * GLcontext.  XXX this will be wrong if drawing attenuated points!
        */
-      dest[0] = ctx->Point._Size;
+      dest[0] = ctx->Point.Size;
    }
    else {
       _mesa_memcpy( dest, ctx->Current.Attrib[attr], 4*sizeof(GLfloat));
diff --git a/src/mesa/tnl/t_vp_build.c b/src/mesa/tnl/t_vp_build.c
index 2a1cae77f2..ee1a2498b3 100644
--- a/src/mesa/tnl/t_vp_build.c
+++ b/src/mesa/tnl/t_vp_build.c
@@ -33,11 +33,11 @@
 #include "glheader.h"
 #include "macros.h"
 #include "enums.h"
-#include "program.h"
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "prog_print.h"
-#include "prog_statevars.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
+#include "shader/prog_statevars.h"
 #include "t_context.h" /* NOTE: very light dependency on this */
 #include "t_vp_build.h"
 
diff --git a/src/mesa/tnl/tnl.h b/src/mesa/tnl/tnl.h
index 20bed5546d..047b764dcb 100644
--- a/src/mesa/tnl/tnl.h
+++ b/src/mesa/tnl/tnl.h
@@ -29,7 +29,7 @@
 #ifndef _TNL_H
 #define _TNL_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 
 
diff --git a/src/mesa/tnl_dd/t_dd_vb.c b/src/mesa/tnl_dd/t_dd_vb.c
index ab3bb37631..3cedd90119 100644
--- a/src/mesa/tnl_dd/t_dd_vb.c
+++ b/src/mesa/tnl_dd/t_dd_vb.c
@@ -184,7 +184,7 @@ void TAG(translate_vertex)(GLcontext *ctx,
       }
    }
 
-   dst->pointSize = ctx->Point._Size;
+   dst->pointSize = ctx->Point.Size;
 }
 
 
diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h
index 874a5f9e0e..04c59c05b2 100644
--- a/src/mesa/vbo/vbo.h
+++ b/src/mesa/vbo/vbo.h
@@ -32,7 +32,7 @@
 #ifndef _VBO_H
 #define _VBO_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 struct _mesa_prim {
    GLuint mode:8;
diff --git a/src/mesa/vbo/vbo_context.c b/src/mesa/vbo/vbo_context.c
index f64f59d11e..656b27f28c 100644
--- a/src/mesa/vbo/vbo_context.c
+++ b/src/mesa/vbo/vbo_context.c
@@ -25,11 +25,11 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "mtypes.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/api_arrayelt.h"
 #include "vbo.h"
 #include "vbo_context.h"
-#include "imports.h"
-#include "api_arrayelt.h"
 
 /* Reach out and grab this to use as the default:
  */
@@ -112,6 +112,8 @@ static void init_mat_currval(GLcontext *ctx)
    struct gl_client_array *arrays = vbo->mat_currval;
    GLuint i;
 
+   ASSERT(NR_MAT_ATTRIBS == MAT_ATTRIB_MAX);
+
    memset(arrays, 0, sizeof(*arrays) * NR_MAT_ATTRIBS);
 
    /* Set up a constant (StrideB == 0) array for each current
@@ -137,11 +139,7 @@ static void init_mat_currval(GLcontext *ctx)
 	 break;
       }
 
-      if (i < MAT_ATTRIB_MAX)
-	 cl->Ptr = (const void *)ctx->Light.Material.Attrib[i];
-      else 
-	 cl->Ptr = (const void *)ctx->Current.Attrib[VERT_ATTRIB_GENERIC0 + i];
-
+      cl->Ptr = (const void *)ctx->Light.Material.Attrib[i];
       cl->Type = GL_FLOAT;
       cl->Stride = 0;
       cl->StrideB = 0;
diff --git a/src/mesa/vbo/vbo_exec.c b/src/mesa/vbo/vbo_exec.c
index 7d95873247..1efa74945d 100644
--- a/src/mesa/vbo/vbo_exec.c
+++ b/src/mesa/vbo/vbo_exec.c
@@ -26,14 +26,14 @@
  */
 
