Merge branch 'i915-unification' of git+ssh://people.freedesktop.org/~anholt/mesa into i915-unification

64 files changed, 1658 insertions, 781 deletions

diff --git a/src/mesa/drivers/dri/common/dri_bufmgr_fake.c b/src/mesa/drivers/dri/common/dri_bufmgr_fake.c
index c6202d969cc..e08b5b3c7fb 100644
--- a/src/mesa/drivers/dri/common/dri_bufmgr_fake.c
+++ b/src/mesa/drivers/dri/common/dri_bufmgr_fake.c
@@ -41,7 +41,11 @@
 #include "mm.h"
 #include "imports.h"
 
+#if 0
+#define DBG(...) _mesa_printf(__VA_ARGS__)
+#else
 #define DBG(...)
+#endif
 
 /* Internal flags:
  */
@@ -570,6 +574,9 @@ dri_fake_bo_alloc(dri_bufmgr *bufmgr, const char *name,
    bo_fake->flags = 0;
    bo_fake->is_static = GL_FALSE;
 
+   DBG("drm_bo_alloc: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
    return &bo_fake->bo;
 }
 
@@ -592,10 +599,14 @@ dri_fake_bo_alloc_static(dri_bufmgr *bufmgr, const char *name,
    bo_fake->bo.virtual = virtual;
    bo_fake->bo.bufmgr = bufmgr;
    bo_fake->refcount = 1;
+   bo_fake->id = ++bufmgr_fake->buf_nr;
    bo_fake->name = name;
    bo_fake->flags = DRM_BO_FLAG_NO_EVICT | DRM_BO_FLAG_NO_MOVE;
    bo_fake->is_static = GL_TRUE;
 
+   DBG("drm_bo_alloc_static: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
    return &bo_fake->bo;
 }
 
@@ -648,7 +659,8 @@ dri_fake_bo_map(dri_bo *bo, GLboolean write_enable)
 
    _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
    {
-      DBG("bmMapBuffer %d\n", bo_fake->id);
+      DBG("drm_bo_map: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+	  bo_fake->bo.size / 1024);
 
       if (bo->virtual != NULL) {
 	 _mesa_printf("%s: already mapped\n", __FUNCTION__);
@@ -698,6 +710,9 @@ dri_fake_bo_unmap(dri_bo *bo)
    if (bo == NULL)
       return 0;
 
+   DBG("drm_bo_unmap: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
    bo->virtual = NULL;
 
    return 0;
@@ -713,22 +728,32 @@ dri_fake_bo_validate(dri_bo *bo, unsigned int flags)
     * different flags.  See drmAddValidateItem().
     */
 
+   DBG("drm_bo_validate: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
    bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
 
    _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
    {
+      if (bo_fake->is_static) {
+	 /* Add it to the needs-fence list */
+	 bufmgr_fake->need_fence = 1;
+	 _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+	 return 0;
+      }
+
       /* Allocate the card memory */
       if (!bo_fake->block && !evict_and_alloc_block(bo)) {
 	 bufmgr_fake->fail = 1;
 	 _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+	 DBG("Failed to validate buf %d:%s\n", bo_fake->id, bo_fake->name);
 	 return -1;
       }
 
       assert(bo_fake->block);
       assert(bo_fake->block->bo == &bo_fake->bo);
 
-      DBG("Add buf %d (block %p, dirty %d) to referenced list\n",
-	  bo_fake->id, bo_fake->block, bo_fake->dirty);
+      DBG("Add buf %d:%s (block %p, dirty %d) to referenced list\n",
+	  bo_fake->id, bo_fake->name, bo_fake->block, bo_fake->dirty);
 
       move_to_tail(&bufmgr_fake->referenced, bo_fake->block);
       bo_fake->block->referenced = 1;
@@ -737,8 +762,8 @@ dri_fake_bo_validate(dri_bo *bo, unsigned int flags)
 
       /* Upload the buffer contents if necessary */
       if (bo_fake->dirty) {
-	 DBG("Upload dirty buf %d (%s) sz %d offset 0x%x\n", bo_fake->id,
-	     bo_fake->name, bo->size, block->mem->ofs);
+	 DBG("Upload dirty buf %d:%s, sz %d offset 0x%x\n", bo_fake->id,
+	     bo_fake->name, bo->size, bo_fake->block->mem->ofs);
 
 	 assert(!(bo_fake->flags &
 		  (BM_NO_BACKING_STORE|DRM_BO_FLAG_NO_EVICT)));
@@ -786,6 +811,8 @@ dri_fake_fence_validated(dri_bufmgr *bufmgr, const char *name,
    fence_blocks(bufmgr_fake, cookie);
    _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
 
+   DBG("drm_fence_validated: 0x%08x cookie\n", fence_fake->fence_cookie);
+
    return &fence_fake->fence;
 }
 
@@ -824,6 +851,8 @@ dri_fake_fence_wait(dri_fence *fence)
    dri_fence_fake *fence_fake = (dri_fence_fake *)fence;
    dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)fence->bufmgr;
 
+   DBG("drm_fence_wait: 0x%08x cookie\n", fence_fake->fence_cookie);
+
    _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
    _fence_wait_internal(bufmgr_fake, fence_fake->fence_cookie);
    _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c b/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c
index 64248a1d871..fd432ba3f72 100644
--- a/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c
+++ b/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c
@@ -150,7 +150,7 @@ dri_ttm_alloc_static(dri_bufmgr *bufmgr, const char *name,
     * pass all of the allocation class flags.
     */
    flags = location_mask | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE |
-      DRM_BO_FLAG_EXE | DRM_BO_FLAG_NO_EVICT | DRM_BO_FLAG_NO_MOVE;
+      DRM_BO_FLAG_EXE | DRM_BO_FLAG_NO_MOVE;
    /* No hints we want to use. */
    hint = 0;
 
@@ -362,6 +362,7 @@ dri_bufmgr_ttm_init(int fd, unsigned int fence_type,
 		    unsigned int fence_type_flush)
 {
    dri_bufmgr_ttm *bufmgr_ttm;
+   dri_bo *test_alloc;
 
    bufmgr_ttm = malloc(sizeof(*bufmgr_ttm));
    bufmgr_ttm->fd = fd;
@@ -381,5 +382,17 @@ dri_bufmgr_ttm_init(int fd, unsigned int fence_type,
    bufmgr_ttm->bufmgr.fence_unreference = dri_ttm_fence_unreference;
    bufmgr_ttm->bufmgr.fence_wait = dri_ttm_fence_wait;
 
+   /* Attempt an allocation to make sure that the DRM was actually set up for
+    * TTM.
+    */
+   test_alloc = dri_bo_alloc((dri_bufmgr *)bufmgr_ttm, "test allocation",
+     4096, 4096, DRM_BO_FLAG_MEM_LOCAL | DRM_BO_FLAG_MEM_TT);
+   if (test_alloc == NULL) {
+      _glthread_DESTROY_MUTEX(bufmgr_ttm->mutex);
+      free(bufmgr_ttm);
+      return NULL;
+   }
+   dri_bo_unreference(test_alloc);
+
    return &bufmgr_ttm->bufmgr;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index 9f0c9491b22..a19d4b65840 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -491,12 +491,19 @@ static void i915SetTexImages( i915ContextPtr i915,
       abort();
    }
 
-
-   if (i915->intel.intelScreen->deviceID == PCI_CHIP_I945_G ||
-       i915->intel.intelScreen->deviceID == PCI_CHIP_I945_GM)
-      i945LayoutTextureImages( i915, tObj );	 
-   else
-      i915LayoutTextureImages( i915, tObj );
+   switch (i915->intel.intelScreen->deviceID) {
+   case PCI_CHIP_I945_G:
+   case PCI_CHIP_I945_GM:
+   case PCI_CHIP_I945_GME:
+   case PCI_CHIP_G33_G:
+   case PCI_CHIP_Q33_G:
+   case PCI_CHIP_Q35_G:
+       i945LayoutTextureImages( i915, tObj );
+       break;
+   default:
+       i915LayoutTextureImages( i915, tObj );
+       break;
+   }
 
    t->Setup[I915_TEXREG_MS3] = 
       (((tObj->Image[0][t->intel.base.firstLevel]->Height - 1) << MS3_HEIGHT_SHIFT) |
diff --git a/src/mesa/drivers/dri/i915/intel_context.c b/src/mesa/drivers/dri/i915/intel_context.c
index e747fc6991b..11c23f24a1b 100644
--- a/src/mesa/drivers/dri/i915/intel_context.c
+++ b/src/mesa/drivers/dri/i915/intel_context.c
@@ -123,6 +123,14 @@ const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
 	 chipset = "Intel(R) 945G"; break;
       case PCI_CHIP_I945_GM:
 	 chipset = "Intel(R) 945GM"; break;
+      case PCI_CHIP_I945_GME:
+	 chipset = "Intel(R) 945GME"; break;
+      case PCI_CHIP_G33_G:
+	 chipset = "Intel(R) G33"; break;
+      case PCI_CHIP_Q35_G:
+	 chipset = "Intel(R) Q35"; break;
+      case PCI_CHIP_Q33_G:
+	 chipset = "Intel(R) Q33"; break;
       default:
 	 chipset = "Unknown Intel Chipset"; break;
       }
diff --git a/src/mesa/drivers/dri/i915/intel_context.h b/src/mesa/drivers/dri/i915/intel_context.h
index c48b074cc51..3b50107d73f 100644
--- a/src/mesa/drivers/dri/i915/intel_context.h
+++ b/src/mesa/drivers/dri/i915/intel_context.h
@@ -454,6 +454,10 @@ extern int INTEL_DEBUG;
 #define PCI_CHIP_I915_GM		0x2592
 #define PCI_CHIP_I945_G			0x2772
 #define PCI_CHIP_I945_GM		0x27A2
+#define PCI_CHIP_I945_GME		0x27AE
+#define PCI_CHIP_G33_G			0x29C2
+#define PCI_CHIP_Q35_G			0x29B2
+#define PCI_CHIP_Q33_G			0x29D2
 
 
 /* ================================================================
diff --git a/src/mesa/drivers/dri/i915/intel_pixel.c b/src/mesa/drivers/dri/i915/intel_pixel.c
index 535cbfcb26e..d175870a0c5 100644
--- a/src/mesa/drivers/dri/i915/intel_pixel.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel.c
@@ -439,10 +439,26 @@ intelDrawPixels( GLcontext *ctx,
    if (INTEL_DEBUG & DEBUG_PIXEL)
       fprintf(stderr, "%s\n", __FUNCTION__);
 
-   if (!intelTryDrawPixels( ctx, x, y, width, height, format, type,
-                            unpack, pixels ))
+   if (intelTryDrawPixels( ctx, x, y, width, height, format, type,
+                           unpack, pixels ))
+      return;
+
+   if (ctx->FragmentProgram._Current == ctx->FragmentProgram._TexEnvProgram) {
+      /*
+       * We don't want the i915 texenv program to be applied to DrawPixels.
+       * This is really just a performance optimization (mesa will other-
+       * wise happily run the fragment program on each pixel in the image).
+       */
+      struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
+      ctx->FragmentProgram._Current = NULL;
+      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
+                          unpack, pixels );
+      ctx->FragmentProgram._Current = fpSave;
+   }
+   else {
       _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-			  unpack, pixels );
+                          unpack, pixels );
+   }
 }
 
 
diff --git a/src/mesa/drivers/dri/i915/intel_screen.c b/src/mesa/drivers/dri/i915/intel_screen.c
index 67e176a1c6f..ca8610b4965 100644
--- a/src/mesa/drivers/dri/i915/intel_screen.c
+++ b/src/mesa/drivers/dri/i915/intel_screen.c
@@ -514,6 +514,10 @@ static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
    case PCI_CHIP_I915_GM:
    case PCI_CHIP_I945_G:
    case PCI_CHIP_I945_GM:
+   case PCI_CHIP_I945_GME:
+   case PCI_CHIP_G33_G:
+   case PCI_CHIP_Q35_G:
+   case PCI_CHIP_Q33_G:
       return i915CreateContext( mesaVis, driContextPriv, 
 			       sharedContextPrivate );
  
diff --git a/src/mesa/drivers/dri/i915/intel_tex.c b/src/mesa/drivers/dri/i915/intel_tex.c
index 98ddc796724..5bd280652af 100644
--- a/src/mesa/drivers/dri/i915/intel_tex.c
+++ b/src/mesa/drivers/dri/i915/intel_tex.c
@@ -677,7 +677,11 @@ static void intelUploadTexImage( intelContextPtr intel,
    /* Time for another vtbl entry:
     */
    else if (intel->intelScreen->deviceID == PCI_CHIP_I945_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_I945_GM) {
+            intel->intelScreen->deviceID == PCI_CHIP_I945_GM ||
+            intel->intelScreen->deviceID == PCI_CHIP_I945_GME ||
+            intel->intelScreen->deviceID == PCI_CHIP_G33_G ||
+            intel->intelScreen->deviceID == PCI_CHIP_Q33_G ||
+            intel->intelScreen->deviceID == PCI_CHIP_Q35_G) {
       GLuint row_len = image->Width * image->TexFormat->TexelBytes;
       GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
       GLubyte *src = (GLubyte *)image->Data;
diff --git a/src/mesa/drivers/dri/i915tex/Makefile b/src/mesa/drivers/dri/i915tex/Makefile
index c6a29149c77..eb704e1538b 100644
--- a/src/mesa/drivers/dri/i915tex/Makefile
+++ b/src/mesa/drivers/dri/i915tex/Makefile
@@ -33,6 +33,7 @@ DRIVER_SOURCES = \
 	intel_pixel_draw.c \
 	intel_buffers.c \
 	intel_blit.c \
+	i915_disasm.c \
 	i915_tex.c \
 	i915_texstate.c \
 	i915_context.c \
diff --git a/src/mesa/drivers/dri/i915tex/i830_vtbl.c b/src/mesa/drivers/dri/i915tex/i830_vtbl.c
index 441dc660acb..dc91af71815 100644
--- a/src/mesa/drivers/dri/i915tex/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915tex/i830_vtbl.c
@@ -487,11 +487,13 @@ i830_emit_state(struct intel_context *intel)
                       DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
                       state->tex_offset[i] | TM0S0_USE_FENCE);
          }
-         else {
-            assert(i == 0);
-            assert(state == &i830->meta);
-            OUT_BATCH(0);
-         }
+	 else if (state == &i830->meta) {
+	    assert(i == 0);
+	    OUT_BATCH(0);
+	 }
+	 else {
+	    OUT_BATCH(state->tex_offset[i]);
+	 }
 
          OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S1]);
          OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S2]);
diff --git a/src/mesa/drivers/dri/i915tex/i915_disasm.c b/src/mesa/drivers/dri/i915tex/i915_disasm.c
new file mode 100644
index 00000000000..6ac34a14804
--- /dev/null
+++ b/src/mesa/drivers/dri/i915tex/i915_disasm.c
@@ -0,0 +1,740 @@
+/* -*- c-basic-offset: 4 -*- */
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * 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 (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 NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
+ *
+ */
+
+#include <stdio.h>
+#include <stdarg.h>
+#include <inttypes.h>
+
+#include "i915_disasm.h"
+#include "i915_reg.h"
+
+#define BUFFER_FAIL(_count, _len, _name) do {			\
+    fprintf(out, "Buffer size too small in %s (%d < %d)\n",	\
+	    (_name), (_count), (_len));				\
+    (*failures)++;						\
+    return count;						\
+} while (0)
+
+static FILE *out;
+static uint32_t saved_s2 = 0, saved_s4 = 0;
+static char saved_s2_set = 0, saved_s4_set = 0;
+
+static float
+int_as_float(uint32_t intval)
+{
+    union intfloat {
+	uint32_t i;
+	float f;
+    } uval;
+
+    uval.i = intval;
+    return uval.f;
+}
+
+static void
+instr_out(uint32_t *data, uint32_t hw_offset, unsigned int index,
+	  char *fmt, ...)
+{
+    va_list va;
+
+    fprintf(out, "0x%08x: 0x%08x: ", hw_offset + index * 4, data[index]);
+    va_start(va, fmt);
+    vfprintf(out, fmt, va);
+    va_end(va);
+}
+
+
+static int
+decode_mi(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_mi[] = {
+	{ 0x08, 1, 1, "MI_ARB_ON_OFF" },
+	{ 0x0a, 1, 1, "MI_BATCH_BUFFER_END" },
+	{ 0x31, 2, 2, "MI_BATCH_BUFFER_START" },
+	{ 0x14, 3, 3, "MI_DISPLAY_BUFFER_INFO" },
+	{ 0x04, 1, 1, "MI_FLUSH" },
+	{ 0x22, 3, 3, "MI_LOAD_REGISTER_IMM" },
+	{ 0x13, 2, 2, "MI_LOAD_SCAN_LINES_EXCL" },
+	{ 0x12, 2, 2, "MI_LOAD_SCAN_LINES_INCL" },
+	{ 0x00, 1, 1, "MI_NOOP" },
+	{ 0x11, 2, 2, "MI_OVERLAY_FLIP" },
+	{ 0x07, 1, 1, "MI_REPORT_HEAD" },
+	{ 0x18, 2, 2, "MI_SET_CONTEXT" },
+	{ 0x20, 3, 4, "MI_STORE_DATA_IMM" },
+	{ 0x21, 3, 4, "MI_STORE_DATA_INDEX" },
+	{ 0x24, 3, 3, "MI_STORE_REGISTER_MEM" },
+	{ 0x02, 1, 1, "MI_USER_INTERRUPT" },
+	{ 0x03, 1, 1, "MI_WAIT_FOR_EVENT" },
+    };
+
+
+    for (opcode = 0; opcode < sizeof(opcodes_mi) / sizeof(opcodes_mi[0]);
+	 opcode++) {
+	if ((data[0] & 0x1e000000) >> 23 == opcodes_mi[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_mi[opcode].name);
+	    if (opcodes_mi[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 2;
+		if (len < opcodes_mi[opcode].min_len ||
+		    len > opcodes_mi[opcode].max_len)
+		{
+		    fprintf(out, "Bad length in %s\n",
+			    opcodes_mi[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_mi[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "MI UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_2d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_2d[] = {
+	{ 0x40, 5, 5, "COLOR_BLT" },
+	{ 0x43, 6, 6, "SRC_COPY_BLT" },
+	{ 0x01, 8, 8, "XY_SETUP_BLT" },
+	{ 0x11, 9, 9, "XY_SETUP_MONO_PATTERN_SL_BLT" },
+	{ 0x03, 3, 3, "XY_SETUP_CLIP_BLT" },
+	{ 0x24, 2, 2, "XY_PIXEL_BLT" },
+	{ 0x25, 3, 3, "XY_SCANLINES_BLT" },
+	{ 0x26, 4, 4, "Y_TEXT_BLT" },
+	{ 0x31, 5, 134, "XY_TEXT_IMMEDIATE_BLT" },
+	{ 0x50, 6, 6, "XY_COLOR_BLT" },
+	{ 0x51, 6, 6, "XY_PAT_BLT" },
+	{ 0x76, 8, 8, "XY_PAT_CHROMA_BLT" },
+	{ 0x72, 7, 135, "XY_PAT_BLT_IMMEDIATE" },
+	{ 0x77, 9, 137, "XY_PAT_CHROMA_BLT_IMMEDIATE" },
+	{ 0x52, 9, 9, "XY_MONO_PAT_BLT" },
+	{ 0x59, 7, 7, "XY_MONO_PAT_FIXED_BLT" },
+	{ 0x53, 8, 8, "XY_SRC_COPY_BLT" },
+	{ 0x54, 8, 8, "XY_MONO_SRC_COPY_BLT" },
+	{ 0x71, 9, 137, "XY_MONO_SRC_COPY_IMMEDIATE_BLT" },
+	{ 0x55, 9, 9, "XY_FULL_BLT" },
+	{ 0x55, 9, 137, "XY_FULL_IMMEDIATE_PATTERN_BLT" },
+	{ 0x56, 9, 9, "XY_FULL_MONO_SRC_BLT" },
+	{ 0x75, 10, 138, "XY_FULL_MONO_SRC_IMMEDIATE_PATTERN_BLT" },
+	{ 0x57, 12, 12, "XY_FULL_MONO_PATTERN_BLT" },
+	{ 0x58, 12, 12, "XY_FULL_MONO_PATTERN_MONO_SRC_BLT" },
+    };
+
+    for (opcode = 0; opcode < sizeof(opcodes_2d) / sizeof(opcodes_2d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1fc00000) >> 22 == opcodes_2d[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_2d[opcode].name);
+	    if (opcodes_2d[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 2;
+		if (len < opcodes_2d[opcode].min_len ||
+		    len > opcodes_2d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_2d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_2d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "2D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1c(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    switch ((data[0] & 0x00f80000) >> 19) {
+    case 0x11:
+	instr_out(data, hw_offset, 0, "3DSTATE_DEPTH_SUBRECTANGLE_DISALBE\n");
+	return 1;
+    case 0x10:
+	instr_out(data, hw_offset, 0, "3DSTATE_SCISSOR_ENABLE\n");
+	return 1;
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int len, i, c, opcode, word, map, sampler, instr;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d_1d[] = {
+	{ 0x8e, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x86, 4, 4, "3DSTATE_CHROMA_KEY" },
+	{ 0x9c, 1, 1, "3DSTATE_CLEAR_PARAMETERS" },
+	{ 0x88, 2, 2, "3DSTATE_CONSTANT_BLEND_COLOR" },
+	{ 0x99, 2, 2, "3DSTATE_DEFAULT_DIFFUSE" },
+	{ 0x9a, 2, 2, "3DSTATE_DEFAULT_SPECULAR" },
+	{ 0x98, 2, 2, "3DSTATE_DEFAULT_Z" },
+	{ 0x97, 2, 2, "3DSTATE_DEPTH_OFFSET_SCALE" },
+	{ 0x85, 2, 2, "3DSTATE_DEST_BUFFER_VARIABLES" },
+	{ 0x80, 5, 5, "3DSTATE_DRAWING_RECTANGLE" },
+	{ 0x8e, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x9d, 65, 65, "3DSTATE_FILTER_COEFFICIENTS_4X4" },
+	{ 0x9e, 4, 4, "3DSTATE_MONO_FILTER" },
+	{ 0x89, 4, 4, "3DSTATE_FOG_MODE" },
+	{ 0x8f, 2, 16, "3DSTATE_MAP_PALLETE_LOAD_32" },
+	{ 0x81, 3, 3, "3DSTATE_SCISSOR_RECTANGLE" },
+	{ 0x83, 2, 2, "3DSTATE_SPAN_STIPPLE" },
+    };
+
+    switch ((data[0] & 0x00ff0000) >> 16) {
+    case 0x07:
+	/* This instruction is unusual.  A 0 length means just 1 DWORD instead of
+	 * 2.  The 0 length is specified in one place to be unsupported, but
+	 * stated to be required in another, and 0 length LOAD_INDIRECTs appear
+	 * to cause no harm at least.
+	 */
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_INDIRECT\n");
+	len = (data[0] & 0x000000ff) + 1;
+	i = 1;
+	if (data[0] & (0x01 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SIS.0\n");
+	    instr_out(data, hw_offset, i++, "SIS.1\n");
+	}
+	if (data[0] & (0x02 << 8)) {
+	    if (i + 1 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "DIS.0\n");
+	}
+	if (data[0] & (0x04 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SSB.0\n");
+	    instr_out(data, hw_offset, i++, "SSB.1\n");
+	}
+	if (data[0] & (0x08 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "MSB.0\n");
+	    instr_out(data, hw_offset, i++, "MSB.1\n");
+	}
+	if (data[0] & (0x10 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSP.0\n");
+	    instr_out(data, hw_offset, i++, "PSP.1\n");
+	}
+	if (data[0] & (0x20 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSC.0\n");
+	    instr_out(data, hw_offset, i++, "PSC.1\n");
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x04:
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_STATE_IMMEDIATE_1\n");
+	len = (data[0] & 0x0000000f) + 2;
+	i = 1;
+	for (word = 0; word <= 7; word++) {
+	    if (data[0] & (1 << (4 + word))) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_LOAD_STATE_IMMEDIATE_1");
+
+		/* save vertex state for decode */
+		if (word == 2) {
+		    saved_s2_set = 1;
+		    saved_s2 = data[i];
+		}
+		if (word == 4) {
+		    saved_s4_set = 1;
+		    saved_s4 = data[i];
+		}
+
+		instr_out(data, hw_offset, i++, "S%d\n", word);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x00:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+
+	i = 1;
+	for (map = 0; map <= 15; map++) {
+	    if (data[1] & (1 << map)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
+		instr_out(data, hw_offset, i++, "map %d MS2\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS3\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS4\n", map);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x06:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_CONSTANTS\n");
+	len = (data[0] & 0x000000ff) + 2;
+
+	i = 1;
+	for (c = 0; c <= 31; c++) {
+	    if (data[1] & (1 << c)) {
+		if (i + 4 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_PIXEL_SHADER_CONSTANTS");
+		instr_out(data, hw_offset, i, "C%d.X = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Y = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Z = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.W = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x05:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	len = (data[0] & 0x000000ff) + 2;
+	if ((len - 1) % 3 != 0 || len > 370) {
+	    fprintf(out, "Bad count in 3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	    (*failures)++;
+	}
+	i = 1;
+	for (instr = 0; instr < (len - 1) / 3; instr++) {
+	    if (i + 3 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	}
+	return len;
+    case 0x01:
+	instr_out(data, hw_offset, 0, "3DSTATE_SAMPLER_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+	i = 1;
+	for (sampler = 0; sampler <= 15; sampler++) {
+	    if (data[1] & (1 << sampler)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_SAMPLER_STATE");
+		instr_out(data, hw_offset, i++, "sampler %d SS2\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS3\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS4\n",
+			  sampler);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_SAMPLER_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d_1d) / sizeof(opcodes_3d_1d[0]);
+	 opcode++)
+    {
+	if (((data[0] & 0x00ff0000) >> 16) == opcodes_3d_1d[opcode].opcode) {
+	    len = 1;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d_1d[opcode].name);
+	    if (opcodes_3d_1d[opcode].max_len > 1) {
+		len = (data[0] & 0x0000ffff) + 2;
+		if (len < opcodes_3d_1d[opcode].min_len ||
+		    len > opcodes_3d_1d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n",
+			    opcodes_3d_1d[opcode].name);
+		    (*failures)++;
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len,  opcodes_3d_1d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_primitive(uint32_t *data, int count, uint32_t hw_offset,
+		    int *failures)
+{
+    char immediate = (data[0] & (1 << 23)) == 0;
+    unsigned int len, i;
+    char *primtype;
+
+    switch ((data[0] >> 18) & 0xf) {
+    case 0x0: primtype = "TRILIST"; break;
+    case 0x1: primtype = "TRISTRIP"; break;
+    case 0x2: primtype = "TRISTRIP_REVERSE"; break;
+    case 0x3: primtype = "TRIFAN"; break;
+    case 0x4: primtype = "POLYGON"; break;
+    case 0x5: primtype = "LINELIST"; break;
+    case 0x6: primtype = "LINESTRIP"; break;
+    case 0x7: primtype = "RECTLIST"; break;
+    case 0x8: primtype = "POINTLIST"; break;
+    case 0x9: primtype = "DIB"; break;
+    case 0xa: primtype = "CLEAR_RECT"; break;
+    default: primtype = "unknown"; break;
+    }
+
+    /* XXX: 3DPRIM_DIB not supported */
+    if (immediate) {
+	len = (data[0] & 0x0003ffff) + 2;
+	instr_out(data, hw_offset, 0, "3DPRIMITIVE inline %s\n", primtype);
+	if (count < len)
+	    BUFFER_FAIL(count, len,  "3DPRIMITIVE inline");
+	if (!saved_s2_set || !saved_s4_set) {
+	    fprintf(out, "unknown vertex format\n");
+	    for (i = 1; i < len; i++) {
+		instr_out(data, hw_offset, i,
+			  "           vertex data (%f float)\n",
+			  int_as_float(data[i]));
+	    }
+	} else {
+	    unsigned int vertex = 0;
+	    for (i = 1; i < len;) {
+		unsigned int tc;
+
+#define VERTEX_OUT(fmt, ...) do {					\
+    if (i < len)							\
+	instr_out(data, hw_offset, i, " V%d."fmt"\n", vertex, __VA_ARGS__); \
+    else								\
+	fprintf(out, " missing data in V%d\n", vertex);			\
+    i++;								\
+} while (0)
+
+		VERTEX_OUT("X = %f", int_as_float(data[i]));
+		VERTEX_OUT("Y = %f", int_as_float(data[i]));
+	        switch (saved_s4 >> 6 & 0x7) {
+		case 0x1:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    break;
+		case 0x2:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		case 0x3:
+		    break;
+		case 0x4:
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		default:
+		    fprintf(out, "bad S4 position mask\n");
+		}
+
+		if (saved_s4 & (1 << 10)) {
+		    VERTEX_OUT("color = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 11)) {
+		    VERTEX_OUT("spec = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 12))
+		    VERTEX_OUT("width = 0x%08x)", data[i]);
+
+		for (tc = 0; tc <= 7; tc++) {
+		    switch ((saved_s2 >> (tc * 4)) & 0xf) {
+		    case 0x0:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x1:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x2:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.W = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x3:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x4:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0x5:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			VERTEX_OUT("T%d.ZW = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0xf:
+			break;
+		    default:
+			fprintf(out, "bad S2.T%d format\n", tc);
+		    }
+		}
+		vertex++;
+	    }
+	}
+    } else {
+	/* indirect vertices */
+	len = data[0] & 0x0000ffff; /* index count */
+	if (data[0] & (1 << 17)) {
+	    /* random vertex access */
+	    if (count < (len + 1) / 2 + 1) {
+		BUFFER_FAIL(count, (len + 1) / 2 + 1,
+			    "3DPRIMITIVE random indirect");
+	    }
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE random indirect %s (%d)\n", primtype, len);
+	    if (len == 0) {
+		/* vertex indices continue until 0xffff is found */
+		for (i = 1; i < count; i++) {
+		    if ((data[i] & 0xffff) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: (terminator)\n");
+			return i;
+		    } else if ((data[i] >> 16) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, "
+				  "(terminator)\n",
+				  data[i] & 0xffff);
+			return i;
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+		fprintf(out,
+			"3DPRIMITIVE: no terminator found in index buffer\n");
+		(*failures)++;
+		return count;
+	    } else {
+		/* fixed size vertex index buffer */
+		for (i = 0; i < len; i += 2) {
+		    if (i * 2 == len - 1) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x\n",
+				  data[i] & 0xffff);
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+	    }
+	    return (len + 1) / 2 + 1;
+	} else {
+	    /* sequential vertex access */
+	    if (count < 2)
+		BUFFER_FAIL(count, 2, "3DPRIMITIVE seq indirect");
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE sequential indirect %s, %d starting from "
+		      "%d\n", primtype, len, data[1] & 0xffff);
+	    instr_out(data, hw_offset, 1, "           start\n");
+	    return 2;
+	}
+    }
+
+    return len;
+}
+
+static int
+decode_3d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d[] = {
+	{ 0x06, 1, 1, "3DSTATE_ANTI_ALIASING" },
+	{ 0x08, 1, 1, "3DSTATE_BACKFACE_STENCIL_OPS" },
+	{ 0x09, 1, 1, "3DSTATE_BACKFACE_STENCIL_MASKS" },
+	{ 0x16, 1, 1, "3DSTATE_COORD_SET_BINDINGS" },
+	{ 0x15, 1, 1, "3DSTATE_FOG_COLOR" },
+	{ 0x0b, 1, 1, "3DSTATE_INDEPENDENT_ALPHA_BLEND" },
+	{ 0x0d, 1, 1, "3DSTATE_MODES_4" },
+	{ 0x0c, 1, 1, "3DSTATE_MODES_5" },
+	{ 0x07, 1, 1, "3DSTATE_RASTERIZATION_RULES" },
+    };
+
+    switch ((data[0] & 0x1f000000) >> 24) {
+    case 0x1f:
+	return decode_3d_primitive(data, count, hw_offset, failures);
+    case 0x1d:
+	return decode_3d_1d(data, count, hw_offset, failures);
+    case 0x1c:
+	return decode_3d_1c(data, count, hw_offset, failures);
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1f000000) >> 24 == opcodes_3d[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d[opcode].name);
+	    if (opcodes_3d[opcode].max_len > 1) {
+		len = (data[0] & 0xff) + 2;
+		if (len < opcodes_3d[opcode].min_len ||
+		    len > opcodes_3d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_3d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_3d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+/**
+ * Decodes an i830-i915 batch buffer, writing the output to stdout.
+ *
+ * \param data batch buffer contents
+ * \param count number of DWORDs to decode in the batch buffer
+ * \param hw_offset hardware address for the buffer
+ */
+int
+i915_disasm(uint32_t *data, int count, uint32_t hw_offset)
+{
+    int index = 0;
+    int failures = 0;
+
+    out = stdout;
+
+    while (index < count) {
+	switch ((data[index] & 0xe0000000) >> 29) {
+	case 0x0:
+	    index += decode_mi(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x2:
+	    index += decode_2d(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x3:
+	    index += decode_3d(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	default:
+	    instr_out(data, hw_offset, index, "UNKNOWN\n");
+	    failures++;
+	    index++;
+	    break;
+	}
+	fflush(out);
+    }
+
+    return failures;
+}
+
+void i915_disasm_context_reset()
+{
+    saved_s2_set = 0;
+    saved_s4_set = 1;
+}
+
diff --git a/src/mesa/drivers/dri/i915tex/i915_disasm.h b/src/mesa/drivers/dri/i915tex/i915_disasm.h
new file mode 100644
index 00000000000..5eb620f9a4e
--- /dev/null
+++ b/src/mesa/drivers/dri/i915tex/i915_disasm.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * 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 (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 NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
+ *
+ */
+
+int i915_disasm(uint32_t *data, int count, uint32_t hw_offset);
+void i915_disasm_context_reset();
diff --git a/src/mesa/drivers/dri/i915tex/i915_vtbl.c b/src/mesa/drivers/dri/i915tex/i915_vtbl.c
index 2a3ccd01a05..5d6ad8d86cf 100644
--- a/src/mesa/drivers/dri/i915tex/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915tex/i915_vtbl.c
@@ -197,7 +197,7 @@ 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 */
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
index 26eecb94409..1edbb618d34 100644
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
@@ -66,27 +66,6 @@
  * modifying cliprects ???
  */
 
-static void
-intel_dump_batchbuffer(GLuint offset, GLuint * ptr, GLuint size)
-{
-   int i;
-   fprintf(stderr, "\n\n\nSTART BATCH (%d dwords):\n", size / 4);
-   for (i = 0; i < size / 4; i += 4) {
-      char dw1[11], dw2[11] = "", dw3[11] = "", dw4[11] = "";
-
-      sprintf(dw1, "0x%08x", ptr[i]);
-      if (i + 1 < size / 4)
-	 sprintf(dw2, "0x%08x", ptr[i + 1]);
-      if (i + 2 < size / 4)
-	 sprintf(dw3, "0x%08x", ptr[i + 2]);
-      if (i + 3 < size / 4)
-	 sprintf(dw4, "0x%08x", ptr[i + 3]);
-      fprintf(stderr, "0x%x:\t%s %s %s %s\n", offset + i * 4,
-	      dw1, dw2, dw3, dw4);
-   }
-   fprintf(stderr, "END BATCH\n\n\n");
-}
-
 /*======================================================================
  * Public functions
  */
@@ -166,7 +145,7 @@ do_flush_locked(struct intel_batchbuffer *batch,
    }
 
    if (INTEL_DEBUG & DEBUG_BATCH)
-      intel_dump_batchbuffer(0, ptr, used);
+      i915_disasm(ptr, used / 4, 0);
 
    dri_bo_unmap(batch->buf);
    batch->map = NULL;
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.c b/src/mesa/drivers/dri/i915tex/intel_context.c
index 696845a7fc9..cd4333d0d34 100644
--- a/src/mesa/drivers/dri/i915tex/intel_context.c
+++ b/src/mesa/drivers/dri/i915tex/intel_context.c
@@ -59,6 +59,7 @@
 #include "intel_regions.h"
 #include "intel_buffer_objects.h"
 #include "intel_fbo.h"
+#include "i915_disasm.h"
 
 #include "drirenderbuffer.h"
 #include "vblank.h"
@@ -130,6 +131,18 @@ intelGetString(GLcontext * ctx, GLenum name)
       case PCI_CHIP_I945_GM:
          chipset = "Intel(R) 945GM";
          break;
+      case PCI_CHIP_I945_GME:
+         chipset = "Intel(R) 945GME";
+         break;
+      case PCI_CHIP_G33_G:
+	 chipset = "Intel(R) G33";
+	 break;
+      case PCI_CHIP_Q35_G:
+	 chipset = "Intel(R) Q35";
+	 break;
+      case PCI_CHIP_Q33_G:
+	 chipset = "Intel(R) Q33";
+	 break;
       default:
          chipset = "Unknown Intel Chipset";
          break;
@@ -646,6 +659,8 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
    if (!intel->intelScreen->ttm && sarea->texAge != intel->hHWContext) {
       sarea->texAge = intel->hHWContext;
       dri_bufmgr_fake_contended_lock_take(intel->intelScreen->bufmgr);
+      if (INTEL_DEBUG & DEBUG_BATCH)
+	 i915_disasm_context_reset();
    }
 
    if (sarea->width != intelScreen->width ||
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.h b/src/mesa/drivers/dri/i915tex/intel_context.h
index a55f7d984e3..8755f5703dc 100644
--- a/src/mesa/drivers/dri/i915tex/intel_context.h
+++ b/src/mesa/drivers/dri/i915tex/intel_context.h
@@ -378,6 +378,10 @@ extern int INTEL_DEBUG;
 #define PCI_CHIP_I915_GM		0x2592
 #define PCI_CHIP_I945_G			0x2772
 #define PCI_CHIP_I945_GM		0x27A2
+#define PCI_CHIP_I945_GME		0x27AE
+#define PCI_CHIP_G33_G			0x29C2
+#define PCI_CHIP_Q35_G			0x29B2
+#define PCI_CHIP_Q33_G			0x29D2
 
 
 /* ================================================================
diff --git a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
index 8e83028b26c..843a78eb82b 100644
--- a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
@@ -79,6 +79,10 @@ intel_miptree_create(struct intel_context *intel,
    switch (intel->intelScreen->deviceID) {
    case PCI_CHIP_I945_G:
    case PCI_CHIP_I945_GM:
+   case PCI_CHIP_I945_GME:
+   case PCI_CHIP_G33_G:
+   case PCI_CHIP_Q33_G:
+   case PCI_CHIP_Q35_G:
       ok = i945_miptree_layout(mt);
       break;
    case PCI_CHIP_I915_G:
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c b/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c
index cbb583d3a20..6d9fc6144a2 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c
@@ -363,5 +363,20 @@ intelDrawPixels(GLcontext * ctx,
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
 
-   _swrast_DrawPixels(ctx, x, y, width, height, format, type, unpack, pixels);
+   if (ctx->FragmentProgram._Current == ctx->FragmentProgram._TexEnvProgram) {
+      /*
+       * We don't want the i915 texenv program to be applied to DrawPixels.
+       * This is really just a performance optimization (mesa will other-
+       * wise happily run the fragment program on each pixel in the image).
+       */
+      struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
+      ctx->FragmentProgram._Current = NULL;
+      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
+                          unpack, pixels );
+      ctx->FragmentProgram._Current = fpSave;
+   }
+   else {
+      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
+                          unpack, pixels );
+   }
 }
diff --git a/src/mesa/drivers/dri/i915tex/intel_screen.c b/src/mesa/drivers/dri/i915tex/intel_screen.c
index e3a631b6961..89cf3ea9139 100644
--- a/src/mesa/drivers/dri/i915tex/intel_screen.c
+++ b/src/mesa/drivers/dri/i915tex/intel_screen.c
@@ -123,8 +123,6 @@ intelMapScreenRegions(__DRIscreenPrivate * sPriv)
    if (0)
       _mesa_printf("TEX 0x%08x ", intelScreen->tex.handle);
    if (intelScreen->tex.size != 0) {
-      intelScreen->ttm = GL_FALSE;
-
       if (drmMap(sPriv->fd,
 		 intelScreen->tex.handle,
 		 intelScreen->tex.size,
@@ -132,8 +130,6 @@ intelMapScreenRegions(__DRIscreenPrivate * sPriv)
 	 intelUnmapScreenRegions(intelScreen);
 	 return GL_FALSE;
       }
-   } else {
-      intelScreen->ttm = GL_TRUE;
    }
 
    if (0)
@@ -530,12 +526,16 @@ intelInitDriver(__DRIscreenPrivate * sPriv)
       (*glx_enable_extension) (psc, "GLX_SGI_make_current_read");
    }
 
-   if (intelScreen->ttm) {
-      intelScreen->bufmgr = dri_bufmgr_ttm_init(sPriv->fd,
-						DRM_FENCE_TYPE_EXE,
-						DRM_FENCE_TYPE_EXE |
-						DRM_I915_FENCE_TYPE_RW);
-   } else {
+   intelScreen->bufmgr = dri_bufmgr_ttm_init(sPriv->fd,
+					     DRM_FENCE_TYPE_EXE,
+					     DRM_FENCE_TYPE_EXE |
+					     DRM_I915_FENCE_TYPE_RW);
+   if (intelScreen->bufmgr == NULL) {
+      if (intelScreen->tex.size == 0) {
+	 fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
+		 __func__, __LINE__);
+	 return GL_FALSE;
+      }
       intelScreen->bufmgr = dri_bufmgr_fake_init(intelScreen->tex.offset,
 						 intelScreen->tex.map,
 						 intelScreen->tex.size,
@@ -740,6 +740,10 @@ intelCreateContext(const __GLcontextModes * mesaVis,
    case PCI_CHIP_I915_GM:
    case PCI_CHIP_I945_G:
    case PCI_CHIP_I945_GM:
+   case PCI_CHIP_I945_GME:
+   case PCI_CHIP_G33_G:
+   case PCI_CHIP_Q35_G:
+   case PCI_CHIP_Q33_G:
       return i915CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
 
    default:
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_image.c b/src/mesa/drivers/dri/i915tex/intel_tex_image.c
index b15569681a5..197cf35ebe6 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_image.c
+++ b/src/mesa/drivers/dri/i915tex/intel_tex_image.c
@@ -377,9 +377,6 @@ intelTexImage(GLcontext * ctx,
       assert(!intelObj->mt);
    }
 
-   if (!pixels)
-      return;
-
    if (!intelObj->mt) {
       guess_and_alloc_mipmap_tree(intel, intelObj, intelImage);
       if (!intelObj->mt) {
@@ -681,6 +678,9 @@ intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
    if (!intelObj)
       return;
 
+   if (intelObj->mt)
+      intel_miptree_release(intel, &intelObj->mt);
+
    intelObj->imageOverride = GL_TRUE;
    intelObj->depthOverride = depth;
    intelObj->pitchOverride = pitch;
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c b/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
index 0ae4fee1ba0..af18c26d55c 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
@@ -116,7 +116,7 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
 
    /* We know/require this is true by now: 
     */
-   assert(intelObj->base.Complete);
+   assert(intelObj->base._Complete);
 
    /* What levels must the tree include at a minimum?
     */
diff --git a/src/mesa/drivers/dri/i965/intel_context.c b/src/mesa/drivers/dri/i965/intel_context.c
index 422eb960979..9617cbebbfd 100644
--- a/src/mesa/drivers/dri/i965/intel_context.c
+++ b/src/mesa/drivers/dri/i965/intel_context.c
@@ -106,20 +106,23 @@ static const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
    case GL_RENDERER:
       switch (intel_context(ctx)->intelScreen->deviceID) {
       case PCI_CHIP_I965_Q:
-	 chipset = "Intel(R) 965Q"; break;
+	 chipset = "Intel(R) 965Q";
          break;
       case PCI_CHIP_I965_G:
       case PCI_CHIP_I965_G_1:
-	 chipset = "Intel(R) 965G"; break;
+	 chipset = "Intel(R) 965G";
          break;
       case PCI_CHIP_I946_GZ:
-	 chipset = "Intel(R) 946GZ"; break;
+	 chipset = "Intel(R) 946GZ";
          break;
       case PCI_CHIP_I965_GM:
-	 chipset = "Intel(R) 965GM"; break;
+	 chipset = "Intel(R) 965GM";
+         break;
+      case PCI_CHIP_I965_GME:
+	 chipset = "Intel(R) 965GME/GLE";
          break;
       default:
-	 chipset = "Unknown Intel Chipset"; break;
+	 chipset = "Unknown Intel Chipset";
       }
 
