- Reintegrate PPC support from old pcigart branchati-pcigart-1-0-0-20010405-freeze

19 files changed, 350 insertions, 121 deletions

diff --git a/xc/config/cf/host.def b/xc/config/cf/host.def
new file mode 100644
index 000000000..13da5d4a9
--- /dev/null
+++ b/xc/config/cf/host.def
@@ -0,0 +1,62 @@
+
+#define DefaultGcc2AxpOpt -O2 -mcpu=ev6
+#define DefaultGcc2AxpOpt -O2 -mcpu=750
+#define DefaultGcc2i386Opt -O2
+#ifdef AlphaArchitecture
+#define LibraryCDebugFlags -O2 -mcpu=ev6
+#else
+#ifdef PpcArchitecture
+#define LibraryCDebugFlags -O2 -mcpu=750
+#else
+#define LibraryCDebugFlags -O2
+#endif
+#endif
+#define BuildServersOnly YES
+#define BuildXFree86ConfigTools NO
+#define XF86CardDrivers vga tdfx i810 mga ati glint
+#define LinuxDistribution LinuxRedHat
+#define GccWarningOptions -Wall -Wpointer-arith -Wstrict-prototypes \
+                          -Wmissing-prototypes -Wmissing-declarations \
+                          -Wnested-externs
+#define DefaultCCOptions -ansi GccWarningOptions -pipe -g
+#define NormalLibGlx NO
+
+#define BuildXF86DRI YES
+#define HasGlide3 YES
+
+#ifdef i386Architecture
+#define MesaUse3DNow YES
+#if HasKatmaiSupport
+#  define MesaUseKatmai YES
+#endif
+#endif
+
+/* To do profiling of the dynamically loaded 'xyz_dri.so' object, turn
+ * this on.
+ * Use 'xc/lib/GL/makeprofile.sh' to make it work.
+ */
+/* #define GlxSoProf YES */
+
+#ifdef GlxSoProf
+#  undef DefaultCCOptions
+#  define DefaultCCOptions -ansi GccWarningOptions -pipe -g -p
+#endif
+
+/* Optionally turn these on for debugging */
+/* #define GlxBuiltInTdfx YES */
+/* #define GlxBuiltInI810 YES */
+/* #define GlxBuiltInMga YES */
+/* #define GlxBuiltInR128 YES */
+/* #define GlxBuiltInRadeon YES */
+/* #define DoLoadableServer NO */
+
+/* Optionally turn this on to change the place where you install the build */
+/* #define ProjectRoot /usr/X11R6-DRI */
+
+/* Optionally turn this on to force the kernel modules to build */
+/* #define BuildXF86DRM YES */
+
+#define XnestServer NO
+#define XVirtualFramebufferServer NO
+#define XprtServer NO
+
diff --git a/xc/lib/GL/mesa/src/drv/Imakefile b/xc/lib/GL/mesa/src/drv/Imakefile
index 6639512c2..dcd78c5e8 100644
--- a/xc/lib/GL/mesa/src/drv/Imakefile
+++ b/xc/lib/GL/mesa/src/drv/Imakefile
@@ -52,7 +52,7 @@ SUBDIRS += sis
 SUBDIRS += tdfx
 #endif
 
-#elif defined(AlphaArchitecture)
+#elif defined(AlphaArchitecture) || defined(PpcArchitecture)
 
 SUBDIRS += common
 SUBDIRS += gamma
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_context.h b/xc/lib/GL/mesa/src/drv/r128/r128_context.h
index 8c5e2fdfc..f0ea2a14e 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_context.h
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_context.h
@@ -88,10 +88,16 @@ typedef struct r128_context *r128ContextPtr;
 #define R128_BLEND_ENV_COLOR	0x1
 #define R128_BLEND_MULTITEX	0x2
 
-/* Subpixel offsets for window coordinates:
+/* Subpixel offsets for window coordinates (triangles):
  */
-#define SUBPIXEL_X		(-0.125F)
-#define SUBPIXEL_Y		( 0.375F)
+#define SUBPIXEL_X  (0.0F)
+#define SUBPIXEL_Y  (0.125F)
+
+/* Offset for points:
+ */
+#define PNT_X_OFFSET  ( 0.125F)
+#define PNT_Y_OFFSET  (-0.125F)
+
 
 typedef void (*r128_interp_func)( GLfloat t,
 				  GLfloat *result,
@@ -127,10 +133,12 @@ struct r128_context {
    GLuint vc_format;
    GLfloat depth_scale;
 
-   CARD32 Color;			/* Current draw color */
-   CARD32 ClearColor;			/* Color used to clear color buffer */
-   CARD32 ClearDepth;			/* Value used to clear depth buffer */
-   CARD32 ClearStencil;			/* Value used to clear stencil */
+   GLuint Color;			/* Current draw color */
+   GLuint ClearColor;			/* Color used to clear color buffer */
+   GLuint ClearDepth;			/* Value used to clear depth buffer */
+   GLuint ClearStencil;			/* Value used to clear stencil */
+   GLuint DepthMask;
+   GLuint StencilMask;
 
