Merge remote branch 'origin/master' into gallium_draw_llvm

79 files changed, 1618 insertions, 2408 deletions

diff --git a/src/gallium/auxiliary/Makefile b/src/gallium/auxiliary/Makefile
index 452eceb7f4..4c629924b9 100644
--- a/src/gallium/auxiliary/Makefile
+++ b/src/gallium/auxiliary/Makefile
@@ -105,7 +105,8 @@ C_SOURCES = \
 	util/u_cpu_detect.c \
 	util/u_dl.c \
 	util/u_draw_quad.c \
-	util/u_format_access.c \
+	util/u_format.c \
+	util/u_format_s3tc.c \
 	util/u_format_table.c \
 	util/u_format_tests.c \
 	util/u_gen_mipmap.c \
@@ -148,8 +149,6 @@ GALLIVM_SOURCES = \
         gallivm/lp_bld_debug.c \
         gallivm/lp_bld_depth.c \
         gallivm/lp_bld_flow.c \
-        gallivm/lp_bld_format_aos.c \
-        gallivm/lp_bld_format_query.c \
         gallivm/lp_bld_format_soa.c \
         gallivm/lp_bld_interp.c \
         gallivm/lp_bld_intr.c \
@@ -169,8 +168,6 @@ GALLIVM_CPP_SOURCES = \
 GENERATED_SOURCES = \
 	indices/u_indices_gen.c \
 	indices/u_unfilled_gen.c \
-	util/u_format_access.c \
-	util/u_format_pack.h \
 	util/u_format_table.c
 
 
@@ -194,12 +191,9 @@ indices/u_indices_gen.c: indices/u_indices_gen.py
 indices/u_unfilled_gen.c: indices/u_unfilled_gen.py
 	python $< > $@
 
-util/u_format_table.c: util/u_format_table.py util/u_format_parse.py util/u_format.csv
+util/u_format_table.c: util/u_format_table.py util/u_format_pack.py util/u_format_parse.py util/u_format.csv
 	python util/u_format_table.py util/u_format.csv > $@
 
-util/u_format_pack.h: util/u_format_pack.py util/u_format_parse.py util/u_format.csv
-	python util/u_format_pack.py util/u_format.csv > $@
-
 util/u_format_access.c: util/u_format_access.py util/u_format_parse.py util/u_format.csv
 	python util/u_format_access.py util/u_format.csv > $@
 
diff --git a/src/gallium/auxiliary/SConscript b/src/gallium/auxiliary/SConscript
index d0785bce16..b4e8eb041a 100644
--- a/src/gallium/auxiliary/SConscript
+++ b/src/gallium/auxiliary/SConscript
@@ -29,20 +29,10 @@ env.CodeGenerate(
     source = ['util/u_format.csv'],
     command = 'python $SCRIPT $SOURCE > $TARGET'
 )
-
-env.CodeGenerate(
-    target = File('util/u_format_pack.h').srcnode(),
-    script = 'util/u_format_pack.py',
-    source = ['util/u_format.csv'],
-    command = 'python $SCRIPT $SOURCE > $TARGET'
-)
-
-env.CodeGenerate(
-    target = 'util/u_format_access.c',
-    script = 'util/u_format_access.py',
-    source = ['util/u_format.csv'],
-    command = 'python $SCRIPT $SOURCE > $TARGET'
-)
+env.Depends('util/u_format_table.c', [
+    'util/u_format_parse.py', 
+    'util/u_format_pack.py', 
+])
 
 source = [
     'cso_cache/cso_context.c',
@@ -149,7 +139,8 @@ source = [
     'util/u_dump_state.c',
     'util/u_dl.c',
     'util/u_draw_quad.c',
-    'util/u_format_access.c',
+    'util/u_format.c',
+    'util/u_format_s3tc.c',
     'util/u_format_table.c',
     'util/u_format_tests.c',
     'util/u_gen_mipmap.c',
@@ -192,8 +183,6 @@ if drawllvm:
     'gallivm/lp_bld_debug.c',
     'gallivm/lp_bld_depth.c',
     'gallivm/lp_bld_flow.c',
-    'gallivm/lp_bld_format_aos.c',
-    'gallivm/lp_bld_format_query.c',
     'gallivm/lp_bld_format_soa.c',
     'gallivm/lp_bld_interp.c',
     'gallivm/lp_bld_intr.c',
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_depth.c b/src/gallium/auxiliary/gallivm/lp_bld_depth.c
index 4ce1a27a06..c6d0e4bc04 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_depth.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_depth.c
@@ -495,8 +495,8 @@ lp_build_depth_stencil_test(LLVMBuilderRef builder,
       assert(format_desc->block.height == 1);
 
       if (stencil[0].enabled) {
-         assert(format_desc->format == PIPE_FORMAT_Z24S8_UNORM ||
-                format_desc->format == PIPE_FORMAT_S8Z24_UNORM);
+         assert(format_desc->format == PIPE_FORMAT_Z24_UNORM_S8_USCALED ||
+                format_desc->format == PIPE_FORMAT_S8_USCALED_Z24_UNORM);
       }
 
       assert(z_swizzle < 4);
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c b/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c
deleted file mode 100644
index a07f7418f2..0000000000
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c
+++ /dev/null
@@ -1,383 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/**
- * @file
- * AoS pixel format manipulation.
- *
- * @author Jose Fonseca <jfonseca@vmware.com>
- */
-
-
-#include "util/u_cpu_detect.h"
-#include "util/u_format.h"
-
-#include "lp_bld_type.h"
-#include "lp_bld_const.h"
-#include "lp_bld_swizzle.h"
-#include "lp_bld_format.h"
-
-
-/**
- * Unpack a single pixel into its RGBA components.
- *
- * @param packed integer.
- *
- * @return RGBA in a 4 floats vector.
- *
- * XXX: This is mostly for reference and testing -- operating a single pixel at
- * a time is rarely if ever needed.
- */
-LLVMValueRef
-lp_build_unpack_rgba_aos(LLVMBuilderRef builder,
-                         const struct util_format_description *desc,
-                         LLVMValueRef packed)
-{
-   LLVMTypeRef type;
-   LLVMValueRef shifted, casted, scaled, masked;
-   LLVMValueRef shifts[4];
-   LLVMValueRef masks[4];
-   LLVMValueRef scales[4];
-   LLVMValueRef swizzles[4];
-   LLVMValueRef aux[4];
-   bool normalized;
-   int empty_channel;
-   unsigned shift;
-   unsigned i;
-
-   /* FIXME: Support more formats */
-   assert(desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
-   assert(desc->block.width == 1);
-   assert(desc->block.height == 1);
-   assert(desc->block.bits <= 32);
-
-   type = LLVMIntType(desc->block.bits);
-
-   /* Do the intermediate integer computations with 32bit integers since it
-    * matches floating point size */
-   if (desc->block.bits < 32)
-      packed = LLVMBuildZExt(builder, packed, LLVMInt32Type(), "");
-
-   /* Broadcast the packed value to all four channels */
-   packed = LLVMBuildInsertElement(builder,
-                                   LLVMGetUndef(LLVMVectorType(LLVMInt32Type(), 4)),
-                                   packed,
-                                   LLVMConstNull(LLVMInt32Type()),
-                                   "");
-   packed = LLVMBuildShuffleVector(builder,
-                                   packed,
-                                   LLVMGetUndef(LLVMVectorType(LLVMInt32Type(), 4)),
-                                   LLVMConstNull(LLVMVectorType(LLVMInt32Type(), 4)),
-                                   "");
-
-   /* Initialize vector constants */
-   normalized = FALSE;
-   empty_channel = -1;
-   shift = 0;
-   for (i = 0; i < 4; ++i) {
-      unsigned bits = desc->channel[i].size;
-
-      if (desc->channel[i].type == UTIL_FORMAT_TYPE_VOID) {
-         shifts[i] = LLVMGetUndef(LLVMInt32Type());
-         masks[i] = LLVMConstNull(LLVMInt32Type());
-         scales[i] =  LLVMConstNull(LLVMFloatType());
-         empty_channel = i;
-      }
-      else {
-         unsigned mask = (1 << bits) - 1;
-
-         assert(desc->channel[i].type == UTIL_FORMAT_TYPE_UNSIGNED);
-         assert(bits < 32);
-
-         shifts[i] = LLVMConstInt(LLVMInt32Type(), shift, 0);
-         masks[i] = LLVMConstInt(LLVMInt32Type(), mask, 0);
-
-         if (desc->channel[i].normalized) {
-            scales[i] = LLVMConstReal(LLVMFloatType(), 1.0/mask);
-            normalized = TRUE;
-         }
-         else
-            scales[i] =  LLVMConstReal(LLVMFloatType(), 1.0);
-      }
-
-      shift += bits;
-   }
-
-   shifted = LLVMBuildLShr(builder, packed, LLVMConstVector(shifts, 4), "");
-   masked = LLVMBuildAnd(builder, shifted, LLVMConstVector(masks, 4), "");
-   /* UIToFP can't be expressed in SSE2 */
-   casted = LLVMBuildSIToFP(builder, masked, LLVMVectorType(LLVMFloatType(), 4), "");
-
-   if (normalized)
-      scaled = LLVMBuildMul(builder, casted, LLVMConstVector(scales, 4), "");
-   else
-      scaled = casted;
-
-   for (i = 0; i < 4; ++i)
-      aux[i] = LLVMGetUndef(LLVMFloatType());
-
-   for (i = 0; i < 4; ++i) {
-      enum util_format_swizzle swizzle = desc->swizzle[i];
-
-      switch (swizzle) {
-      case UTIL_FORMAT_SWIZZLE_X:
-      case UTIL_FORMAT_SWIZZLE_Y:
-      case UTIL_FORMAT_SWIZZLE_Z:
-      case UTIL_FORMAT_SWIZZLE_W:
-         swizzles[i] = LLVMConstInt(LLVMInt32Type(), swizzle, 0);
-         break;
-      case UTIL_FORMAT_SWIZZLE_0:
-         assert(empty_channel >= 0);
-         swizzles[i] = LLVMConstInt(LLVMInt32Type(), empty_channel, 0);
-         break;
-      case UTIL_FORMAT_SWIZZLE_1:
-         swizzles[i] = LLVMConstInt(LLVMInt32Type(), 4, 0);
-         aux[0] = LLVMConstReal(LLVMFloatType(), 1.0);
-         break;
-      case UTIL_FORMAT_SWIZZLE_NONE:
-         swizzles[i] = LLVMGetUndef(LLVMFloatType());
-         assert(0);
-         break;
-      }
-   }
-
-   return LLVMBuildShuffleVector(builder, scaled, LLVMConstVector(aux, 4), LLVMConstVector(swizzles, 4), "");
-}
-
-
-/**
- * Take a vector with packed pixels and unpack into a rgba8 vector.
- *
- * Formats with bit depth smaller than 32bits are accepted, but they must be
- * padded to 32bits.
- */
-LLVMValueRef
-lp_build_unpack_rgba8_aos(LLVMBuilderRef builder,
-                          const struct util_format_description *desc,
-                          struct lp_type type,
-                          LLVMValueRef packed)
-{
-   struct lp_build_context bld;
-   bool rgba8;
-   LLVMValueRef res;
-   unsigned i;
-
-   lp_build_context_init(&bld, builder, type);
-
-   /* FIXME: Support more formats */
-   assert(desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
-   assert(desc->block.width == 1);
-   assert(desc->block.height == 1);
-   assert(desc->block.bits <= 32);
-
-   assert(!type.floating);
-   assert(!type.fixed);
-   assert(type.norm);
-   assert(type.width == 8);
-   assert(type.length % 4 == 0);
-
-   rgba8 = TRUE;
-   for(i = 0; i < 4; ++i) {
-      assert(desc->channel[i].type == UTIL_FORMAT_TYPE_UNSIGNED ||
-             desc->channel[i].type == UTIL_FORMAT_TYPE_VOID);
-      if(desc->channel[0].size != 8)
-         rgba8 = FALSE;
-   }
-
-   if(rgba8) {
-      /*
-       * The pixel is already in a rgba8 format variant. All it is necessary
-       * is to swizzle the channels.
-       */
-
-      unsigned char swizzles[4];
-      boolean zeros[4]; /* bitwise AND mask */
-      boolean ones[4]; /* bitwise OR mask */
-      boolean swizzles_needed = FALSE;
-      boolean zeros_needed = FALSE;
-      boolean ones_needed = FALSE;
-
-      for(i = 0; i < 4; ++i) {
-         enum util_format_swizzle swizzle = desc->swizzle[i];
-
-         /* Initialize with the no-op case */
-         swizzles[i] = util_cpu_caps.little_endian ? 3 - i : i;
-         zeros[i] = TRUE;
-         ones[i] = FALSE;
-
-         switch (swizzle) {
-         case UTIL_FORMAT_SWIZZLE_X:
-         case UTIL_FORMAT_SWIZZLE_Y:
-         case UTIL_FORMAT_SWIZZLE_Z:
-         case UTIL_FORMAT_SWIZZLE_W:
-            if(swizzle != swizzles[i]) {
-               swizzles[i] = swizzle;
-               swizzles_needed = TRUE;
-            }
-            break;
-         case UTIL_FORMAT_SWIZZLE_0:
-            zeros[i] = FALSE;
-            zeros_needed = TRUE;
-            break;
-         case UTIL_FORMAT_SWIZZLE_1:
-            ones[i] = TRUE;
-            ones_needed = TRUE;
-            break;
-         case UTIL_FORMAT_SWIZZLE_NONE:
-            assert(0);
-            break;
-         }
-      }
-
-      res = packed;
-
-      if(swizzles_needed)
-         res = lp_build_swizzle1_aos(&bld, res, swizzles);
-
-      if(zeros_needed) {
-         /* Mask out zero channels */
-         LLVMValueRef mask = lp_build_const_mask_aos(type, zeros);
-         res = LLVMBuildAnd(builder, res, mask, "");
-      }
-
-      if(ones_needed) {
-         /* Or one channels */
-         LLVMValueRef mask = lp_build_const_mask_aos(type, ones);
-         res = LLVMBuildOr(builder, res, mask, "");
-      }
-   }
-   else {
-      /* FIXME */
-      assert(0);
-      res = lp_build_undef(type);
-   }
-
-   return res;
-}
-
-
-/**
- * Pack a single pixel.
- *
- * @param rgba 4 float vector with the unpacked components.
- *
- * XXX: This is mostly for reference and testing -- operating a single pixel at
- * a time is rarely if ever needed.
- */
-LLVMValueRef
-lp_build_pack_rgba_aos(LLVMBuilderRef builder,
-                       const struct util_format_description *desc,
-                       LLVMValueRef rgba)
-{
-   LLVMTypeRef type;
-   LLVMValueRef packed = NULL;
-   LLVMValueRef swizzles[4];
-   LLVMValueRef shifted, casted, scaled, unswizzled;
-   LLVMValueRef shifts[4];
-   LLVMValueRef scales[4];
-   bool normalized;
-   unsigned shift;
-   unsigned i, j;
-
-   assert(desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
-   assert(desc->block.width == 1);
-   assert(desc->block.height == 1);
-
-   type = LLVMIntType(desc->block.bits);
-
-   /* Unswizzle the color components into the source vector. */
-   for (i = 0; i < 4; ++i) {
-      for (j = 0; j < 4; ++j) {
-         if (desc->swizzle[j] == i)
-            break;
-      }
-      if (j < 4)
-         swizzles[i] = LLVMConstInt(LLVMInt32Type(), j, 0);
-      else
-         swizzles[i] = LLVMGetUndef(LLVMInt32Type());
-   }
-
-   unswizzled = LLVMBuildShuffleVector(builder, rgba,
-                                       LLVMGetUndef(LLVMVectorType(LLVMFloatType(), 4)),
-                                       LLVMConstVector(swizzles, 4), "");
-
-   normalized = FALSE;
-   shift = 0;
-   for (i = 0; i < 4; ++i) {
-      unsigned bits = desc->channel[i].size;
-
-      if (desc->channel[i].type == UTIL_FORMAT_TYPE_VOID) {
-         shifts[i] = LLVMGetUndef(LLVMInt32Type());
-         scales[i] =  LLVMGetUndef(LLVMFloatType());
-      }
-      else {
-         unsigned mask = (1 << bits) - 1;
-
-         assert(desc->channel[i].type == UTIL_FORMAT_TYPE_UNSIGNED);
-         assert(bits < 32);
-
-         shifts[i] = LLVMConstInt(LLVMInt32Type(), shift, 0);
-
-         if (desc->channel[i].normalized) {
-            scales[i] = LLVMConstReal(LLVMFloatType(), mask);
-            normalized = TRUE;
-         }
-         else
-            scales[i] =  LLVMConstReal(LLVMFloatType(), 1.0);
-      }
-
-      shift += bits;
-   }
-
-   if (normalized)
-      scaled = LLVMBuildMul(builder, unswizzled, LLVMConstVector(scales, 4), "");
-   else
-      scaled = unswizzled;
-
-   casted = LLVMBuildFPToSI(builder, scaled, LLVMVectorType(LLVMInt32Type(), 4), "");
-
-   shifted = LLVMBuildShl(builder, casted, LLVMConstVector(shifts, 4), "");
-   
-   /* Bitwise or all components */
-   for (i = 0; i < 4; ++i) {
-      if (desc->channel[i].type == UTIL_FORMAT_TYPE_UNSIGNED) {
-         LLVMValueRef component = LLVMBuildExtractElement(builder, shifted, LLVMConstInt(LLVMInt32Type(), i, 0), "");
-         if (packed)
-            packed = LLVMBuildOr(builder, packed, component, "");
-         else
-            packed = component;
-      }
-   }
-
-   if (!packed)
-      packed = LLVMGetUndef(LLVMInt32Type());
-
-   if (desc->block.bits < 32)
-      packed = LLVMBuildTrunc(builder, packed, type, "");
-
-   return packed;
-}
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_query.c b/src/gallium/auxiliary/gallivm/lp_bld_format_query.c
deleted file mode 100644
index f3832d07ff..0000000000
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_query.c
+++ /dev/null
@@ -1,72 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/**
- * @file
- * Utility functions to make assertions about formats.
- *
- * This module centralizes most of logic used when determining what algorithm
- * is most suitable (i.e., most efficient yet correct) for a given format.
- *
- * It might be possible to move some of these functions to u_format module,
- * but since tiny differences in the format my render it more/less
- * appropriate to a given algorithm it is impossible to make any long term
- * guarantee about the semantics of these functions.
- *
- * @author Jose Fonseca <jfonseca@vmware.com>
- */
-
-
-#include "util/u_format.h"
-
-#include "lp_bld_format.h"
-
-
-/**
- * Whether this format is a 4 rgba8 variant
- */
-boolean
-lp_format_is_rgba8(const struct util_format_description *desc)
-{
-   unsigned chan;
-
-   if(desc->block.width != 1 ||
-      desc->block.height != 1 ||
-      desc->block.bits != 32)
-      return FALSE;
-
-   for(chan = 0; chan < 4; ++chan) {
-      if(desc->channel[chan].type != UTIL_FORMAT_TYPE_UNSIGNED &&
-         desc->channel[chan].type != UTIL_FORMAT_TYPE_SIGNED &&
-         desc->channel[chan].type != UTIL_FORMAT_TYPE_VOID)
-         return FALSE;
-      if(desc->channel[chan].size != 8)
-         return FALSE;
-   }
-
-   return TRUE;
-}
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
index 45ee4b12ce..c9fbe0ad10 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
@@ -92,14 +92,14 @@ lp_build_unpack_rgba_soa(LLVMBuilderRef builder,
    unsigned chan;
 