       (void) driGetRendererString( buffer, chipset, DRIVER_VERSION, 0 );
diff --git a/src/mesa/drivers/dri/i965/intel_context.h b/src/mesa/drivers/dri/i965/intel_context.h
index a3c65b66e08..406f8483dce 100644
--- a/src/mesa/drivers/dri/i965/intel_context.h
+++ b/src/mesa/drivers/dri/i965/intel_context.h
@@ -385,6 +385,7 @@ extern int INTEL_DEBUG;
 #define PCI_CHIP_I965_G_1		0x2982
 #define PCI_CHIP_I946_GZ		0x2972
 #define PCI_CHIP_I965_GM                0x2A02
+#define PCI_CHIP_I965_GME               0x2A12
 
 
 /* ================================================================
diff --git a/src/mesa/drivers/dri/i965/intel_tex_validate.c b/src/mesa/drivers/dri/i965/intel_tex_validate.c
index cb23b9dd879..44ee94614d5 100644
--- a/src/mesa/drivers/dri/i965/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i965/intel_tex_validate.c
@@ -138,7 +138,7 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
    
    /* We know/require this is true by now: 
     */
-   assert(intelObj->base.Complete);
+   assert(intelObj->base._Complete);
 
    /* What levels must the tree include at a minimum?
     */
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
index 3bc84d862d3..4576c1ede4d 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
@@ -392,15 +392,6 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	int i;
 	int slots=0;
 	int total_size=0;
-	/* t_vertex_generic dereferences a NULL pointer if we
-	 * pass NULL as the vp transform...
-	 */
-	const GLfloat ident_vp[16] = {
-	   1.0, 0.0, 0.0, 0.0,
-	   0.0, 1.0, 0.0, 0.0,
-	   0.0, 0.0, 1.0, 0.0,
-	   0.0, 0.0, 0.0, 1.0
-	};
 
 	nmesa->vertex_attr_count = 0;
 	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
@@ -431,28 +422,20 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 		if (RENDERINPUTS_TEST(index, i))
 		{
 			slots=i+1;
-			if (i==_TNL_ATTRIB_POS)
-			{
-				/* special-case POS */
-				EMIT_ATTR(_TNL_ATTRIB_POS,EMIT_3F_VIEWPORT);
-			}
-			else
+			switch(attr_size[i])
 			{
-				switch(attr_size[i])
-				{
-					case 1:
-						EMIT_ATTR(i,EMIT_1F);
-						break;
-					case 2:
-						EMIT_ATTR(i,EMIT_2F);
-						break;
-					case 3:
-						EMIT_ATTR(i,EMIT_3F);
-						break;
-					case 4:
-						EMIT_ATTR(i,EMIT_4F);
-						break;
-				}
+				case 1:
+					EMIT_ATTR(i,EMIT_1F);
+					break;
+				case 2:
+					EMIT_ATTR(i,EMIT_2F);
+					break;
+				case 3:
+					EMIT_ATTR(i,EMIT_3F);
+					break;
+				case 4:
+					EMIT_ATTR(i,EMIT_4F);
+					break;
 			}
 			if (i==_TNL_ATTRIB_COLOR0)
 				nmesa->color_offset=total_size;
@@ -465,7 +448,7 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	nmesa->vertex_size=_tnl_install_attrs( ctx,
 			nmesa->vertex_attrs, 
 			nmesa->vertex_attr_count,
-			ident_vp, 0 );
+			NULL, 0 );
 	assert(nmesa->vertex_size==total_size*4);
 
 	/* 
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
index 0351989b2e4..7055286ba95 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
@@ -133,13 +133,15 @@ static void r300PrintStateAtom(r300ContextPtr r300, struct r300_state_atom *stat
 	int i;
 	int dwords = (*state->check) (r300, state);
 
-	fprintf(stderr, "  emit %s/%d/%d\n", state->name, dwords,
+	fprintf(stderr, "  emit %s %d/%d\n", state->name, dwords,
 		state->cmd_size);
 
-	if (RADEON_DEBUG & DEBUG_VERBOSE)
-		for (i = 0; i < dwords; i++)
-			fprintf(stderr, "      %s[%d]: %08X\n",
+	if (RADEON_DEBUG & DEBUG_VERBOSE) {
+		for (i = 0; i < dwords; i++) {
+			fprintf(stderr, "      %s[%d]: %08x\n",
 				state->name, i, state->cmd[i]);
+		}
+	}
 }
 
 /**
@@ -152,24 +154,10 @@ static inline void r300EmitAtoms(r300ContextPtr r300, GLboolean dirty)
 {
 	struct r300_state_atom *atom;
 	uint32_t *dest;
+	int dwords;
 
 	dest = r300->cmdbuf.cmd_buf + r300->cmdbuf.count_used;
 
-	if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
-		foreach(atom, &r300->hw.atomlist) {
-			if ((atom->dirty || r300->hw.all_dirty) == dirty) {
-				int dwords = (*atom->check) (r300, atom);
-
-				if (dwords)
-					r300PrintStateAtom(r300, atom);
-				else
-					fprintf(stderr,
-						"  skip state %s\n",
-						atom->name);
-			}
-		}
-	}
-
 	/* Emit WAIT */
 	*dest = cmdwait(R300_WAIT_3D | R300_WAIT_3D_CLEAN);
 	dest++;
@@ -193,13 +181,20 @@ static inline void r300EmitAtoms(r300ContextPtr r300, GLboolean dirty)
 
 	foreach(atom, &r300->hw.atomlist) {
 		if ((atom->dirty || r300->hw.all_dirty) == dirty) {
-			int dwords = (*atom->check) (r300, atom);
-
+			dwords = (*atom->check) (r300, atom);
 			if (dwords) {
+				if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
+					r300PrintStateAtom(r300, atom);
+				}
 				memcpy(dest, atom->cmd, dwords * 4);
 				dest += dwords;
 				r300->cmdbuf.count_used += dwords;
 				atom->dirty = GL_FALSE;
+			} else {
+				if (DEBUG_CMDBUF && RADEON_DEBUG & DEBUG_STATE) {
+					fprintf(stderr, "  skip state %s\n",
+						atom->name);
+				}
 			}
 		}
 	}
@@ -245,22 +240,28 @@ void r300EmitState(r300ContextPtr r300)
 	r300->hw.all_dirty = GL_FALSE;
 }
 
-#define CHECK( NM, COUNT )				\
-static int check_##NM( r300ContextPtr r300, 		\
-			struct r300_state_atom* atom )	\
-{							\
-   (void) atom;	(void) r300;				\
-   return (COUNT);					\
-}
-
 #define packet0_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->packet0.count)
 #define vpu_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->vpu.count)
 
-CHECK(always, atom->cmd_size)
-    CHECK(variable, packet0_count(atom->cmd) ? (1 + packet0_count(atom->cmd)) : 0)
-    CHECK(vpu, vpu_count(atom->cmd) ? (1 + vpu_count(atom->cmd) * 4) : 0)
-#undef packet0_count
-#undef vpu_count
+static int check_always(r300ContextPtr r300, struct r300_state_atom *atom)
+{
+	return atom->cmd_size;
+}
+
+static int check_variable(r300ContextPtr r300, struct r300_state_atom *atom)
+{
+	int cnt;
+	cnt = packet0_count(atom->cmd);
+	return cnt ? cnt + 1 : 0;
+}
+
+static int check_vpu(r300ContextPtr r300, struct r300_state_atom *atom)
+{
+	int cnt;
+	cnt = vpu_count(atom->cmd);
+	return cnt ? (cnt * 4) + 1 : 0;
+}
+
 #define ALLOC_STATE( ATOM, CHK, SZ, IDX )				\
    do {									\
       r300->hw.ATOM.cmd_size = (SZ);					\
@@ -318,8 +319,8 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	r300->hw.unk21DC.cmd[0] = cmdpacket0(0x21DC, 1);
 	ALLOC_STATE(unk221C, always, 2, 0);
 	r300->hw.unk221C.cmd[0] = cmdpacket0(R300_VAP_UNKNOWN_221C, 1);
-	ALLOC_STATE(unk2220, always, 5, 0);
-	r300->hw.unk2220.cmd[0] = cmdpacket0(0x2220, 4);
+	ALLOC_STATE(vap_clip, always, 5, 0);
+	r300->hw.vap_clip.cmd[0] = cmdpacket0(R300_VAP_CLIP_X_0, 4);
 	ALLOC_STATE(unk2288, always, 2, 0);
 	r300->hw.unk2288.cmd[0] = cmdpacket0(R300_VAP_UNKNOWN_2288, 1);
 	ALLOC_STATE(vof, always, R300_VOF_CMDSIZE, 0);
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 01caa617667..076bb49a006 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -49,8 +49,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define USER_BUFFERS
 
-//#define OPTIMIZE_ELTS
-
 struct r300_context;
 typedef struct r300_context r300ContextRec;
 typedef struct r300_context *r300ContextPtr;
@@ -149,7 +147,6 @@ struct r300_dma_region {
 	int aos_offset;		/* address in GART memory */
 	int aos_stride;		/* distance between elements, in dwords */
 	int aos_size;		/* number of components (1-4) */
-	int aos_reg;		/* VAP register assignment */
 };
 
 struct r300_dma {
@@ -455,7 +452,7 @@ struct r300_hw_state {
 	struct r300_state_atom vic;	/* vap input control (2180) */
 	struct r300_state_atom unk21DC;	/* (21DC) */
 	struct r300_state_atom unk221C;	/* (221C) */
-	struct r300_state_atom unk2220;	/* (2220) */
+	struct r300_state_atom vap_clip;
 	struct r300_state_atom unk2288;	/* (2288) */
 	struct r300_state_atom pvs;	/* pvs_cntl (22D0) */
 	struct r300_state_atom gb_enable;	/* (4008) */
@@ -783,11 +780,6 @@ struct r300_fragment_program {
 
 #define R300_MAX_AOS_ARRAYS		16
 
-#define AOS_FORMAT_USHORT	0
-#define AOS_FORMAT_FLOAT	1
-#define AOS_FORMAT_UBYTE	2
-#define AOS_FORMAT_FLOAT_COLOR	3
-
 #define REG_COORDS	0
 #define REG_COLOR0	1
 #define REG_TEX0	2
diff --git a/src/mesa/drivers/dri/r300/r300_emit.c b/src/mesa/drivers/dri/r300/r300_emit.c
index 9fb712f7b8f..4670c28a02f 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.c
+++ b/src/mesa/drivers/dri/r300/r300_emit.c
@@ -86,16 +86,15 @@ do {						\
 } while (0)
 #endif
 
-static void r300EmitVec4(GLcontext * ctx,
-			 struct r300_dma_region *rvb,
+static void r300EmitVec4(GLcontext * ctx, struct r300_dma_region *rvb,
 			 GLvoid * data, int stride, int count)
 {
 	int i;
 	int *out = (int *)(rvb->address + rvb->start);
 
 	if (RADEON_DEBUG & DEBUG_VERTS)
-		fprintf(stderr, "%s count %d stride %d\n",
-			__FUNCTION__, count, stride);
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
 
 	if (stride == 4)
 		COPY_DWORDS(out, data, count);
@@ -107,16 +106,15 @@ static void r300EmitVec4(GLcontext * ctx,
 		}
 }
 
-static void r300EmitVec8(GLcontext * ctx,
-			 struct r300_dma_region *rvb,
+static void r300EmitVec8(GLcontext * ctx, struct r300_dma_region *rvb,
 			 GLvoid * data, int stride, int count)
 {
 	int i;
 	int *out = (int *)(rvb->address + rvb->start);
 
 	if (RADEON_DEBUG & DEBUG_VERTS)
-		fprintf(stderr, "%s count %d stride %d\n",
-			__FUNCTION__, count, stride);
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
 
 	if (stride == 8)
 		COPY_DWORDS(out, data, count * 2);
@@ -129,8 +127,7 @@ static void r300EmitVec8(GLcontext * ctx,
 		}
 }
 
-static void r300EmitVec12(GLcontext * ctx,
-			  struct r300_dma_region *rvb,
+static void r300EmitVec12(GLcontext * ctx, struct r300_dma_region *rvb,
 			  GLvoid * data, int stride, int count)
 {
 	int i;
@@ -152,16 +149,15 @@ static void r300EmitVec12(GLcontext * ctx,
 		}
 }
 
-static void r300EmitVec16(GLcontext * ctx,
-			  struct r300_dma_region *rvb,
+static void r300EmitVec16(GLcontext * ctx, struct r300_dma_region *rvb,
 			  GLvoid * data, int stride, int count)
 {
 	int i;
 	int *out = (int *)(rvb->address + rvb->start);
 
 	if (RADEON_DEBUG & DEBUG_VERTS)
-		fprintf(stderr, "%s count %d stride %d\n",
-			__FUNCTION__, count, stride);
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
 
 	if (stride == 16)
 		COPY_DWORDS(out, data, count * 4);
@@ -176,32 +172,22 @@ static void r300EmitVec16(GLcontext * ctx,
 		}
 }
 
-static void r300EmitVec(GLcontext * ctx,
-			struct r300_dma_region *rvb,
+static void r300EmitVec(GLcontext * ctx, struct r300_dma_region *rvb,
 			GLvoid * data, int size, int stride, int count)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 
-	if (RADEON_DEBUG & DEBUG_VERTS)
-		fprintf(stderr, "%s count %d size %d stride %d\n",
-			__FUNCTION__, count, size, stride);
-
-	/* Gets triggered when playing with future_hw_tcl_on ... */
-	//assert(!rvb->buf);
-
 	if (stride == 0) {
 		r300AllocDmaRegion(rmesa, rvb, size * 4, 4);
 		count = 1;
 		rvb->aos_offset = GET_START(rvb);
 		rvb->aos_stride = 0;
 	} else {
-		r300AllocDmaRegion(rmesa, rvb, size * count * 4, 4);	/* alignment? */
+		r300AllocDmaRegion(rmesa, rvb, size * count * 4, 4);
 		rvb->aos_offset = GET_START(rvb);
 		rvb->aos_stride = size;
 	}
 
-	/* Emit the data
-	 */
 	switch (size) {
 	case 1:
 		r300EmitVec4(ctx, rvb, data, stride, count);
@@ -217,57 +203,35 @@ static void r300EmitVec(GLcontext * ctx,
 		break;
 	default:
 		assert(0);
-		_mesa_exit(-1);
 		break;
 	}
-
-}
-
-#define R300_VIR0_AOS_SIZE_SHIFT 0
-#define R300_VIR0_AOS_INPUT_SHIFT 8
-#define R300_VIR0_AOS_STOP_SHIFT 13
-#define R300_VIR0_AOS_TYPE_SHIFT 14
-#define R300_VIR0_HIGH_SHIFT 16
-
-// Pack 4 elemets in a 16 bit (aos_size first 8, input next 5, 1 stop bit(Whild gues), aos_type last 2);
-static inline GLuint t_vir_pack(GLvector4f ** dt, int *inputs, int i)
-{
-	GLuint dw;
-	dw = (dt[i]->size - 1) << R300_VIR0_AOS_SIZE_SHIFT;
-	dw |= inputs[i] << R300_VIR0_AOS_INPUT_SHIFT;
-	//dw |= t_type(&dt[i]) << R300_VIR0_AOS_TYPE_SHIFT;
-	return dw;
 }
 
-static GLuint t_vir0(uint32_t * dst, GLvector4f ** dt, int *inputs,
-		     GLint * tab, GLuint nr)
+static GLuint r300VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
+				 int *inputs, GLint * tab, GLuint nr)
 {
-	GLuint i, dw, dwInternel;
+	GLuint i, dw;
 
+	/* type, inputs, stop bit, size */
 	for (i = 0; i + 1 < nr; i += 2) {
-		dw = t_vir_pack(dt, inputs, tab[i]);
-		dwInternel = t_vir_pack(dt, inputs, tab[i + 1]);
-		dw |= dwInternel << R300_VIR0_HIGH_SHIFT;
-
+		dw = R300_INPUT_ROUTE_FLOAT | (inputs[tab[i]] << 8) | (attribptr[tab[i]]->size - 1);
+		dw |= (R300_INPUT_ROUTE_FLOAT | (inputs[tab[i + 1]] << 8) | (attribptr[tab[i + 1]]->size - 1)) << 16;
 		if (i + 2 == nr) {
-			dw |=
-			    (1 <<
-			     (R300_VIR0_AOS_STOP_SHIFT + R300_VIR0_HIGH_SHIFT));
+			dw |= (1 << (13 + 16));
 		}
-		dst[i >> 1] = dw;	// Is the same as i/2
+		dst[i >> 1] = dw;
 	}
 
 	if (nr & 1) {
-		dw = t_vir_pack(dt, inputs, tab[nr - 1]);
-		dw |= 1 << R300_VIR0_AOS_STOP_SHIFT;
-
+		dw = R300_INPUT_ROUTE_FLOAT | (inputs[tab[nr - 1]] << 8) | (attribptr[tab[nr - 1]]->size - 1);
+		dw |= 1 << 13;
 		dst[nr >> 1] = dw;
 	}
 
-	return (nr + 1) >> 1;	// Is the same as (nr+1)/2
+	return (nr + 1) >> 1;
 }
 
-static GLuint t_swizzle(int swizzle[4])
+static GLuint r300VAPInputRoute1Swizzle(int swizzle[4])
 {
 	return (swizzle[0] << R300_INPUT_ROUTE_X_SHIFT) |
 	    (swizzle[1] << R300_INPUT_ROUTE_Y_SHIFT) |
@@ -275,27 +239,32 @@ static GLuint t_swizzle(int swizzle[4])
 	    (swizzle[3] << R300_INPUT_ROUTE_W_SHIFT);
 }
 
-static GLuint t_vir1(uint32_t * dst, int swizzle[][4], GLuint nr)
+static GLuint r300VAPInputRoute1(uint32_t * dst, int swizzle[][4], GLuint nr)
 {
 	GLuint i;
 
 	for (i = 0; i + 1 < nr; i += 2) {
-		dst[i >> 1] = t_swizzle(swizzle[i]) | R300_INPUT_ROUTE_ENABLE;
-		dst[i >> 1] |=
-		    (t_swizzle(swizzle[i + 1]) | R300_INPUT_ROUTE_ENABLE)
-		    << 16;
+		dst[i >> 1] = r300VAPInputRoute1Swizzle(swizzle[i]) | R300_INPUT_ROUTE_ENABLE;
+		dst[i >> 1] |= (r300VAPInputRoute1Swizzle(swizzle[i + 1]) | R300_INPUT_ROUTE_ENABLE) << 16;
 	}
 
-	if (nr & 1)
-		dst[nr >> 1] =
-		    t_swizzle(swizzle[nr - 1]) | R300_INPUT_ROUTE_ENABLE;
+	if (nr & 1) {
+		dst[nr >> 1] = r300VAPInputRoute1Swizzle(swizzle[nr - 1]) | R300_INPUT_ROUTE_ENABLE;
+	}
 
 	return (nr + 1) >> 1;
 }
 
-static GLuint t_vic(GLcontext * ctx, GLuint InputsRead)
+static GLuint r300VAPInputCntl0(GLcontext * ctx, GLuint InputsRead)
+{
+	/* No idea what this value means. I have seen other values written to
+	 * this register... */
+	return 0x5555;
+}
+
+static GLuint r300VAPInputCntl1(GLcontext * ctx, GLuint InputsRead)
 {
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	GLuint i, vic_1 = 0;
 
 	if (InputsRead & (1 << VERT_ATTRIB_POS))
@@ -307,25 +276,65 @@ static GLuint t_vic(GLcontext * ctx, GLuint InputsRead)
 	if (InputsRead & (1 << VERT_ATTRIB_COLOR0))
 		vic_1 |= R300_INPUT_CNTL_COLOR;
 
-	r300->state.texture.tc_count = 0;
+	rmesa->state.texture.tc_count = 0;
 	for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
 		if (InputsRead & (1 << (VERT_ATTRIB_TEX0 + i))) {
-			r300->state.texture.tc_count++;
+			rmesa->state.texture.tc_count++;
 			vic_1 |= R300_INPUT_CNTL_TC0 << i;
 		}
 
 	return vic_1;
 }
 
+static GLuint r300VAPOutputCntl0(GLcontext * ctx, GLuint OutputsWritten)
+{
+	GLuint ret = 0;
+
+	if (OutputsWritten & (1 << VERT_RESULT_HPOS))
+		ret |= R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT;
+
+	if (OutputsWritten & (1 << VERT_RESULT_COL0))
+		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT;
+
+	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_FOGC)) ;
+#endif
+
+	if (OutputsWritten & (1 << VERT_RESULT_PSIZ))
+		ret |= R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT;
+
+	return ret;
+}
+
+static GLuint r300VAPOutputCntl1(GLcontext * ctx, GLuint OutputsWritten)
+{
+	GLuint i, ret = 0;
+
+	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+		if (OutputsWritten & (1 << (VERT_RESULT_TEX0 + i))) {
+			ret |= (4 << (3 * i));
+		}
+	}
+
+	return ret;
+}
+
 /* Emit vertex data to GART memory
  * Route inputs to the vertex processor
  * This function should never return R300_FALLBACK_TCL when using software tcl.
  */
-
 int r300EmitArrays(GLcontext * ctx)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
-	r300ContextPtr r300 = rmesa;
 	TNLcontext *tnl = TNL_CONTEXT(ctx);
 	struct vertex_buffer *vb = &tnl->vb;
 	GLuint nr;
@@ -336,114 +345,105 @@ int r300EmitArrays(GLcontext * ctx)
 	int vir_inputs[VERT_ATTRIB_MAX];
 	GLint tab[VERT_ATTRIB_MAX];
 	int swizzle[VERT_ATTRIB_MAX][4];
+	struct r300_vertex_program *prog =
+	    (struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
 
 	if (hw_tcl_on) {
-		struct r300_vertex_program *prog =
-		    (struct r300_vertex_program *)
-		    CURRENT_VERTEX_SHADER(ctx);
 		inputs = prog->inputs;
-		InputsRead = CURRENT_VERTEX_SHADER(ctx)->key.InputsRead;
-		OutputsWritten = CURRENT_VERTEX_SHADER(ctx)->key.OutputsWritten;
+		InputsRead = prog->key.InputsRead;
+		OutputsWritten = prog->key.OutputsWritten;
 	} else {
-		DECLARE_RENDERINPUTS(inputs_bitset);
-		inputs = r300->state.sw_tcl_inputs;
+		inputs = rmesa->state.sw_tcl_inputs;
 