    /* Map GL texture units onto hardware
     */
@@ -210,6 +218,8 @@ struct r128_context {
    __DRIscreenPrivate	*driScreen;	/* DRI screen */
    __DRIdrawablePrivate	*driDrawable;	/* DRI drawable bound to this ctx */
 
+   int lastStamp;		        /* mirror driDrawable->lastStamp */
+
    drmContext hHWContext;
    drmLock *driHwLock;
    int driFd;
@@ -245,14 +255,32 @@ extern r128ContextPtr r128MakeCurrent( r128ContextPtr oldCtx,
 				       r128ContextPtr newCtx,
 				       __DRIdrawablePrivate *dPriv );
 
+/* ================================================================
+ * Byte ordering
+ */
+#include <endian.h>
+
+#if __BYTE_ORDER == __LITTLE_ENDIAN
+#define LE32_OUT( x, y )	do { x = y; } while (0)
+#define LE32_OUT_FLOAT( x, y )	do { *(GLfloat *)&(x) = y; } while (0)
+#else
+#include <byteswap.h>
+#define LE32_OUT( x, y )	do { x = bswap_32( y ); } while (0)
+#define LE32_OUT_FLOAT( x, y )						\
+do {									\
+   GLuint __tmp;							\
+   *(GLfloat *)&__tmp = y;						\
+   x = bswap_32( __tmp );						\
+} while (0)
+#endif
 
 /* ================================================================
  * Debugging:
  */
-#define DEBUG			0
+#define DO_DEBUG		0
 #define ENABLE_PERF_BOXES	0
 
-#if DEBUG
+#if DO_DEBUG
 extern int R128_DEBUG;
 #else
 #define R128_DEBUG		0
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_dd.c b/xc/lib/GL/mesa/src/drv/r128/r128_dd.c
index 60b9c8606..5b595120e 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_dd.c
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_dd.c
@@ -45,7 +45,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "X86/common_x86_asm.h"
 #endif
 
-#define R128_DATE	"20010326"
+#define R128_DATE	"20010405"
 
 
 /* Return the width and height of the current color buffer.
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_ioctl.c b/xc/lib/GL/mesa/src/drv/r128/r128_ioctl.c
index 51a68f53e..426425583 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_ioctl.c
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_ioctl.c
@@ -316,6 +316,11 @@ static int r128WaitForFrameCompletion( r128ContextPtr rmesa )
 #else
       frame = INREG( R128_LAST_FRAME_REG );
 #endif
+
+      if ( 0 )
+	 fprintf( stderr, " last=0x%08x frame=0x%08x\n",
+		  rmesa->sarea->last_frame, frame );
+
       if ( rmesa->sarea->last_frame - frame <= R128_MAX_OUTSTANDING ) {
 	 break;
       }
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_tex.c b/xc/lib/GL/mesa/src/drv/r128/r128_tex.c
index f518720f5..efdd025f4 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_tex.c
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_tex.c
@@ -30,6 +30,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Authors:
  *   Gareth Hughes <gareth@valinux.com>
  *   Kevin E. Martin <martin@valinux.com>
+ *   Michel Dänzer <michdaen@iiic.ethz.ch>
  *
  */
 
@@ -572,8 +573,8 @@ static void r128ConvertTexture16bpp( CARD32 *dst,
       for ( i = 0 ; i < height ; i++ ) {
 	 src = (CARD8 *)image->Data + ((y + i) * pitch + x) * 4;
 	 for ( j = width >> 1 ; j ; j-- ) {
-	    *dst++ = ((R128PACKCOLOR4444( src[0], src[1], src[2], src[3] )) |
-		      (R128PACKCOLOR4444( src[4], src[5], src[6], src[7] ) << 16));
+	    *dst++ = R128PACKCOLORS4444( src[0], src[1], src[2], src[3],
+					 src[4], src[5], src[6], src[7] );
 	    src += 8;
 	 }
       }
@@ -583,8 +584,8 @@ static void r128ConvertTexture16bpp( CARD32 *dst,
       for ( i = 0 ; i < height ; i++ ) {
 	 src = (CARD8 *)image->Data + ((y + i) * pitch + x) * 3;
 	 for ( j = width >> 1 ; j ; j-- ) {
-	    *dst++ = ((R128PACKCOLOR565( src[0], src[1], src[2] )) |
-		      (R128PACKCOLOR565( src[3], src[4], src[5] ) << 16));
+	    *dst++ = R128PACKCOLORS565( src[0], src[1], src[2],
+					src[3], src[4], src[5] );
 	    src += 6;
 	 }
       }
@@ -594,8 +595,8 @@ static void r128ConvertTexture16bpp( CARD32 *dst,
       for ( i = 0 ; i < height ; i++ ) {
 	 src = (CARD8 *)image->Data + ((y + i) * pitch + x);
 	 for ( j = width >> 1 ; j ; j-- ) {
-	    *dst++ = ((R128PACKCOLOR4444( 0xff, 0xff, 0xff, src[0] )) |
-		      (R128PACKCOLOR4444( 0xff, 0xff, 0xff, src[1] ) << 16));
+	    *dst++ = R128PACKCOLORS4444( 0xff, 0xff, 0xff, src[0],
+					 0xff, 0xff, 0xff, src[1] );
 	    src += 2;
 	 }
       }
@@ -605,8 +606,8 @@ static void r128ConvertTexture16bpp( CARD32 *dst,
       for ( i = 0 ; i < height ; i++ ) {
 	 src = (CARD8 *)image->Data + ((y + i) * pitch + x) * 2;
 	 for ( j = width >> 1 ; j ; j-- ) {
-	    *dst++ = ((R128PACKCOLOR4444( src[0], src[0], src[0], src[1] )) |
-		      (R128PACKCOLOR4444( src[2], src[2], src[2], src[3] ) << 16));
+	    *dst++ = R128PACKCOLORS4444( src[0], src[0], src[0], src[1],
+					 src[2], src[2], src[2], src[3] );
 	    src += 4;
 	 }
       }
@@ -877,7 +878,8 @@ int r128UploadTexImages( r128ContextPtr rmesa, r128TexObjPtr t )
       t->memBlock = mmAllocMem( rmesa->texHeap[heap], t->totalSize, 12, 0 );
 