    /* FIXME: Support more formats */
-   assert(format_desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
+   assert(format_desc->is_bitmask);
    assert(format_desc->block.width == 1);
    assert(format_desc->block.height == 1);
    assert(format_desc->block.bits <= 32);
 
    /* Decode the input vector components */
    start = 0;
-   for (chan = 0; chan < 4; ++chan) {
+   for (chan = 0; chan < format_desc->nr_channels; ++chan) {
       unsigned width = format_desc->channel[chan].size;
       unsigned stop = start + width;
       LLVMValueRef input;
@@ -123,7 +123,7 @@ lp_build_unpack_rgba_soa(LLVMBuilderRef builder,
             if(format_desc->channel[chan].normalized)
                input = lp_build_unsigned_norm_to_float(builder, width, type, input);
             else
-               input = LLVMBuildFPToSI(builder, input, lp_build_vec_type(type), "");
+               input = LLVMBuildSIToFP(builder, input, lp_build_vec_type(type), "");
          }
          else {
             /* FIXME */
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_printf.c b/src/gallium/auxiliary/gallivm/lp_bld_printf.c
index 78c9ec778b..e4fa2c264c 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_printf.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_printf.c
@@ -104,8 +104,15 @@ lp_build_printf(LLVMBuilderRef builder, const char *fmt, ...)
    params[0] = LLVMBuildGEP(builder, fmtarg, index, 2, "");
 
    va_start(arglist, fmt);
-   for (i = 1; i <= argcount; i++)
-      params[i] = va_arg(arglist, LLVMValueRef);
+   for (i = 1; i <= argcount; i++) {
+      LLVMValueRef val = va_arg(arglist, LLVMValueRef);
+      LLVMTypeRef type = LLVMTypeOf(val);
+      /* printf wants doubles, so lets convert so that
+       * we can actually print them */
+      if (LLVMGetTypeKind(type) == LLVMFloatTypeKind)
+         val = LLVMBuildFPExt(builder, val, LLVMDoubleType(), "");
+      params[i] = val;
+   }
    va_end(arglist);
 
    return LLVMBuildCall(builder, func_printf, params, argcount + 1, "");
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
index 995c016b9d..5b56f2cf3b 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
@@ -2037,7 +2037,7 @@ lp_build_sample_soa(LLVMBuilderRef builder,
    height_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, height);
    depth_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, depth);
 
-   if (lp_format_is_rgba8(bld.format_desc) &&
+   if (util_format_is_rgba8_variant(bld.format_desc) &&
        static_state->target == PIPE_TEXTURE_2D &&
        static_state->min_img_filter == PIPE_TEX_FILTER_LINEAR &&
        static_state->mag_img_filter == PIPE_TEX_FILTER_LINEAR &&
diff --git a/src/gallium/auxiliary/os/os_thread.h b/src/gallium/auxiliary/os/os_thread.h
index a04df4106f..004dad6b63 100644
--- a/src/gallium/auxiliary/os/os_thread.h
+++ b/src/gallium/auxiliary/os/os_thread.h
@@ -40,7 +40,7 @@
 #include "util/u_debug.h" /* for assert */
 
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
 
 #include <pthread.h> /* POSIX threads headers */
 #include <stdio.h> /* for perror() */
@@ -257,7 +257,7 @@ typedef unsigned pipe_condvar;
  * pipe_barrier
  */
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
 
 typedef pthread_barrier_t pipe_barrier;
 
@@ -377,7 +377,7 @@ pipe_semaphore_wait(pipe_semaphore *sema)
  */
 
 typedef struct {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
    pthread_key_t key;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    DWORD key;
@@ -392,7 +392,7 @@ typedef struct {
 static INLINE void
 pipe_tsd_init(pipe_tsd *tsd)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
    if (pthread_key_create(&tsd->key, NULL/*free*/) != 0) {
       perror("pthread_key_create(): failed to allocate key for thread specific data");
       exit(-1);
@@ -409,7 +409,7 @@ pipe_tsd_get(pipe_tsd *tsd)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
    return pthread_getspecific(tsd->key);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    assert(0);
@@ -426,7 +426,7 @@ pipe_tsd_set(pipe_tsd *tsd, void *value)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
    if (pthread_setspecific(tsd->key, value) != 0) {
       perror("pthread_set_specific() failed");
       exit(-1);
diff --git a/src/gallium/auxiliary/util/.gitignore b/src/gallium/auxiliary/util/.gitignore
index 448d2f304f..53fe9b8a65 100644
--- a/src/gallium/auxiliary/util/.gitignore
+++ b/src/gallium/auxiliary/util/.gitignore
@@ -1,3 +1 @@
-u_format_access.c
 u_format_table.c
-u_format_pack.h
diff --git a/src/gallium/auxiliary/util/u_format.c b/src/gallium/auxiliary/util/u_format.c
new file mode 100644
index 0000000000..7f16cf7d01
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format.c
@@ -0,0 +1,118 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Pixel format accessor functions.
+ *
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#include "u_format.h"
+
+
+void
+util_format_read_4f(enum pipe_format format,
+                    float *dst, unsigned dst_stride,
+                    const void *src, unsigned src_stride,
+                    unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   const uint8_t *src_row;
+   float *dst_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   src_row = (const uint8_t *)src + y*src_stride + x*(format_desc->block.bits/8);
+   dst_row = dst;
+
+   format_desc->unpack_float(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_write_4f(enum pipe_format format,
+                     const float *src, unsigned src_stride,
+                     void *dst, unsigned dst_stride,
+                     unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   uint8_t *dst_row;
+   const float *src_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   dst_row = (uint8_t *)dst + y*dst_stride + x*(format_desc->block.bits/8);
+   src_row = src;
+
+   format_desc->pack_float(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_read_4ub(enum pipe_format format, uint8_t *dst, unsigned dst_stride, const void *src, unsigned src_stride, unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   const uint8_t *src_row;
+   uint8_t *dst_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   src_row = (const uint8_t *)src + y*src_stride + x*(format_desc->block.bits/8);
+   dst_row = dst;
+
+   format_desc->unpack_8unorm(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_write_4ub(enum pipe_format format, const uint8_t *src, unsigned src_stride, void *dst, unsigned dst_stride, unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   uint8_t *dst_row;
+   const uint8_t *src_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   dst_row = (uint8_t *)dst + y*dst_stride + x*(format_desc->block.bits/8);
+   src_row = src;
+
+   format_desc->pack_8unorm(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index d819bbbde1..0cf539247e 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -56,14 +56,18 @@
 # internal formats to base in- ternal formats, and desired component
 # resolutions for each sized internal format."
 
+# None
+# Described as regular uint_8 bytes, i.e. PIPE_FORMAT_R8_USCALED
+PIPE_FORMAT_NONE                  , plain, 1, 1, u8  ,     ,     ,     , x001, rgb
+
 # Typical rendertarget formats
 PIPE_FORMAT_B8G8R8A8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , zyxw, rgb
-PIPE_FORMAT_B8G8R8X8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , zyx1, rgb
+PIPE_FORMAT_B8G8R8X8_UNORM        , plain, 1, 1, un8 , un8 , un8 , x8  , zyx1, rgb
 PIPE_FORMAT_A8R8G8B8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , yzwx, rgb
-PIPE_FORMAT_X8R8G8B8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , yzw1, rgb
+PIPE_FORMAT_X8R8G8B8_UNORM        , plain, 1, 1, x8  , un8 , un8 , un8 , yzw1, rgb
 PIPE_FORMAT_A8B8G8R8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , wzyx, rgb
-PIPE_FORMAT_X8B8G8R8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , wzy1, rgb
-PIPE_FORMAT_B5G5R5X1_UNORM        , plain, 1, 1, un5 , un5 , un5 , un1 , zyx1, rgb
+PIPE_FORMAT_X8B8G8R8_UNORM        , plain, 1, 1, x8  , un8 , un8 , un8 , wzy1, rgb
+PIPE_FORMAT_B5G5R5X1_UNORM        , plain, 1, 1, un5 , un5 , un5 , x1  , zyx1, rgb
 PIPE_FORMAT_B5G5R5A1_UNORM        , plain, 1, 1, un5 , un5 , un5 , un1 , zyxw, rgb
 PIPE_FORMAT_B4G4R4A4_UNORM        , plain, 1, 1, un4 , un4 , un4 , un4 , zyxw, rgb
 PIPE_FORMAT_B5G6R5_UNORM          , plain, 1, 1, un5 , un6 , un5 ,     , zyx1, rgb
@@ -80,26 +84,28 @@ PIPE_FORMAT_L16_UNORM             , plain, 1, 1, un16,     ,     ,     , xxx1, r
 PIPE_FORMAT_L8_SRGB               , plain, 1, 1, un8 ,     ,     ,     , xxx1, srgb 
 PIPE_FORMAT_L8A8_SRGB             , plain, 1, 1, un8 , un8 ,     ,     , xxxy, srgb 
 PIPE_FORMAT_R8G8B8_SRGB           , plain, 1, 1, un8 , un8 , un8 ,     , xyz1, srgb 
+PIPE_FORMAT_R8G8B8A8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , xyzw, srgb 
 PIPE_FORMAT_A8B8G8R8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , wzyx, srgb
-PIPE_FORMAT_X8B8G8R8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , wzy1, srgb
+PIPE_FORMAT_X8B8G8R8_SRGB         , plain, 1, 1, x8  , un8 , un8 , un8 , wzy1, srgb
 PIPE_FORMAT_B8G8R8A8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , zyxw, srgb
-PIPE_FORMAT_B8G8R8X8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , zyx1, srgb
+PIPE_FORMAT_B8G8R8X8_SRGB         , plain, 1, 1, un8 , un8 , un8 , x8  , zyx1, srgb
 PIPE_FORMAT_A8R8G8B8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , yzwx, srgb
-PIPE_FORMAT_X8R8G8B8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , yzw1, srgb
+PIPE_FORMAT_X8R8G8B8_SRGB         , plain, 1, 1, x8  , un8 , un8 , un8 , yzw1, srgb
 
 # Mixed-sign formats (typically used for bump map textures)
 PIPE_FORMAT_R8SG8SB8UX8U_NORM     , plain, 1, 1, sn8 , sn8 , un8 , x8  , xyz1, rgb
 PIPE_FORMAT_R5SG5SB6U_NORM        , plain, 1, 1, sn5 , sn5 , un6 ,     , xyz1, rgb
 
 # Depth-stencil formats
-PIPE_FORMAT_S8_UNORM              , plain, 1, 1, un8 ,     ,     ,     , _x__, zs 
-PIPE_FORMAT_Z16_UNORM             , plain, 1, 1, un16,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z32_UNORM             , plain, 1, 1, un32,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z32_FLOAT             , plain, 1, 1, f32 ,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z24S8_UNORM           , plain, 1, 1, un24, un8 ,     ,     , xy__, zs 
-PIPE_FORMAT_S8Z24_UNORM           , plain, 1, 1, un8 , un24,     ,     , yx__, zs 
-PIPE_FORMAT_Z24X8_UNORM           , plain, 1, 1, un24, un8 ,     ,     , x___, zs 
-PIPE_FORMAT_X8Z24_UNORM           , plain, 1, 1, un8 , un24,     ,     , y___, zs 
+PIPE_FORMAT_S8_USCALED              , plain, 1, 1, u8  ,     ,     ,     , _x__, zs
+PIPE_FORMAT_Z16_UNORM               , plain, 1, 1, un16,     ,     ,     , x___, zs
+PIPE_FORMAT_Z32_UNORM               , plain, 1, 1, un32,     ,     ,     , x___, zs
+PIPE_FORMAT_Z32_FLOAT               , plain, 1, 1, f32 ,     ,     ,     , x___, zs
+PIPE_FORMAT_Z24_UNORM_S8_USCALED    , plain, 1, 1, un24, u8  ,     ,     , xy__, zs
+PIPE_FORMAT_S8_USCALED_Z24_UNORM    , plain, 1, 1, u8 ,  un24,     ,     , yx__, zs
+PIPE_FORMAT_Z24X8_UNORM             , plain, 1, 1, un24, x8  ,     ,     , x___, zs
+PIPE_FORMAT_X8Z24_UNORM             , plain, 1, 1, x8  , un24,     ,     , y___, zs
+PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED , plain, 1, 1, f32,  u8  , x24 ,     , xy__, zs
 
 # YUV formats
 # http://www.fourcc.org/yuv.php#UYVY
@@ -108,6 +114,15 @@ PIPE_FORMAT_UYVY                 , subsampled, 2, 1, x32 ,     ,     ,     , xyz
 # XXX: u_tile.c's ycbcr_get_tile_rgba actually interprets it as VYUY but the 
 # intent should be to match D3DFMT_YUY2
 PIPE_FORMAT_YUYV                 , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, yuv
+# same subsampling but with rgb channels
+PIPE_FORMAT_R8G8_B8G8_UNORM      , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_G8R8_G8B8_UNORM      , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, rgb
+
+# some special formats not fitting anywhere else
+PIPE_FORMAT_R10G10B10A2_USCALED   , plain,      1,  1, u10 , u10 , u10 , u2  , xyzw, rgb
+PIPE_FORMAT_R11G11B10_FLOAT       , plain,      1,  1, f11 , f11 , f10 ,     , xyz1, rgb
+PIPE_FORMAT_R9G9B9E5_FLOAT        , other,      1,  1, x32 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_R1_UNORM              , other,      8,  1, x8  ,     ,     ,     , x001, rgb
 
 # Compressed formats
 # - http://en.wikipedia.org/wiki/S3_Texture_Compression
@@ -123,6 +138,11 @@ PIPE_FORMAT_DXT1_SRGBA            , s3tc, 4, 4, x64 ,     ,     ,     , xyzw, sr
 PIPE_FORMAT_DXT3_SRGBA            , s3tc, 4, 4, x128,     ,     ,     , xyzw, srgb
 PIPE_FORMAT_DXT5_SRGBA            , s3tc, 4, 4, x128,     ,     ,     , xyzw, srgb
 