-		RENDERINPUTS_COPY(inputs_bitset,
-				  TNL_CONTEXT(ctx)->render_inputs_bitset);
+		DECLARE_RENDERINPUTS(render_inputs_bitset);
+		RENDERINPUTS_COPY(render_inputs_bitset, tnl->render_inputs_bitset);
 
-		assert(RENDERINPUTS_TEST(inputs_bitset, _TNL_ATTRIB_POS));
-		InputsRead |= 1 << VERT_ATTRIB_POS;
-		OutputsWritten |= 1 << VERT_RESULT_HPOS;
+		assert(RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_POS));
+		assert(RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_NORMAL) == 0);
+		assert(RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_COLOR0));
 
-		assert(RENDERINPUTS_TEST(inputs_bitset, _TNL_ATTRIB_NORMAL)
-		       == 0);
+		if (RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_POS)) {
+			InputsRead |= 1 << VERT_ATTRIB_POS;
+			OutputsWritten |= 1 << VERT_RESULT_HPOS;
+		}
 
-		assert(RENDERINPUTS_TEST(inputs_bitset, _TNL_ATTRIB_COLOR0));
-		InputsRead |= 1 << VERT_ATTRIB_COLOR0;
-		OutputsWritten |= 1 << VERT_RESULT_COL0;
+		if (RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_COLOR0)) {
+			InputsRead |= 1 << VERT_ATTRIB_COLOR0;
+			OutputsWritten |= 1 << VERT_RESULT_COL0;
+		}
 
-		if (RENDERINPUTS_TEST(inputs_bitset, _TNL_ATTRIB_COLOR1)) {
+		if (RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_COLOR1)) {
 			InputsRead |= 1 << VERT_ATTRIB_COLOR1;
 			OutputsWritten |= 1 << VERT_RESULT_COL1;
 		}
 
-		for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
-			if (RENDERINPUTS_TEST
-			    (inputs_bitset, _TNL_ATTRIB_TEX(i))) {
+		for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+			if (RENDERINPUTS_TEST(render_inputs_bitset, _TNL_ATTRIB_TEX(i))) {
 				InputsRead |= 1 << (VERT_ATTRIB_TEX0 + i);
 				OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i);
 			}
+		}
 
-		for (i = 0, nr = 0; i < VERT_ATTRIB_MAX; i++)
-			if (InputsRead & (1 << i))
+		for (i = 0, nr = 0; i < VERT_ATTRIB_MAX; i++) {
+			if (InputsRead & (1 << i)) {
 				inputs[i] = nr++;
-			else
+			} else {
 				inputs[i] = -1;
+			}
+		}
 
-		if (!
-		    (r300->radeon.radeonScreen->
-		     chip_flags & RADEON_CHIPSET_TCL)) {
+		if (!(rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL)) {
 			/* Fixed, apply to vir0 only */
-			memcpy(vir_inputs, inputs,
-			       VERT_ATTRIB_MAX * sizeof(int));
+			memcpy(vir_inputs, inputs, VERT_ATTRIB_MAX * sizeof(int));
 			inputs = vir_inputs;
-
 			if (InputsRead & VERT_ATTRIB_POS)
 				inputs[VERT_ATTRIB_POS] = 0;
-
 			if (InputsRead & (1 << VERT_ATTRIB_COLOR0))
 				inputs[VERT_ATTRIB_COLOR0] = 2;
-
 			if (InputsRead & (1 << VERT_ATTRIB_COLOR1))
 				inputs[VERT_ATTRIB_COLOR1] = 3;
-
 			for (i = VERT_ATTRIB_TEX0; i <= VERT_ATTRIB_TEX7; i++)
 				if (InputsRead & (1 << i))
 					inputs[i] = 6 + (i - VERT_ATTRIB_TEX0);
 		}
 
-		RENDERINPUTS_COPY(rmesa->state.render_inputs_bitset,
-				  inputs_bitset);
+		RENDERINPUTS_COPY(rmesa->state.render_inputs_bitset, render_inputs_bitset);
 	}
+
 	assert(InputsRead);
 	assert(OutputsWritten);
 
-	for (i = 0, nr = 0; i < VERT_ATTRIB_MAX; i++)
-		if (InputsRead & (1 << i))
+	for (i = 0, nr = 0; i < VERT_ATTRIB_MAX; i++) {
+		if (InputsRead & (1 << i)) {
 			tab[nr++] = i;
+		}
+	}
 
-	if (nr > R300_MAX_AOS_ARRAYS)
+	if (nr > R300_MAX_AOS_ARRAYS) {
 		return R300_FALLBACK_TCL;
+	}
 
 	for (i = 0; i < nr; i++) {
-		int ci;
-		int comp_size, fix, found = 0;
+		int ci, fix, found = 0;
 
 		swizzle[i][0] = SWIZZLE_ZERO;
 		swizzle[i][1] = SWIZZLE_ZERO;
 		swizzle[i][2] = SWIZZLE_ZERO;
 		swizzle[i][3] = SWIZZLE_ONE;
 
-		for (ci = 0; ci < vb->AttribPtr[tab[i]]->size; ci++)
+		for (ci = 0; ci < vb->AttribPtr[tab[i]]->size; ci++) {
 			swizzle[i][ci] = ci;
+		}
 
-		if (r300IsGartMemory(rmesa, vb->AttribPtr[tab[i]]->data,
-				     /*(count-1)*stride */ 4)) {
-			if (vb->AttribPtr[tab[i]]->stride % 4)
+		if (r300IsGartMemory(rmesa, vb->AttribPtr[tab[i]]->data, 4)) {
+			if (vb->AttribPtr[tab[i]]->stride % 4) {
 				return R300_FALLBACK_TCL;
-
-			rmesa->state.aos[i].address =
-			    (void *)(vb->AttribPtr[tab[i]]->data);
+			}
+			rmesa->state.aos[i].address = (void *)(vb->AttribPtr[tab[i]]->data);
 			rmesa->state.aos[i].start = 0;
-			rmesa->state.aos[i].aos_offset =
-			    r300GartOffsetFromVirtual(rmesa,
-						      vb->
-						      AttribPtr[tab[i]]->data);
-			rmesa->state.aos[i].aos_stride =
-			    vb->AttribPtr[tab[i]]->stride / 4;
-
-			rmesa->state.aos[i].aos_size =
-			    vb->AttribPtr[tab[i]]->size;
+			rmesa->state.aos[i].aos_offset = r300GartOffsetFromVirtual(rmesa, vb->AttribPtr[tab[i]]->data);
+			rmesa->state.aos[i].aos_stride = vb->AttribPtr[tab[i]]->stride / 4;
+			rmesa->state.aos[i].aos_size = vb->AttribPtr[tab[i]]->size;
 		} else {
 			r300EmitVec(ctx, &rmesa->state.aos[i],
 				    vb->AttribPtr[tab[i]]->data,
@@ -453,13 +453,10 @@ int r300EmitArrays(GLcontext * ctx)
 
 		rmesa->state.aos[i].aos_size = vb->AttribPtr[tab[i]]->size;
 
-		comp_size = _mesa_sizeof_type(GL_FLOAT);
-
 		for (fix = 0; fix <= 4 - vb->AttribPtr[tab[i]]->size; fix++) {
-			if ((rmesa->state.aos[i].aos_offset -
-			     comp_size * fix) % 4)
+			if ((rmesa->state.aos[i].aos_offset - _mesa_sizeof_type(GL_FLOAT) * fix) % 4) {
 				continue;
-
+			}
 			found = 1;
 			break;
 		}
@@ -468,11 +465,10 @@ int r300EmitArrays(GLcontext * ctx)
 			if (fix > 0) {
 				WARN_ONCE("Feeling lucky?\n");
 			}
-
-			rmesa->state.aos[i].aos_offset -= comp_size * fix;
-
-			for (ci = 0; ci < vb->AttribPtr[tab[i]]->size; ci++)
+			rmesa->state.aos[i].aos_offset -= _mesa_sizeof_type(GL_FLOAT) * fix;
+			for (ci = 0; ci < vb->AttribPtr[tab[i]]->size; ci++) {
 				swizzle[i][ci] += fix;
+			}
 		} else {
 			WARN_ONCE
 			    ("Cannot handle offset %x with stride %d, comp %d\n",
@@ -483,55 +479,27 @@ int r300EmitArrays(GLcontext * ctx)
 		}
 	}
 
-	/* setup INPUT_ROUTE */
-	R300_STATECHANGE(r300, vir[0]);
-	((drm_r300_cmd_header_t *) r300->hw.vir[0].cmd)->packet0.count =
-	    t_vir0(&r300->hw.vir[0].cmd[R300_VIR_CNTL_0], vb->AttribPtr,
-		   inputs, tab, nr);
-
-	R300_STATECHANGE(r300, vir[1]);
-	((drm_r300_cmd_header_t *) r300->hw.vir[1].cmd)->packet0.count =
-	    t_vir1(&r300->hw.vir[1].cmd[R300_VIR_CNTL_0], swizzle, nr);
-
-	/* Set up input_cntl */
-	/* I don't think this is needed for vertex buffers, but it doesn't hurt anything */
-	R300_STATECHANGE(r300, vic);
-	r300->hw.vic.cmd[R300_VIC_CNTL_0] = 0x5555;	/* Hard coded value, no idea what it means */
-	r300->hw.vic.cmd[R300_VIC_CNTL_1] = t_vic(ctx, InputsRead);
-
-	/* Stage 3: VAP output */
-
-	R300_STATECHANGE(r300, vof);
-
-	r300->hw.vof.cmd[R300_VOF_CNTL_0] = 0;
-	r300->hw.vof.cmd[R300_VOF_CNTL_1] = 0;
-
-	if (OutputsWritten & (1 << VERT_RESULT_HPOS))
-		r300->hw.vof.cmd[R300_VOF_CNTL_0] |=
-		    R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT;
-
-	if (OutputsWritten & (1 << VERT_RESULT_COL0))
-		r300->hw.vof.cmd[R300_VOF_CNTL_0] |=
-		    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT;
-
-	if (OutputsWritten & (1 << VERT_RESULT_COL1))
-		r300->hw.vof.cmd[R300_VOF_CNTL_0] |=
-		    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT;
-
-	/*if(OutputsWritten & (1 << VERT_RESULT_BFC0))
-	   r300->hw.vof.cmd[R300_VOF_CNTL_0] |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT;
-
-	   if(OutputsWritten & (1 << VERT_RESULT_BFC1))
-	   r300->hw.vof.cmd[R300_VOF_CNTL_0] |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT; */
-	//if(OutputsWritten & (1 << VERT_RESULT_FOGC))
-
-	if (OutputsWritten & (1 << VERT_RESULT_PSIZ))
-		r300->hw.vof.cmd[R300_VOF_CNTL_0] |=
-		    R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT;
-
-	for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
-		if (OutputsWritten & (1 << (VERT_RESULT_TEX0 + i)))
-			r300->hw.vof.cmd[R300_VOF_CNTL_1] |= (4 << (3 * i));
+	/* Setup INPUT_ROUTE. */
+	R300_STATECHANGE(rmesa, vir[0]);
+	((drm_r300_cmd_header_t *) rmesa->hw.vir[0].cmd)->packet0.count =
+	    r300VAPInputRoute0(&rmesa->hw.vir[0].cmd[R300_VIR_CNTL_0],
+			       vb->AttribPtr, inputs, tab, nr);
+	R300_STATECHANGE(rmesa, vir[1]);
+	((drm_r300_cmd_header_t *) rmesa->hw.vir[1].cmd)->packet0.count =
+	    r300VAPInputRoute1(&rmesa->hw.vir[1].cmd[R300_VIR_CNTL_0], swizzle,
+			       nr);
+
+	/* Setup INPUT_CNTL. */
+	R300_STATECHANGE(rmesa, vic);
+	rmesa->hw.vic.cmd[R300_VIC_CNTL_0] = r300VAPInputCntl0(ctx, InputsRead);
+	rmesa->hw.vic.cmd[R300_VIC_CNTL_1] = r300VAPInputCntl1(ctx, InputsRead);
+
+	/* Setup OUTPUT_VTX_FMT. */
+	R300_STATECHANGE(rmesa, vof);
+	rmesa->hw.vof.cmd[R300_VOF_CNTL_0] =
+	    r300VAPOutputCntl0(ctx, OutputsWritten);
+	rmesa->hw.vof.cmd[R300_VOF_CNTL_1] =
+	    r300VAPOutputCntl1(ctx, OutputsWritten);
 
 	rmesa->state.aos_count = nr;
 
diff --git a/src/mesa/drivers/dri/r300/r300_emit.h b/src/mesa/drivers/dri/r300/r300_emit.h
index 4f841a54139..2f79ee3a23a 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.h
+++ b/src/mesa/drivers/dri/r300/r300_emit.h
@@ -44,20 +44,11 @@
 #include "r300_cmdbuf.h"
 #include "radeon_reg.h"
 
-/*
- * CP type-3 packets
- */
-#define RADEON_CP_PACKET3_UNK1B                     0xC0001B00
-#define RADEON_CP_PACKET3_INDX_BUFFER               0xC0003300
-#define RADEON_CP_PACKET3_3D_DRAW_VBUF_2            0xC0003400
-#define RADEON_CP_PACKET3_3D_DRAW_IMMD_2            0xC0003500
-#define RADEON_CP_PACKET3_3D_DRAW_INDX_2            0xC0003600
-#define RADEON_CP_PACKET3_3D_LOAD_VBPNTR            0xC0002F00
-#define RADEON_CP_PACKET3_3D_CLEAR_ZMASK            0xC0003202
-#define RADEON_CP_PACKET3_3D_CLEAR_CMASK            0xC0003802
-#define RADEON_CP_PACKET3_3D_CLEAR_HIZ              0xC0003702
-
+/* TODO: move these defines (and the ones from DRM) into r300_reg.h and sync up
+ * with DRM */
 #define CP_PACKET0(reg, n)	(RADEON_CP_PACKET0 | ((n)<<16) | ((reg)>>2))
+#define CP_PACKET3( pkt, n )						\
+	(RADEON_CP_PACKET3 | (pkt) | ((n) << 16))
 
 static inline uint32_t cmdpacket0(int reg, int count)
 {
diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.c b/src/mesa/drivers/dri/r300/r300_ioctl.c
index 416ea7f231d..15c2cf3ad7b 100644
--- a/src/mesa/drivers/dri/r300/r300_ioctl.c
+++ b/src/mesa/drivers/dri/r300/r300_ioctl.c
@@ -224,25 +224,23 @@ static void r300EmitClearState(GLcontext * ctx)
 	e32(R300_INPUT_CNTL_0_COLOR);
 	e32(R300_INPUT_CNTL_POS | R300_INPUT_CNTL_COLOR | R300_INPUT_CNTL_TC0);
 
-	if (!has_tcl) {
-		R300_STATECHANGE(r300, vte);
-		/* comes from fglrx startup of clear */
-		reg_start(R300_SE_VTE_CNTL, 1);
-		e32(R300_VTX_W0_FMT | R300_VPORT_X_SCALE_ENA |
-		    R300_VPORT_X_OFFSET_ENA | R300_VPORT_Y_SCALE_ENA |
-		    R300_VPORT_Y_OFFSET_ENA | R300_VPORT_Z_SCALE_ENA |
-		    R300_VPORT_Z_OFFSET_ENA);
-		e32(0x8);
-
-		reg_start(0x21dc, 0);
-		e32(0xaaaaaaaa);
-	}
+	R300_STATECHANGE(r300, vte);
+	/* comes from fglrx startup of clear */
+	reg_start(R300_SE_VTE_CNTL, 1);
+	e32(R300_VTX_W0_FMT | R300_VPORT_X_SCALE_ENA |
+	    R300_VPORT_X_OFFSET_ENA | R300_VPORT_Y_SCALE_ENA |
+	    R300_VPORT_Y_OFFSET_ENA | R300_VPORT_Z_SCALE_ENA |
+	    R300_VPORT_Z_OFFSET_ENA);
+	e32(0x8);
+
+	reg_start(0x21dc, 0);
+	e32(0xaaaaaaaa);
 
 	R300_STATECHANGE(r300, vof);
 	reg_start(R300_VAP_OUTPUT_VTX_FMT_0, 1);
 	e32(R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT |
 	    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT);
-	e32(0x0);			/* no textures */
+	e32(0x0);		/* no textures */
 
 	R300_STATECHANGE(r300, txe);
 	reg_start(R300_TX_ENABLE, 0);
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index e64f5095bc5..3ce09c16d3f 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -116,6 +116,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT (1<<16) /* GUESS */
 
 #define R300_VAP_OUTPUT_VTX_FMT_1           0x2094
+	/* each of the following is 3 bits wide, specifies number
+	   of components */
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_0_COMP_CNT_SHIFT 0
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_1_COMP_CNT_SHIFT 3
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_2_COMP_CNT_SHIFT 6
@@ -299,6 +301,18 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_221C_NORMAL                  0x00000000
 #       define R300_221C_CLEAR                   0x0001C000
 
+/* These seem to be per-pixel and per-vertex X and Y clipping planes. The first
+ * plane is per-pixel and the second plane is per-vertex.
+ *
+ * This was determined by experimentation alone but I believe it is correct.
+ *
+ * These registers are called X_QUAD0_1_FL to X_QUAD0_4_FL by glxtest.
+ */
+#define R300_VAP_CLIP_X_0                   0x2220
+#define R300_VAP_CLIP_X_1                   0x2224
+#define R300_VAP_CLIP_Y_0                   0x2228
+#define R300_VAP_CLIP_Y_1                   0x2230
+
 /* gap */
 
 /* Sometimes, END_OF_PKT and 0x2284=0 are the only commands sent between
@@ -628,11 +642,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Set INTERP_USED on all interpolators that produce data used by
  * the fragment program. INTERP_USED looks like a swizzling mask,
  * but I haven't seen it used that way.
+ *
+ * Note: The _UNKNOWN constants are always set in their respective
+ * register. I don't know if this is necessary.
  */
 #define R300_RS_INTERP_0                    0x4310
 #define R300_RS_INTERP_1                    0x4314
+#       define R300_RS_INTERP_1_UNKNOWN          0x40
 #define R300_RS_INTERP_2                    0x4318
+#       define R300_RS_INTERP_2_UNKNOWN          0x80
 #define R300_RS_INTERP_3                    0x431C
+#       define R300_RS_INTERP_3_UNKNOWN          0xC0
 #define R300_RS_INTERP_4                    0x4320
 #define R300_RS_INTERP_5                    0x4324
 #define R300_RS_INTERP_6                    0x4328
@@ -961,7 +981,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * first node is stored in NODE_2, the second node is stored in NODE_3.
  *
  * Offsets are relative to the master offset from PFS_CNTL_2.
- * LAST_NODE is set for the last node, and only for the last node.
  */
 #define R300_PFS_NODE_0                     0x4610
 #define R300_PFS_NODE_1                     0x4614
@@ -975,7 +994,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_PFS_NODE_TEX_OFFSET_MASK     (31 << 12)
 #       define R300_PFS_NODE_TEX_END_SHIFT       17
 #       define R300_PFS_NODE_TEX_END_MASK        (31 << 17)
-/*#       define R300_PFS_NODE_LAST_NODE           (1 << 22) */
 #		define R300_PFS_NODE_OUTPUT_COLOR        (1 << 22)
 #		define R300_PFS_NODE_OUTPUT_DEPTH        (1 << 23)
 
@@ -1585,6 +1603,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #    define R300_EB_UNK1_SHIFT                      24
 #    define R300_EB_UNK1                    (0x80<<24)
 #    define R300_EB_UNK2                        0x0810
+#define R300_PACKET3_3D_DRAW_VBUF_2         0x00003400
 #define R300_PACKET3_3D_DRAW_INDX_2         0x00003600
 
 /* END: Packet 3 commands */
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 143fe9fd35d..83999307b56 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -171,16 +171,13 @@ static int r300NumVerts(r300ContextPtr rmesa, int num_verts, int prim)
 	return num_verts - verts_off;
 }
 
-static void r300EmitElts(GLcontext * ctx, void *elts, unsigned long n_elts,
-			 int elt_size)
+static void r300EmitElts(GLcontext * ctx, void *elts, unsigned long n_elts)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct r300_dma_region *rvb = &rmesa->state.elt_dma;
 	void *out;
 
-	assert(elt_size == 2 || elt_size == 4);
-
-	if (r300IsGartMemory(rmesa, elts, n_elts * elt_size)) {
+	if (r300IsGartMemory(rmesa, elts, n_elts * 4)) {
 		rvb->address = rmesa->radeon.radeonScreen->gartTextures.map;
 		rvb->start = ((char *)elts) - rvb->address;
 		rvb->aos_offset =
@@ -192,66 +189,27 @@ static void r300EmitElts(GLcontext * ctx, void *elts, unsigned long n_elts,
 		_mesa_exit(-1);
 	}
 
-	r300AllocDmaRegion(rmesa, rvb, n_elts * elt_size, elt_size);
+	r300AllocDmaRegion(rmesa, rvb, n_elts * 4, 4);
 	rvb->aos_offset = GET_START(rvb);
 
 	out = rvb->address + rvb->start;
-	memcpy(out, elts, n_elts * elt_size);
+	memcpy(out, elts, n_elts * 4);
 }
 
 static void r300FireEB(r300ContextPtr rmesa, unsigned long addr,
-		       int vertex_count, int type, int elt_size)
+		       int vertex_count, int type)
 {
 	int cmd_reserved = 0;
 	int cmd_written = 0;
 	drm_radeon_cmd_header_t *cmd = NULL;
-	unsigned long t_addr;
-	unsigned long magic_1, magic_2;
-
-	assert(elt_size == 2 || elt_size == 4);
-
-	if (addr & (elt_size - 1)) {
-		WARN_ONCE("Badly aligned buffer\n");
-		return;
-	}
 
-	magic_1 = (addr % 32) / 4;
-	t_addr = addr & ~0x1d;
-	magic_2 = (vertex_count + 1 + (t_addr & 0x2)) / 2 + magic_1;
+	start_packet3(CP_PACKET3(R300_PACKET3_3D_DRAW_INDX_2, 0), 0);
+	e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count << 16) | type | R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
 