       /* Try AGP before kicking anything out of local mem */
-      if ( !t->memBlock && heap == R128_CARD_HEAP ) {
+      if ( !rmesa->r128Screen->IsPCI &&
+	   !t->memBlock && heap == R128_CARD_HEAP ) {
 	 t->memBlock = mmAllocMem( rmesa->texHeap[R128_AGP_HEAP],
 				   t->totalSize, 12, 0 );
 
@@ -1572,6 +1574,9 @@ static void r128DDTexSubImage( GLcontext *ctx, GLenum target,
    if ( t ) {
       if ( t->bound ) FLUSH_BATCH( rmesa );
 
+#if 0
+      /* FIXME: Only upload textures if we already have space in the heap.
+       */
       LOCK_HARDWARE( rmesa );
       r128UploadSubImage( rmesa, t, level,
 			  xoffset, yoffset, width, height );
@@ -1579,6 +1584,9 @@ static void r128DDTexSubImage( GLcontext *ctx, GLenum target,
 
       /* Update the context state */
       rmesa->setup.tex_cntl_c |= R128_TEX_CACHE_FLUSH;
+#else
+      r128DestroyTexObj( rmesa, t );
+#endif
 
       rmesa->new_state |= R128_NEW_TEXTURE;
    }
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_tex.h b/xc/lib/GL/mesa/src/drv/r128/r128_tex.h
index e1f67a56d..a0090f389 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_tex.h
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_tex.h
@@ -74,6 +74,23 @@ extern void r128DDInitTextureFuncs( GLcontext *ctx );
 #define R128PACKCOLOR4444( r, g, b, a )					\
    ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
 
+#if __BYTE_ORDER == __LITTLE_ENDIAN
+#define R128PACKCOLORS565( r0, g0, b0, r1, g1, b1 )			\
+		((R128PACKCOLOR565( r0, g0, b0 )) |			\
+		 (R128PACKCOLOR565( r1, g1, b1 ) << 16))
+#define R128PACKCOLORS4444( r0, g0, b0, a0, r1, g1, b1, a1 )		\
+		((R128PACKCOLOR4444( r0, g0, b0, a0 )) |		\
+		 (R128PACKCOLOR4444( r1, g1, b1, a1 ) << 16))
+#else
+#define R128PACKCOLORS565( r0, g0, b0, r1, g1, b1 )			\
+		((R128PACKCOLOR565( r1, g1, b1 )) |			\
+		 (R128PACKCOLOR565( r0, g0, b0 ) << 16))
+#define R128PACKCOLORS4444( r0, g0, b0, a0, r1, g1, b1, a1 )		\
+		((R128PACKCOLOR4444( r1, g1, b1, a1 )) |		\
+		 (R128PACKCOLOR4444( r0, g0, b0, a0 ) << 16))
+
+#endif
+
 static __inline__ CARD32 r128PackColor( GLuint cpp,
 					GLubyte r, GLubyte g,
 					GLubyte b, GLubyte a )
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_tris.c b/xc/lib/GL/mesa/src/drv/r128/r128_tris.c
index 4d83e27e0..52d1ad032 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_tris.c
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_tris.c
@@ -30,6 +30,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Authors:
  *   Gareth Hughes <gareth@valinux.com>
  *   Kevin E. Martin <martin@valinux.com>
+ *   Michel Dänzer <michdaen@iiic.ethz.ch>
  *
  */
 
@@ -49,6 +50,7 @@ static struct {
    quad_func		quad;
 } rast_tab[R128_MAX_TRIFUNC];
 