 
-#include "api_arrayelt.h"
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "dlist.h"
-#include "vtxfmt.h"
+#include "main/api_arrayelt.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/dlist.h"
+#include "main/vtxfmt.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_exec.h b/src/mesa/vbo/vbo_exec.h
index a9b01e08e6..b7e8c9fe79 100644
--- a/src/mesa/vbo/vbo_exec.h
+++ b/src/mesa/vbo/vbo_exec.h
@@ -34,7 +34,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __VBO_EXEC_H__
 #define __VBO_EXEC_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "vbo.h"
 #include "vbo_attrib.h"
 
diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c
index 2d4ded0f98..7f56b3b629 100644
--- a/src/mesa/vbo/vbo_exec_api.c
+++ b/src/mesa/vbo/vbo_exec_api.c
@@ -30,16 +30,16 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "vtxfmt.h"
-#include "dlist.h"
-#include "state.h"
-#include "light.h"
-#include "api_arrayelt.h"
-#include "api_noop.h"
-#include "dispatch.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/vtxfmt.h"
+#include "main/dlist.h"
+#include "main/state.h"
+#include "main/light.h"
+#include "main/api_arrayelt.h"
+#include "main/api_noop.h"
+#include "glapi/dispatch.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c
index 1e4c310203..77f3cf1455 100644
--- a/src/mesa/vbo/vbo_exec_array.c
+++ b/src/mesa/vbo/vbo_exec_array.c
@@ -25,12 +25,12 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "context.h"
-#include "state.h"
-#include "api_validate.h"
-#include "api_noop.h"
-#include "dispatch.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/state.h"
+#include "main/api_validate.h"
+#include "main/api_noop.h"
+#include "glapi/dispatch.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index d8f167b357..0ef26cdfe3 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -25,11 +25,11 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "enums.h"
-#include "state.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/state.h"
+#include "main/macros.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_exec_eval.c b/src/mesa/vbo/vbo_exec_eval.c
index fe533290bd..0ba5585d24 100644
--- a/src/mesa/vbo/vbo_exec_eval.c
+++ b/src/mesa/vbo/vbo_exec_eval.c
@@ -25,13 +25,13 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "api_eval.h"
-#include "context.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/api_eval.h"
+#include "main/context.h"
+#include "main/macros.h"
 #include "math/m_eval.h"
+#include "glapi/dispatch.h"
 #include "vbo_exec.h"
-#include "dispatch.h"
 
 
 static void clear_active_eval1( struct vbo_exec_context *exec, GLuint attr ) 
diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c
index bc4211d852..dae778e741 100644
--- a/src/mesa/vbo/vbo_rebase.c
+++ b/src/mesa/vbo/vbo_rebase.c
@@ -46,9 +46,9 @@
  * of zero.
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "vbo.h"
 
diff --git a/src/mesa/vbo/vbo_save.c b/src/mesa/vbo/vbo_save.c
index e7f4687963..87248e10f3 100644
--- a/src/mesa/vbo/vbo_save.c
+++ b/src/mesa/vbo/vbo_save.c
@@ -26,10 +26,10 @@
  */
 
 
-#include "mtypes.h"
-#include "dlist.h"
-#include "vtxfmt.h"
-#include "imports.h"
+#include "main/mtypes.h"
+#include "main/dlist.h"
+#include "main/vtxfmt.h"
+#include "main/imports.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_save.h b/src/mesa/vbo/vbo_save.h
index b81f275a60..b7e9baabf8 100644
--- a/src/mesa/vbo/vbo_save.h
+++ b/src/mesa/vbo/vbo_save.h
@@ -34,7 +34,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef VBO_SAVE_H
 #define VBO_SAVE_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "vbo.h"
 #include "vbo_attrib.h"
 
diff --git a/src/mesa/vbo/vbo_save_api.c b/src/mesa/vbo/vbo_save_api.c
index e7794c2a6c..f62be5c14c 100644
--- a/src/mesa/vbo/vbo_save_api.c
+++ b/src/mesa/vbo/vbo_save_api.c
@@ -67,15 +67,15 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
 
-#include "glheader.h"
-#include "context.h"
-#include "dlist.h"
-#include "enums.h"
-#include "macros.h"
-#include "api_validate.h"
-#include "api_arrayelt.h"
-#include "vtxfmt.h"
-#include "dispatch.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/dlist.h"
+#include "main/enums.h"
+#include "main/macros.h"
+#include "main/api_validate.h"
+#include "main/api_arrayelt.h"
+#include "main/vtxfmt.h"
+#include "glapi/dispatch.h"
 
 #include "vbo_context.h"
 
@@ -864,6 +864,9 @@ static void GLAPIENTRY _save_OBE_DrawElements(GLenum mode, GLsizei count, GLenum
 
    _ae_map_vbos( ctx );
 