-	start_packet3(RADEON_CP_PACKET3_3D_DRAW_INDX_2, 0);
-	if (elt_size == 4) {
-		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES |
-		    (vertex_count << 16) | type |
-		    R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
-	} else {
-		e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES |
-		    (vertex_count << 16) | type);
-	}
-
-	start_packet3(RADEON_CP_PACKET3_INDX_BUFFER, 2);
-#ifdef OPTIMIZE_ELTS
-	if (elt_size == 4) {
-		e32(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
-		e32(addr);
-	} else {
-		e32(R300_EB_UNK1 | (magic_1 << 16) | R300_EB_UNK2);
-		e32(t_addr);
-	}
-#else
+	start_packet3(CP_PACKET3(R300_PACKET3_INDX_BUFFER, 2), 2);
 	e32(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
 	e32(addr);
-#endif
-
-	if (elt_size == 4) {
-		e32(vertex_count);
-	} else {
-#ifdef OPTIMIZE_ELTS
-		e32(magic_2);
-#else
-		e32((vertex_count + 1) / 2);
-#endif
-	}
+	e32(vertex_count);
 }
 
 static void r300EmitAOS(r300ContextPtr rmesa, GLuint nr, GLuint offset)
@@ -266,26 +224,23 @@ static void r300EmitAOS(r300ContextPtr rmesa, GLuint nr, GLuint offset)
 		fprintf(stderr, "%s: nr=%d, ofs=0x%08x\n", __FUNCTION__, nr,
 			offset);
 
-	start_packet3(RADEON_CP_PACKET3_3D_LOAD_VBPNTR, sz - 1);
+	start_packet3(CP_PACKET3(R300_PACKET3_3D_LOAD_VBPNTR, sz - 1), sz - 1);
 	e32(nr);
+
 	for (i = 0; i + 1 < nr; i += 2) {
-		e32((rmesa->state.aos[i].aos_size << 0)
-		    | (rmesa->state.aos[i].aos_stride << 8)
-		    | (rmesa->state.aos[i + 1].aos_size << 16)
-		    | (rmesa->state.aos[i + 1].aos_stride << 24)
-		    );
-		e32(rmesa->state.aos[i].aos_offset +
-		    offset * 4 * rmesa->state.aos[i].aos_stride);
-		e32(rmesa->state.aos[i + 1].aos_offset +
-		    offset * 4 * rmesa->state.aos[i + 1].aos_stride);
+		e32((rmesa->state.aos[i].aos_size << 0) |
+		    (rmesa->state.aos[i].aos_stride << 8) |
+		    (rmesa->state.aos[i + 1].aos_size << 16) |
+		    (rmesa->state.aos[i + 1].aos_stride << 24));
+
+		e32(rmesa->state.aos[i].aos_offset + offset * 4 * rmesa->state.aos[i].aos_stride);
+		e32(rmesa->state.aos[i + 1].aos_offset + offset * 4 * rmesa->state.aos[i + 1].aos_stride);
 	}
 
 	if (nr & 1) {
-		e32((rmesa->state.aos[nr - 1].aos_size << 0)
-		    | (rmesa->state.aos[nr - 1].aos_stride << 8)
-		    );
-		e32(rmesa->state.aos[nr - 1].aos_offset +
-		    offset * 4 * rmesa->state.aos[nr - 1].aos_stride);
+		e32((rmesa->state.aos[nr - 1].aos_size << 0) |
+		    (rmesa->state.aos[nr - 1].aos_stride << 8));
+		e32(rmesa->state.aos[nr - 1].aos_offset + offset * 4 * rmesa->state.aos[nr - 1].aos_stride);
 	}
 }
 
@@ -295,9 +250,8 @@ static void r300FireAOS(r300ContextPtr rmesa, int vertex_count, int type)
 	int cmd_written = 0;
 	drm_radeon_cmd_header_t *cmd = NULL;
 
-	start_packet3(RADEON_CP_PACKET3_3D_DRAW_VBUF_2, 0);
-	e32(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (vertex_count << 16)
-	    | type);
+	start_packet3(CP_PACKET3(R300_PACKET3_3D_DRAW_VBUF_2, 0), 0);
+	e32(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (vertex_count << 16) | type);
 }
 
 static void r300RunRenderPrimitive(r300ContextPtr rmesa, GLcontext * ctx,
@@ -320,9 +274,8 @@ static void r300RunRenderPrimitive(r300ContextPtr rmesa, GLcontext * ctx,
 			WARN_ONCE("Too many elts\n");
 			return;
 		}
-		r300EmitElts(ctx, vb->Elts, num_verts, 4);
-		r300FireEB(rmesa, rmesa->state.elt_dma.aos_offset,
-			   num_verts, type, 4);
+		r300EmitElts(ctx, vb->Elts, num_verts);
+		r300FireEB(rmesa, rmesa->state.elt_dma.aos_offset, num_verts, type);
 	} else {
 		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
 		r300FireAOS(rmesa, num_verts, type);
@@ -343,6 +296,8 @@ static GLboolean r300RunRender(GLcontext * ctx,
 	if (RADEON_DEBUG & DEBUG_PRIMS)
 		fprintf(stderr, "%s\n", __FUNCTION__);
 
+	if (hw_tcl_on == GL_FALSE)
+	  vb->AttribPtr[VERT_ATTRIB_POS] = vb->ClipPtr;
 	r300UpdateShaders(rmesa);
 	if (r300EmitArrays(ctx))
 		return GL_TRUE;
@@ -416,8 +371,6 @@ static int r300Fallback(GLcontext * ctx)
 		FALLBACK_IF(ctx->Point.PointSprite);
 
 	if (!r300->disable_lowimpact_fallback) {
-		FALLBACK_IF(ctx->Polygon.OffsetPoint);
-		FALLBACK_IF(ctx->Polygon.OffsetLine);
 		FALLBACK_IF(ctx->Polygon.StippleFlag);
 		FALLBACK_IF(ctx->Multisample.Enabled);
 		FALLBACK_IF(ctx->Line.StippleFlag);
diff --git a/src/mesa/drivers/dri/r300/r300_shader.c b/src/mesa/drivers/dri/r300/r300_shader.c
index 59fe17ba10e..5f5ac7c4c71 100644
--- a/src/mesa/drivers/dri/r300/r300_shader.c
+++ b/src/mesa/drivers/dri/r300/r300_shader.c
@@ -54,6 +54,7 @@ r300ProgramStringNotify(GLcontext * ctx, GLenum target, struct gl_program *prog)
 		fp->translated = GL_FALSE;
 		break;
 	}
+
 	/* need this for tcl fallbacks */
 	_tnl_program_string(ctx, target, prog);
 }
@@ -61,7 +62,7 @@ r300ProgramStringNotify(GLcontext * ctx, GLenum target, struct gl_program *prog)
 static GLboolean
 r300IsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
-	return 1;
+	return GL_TRUE;
 }
 
 void r300InitShaderFuncs(struct dd_function_table *functions)
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index b235baaf10f..bdd6855802b 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -715,8 +715,8 @@ static void r300LineWidth(GLcontext * ctx, GLfloat widthf)
 	widthf = ctx->Line._Width;
 
 	R300_STATECHANGE(r300, lcntl);
-	r300->hw.lcntl.cmd[1] = (int)(widthf * 6.0);
-	r300->hw.lcntl.cmd[1] |= R300_LINE_CNT_VE;
+	r300->hw.lcntl.cmd[1] =
+	    R300_LINE_CNT_HO | R300_LINE_CNT_VE | (int)(widthf * 6.0);
 }
 
 static void r300PolygonMode(GLcontext * ctx, GLenum face, GLenum mode)
@@ -1354,6 +1354,17 @@ union r300_outputs_written {
 static void r300SetupRSUnit(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	/* I'm still unsure if these are needed */
+	GLuint interp_magic[8] = {
+		0x00,
+		R300_RS_INTERP_1_UNKNOWN,
+		R300_RS_INTERP_2_UNKNOWN,
+		R300_RS_INTERP_3_UNKNOWN,
+		0x00,
+		0x00,
+		0x00,
+		0x00
+	};
 	union r300_outputs_written OutputsWritten;
 	GLuint InputsRead;
 	int fp_reg, high_rr;
@@ -1361,11 +1372,9 @@ static void r300SetupRSUnit(GLcontext * ctx)
 	int i;
 
 	if (hw_tcl_on)
-		OutputsWritten.vp_outputs =
-		    CURRENT_VERTEX_SHADER(ctx)->key.OutputsWritten;
+		OutputsWritten.vp_outputs = CURRENT_VERTEX_SHADER(ctx)->key.OutputsWritten;
 	else
-		RENDERINPUTS_COPY(OutputsWritten.index_bitset,
-				  r300->state.render_inputs_bitset);
+		RENDERINPUTS_COPY(OutputsWritten.index_bitset, r300->state.render_inputs_bitset);
 
 	if (ctx->FragmentProgram._Current)
 		InputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
@@ -1397,9 +1406,8 @@ static void r300SetupRSUnit(GLcontext * ctx)
 	}
 
 	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] = 0
-		    | R300_RS_INTERP_USED
-		    | (in_texcoords << R300_RS_INTERP_SRC_SHIFT);
+		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] = 0 | R300_RS_INTERP_USED | (in_texcoords << R300_RS_INTERP_SRC_SHIFT)
+		    | interp_magic[i];
 
 		r300->hw.rr.cmd[R300_RR_ROUTE_0 + fp_reg] = 0;
 		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
@@ -1408,65 +1416,45 @@ static void r300SetupRSUnit(GLcontext * ctx)
 			    | (fp_reg << R300_RS_ROUTE_DEST_SHIFT);
 			high_rr = fp_reg;
 
-			if (!R300_OUTPUTS_WRITTEN_TEST
-			    (OutputsWritten, VERT_RESULT_TEX0 + i,
-			     _TNL_ATTRIB_TEX(i))) {
-				/* Passing invalid data here can lock the GPU. */
-				WARN_ONCE
-				    ("fragprog wants coords for tex%d, vp doesn't provide them!\n",
-				     i);
-				//_mesa_print_program(&CURRENT_VERTEX_SHADER(ctx)->Base);
-				//_mesa_exit(-1);
+			/* Passing invalid data here can lock the GPU. */
+			if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_TEX0 + i, _TNL_ATTRIB_TEX(i))) {
+				InputsRead &= ~(FRAG_BIT_TEX0 << i);
+				fp_reg++;
+			} else {
+				WARN_ONCE("fragprog wants coords for tex%d, vp doesn't provide them!\n", i);
 			}
-			InputsRead &= ~(FRAG_BIT_TEX0 << i);
-			fp_reg++;
 		}
 		/* Need to count all coords enabled at vof */
-		if (R300_OUTPUTS_WRITTEN_TEST
-		    (OutputsWritten, VERT_RESULT_TEX0 + i, _TNL_ATTRIB_TEX(i)))
+		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_TEX0 + i, _TNL_ATTRIB_TEX(i))) {
 			in_texcoords++;
+		}
 	}
 
 	if (InputsRead & FRAG_BIT_COL0) {
-		if (!R300_OUTPUTS_WRITTEN_TEST
-		    (OutputsWritten, VERT_RESULT_COL0, _TNL_ATTRIB_COLOR0)) {
-			WARN_ONCE
-			    ("fragprog wants col0, vp doesn't provide it\n");
-			goto out;	/* FIXME */
-			//_mesa_print_program(&CURRENT_VERTEX_SHADER(ctx)->Base);
-			//_mesa_exit(-1);
+		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL0, _TNL_ATTRIB_COLOR0)) {
+			r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0 | R300_RS_ROUTE_0_COLOR | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
+			InputsRead &= ~FRAG_BIT_COL0;
+			col_interp_nr++;
+		} else {
+			WARN_ONCE("fragprog wants col0, vp doesn't provide it\n");
 		}
-
-		r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0
-		    | R300_RS_ROUTE_0_COLOR
-		    | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
-		InputsRead &= ~FRAG_BIT_COL0;
-		col_interp_nr++;
 	}
-      out:
 
 	if (InputsRead & FRAG_BIT_COL1) {
-		if (!R300_OUTPUTS_WRITTEN_TEST
-		    (OutputsWritten, VERT_RESULT_COL1, _TNL_ATTRIB_COLOR1)) {
-			WARN_ONCE
-			    ("fragprog wants col1, vp doesn't provide it\n");
-			//_mesa_exit(-1);
+		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL1, _TNL_ATTRIB_COLOR1)) {
+			r300->hw.rr.cmd[R300_RR_ROUTE_1] |= R300_RS_ROUTE_1_UNKNOWN11 | R300_RS_ROUTE_1_COLOR1 | (fp_reg++ << R300_RS_ROUTE_1_COLOR1_DEST_SHIFT);
+			InputsRead &= ~FRAG_BIT_COL1;
+			if (high_rr < 1)
+				high_rr = 1;
+			col_interp_nr++;
+		} else {
+			WARN_ONCE("fragprog wants col1, vp doesn't provide it\n");
 		}
-
-		r300->hw.rr.cmd[R300_RR_ROUTE_1] |=
-		    R300_RS_ROUTE_1_UNKNOWN11 | R300_RS_ROUTE_1_COLOR1 |
-		    (fp_reg++ << R300_RS_ROUTE_1_COLOR1_DEST_SHIFT);
-		InputsRead &= ~FRAG_BIT_COL1;
-		if (high_rr < 1)
-			high_rr = 1;
-		col_interp_nr++;
 	}
 
 	/* Need at least one. This might still lock as the values are undefined... */
 	if (in_texcoords == 0 && col_interp_nr == 0) {
-		r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0
-		    | R300_RS_ROUTE_0_COLOR
-		    | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
+		r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0 | R300_RS_ROUTE_0_COLOR | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
 		col_interp_nr++;
 	}
 
@@ -1475,17 +1463,13 @@ static void r300SetupRSUnit(GLcontext * ctx)
 	    | R300_RS_CNTL_0_UNKNOWN_18;
 
 	assert(high_rr >= 0);
-	r300->hw.rr.cmd[R300_RR_CMD_0] =
-	    cmdpacket0(R300_RS_ROUTE_0, high_rr + 1);
+	r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R300_RS_ROUTE_0, high_rr + 1);
 	r300->hw.rc.cmd[2] = 0xC0 | high_rr;
 
 	if (InputsRead)
-		WARN_ONCE("Don't know how to satisfy InputsRead=0x%08x\n",
-			  InputsRead);
+		WARN_ONCE("Don't know how to satisfy InputsRead=0x%08x\n", InputsRead);
 }
 
-#define vpucount(ptr) (((drm_r300_cmd_header_t*)(ptr))->vpu.count)
-
 #define bump_vpu_count(ptr, new_count)   do{\
 	drm_r300_cmd_header_t* _p=((drm_r300_cmd_header_t*)(ptr));\
 	int _nc=(new_count)/4; \
@@ -1785,8 +1769,6 @@ static void r300Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 
 	case GL_POLYGON_OFFSET_POINT:
 	case GL_POLYGON_OFFSET_LINE:
-		break;
-
 	case GL_POLYGON_OFFSET_FILL:
 		R300_STATECHANGE(r300, occlusion_cntl);
 		if (state) {
@@ -1859,10 +1841,12 @@ static void r300ResetHwState(r300ContextPtr r300)
 
 	r300->hw.unk2134.cmd[1] = 0x00FFFFFF;
 	r300->hw.unk2134.cmd[2] = 0x00000000;
-	if (_mesa_little_endian())
-		r300->hw.vap_cntl_status.cmd[1] = R300_VC_NO_SWAP;
-	else
-		r300->hw.vap_cntl_status.cmd[1] = R300_VC_32BIT_SWAP;
+
+#ifdef MESA_LITTLE_ENDIAN
+	r300->hw.vap_cntl_status.cmd[1] = R300_VC_NO_SWAP;
+#else
+	r300->hw.vap_cntl_status.cmd[1] = R300_VC_32BIT_SWAP;
+#endif
 
 	/* disable VAP/TCL on non-TCL capable chips */
 	if (!has_tcl)
@@ -1872,10 +1856,10 @@ static void r300ResetHwState(r300ContextPtr r300)
 
 	r300->hw.unk221C.cmd[1] = R300_221C_NORMAL;
 
-	r300->hw.unk2220.cmd[1] = r300PackFloat32(1.0);
-	r300->hw.unk2220.cmd[2] = r300PackFloat32(1.0);
-	r300->hw.unk2220.cmd[3] = r300PackFloat32(1.0);
-	r300->hw.unk2220.cmd[4] = r300PackFloat32(1.0);
+	r300->hw.vap_clip.cmd[1] = r300PackFloat32(1.0); /* X */
+	r300->hw.vap_clip.cmd[2] = r300PackFloat32(1.0); /* X */
+	r300->hw.vap_clip.cmd[3] = r300PackFloat32(1.0); /* Y */
+	r300->hw.vap_clip.cmd[4] = r300PackFloat32(1.0); /* Y */
 
 	/* XXX: Other families? */
 	switch (r300->radeon.radeonScreen->chip_family) {
@@ -1930,13 +1914,13 @@ static void r300ResetHwState(r300ContextPtr r300)
 
 	r300->hw.unk4214.cmd[1] = 0x00050005;
 
-	r300PointSize(ctx, 0.0);
+	r300PointSize(ctx, 1.0);
 
 	r300->hw.unk4230.cmd[1] = 0x18000006;
 	r300->hw.unk4230.cmd[2] = 0x00020006;
 	r300->hw.unk4230.cmd[3] = r300PackFloat32(1.0 / 192.0);
 
-	r300LineWidth(ctx, 0.0);
+	r300LineWidth(ctx, 1.0);
 
 	r300->hw.unk4260.cmd[1] = 0;
 	r300->hw.unk4260.cmd[2] = r300PackFloat32(0.0);
@@ -2152,7 +2136,7 @@ static void r300SetupPixelShader(r300ContextPtr rmesa)
 			       tex_offset << R300_PFS_NODE_TEX_OFFSET_SHIFT)
 			    | (fp->node[i].
 			       tex_end << R300_PFS_NODE_TEX_END_SHIFT)
-			    | fp->node[i].flags;	/*  ( (k==3) ? R300_PFS_NODE_LAST_NODE : 0); */
+			    | fp->node[i].flags;
 		} else {
 			rmesa->hw.fp.cmd[R300_FP_NODE0 + (3 - i)] = 0;
 		}
diff --git a/src/mesa/drivers/dri/r300/r300_tex.c b/src/mesa/drivers/dri/r300/r300_tex.c
index 2a21c611629..1805cecd0af 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.c
+++ b/src/mesa/drivers/dri/r300/r300_tex.c
@@ -294,27 +294,20 @@ static const struct gl_texture_format *r300Choose8888TexFormat(GLenum srcFormat,
 	const GLubyte littleEndian = *((const GLubyte *)&ui);
 
 	if ((srcFormat == GL_RGBA && srcType == GL_UNSIGNED_INT_8_8_8_8) ||
-	    (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE
-	     && !littleEndian) || (srcFormat == GL_ABGR_EXT
-				   && srcType == GL_UNSIGNED_INT_8_8_8_8_REV)
-	    || (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE
-		&& littleEndian)) {
+	    (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
+	    (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_INT_8_8_8_8_REV) ||
+	    (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE && littleEndian)) {
 		return &_mesa_texformat_rgba8888;
-	} else
-	    if ((srcFormat == GL_RGBA && srcType == GL_UNSIGNED_INT_8_8_8_8_REV)
-		|| (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE
-		    && littleEndian) || (srcFormat == GL_ABGR_EXT
-					 && srcType == GL_UNSIGNED_INT_8_8_8_8)
-		|| (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE
-		    && !littleEndian)) {
+	} else if ((srcFormat == GL_RGBA && srcType == GL_UNSIGNED_INT_8_8_8_8_REV) ||
+		   (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE && littleEndian) ||
+		   (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_INT_8_8_8_8) ||
+		   (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE && !littleEndian)) {
 		return &_mesa_texformat_rgba8888_rev;
-	} else if (srcFormat == GL_BGRA &&
-		   ((srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
-		    srcType == GL_UNSIGNED_INT_8_8_8_8)) {
+	} else if (srcFormat == GL_BGRA && ((srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
+					    srcType == GL_UNSIGNED_INT_8_8_8_8)) {
 		return &_mesa_texformat_argb8888_rev;
-	} else if (srcFormat == GL_BGRA &&
-		   ((srcType == GL_UNSIGNED_BYTE && littleEndian) ||
-		    srcType == GL_UNSIGNED_INT_8_8_8_8_REV)) {
+	} else if (srcFormat == GL_BGRA && ((srcType == GL_UNSIGNED_BYTE && littleEndian) ||
+					    srcType == GL_UNSIGNED_INT_8_8_8_8_REV)) {
 		return &_mesa_texformat_argb8888;
 	} else
 		return _dri_texformat_argb8888;
@@ -563,34 +556,31 @@ r300ValidateClientStorage(GLcontext * ctx, GLenum target,
 		return 0;
 	}
 
-	{
-		GLint srcRowStride = _mesa_image_row_stride(packing, srcWidth,
-							    format, type);
+	GLint srcRowStride = _mesa_image_row_stride(packing, srcWidth,
+						    format, type);
 
-		if (RADEON_DEBUG & DEBUG_TEXTURE)
-			fprintf(stderr, "%s: srcRowStride %d/%x\n",
-				__FUNCTION__, srcRowStride, srcRowStride);
+	if (RADEON_DEBUG & DEBUG_TEXTURE)
+		fprintf(stderr, "%s: srcRowStride %d/%x\n",
+			__FUNCTION__, srcRowStride, srcRowStride);
 
-		/* Could check this later in upload, pitch restrictions could be
-		 * relaxed, but would need to store the image pitch somewhere,
-		 * as packing details might change before image is uploaded:
-		 */
-		if (!r300IsGartMemory(rmesa, pixels, srcHeight * srcRowStride)
-		    || (srcRowStride & 63))
-			return 0;
+	/* Could check this later in upload, pitch restrictions could be
+	 * relaxed, but would need to store the image pitch somewhere,
+	 * as packing details might change before image is uploaded:
+	 */
+	if (!r300IsGartMemory(rmesa, pixels, srcHeight * srcRowStride)
+	    || (srcRowStride & 63))
+		return 0;
 