+PIPE_FORMAT_RGTC1_UNORM           , rgtc, 4, 4, x64,      ,     ,     , x001, rgb
+PIPE_FORMAT_RGTC1_SNORM           , rgtc, 4, 4, x64,      ,     ,     , x001, rgb
+PIPE_FORMAT_RGTC2_UNORM           , rgtc, 4, 4, x128,     ,     ,     , xy01, rgb
+PIPE_FORMAT_RGTC2_SNORM           , rgtc, 4, 4, x128,     ,     ,     , xy01, rgb
+
 # Straightforward D3D10-like formats (also used for 
 # vertex buffer element description)
 # 
@@ -157,6 +177,10 @@ PIPE_FORMAT_R32_FIXED             , plain, 1, 1, h32 ,     ,     ,     , x001, r
 PIPE_FORMAT_R32G32_FIXED          , plain, 1, 1, h32 , h32 ,     ,     , xy01, rgb
 PIPE_FORMAT_R32G32B32_FIXED       , plain, 1, 1, h32 , h32 , h32 ,     , xyz1, rgb
 PIPE_FORMAT_R32G32B32A32_FIXED    , plain, 1, 1, h32 , h32 , h32 , h32 , xyzw, rgb
+PIPE_FORMAT_R16_FLOAT             , plain, 1, 1, f16 ,     ,     ,     , x001, rgb
+PIPE_FORMAT_R16G16_FLOAT          , plain, 1, 1, f16 , f16 ,     ,     , xy01, rgb
+PIPE_FORMAT_R16G16B16_FLOAT       , plain, 1, 1, f16 , f16 , f16 ,     , xyz1, rgb
+PIPE_FORMAT_R16G16B16A16_FLOAT    , plain, 1, 1, f16 , f16 , f16 , f16 , xyzw, rgb
 PIPE_FORMAT_R16_UNORM             , plain, 1, 1, un16,     ,     ,     , x001, rgb
 PIPE_FORMAT_R16G16_UNORM          , plain, 1, 1, un16, un16,     ,     , xy01, rgb
 PIPE_FORMAT_R16G16B16_UNORM       , plain, 1, 1, un16, un16, un16,     , xyz1, rgb
diff --git a/src/gallium/auxiliary/util/u_format.h b/src/gallium/auxiliary/util/u_format.h
index 98d4b98ebb..93818a3161 100644
--- a/src/gallium/auxiliary/util/u_format.h
+++ b/src/gallium/auxiliary/util/u_format.h
@@ -189,6 +189,46 @@ struct util_format_description
     * Colorspace transformation.
     */
    enum util_format_colorspace colorspace;
+
+   /**
+    * Unpack pixel blocks to R8G8B8A8_UNORM.
+    */
+   void
+   (*unpack_8unorm)(uint8_t *dst, unsigned dst_stride,
+                    const uint8_t *src, unsigned src_stride,
+                    unsigned width, unsigned height);
+
+   /**
+    * Pack pixel blocks from R8G8B8A8_UNORM.
+    */
+   void
+   (*pack_8unorm)(uint8_t *dst, unsigned dst_stride,
+                  const uint8_t *src, unsigned src_stride,
+                  unsigned width, unsigned height);
+
+   /**
+    * Unpack pixel blocks to R32G32B32A32_FLOAT.
+    */
+   void
+   (*unpack_float)(float *dst, unsigned dst_stride,
+                   const uint8_t *src, unsigned src_stride,
+                   unsigned width, unsigned height);
+
+   /**
+    * Pack pixel blocks from R32G32B32A32_FLOAT.
+    */
+   void
+   (*pack_float)(uint8_t *dst, unsigned dst_stride,
+                 const float *src, unsigned src_stride,
+                 unsigned width, unsigned height);
+
+   /**
+    * Fetch a single pixel (i, j) from a block.
+    */
+   void
+   (*fetch_float)(float *dst,
+                  const uint8_t *src,
+                  unsigned i, unsigned j);
 };
 
 
@@ -209,8 +249,8 @@ util_format_name(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return "???";
    }
 
@@ -222,8 +262,8 @@ util_format_is_s3tc(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
@@ -235,8 +275,8 @@ util_format_is_depth_or_stencil(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
@@ -248,8 +288,8 @@ util_format_is_depth_and_stencil(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
@@ -261,6 +301,34 @@ util_format_is_depth_and_stencil(enum pipe_format format)
            desc->swizzle[1] != UTIL_FORMAT_SWIZZLE_NONE) ? TRUE : FALSE;
 }
 
+/**
+ * Whether this format is a rgab8 variant.
+ *
+ * That is, any format that matches the
+ *
+ *   PIPE_FORMAT_?8?8?8?8_UNORM
+ */
+static INLINE boolean
+util_format_is_rgba8_variant(const struct util_format_description *desc)
+{
+   unsigned chan;
+
+   if(desc->block.width != 1 ||
+      desc->block.height != 1 ||
+      desc->block.bits != 32)
+      return FALSE;
+
+   for(chan = 0; chan < 4; ++chan) {
+      if(desc->channel[chan].type != UTIL_FORMAT_TYPE_UNSIGNED &&
+         desc->channel[chan].type != UTIL_FORMAT_TYPE_VOID)
+         return FALSE;
+      if(desc->channel[chan].size != 8)
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
 
 /**
  * Return total bits needed for the pixel format per block.
@@ -270,8 +338,8 @@ util_format_get_blocksizebits(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 0;
    }
 
@@ -296,8 +364,8 @@ util_format_get_blockwidth(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 1;
    }
 
@@ -309,8 +377,8 @@ util_format_get_blockheight(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 1;
    }
 
diff --git a/src/gallium/auxiliary/util/u_format_access.py b/src/gallium/auxiliary/util/u_format_access.py
deleted file mode 100644
index 00424779d2..0000000000
--- a/src/gallium/auxiliary/util/u_format_access.py
+++ /dev/null
@@ -1,341 +0,0 @@
-#!/usr/bin/env python
-
-'''
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/**
- * @file
- * Pixel format accessor functions.
- *
- * @author Jose Fonseca <jfonseca@vmware.com>
- */
-'''
-
-
-import math
-import sys
-
-from u_format_pack import *
-
-
-def is_format_supported(format):
-    '''Determines whether we actually have the plumbing necessary to generate the 
-    to read/write to/from this format.'''
-
-    # FIXME: Ideally we would support any format combination here.
-
-    # XXX: It should be straightforward to support srgb
-    if format.colorspace not in ('rgb', 'zs'):
-        return False
-
-    if format.layout != PLAIN:
-        return False
-
-    for i in range(4):
-        channel = format.channels[i]
-        if channel.type not in (VOID, UNSIGNED, FLOAT):
-            return False
-
-    # We can only read a color from a depth/stencil format if the depth channel is present
-    if format.colorspace == 'zs' and format.swizzles[0] == SWIZZLE_NONE:
-        return False
-
-    return True
-
-
-def native_type(format):
-    '''Get the native appropriate for a format.'''
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            # For arithmetic pixel formats return the integer type that matches the whole pixel
-            return 'uint%u_t' % format.block_size()
-        else:
-            # For array pixel formats return the integer type that matches the color channel
-            channel = format.channels[0]
-            if channel.type == UNSIGNED:
-                return 'uint%u_t' % channel.size
-            elif channel.type == SIGNED:
-                return 'int%u_t' % channel.size
-            elif channel.type == FLOAT:
-                if channel.size == 32:
-                    return 'float'
-                elif channel.size == 64:
-                    return 'double'
-                else:
-                    assert False
-            else:
-                assert False
-    else:
-        assert False
-
-
-def generate_srgb_tables():
-    print 'static ubyte srgb_to_linear[256] = {'
-    for i in range(256):
-        print '   %s,' % (int(math.pow((i / 255.0 + 0.055) / 1.055, 2.4) * 255))
-    print '};'
-    print
-    print 'static ubyte linear_to_srgb[256] = {'
-    print '   0,'
-    for i in range(1, 256):
-        print '   %s,' % (int((1.055 * math.pow(i / 255.0, 0.41666) - 0.055) * 255))
-    print '};'
-    print
-
-
-def generate_format_read(format, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the function to read pixels from a particular format'''
-
-    name = format.short_name()
-
-    src_native_type = native_type(format)
-
-    print 'static void'
-    print 'util_format_%s_read_%s(%s *dst, unsigned dst_stride, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, dst_suffix, dst_native_type)
-    print '{'
-    print '   unsigned x, y;'
-    print '   const uint8_t *src_row = src + y0*src_stride;'
-    print '   %s *dst_row = dst;' % dst_native_type
-    print '   for (y = 0; y < h; ++y) {'
-    print '      const %s *src_pixel = (const %s *)(src_row + x0*%u);' % (src_native_type, src_native_type, format.stride())
-    print '      %s *dst_pixel = dst_row;' %dst_native_type
-    print '      for (x = 0; x < w; ++x) {'
-
-    names = ['']*4
-    if format.colorspace == 'rgb':
-        for i in range(4):
-            swizzle = format.swizzles[i]
-            if swizzle < 4:
-                names[swizzle] += 'rgba'[i]
-    elif format.colorspace == 'zs':
-        swizzle = format.swizzles[0]
-        if swizzle < 4:
-            names[swizzle] = 'z'
-        else:
-            assert False
-    else:
-        assert False
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            print '         %s pixel = *src_pixel++;' % src_native_type
-            shift = 0;
-            for i in range(4):
-                src_channel = format.channels[i]
-                width = src_channel.size
-                if names[i]:
-                    value = 'pixel'
-                    mask = (1 << width) - 1
-                    if shift:
-                        value = '(%s >> %u)' % (value, shift)
-                    if shift + width < format.block_size():
-                        value = '(%s & 0x%x)' % (value, mask)
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         %s %s = %s;' % (dst_native_type, names[i], value)
-                shift += width
-        else:
-            for i in range(4):
-                src_channel = format.channels[i]
-                if names[i]:
-                    value = 'src_pixel[%u]' % i
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         %s %s = %s;' % (dst_native_type, names[i], value)
-            print '         src_pixel += %u;' % (format.nr_channels())
-    else:
-        assert False
-
-    for i in range(4):
-        if format.colorspace == 'rgb':
-            swizzle = format.swizzles[i]
-            if swizzle < 4:
-                value = names[swizzle]
-            elif swizzle == SWIZZLE_0:
-                value = '0'
-            elif swizzle == SWIZZLE_1:
-                value = get_one(dst_channel)
-            else:
-                assert False
-        elif format.colorspace == 'zs':
-            if i < 3:
-                value = 'z'
-            else:
-                value = get_one(dst_channel)
-        else:
-            assert False
-        print '         *dst_pixel++ = %s; /* %s */' % (value, 'rgba'[i])
-
-    print '      }'
-    print '      src_row += src_stride;'
-    print '      dst_row += dst_stride/sizeof(*dst_row);'
-    print '   }'
-    print '}'
-    print
-    
-
-def generate_format_write(format, src_channel, src_native_type, src_suffix):
-    '''Generate the function to write pixels to a particular format'''
-
-    name = format.short_name()
-
-    dst_native_type = native_type(format)
-
-    print 'static void'
-    print 'util_format_%s_write_%s(const %s *src, unsigned src_stride, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, src_suffix, src_native_type)
-    print '{'
-    print '   unsigned x, y;'
-    print '   uint8_t *dst_row = dst + y0*dst_stride;'
-    print '   const %s *src_row = src;' % src_native_type
-    print '   for (y = 0; y < h; ++y) {'
-    print '      %s *dst_pixel = (%s *)(dst_row + x0*%u);' % (dst_native_type, dst_native_type, format.stride())
-    print '      const %s *src_pixel = src_row;' %src_native_type
-    print '      for (x = 0; x < w; ++x) {'
-
-    inv_swizzle = format.inv_swizzles()
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            print '         %s pixel = 0;' % dst_native_type
-            shift = 0;
-            for i in range(4):
-                dst_channel = format.channels[i]
-                width = dst_channel.size
-                if inv_swizzle[i] is not None:
-                    value = 'src_pixel[%u]' % inv_swizzle[i]
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    if shift:
-                        value = '(%s << %u)' % (value, shift)
-                    print '         pixel |= %s;' % value
-                shift += width
-            print '         *dst_pixel++ = pixel;'
-        else:
-            for i in range(4):
-                dst_channel = format.channels[i]
-                if inv_swizzle[i] is not None:
-                    value = 'src_pixel[%u]' % inv_swizzle[i]
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         *dst_pixel++ = %s;' % value
-    else:
-        assert False
-    print '         src_pixel += 4;'
-
-    print '      }'
-    print '      dst_row += dst_stride;'
-    print '      src_row += src_stride/sizeof(*src_row);'
-    print '   }'
-    print '}'
-    print
-    
-
-def generate_read(formats, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the dispatch function to read pixels from any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_read(format, dst_channel, dst_native_type, dst_suffix)
-
-    print 'void'
-    print 'util_format_read_%s(enum pipe_format format, %s *dst, unsigned dst_stride, const void *src, unsigned src_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (dst_suffix, dst_native_type)
-    print '{'
-    print '   void (*func)(%s *dst, unsigned dst_stride, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % dst_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_read_%s;' % (format.short_name(), dst_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(dst, dst_stride, (const uint8_t *)src, src_stride, x, y, w, h);'
-    print '}'
-    print
-
-
-def generate_write(formats, src_channel, src_native_type, src_suffix):
-    '''Generate the dispatch function to write pixels to any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_write(format, src_channel, src_native_type, src_suffix)
-
-    print 'void'
-    print 'util_format_write_%s(enum pipe_format format, const %s *src, unsigned src_stride, void *dst, unsigned dst_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (src_suffix, src_native_type)
-    
-    print '{'
-    print '   void (*func)(const %s *src, unsigned src_stride, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % src_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_write_%s;' % (format.short_name(), src_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(src, src_stride, (uint8_t *)dst, dst_stride, x, y, w, h);'
-    print '}'
-    print
-
-
-def main():
-    formats = []
-    for arg in sys.argv[1:]:
-        formats.extend(parse(arg))
-
-    print '/* This file is autogenerated by u_format_access.py from u_format.csv. Do not edit directly. */'
-    print
-    # This will print the copyright message on the top of this file
-    print __doc__.strip()
-    print
-    print '#include "pipe/p_compiler.h"'
-    print '#include "u_math.h"'
-    print '#include "u_format_pack.h"'
-    print
-
-    generate_srgb_tables()
-
-    type = Channel(FLOAT, False, 32)
-    native_type = 'float'
-    suffix = '4f'
-
-    generate_read(formats, type, native_type, suffix)
-    generate_write(formats, type, native_type, suffix)
-
-    type = Channel(UNSIGNED, True, 8)
-    native_type = 'uint8_t'
-    suffix = '4ub'
-
-    generate_read(formats, type, native_type, suffix)
-    generate_write(formats, type, native_type, suffix)
-
-
-if __name__ == '__main__':
-    main()
diff --git a/src/gallium/auxiliary/util/u_format_pack.py b/src/gallium/auxiliary/util/u_format_pack.py
index 409d024c63..95c0c79be9 100644
--- a/src/gallium/auxiliary/util/u_format_pack.py
+++ b/src/gallium/auxiliary/util/u_format_pack.py
@@ -38,19 +38,59 @@
 
 
 import sys
+import math
 
 from u_format_parse import *
 
 
+def generate_f16_to_f32():
+    '''Naive implementation, need something faster that operates on bits'''
+
+    print '''
+static float
+f16_to_f32(uint16_t h)
+{
+    unsigned mantissa = h & 0x3ff;
+    unsigned exponent = (h >> 10) & 0x1f;
+    float sign = (h & 0x8000) ? -1.0f : 1.0f;
+
+    if (exponent == 0) {
+        if (mantissa == 0) {
+            return sign * 0.0f;
+        }
+        return sign * powf(2.0f, -14.0f) * (float)mantissa / 1024.0f;
+    }
+    if (exponent == 31) {
+        if (mantissa == 0) {
+            /* XXX: infinity */
+            return sign * 100000.0f;
+        }
+        /* XXX: NaN */
+        return 1000.0f;
+    }
+    return sign * powf(2.0f, (float)exponent - 15.0f) * (1.0f + (float)mantissa / 1024.0f);
+}
+'''
+
+def generate_f32_to_f16():
+    print '''
+static uint16_t
+f32_to_f16(float f)
+{
+    /* TODO */
+    return 0;
+}
+'''
+
 def generate_format_type(format):
     '''Generate a structure that describes the format.'''
 
     print 'union util_format_%s {' % format.short_name()
-    if format.is_bitmask():
+    if format.is_bitmask() or format.short_name() == "r11g11b10_float":
         print '   uint%u_t value;' % (format.block_size(),)
     print '   struct {'
     for channel in format.channels:
-        if format.is_bitmask() and not format.is_array():
+        if (format.is_bitmask() or format.is_mixed()) and not format.is_array() or format.short_name() == "r11g11b10_float":
             if channel.type == VOID:
                 if channel.size:
                     print '      unsigned %s:%u;' % (channel.name, channel.size)
@@ -58,6 +98,11 @@ def generate_format_type(format):
                 print '      unsigned %s:%u;' % (channel.name, channel.size)
             elif channel.type == SIGNED:
                 print '      int %s:%u;' % (channel.name, channel.size)
+            elif channel.type == FLOAT:
+                if channel.size == 32:
+                    print '      float %s;' % (channel.name)
+                else:
+                    print '      unsigned %s:%u;' % (channel.name, channel.size)
             else:
                 assert 0
         else:
@@ -85,6 +130,20 @@ def generate_format_type(format):
     print
 
 
+def generate_srgb_tables():
+    print 'static ubyte srgb_to_linear[256] = {'
+    for i in range(256):
+        print '   %s,' % (int(math.pow((i / 255.0 + 0.055) / 1.055, 2.4) * 255))
+    print '};'
+    print
+    print 'static ubyte linear_to_srgb[256] = {'
+    print '   0,'
+    for i in range(1, 256):
+        print '   %s,' % (int((1.055 * math.pow(i / 255.0, 0.41666) - 0.055) * 255))
+    print '};'
+    print
+
+
 def bswap_format(format):
     '''Generate a structure that describes the format.'''
 