+#if __BYTE_ORDER ==  __LITTLE_ENDIAN
 #define R128_COLOR( to, from )						\
 do {									\
    (to)[0] = (from)[2];							\
@@ -56,6 +58,15 @@ do {									\
    (to)[2] = (from)[0];							\
    (to)[3] = (from)[3];							\
 } while (0)
+#else
+#define R128_COLOR( to, from )						\
+do {									\
+   (to)[0] = (from)[3];							\
+   (to)[1] = (from)[0];							\
+   (to)[2] = (from)[1];							\
+   (to)[3] = (from)[2];							\
+} while (0)
+#endif
 
 
 static void r128_null_quad( GLcontext *ctx, GLuint v0,
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_tris.h b/xc/lib/GL/mesa/src/drv/r128/r128_tris.h
index 79b3fb69e..320ace740 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_tris.h
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_tris.h
@@ -56,7 +56,7 @@ static __inline void r128_draw_triangle( r128ContextPtr rmesa,
 					 r128VertexPtr v1,
 					 r128VertexPtr v2 )
 {
-   int vertsize = rmesa->vertsize;
+   GLuint vertsize = rmesa->vertsize;
    CARD32 *vb = r128AllocVerticesInline( rmesa, 3 );
    int j;
 
@@ -80,15 +80,15 @@ static __inline void r128_draw_triangle( r128ContextPtr rmesa,
 			 : "memory" );
 #else
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v0->ui[j];
+      LE32_OUT( vb[j], v0->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v1->ui[j];
+      LE32_OUT( vb[j], v1->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v2->ui[j];
+      LE32_OUT( vb[j], v2->ui[j] );
 #endif
 }
 
@@ -98,7 +98,7 @@ static __inline void r128_draw_quad( r128ContextPtr rmesa,
 				     r128VertexPtr v2,
 				     r128VertexPtr v3 )
 {
-   int vertsize = rmesa->vertsize;
+   GLuint vertsize = rmesa->vertsize;
    CARD32 *vb = r128AllocVerticesInline( rmesa, 6 );
    int j;
 
@@ -134,87 +134,110 @@ static __inline void r128_draw_quad( r128ContextPtr rmesa,
 			 : "memory" );
 #else
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v0->ui[j];
+      LE32_OUT( vb[j], v0->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v1->ui[j];
+      LE32_OUT( vb[j], v1->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v3->ui[j];
+      LE32_OUT( vb[j], v3->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v1->ui[j];
+      LE32_OUT( vb[j], v1->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v2->ui[j];
+      LE32_OUT( vb[j], v2->ui[j] );
 
    vb += vertsize;
    for ( j = 0 ; j < vertsize ; j++ )
-      vb[j] = v3->ui[j];
+      LE32_OUT( vb[j], v3->ui[j] );
 #endif
 }
 
 static __inline void r128_draw_line( r128ContextPtr rmesa,
 				     r128VertexPtr tmp0,
 				     r128VertexPtr tmp1,
-				     float width )
+				     GLfloat width )
 {
 #if 1
-   int vertsize = rmesa->vertsize;
+   GLuint vertsize = rmesa->vertsize;
    CARD32 *vb = r128AllocVerticesInline( rmesa, 6 );
-   float dx, dy, ix, iy;
-   int j;
+   GLfloat hw, dx, dy, ix, iy;
+   GLuint j;
+   GLfloat x0 = tmp0->v.x;
+   GLfloat y0 = tmp0->v.y;
+   GLfloat x1 = tmp1->v.x;
+   GLfloat y1 = tmp1->v.y;
+
+   hw = 0.5F * width;
+   if (hw > 0.1F && hw < 0.5F) {
+      hw = 0.5F;
+   }
 
+   /* adjust vertices depending on line direction */
    dx = tmp0->v.x - tmp1->v.x;
    dy = tmp0->v.y - tmp1->v.y;
-
-   ix = width * .5; iy = 0;
-
-   if ((ix<.5) && (ix>0.1)) ix = .5; /* I want to see lines with width
-					0.5 also */
-
    if (dx * dx > dy * dy) {
-      iy = ix; ix = 0;
+      /* X-major line */
+      ix = 0.0F;
+      iy = hw;
+      if (x1 < x0) {
+         x0 += 0.5F;
+         x1 += 0.5F;
+      }
+      y0 -= 0.5F;
+      y1 -= 0.5F;
+   }
+   else {
+      /* Y-major line */
+      ix = hw;
+      iy = 0.0F;
+      if (y1 > y0) {
+         y0 -= 0.5F;
+         y1 -= 0.5F;
+      }
+      x0 += 0.5F;
+      x1 += 0.5F;
    }
 