-		/* Have validated that _mesa_transfer_teximage would be a straight
-		 * memcpy at this point.  NOTE: future calls to TexSubImage will
-		 * overwrite the client data.  This is explicitly mentioned in the
-		 * extension spec.
-		 */
-		texImage->Data = (void *)pixels;
-		texImage->IsClientData = GL_TRUE;
-		texImage->RowStride =
-		    srcRowStride / texImage->TexFormat->TexelBytes;
+	/* Have validated that _mesa_transfer_teximage would be a straight
+	 * memcpy at this point.  NOTE: future calls to TexSubImage will
+	 * overwrite the client data.  This is explicitly mentioned in the
+	 * extension spec.
+	 */
+	texImage->Data = (void *)pixels;
+	texImage->IsClientData = GL_TRUE;
+	texImage->RowStride = srcRowStride / texImage->TexFormat->TexelBytes;
 
-		return 1;
-	}
+	return 1;
 }
 
 static void r300TexImage1D(GLcontext * ctx, GLenum target, GLint level,
diff --git a/src/mesa/drivers/dri/r300/r300_texmem.c b/src/mesa/drivers/dri/r300/r300_texmem.c
index e2e8355d275..38f0da8b7c0 100644
--- a/src/mesa/drivers/dri/r300/r300_texmem.c
+++ b/src/mesa/drivers/dri/r300/r300_texmem.c
@@ -63,29 +63,16 @@ SOFTWARE.
  */
 void r300DestroyTexObj(r300ContextPtr rmesa, r300TexObjPtr t)
 {
+	int i;
+
 	if (RADEON_DEBUG & DEBUG_TEXTURE) {
 		fprintf(stderr, "%s( %p, %p )\n", __FUNCTION__,
 			(void *)t, (void *)t->base.tObj);
 	}
 
-	if (rmesa != NULL) {
-		unsigned i;
-
-		for (i = 0; i < rmesa->radeon.glCtx->Const.MaxTextureUnits; i++) {
-			if (t == rmesa->state.texture.unit[i].texobj) {
-				rmesa->state.texture.unit[i].texobj = NULL;
-				/* This code below is meant to shorten state
-				   pushed to the hardware by not programming
-				   unneeded units.
-
-				   This does not appear to be worthwhile on R300 */
-#if 0
-				remove_from_list(&rmesa->hw.tex[i]);
-				make_empty_list(&rmesa->hw.tex[i]);
-				remove_from_list(&rmesa->hw.cube[i]);
-				make_empty_list(&rmesa->hw.cube[i]);
-#endif
-			}
+	for (i = 0; i < rmesa->radeon.glCtx->Const.MaxTextureUnits; i++) {
+		if (rmesa->state.texture.unit[i].texobj == t) {
+			rmesa->state.texture.unit[i].texobj = NULL;
 		}
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/r300_texstate.c b/src/mesa/drivers/dri/r300/r300_texstate.c
index eeaba584df8..1d2909fd214 100644
--- a/src/mesa/drivers/dri/r300/r300_texstate.c
+++ b/src/mesa/drivers/dri/r300/r300_texstate.c
@@ -480,11 +480,11 @@ static GLboolean r300UpdateTexture(GLcontext * ctx, int unit)
 			 */
 
 			rmesa->state.texture.unit[unit].texobj->base.bound &=
-			    ~(1UL << unit);
+			    ~(1 << unit);
 		}
 
 		rmesa->state.texture.unit[unit].texobj = t;
-		t->base.bound |= (1UL << unit);
+		t->base.bound |= (1 << unit);
 		t->dirty_state |= 1 << unit;
 		driUpdateTextureLRU((driTextureObject *) t);	/* XXX: should be locked! */
 	}
@@ -501,7 +501,6 @@ void r300SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	struct gl_texture_object *tObj =
 	    _mesa_lookup_texture(rmesa->radeon.glCtx, texname);
 	r300TexObjPtr t;
-	int idx;
 
 	if (!tObj)
 		return;
@@ -518,24 +517,24 @@ void r300SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 
 	switch (depth) {
 	case 32:
-		idx = 2;
+		t->format = R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
+		t->filter |= tx_table[2].filter;
 		t->pitch_reg /= 4;
 		break;
 	case 24:
 	default:
-		idx = 4;
+		t->format = R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
+		t->filter |= tx_table[4].filter;
 		t->pitch_reg /= 4;
 		break;
 	case 16:
-		idx = 5;
+		t->format = R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
+		t->filter |= tx_table[5].filter;
 		t->pitch_reg /= 2;
 		break;
 	}
 
 	t->pitch_reg--;
-
-	t->format = tx_table[idx].format;
-	t->filter |= tx_table[idx].filter;
 }
 
 static GLboolean r300UpdateTextureUnit(GLcontext * ctx, int unit)
diff --git a/src/mesa/drivers/x11/glxapi.c b/src/mesa/drivers/x11/glxapi.c
index 5f11c90c13c..309a0008d76 100644
--- a/src/mesa/drivers/x11/glxapi.c
+++ b/src/mesa/drivers/x11/glxapi.c
@@ -141,7 +141,7 @@ static void
 SetCurrentContext(GLXContext c)
 {
 #if defined(GLX_USE_TLS)
-   CurrentContext = context;
+   CurrentContext = c;
 #elif defined(THREADS)
    _glthread_SetTSD(&ContextTSD, c);
 #else
@@ -1169,7 +1169,7 @@ _glxapi_get_extensions(void)
 #ifdef GLX_SGIX_pbuffer
       "GLX_SGIX_pbuffer",
 #endif
-#ifdef GLX_EXT_texture_from_pixmap,
+#ifdef GLX_EXT_texture_from_pixmap
       "GLX_EXT_texture_from_pixmap",
 #endif
       NULL
diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c
index cff64d17ade..eaa277db4ad 100644
--- a/src/mesa/drivers/x11/xm_api.c
+++ b/src/mesa/drivers/x11/xm_api.c
@@ -63,7 +63,6 @@
 #endif
 
 #include "glxheader.h"
-#include "GL/glxtokens.h"
 #include "GL/xmesa.h"
 #include "xmesaP.h"
 #include "context.h"
diff --git a/src/mesa/glapi/gl_x86_asm.py b/src/mesa/glapi/gl_x86_asm.py
index 650331a0c5a..403e87261b0 100644
--- a/src/mesa/glapi/gl_x86_asm.py
+++ b/src/mesa/glapi/gl_x86_asm.py
@@ -82,12 +82,18 @@ class PrintGenericStubs(gl_XML.gl_print_base):
 		print ''
 		print '#ifdef GLX_USE_TLS'
 		print ''
+		print '#ifdef GLX_X86_READONLY_TEXT'
+		print '# define CTX_INSNS MOV_L(GS:(EAX), EAX)'
+		print '#else'
+		print '# define CTX_INSNS NOP /* Pad for init_glapi_relocs() */'
+		print '#endif'
+		print ''
 		print '#  define GL_STUB(fn,off,fn_alt)\t\t\t\\'
 		print 'ALIGNTEXT16;\t\t\t\t\t\t\\'
 		print 'GLOBL_FN(GL_PREFIX(fn, fn_alt));\t\t\t\\'
 		print 'GL_PREFIX(fn, fn_alt):\t\t\t\t\t\\'
 		print '\tCALL(_x86_get_dispatch) ;\t\t\t\\'
-		print '\tNOP ;\t\t\t\t\t\t\\'
+		print '\tCTX_INSNS ;					\\'
 		print '\tJMP(GL_OFFSET(off))'
 		print ''
 		print '#elif defined(PTHREADS)'
@@ -138,7 +144,10 @@ class PrintGenericStubs(gl_XML.gl_print_base):
 		print '\tHIDDEN(GLNAME(_x86_get_dispatch))'
 		print 'ALIGNTEXT16'
 		print 'GLNAME(_x86_get_dispatch):'
-		print '\tmovl\t%gs:_glapi_tls_Dispatch@NTPOFF, %eax'
+		print '\tcall	1f'
+		print '1:\tpopl	%eax'
+		print '\taddl	$_GLOBAL_OFFSET_TABLE_+[.-1b], %eax'
+		print '\tmovl	_glapi_tls_Dispatch@GOTNTPOFF(%eax), %eax'
 		print '\tret'
 		print ''
 		print '#elif defined(PTHREADS)'
@@ -158,7 +167,7 @@ class PrintGenericStubs(gl_XML.gl_print_base):
 		print '#endif'
 		print ''
 
-		print '#if defined( GLX_USE_TLS )'
+		print '#if defined( GLX_USE_TLS ) && !defined( GLX_X86_READONLY_TEXT )'
 		print '\t\t.section\twtext, "awx", @progbits'
 		print '#endif /* defined( GLX_USE_TLS ) */'
 
diff --git a/src/mesa/glapi/glapi.c b/src/mesa/glapi/glapi.c
index 5815dbff842..47c57822732 100644
--- a/src/mesa/glapi/glapi.c
+++ b/src/mesa/glapi/glapi.c
@@ -1028,22 +1028,24 @@ _glapi_check_table(const struct _glapi_table *table)
 #if defined(PTHREADS) || defined(GLX_USE_TLS)
 /**
  * Perform platform-specific GL API entry-point fixups.
- * 
- * 
  */
 static void
 init_glapi_relocs( void )
 {
-#if defined( USE_X86_ASM ) && defined( GLX_USE_TLS )
-    extern void * _x86_get_dispatch(void);
-    const GLubyte * const get_disp = (const GLubyte *) _x86_get_dispatch;
+#if defined(USE_X86_ASM) && defined(GLX_USE_TLS) && !defined(GLX_X86_READONLY_TEXT)
+    extern unsigned long _x86_get_dispatch(void);
+    char run_time_patch[] = {
+       0x65, 0xa1, 0, 0, 0, 0 /* movl %gs:0,%eax */
+    };
+    GLuint *offset = (GLuint *) &run_time_patch[2]; /* 32-bits for x86/32 */
+    const GLubyte * const get_disp = (const GLubyte *) run_time_patch;
     GLubyte * curr_func = (GLubyte *) gl_dispatch_functions_start;
 
-
+    *offset = _x86_get_dispatch();
     while ( curr_func != (GLubyte *) gl_dispatch_functions_end ) {
-	(void) memcpy( curr_func, get_disp, 6 );
+	(void) memcpy( curr_func, get_disp, sizeof(run_time_patch));
 	curr_func += DISPATCH_FUNCTION_SIZE;
     }
-#endif /* defined( USE_X86_ASM ) && defined( GLX_USE_TLS ) */
-}
 #endif
+}
+#endif /* defined(PTHREADS) || defined(GLX_USE_TLS) */
diff --git a/src/mesa/main/buffers.c b/src/mesa/main/buffers.c
index 11bd173e35c..c280f89e1d1 100644
--- a/src/mesa/main/buffers.c
+++ b/src/mesa/main/buffers.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"),
@@ -184,17 +184,19 @@ _mesa_Clear( GLbitfield mask )
 
 /**
  * Return bitmask of BUFFER_BIT_* flags indicating which color buffers are
- * available to the rendering context.
- * This depends on the framebuffer we're writing to.  For window system
- * framebuffers we look at the framebuffer's visual.  But for user-
- * create framebuffers we look at the number of supported color attachments.
+ * available to the rendering context (for drawing or reading).
+ * This depends on the type of framebuffer.  For window system framebuffers
+ * we look at the framebuffer's visual.  But for user-create framebuffers we
+ * look at the number of supported color attachments.
+ * \param fb  the framebuffer to draw to, or read from
+ * \return  bitmask of BUFFER_BIT_* flags
  */
 static GLbitfield
-supported_buffer_bitmask(const GLcontext *ctx, GLuint framebufferID)
+supported_buffer_bitmask(const GLcontext *ctx, const struct gl_framebuffer *fb)
 {
    GLbitfield mask = 0x0;
 
-   if (framebufferID > 0) {
+   if (fb->Name > 0) {
       /* A user-created renderbuffer */
       GLuint i;
       ASSERT(ctx->Extensions.EXT_framebuffer_object);
@@ -203,20 +205,20 @@ supported_buffer_bitmask(const GLcontext *ctx, GLuint framebufferID)
       }
    }
    else {
-      /* A window system renderbuffer */
+      /* A window system framebuffer */
       GLint i;
       mask = BUFFER_BIT_FRONT_LEFT; /* always have this */
-      if (ctx->Visual.stereoMode) {
+      if (fb->Visual.stereoMode) {
          mask |= BUFFER_BIT_FRONT_RIGHT;
-         if (ctx->Visual.doubleBufferMode) {
+         if (fb->Visual.doubleBufferMode) {
             mask |= BUFFER_BIT_BACK_LEFT | BUFFER_BIT_BACK_RIGHT;
          }
       }
-      else if (ctx->Visual.doubleBufferMode) {
+      else if (fb->Visual.doubleBufferMode) {
          mask |= BUFFER_BIT_BACK_LEFT;
       }
 
-      for (i = 0; i < ctx->Visual.numAuxBuffers; i++) {
+      for (i = 0; i < fb->Visual.numAuxBuffers; i++) {
          mask |= (BUFFER_BIT_AUX0 << i);
       }
    }
@@ -338,7 +340,6 @@ read_buffer_enum_to_index(GLenum buffer)
 void GLAPIENTRY
 _mesa_DrawBuffer(GLenum buffer)
 {
-   GLuint bufferID;
    GLbitfield destMask;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); /* too complex... */
@@ -347,13 +348,12 @@ _mesa_DrawBuffer(GLenum buffer)
       _mesa_debug(ctx, "glDrawBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
    }
 
-   bufferID = ctx->DrawBuffer->Name;
-
    if (buffer == GL_NONE) {
       destMask = 0x0;
    }
    else {
-      const GLbitfield supportedMask = supported_buffer_bitmask(ctx, bufferID);
+      const GLbitfield supportedMask
+         = supported_buffer_bitmask(ctx, ctx->DrawBuffer);
       destMask = draw_buffer_enum_to_bitmask(buffer);
       if (destMask == BAD_MASK) {
          /* totally bogus buffer */
@@ -386,7 +386,6 @@ void GLAPIENTRY
 _mesa_DrawBuffersARB(GLsizei n, const GLenum *buffers)
 {
    GLint output;
-   GLuint bufferID;
    GLbitfield usedBufferMask, supportedMask;
    GLbitfield destMask[MAX_DRAW_BUFFERS];
    GET_CURRENT_CONTEXT(ctx);
@@ -401,9 +400,7 @@ _mesa_DrawBuffersARB(GLsizei n, const GLenum *buffers)
       return;
    }
 
-   bufferID = ctx->DrawBuffer->Name;
-
-   supportedMask = supported_buffer_bitmask(ctx, bufferID);
+   supportedMask = supported_buffer_bitmask(ctx, ctx->DrawBuffer);
    usedBufferMask = 0x0;
 
    /* complicated error checking... */
@@ -492,8 +489,8 @@ _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
 
    if (!destMask) {
       /* compute destMask values now */
-      const GLuint bufferID = ctx->DrawBuffer->Name;
-      const GLbitfield supportedMask = supported_buffer_bitmask(ctx, bufferID);
+      const GLbitfield supportedMask
+         = supported_buffer_bitmask(ctx, ctx->DrawBuffer);
       for (output = 0; output < n; output++) {
          mask[output] = draw_buffer_enum_to_bitmask(buffers[output]);
          ASSERT(mask[output] != BAD_MASK);
@@ -534,17 +531,15 @@ _mesa_ReadBuffer(GLenum buffer)
    struct gl_framebuffer *fb;
    GLbitfield supportedMask;
    GLint srcBuffer;
-   GLuint bufferID;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    fb = ctx->ReadBuffer;
-   bufferID = fb->Name;
 
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
 
-   if (bufferID > 0 && buffer == GL_NONE) {
+   if (fb->Name > 0 && buffer == GL_NONE) {
       /* This is legal for user-created framebuffer objects */
       srcBuffer = -1;
    }
@@ -555,14 +550,14 @@ _mesa_ReadBuffer(GLenum buffer)
          _mesa_error(ctx, GL_INVALID_ENUM, "glReadBuffer(buffer=0x%x)", buffer);
          return;
       }
-      supportedMask = supported_buffer_bitmask(ctx, bufferID);
+      supportedMask = supported_buffer_bitmask(ctx, fb);
       if (((1 << srcBuffer) & supportedMask) == 0) {
          _mesa_error(ctx, GL_INVALID_OPERATION, "glReadBuffer(buffer=0x%x)", buffer);
          return;
       }
    }
 
-   if (bufferID == 0) {
+   if (fb->Name == 0) {
       ctx->Pixel.ReadBuffer = buffer;
    }
    fb->ColorReadBuffer = buffer;
diff --git a/src/mesa/main/config.h b/src/mesa/main/config.h
index 9e4d1838ade..55199245568 100644
--- a/src/mesa/main/config.h
+++ b/src/mesa/main/config.h
@@ -5,9 +5,9 @@
 
 /*
  * 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"),
@@ -243,10 +243,7 @@
 
 
 /**
- * Bits per depth buffer value.  
- *
- * Any reasonable value up to 31 will work.  32 doesn't work because of integer
- * overflow problems in the rasterizer code.
+ * Bits per depth buffer value (max is 32).
  */
 #ifndef DEFAULT_SOFTWARE_DEPTH_BITS
 #define DEFAULT_SOFTWARE_DEPTH_BITS 16
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index ccaf6f64280..255023c0fa9 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -978,7 +978,7 @@ init_attrib_groups(GLcontext *ctx)
 static int
 generic_nop(void)
 {
-   _mesa_problem(NULL, "User called no-op dispatch function (an unsupported extension function?)");
+   _mesa_warning(NULL, "User called no-op dispatch function (an unsupported extension function?)");
    return 0;
 }
 
@@ -1357,9 +1357,9 @@ _mesa_copy_context( const GLcontext *src, GLcontext *dst, GLuint mask )
  * Check if the given context can render into the given framebuffer
  * by checking visual attributes.
  *
- * XXX this may go away someday because we're moving toward more freedom
- * in binding contexts to drawables with different visual attributes.
- * The GL_EXT_f_b_o extension is prompting some of that.
+ * Most of these tests could go away because Mesa is now pretty flexible
+ * in terms of mixing rendering contexts with framebuffers.  As long
+ * as RGB vs. CI mode agree, we're probably good.
  *
  * \return GL_TRUE if compatible, GL_FALSE otherwise.
  */
@@ -1393,8 +1393,11 @@ check_compatible(const GLcontext *ctx, const GLframebuffer *buffer)
       return GL_FALSE;
    if (ctxvis->blueMask && ctxvis->blueMask != bufvis->blueMask)
       return GL_FALSE;
+#if 0
+   /* disabled (see bug 11161) */
    if (ctxvis->depthBits && ctxvis->depthBits != bufvis->depthBits)
       return GL_FALSE;
+#endif
    if (ctxvis->stencilBits && ctxvis->stencilBits != bufvis->stencilBits)
       return GL_FALSE;
 
diff --git a/src/mesa/main/depth.c b/src/mesa/main/depth.c
index f5511ce2fbf..91c036ef968 100644
--- a/src/mesa/main/depth.c
+++ b/src/mesa/main/depth.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"),
@@ -28,7 +28,6 @@
 #include "context.h"
 #include "depth.h"
 #include "enums.h"
-#include "hash.h"
 #include "macros.h"
 #include "mtypes.h"
 
@@ -153,36 +152,11 @@ _mesa_DepthBoundsEXT( GLclampd zmin, GLclampd zmax )
 /**
  * Initialize the depth buffer attribute group in the given context.
  */
-void _mesa_init_depth( GLcontext * ctx )
+void
+_mesa_init_depth(GLcontext *ctx)
 {
-   /* Depth buffer group */
    ctx->Depth.Test = GL_FALSE;
    ctx->Depth.Clear = 1.0;
    ctx->Depth.Func = GL_LESS;
    ctx->Depth.Mask = GL_TRUE;
-
-   /* XXX this is now per-framebuffer state */
-#if 00
-   /* Z buffer stuff */
-   if (ctx->Visual.depthBits == 0) {
-      /* Special case.  Even if we don't have a depth buffer we need
-       * good values for DepthMax for Z vertex transformation purposes
-       * and for per-fragment fog computation.
-       */
-      ctx->DepthMax = (1 << 16) - 1;
-      ctx->DepthMaxF = (GLfloat) ctx->DepthMax;
-   }
-   else if (ctx->Visual.depthBits < 32) {
-      ctx->DepthMax = (1 << ctx->Visual.depthBits) - 1;
-      ctx->DepthMaxF = (GLfloat) ctx->DepthMax;
-   }
-   else {
-      /* Special case since shift values greater than or equal to the
-       * number of bits in the left hand expression's type are undefined.
-       */
-      ctx->DepthMax = 0xffffffff;
-      ctx->DepthMaxF = (GLfloat) ctx->DepthMax;
-   }
-   ctx->MRD = 1.0;  /* Minimum resolvable depth value, for polygon offset */
-#endif
 }
diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
index a4a6cdae366..80dce56c0c1 100644
--- a/src/mesa/main/extensions.c
+++ b/src/mesa/main/extensions.c
@@ -47,6 +47,7 @@ static const struct {
    { OFF, "GL_ARB_depth_texture",              F(ARB_depth_texture) },
    { OFF, "GL_ARB_draw_buffers",               F(ARB_draw_buffers) },
    { OFF, "GL_ARB_fragment_program",           F(ARB_fragment_program) },
+   { OFF, "GL_ARB_fragment_program_shadow",    F(ARB_fragment_program_shadow) },
    { OFF, "GL_ARB_fragment_shader",            F(ARB_fragment_shader) },
    { OFF, "GL_ARB_half_float_pixel",           F(ARB_half_float_pixel) },
    { OFF, "GL_ARB_imaging",                    F(ARB_imaging) },
@@ -184,6 +185,7 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
    ctx->Extensions.ARB_draw_buffers = GL_TRUE;
 #if FEATURE_ARB_fragment_program
    ctx->Extensions.ARB_fragment_program = GL_TRUE;
+   ctx->Extensions.ARB_fragment_program_shadow = GL_TRUE;
 #endif
 #if FEATURE_ARB_fragment_shader
    ctx->Extensions.ARB_fragment_shader = GL_TRUE;
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index fc712def576..05c08c19fec 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1290,15 +1290,22 @@ struct gl_texture_format
 				 *   GL_DEPTH_COMPONENT.
 				 */
    GLenum DataType;		/**< GL_FLOAT or GL_UNSIGNED_NORMALIZED_ARB */
-   GLubyte RedBits;		/**< Bits per texel component */
-   GLubyte GreenBits;		/**< These are just rough approximations for */
-   GLubyte BlueBits;		/**< compressed texture formats. */
+
+   /**
+    * Bits per texel component.  These are just rough approximations
+    * for compressed texture formats.
+    */
+   /*@{*/
+   GLubyte RedBits;
+   GLubyte GreenBits;
+   GLubyte BlueBits;
    GLubyte AlphaBits;
    GLubyte LuminanceBits;
    GLubyte IntensityBits;
    GLubyte IndexBits;
    GLubyte DepthBits;
    GLubyte StencilBits; 	/**< GL_EXT_packed_depth_stencil */
+   /*@}*/
 