@@ -124,15 +183,17 @@ def native_type(format):
             return 'uint%u_t' % format.block_size()
         else:
             # For array pixel formats return the integer type that matches the color channel
-            type = format.channels[0]
-            if type.type == UNSIGNED:
-                return 'uint%u_t' % type.size
-            elif type.type == SIGNED:
-                return 'int%u_t' % type.size
-            elif type.type == FLOAT:
-                if type.size == 32:
+            channel = format.channels[0]
+            if channel.type in (UNSIGNED, VOID):
+                return 'uint%u_t' % channel.size
+            elif channel.type == SIGNED:
+                return 'int%u_t' % channel.size
+            elif channel.type == FLOAT:
+                if channel.size == 16:
+                    return 'uint16_t'
+                elif channel.size == 32:
                     return 'float'
-                elif type.size == 64:
+                elif channel.size == 64:
                     return 'double'
                 else:
                     assert False
@@ -179,31 +240,6 @@ def get_one(type):
         return (1 << get_one_shift(type)) - 1
 
 
-def generate_clamp():
-    '''Code generate the clamping functions for each type.
-
-    We don't use a macro so that arguments with side effects, 
-    like *src_pixel++ are correctly handled.
-    '''
-
-    for suffix, native_type in [
-        ('', 'double'),
-        ('f', 'float'),
-        ('ui', 'unsigned int'),
-        ('si', 'int'),
-    ]:
-        print 'static INLINE %s' % native_type
-        print 'clamp%s(%s value, %s lbound, %s ubound)' % (suffix, native_type, native_type, native_type)
-        print '{'
-        print '   if(value < lbound)'
-        print '      return lbound;'
-        print '   if(value > ubound)'
-        print '      return ubound;'
-        print '   return value;'
-        print '}'
-        print
-
-
 def clamp_expr(src_channel, dst_channel, dst_native_type, value):
     '''Generate the expression to clamp the value in the source type to the
     destination type range.'''
@@ -211,21 +247,6 @@ def clamp_expr(src_channel, dst_channel, dst_native_type, value):
     if src_channel == dst_channel:
         return value
 
-    # Pick the approriate clamp function
-    if src_channel.type == FLOAT:
-        if src_channel.size == 32:
-            func = 'clampf'
-        elif src_channel.size == 64:
-            func = 'clamp'
-        else:
-            assert False
-    elif src_channel.type == UNSIGNED:
-        func = 'clampui'
-    elif src_channel.type == SIGNED:
-        func = 'clampsi'
-    else:
-        assert False
-
     src_min = src_channel.min()
     src_max = src_channel.max()
     dst_min = dst_channel.min()
@@ -250,7 +271,17 @@ def conversion_expr(src_channel, dst_channel, dst_native_type, value, clamp=True
         return value
 
     if src_channel.type == FLOAT and dst_channel.type == FLOAT:
-        return '(%s)%s' % (dst_native_type, value)
+        if src_channel.size == dst_channel.size:
+            return value
+        if src_channel.size == 64:
+            value = '(float)%s' % (value)
+        elif src_channel.size == 16:
+            value = 'f16_to_f32(%s)' % (value)
+        if dst_channel.size == 16:
+            value = 'f32_to_f16(%s)' % (value)
+        elif dst_channel.size == 64:
+            value = '(double)%s' % (value)
+        return value
     
     if clamp:
         value = clamp_expr(src_channel, dst_channel, dst_native_type, value)
@@ -301,33 +332,29 @@ def conversion_expr(src_channel, dst_channel, dst_native_type, value, clamp=True
     assert False
 
 
-def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the function to unpack pixels from a particular format'''
+def generate_unpack_kernel(format, dst_channel, dst_native_type):
 
+    if not is_format_supported(format):
+        return
+    
     assert format.layout == PLAIN
 
-    name = format.short_name()
-
     src_native_type = native_type(format)
 
-    print 'static INLINE void'
-    print 'util_format_%s_unpack_%s(%s *dst, const void *src)' % (name, dst_suffix, dst_native_type)
-    print '{'
-    
     if format.is_bitmask():
         depth = format.block_size()
-        print '   uint%u_t value = *(uint%u_t *)src;' % (depth, depth) 
+        print '         uint%u_t value = *(uint%u_t *)src;' % (depth, depth) 
 
         # Declare the intermediate variables
         for i in range(format.nr_channels()):
             src_channel = format.channels[i]
             if src_channel.type == UNSIGNED:
-                print '   uint%u_t %s;' % (depth, src_channel.name)
+                print '         uint%u_t %s;' % (depth, src_channel.name)
             elif src_channel.type == SIGNED:
-                print '   int%u_t %s;' % (depth, src_channel.name)
+                print '         int%u_t %s;' % (depth, src_channel.name)
 
         print '#ifdef PIPE_ARCH_BIG_ENDIAN'
-        print '   value = util_bswap%u(value);' % depth
+        print '         value = util_bswap%u(value);' % depth
         print '#endif'
 
         # Compute the intermediate unshifted values 
@@ -355,7 +382,7 @@ def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
                 value = None
                 
             if value is not None:
-                print '   %s = %s;' % (src_channel.name, value)
+                print '         %s = %s;' % (src_channel.name, value)
                 
             shift += src_channel.size
 
@@ -379,11 +406,11 @@ def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
                     value = get_one(dst_channel)
                 elif i >= 1:
                     value = 'dst[0]'
-            print '   dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
+            print '         dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
         
     else:
-        print '   union util_format_%s pixel;' % format.short_name()
-        print '   memcpy(&pixel, src, sizeof pixel);'
+        print '         union util_format_%s pixel;' % format.short_name()
+        print '         memcpy(&pixel, src, sizeof pixel);'
         bswap_format(format)
     
         for i in range(4):
@@ -405,36 +432,29 @@ def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
                     value = get_one(dst_channel)
                 elif i >= 1:
                     value = 'dst[0]'
-            print '   dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
-
-    print '}'
-    print
+            print '         dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
     
 
-def generate_format_pack(format, src_channel, src_native_type, src_suffix):
-    '''Generate the function to pack pixels to a particular format'''
-
-    name = format.short_name()
+def generate_pack_kernel(format, src_channel, src_native_type):
 
+    if not is_format_supported(format):
+        return
+    
     dst_native_type = native_type(format)
 
     assert format.layout == PLAIN
 
     inv_swizzle = format.inv_swizzles()
-    
-    print 'static INLINE void'
-    print 'util_format_%s_pack_%s(void *dst, %s r, %s g, %s b, %s a)' % (name, src_suffix, src_native_type, src_native_type, src_native_type, src_native_type)
-    print '{'
-    
+
     if format.is_bitmask():
         depth = format.block_size()
-        print '   uint%u_t value = 0;' % depth 
+        print '         uint%u_t value = 0;' % depth 
 
         shift = 0
         for i in range(4):
             dst_channel = format.channels[i]
             if inv_swizzle[i] is not None:
-                value = 'rgba'[inv_swizzle[i]]
+                value ='src[%u]' % inv_swizzle[i]
                 value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
                 if format.colorspace == ZS:
                     if i == 3:
@@ -452,138 +472,141 @@ def generate_format_pack(format, src_channel, src_native_type, src_suffix):
                 else:
                     value = None
                 if value is not None:
-                    print '   value |= %s;' % (value)
+                    print '         value |= %s;' % (value)
                 
             shift += dst_channel.size
 
         print '#ifdef PIPE_ARCH_BIG_ENDIAN'
-        print '   value = util_bswap%u(value);' % depth
+        print '         value = util_bswap%u(value);' % depth
         print '#endif'
         
-        print '   *(uint%u_t *)dst = value;' % depth 
+        print '         *(uint%u_t *)dst = value;' % depth 
 
     else:
-        print '   union util_format_%s pixel;' % format.short_name()
+        print '         union util_format_%s pixel;' % format.short_name()
     
         for i in range(4):
             dst_channel = format.channels[i]
             width = dst_channel.size
             if inv_swizzle[i] is None:
                 continue
-            value = 'rgba'[inv_swizzle[i]]
+            value ='src[%u]' % inv_swizzle[i]
             value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
             if format.colorspace == ZS:
                 if i == 3:
                     value = get_one(dst_channel)
                 elif i >= 1:
                     value = '0'
-            print '   pixel.chan.%s = %s;' % (dst_channel.name, value)
+            print '         pixel.chan.%s = %s;' % (dst_channel.name, value)
     
         bswap_format(format)
-        print '   memcpy(dst, &pixel, sizeof pixel);'
-        
-    print '}'
-    print
+        print '         memcpy(dst, &pixel, sizeof pixel);'
     
 
-def generate_unpack(formats, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the dispatch function to unpack pixels from any format'''
+def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
+    '''Generate the function to unpack pixels from a particular format'''
 
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix)
+    name = format.short_name()
 
     print 'static INLINE void'
-    print 'util_format_unpack_%s(enum pipe_format format, %s *dst, const void *src)' % (dst_suffix, dst_native_type)
+    print 'util_format_%s_unpack_%s(%s *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, dst_suffix, dst_native_type)
     print '{'
-    print '   void (*func)(%s *dst, const void *src);' % dst_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_unpack_%s;' % (format.short_name(), dst_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(dst, src);'
+
+    if is_format_supported(format):
+        print '   unsigned x, y;'
+        print '   for(y = 0; y < height; y += %u) {' % (format.block_height,)
+        print '      %s *dst = dst_row;' % (dst_native_type)
+        print '      const uint8_t *src = src_row;'
+        print '      for(x = 0; x < width; x += %u) {' % (format.block_width,)
+        
+        generate_unpack_kernel(format, dst_channel, dst_native_type)
+    
+        print '         src += %u;' % (format.block_size() / 8,)
+        print '         dst += 4;'
+        print '      }'
+        print '      src_row += src_stride;'
+        print '      dst_row += dst_stride/sizeof(*dst_row);'
+        print '   }'
+
     print '}'
     print
+    
 
+def generate_format_pack(format, src_channel, src_native_type, src_suffix):
+    '''Generate the function to pack pixels to a particular format'''
 
-def generate_pack(formats, src_channel, src_native_type, src_suffix):
-    '''Generate the dispatch function to pack pixels to any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_pack(format, src_channel, src_native_type, src_suffix)
+    name = format.short_name()
 
     print 'static INLINE void'
-    print 'util_format_pack_%s(enum pipe_format format, void *dst, %s r, %s g, %s b, %s a)' % (src_suffix, src_native_type, src_native_type, src_native_type, src_native_type)
+    print 'util_format_%s_pack_%s(uint8_t *dst_row, unsigned dst_stride, const %s *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, src_suffix, src_native_type)
     print '{'
-    print '   void (*func)(void *dst, %s r, %s g, %s b, %s a);' % (src_native_type, src_native_type, src_native_type, src_native_type)
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_pack_%s;' % (format.short_name(), src_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("%s: unsupported format\\n", __FUNCTION__);'
-    print '      return;'
-    print '   }'
-    print '   func(dst, r, g, b, a);'
+    
+    if is_format_supported(format):
+        print '   unsigned x, y;'
+        print '   for(y = 0; y < height; y += %u) {' % (format.block_height,)
+        print '      const %s *src = src_row;' % (src_native_type)
+        print '      uint8_t *dst = dst_row;'
+        print '      for(x = 0; x < width; x += %u) {' % (format.block_width,)
+    
+        generate_pack_kernel(format, src_channel, src_native_type)
+            
+        print '         src += 4;'
+        print '         dst += %u;' % (format.block_size() / 8,)
+        print '      }'
+        print '      dst_row += dst_stride;'
+        print '      src_row += src_stride/sizeof(*src_row);'
+        print '   }'
+        
     print '}'
     print
+    
 
+def generate_format_fetch(format, dst_channel, dst_native_type, dst_suffix):
+    '''Generate the function to unpack pixels from a particular format'''
 
-def main():
-    formats = []
-    for arg in sys.argv[1:]:
-        formats.extend(parse(arg))
+    name = format.short_name()
 
-    print '/* This file is autogenerated by u_format_pack.py from u_format.csv. Do not edit directly. */'
-    print
-    # This will print the copyright message on the top of this file
-    print __doc__.strip()
+    print 'static INLINE void'
+    print 'util_format_%s_fetch_%s(%s *dst, const uint8_t *src, unsigned i, unsigned j)' % (name, dst_suffix, dst_native_type)
+    print '{'
+
+    if is_format_supported(format):
+        generate_unpack_kernel(format, dst_channel, dst_native_type)
 
+    print '}'
     print
-    print '#ifndef U_FORMAT_PACK_H'
-    print '#define U_FORMAT_PACK_H'
+
+
+def generate(formats):
     print
     print '#include "pipe/p_compiler.h"'
     print '#include "u_math.h"'
     print '#include "u_format.h"'
     print
 
-    generate_clamp()
+    generate_f16_to_f32()
+    generate_f32_to_f16()
 
     for format in formats:
-        if format.layout == PLAIN:
+        if is_format_supported(format):
             generate_format_type(format)
 
     channel = Channel(FLOAT, False, 32)
     native_type = 'float'
-    suffix = '4f'
+    suffix = 'float'
 
-    generate_unpack(formats, channel, native_type, suffix)
-    generate_pack(formats, channel, native_type, suffix)
+    for format in formats:
+        if format.layout != 's3tc':
+            generate_format_unpack(format, channel, native_type, suffix)
+            generate_format_pack(format, channel, native_type, suffix)
+            generate_format_fetch(format, channel, native_type, suffix)
 
     channel = Channel(UNSIGNED, True, 8)
     native_type = 'uint8_t'
-    suffix = '4ub'
-
-    generate_unpack(formats, channel, native_type, suffix)
-    generate_pack(formats, channel, native_type, suffix)
-
-    print
-    print '#ifdef __cplusplus'
-    print '}'
-    print '#endif'
-    print
-    print '#endif /* ! U_FORMAT_PACK_H */'
+    suffix = '8unorm'
 
+    for format in formats:
+        if format.layout != 's3tc':
+            generate_format_unpack(format, channel, native_type, suffix)
+            generate_format_pack(format, channel, native_type, suffix)
 
-if __name__ == '__main__':
-    main()
diff --git a/src/gallium/auxiliary/util/u_format_parse.py b/src/gallium/auxiliary/util/u_format_parse.py
index f74dc5e88a..03dc62d329 100755
--- a/src/gallium/auxiliary/util/u_format_parse.py
+++ b/src/gallium/auxiliary/util/u_format_parse.py
@@ -142,6 +142,8 @@ class Format:
 