-   *(float *)&vb[0] = tmp0->v.x - ix;
-   *(float *)&vb[1] = tmp0->v.y - iy;
+   LE32_OUT_FLOAT( vb[0], x0 - ix );
+   LE32_OUT_FLOAT( vb[1], y0 - iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp0->ui[j];
+      LE32_OUT( vb[j], tmp0->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp1->v.x + ix;
-   *(float *)&vb[1] = tmp1->v.y + iy;
+   LE32_OUT_FLOAT( vb[0], x1 + ix );
+   LE32_OUT_FLOAT( vb[1], y1 + iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp1->ui[j];
+      LE32_OUT( vb[j], tmp1->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp0->v.x + ix;
-   *(float *)&vb[1] = tmp0->v.y + iy;
+   LE32_OUT_FLOAT( vb[0], x0 + ix );
+   LE32_OUT_FLOAT( vb[1], y0 + iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp0->ui[j];
+      LE32_OUT( vb[j], tmp0->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp0->v.x - ix;
-   *(float *)&vb[1] = tmp0->v.y - iy;
+   LE32_OUT_FLOAT( vb[0], x0 - ix );
+   LE32_OUT_FLOAT( vb[1], y0 - iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp0->ui[j];
+      LE32_OUT( vb[j], tmp0->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp1->v.x - ix;
-   *(float *)&vb[1] = tmp1->v.y - iy;
+   LE32_OUT_FLOAT( vb[0], x1 - ix );
+   LE32_OUT_FLOAT( vb[1], y1 - iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp1->ui[j];
+      LE32_OUT( vb[j], tmp1->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp1->v.x + ix;
-   *(float *)&vb[1] = tmp1->v.y + iy;
+   LE32_OUT_FLOAT( vb[0], x1 + ix );
+   LE32_OUT_FLOAT( vb[1], y1 + iy );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp1->ui[j];
+      LE32_OUT( vb[j], tmp1->ui[j] );
 
 #else
 
@@ -248,47 +271,50 @@ static __inline void r128_draw_line( r128ContextPtr rmesa,
 }
 
 static __inline void r128_draw_point( r128ContextPtr rmesa,
-				      r128VertexPtr tmp, float sz )
+				      r128VertexPtr tmp, GLfloat sz )
 {
 #if 1
-   int vertsize = rmesa->vertsize;
+   GLuint vertsize = rmesa->vertsize;
    CARD32 *vb = r128AllocVerticesInline( rmesa, 6 );
    int j;
+   const float x = tmp->v.x + PNT_X_OFFSET;
+   const float y = tmp->v.y + PNT_Y_OFFSET;
 
-   *(float *)&vb[0] = tmp->v.x - sz;
-   *(float *)&vb[1] = tmp->v.y - sz;
+   LE32_OUT_FLOAT( vb[0], x - sz );
+   LE32_OUT_FLOAT( vb[1], y - sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp->v.x + sz;
-   *(float *)&vb[1] = tmp->v.y - sz;
+   LE32_OUT_FLOAT( vb[0], x + sz );
+   LE32_OUT_FLOAT( vb[1], y - sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp->v.x + sz;
-   *(float *)&vb[1] = tmp->v.y + sz;
+   LE32_OUT_FLOAT( vb[0], x + sz );
+   LE32_OUT_FLOAT( vb[1], y + sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp->v.x + sz;
-   *(float *)&vb[1] = tmp->v.y + sz;
+   LE32_OUT_FLOAT( vb[0], x + sz );
+   LE32_OUT_FLOAT( vb[1], y + sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp->v.x - sz;
-   *(float *)&vb[1] = tmp->v.y + sz;
+   LE32_OUT_FLOAT( vb[0], x - sz );
+   LE32_OUT_FLOAT( vb[1], y + sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
    vb += vertsize;
 
-   *(float *)&vb[0] = tmp->v.x - sz;
-   *(float *)&vb[1] = tmp->v.y - sz;
+   LE32_OUT_FLOAT( vb[0], x - sz );
+   LE32_OUT_FLOAT( vb[1], y - sz );
    for (j = 2 ; j < vertsize ; j++)
-      vb[j] = tmp->ui[j];
+      LE32_OUT( vb[j], tmp->ui[j] );
+
 #else
 
    int vertsize = rmesa->vertsize;
diff --git a/xc/lib/GL/mesa/src/drv/r128/r128_vb.h b/xc/lib/GL/mesa/src/drv/r128/r128_vb.h
index 1a00d9eea..7bb314692 100644
--- a/xc/lib/GL/mesa/src/drv/r128/r128_vb.h
+++ b/xc/lib/GL/mesa/src/drv/r128/r128_vb.h
@@ -30,6 +30,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Authors:
  *   Gareth Hughes <gareth@valinux.com>
  *   Kevin E. Martin <martin@valinux.com>
+ *   Michel Dänzer <michdaen@iiic.ethz.ch>
  *
  */
 
@@ -38,13 +39,21 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #ifdef GLX_DIRECT_RENDERING
 
-/* FIXME: This is endian-specific */
+#if __BYTE_ORDER == __LITTLE_ENDIAN
 typedef struct {
    GLubyte	blue;
    GLubyte	green;
    GLubyte	red;
    GLubyte	alpha;
 } r128_color_t;
+#else
+typedef struct {
+   GLubyte	alpha;
+   GLubyte	red;
+   GLubyte	green;
+   GLubyte	blue;
+} r128_color_t;
+#endif
 
 /* The vertex structure.  The final tu1/tv1 values are only used in
  * multitexture modes, and the rhw2 value is currently never used.
diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c
index 4d5c6a5cf..b94669ab6 100644
--- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c
+++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c
@@ -871,11 +871,11 @@ static Bool RADEONDRIAgpInit(RADEONInfoPtr info, ScreenPtr pScreen)
     return TRUE;
 }
 