    GLuint TexelBytes;		/**< Bytes per texel, 0 if compressed format */
 
@@ -1419,11 +1426,15 @@ struct gl_texture_object
    GLfloat ShadowAmbient;       /**< GL_ARB_shadow_ambient */
    GLenum CompareMode;		/**< GL_ARB_shadow */
    GLenum CompareFunc;		/**< GL_ARB_shadow */
+   GLenum _Function;		/**< Comparison function derived from 
+				 * \c CompareOperator, \c CompareMode, and
+				 * \c CompareFunc.
+				 */
    GLenum DepthMode;		/**< GL_ARB_depth_texture */
    GLint _MaxLevel;		/**< actual max mipmap level (q in the spec) */
    GLfloat _MaxLambda;		/**< = _MaxLevel - BaseLevel (q - b in spec) */
    GLboolean GenerateMipmap;    /**< GL_SGIS_generate_mipmap */
-   GLboolean Complete;		/**< Is texture object complete? */
+   GLboolean _Complete;		/**< Is texture object complete? */
 
    /** Actual texture images, indexed by [cube face] and [mipmap level] */
    struct gl_texture_image *Image[MAX_FACES][MAX_TEXTURE_LEVELS];
@@ -1490,7 +1501,7 @@ struct gl_texture_unit
    GLbitfield _GenBitT;
    GLbitfield _GenBitR;
    GLbitfield _GenBitQ;
-   GLbitfield _GenFlags;	/**< bitwise or of GenBit[STRQ] */
+   GLbitfield _GenFlags;	/**< bitwise or of _GenBit[STRQ] */
    GLfloat ObjectPlaneS[4];
    GLfloat ObjectPlaneT[4];
    GLfloat ObjectPlaneR[4];
@@ -1528,18 +1539,23 @@ struct gl_texture_unit
 
    struct gl_texture_object *_Current; /**< Points to really enabled tex obj */
 
-   struct gl_texture_object Saved1D;  /**< only used by glPush/PopAttrib */
+   /** 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 */
+   /** GL_SGI_texture_color_table */
+   /*@{*/
    struct gl_color_table ColorTable;
    struct gl_color_table ProxyColorTable;
    GLboolean ColorTableEnabled;
+   /*@}*/
 };
 
 struct texenvprog_cache_item {
@@ -1887,6 +1903,7 @@ struct gl_program
    GLbitfield InputsRead;     /**< Bitmask of which input regs are read */
    GLbitfield OutputsWritten; /**< Bitmask of which output regs are written to */
    GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS];  /**< TEXTURE_x_BIT bitmask */
+   GLbitfield ShadowSamplers; /**< Texture units used for shadow sampling. */
 
    /** Named parameters, constants, etc. from program text */
    struct gl_program_parameter_list *Parameters;
@@ -2516,6 +2533,7 @@ struct gl_extensions
    GLboolean ARB_depth_texture;
    GLboolean ARB_draw_buffers;
    GLboolean ARB_fragment_program;
+   GLboolean ARB_fragment_program_shadow;
    GLboolean ARB_fragment_shader;
    GLboolean ARB_half_float_pixel;
    GLboolean ARB_imaging;
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index 1f4c9f4722c..f315c3de74e 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -2463,7 +2463,7 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat,
 	 update_fbo_texture(ctx, texObj, face, level);
 	 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
@@ -2566,7 +2566,7 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat,
 	 update_fbo_texture(ctx, texObj, face, level);
 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
@@ -2667,7 +2667,7 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat,
 	 update_fbo_texture(ctx, texObj, face, level);
 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
@@ -2938,7 +2938,7 @@ _mesa_CopyTexImage1D( GLenum target, GLint level,
       update_fbo_texture(ctx, texObj, face, level);
 
       /* state update */
-      texObj->Complete = GL_FALSE;
+      texObj->_Complete = GL_FALSE;
       ctx->NewState |= _NEW_TEXTURE;
    }
  out:
@@ -3004,7 +3004,7 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat,
       update_fbo_texture(ctx, texObj, face, level);
 
       /* state update */
-      texObj->Complete = GL_FALSE;
+      texObj->_Complete = GL_FALSE;
       ctx->NewState |= _NEW_TEXTURE;
    }
  out:
@@ -3398,7 +3398,7 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level,
 					     texObj, texImage);
 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
@@ -3495,7 +3495,7 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level,
 					     texObj, texImage);
 	 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
@@ -3591,7 +3591,7 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level,
 					     texObj, texImage);
 	 
 	 /* state update */
-	 texObj->Complete = GL_FALSE;
+	 texObj->_Complete = GL_FALSE;
 	 ctx->NewState |= _NEW_TEXTURE;
       }
    out:
diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c
index 0d2946e95a7..df64002f994 100644
--- a/src/mesa/main/texobj.c
+++ b/src/mesa/main/texobj.c
@@ -209,7 +209,7 @@ _mesa_copy_texture_object( struct gl_texture_object *dest,
    dest->_MaxLambda = src->_MaxLambda;
    dest->GenerateMipmap = src->GenerateMipmap;
    dest->Palette = src->Palette;
-   dest->Complete = src->Complete;
+   dest->_Complete = src->_Complete;
 }
 
 
@@ -251,7 +251,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
    const GLint baseLevel = t->BaseLevel;
    GLint maxLog2 = 0, maxLevels = 0;
 
-   t->Complete = GL_TRUE;  /* be optimistic */
+   t->_Complete = GL_TRUE;  /* be optimistic */
 
    /* Always need the base level image */
    if (!t->Image[0][baseLevel]) {
@@ -259,7 +259,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
       _mesa_sprintf(s, "obj %p (%d) Image[baseLevel=%d] == NULL",
               (void *) t, t->Name, baseLevel);
       incomplete(t, s);
-      t->Complete = GL_FALSE;
+      t->_Complete = GL_FALSE;
       return;
    }
 
@@ -268,7 +268,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
        t->Image[0][baseLevel]->Height == 0 ||
        t->Image[0][baseLevel]->Depth == 0) {
       incomplete(t, "texture width = 0");
-      t->Complete = GL_FALSE;
+      t->_Complete = GL_FALSE;
       return;
    }
 
@@ -322,7 +322,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
 	 if (t->Image[face][baseLevel] == NULL ||
 	     t->Image[face][baseLevel]->Width2 != w ||
 	     t->Image[face][baseLevel]->Height2 != h) {
-	    t->Complete = GL_FALSE;
+	    t->_Complete = GL_FALSE;
 	    incomplete(t, "Non-quare cubemap image");
 	    return;
 	 }
@@ -339,7 +339,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
       GLint maxLevel = t->_MaxLevel;
 
       if (minLevel > maxLevel) {
-         t->Complete = GL_FALSE;
+         t->_Complete = GL_FALSE;
          incomplete(t, "minLevel > maxLevel");
          return;
       }
@@ -348,12 +348,12 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
       for (i = minLevel; i <= maxLevel; i++) {
          if (t->Image[0][i]) {
             if (t->Image[0][i]->TexFormat != t->Image[0][baseLevel]->TexFormat) {
-               t->Complete = GL_FALSE;
+               t->_Complete = GL_FALSE;
                incomplete(t, "Format[i] != Format[baseLevel]");
                return;
             }
             if (t->Image[0][i]->Border != t->Image[0][baseLevel]->Border) {
-               t->Complete = GL_FALSE;
+               t->_Complete = GL_FALSE;
                incomplete(t, "Border[i] != Border[baseLevel]");
                return;
             }
@@ -371,12 +371,12 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
             }
             if (i >= minLevel && i <= maxLevel) {
                if (!t->Image[0][i]) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "1D Image[0][i] == NULL");
                   return;
                }
                if (t->Image[0][i]->Width2 != width ) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "1D Image[0][i] bad width");
                   return;
                }
@@ -400,17 +400,17 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
             }
             if (i >= minLevel && i <= maxLevel) {
                if (!t->Image[0][i]) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "2D Image[0][i] == NULL");
                   return;
                }
                if (t->Image[0][i]->Width2 != width) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "2D Image[0][i] bad width");
                   return;
                }
                if (t->Image[0][i]->Height2 != height) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "2D Image[0][i] bad height");
                   return;
                }
@@ -438,26 +438,26 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
             if (i >= minLevel && i <= maxLevel) {
                if (!t->Image[0][i]) {
                   incomplete(t, "3D Image[0][i] == NULL");
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   return;
                }
                if (t->Image[0][i]->_BaseFormat == GL_DEPTH_COMPONENT) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "GL_DEPTH_COMPONENT only works with 1/2D tex");
                   return;
                }
                if (t->Image[0][i]->Width2 != width) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "3D Image[0][i] bad width");
                   return;
                }
                if (t->Image[0][i]->Height2 != height) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "3D Image[0][i] bad height");
                   return;
                }
                if (t->Image[0][i]->Depth2 != depth) {
-                  t->Complete = GL_FALSE;
+                  t->_Complete = GL_FALSE;
                   incomplete(t, "3D Image[0][i] bad depth");
                   return;
                }
@@ -483,20 +483,20 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
 	       for (face = 0; face < 6; face++) {
 		  /* check that we have images defined */
 		  if (!t->Image[face][i]) {
-		     t->Complete = GL_FALSE;
+		     t->_Complete = GL_FALSE;
 		     incomplete(t, "CubeMap Image[n][i] == NULL");
 		     return;
 		  }
 		  /* Don't support GL_DEPTH_COMPONENT for cube maps */
 		  if (t->Image[face][i]->_BaseFormat == GL_DEPTH_COMPONENT) {
-		     t->Complete = GL_FALSE;
+		     t->_Complete = GL_FALSE;
 		     incomplete(t, "GL_DEPTH_COMPONENT only works with 1/2D tex");
 		     return;
 		  }
 		  /* check that all six images have same size */
 		  if (t->Image[face][i]->Width2!=width || 
 		      t->Image[face][i]->Height2!=height) {
-		     t->Complete = GL_FALSE;
+		     t->_Complete = GL_FALSE;
 		     incomplete(t, "CubeMap Image[n][i] bad size");
 		     return;
 		  }
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index d15af22b7db..c9f8a0656e8 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.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.
  *
@@ -1141,26 +1141,29 @@ _mesa_GetTexEnviv( GLenum target, GLenum pname, GLint *params )
 /*                       Texture Parameters                           */
 /**********************************************************************/
 
+/**
+ * Check if a coordinate wrap mode is supported for the texture target.
+ * \return GL_TRUE if legal, GL_FALSE otherwise
+ */
 static GLboolean 
-_mesa_validate_texture_wrap_mode(GLcontext * ctx,
-				 GLenum target, GLenum eparam)
+validate_texture_wrap_mode(GLcontext * ctx, GLenum target, GLenum wrap)
 {
    const struct gl_extensions * const e = & ctx->Extensions;
 
-   if (eparam == GL_CLAMP || eparam == GL_CLAMP_TO_EDGE ||
-       (eparam == GL_CLAMP_TO_BORDER && e->ARB_texture_border_clamp)) {
+   if (wrap == GL_CLAMP || wrap == GL_CLAMP_TO_EDGE ||
+       (wrap == GL_CLAMP_TO_BORDER && e->ARB_texture_border_clamp)) {
       /* any texture target */
       return GL_TRUE;
    }
    else if (target != GL_TEXTURE_RECTANGLE_NV &&
-	    (eparam == GL_REPEAT ||
-	     (eparam == GL_MIRRORED_REPEAT &&
+	    (wrap == GL_REPEAT ||
+	     (wrap == GL_MIRRORED_REPEAT &&
 	      e->ARB_texture_mirrored_repeat) ||
-	     (eparam == GL_MIRROR_CLAMP_EXT &&
+	     (wrap == GL_MIRROR_CLAMP_EXT &&
 	      (e->ATI_texture_mirror_once || e->EXT_texture_mirror_clamp)) ||
-	     (eparam == GL_MIRROR_CLAMP_TO_EDGE_EXT &&
+	     (wrap == GL_MIRROR_CLAMP_TO_EDGE_EXT &&
 	      (e->ATI_texture_mirror_once || e->EXT_texture_mirror_clamp)) ||
-	     (eparam == GL_MIRROR_CLAMP_TO_BORDER_EXT &&
+	     (wrap == GL_MIRROR_CLAMP_TO_BORDER_EXT &&
 	      (e->EXT_texture_mirror_clamp)))) {
       /* non-rectangle texture */
       return GL_TRUE;
@@ -1283,7 +1286,7 @@ _mesa_TexParameterfv( GLenum target, GLenum pname, const GLfloat *params )
       case GL_TEXTURE_WRAP_S:
          if (texObj->WrapS == eparam)
             return;
-         if (_mesa_validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
+         if (validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
             FLUSH_VERTICES(ctx, _NEW_TEXTURE);
             texObj->WrapS = eparam;
          }
@@ -1294,7 +1297,7 @@ _mesa_TexParameterfv( GLenum target, GLenum pname, const GLfloat *params )
       case GL_TEXTURE_WRAP_T:
          if (texObj->WrapT == eparam)
             return;
-         if (_mesa_validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
+         if (validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
             FLUSH_VERTICES(ctx, _NEW_TEXTURE);
             texObj->WrapT = eparam;
          }
@@ -1305,7 +1308,7 @@ _mesa_TexParameterfv( GLenum target, GLenum pname, const GLfloat *params )
       case GL_TEXTURE_WRAP_R:
          if (texObj->WrapR == eparam)
             return;
-         if (_mesa_validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
+         if (validate_texture_wrap_mode(ctx, texObj->Target, eparam)) {
             FLUSH_VERTICES(ctx, _NEW_TEXTURE);
             texObj->WrapR = eparam;
          }
@@ -1515,7 +1518,7 @@ _mesa_TexParameterfv( GLenum target, GLenum pname, const GLfloat *params )
          return;
    }
 
-   texObj->Complete = GL_FALSE;
+   texObj->_Complete = GL_FALSE;
 
    if (ctx->Driver.TexParameter) {
       (*ctx->Driver.TexParameter)( ctx, target, texObj, pname, params );
@@ -2793,6 +2796,47 @@ update_texture_matrices( GLcontext *ctx )
 
 
 /**
+ * Update texture object's _Function field.  We need to do this
+ * whenever any of the texture object's shadow-related fields change
+ * or when we start/stop using a fragment program.
+ *
+ * This function could be expanded someday to update additional per-object
+ * fields that depend on assorted state changes.
+ */
+static void
+update_texture_compare_function(GLcontext *ctx,
+                                struct gl_texture_object *tObj)
+{
+   /* XXX temporarily disable this test since it breaks the GLSL
+    * shadow2D(), etc. functions.
+    */
+   if (0 /*ctx->FragmentProgram._Current*/) {
+      /* Texel/coordinate comparison is ignored for programs.
+       * See GL_ARB_fragment_program/shader spec for details.
+       */
+      tObj->_Function = GL_NONE;
+   }
+   else if (tObj->CompareFlag) {
+      /* GL_SGIX_shadow */
+      if (tObj->CompareOperator == GL_TEXTURE_LEQUAL_R_SGIX) {
+         tObj->_Function = GL_LEQUAL;
+      }
+      else {
+         ASSERT(tObj->CompareOperator == GL_TEXTURE_GEQUAL_R_SGIX);
+         tObj->_Function = GL_GEQUAL;
+      }
+   }
+   else if (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) {
+      /* GL_ARB_shadow */
+      tObj->_Function = tObj->CompareFunc;
+   }
+   else {
+      tObj->_Function = GL_NONE;  /* pass depth through as grayscale */
+   }
+}
+
+
+/**
  * Helper function for determining which texture object (1D, 2D, cube, etc)
  * should actually be used.
  */
@@ -2802,12 +2846,13 @@ texture_override(GLcontext *ctx,
                  struct gl_texture_object *texObj, GLuint textureBit)
 {
    if (!texUnit->_ReallyEnabled && (enableBits & textureBit)) {
-      if (!texObj->Complete) {
+      if (!texObj->_Complete) {
          _mesa_test_texobj_completeness(ctx, texObj);
       }
-      if (texObj->Complete) {
+      if (texObj->_Complete) {
          texUnit->_ReallyEnabled = textureBit;
          texUnit->_Current = texObj;
+         update_texture_compare_function(ctx, texObj);
       }
    }
 }
diff --git a/src/mesa/main/texstate.h b/src/mesa/main/texstate.h
index ca29c6a23ff..60145691b8c 100644
--- a/src/mesa/main/texstate.h
+++ b/src/mesa/main/texstate.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"),
@@ -42,6 +42,7 @@ extern void
 _mesa_print_texunit_state( GLcontext *ctx, GLuint unit );
 
 
+
 /**
  * \name Called from API
  */
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index 7da3c19a89a..5d8f763741f 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -71,6 +71,7 @@ struct arb_program
 
    /* ARB_fragment_program specifics */
    GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS]; 
+   GLbitfield ShadowSamplers;
    GLuint NumAluInstructions; 
    GLuint NumTexInstructions;
    GLuint NumTexIndirections;
@@ -374,6 +375,8 @@ LONGSTRING static char arb_grammar_text[] =
 /* GL_MESA_texture_array */
 #define  TEXTARGET_1D_ARRAY                         0x09
 #define  TEXTARGET_2D_ARRAY                         0x0a
+#define  TEXTARGET_SHADOW1D_ARRAY                   0x0b
+#define  TEXTARGET_SHADOW2D_ARRAY                   0x0c
 
 /* face type */
 #define  FACE_FRONT                                 0x00
@@ -2659,6 +2662,7 @@ parse_fp_instruction (GLcontext * ctx, const GLubyte ** inst,
    GLuint texcoord;
    GLubyte instClass, type, code;
    GLboolean rel;
+   GLuint shadow_tex = 0;
 
    _mesa_init_instructions(fp, 1);
 
@@ -2976,33 +2980,54 @@ parse_fp_instruction (GLcontext * ctx, const GLubyte ** inst,
 
          /* texTarget */
          switch (*(*inst)++) {
+            case TEXTARGET_SHADOW1D:
+               shadow_tex = 1 << texcoord;
+               /* FALLTHROUGH */
             case TEXTARGET_1D:
                fp->TexSrcTarget = TEXTURE_1D_INDEX;
                break;
+            case TEXTARGET_SHADOW2D:
+               shadow_tex = 1 << texcoord;
+               /* FALLTHROUGH */
             case TEXTARGET_2D:
                fp->TexSrcTarget = TEXTURE_2D_INDEX;
                break;
             case TEXTARGET_3D:
                fp->TexSrcTarget = TEXTURE_3D_INDEX;
                break;
+            case TEXTARGET_SHADOWRECT:
+               shadow_tex = 1 << texcoord;
+               /* FALLTHROUGH */
             case TEXTARGET_RECT:
                fp->TexSrcTarget = TEXTURE_RECT_INDEX;
                break;
             case TEXTARGET_CUBE:
                fp->TexSrcTarget = TEXTURE_CUBE_INDEX;
                break;
-	    case TEXTARGET_SHADOW1D:
-	    case TEXTARGET_SHADOW2D:
-	    case TEXTARGET_SHADOWRECT:
-	       /* TODO ARB_fragment_program_shadow code */
-	       break;
+            case TEXTARGET_SHADOW1D_ARRAY:
+               shadow_tex = 1 << texcoord;
+               /* FALLTHROUGH */
             case TEXTARGET_1D_ARRAY:
                fp->TexSrcTarget = TEXTURE_1D_ARRAY_INDEX;
                break;
+            case TEXTARGET_SHADOW2D_ARRAY:
+               shadow_tex = 1 << texcoord;
+               /* FALLTHROUGH */
             case TEXTARGET_2D_ARRAY:
                fp->TexSrcTarget = TEXTURE_2D_ARRAY_INDEX;
                break;
          }
+
+         /* Don't test the first time a particular sampler is seen.  Each time
+          * after that, make sure the shadow state is the same.
+          */
+         if ((_mesa_bitcount(Program->TexturesUsed[texcoord]) > 0)
+             && ((Program->ShadowSamplers & (1 << texcoord)) != shadow_tex)) {
+            program_error(ctx, Program->Position,
+                          "texture image unit used for shadow sampling and non-shadow sampling");
+            return 1;
+         }
+
          Program->TexturesUsed[texcoord] |= (1 << fp->TexSrcTarget);
          /* Check that both "2D" and "CUBE" (for example) aren't both used */
          if (_mesa_bitcount(Program->TexturesUsed[texcoord]) > 1) {
@@ -3010,6 +3035,9 @@ parse_fp_instruction (GLcontext * ctx, const GLubyte ** inst,
                           "multiple targets used on one texture image unit");
             return 1;
          }
+      
+
+         Program->ShadowSamplers |= shadow_tex;
          break;
 
       case OP_TEX_KIL:
@@ -3600,10 +3628,10 @@ enable_parser_extensions(GLcontext *ctx, grammar id)
    if (ctx->Extensions.ARB_matrix_palette
        && !enable_ext(ctx, id, "matrix_palette"))
       return GL_FALSE;
+#endif
    if (ctx->Extensions.ARB_fragment_program_shadow
        && !enable_ext(ctx, id, "fragment_program_shadow"))
       return GL_FALSE;
-#endif
    if (ctx->Extensions.EXT_point_parameters
        && !enable_ext(ctx, id, "point_parameters"))
       return GL_FALSE;
@@ -3800,6 +3828,7 @@ _mesa_parse_arb_program(GLcontext *ctx, GLenum target,
    program->HintPositionInvariant = GL_FALSE;
    for (a = 0; a < MAX_TEXTURE_IMAGE_UNITS; a++)
       program->TexturesUsed[a] = 0x0;
+   program->ShadowSamplers = 0x0;
    program->NumAluInstructions =
    program->NumTexInstructions =
    program->NumTexIndirections = 0;
@@ -3880,6 +3909,7 @@ _mesa_parse_arb_fragment_program(GLcontext* ctx, GLenum target,
    program->Base.OutputsWritten  = ap.Base.OutputsWritten;
    for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++)
       program->Base.TexturesUsed[i] = ap.TexturesUsed[i];
+   program->Base.ShadowSamplers = ap.ShadowSamplers;
    program->FogOption          = ap.FogOption;
    program->UsesKill          = ap.UsesKill;
 
diff --git a/src/mesa/shader/arbprogram.syn b/src/mesa/shader/arbprogram.syn
index 4f82717873e..1746a876c30 100644
--- a/src/mesa/shader/arbprogram.syn
+++ b/src/mesa/shader/arbprogram.syn
@@ -21,13 +21,13 @@
  * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
- 
+
  /**
  * \file arbprogram.syn
  * ARB_fragment/vertex_program syntax
  * \author Michal Krol
  */
- 
+
 .syntax program;
 