     def is_mixed(self):
         ref_channel = self.channels[0]
+        if ref_channel.type == VOID:
+           ref_channel = self.channels[1]
         for channel in self.channels[1:]:
             if channel.type != VOID:
                 if channel.type != ref_channel.type:
diff --git a/src/gallium/auxiliary/util/u_format_s3tc.c b/src/gallium/auxiliary/util/u_format_s3tc.c
new file mode 100644
index 0000000000..4fe00b0e43
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_s3tc.c
@@ -0,0 +1,510 @@
+/**************************************************************************
+ *
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#include "u_dl.h"
+#include "u_math.h"
+#include "u_format.h"
+#include "u_format_s3tc.h"
+
+
+#if defined(_WIN32) || defined(WIN32)
+#define DXTN_LIBNAME "dxtn.dll"
+#else
+#define DXTN_LIBNAME "libtxc_dxtn.so"
+#endif
+
+
+boolean util_format_s3tc_enabled = FALSE;
+
+util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch = NULL;
+util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch = NULL;
+util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch = NULL;
+util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch = NULL;
+
+util_format_dxtn_pack_t util_format_dxtn_pack = NULL;
+
+
+void
+util_format_s3tc_init(void)
+{
+   static struct util_dl_library *
+   library = NULL;
+
+   if (util_format_s3tc_enabled)
+      return;
+
+   if (!library) {
+      library = util_dl_open(DXTN_LIBNAME);
+      if (!library) {
+	 debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
+	    "compression/decompression unavailable");
+      }
+      else {
+         util_format_dxt1_rgb_fetch = (util_format_dxtn_fetch_t)
+            util_dl_get_proc_address(library, "fetch_2d_texel_rgb_dxt1");
+         util_format_dxt1_rgba_fetch = (util_format_dxtn_fetch_t)
+            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt1");
+         util_format_dxt3_rgba_fetch = (util_format_dxtn_fetch_t)
+            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt3");
+         util_format_dxt5_rgba_fetch = (util_format_dxtn_fetch_t)
+            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt5");
+         util_format_dxtn_pack = (util_format_dxtn_pack_t)
+            util_dl_get_proc_address(library, "tx_compress_dxtn");
+
+         if (util_format_dxt1_rgb_fetch ||
+             util_format_dxt1_rgba_fetch ||
+             util_format_dxt3_rgba_fetch ||
+             util_format_dxt5_rgba_fetch ||
+             util_format_dxtn_pack) {
+            util_format_s3tc_enabled = TRUE;
+            debug_printf("software DXTn compression/decompression available");
+
+         } else {
+	    debug_printf("couldn't reference all symbols in "
+	       DXTN_LIBNAME ", software DXTn compression/decompression "
+	       "unavailable");
+         }
+      }
+   }
+}
+
+
+/*
+ * Pixel fetch.
+ */
+
+void
+util_format_dxt1_rgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt1_rgb_fetch) {
+      util_format_dxt1_rgb_fetch(0, src, i, j, dst);
+   }
+}
+
+void
+util_format_dxt1_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt1_rgba_fetch) {
+      util_format_dxt1_rgba_fetch(0, src, i, j, dst);
+   }
+}
+
+void
+util_format_dxt3_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt3_rgba_fetch) {
+      util_format_dxt3_rgba_fetch(0, src, i, j, dst);
+   }
+}
+
+void
+util_format_dxt5_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt5_rgba_fetch) {
+      util_format_dxt5_rgba_fetch(0, src, i, j, dst);
+   }
+}
+
+void
+util_format_dxt1_rgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt1_rgb_fetch) {
+      uint8_t tmp[4];
+      util_format_dxt1_rgb_fetch(0, src, i, j, tmp);
+      dst[0] = ubyte_to_float(tmp[0]);
+      dst[1] = ubyte_to_float(tmp[1]);
+      dst[2] = ubyte_to_float(tmp[2]);
+      dst[3] = 1.0;
+   }
+}
+
+void
+util_format_dxt1_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt1_rgba_fetch) {
+      uint8_t tmp[4];
+      util_format_dxt1_rgba_fetch(0, src, i, j, tmp);
+      dst[0] = ubyte_to_float(tmp[0]);
+      dst[1] = ubyte_to_float(tmp[1]);
+      dst[2] = ubyte_to_float(tmp[2]);
+      dst[3] = ubyte_to_float(tmp[3]);
+   }
+}
+
+void
+util_format_dxt3_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt3_rgba_fetch) {
+      uint8_t tmp[4];
+      util_format_dxt3_rgba_fetch(0, src, i, j, tmp);
+      dst[0] = ubyte_to_float(tmp[0]);
+      dst[1] = ubyte_to_float(tmp[1]);
+      dst[2] = ubyte_to_float(tmp[2]);
+      dst[3] = ubyte_to_float(tmp[3]);
+   }
+}
+
+void
+util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   if (util_format_dxt5_rgba_fetch) {
+      uint8_t tmp[4];
+      util_format_dxt5_rgba_fetch(0, src, i, j, tmp);
+      dst[0] = ubyte_to_float(tmp[0]);
+      dst[1] = ubyte_to_float(tmp[1]);
+      dst[2] = ubyte_to_float(tmp[2]);
+      dst[3] = ubyte_to_float(tmp[3]);
+   }
+}
+
+
+/*
+ * Block decompression.
+ *
+ * FIXME
+ */
+
+void
+util_format_dxt1_rgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt1_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+
+}
+
+void
+util_format_dxt3_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt5_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt1_rgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   if (util_format_dxt1_rgb_fetch) {
+      unsigned x, y, i, j;
+      for(y = 0; y < height; y += 4) {
+         const uint8_t *src = src_row;
+         for(x = 0; x < width; x += 4) {
+            for(j = 0; j < 4; ++j) {
+               for(i = 0; i < 4; ++i) {
+                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+                  uint8_t tmp[4];
+                  util_format_dxt1_rgb_fetch(0, src, i, j, tmp);
+                  dst[0] = ubyte_to_float(tmp[0]);
+                  dst[1] = ubyte_to_float(tmp[1]);
+                  dst[2] = ubyte_to_float(tmp[2]);
+                  dst[3] = 1.0;
+               }
+            }
+            src += 8;
+         }
+         src_row += src_stride;
+      }
+   }
+}
+
+void
+util_format_dxt1_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   if (util_format_dxt1_rgba_fetch) {
+      unsigned x, y, i, j;
+      for(y = 0; y < height; y += 4) {
+         const uint8_t *src = src_row;
+         for(x = 0; x < width; x += 4) {
+            for(j = 0; j < 4; ++j) {
+               for(i = 0; i < 4; ++i) {
+                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+                  uint8_t tmp[4];
+                  util_format_dxt1_rgba_fetch(0, src, i, j, tmp);
+                  dst[0] = ubyte_to_float(tmp[0]);
+                  dst[1] = ubyte_to_float(tmp[1]);
+                  dst[2] = ubyte_to_float(tmp[2]);
+                  dst[3] = ubyte_to_float(tmp[3]);
+               }
+            }
+            src += 8;
+         }
+         src_row += src_stride;
+      }
+   }
+}
+
+void
+util_format_dxt3_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   if (util_format_dxt3_rgba_fetch) {
+      unsigned x, y, i, j;
+      for(y = 0; y < height; y += 4) {
+         const uint8_t *src = src_row;
+         for(x = 0; x < width; x += 4) {
+            for(j = 0; j < 4; ++j) {
+               for(i = 0; i < 4; ++i) {
+                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+                  uint8_t tmp[4];
+                  util_format_dxt3_rgba_fetch(0, src, i, j, tmp);
+                  dst[0] = ubyte_to_float(tmp[0]);
+                  dst[1] = ubyte_to_float(tmp[1]);
+                  dst[2] = ubyte_to_float(tmp[2]);
+                  dst[3] = ubyte_to_float(tmp[3]);
+               }
+            }
+            src += 16;
+         }
+         src_row += src_stride;
+      }
+   }
+}
+
+void
+util_format_dxt5_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   if (util_format_dxt5_rgba_fetch) {
+      unsigned x, y, i, j;
+      for(y = 0; y < height; y += 4) {
+         const uint8_t *src = src_row;
+         for(x = 0; x < width; x += 4) {
+            for(j = 0; j < 4; ++j) {
+               for(i = 0; i < 4; ++i) {
+                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+                  uint8_t tmp[4];
+                  util_format_dxt5_rgba_fetch(0, src, i, j, tmp);
+                  dst[0] = ubyte_to_float(tmp[0]);
+                  dst[1] = ubyte_to_float(tmp[1]);
+                  dst[2] = ubyte_to_float(tmp[2]);
+                  dst[3] = ubyte_to_float(tmp[3]);
+               }
+            }
+            src += 16;
+         }
+         src_row += src_stride;
+      }
+   }
+}
+
+
+/*
+ * Block compression.
+ *
+ * FIXME
+ */
+
+void
+util_format_dxt1_rgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt1_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt3_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt5_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt1_rgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt1_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt3_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+void
+util_format_dxt5_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+}
+
+
+/*
+ * SRGB variants.
+ *
+ * FIXME: shunts to RGB for now
+ */
+
+void
+util_format_dxt1_srgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgb_fetch_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgba_fetch_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt3_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt3_rgba_fetch_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt5_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt5_rgba_fetch_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgb_fetch_float(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgba_fetch_float(dst, src, i, j);
+}
+
+void
+util_format_dxt3_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt3_rgba_fetch_float(dst, src, i, j);
+}
+
+void
+util_format_dxt5_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt5_rgba_fetch_float(dst, src, i, j);
+}
+
diff --git a/src/gallium/auxiliary/util/u_format_s3tc.h b/src/gallium/auxiliary/util/u_format_s3tc.h
new file mode 100644
index 0000000000..5c3a9dd22c
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_s3tc.h
@@ -0,0 +1,218 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#ifndef U_FORMAT_S3TC_H_
+#define U_FORMAT_S3TC_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+enum util_format_dxtn {
+  UTIL_FORMAT_DXT1_RGB = 0x83F0,
+  UTIL_FORMAT_DXT1_RGBA = 0x83F1,
+  UTIL_FORMAT_DXT3_RGBA = 0x83F2,
+  UTIL_FORMAT_DXT5_RGBA = 0x83F3
+};
+
+
+typedef void
+(*util_format_dxtn_fetch_t)( int src_stride,
+                             const uint8_t *src,
+                             int col, int row,
+                             uint8_t *dst );
+
+typedef void
+(*util_format_dxtn_pack_t)( int src_comps,
+                            int width, int height,
+                            const uint8_t *src,
+                            enum util_format_dxtn dst_format,
+                            uint8_t *dst,
+                            int dst_stride);
+
+extern boolean util_format_s3tc_enabled;
+
+extern util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch;
+
+extern util_format_dxtn_pack_t util_format_dxtn_pack;
+
+
+void
+util_format_s3tc_init(void);
+
+
+void
+util_format_dxt1_rgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+
+void
+util_format_dxt1_rgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+
+#endif /* U_FORMAT_S3TC_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_table.py b/src/gallium/auxiliary/util/u_format_table.py
index fb68852a53..94a4331b15 100755
--- a/src/gallium/auxiliary/util/u_format_table.py
+++ b/src/gallium/auxiliary/util/u_format_table.py
@@ -3,7 +3,7 @@
 '''
 /**************************************************************************
  *
- * Copyright 2009 VMware, Inc.
+ * Copyright 2010 VMware, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -33,6 +33,7 @@
 import sys
 
 from u_format_parse import *
+import u_format_pack
 
 
 def layout_map(layout):
@@ -85,23 +86,11 @@ def write_format_table(formats):
     print __doc__.strip()
     print
     print '#include "u_format.h"'
+    print '#include "u_format_s3tc.h"'
     print
-    print 'const struct util_format_description'
-    print 'util_format_none_description = {'
-    print "   PIPE_FORMAT_NONE,"
-    print "   \"PIPE_FORMAT_NONE\","
-    print "   \"none\","
-    print "   {0, 0, 0},"
-    print "   0,"
-    print "   0,"
-    print "   0,"
-    print "   0,"
-    print "   0,"
-    print "   {{0, 0, 0}, {0, 0, 0}, {0, 0, 0}, {0, 0, 0}},"
-    print "   {0, 0, 0, 0},"
-    print "   0"
-    print "};"
-    print
+    
+    u_format_pack.generate(formats)
+    
     for format in formats:
         print 'const struct util_format_description'
         print 'util_format_%s_description = {' % (format.short_name(),)
@@ -140,8 +129,14 @@ def write_format_table(formats):
             print "      %s%s\t/* %s */" % (swizzle_map[swizzle], sep, comment)
         print "   },"
         print "   %s," % (colorspace_map(format.colorspace),)
+        print "   &util_format_%s_unpack_8unorm," % format.short_name() 
+        print "   &util_format_%s_pack_8unorm," % format.short_name() 
+        print "   &util_format_%s_unpack_float," % format.short_name() 
+        print "   &util_format_%s_pack_float," % format.short_name() 
+        print "   &util_format_%s_fetch_float" % format.short_name() 
         print "};"
         print
+        
     print "const struct util_format_description *"
     print "util_format_description(enum pipe_format format)"
     print "{"
@@ -150,8 +145,6 @@ def write_format_table(formats):
     print "   }"
     print
     print "   switch (format) {"
-    print "   case PIPE_FORMAT_NONE:"
-    print "      return &util_format_none_description;"
     for format in formats:
         print "   case %s:" % format.name
         print "      return &util_format_%s_description;" % (format.short_name(),)
diff --git a/src/gallium/auxiliary/util/u_format_tests.c b/src/gallium/auxiliary/util/u_format_tests.c
index 9d6debcd8c..4c4cc067a5 100644
--- a/src/gallium/auxiliary/util/u_format_tests.c
+++ b/src/gallium/auxiliary/util/u_format_tests.c
@@ -120,12 +120,11 @@ util_format_test_cases[] =
     * 16-bit rendertarget formats
     */
 
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x0000), {0.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x001f), {0.0, 0.0, 1.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x03e0), {0.0, 1.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7c00), {1.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x8000), {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0xffff), {1.0, 1.0, 1.0, 1.0}},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x0000), {0.0, 0.0, 0.0, 1.0}},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x001f), {0.0, 0.0, 1.0, 1.0}},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x03e0), {0.0, 1.0, 0.0, 1.0}},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7c00), {1.0, 0.0, 0.0, 1.0}},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7fff), {1.0, 1.0, 1.0, 1.0}},
 
    {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), {0.0, 0.0, 0.0, 0.0}},
    {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x001f), {0.0, 0.0, 1.0, 0.0}},
diff --git a/src/gallium/auxiliary/util/u_memory.h b/src/gallium/auxiliary/util/u_memory.h
index a2fc597356..53d56599fe 100644
--- a/src/gallium/auxiliary/util/u_memory.h
+++ b/src/gallium/auxiliary/util/u_memory.h
@@ -88,7 +88,7 @@ mem_dup(const void *src, uint size)
 /**
  * Offset of a field in a struct, in bytes.
  */
-#define Offset(TYPE, MEMBER) ((unsigned)&(((TYPE *)NULL)->MEMBER))
+#define Offset(TYPE, MEMBER) ((uintptr_t)&(((TYPE *)NULL)->MEMBER))
 
 
 
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index c5fd7a6783..ad4a06dc77 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -446,17 +446,17 @@ util_pack_z(enum pipe_format format, double z)
       return (uint) (z * 0xffffffff);
    case PIPE_FORMAT_Z32_FLOAT:
       return (uint)z;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       if (z == 1.0)
          return 0xffffff;
       return (uint) (z * 0xffffff);
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       if (z == 1.0)
          return 0xffffff00;
       return ((uint) (z * 0xffffff)) << 8;
-   case PIPE_FORMAT_S8_UNORM:
+   case PIPE_FORMAT_S8_USCALED:
       /* this case can get it via util_pack_z_stencil() */
       return 0;
    default:
@@ -477,13 +477,13 @@ util_pack_z_stencil(enum pipe_format format, double z, uint s)
    unsigned packed = util_pack_z(format, z);
 
    switch (format) {
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       packed |= s << 24;
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       packed |= s;
       break;
-   case PIPE_FORMAT_S8_UNORM:
+   case PIPE_FORMAT_S8_USCALED:
       packed |= s;
       break;
    default:
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 09b2382733..43b45eb926 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -108,436 +108,6 @@ pipe_put_tile_raw(struct pipe_context *pipe,
 
 
 
-/*** PIPE_FORMAT_B8G8R8A8_UNORM ***/
-
-static void
-a8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  0) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >> 24) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B8G8R8X8_UNORM ***/
-
-static void
-x8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  0) & 0xff);
-         pRow[3] = 1.0F;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-x8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         *dst++ = (0xff << 24) | (r << 16) | (g << 8) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
-
-static void
-b8g8r8a8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >> 24) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >>  0) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-b8g8r8a8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (b << 24) | (g << 16) | (r << 8) | a;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_A8B8G8R8_UNORM ***/
-
-static void
-r8g8b8a8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 24) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >>  0) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r8g8b8a8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (r << 24) | (g << 16) | (b << 8) | a;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B5G5R5X1_UNORM ***/
-
-static void
-x1r5g5b5_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >> 10) & 0x1f) * (1.0f / 31.0f);
-         pRow[1] = ((pixel >>  5) & 0x1f) * (1.0f / 31.0f);
-         pRow[2] = ((pixel      ) & 0x1f) * (1.0f / 31.0f);
-         pRow[3] = 1.0f;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-x1r5g5b5_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         r = r >> 3;  /* 5 bits */
-         g = g >> 3;  /* 5 bits */
-         b = b >> 3;  /* 5 bits */
-         *dst++ = (1 << 15) | (r << 10) | (g << 5) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B5G5R5A1_UNORM ***/
-
-static void
-a1r5g5b5_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >> 10) & 0x1f) * (1.0f / 31.0f);
-         pRow[1] = ((pixel >>  5) & 0x1f) * (1.0f / 31.0f);
-         pRow[2] = ((pixel      ) & 0x1f) * (1.0f / 31.0f);
-         pRow[3] = ((pixel >> 15)       ) * 1.0f;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a1r5g5b5_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         r = r >> 3;  /* 5 bits */
-         g = g >> 3;  /* 5 bits */
-         b = b >> 3;  /* 5 bits */
-         a = a >> 7;  /* 1 bit */
-         *dst++ = (a << 15) | (r << 10) | (g << 5) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B4G4R4A4_UNORM ***/
-
-static void
-a4r4g4b4_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >>  8) & 0xf) * (1.0f / 15.0f);
-         pRow[1] = ((pixel >>  4) & 0xf) * (1.0f / 15.0f);
-         pRow[2] = ((pixel      ) & 0xf) * (1.0f / 15.0f);
-         pRow[3] = ((pixel >> 12)      ) * (1.0f / 15.0f);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a4r4g4b4_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         r >>= 4;
-         g >>= 4;
-         b >>= 4;
-         a >>= 4;
-         *dst++ = (a << 12) | (r << 8) | (g << 4) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B5G6R5_UNORM ***/
-
-static void
-r5g6b5_get_tile_rgba(const ushort *src,
-                     unsigned w, unsigned h,
-                     float *p,
-                     unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >> 11) & 0x1f) * (1.0f / 31.0f);
-         pRow[1] = ((pixel >>  5) & 0x3f) * (1.0f / 63.0f);
-         pRow[2] = ((pixel      ) & 0x1f) * (1.0f / 31.0f);
-         pRow[3] = 1.0f;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r5g6b5_put_tile_rgba(ushort *dst,
-                     unsigned w, unsigned h,
-                     const float *p,
-                     unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         uint r = (uint) (CLAMP(pRow[0], 0.0, 1.0) * 31.0);
-         uint g = (uint) (CLAMP(pRow[1], 0.0, 1.0) * 63.0);
-         uint b = (uint) (CLAMP(pRow[2], 0.0, 1.0) * 31.0);
-         *dst++ = (r << 11) | (g << 5) | (b);
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_R8G8B8_UNORM ***/
-
-static void
-r8g8b8_get_tile_rgba(const ubyte *src,
-                     unsigned w, unsigned h,
-                     float *p,
-                     unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         pRow[0] = ubyte_to_float(src[0]);
-         pRow[1] = ubyte_to_float(src[1]);
-         pRow[2] = ubyte_to_float(src[2]);
-         pRow[3] = 1.0f;
-         src += 3;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r8g8b8_put_tile_rgba(ubyte *dst,
-                     unsigned w, unsigned h,
-                     const float *p,
-                     unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         dst[0] = float_to_ubyte(pRow[0]);
-         dst[1] = float_to_ubyte(pRow[1]);
-         dst[2] = float_to_ubyte(pRow[2]);
-         dst += 3;
-      }
-      p += src_stride;
-   }
-}
-
-
-
 /*** PIPE_FORMAT_Z16_UNORM ***/
 
 /**
@@ -567,179 +137,6 @@ z16_get_tile_rgba(const ushort *src,
 
 
 