+#if 0
 /* Initialize the PCIGART state.  Request memory for use in PCI space,
    and initialize the Radeon registers to point to that memory. */
 static Bool RADEONDRIPciInit(RADEONInfoPtr info, ScreenPtr pScreen)
 {
-    unsigned char *RADEONMMIO = info->MMIO;
     int ret;
     int flags;
 
@@ -974,6 +974,7 @@ static Bool RADEONDRIPciInit(RADEONInfoPtr info, ScreenPtr pScreen)
 
     return TRUE;
 }
+#endif
 
 /* Add a map for the MMIO registers that will be accessed by any
    DRI-based clients. */
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drmP.h b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drmP.h
index 274751331..085d1cb83 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drmP.h
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drmP.h
@@ -53,7 +53,7 @@
 #include <linux/sched.h>
 #include <linux/smp_lock.h>	/* For (un)lock_kernel */
 #include <linux/mm.h>
-#ifdef __alpha__
+#if defined(__alpha__) || defined(__powerpc__)
 #include <asm/pgtable.h> /* For pte_wrprotect */
 #endif
 #include <asm/io.h>
@@ -287,12 +287,43 @@ static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
 	return old;
 }
 
+#elif defined(__powerpc__)
+extern void __cmpxchg_called_with_bad_pointer(void);
+static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
+                                      unsigned long new, int size)
+{
+	unsigned long prev;
+
+	switch (size) {
+	case 4:
+		__asm__ __volatile__(
+			"sync;"
+			"0:    lwarx %0,0,%1 ;"
+			"      cmpl 0,%0,%3;"
+			"      bne 1f;"
+			"      stwcx. %2,0,%1;"
+			"      bne- 0b;"
+			"1:    "
+			"sync;"
+			: "=&r"(prev)
+			: "r"(ptr), "r"(new), "r"(old)
+			: "cr0", "memory");
+		return prev;
+	}
+	__cmpxchg_called_with_bad_pointer();
+	return old;
+}
+
+#endif /* i386, powerpc & alpha */
+
+#ifndef __alpha__
 #define cmpxchg(ptr,o,n)						\
   ((__typeof__(*(ptr)))__cmpxchg((ptr),(unsigned long)(o),		\
 				 (unsigned long)(n),sizeof(*(ptr))))
-#endif /* i386 & alpha */
 #endif
 
+#endif /* !__HAVE_ARCH_CMPXCHG */
+
 				/* Macros to make printk easier */
 #define DRM_ERROR(fmt, arg...) \
 	printk(KERN_ERR "[" DRM_NAME ":" __FUNCTION__ "] *ERROR* " fmt , ##arg)
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drm_vm.h b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drm_vm.h
index 26906aee2..771c11bd2 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drm_vm.h
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/drm_vm.h
@@ -424,7 +424,10 @@ int DRM(mmap)(struct file *filp, struct vm_area_struct *vma)
 			}
 #elif defined(__ia64__)
 			if (map->type != _DRM_AGP)
-				vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
+				vma->vm_page_prot =
+					pgprot_writecombine(vma->vm_page_prot);
+#elif defined(__powerpc__)
+			pgprot_val(vma->vm_page_prot) |= _PAGE_NO_CACHE | _PAGE_GUARDED;
 #endif
 			vma->vm_flags |= VM_IO;	/* not in core dump */
 		}
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_cce.c b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_cce.c
index da44a496a..1ced05fe9 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_cce.c
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_cce.c
@@ -210,7 +210,7 @@ int r128_do_cce_idle( drm_r128_private_t *dev_priv )
 	int i;
 
 	for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) {
-		if ( *dev_priv->ring.head == dev_priv->ring.tail ) {
+		if ( GET_RING_HEAD( &dev_priv->ring ) == dev_priv->ring.tail ) {
 			int pm4stat = R128_READ( R128_PM4_STAT );
 			if ( ( (pm4stat & R128_PM4_FIFOCNT_MASK) >=
 			       dev_priv->cce_fifo_size ) &&
@@ -251,7 +251,7 @@ static void r128_do_cce_reset( drm_r128_private_t *dev_priv )
 {
 	R128_WRITE( R128_PM4_BUFFER_DL_WPTR, 0 );
 	R128_WRITE( R128_PM4_BUFFER_DL_RPTR, 0 );
-	*dev_priv->ring.head = 0;
+	SET_RING_HEAD( &dev_priv->ring, 0 );
 	dev_priv->ring.tail = 0;
 }
 
@@ -332,7 +332,7 @@ static void r128_cce_init_ring_buffer( drm_device_t *dev )
 	R128_WRITE( R128_PM4_BUFFER_DL_RPTR, 0 );
 