 /*
@@ -226,8 +226,11 @@
 .emtcode TEXTARGET_SHADOW1D                         0x06
 .emtcode TEXTARGET_SHADOW2D                         0x07
 .emtcode TEXTARGET_SHADOWRECT                       0x08
+/* GL_MESA_texture_array */
 .emtcode TEXTARGET_1D_ARRAY                         0x09
 .emtcode TEXTARGET_2D_ARRAY                         0x0a
+.emtcode TEXTARGET_SHADOW1D_ARRAY                   0x0b
+.emtcode TEXTARGET_SHADOW2D_ARRAY                   0x0c
 
 /* face type */
 .emtcode FACE_FRONT                                 0x00
@@ -912,6 +915,7 @@ fragment program
                              | "CUBE"
                              | "RECT"
                              | <shadowTarget> (if option ARB_fragment_program_shadow present)
+                             | <arrayTarget> (if option MESA_texture_array present)
 */
 texTarget
     "1D" .emit TEXTARGET_1D .or
@@ -920,19 +924,38 @@ texTarget
     .if (texture_rectangle != 0x00) "RECT" .emit TEXTARGET_RECT .or
     "CUBE" .emit TEXTARGET_CUBE .or
     .if (ARB_fragment_program_shadow != 0x00) shadowTarget .or
-    .if (MESA_texture_array != 0x00) "ARRAY1D" .emit TEXTARGET_1D_ARRAY .or
-    .if (MESA_texture_array != 0x00) "ARRAY2D" .emit TEXTARGET_2D_ARRAY;
+    .if (MESA_texture_array != 0x00) arrayTarget;
 
 /*
 GL_ARB_fragment_program_shadow
     <shadowTarget>         ::= "SHADOW1D"
                              | "SHADOW2D"
                              | "SHADOWRECT"
+                             | <shadowArrayTarget> (if option MESA_texture_array present)
 */
 shadowTarget
     "SHADOW1D" .emit TEXTARGET_SHADOW1D .or
     "SHADOW2D" .emit TEXTARGET_SHADOW2D .or
-    .if (texture_rectangle != 0x00) "SHADOWRECT" .emit TEXTARGET_SHADOWRECT;
+    .if (texture_rectangle != 0x00) "SHADOWRECT" .emit TEXTARGET_SHADOWRECT .or
+    .if (MESA_texture_array != 0x00) shadowArrayTarget;
+
+/*
+GL_MESA_texture_array
+
+    <arrayTarget>         ::= "ARRAY1D"
+                             | "ARRAY2D"
+
+    <shadowArrayTarget>   ::= "SHADOWARRAY1D"
+                             | "SHADOWARRAY2D"
+*/
+
+arrayTarget
+    "ARRAY1D" .emit TEXTARGET_1D_ARRAY .or
+    "ARRAY2D" .emit TEXTARGET_2D_ARRAY;
+
+shadowArrayTarget
+    "SHADOWARRAY1D" .emit TEXTARGET_SHADOW1D_ARRAY .or
+    "SHADOWARRAY2D" .emit TEXTARGET_SHADOW2D_ARRAY;
 
 /*
 fragment program
diff --git a/src/mesa/shader/arbprogram_syn.h b/src/mesa/shader/arbprogram_syn.h
index 30dc9f4594e..5f3f7d6cf46 100644
--- a/src/mesa/shader/arbprogram_syn.h
+++ b/src/mesa/shader/arbprogram_syn.h
@@ -1,3 +1,7 @@
+
+/* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE .syn FILE */
+
+" \n"
 ".syntax program;\n"
 ".emtcode REVISION 0x0a\n"
 ".emtcode FRAGMENT_PROGRAM 0x01\n"
@@ -122,7 +126,9 @@
 ".emtcode TEXTARGET_SHADOW2D 0x07\n"
 ".emtcode TEXTARGET_SHADOWRECT 0x08\n"
 ".emtcode TEXTARGET_1D_ARRAY 0x09\n"
-".emtcode TEXTARGET_2D_ARRAY 0x0A\n"
+".emtcode TEXTARGET_2D_ARRAY 0x0a\n"
+".emtcode TEXTARGET_SHADOW1D_ARRAY 0x0b\n"
+".emtcode TEXTARGET_SHADOW2D_ARRAY 0x0c\n"
 ".emtcode FACE_FRONT 0x00\n"
 ".emtcode FACE_BACK 0x01\n"
 ".emtcode COLOR_PRIMARY 0x00\n"
@@ -479,12 +485,18 @@
 " .if (texture_rectangle != 0x00) \"RECT\" .emit TEXTARGET_RECT .or\n"
 " \"CUBE\" .emit TEXTARGET_CUBE .or\n"
 " .if (ARB_fragment_program_shadow != 0x00) shadowTarget .or\n"
-" .if (MESA_texture_array != 0x00) \"ARRAY1D\" .emit TEXTARGET_1D_ARRAY .or\n"
-" .if (MESA_texture_array != 0x00) \"ARRAY2D\" .emit TEXTARGET_2D_ARRAY;\n"
+" .if (MESA_texture_array != 0x00) arrayTarget;\n"
 "shadowTarget\n"
 " \"SHADOW1D\" .emit TEXTARGET_SHADOW1D .or\n"
 " \"SHADOW2D\" .emit TEXTARGET_SHADOW2D .or\n"
-" .if (texture_rectangle != 0x00) \"SHADOWRECT\" .emit TEXTARGET_SHADOWRECT;\n"
+" .if (texture_rectangle != 0x00) \"SHADOWRECT\" .emit TEXTARGET_SHADOWRECT .or\n"
+" .if (MESA_texture_array != 0x00) shadowArrayTarget;\n"
+"arrayTarget\n"
+" \"ARRAY1D\" .emit TEXTARGET_1D_ARRAY .or\n"
+" \"ARRAY2D\" .emit TEXTARGET_2D_ARRAY;\n"
+"shadowArrayTarget\n"
+" \"SHADOWARRAY1D\" .emit TEXTARGET_SHADOW1D_ARRAY .or\n"
+" \"SHADOWARRAY2D\" .emit TEXTARGET_SHADOW2D_ARRAY;\n"
 "optTexImageUnitNum\n"
 " optTexImageUnitNum_1 .or .true .emit 0x00;\n"
 "optTexImageUnitNum_1\n"
diff --git a/src/mesa/swrast/s_aalinetemp.h b/src/mesa/swrast/s_aalinetemp.h
index 8756f122d09..3d3511823d7 100644
--- a/src/mesa/swrast/s_aalinetemp.h
+++ b/src/mesa/swrast/s_aalinetemp.h
@@ -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.
  *
@@ -185,10 +185,18 @@ NAME(line)(GLcontext *ctx, const SWvertex *v0, const SWvertex *v1)
       compute_plane(line.x0, line.y0, line.x1, line.y1, invW0, invW1, line.wPlane);
       ATTRIB_LOOP_BEGIN
          GLuint c;
-         for (c = 0; c < 4; c++) {
-            const GLfloat a0 = v0->attrib[attr][c] * invW0;
-            const GLfloat a1 = v1->attrib[attr][c] * invW1;
-            compute_plane(line.x0, line.y0, line.x1, line.y1, a0, a1, line.attrPlane[attr][c]);
+         if (swrast->_InterpMode[attr] == GL_FLAT) {
+            for (c = 0; c < 4; c++) {
+               constant_plane(v1->attrib[attr][c], line.attrPlane[attr][c]);
+            }
+         }
+         else {
+            for (c = 0; c < 4; c++) {
+               const GLfloat a0 = v0->attrib[attr][c] * invW0;
+               const GLfloat a1 = v1->attrib[attr][c] * invW1;
+               compute_plane(line.x0, line.y0, line.x1, line.y1, a0, a1,
+                             line.attrPlane[attr][c]);
+            }
          }
          line.span.arrayAttribs |= (1 << attr);
          if (attr >= FRAG_ATTRIB_TEX0 && attr < FRAG_ATTRIB_VAR0) {
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index f373fde781b..791850cb502 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.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.
  *
@@ -535,10 +535,11 @@ _swrast_update_texture_samplers(GLcontext *ctx)
 
 
 /**
- * Update swrast->_ActiveAttribs, swrast->_NumActiveAttribs, swrast->_ActiveAtttribMask.
+ * Update swrast->_ActiveAttribs, swrast->_NumActiveAttribs,
+ * swrast->_ActiveAtttribMask.
  */
 static void
-_swrast_update_fragment_attribs(GLcontext *ctx)
+_swrast_update_active_attribs(GLcontext *ctx)
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLuint attribsMask;
@@ -663,16 +664,14 @@ _swrast_validate_derived( GLcontext *ctx )
                               _NEW_PROGRAM))
 	 _swrast_update_fragment_program( ctx, swrast->NewState );
 
-      if (swrast->NewState & (_NEW_TEXTURE | _NEW_PROGRAM))
-         _swrast_update_texture_samplers( ctx );
-
       if (swrast->NewState & (_NEW_TEXTURE | _NEW_PROGRAM)) {
+         _swrast_update_texture_samplers( ctx );
          _swrast_validate_texture_images(ctx);
-         if (swrast->NewState & (_NEW_COLOR)) {
-            _swrast_update_deferred_texture(ctx);
-         }
       }
 
+      if (swrast->NewState & (_NEW_COLOR | _NEW_PROGRAM))
+         _swrast_update_deferred_texture(ctx);
+
       if (swrast->NewState & _SWRAST_NEW_RASTERMASK)
  	 _swrast_update_rasterflags( ctx );
 
@@ -681,7 +680,7 @@ _swrast_validate_derived( GLcontext *ctx )
                               _NEW_LIGHT |
                               _NEW_PROGRAM |
                               _NEW_TEXTURE))
-         _swrast_update_fragment_attribs(ctx);
+         _swrast_update_active_attribs(ctx);
 
       if (swrast->NewState & (_NEW_PROGRAM | _NEW_BUFFERS))
          _swrast_update_color_outputs(ctx);
diff --git a/src/mesa/swrast/s_copypix.c b/src/mesa/swrast/s_copypix.c
index 53e584b3b6b..2383015000a 100644
--- a/src/mesa/swrast/s_copypix.c
+++ b/src/mesa/swrast/s_copypix.c
@@ -188,6 +188,8 @@ static void
 copy_rgba_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
                  GLint width, GLint height, GLint destx, GLint desty)
 {
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const GLbitfield prevActiveAttribs = swrast->_ActiveAttribMask;
    GLfloat *tmpImage, *p;
    GLint sy, dy, stepy, row;
    const GLboolean zoom = ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F;
@@ -197,12 +199,15 @@ copy_rgba_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
 
    if (!ctx->ReadBuffer->_ColorReadBuffer) {
       /* no readbuffer - OK */
-      return;
+      goto end;
    }
 
+   /* don't interpolate COL0 and overwrite the glDrawPixel colors! */
+   swrast->_ActiveAttribMask &= ~FRAG_BIT_COL0;
+
    if (ctx->Pixel.Convolution2DEnabled || ctx->Pixel.Separable2DEnabled) {
       copy_conv_rgba_pixels(ctx, srcx, srcy, width, height, destx, desty);
-      return;
+      goto end;
    }
    else if (ctx->Pixel.Convolution1DEnabled) {
       /* make sure we don't apply 1D convolution */
@@ -239,7 +244,7 @@ copy_rgba_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
       tmpImage = (GLfloat *) _mesa_malloc(width * height * sizeof(GLfloat) * 4);
       if (!tmpImage) {
          _mesa_error( ctx, GL_OUT_OF_MEMORY, "glCopyPixels" );
-         return;
+         goto end;
       }
       /* read the source image as RGBA/float */
       p = tmpImage;
@@ -294,6 +299,9 @@ copy_rgba_pixels(GLcontext *ctx, GLint srcx, GLint srcy,
 
    if (overlapping)
       _mesa_free(tmpImage);
+
+end:
+   swrast->_ActiveAttribMask = prevActiveAttribs;
 }
 
 
diff --git a/src/mesa/swrast/s_drawpix.c b/src/mesa/swrast/s_drawpix.c
index d971d90fb9a..925358d77eb 100644
--- a/src/mesa/swrast/s_drawpix.c
+++ b/src/mesa/swrast/s_drawpix.c
@@ -532,16 +532,22 @@ draw_rgba_pixels( GLcontext *ctx, GLint x, GLint y,
                   const struct gl_pixelstore_attrib *unpack,
                   const GLvoid *pixels )
 {
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const GLbitfield prevActiveAttribs = swrast->_ActiveAttribMask;
    const GLint imgX = x, imgY = y;
    const GLboolean zoom = ctx->Pixel.ZoomX!=1.0 || ctx->Pixel.ZoomY!=1.0;
    GLfloat *convImage = NULL;
    GLbitfield transferOps = ctx->_ImageTransferState;
    SWspan span;
 
+   /* don't interpolate COL0 and overwrite the glDrawPixel colors! */
+   swrast->_ActiveAttribMask &= ~FRAG_BIT_COL0;
+
    /* Try an optimized glDrawPixels first */
    if (fast_draw_rgba_pixels(ctx, x, y, width, height, format, type,
-                             unpack, pixels))
-      return;
+                             unpack, pixels)) {
+      goto end;
+   }
 
    INIT_SPAN(span, GL_BITMAP, 0, 0x0, SPAN_RGBA);
    _swrast_span_default_attribs(ctx, &span);
@@ -559,13 +565,13 @@ draw_rgba_pixels( GLcontext *ctx, GLint x, GLint y,
       tmpImage = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat));
       if (!tmpImage) {
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
-         return;
+         goto end;
       }
       convImage = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat));
       if (!convImage) {
          _mesa_free(tmpImage);
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
-         return;
+         goto end;
       }
 
       /* Unpack the image and apply transfer ops up to convolution */
@@ -669,6 +675,9 @@ draw_rgba_pixels( GLcontext *ctx, GLint x, GLint y,
    if (convImage) {
       _mesa_free(convImage);
    }
+
+end:
+   swrast->_ActiveAttribMask = prevActiveAttribs;
 }
 
 
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index e47dbbdaf35..923b67e78e6 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -26,6 +26,7 @@
 #include "colormac.h"
 #include "context.h"
 #include "prog_instruction.h"
+#include "texstate.h"
 
 #include "s_fragprog.h"
 #include "s_span.h"
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 90a3d5545f8..431629efb1f 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -171,6 +171,11 @@ interpolate_active_attribs(GLcontext *ctx, SWspan *span, GLbitfield attrMask)
 {
    const SWcontext *swrast = SWRAST_CONTEXT(ctx);
 
+   /* for glDraw/CopyPixels() we may have turned off some bits in
+    * the _ActiveAttribMask - be sure to obey that mask now.
+    */
+   attrMask &= swrast->_ActiveAttribMask;
+
    ATTRIB_LOOP_BEGIN
       if (attrMask & (1 << attr)) {
          const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
@@ -1169,9 +1174,11 @@ shade_texture_span(GLcontext *ctx, SWspan *span)
    if (ctx->FragmentProgram._Current ||
        ctx->ATIFragmentShader._Enabled) {
       /* programmable shading */
-      span->array->ChanType = GL_FLOAT;
-
+      if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {
+         convert_color_type(span, GL_FLOAT, 0);
+      }
       interpolate_active_attribs(ctx, span, ~0);
+      span->array->ChanType = GL_FLOAT;
 
       if (!(span->arrayMask & SPAN_Z))
          _swrast_span_interpolate_z (ctx, span);
diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c
index 2c8e443daf9..c2a7512388e 100644
--- a/src/mesa/swrast/s_texfilter.c
+++ b/src/mesa/swrast/s_texfilter.c
@@ -2893,25 +2893,7 @@ sample_depth_texture( GLcontext *ctx,
 
    /* XXXX if tObj->MinFilter != tObj->MagFilter, we're ignoring lambda */
 
-   /* XXX this could be precomputed and saved in the texture object */
-   if (tObj->CompareFlag) {
-      /* GL_SGIX_shadow */
-      if (tObj->CompareOperator == GL_TEXTURE_LEQUAL_R_SGIX) {
-         function = GL_LEQUAL;
-      }
-      else {
-         ASSERT(tObj->CompareOperator == GL_TEXTURE_GEQUAL_R_SGIX);
-         function = GL_GEQUAL;
-      }
-   }
-   else if (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) {
-      /* GL_ARB_shadow */
-      function = tObj->CompareFunc;
-   }
-   else {
-      function = GL_NONE;  /* pass depth through as grayscale */
-   }
-
+   function = tObj->_Function;
    if (tObj->MagFilter == GL_NEAREST) {
       GLuint i;
       for (i = 0; i < n; i++) {
@@ -3365,7 +3347,7 @@ texture_sample_func
 _swrast_choose_texture_sample_func( GLcontext *ctx,
 				    const struct gl_texture_object *t )
 {
-   if (!t || !t->Complete) {
+   if (!t || !t->_Complete) {
       return &null_sample_func;
    }
    else {
diff --git a/src/mesa/swrast/s_zoom.c b/src/mesa/swrast/s_zoom.c
index 78fa137d3fd..1ab5911f2f1 100644
--- a/src/mesa/swrast/s_zoom.c
+++ b/src/mesa/swrast/s_zoom.c
@@ -162,12 +162,14 @@ zoom_span( GLcontext *ctx, GLint imgX, GLint imgY, const SWspan *span,
    zoomed_arrays.rgba = zoomed_arrays.attribs[FRAG_ATTRIB_COL0];
 #endif
 
+   /* copy attribute info (XXX copy all attribs?) */
+   COPY_4V(zoomed.attrStart[FRAG_ATTRIB_WPOS], span->attrStart[FRAG_ATTRIB_WPOS]);
+   COPY_4V(zoomed.attrStepX[FRAG_ATTRIB_WPOS], span->attrStepX[FRAG_ATTRIB_WPOS]);
+   COPY_4V(zoomed.attrStepY[FRAG_ATTRIB_WPOS], span->attrStepY[FRAG_ATTRIB_WPOS]);
 
-   /* copy fog interp info */
    zoomed.attrStart[FRAG_ATTRIB_FOGC][0] = span->attrStart[FRAG_ATTRIB_FOGC][0];
    zoomed.attrStepX[FRAG_ATTRIB_FOGC][0] = span->attrStepX[FRAG_ATTRIB_FOGC][0];
    zoomed.attrStepY[FRAG_ATTRIB_FOGC][0] = span->attrStepY[FRAG_ATTRIB_FOGC][0];
-   /* XXX copy texcoord info? */
 
    if (format == GL_RGBA || format == GL_RGB) {
       /* copy Z info */
diff --git a/src/mesa/x86/common_x86.c b/src/mesa/x86/common_x86.c
index 889b40a89f2..0b2af0a3706 100644
--- a/src/mesa/x86/common_x86.c
+++ b/src/mesa/x86/common_x86.c
@@ -104,12 +104,7 @@ static LONG WINAPI ExceptionFilter(LPEXCEPTION_POINTERS exp)
 
 static void check_os_sse_support( void )
 {
-#if defined(__linux__)
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,0)
-   _mesa_debug(NULL, "Cannot safely enable SSE on pre-2.4 kernels.\n");
-   _mesa_x86_cpu_features &= ~(X86_FEATURE_XMM);
-#endif
-#elif defined(__FreeBSD__)
+#if defined(__FreeBSD__)
    {
       int ret, enabled;
       unsigned int len;
@@ -160,7 +155,7 @@ static void check_os_sse_support( void )
    /* Do nothing on other platforms for now.
     */
    _mesa_debug(NULL, "Not testing OS support for SSE, leaving enabled.\n");
-#endif /* __linux__ */
+#endif /* __FreeBSD__ */
 }
 
 #endif /* USE_SSE_ASM */
diff --git a/src/mesa/x86/glapi_x86.S b/src/mesa/x86/glapi_x86.S
index bdf42ac0887..74e93721bc3 100644
--- a/src/mesa/x86/glapi_x86.S
+++ b/src/mesa/x86/glapi_x86.S
@@ -58,12 +58,17 @@
 
 #ifdef GLX_USE_TLS
 
+#ifdef GLX_X86_READONLY_TEXT
+# define CTX_INSNS MOV_L(GS:(EAX), EAX)
+#else
+# define CTX_INSNS NOP /* Pad for init_glapi_relocs() */
+#endif
 #  define GL_STUB(fn,off,fn_alt)			\
 ALIGNTEXT16;						\
 GLOBL_FN(GL_PREFIX(fn, fn_alt));			\
 GL_PREFIX(fn, fn_alt):					\
 	CALL(_x86_get_dispatch) ;			\
-	NOP ;						\
+	CTX_INSNS ;					\
 	JMP(GL_OFFSET(off))
 
 #elif defined(PTHREADS)
@@ -114,7 +119,10 @@ SEG_TEXT
 	HIDDEN(GLNAME(_x86_get_dispatch))
 ALIGNTEXT16
 GLNAME(_x86_get_dispatch):
-	movl	%gs:_glapi_tls_Dispatch@NTPOFF, %eax
+	call	1f
+1:	popl	%eax
+	addl	$_GLOBAL_OFFSET_TABLE_+[.-1b], %eax
+	movl	_glapi_tls_Dispatch@GOTNTPOFF(%eax), %eax
 	ret
 
 #elif defined(PTHREADS)
@@ -133,7 +141,7 @@ GLNAME(_x86_get_dispatch):
 EXTERN GLNAME(_glapi_get_dispatch)
 #endif
 
-#if defined( GLX_USE_TLS )
+#if defined( GLX_USE_TLS ) && !defined( GLX_X86_READONLY_TEXT )
 		.section	wtext, "awx", @progbits
 #endif /* defined( GLX_USE_TLS ) */