-/*** PIPE_FORMAT_L8_UNORM ***/
-
-static void
-l8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = ubyte_to_float(*src);
-         pRow[3] = 1.0;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-l8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r;
-         r = float_to_ubyte(pRow[0]);
-         *dst++ = (ubyte) r;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_A8_UNORM ***/
-
-static void
-a8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = 0.0;
-         pRow[3] = ubyte_to_float(*src);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned a;
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (ubyte) a;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_R16_SNORM ***/
-
-static void
-r16_get_tile_rgba(const short *src,
-                  unsigned w, unsigned h,
-                  float *p,
-                  unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] = SHORT_TO_FLOAT(src[0]);
-         pRow[1] =
-         pRow[2] = 0.0;
-         pRow[3] = 1.0;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r16_put_tile_rgba(short *dst,
-                  unsigned w, unsigned h,
-                  const float *p,
-                  unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, dst++, pRow += 4) {
-         UNCLAMPED_FLOAT_TO_SHORT(dst[0], pRow[0]);
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_R16G16B16A16_SNORM ***/
-
-static void
-r16g16b16a16_get_tile_rgba(const short *src,
-                           unsigned w, unsigned h,
-                           float *p,
-                           unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src += 4, pRow += 4) {
-         pRow[0] = SHORT_TO_FLOAT(src[0]);
-         pRow[1] = SHORT_TO_FLOAT(src[1]);
-         pRow[2] = SHORT_TO_FLOAT(src[2]);
-         pRow[3] = SHORT_TO_FLOAT(src[3]);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r16g16b16a16_put_tile_rgba(short *dst,
-                           unsigned w, unsigned h,
-                           const float *p,
-                           unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, dst += 4, pRow += 4) {
-         UNCLAMPED_FLOAT_TO_SHORT(dst[0], pRow[0]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[1], pRow[1]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[2], pRow[2]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[3], pRow[3]);
-      }
-      p += src_stride;
-   }
-}
-
-
 /*** PIPE_FORMAT_A8B8G8R8_SRGB ***/
 
 /**
@@ -919,96 +316,6 @@ l8_srgb_put_tile_rgba(ubyte *dst,
 }
 
 
-/*** PIPE_FORMAT_I8_UNORM ***/
-
-static void
-i8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] =
-         pRow[3] = ubyte_to_float(*src);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-i8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r;
-         r = float_to_ubyte(pRow[0]);
-         *dst++ = (ubyte) r;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_L8A8_UNORM ***/
-
-static void
-a8l8_get_tile_rgba(const ushort *src,
-                   unsigned w, unsigned h,
-                   float *p,
-                   unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         ushort p = *src++;
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = ubyte_to_float(p & 0xff);
-         pRow[3] = ubyte_to_float(p >> 8);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8l8_put_tile_rgba(ushort *dst,
-                   unsigned w, unsigned h,
-                   const float *p,
-                   unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, a;
-         r = float_to_ubyte(pRow[0]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 8) | r;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-
 /*** PIPE_FORMAT_Z32_UNORM ***/
 
 /**
@@ -1036,7 +343,7 @@ z32_get_tile_rgba(const unsigned *src,
 }
 
 
-/*** PIPE_FORMAT_Z24S8_UNORM ***/
+/*** PIPE_FORMAT_Z24_UNORM_S8_USCALED ***/
 
 /**
  * Return Z component as four float in [0,1].  Stencil part ignored.
@@ -1063,7 +370,7 @@ s8z24_get_tile_rgba(const unsigned *src,
 }
 
 
-/*** PIPE_FORMAT_S8Z24_UNORM ***/
+/*** PIPE_FORMAT_S8_USCALED_Z24_UNORM ***/
 
 /**
  * Return Z component as four float in [0,1].  Stencil part ignored.
@@ -1211,51 +518,6 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
                       float *dst, unsigned dst_stride)
 {
    switch (format) {
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      a8r8g8b8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
-      x8r8g8b8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      b8g8r8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
-      r8g8b8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5X1_UNORM:
-      x1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5A1_UNORM:
-      a1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B4G4R4A4_UNORM:
-      a4r4g4b4_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B5G6R5_UNORM:
-      r5g6b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      r8g8b8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8A8_UNORM:
-      a8l8_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
-      break;
    case PIPE_FORMAT_B8G8R8A8_SRGB:
       a8r8g8b8_srgb_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
@@ -1271,11 +533,11 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
    case PIPE_FORMAT_Z32_UNORM:
       z32_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       s8z24_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       z24s8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
@@ -1408,51 +670,6 @@ pipe_put_tile_rgba(struct pipe_context *pipe,
       return;
 
    switch (format) {
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      a8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
-      x8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      b8g8r8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
-      r8g8b8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5X1_UNORM:
-      x1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5A1_UNORM:
-      a1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B5G6R5_UNORM:
-      r5g6b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      r8g8b8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B4G4R4A4_UNORM:
-      a4r4g4b4_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8A8_UNORM:
-      a8l8_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_put_tile_rgba((short *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_put_tile_rgba((short *) packed, w, h, p, src_stride);
-      break;
    case PIPE_FORMAT_B8G8R8A8_SRGB:
       a8r8g8b8_srgb_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
       break;
@@ -1468,11 +685,11 @@ pipe_put_tile_rgba(struct pipe_context *pipe,
    case PIPE_FORMAT_Z32_UNORM:
       /*z32_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       /*s8z24_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       /*z24s8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
@@ -1525,7 +742,7 @@ pipe_get_tile_z(struct pipe_context *pipe,
          }
       }
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       {
          const uint *ptrc
@@ -1540,7 +757,7 @@ pipe_get_tile_z(struct pipe_context *pipe,
          }
       }
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       {
          const uint *ptrc
@@ -1609,7 +826,7 @@ pipe_put_tile_z(struct pipe_context *pipe,
          }
       }
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
          assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
@@ -1636,7 +853,7 @@ pipe_put_tile_z(struct pipe_context *pipe,
          }
       }
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
          assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
diff --git a/src/gallium/docs/d3d11ddi.txt b/src/gallium/docs/d3d11ddi.txt
index 0baf268287..8f2509ce03 100644
--- a/src/gallium/docs/d3d11ddi.txt
+++ b/src/gallium/docs/d3d11ddi.txt
@@ -138,7 +138,7 @@ ClearUnorderedAccessViewUint (D3D11 only)
 	- Gallium does not support unordered access views (views that can be written to arbitrarily from compute shaders)
 
 CreateBlendState (extended in D3D10.1) -> create_blend_state
-	# D3D10 does not support per-RT blending, only D3D10.1 does
+	# D3D10 does not support per-RT blend modes (but per-RT blending), only D3D10.1 does
 	- Gallium lacks alpha-to-coverage
 	+ Gallium supports logic ops
 	+ Gallium supports dithering
diff --git a/src/gallium/docs/source/cso/rasterizer.rst b/src/gallium/docs/source/cso/rasterizer.rst
index e8dc82964f..56a601a8d0 100644
--- a/src/gallium/docs/source/cso/rasterizer.rst
+++ b/src/gallium/docs/source/cso/rasterizer.rst
@@ -81,8 +81,8 @@ point_quad_rasterization
 Determines if points should be rasterized as quads or points. Certain APIs,
 like Direct3D, always use quad rasterization for points, regardless of
 whether point sprites are enabled or not. If this state is enabled, point
-smoothing and antialiasing are disabled, and sprite coordinates are not
-generated.
+smoothing and antialiasing are disabled. If it is disabled, point sprite
+coordinates are not generated.
 
 .. note::
 
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
index c54576b3c3..628bc1c694 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
@@ -1859,7 +1859,7 @@ gen_depth_stencil(struct cell_context *cell,
    spe_comment(f, 0, "Fetch Z/stencil quad from tile");
 
    switch(zs_format) {
-   case PIPE_FORMAT_Z24S8_UNORM: /* fall through */
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED: /* fall through */
    case PIPE_FORMAT_Z24X8_UNORM:
       /* prepare mask to extract Z vals from ZS vals */
       spe_load_uint(f, zmask_reg, 0x00ffffff);
@@ -1880,7 +1880,7 @@ gen_depth_stencil(struct cell_context *cell,
       spe_rotmi(f, fbS_reg, fbZS_reg, -24);
       break;
 
-   case PIPE_FORMAT_S8Z24_UNORM: /* fall through */
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM: /* fall through */
    case PIPE_FORMAT_X8Z24_UNORM:
       /* convert fragment Z from [0,1] to 32-bit ints */
       spe_cfltu(f, fragZ_reg, fragZ_reg, 32);
@@ -1969,12 +1969,12 @@ gen_depth_stencil(struct cell_context *cell,
        * fbS_reg has four 8-bit Z values in bits [7..0].
        */
       spe_comment(f, 0, "Store quad's depth/stencil values in tile");
-      if (zs_format == PIPE_FORMAT_Z24S8_UNORM ||
+      if (zs_format == PIPE_FORMAT_Z24_UNORM_S8_USCALED ||
           zs_format == PIPE_FORMAT_Z24X8_UNORM) {
          spe_shli(f, fbS_reg, fbS_reg, 24); /* fbS = fbS << 24 */
          spe_or(f, fbZS_reg, fbS_reg, fbZ_reg); /* fbZS = fbS | fbZ */
       }
-      else if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
+      else if (zs_format == PIPE_FORMAT_S8_USCALED_Z24_UNORM ||
                zs_format == PIPE_FORMAT_X8Z24_UNORM) {
          spe_shli(f, fbZ_reg, fbZ_reg, 8); /* fbZ = fbZ << 8 */
          spe_or(f, fbZS_reg, fbS_reg, fbZ_reg); /* fbZS = fbS | fbZ */
@@ -1985,7 +1985,7 @@ gen_depth_stencil(struct cell_context *cell,
       else if (zs_format == PIPE_FORMAT_Z16_UNORM) {
          spe_move(f, fbZS_reg, fbZ_reg); /* fbZS = fbZ */
       }
-      else if (zs_format == PIPE_FORMAT_S8_UNORM) {
+      else if (zs_format == PIPE_FORMAT_S8_USCALED) {
          ASSERT(0);   /* XXX to do */
       }
       else {
@@ -2015,7 +2015,7 @@ gen_depth_stencil(struct cell_context *cell,
  * code before the fragment shader to cull fragments/quads that are
  * totally occluded/discarded.
  *
- * XXX we only support PIPE_FORMAT_S8Z24_UNORM z/stencil buffer right now.
+ * XXX we only support PIPE_FORMAT_S8_USCALED_Z24_UNORM z/stencil buffer right now.
  *
  * See the spu_default_fragment_ops() function to see how the per-fragment
  * operations would be done with ordinary C code.
diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c
index 4204574406..363d1e65e7 100644
--- a/src/gallium/drivers/cell/ppu/cell_screen.c
+++ b/src/gallium/drivers/cell/ppu/cell_screen.c
@@ -147,7 +147,7 @@ cell_is_format_supported( struct pipe_screen *screen,
 
    /* only a few formats are known to work at this time */
    switch (format) {
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
    case PIPE_FORMAT_B8G8R8A8_UNORM:
    case PIPE_FORMAT_I8_UNORM:
diff --git a/src/gallium/drivers/cell/spu/spu_command.c b/src/gallium/drivers/cell/spu/spu_command.c
index 79f1fb7fb2..f16cabc027 100644
--- a/src/gallium/drivers/cell/spu/spu_command.c
+++ b/src/gallium/drivers/cell/spu/spu_command.c
@@ -337,8 +337,8 @@ cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
       spu.fb.zsize = 4;
       spu.fb.zscale = (float) 0xffffffffu;
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_X8Z24_UNORM:
    case PIPE_FORMAT_Z24X8_UNORM:
       spu.fb.zsize = 4;
diff --git a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
index 2c9e7458af..3b9566042a 100644
--- a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
+++ b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
@@ -138,13 +138,13 @@ spu_fallback_fragment_ops(uint x, uint y,
 
       if (spu.depth_stencil_alpha.stencil[0].enabled) {
          /* do stencil test */
-         ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z24S8_UNORM);
+         ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z24_UNORM_S8_USCALED);
 
       }
       else if (spu.depth_stencil_alpha.depth.enabled) {
          /* do depth test */
 
-         ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z24S8_UNORM ||
+         ASSERT(spu.fb.depth_format == PIPE_FORMAT_Z24_UNORM_S8_USCALED ||
                 spu.fb.depth_format == PIPE_FORMAT_Z24X8_UNORM);
 
          vector unsigned int ifragZ;
diff --git a/src/gallium/drivers/i915/i915_screen.c b/src/gallium/drivers/i915/i915_screen.c
index 53ef6e50f6..4ccc66b037 100644
--- a/src/gallium/drivers/i915/i915_screen.c
+++ b/src/gallium/drivers/i915/i915_screen.c
@@ -178,13 +178,13 @@ i915_is_format_supported(struct pipe_screen *screen,
       PIPE_FORMAT_L8A8_UNORM,
       PIPE_FORMAT_UYVY,
       PIPE_FORMAT_YUYV,
-      PIPE_FORMAT_Z24S8_UNORM,
+      PIPE_FORMAT_Z24_UNORM_S8_USCALED,
       PIPE_FORMAT_NONE  /* list terminator */
    };
    static const enum pipe_format surface_supported[] = {
       PIPE_FORMAT_B8G8R8A8_UNORM,
       PIPE_FORMAT_B5G6R5_UNORM,
-      PIPE_FORMAT_Z24S8_UNORM,
+      PIPE_FORMAT_Z24_UNORM_S8_USCALED,
       PIPE_FORMAT_NONE  /* list terminator */
    };
    const enum pipe_format *list;
diff --git a/src/gallium/drivers/i915/i915_state_emit.c b/src/gallium/drivers/i915/i915_state_emit.c
index 04a3924aaf..2d212abfb9 100644
--- a/src/gallium/drivers/i915/i915_state_emit.c
+++ b/src/gallium/drivers/i915/i915_state_emit.c
@@ -50,7 +50,7 @@ static unsigned translate_format( enum pipe_format format )
 static unsigned translate_depth_format( enum pipe_format zformat )
 {
    switch (zformat) {
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       return DEPTH_FRMT_24_FIXED_8_OTHER;
    case PIPE_FORMAT_Z16_UNORM:
       return DEPTH_FRMT_16_FIXED;
diff --git a/src/gallium/drivers/i915/i915_state_sampler.c b/src/gallium/drivers/i915/i915_state_sampler.c
index 4c326561cb..270d4f21e1 100644
--- a/src/gallium/drivers/i915/i915_state_sampler.c
+++ b/src/gallium/drivers/i915/i915_state_sampler.c
@@ -220,7 +220,7 @@ translate_texture_format(enum pipe_format pipeFormat)
    case PIPE_FORMAT_RGBA_DXT5:
       return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
 #endif
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       return (MAPSURF_32BIT | MT_32BIT_xI824);
    default:
       debug_printf("i915: translate_texture_format() bad image format %x\n",
diff --git a/src/gallium/drivers/i965/brw_misc_state.c b/src/gallium/drivers/i965/brw_misc_state.c
index e3f25bdf62..b5029ceb69 100644
--- a/src/gallium/drivers/i965/brw_misc_state.c
+++ b/src/gallium/drivers/i965/brw_misc_state.c
@@ -266,7 +266,7 @@ static int emit_depthbuffer(struct brw_context *brw)
 	 cpp = 2;
 	 break;
       case PIPE_FORMAT_Z24X8_UNORM:
-      case PIPE_FORMAT_Z24S8_UNORM:
+      case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 	 format = BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
 	 cpp = 4;
 	 break;
diff --git a/src/gallium/drivers/i965/brw_pipe_clear.c b/src/gallium/drivers/i965/brw_pipe_clear.c
index d7048f769b..abf507991e 100644
--- a/src/gallium/drivers/i965/brw_pipe_clear.c
+++ b/src/gallium/drivers/i965/brw_pipe_clear.c
@@ -140,7 +140,7 @@ static void zstencil_clear(struct brw_context *brw,
 
    switch (bsurface->base.format) {
    case PIPE_FORMAT_Z24X8_UNORM:
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       value = ((unsigned)(depth * MASK24) & MASK24);
       break;
    case PIPE_FORMAT_Z16_UNORM:
@@ -153,7 +153,7 @@ static void zstencil_clear(struct brw_context *brw,
 
    switch (bsurface->base.format) {
    case PIPE_FORMAT_Z24X8_UNORM:
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       value = value | (stencil << 24);
       break;
 
diff --git a/src/gallium/drivers/i965/brw_screen.c b/src/gallium/drivers/i965/brw_screen.c
index cef83ffea8..ba30e63f20 100644
--- a/src/gallium/drivers/i965/brw_screen.c
+++ b/src/gallium/drivers/i965/brw_screen.c
@@ -252,7 +252,7 @@ brw_is_format_supported(struct pipe_screen *screen,
       /* depth */
       PIPE_FORMAT_Z32_FLOAT,
       PIPE_FORMAT_Z24X8_UNORM,
-      PIPE_FORMAT_Z24S8_UNORM,
+      PIPE_FORMAT_Z24_UNORM_S8_USCALED,
       PIPE_FORMAT_Z16_UNORM,
       /* signed */
       PIPE_FORMAT_R8G8_SNORM,
@@ -268,7 +268,7 @@ brw_is_format_supported(struct pipe_screen *screen,
    static const enum pipe_format depth_supported[] = {
       PIPE_FORMAT_Z32_FLOAT,
       PIPE_FORMAT_Z24X8_UNORM,
-      PIPE_FORMAT_Z24S8_UNORM,
+      PIPE_FORMAT_Z24_UNORM_S8_USCALED,
       PIPE_FORMAT_Z16_UNORM,
       PIPE_FORMAT_NONE  /* list terminator */
    };
diff --git a/src/gallium/drivers/i965/brw_screen_texture.c b/src/gallium/drivers/i965/brw_screen_texture.c
index f9f17bdaba..17bf3152dc 100644
--- a/src/gallium/drivers/i965/brw_screen_texture.c
+++ b/src/gallium/drivers/i965/brw_screen_texture.c
@@ -158,7 +158,7 @@ static GLuint translate_tex_format( enum pipe_format pf )
    case PIPE_FORMAT_Z16_UNORM:
          return BRW_SURFACEFORMAT_I16_UNORM;
 
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
          return BRW_SURFACEFORMAT_I24X8_UNORM;
 