 	/* DL_RPTR_ADDR is a physical address in AGP space. */
-	*dev_priv->ring.head = 0;
+	SET_RING_HEAD( &dev_priv->ring, 0 );
 
 	if ( !dev_priv->is_pci ) {
 		R128_WRITE( R128_PM4_BUFFER_DL_RPTR_ADDR,
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.c b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.c
index 7e736541c..584cb29c3 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.c
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.c
@@ -39,11 +39,11 @@
 
 #define DRIVER_NAME		"r128"
 #define DRIVER_DESC		"ATI Rage 128"
-#define DRIVER_DATE		"20010308"
+#define DRIVER_DATE		"20010405"
 
 #define DRIVER_MAJOR		2
 #define DRIVER_MINOR		1
-#define DRIVER_PATCHLEVEL	5
+#define DRIVER_PATCHLEVEL	6
 
 #define DRIVER_IOCTLS							    \
    [DRM_IOCTL_NR(DRM_IOCTL_DMA)]             = { r128_cce_buffers,  1, 0 }, \
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.h b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.h
index bf682998b..fbc692298 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.h
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_drv.h
@@ -28,11 +28,15 @@
  *    Rickard E. (Rik) Faith <faith@valinux.com>
  *    Kevin E. Martin <martin@valinux.com>
  *    Gareth Hughes <gareth@valinux.com>
+ *    Michel Dänzer <daenzerm@student.ethz.ch>
  */
 
 #ifndef __R128_DRV_H__
 #define __R128_DRV_H__
 
+#define GET_RING_HEAD( ring )		le32_to_cpu( *(ring)->head )
+#define SET_RING_HEAD( ring, val )	*(ring)->head = cpu_to_le32( val )
+
 typedef struct drm_r128_freelist {
    	unsigned int age;
    	drm_buf_t *buf;
@@ -136,7 +140,7 @@ extern int r128_wait_ring( drm_r128_private_t *dev_priv, int n );
 static inline void
 r128_update_ring_snapshot( drm_r128_ring_buffer_t *ring )
 {
-	ring->space = (*(volatile int *)ring->head - ring->tail) * sizeof(u32);
+	ring->space = (GET_RING_HEAD( ring ) - ring->tail) * sizeof(u32);
 	if ( ring->space <= 0 )
 		ring->space += ring->size;
 }
@@ -368,7 +372,7 @@ extern int r128_cce_indirect( struct inode *inode, struct file *filp,
 
 #define R128_LAST_FRAME_REG		R128_GUI_SCRATCH_REG0
 #define R128_LAST_DISPATCH_REG		R128_GUI_SCRATCH_REG1
-#define R128_MAX_VB_AGE			0xffffffff
+#define R128_MAX_VB_AGE			0x7fffffff
 #define R128_MAX_VB_VERTS		(0xffff)
 
 #define R128_RING_HIGH_MARK		128
@@ -389,8 +393,8 @@ static inline u32 _R128_READ(u32 *addr) {
 #define R128_WRITE(reg,val)	\
 	do { wmb(); R128_DEREF(reg) = val; } while (0)
 #else
-#define R128_READ(reg)		R128_DEREF( reg )
-#define R128_WRITE(reg,val)	do { R128_DEREF( reg ) = val; } while (0)
+#define R128_READ(reg)		le32_to_cpu( R128_DEREF( reg ) )
+#define R128_WRITE(reg,val)	do { R128_DEREF( reg ) = cpu_to_le32( val ); } while (0)
 #endif
 
 #define R128_DEREF8(reg)	*(volatile u8 *)R128_ADDR( reg )
@@ -525,7 +529,7 @@ do {									\
 		DRM_INFO( "   OUT_RING( 0x%08x ) at 0x%x\n",		\
 			   (unsigned int)(x), write );			\
 	}								\
-	ring[write++] = (x);						\
+	ring[write++] = cpu_to_le32( x );				\
 	write &= tail_mask;						\
 } while (0)
 
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_state.c b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_state.c
index 1006a77c1..7eafd72d6 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_state.c
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/r128_state.c
@@ -660,7 +660,7 @@ static void r128_cce_dispatch_indirect( drm_device_t *dev,
 			u32 *data = (u32 *)
 				((char *)dev_priv->buffers->handle
 				 + buf->offset + start);
-			data[dwords++] = R128_CCE_PACKET2;
+			data[dwords++] = cpu_to_le32( R128_CCE_PACKET2 );
 		}
 
 		buf_priv->dispatched = 1;
@@ -727,16 +727,21 @@ static void r128_cce_dispatch_indices( drm_device_t *dev,
 		data = (u32 *)((char *)dev_priv->buffers->handle
 			       + buf->offset + start);
 
-		data[0] = CCE_PACKET3( R128_3D_RNDR_GEN_INDX_PRIM, dwords-2 );
+		data[0] = cpu_to_le32( CCE_PACKET3( R128_3D_RNDR_GEN_INDX_PRIM,
+						    dwords-2 ) );
 