+   if (ctx->Array.ElementArrayBufferObj->Name)
+      indices = ADD_POINTERS(ctx->Array.ElementArrayBufferObj->Pointer, indices);
+
    vbo_save_NotifyBegin( ctx, mode | VBO_SAVE_PRIM_WEAK );
 
    switch (type) {
diff --git a/src/mesa/vbo/vbo_save_draw.c b/src/mesa/vbo/vbo_save_draw.c
index 8940551d08..3c6f0fccd9 100644
--- a/src/mesa/vbo/vbo_save_draw.c
+++ b/src/mesa/vbo/vbo_save_draw.c
@@ -26,13 +26,13 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "light.h"
-#include "state.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/light.h"
+#include "main/state.h"
 
 #include "vbo_context.h"
 
diff --git a/src/mesa/vbo/vbo_save_loopback.c b/src/mesa/vbo/vbo_save_loopback.c
index 430333b84d..f2cef698fb 100644
--- a/src/mesa/vbo/vbo_save_loopback.c
+++ b/src/mesa/vbo/vbo_save_loopback.c
@@ -28,17 +28,17 @@
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/swrast.h"
 #include "tnl/tnl.h"
-#include "context.h"
+#include "main/context.h"
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "glapi/dispatch.h"
+#include "glapi/glapi.h"
 
 #include "vbo_context.h"
 
-#include "glheader.h"
-#include "enums.h"
-#include "glapi.h"
-#include "imports.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "dispatch.h"
 
 
 typedef void (*attr_func)( GLcontext *ctx, GLint target, const GLfloat * );
diff --git a/src/mesa/vbo/vbo_split.c b/src/mesa/vbo/vbo_split.c
index ef205a3bb1..58e879628d 100644
--- a/src/mesa/vbo/vbo_split.c
+++ b/src/mesa/vbo/vbo_split.c
@@ -47,9 +47,9 @@
  * limitations on drivers which want to use it as a fallback path.
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "vbo_split.h"
 #include "vbo.h"
diff --git a/src/mesa/vbo/vbo_split_copy.c b/src/mesa/vbo/vbo_split_copy.c
index e142dde680..685cc0fdf6 100644
--- a/src/mesa/vbo/vbo_split_copy.c
+++ b/src/mesa/vbo/vbo_split_copy.c
@@ -29,11 +29,11 @@
 /* Split indexed primitives with per-vertex copying.
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "macros.h"
-#include "enums.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
 
 #include "vbo_split.h"
 #include "vbo.h"
@@ -129,6 +129,13 @@ static GLuint attr_size( const struct gl_client_array *array )
  */
 static GLboolean check_flush( struct copy_context *copy )
 {
+   GLenum mode = copy->dstprim[copy->dstprim_nr].mode; 
+
+   if (GL_TRIANGLE_STRIP == mode &&
+       copy->dstelt_nr & 1) { /* see bug9962 */
+       return GL_FALSE;
+   }
+
    if (copy->dstbuf_nr + 4 > copy->dstbuf_size)
       return GL_TRUE;
 
@@ -458,7 +465,7 @@ static void replay_init( struct copy_context *copy )
       dst->StrideB = copy->vertex_size;
       dst->Ptr = copy->dstbuf + offset;
       dst->Enabled = GL_TRUE;
-      dst->Normalized = GL_TRUE;
+      dst->Normalized = src->Normalized; 
       dst->BufferObj = ctx->Array.NullBufferObj;
       dst->_MaxElement = copy->dstbuf_size; /* may be less! */
 
diff --git a/src/mesa/vbo/vbo_split_inplace.c b/src/mesa/vbo/vbo_split_inplace.c
index ea62866e7c..958afccd0c 100644
--- a/src/mesa/vbo/vbo_split_inplace.c
+++ b/src/mesa/vbo/vbo_split_inplace.c
@@ -27,9 +27,9 @@
  */
 
 
-#include "mtypes.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "vbo_split.h"
author	Roland Scheidegger <sroland@tungstengraphics.com>	2007-08-18 22:31:43 +0200
committer	Roland Scheidegger <sroland@tungstengraphics.com>	2007-08-18 22:31:43 +0200
commit	4b4fc25c3fab92bf292a78cb9c16a5a32a8bb01b (patch)
tree	a7fb45a41c6558a3304210fd72cf211258996dbc
parent	7aefdd518578687f961bb512a2b408857f0b9824 (diff)
parent	53cf87be1b93c760228e6a9af8115d2a9ff99337 (diff)