@@ -315,6 +315,7 @@ brw_texture_from_handle(struct pipe_screen *screen,
    struct brw_winsys_buffer *buffer;
    unsigned tiling;
    unsigned pitch;
+   GLuint format;
 
    if (templ->target != PIPE_TEXTURE_2D ||
        templ->last_level != 0 ||
@@ -362,8 +363,10 @@ brw_texture_from_handle(struct pipe_screen *screen,
 
    tex->ss.ss0.mipmap_layout_mode = BRW_SURFACE_MIPMAPLAYOUT_BELOW;
    tex->ss.ss0.surface_type = translate_tex_target(tex->base.target);
-   tex->ss.ss0.surface_format = translate_tex_format(tex->base.format);
-   assert(tex->ss.ss0.surface_format != BRW_SURFACEFORMAT_INVALID);
+
+   format = translate_tex_format(tex->base.format);
+   assert(format != BRW_SURFACEFORMAT_INVALID);
+   tex->ss.ss0.surface_format = format;
 
    /* This is ok for all textures with channel width 8bit or less:
     */
diff --git a/src/gallium/drivers/identity/id_drm.c b/src/gallium/drivers/identity/id_drm.c
index 26920fed08..d332c36af2 100644
--- a/src/gallium/drivers/identity/id_drm.c
+++ b/src/gallium/drivers/identity/id_drm.c
@@ -31,7 +31,6 @@
 #include "id_drm.h"
 #include "id_screen.h"
 #include "id_public.h"
-#include "id_objects.h"
 
 struct identity_drm_api
 {
diff --git a/src/gallium/drivers/llvmpipe/Makefile b/src/gallium/drivers/llvmpipe/Makefile
index b37c6754c0..7bfce15c2e 100644
--- a/src/gallium/drivers/llvmpipe/Makefile
+++ b/src/gallium/drivers/llvmpipe/Makefile
@@ -42,7 +42,7 @@ C_SOURCES = \
 
 CPP_SOURCES = \
 
-PROGS := lp_test_format	\
+PROGS := \
 	 lp_test_blend	\
 	 lp_test_conv	\
 	 lp_test_printf
@@ -50,7 +50,7 @@ PROGS := lp_test_format	\
 
 include ../../Makefile.template
 
-lp_tile_soa.c: lp_tile_soa.py ../../auxiliary/util/u_format_parse.py ../../auxiliary/util/u_format_access.py ../../auxiliary/util/u_format.csv
+lp_tile_soa.c: lp_tile_soa.py ../../auxiliary/util/u_format_parse.py ../../auxiliary/util/u_format_pack.py ../../auxiliary/util/u_format.csv
 	python lp_tile_soa.py ../../auxiliary/util/u_format.csv > $@
 
 
diff --git a/src/gallium/drivers/llvmpipe/SConscript b/src/gallium/drivers/llvmpipe/SConscript
index 13c1a13e87..07e6ccfce4 100644
--- a/src/gallium/drivers/llvmpipe/SConscript
+++ b/src/gallium/drivers/llvmpipe/SConscript
@@ -22,7 +22,6 @@ env.CodeGenerate(
 env.Depends('lp_tile_soa.c', [
     '#src/gallium/auxiliary/util/u_format_parse.py', 
     '#src/gallium/auxiliary/util/u_format_pack.py', 
-    '#src/gallium/auxiliary/util/u_format_access.py',
 ])
 
 llvmpipe = env.ConvenienceLibrary(
@@ -70,7 +69,6 @@ if env['platform'] != 'embedded':
     env.Prepend(LIBS = [llvmpipe] + gallium)
 
     tests = [
-        'format',
         'blend',
         'conv',
     ]
diff --git a/src/gallium/drivers/llvmpipe/lp_rast.c b/src/gallium/drivers/llvmpipe/lp_rast.c
index 8352f17559..e629a3e5f1 100644
--- a/src/gallium/drivers/llvmpipe/lp_rast.c
+++ b/src/gallium/drivers/llvmpipe/lp_rast.c
@@ -769,7 +769,11 @@ create_rast_threads(struct lp_rasterizer *rast)
     * properly implemented. */
    rast->num_threads = 0;
 #else
+#ifdef PIPE_OS_EMBEDDED
+   rast->num_threads = 0;
+#else
    rast->num_threads = util_cpu_caps.nr_cpus;
+#endif
    rast->num_threads = debug_get_num_option("LP_NUM_THREADS", rast->num_threads);
    rast->num_threads = MIN2(rast->num_threads, MAX_THREADS);
 #endif
diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index f1bbc2092c..3d6c7930a0 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -222,11 +222,7 @@ llvmpipe_is_format_supported( struct pipe_screen *_screen,
 
    /* FIXME: Temporary restrictions. See lp_bld_sample_soa.c */
    if(tex_usage & PIPE_TEXTURE_USAGE_SAMPLER) {
-      if(format_desc->block.width != 1 ||
-         format_desc->block.height != 1)
-         return FALSE;
-
-      if(format_desc->layout != UTIL_FORMAT_LAYOUT_PLAIN)
+      if(!format_desc->is_bitmask)
          return FALSE;
 
       if(format_desc->colorspace != UTIL_FORMAT_COLORSPACE_RGB &&
diff --git a/src/gallium/drivers/llvmpipe/lp_test_format.c b/src/gallium/drivers/llvmpipe/lp_test_format.c
deleted file mode 100644
index fb595893bd..0000000000
--- a/src/gallium/drivers/llvmpipe/lp_test_format.c
+++ /dev/null
@@ -1,314 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "gallivm/lp_bld.h"
-#include <llvm-c/Analysis.h>
-#include <llvm-c/ExecutionEngine.h>
-#include <llvm-c/Target.h>
-#include <llvm-c/Transforms/Scalar.h>
-
-#include "util/u_cpu_detect.h"
-#include "util/u_format.h"
-
-#include "gallivm/lp_bld_format.h"
-#include "lp_test.h"
-
-
-struct pixel_test_case
-{
-   enum pipe_format format;
-   uint32_t packed;
-   double unpacked[4];
-};
-
-
-struct pixel_test_case test_cases[] =
-{
-   {PIPE_FORMAT_B5G6R5_UNORM,   0x0000, {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B5G6R5_UNORM,   0x001f, {0.0, 0.0, 1.0, 1.0}},
-   {PIPE_FORMAT_B5G6R5_UNORM,   0x07e0, {0.0, 1.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B5G6R5_UNORM,   0xf800, {1.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B5G6R5_UNORM,   0xffff, {1.0, 1.0, 1.0, 1.0}},
-
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0x0000, {0.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0x001f, {0.0, 0.0, 1.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0x03e0, {0.0, 1.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0x7c00, {1.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0x8000, {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B5G5R5A1_UNORM, 0xffff, {1.0, 1.0, 1.0, 1.0}},
-
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0x00000000, {0.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0x000000ff, {0.0, 0.0, 1.0, 0.0}},
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0x0000ff00, {0.0, 1.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0x00ff0000, {1.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0xff000000, {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_B8G8R8A8_UNORM, 0xffffffff, {1.0, 1.0, 1.0, 1.0}},
-
-#if 0
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0x00000000, {0.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0x000000ff, {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0x0000ff00, {0.0, 0.0, 1.0, 0.0}},
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0x00ff0000, {0.0, 1.0, 0.0, 0.0}},
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0xff000000, {1.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_R8G8B8A8_UNORM, 0xffffffff, {1.0, 1.0, 1.0, 1.0}},
-#endif
-
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0x00000000, {0.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0x000000ff, {0.0, 0.0, 0.0, 1.0}},
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0x0000ff00, {1.0, 0.0, 0.0, 0.0}},
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0x00ff0000, {0.0, 1.0, 0.0, 0.0}},
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0xff000000, {0.0, 0.0, 1.0, 0.0}},
-   {PIPE_FORMAT_A8R8G8B8_UNORM, 0xffffffff, {1.0, 1.0, 1.0, 1.0}},
-};
-
-
-void
-write_tsv_header(FILE *fp)
-{
-   fprintf(fp,
-           "result\t"
-           "format\n");
-
-   fflush(fp);
-}
-
-
-static void
-write_tsv_row(FILE *fp,
-              const struct util_format_description *desc,
-              boolean success)
-{
-   fprintf(fp, "%s\t", success ? "pass" : "fail");
-
-   fprintf(fp, "%s\n", desc->name);
-
-   fflush(fp);
-}
-
-
-typedef void (*load_ptr_t)(const uint32_t packed, float *);
-
-
-static LLVMValueRef
-add_load_rgba_test(LLVMModuleRef module,
-                   const struct util_format_description *desc)
-{
-   LLVMTypeRef args[2];
-   LLVMValueRef func;
-   LLVMValueRef packed;
-   LLVMValueRef rgba_ptr;
-   LLVMBasicBlockRef block;
-   LLVMBuilderRef builder;
-   LLVMValueRef rgba;
-
-   args[0] = LLVMInt32Type();
-   args[1] = LLVMPointerType(LLVMVectorType(LLVMFloatType(), 4), 0);
-
-   func = LLVMAddFunction(module, "load", LLVMFunctionType(LLVMVoidType(), args, 2, 0));
-   LLVMSetFunctionCallConv(func, LLVMCCallConv);
-   packed = LLVMGetParam(func, 0);
-   rgba_ptr = LLVMGetParam(func, 1);
-
-   block = LLVMAppendBasicBlock(func, "entry");
-   builder = LLVMCreateBuilder();
-   LLVMPositionBuilderAtEnd(builder, block);
-
-   if(desc->block.bits < 32)
-      packed = LLVMBuildTrunc(builder, packed, LLVMIntType(desc->block.bits), "");
-
-   rgba = lp_build_unpack_rgba_aos(builder, desc, packed);
-
-   LLVMBuildStore(builder, rgba, rgba_ptr);
-
-   LLVMBuildRetVoid(builder);
-
-   LLVMDisposeBuilder(builder);
-   return func;
-}
-
-
-typedef void (*store_ptr_t)(uint32_t *, const float *);
-
-
-static LLVMValueRef
-add_store_rgba_test(LLVMModuleRef module,
-                    const struct util_format_description *desc)
-{
-   LLVMTypeRef args[2];
-   LLVMValueRef func;
-   LLVMValueRef packed_ptr;
-   LLVMValueRef rgba_ptr;
-   LLVMBasicBlockRef block;
-   LLVMBuilderRef builder;
-   LLVMValueRef rgba;
-   LLVMValueRef packed;
-
-   args[0] = LLVMPointerType(LLVMInt32Type(), 0);
-   args[1] = LLVMPointerType(LLVMVectorType(LLVMFloatType(), 4), 0);
-
-   func = LLVMAddFunction(module, "store", LLVMFunctionType(LLVMVoidType(), args, 2, 0));
-   LLVMSetFunctionCallConv(func, LLVMCCallConv);
-   packed_ptr = LLVMGetParam(func, 0);
-   rgba_ptr = LLVMGetParam(func, 1);
-
-   block = LLVMAppendBasicBlock(func, "entry");
-   builder = LLVMCreateBuilder();
-   LLVMPositionBuilderAtEnd(builder, block);
-
-   rgba = LLVMBuildLoad(builder, rgba_ptr, "");
-
-   packed = lp_build_pack_rgba_aos(builder, desc, rgba);
-
-   if(desc->block.bits < 32)
-      packed = LLVMBuildZExt(builder, packed, LLVMInt32Type(), "");
-
-   LLVMBuildStore(builder, packed, packed_ptr);
-
-   LLVMBuildRetVoid(builder);
-
-   LLVMDisposeBuilder(builder);
-   return func;
-}
-
-
-PIPE_ALIGN_STACK
-static boolean
-test_format(unsigned verbose, FILE *fp, const struct pixel_test_case *test)
-{
-   LLVMModuleRef module = NULL;
-   LLVMValueRef load = NULL;
-   LLVMValueRef store = NULL;
-   LLVMExecutionEngineRef engine = NULL;
-   LLVMModuleProviderRef provider = NULL;
-   LLVMPassManagerRef pass = NULL;
-   char *error = NULL;
-   const struct util_format_description *desc;
-   load_ptr_t load_ptr;
-   store_ptr_t store_ptr;
-   float unpacked[4];
-   unsigned packed;
-   boolean success;
-   unsigned i;
-
-   desc = util_format_description(test->format);
-   fprintf(stderr, "%s\n", desc->name);
-
-   module = LLVMModuleCreateWithName("test");
-
-   load = add_load_rgba_test(module, desc);
-   store = add_store_rgba_test(module, desc);
-
-   if(LLVMVerifyModule(module, LLVMPrintMessageAction, &error)) {
-      LLVMDumpModule(module);
-      abort();
-   }
-   LLVMDisposeMessage(error);
-
-   provider = LLVMCreateModuleProviderForExistingModule(module);
-   if (LLVMCreateJITCompiler(&engine, provider, 1, &error)) {
-      fprintf(stderr, "%s\n", error);
-      LLVMDisposeMessage(error);
-      abort();
-   }
-
-#if 0
-   pass = LLVMCreatePassManager();
-   LLVMAddTargetData(LLVMGetExecutionEngineTargetData(engine), pass);
-   /* These are the passes currently listed in llvm-c/Transforms/Scalar.h,
-    * but there are more on SVN. */
-   LLVMAddConstantPropagationPass(pass);
-   LLVMAddInstructionCombiningPass(pass);
-   LLVMAddPromoteMemoryToRegisterPass(pass);
-   LLVMAddGVNPass(pass);
-   LLVMAddCFGSimplificationPass(pass);
-   LLVMRunPassManager(pass, module);
-#else
-   (void)pass;
-#endif
-
-   load_ptr  = (load_ptr_t) LLVMGetPointerToGlobal(engine, load);
-   store_ptr = (store_ptr_t)LLVMGetPointerToGlobal(engine, store);
-
-   memset(unpacked, 0, sizeof unpacked);
-   packed = 0;
-
-   load_ptr(test->packed, unpacked);
-   store_ptr(&packed, unpacked);
-
-   success = TRUE;
-   if(test->packed != packed)
-      success = FALSE;
-   for(i = 0; i < 4; ++i)
-      if(test->unpacked[i] != unpacked[i])
-         success = FALSE;
-
-   if (!success) {
-      printf("FAILED\n");
-      printf("  Packed: %08x\n", test->packed);
-      printf("          %08x\n", packed);
-      printf("  Unpacked: %f %f %f %f\n", unpacked[0], unpacked[1], unpacked[2], unpacked[3]);
-      printf("            %f %f %f %f\n", test->unpacked[0], test->unpacked[1], test->unpacked[2], test->unpacked[3]);
-      LLVMDumpModule(module);
-   }
-
-   LLVMFreeMachineCodeForFunction(engine, store);
-   LLVMFreeMachineCodeForFunction(engine, load);
-
-   LLVMDisposeExecutionEngine(engine);
-   if(pass)
-      LLVMDisposePassManager(pass);
-
-   if(fp)
-      write_tsv_row(fp, desc, success);
-
-   return success;
-}
-
-
-boolean
-test_all(unsigned verbose, FILE *fp)
-{
-   unsigned i;
-   bool success = TRUE;
-
-   for (i = 0; i < sizeof(test_cases)/sizeof(test_cases[0]); ++i)
-      if(!test_format(verbose, fp, &test_cases[i]))
-        success = FALSE;
-
-   return success;
-}
-
-
-boolean
-test_some(unsigned verbose, FILE *fp, unsigned long n)
-{
-   return test_all(verbose, fp);
-}
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_soa.py b/src/gallium/drivers/llvmpipe/lp_tile_soa.py
index 00b8d4fc38..6a3ede02a4 100644
--- a/src/gallium/drivers/llvmpipe/lp_tile_soa.py
+++ b/src/gallium/drivers/llvmpipe/lp_tile_soa.py
@@ -42,7 +42,7 @@ import os.path
 
 sys.path.insert(0, os.path.join(os.path.dirname(sys.argv[0]), '../../auxiliary/util'))
 
-from u_format_access import *
+from u_format_pack import *
 
 
 def generate_format_read(format, dst_channel, dst_native_type, dst_suffix):
@@ -62,7 +62,7 @@ def generate_format_read(format, dst_channel, dst_native_type, dst_suffix):
     print '      for (x = 0; x < w; ++x) {'
 
     names = ['']*4
-    if format.colorspace == 'rgb':
+    if format.colorspace in ('rgb', 'srgb'):
         for i in range(4):
             swizzle = format.swizzles[i]
             if swizzle < 4:
@@ -104,7 +104,7 @@ def generate_format_read(format, dst_channel, dst_native_type, dst_suffix):
         assert False
 
     for i in range(4):
-        if format.colorspace == 'rgb':
+        if format.colorspace in ('rgb', 'srgb'):
             swizzle = format.swizzles[i]
             if swizzle < 4:
                 value = names[swizzle]
@@ -134,7 +134,7 @@ def pack_rgba(format, src_channel, r, g, b, a):
     """Return an expression for packing r, g, b, a into a pixel of the
     given format.  Ex: '(b << 24) | (g << 16) | (r << 8) | (a << 0)'
     """
-    assert format.colorspace == 'rgb'
+    assert format.colorspace in ('rgb', 'srgb')
     inv_swizzle = format.inv_swizzles()
     shift = 0
     expr = None
@@ -277,7 +277,7 @@ def generate_read(formats, dst_channel, dst_native_type, dst_suffix):
             print '      func = &lp_tile_%s_read_%s;' % (format.short_name(), dst_suffix)
             print '      break;'
     print '   default:'
-    print '      debug_printf("unsupported format\\n");'
+    print '      debug_printf("%s: unsupported format %s\\n", __FUNCTION__, util_format_name(format));'
     print '      return;'
     print '   }'
     print '   func(dst, (const uint8_t *)src, src_stride, x, y, w, h);'
@@ -304,7 +304,7 @@ def generate_write(formats, src_channel, src_native_type, src_suffix):
             print '      func = &lp_tile_%s_write_%s;' % (format.short_name(), src_suffix)
             print '      break;'
     print '   default:'
-    print '      debug_printf("unsupported format\\n");'
+    print '      debug_printf("%s: unsupported format %s\\n", __FUNCTION__, util_format_name(format));'
     print '      return;'
     print '   }'
     print '   func(src, (uint8_t *)dst, dst_stride, x, y, w, h);'
@@ -349,8 +349,6 @@ def main():
     print '};'
     print
 