-		data[1] = offset;
-		data[2] = R128_MAX_VB_VERTS;
-		data[3] = format;
-		data[4] = (prim | R128_CCE_VC_CNTL_PRIM_WALK_IND |
-			   (count << 16));
+		data[1] = cpu_to_le32( offset );
+		data[2] = cpu_to_le32( R128_MAX_VB_VERTS );
+		data[3] = cpu_to_le32( format );
+		data[4] = cpu_to_le32( (prim | R128_CCE_VC_CNTL_PRIM_WALK_IND |
+					(count << 16)) );
 
 		if ( count & 0x1 ) {
+#ifdef __LITTLE_ENDIAN
 			data[dwords-1] &= 0x0000ffff;
+#else
+			data[dwords-1] &= 0xffff0000;
+#endif
 		}
 
 		do {
@@ -842,23 +847,23 @@ static int r128_cce_dispatch_blit( drm_device_t *dev,
 
 	data = (u32 *)((char *)dev_priv->buffers->handle + buf->offset);
 
-	data[0] = CCE_PACKET3( R128_CNTL_HOSTDATA_BLT, dwords + 6 );
-	data[1] = (R128_GMC_DST_PITCH_OFFSET_CNTL |
-		   R128_GMC_BRUSH_NONE |
-		   (blit->format << 8) |
-		   R128_GMC_SRC_DATATYPE_COLOR |
-		   R128_ROP3_S |
-		   R128_DP_SRC_SOURCE_HOST_DATA |
-		   R128_GMC_CLR_CMP_CNTL_DIS |
-		   R128_GMC_AUX_CLIP_DIS |
-		   R128_GMC_WR_MSK_DIS);
-
-	data[2] = (blit->pitch << 21) | (blit->offset >> 5);
-	data[3] = 0xffffffff;
-	data[4] = 0xffffffff;
-	data[5] = (blit->y << 16) | blit->x;
-	data[6] = (blit->height << 16) | blit->width;
-	data[7] = dwords;
+	data[0] = cpu_to_le32( CCE_PACKET3( R128_CNTL_HOSTDATA_BLT, dwords + 6 ) );
+	data[1] = cpu_to_le32( (R128_GMC_DST_PITCH_OFFSET_CNTL |
+				R128_GMC_BRUSH_NONE |
+				(blit->format << 8) |
+				R128_GMC_SRC_DATATYPE_COLOR |
+				R128_ROP3_S |
+				R128_DP_SRC_SOURCE_HOST_DATA |
+				R128_GMC_CLR_CMP_CNTL_DIS |
+				R128_GMC_AUX_CLIP_DIS |
+				R128_GMC_WR_MSK_DIS) );
+
+	data[2] = cpu_to_le32( (blit->pitch << 21) | (blit->offset >> 5) );
+	data[3] = cpu_to_le32( 0xffffffff );
+	data[4] = cpu_to_le32( 0xffffffff );
+	data[5] = cpu_to_le32( (blit->y << 16) | blit->x );
+	data[6] = cpu_to_le32( (blit->height << 16) | blit->width );
+	data[7] = cpu_to_le32( dwords );
 
 	buf->used = (dwords + 8) * sizeof(u32);
 
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/radeon_drv.c b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/radeon_drv.c
index 3791c5ecb..59497acae 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/radeon_drv.c
+++ b/xc/programs/Xserver/hw/xfree86/os-support/linux/drm/kernel/radeon_drv.c
@@ -37,11 +37,11 @@
 
 #define DRIVER_NAME		"radeon"
 #define DRIVER_DESC		"ATI Radeon"
-#define DRIVER_DATE		"20010308"
+#define DRIVER_DATE		"20010405"
 
 #define DRIVER_MAJOR		1
 #define DRIVER_MINOR		0
-#define DRIVER_PATCHLEVEL	1
+#define DRIVER_PATCHLEVEL	2
 
 #define DRIVER_IOCTLS							     \
  [DRM_IOCTL_NR(DRM_IOCTL_DMA)]               = { radeon_cp_buffers,  1, 0 }, \
diff --git a/xc/programs/Xserver/hw/xfree86/os-support/xf86drm.h b/xc/programs/Xserver/hw/xfree86/os-support/xf86drm.h
index e3702c2cf..7a07c7e0f 100644
--- a/xc/programs/Xserver/hw/xfree86/os-support/xf86drm.h
+++ b/xc/programs/Xserver/hw/xfree86/os-support/xf86drm.h
@@ -315,6 +315,25 @@ do {	register unsigned int __old __asm("o0");		\
 	} while (0)
 
 #endif
+
+#elif defined(__powerpc__)
+
+#define DRM_CAS(lock,old,new,__ret)			\
+	do {						\
+		__asm__ __volatile__(			\
+			"sync;"				\
+			"0:    lwarx %0,0,%1;"		\
+			"      xor. %0,%3,%0;"		\
+			"      bne 1f;"			\
+			"      stwcx. %2,0,%1;"		\
+			"      bne- 0b;"		\
+			"1:    "			\
+			"sync;"				\
+		: "=&r"(__ret)				\
+		: "r"(lock), "r"(new), "r"(old)		\
+		: "cr0", "memory");			\
+	} while (0)
+
 #endif /* architecture */
 #endif /* __GNUC__ >= 2 */