-    generate_clamp()
-
     channel = Channel(UNSIGNED, True, 8)
     native_type = 'uint8_t'
     suffix = '4ub'
diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c
index e091cae602..85b0aef20c 100644
--- a/src/gallium/drivers/nv50/nv50_miptree.c
+++ b/src/gallium/drivers/nv50/nv50_miptree.c
@@ -89,14 +89,14 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *tmp)
 	case PIPE_FORMAT_Z32_FLOAT:
 		tile_flags = 0x4800;
 		break;
-	case PIPE_FORMAT_S8Z24_UNORM:
+	case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 		tile_flags = 0x1800;
 		break;
 	case PIPE_FORMAT_Z16_UNORM:
 		tile_flags = 0x6c00;
 		break;
 	case PIPE_FORMAT_Z24X8_UNORM:
-	case PIPE_FORMAT_Z24S8_UNORM:
+	case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 		tile_flags = 0x2800;
 		break;
 	case PIPE_FORMAT_R32G32B32A32_FLOAT:
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index 1a4606d9e2..801548d8a1 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -51,9 +51,9 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen,
 	if (tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL) {
 		switch (format) {
 		case PIPE_FORMAT_Z32_FLOAT:
-		case PIPE_FORMAT_S8Z24_UNORM:
+		case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 		case PIPE_FORMAT_Z24X8_UNORM:
-		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 			return TRUE;
 		default:
 			break;
@@ -75,8 +75,8 @@ nv50_screen_is_format_supported(struct pipe_screen *pscreen,
 		case PIPE_FORMAT_DXT1_RGBA:
 		case PIPE_FORMAT_DXT3_RGBA:
 		case PIPE_FORMAT_DXT5_RGBA:
-		case PIPE_FORMAT_S8Z24_UNORM:
-		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
+		case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 		case PIPE_FORMAT_Z32_FLOAT:
 		case PIPE_FORMAT_R16G16B16A16_SNORM:
 		case PIPE_FORMAT_R16G16B16A16_UNORM:
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index b7e355283c..402045416c 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -122,13 +122,13 @@ validate_fb(struct nv50_context *nv50)
 		so_reloc (so, bo, fb->zsbuf->offset, NOUVEAU_BO_VRAM |
 			      NOUVEAU_BO_LOW | NOUVEAU_BO_RDWR, 0, 0);
 		switch (fb->zsbuf->format) {
-		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 			so_data(so, NV50TCL_ZETA_FORMAT_S8Z24_UNORM);
 			break;
 		case PIPE_FORMAT_Z24X8_UNORM:
 			so_data(so, NV50TCL_ZETA_FORMAT_X8Z24_UNORM);
 			break;
-		case PIPE_FORMAT_S8Z24_UNORM:
+		case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 			so_data(so, NV50TCL_ZETA_FORMAT_Z24S8_UNORM);
 			break;
 		case PIPE_FORMAT_Z32_FLOAT:
diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c
index 6467c48a32..7a9dada531 100644
--- a/src/gallium/drivers/nv50/nv50_surface.c
+++ b/src/gallium/drivers/nv50/nv50_surface.c
@@ -60,8 +60,8 @@ nv50_format(enum pipe_format format)
 	/* only because we require src format == dst format: */
 	case PIPE_FORMAT_R16G16_SNORM:
 	case PIPE_FORMAT_R16G16_UNORM:
-	case PIPE_FORMAT_S8Z24_UNORM:
-	case PIPE_FORMAT_Z24S8_UNORM:
+	case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
+	case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
 		return NV50_2D_DST_FORMAT_A8R8G8B8_UNORM;
 	case PIPE_FORMAT_L8A8_UNORM:
 	case PIPE_FORMAT_B4G4R4A4_UNORM:
diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c
index 85ab947c00..d094b490ed 100644
--- a/src/gallium/drivers/nv50/nv50_tex.c
+++ b/src/gallium/drivers/nv50/nv50_tex.c
@@ -62,8 +62,8 @@ static const uint32_t nv50_texture_formats[PIPE_FORMAT_COUNT] =
 	_(DXT3_RGBA, UNORM, C0, C1, C2, C3, DXT3),
 	_(DXT5_RGBA, UNORM, C0, C1, C2, C3, DXT5),
 
-	_MIXED(S8Z24_UNORM, UINT, UNORM, UINT, UINT, C1, C1, C1, ONE, 24_8),
-	_MIXED(Z24S8_UNORM, UNORM, UINT, UINT, UINT, C0, C0, C0, ONE, 8_24),
+	_MIXED(S8_USCALED_Z24_UNORM, UINT, UNORM, UINT, UINT, C1, C1, C1, ONE, 24_8),
+	_MIXED(Z24_UNORM_S8_USCALED, UNORM, UINT, UINT, UINT, C0, C0, C0, ONE, 8_24),
 
 	_(R16G16B16A16_SNORM, UNORM, C0, C1, C2, C3, 16_16_16_16),
 	_(R16G16B16A16_UNORM, SNORM, C0, C1, C2, C3, 16_16_16_16),
diff --git a/src/gallium/drivers/nvfx/nv04_surface_2d.c b/src/gallium/drivers/nvfx/nv04_surface_2d.c
index ed18c9f24d..6784170c00 100644
--- a/src/gallium/drivers/nvfx/nv04_surface_2d.c
+++ b/src/gallium/drivers/nvfx/nv04_surface_2d.c
@@ -25,7 +25,7 @@ nv04_surface_format(enum pipe_format format)
 	case PIPE_FORMAT_B8G8R8X8_UNORM:
 	case PIPE_FORMAT_B8G8R8A8_UNORM:
 		return NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8;
-	case PIPE_FORMAT_S8Z24_UNORM:
+	case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 	case PIPE_FORMAT_X8Z24_UNORM:
 		return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32;
 	default:
@@ -45,7 +45,7 @@ nv04_rect_format(enum pipe_format format)
 		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5;
 	case PIPE_FORMAT_B8G8R8X8_UNORM:
 	case PIPE_FORMAT_B8G8R8A8_UNORM:
-	case PIPE_FORMAT_S8Z24_UNORM:
+	case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 	case PIPE_FORMAT_X8Z24_UNORM:
 		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8;
 	default:
diff --git a/src/gallium/drivers/nvfx/nv30_fragtex.c b/src/gallium/drivers/nvfx/nv30_fragtex.c
index 54e47574c4..d3f3edb327 100644
--- a/src/gallium/drivers/nvfx/nv30_fragtex.c
+++ b/src/gallium/drivers/nvfx/nv30_fragtex.c
@@ -63,7 +63,7 @@ nv30_texture_formats[] = {
 	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X),
 	_(L8A8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y),
 	_(Z16_UNORM     , R5G6B5  ,   S1,   S1,   S1,  ONE, X, X, X, X),
-	_(S8Z24_UNORM   , A8R8G8B8,   S1,   S1,   S1,  ONE, X, X, X, X),
+	_(S8_USCALED_Z24_UNORM   , A8R8G8B8,   S1,   S1,   S1,  ONE, X, X, X, X),
 	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
 	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
 	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
diff --git a/src/gallium/drivers/nvfx/nv40_fragtex.c b/src/gallium/drivers/nvfx/nv40_fragtex.c
index 05506e2809..fe87cebbb6 100644
--- a/src/gallium/drivers/nvfx/nv40_fragtex.c
+++ b/src/gallium/drivers/nvfx/nv40_fragtex.c
@@ -81,7 +81,7 @@ nv40_texture_formats[] = {
 	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X, 0, 0, 0, 0),
 	_(L8A8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y, 0, 0, 0, 0),
 	_(Z16_UNORM     , Z16     ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
-	_(S8Z24_UNORM   , Z24     ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
+	_(S8_USCALED_Z24_UNORM   , Z24     ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
 	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W, 0, 0, 0, 0),
 	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
 	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
diff --git a/src/gallium/drivers/nvfx/nvfx_screen.c b/src/gallium/drivers/nvfx/nvfx_screen.c
index f7f3921894..1a103520a3 100644
--- a/src/gallium/drivers/nvfx/nvfx_screen.c
+++ b/src/gallium/drivers/nvfx/nvfx_screen.c
@@ -133,7 +133,7 @@ nvfx_screen_surface_format_supported(struct pipe_screen *pscreen,
 	} else
 	if (tex_usage & PIPE_TEXTURE_USAGE_DEPTH_STENCIL) {
 		switch (format) {
-		case PIPE_FORMAT_S8Z24_UNORM:
+		case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 		case PIPE_FORMAT_X8Z24_UNORM:
 			return TRUE;
 		case PIPE_FORMAT_Z16_UNORM:
@@ -155,7 +155,7 @@ nvfx_screen_surface_format_supported(struct pipe_screen *pscreen,
 		case PIPE_FORMAT_I8_UNORM:
 		case PIPE_FORMAT_L8A8_UNORM:
 		case PIPE_FORMAT_Z16_UNORM:
-		case PIPE_FORMAT_S8Z24_UNORM:
+		case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 		case PIPE_FORMAT_DXT1_RGB:
 		case PIPE_FORMAT_DXT1_RGBA:
 		case PIPE_FORMAT_DXT3_RGBA:
diff --git a/src/gallium/drivers/nvfx/nvfx_state_fb.c b/src/gallium/drivers/nvfx/nvfx_state_fb.c
index dd64ba4193..1923184163 100644
--- a/src/gallium/drivers/nvfx/nvfx_state_fb.c
+++ b/src/gallium/drivers/nvfx/nvfx_state_fb.c
@@ -102,7 +102,7 @@ nvfx_state_framebuffer_validate(struct nvfx_context *nvfx)
 		rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16;
 		zeta_bits = 16;
 		break;
-	case PIPE_FORMAT_S8Z24_UNORM:
+	case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
 	case PIPE_FORMAT_X8Z24_UNORM:
 	case 0:
 		rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8;
diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h
index 1c2b252887..0d6b7654f3 100644
--- a/src/gallium/drivers/r300/r300_reg.h
+++ b/src/gallium/drivers/r300/r300_reg.h
@@ -1556,6 +1556,26 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_FORMAT_32F_32F 	    	    0x1C
 #	define R300_TX_FORMAT_32F_32F_32F_32F  	    0x1D
 #       define R300_TX_FORMAT_W24_FP                0x1E
+#       define R400_TX_FORMAT_ATI2N                 0x1F
+
+/* These need TX_FORMAT2_[0-15].TXFORMAT_MSB set.
+
+   My guess is the 10-bit formats are the 8-bit ones but with filtering being
+   performed with the precision of 10 bits per channel. This makes sense
+   with sRGB textures since the conversion to linear space reduces the precision
+   significantly so the shader gets approximately the 8-bit precision
+   in the end. It might also improve the quality of HDR rendering where
+   high-precision filtering is desirable.
+
+   Again, this is guessed, the formats might mean something entirely else.
+   The others should be fine. */
+#       define R500_TX_FORMAT_X1                    0x0
+#       define R500_TX_FORMAT_X1_REV                0x1
+#       define R500_TX_FORMAT_X10                   0x2
+#       define R500_TX_FORMAT_Y10X10                0x3
+#       define R500_TX_FORMAT_W10Z10Y10X10          0x4
+#       define R500_TX_FORMAT_ATI1N                 0x5
+
 
 #       define R300_TX_FORMAT_SIGNED_W             (1 << 5)
 #       define R300_TX_FORMAT_SIGNED_Z             (1 << 6)
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index e46f836dd2..50e5e9307e 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -207,9 +207,14 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
 {
     uint32_t retval = 0;
     boolean is_r500 = r300_screen(screen)->caps->is_r500;
+    boolean is_r400 = r300_screen(screen)->caps->is_r400;
     boolean is_z24 = format == PIPE_FORMAT_X8Z24_UNORM ||
-                     format == PIPE_FORMAT_S8Z24_UNORM;
+                     format == PIPE_FORMAT_S8_USCALED_Z24_UNORM;
     boolean is_color2101010 = format == PIPE_FORMAT_R10G10B10A2_UNORM;
+    boolean is_ati1n = format == PIPE_FORMAT_RGTC1_UNORM ||
+                       format == PIPE_FORMAT_RGTC1_SNORM;
+    boolean is_ati2n = format == PIPE_FORMAT_RGTC2_UNORM ||
+                       format == PIPE_FORMAT_RGTC2_SNORM;
 
     if (target >= PIPE_MAX_TEXTURE_TYPES) {
         fprintf(stderr, "r300: Implementation error: Received bogus texture "
@@ -221,6 +226,10 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
     if ((usage & PIPE_TEXTURE_USAGE_SAMPLER) &&
         /* Z24 cannot be sampled from on non-r5xx. */
         (is_r500 || !is_z24) &&
+        /* ATI1N is r5xx-only. */
+        (is_r500 || !is_ati1n) &&
+        /* ATI2N is supported on r4xx-r5xx. */
+        (is_r400 || is_r500 || !is_ati2n) &&
         r300_is_sampler_format_supported(format)) {
         retval |= PIPE_TEXTURE_USAGE_SAMPLER;
     }
diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index 02d09c008c..8a690039b6 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -367,7 +367,6 @@ static INLINE uint16_t
 r300_translate_vertex_data_type(enum pipe_format format) {
     uint32_t result = 0;
     const struct util_format_description *desc;
-    unsigned components = util_format_get_nr_components(format);
 
     desc = util_format_description(format);
 
@@ -380,17 +379,17 @@ r300_translate_vertex_data_type(enum pipe_format format) {
     switch (desc->channel[0].type) {
         /* Half-floats, floats, doubles */
         case UTIL_FORMAT_TYPE_FLOAT:
-            switch (util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 0)) {
+            switch (desc->channel[0].size) {
                 case 16:
                     /* XXX Supported only on RV350 and later. */
-                    if (components > 2) {
+                    if (desc->nr_channels > 2) {
                         result = R300_DATA_TYPE_FLT16_4;
                     } else {
                         result = R300_DATA_TYPE_FLT16_2;
                     }
                     break;
                 case 32:
-                    result = R300_DATA_TYPE_FLOAT_1 + (components - 1);
+                    result = R300_DATA_TYPE_FLOAT_1 + (desc->nr_channels - 1);
                     break;
                 default:
                     fprintf(stderr, "r300: Bad format %s in %s:%d\n",
@@ -402,12 +401,12 @@ r300_translate_vertex_data_type(enum pipe_format format) {
         case UTIL_FORMAT_TYPE_UNSIGNED:
         /* Signed ints */
         case UTIL_FORMAT_TYPE_SIGNED:
-            switch (util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 0)) {
+            switch (desc->channel[0].size) {
                 case 8:
                     result = R300_DATA_TYPE_BYTE;
                     break;
                 case 16:
-                    if (components > 2) {
+                    if (desc->nr_channels > 2) {
                         result = R300_DATA_TYPE_SHORT_4;
                     } else {
                         result = R300_DATA_TYPE_SHORT_2;
@@ -416,8 +415,8 @@ r300_translate_vertex_data_type(enum pipe_format format) {
                 default:
                     fprintf(stderr, "r300: Bad format %s in %s:%d\n",
                         util_format_name(format), __FUNCTION__, __LINE__);
-                    fprintf(stderr, "r300: util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 0) == %d\n",
-                        util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 0));
+                    fprintf(stderr, "r300: desc->channel[0].size == %d\n",
+                        desc->channel[0].size);
                     assert(0);
             }
             break;
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index 22ccadfe3d..e0dee44b54 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -33,6 +33,9 @@
 #include "r300_state_inlines.h"
 #include "r300_winsys.h"
 
+/* XXX Enable float textures here. */
+/*#define ENABLE_FLOAT_TEXTURES*/
+
 #define TILE_WIDTH 0
 #define TILE_HEIGHT 1
 
@@ -73,7 +76,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
 {
     uint32_t result = 0;
     const struct util_format_description *desc;
-    unsigned components = 0, i;
+    unsigned i;
     boolean uniform = TRUE;
     const uint32_t swizzle_shift[4] = {
         R300_TX_FORMAT_R_SHIFT,
@@ -104,7 +107,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
                 case PIPE_FORMAT_Z16_UNORM:
                     return R300_EASY_TX_FORMAT(X, X, X, X, X16);
                 case PIPE_FORMAT_X8Z24_UNORM:
-                case PIPE_FORMAT_S8Z24_UNORM:
+                case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
                     return R300_EASY_TX_FORMAT(X, X, X, X, W24_FP);
                 default:
                     return ~0; /* Unsupported. */
@@ -158,7 +161,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
         }
     }
 
-    /* Compressed formats. */
+    /* S3TC formats. */
     if (desc->layout == UTIL_FORMAT_LAYOUT_S3TC) {
         switch (format) {
             case PIPE_FORMAT_DXT1_RGB:
@@ -177,28 +180,35 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
         }
     }
 
-    /* Get the number of components. */
-    for (i = 0; i < 4; i++) {
-        if (desc->channel[i].type != UTIL_FORMAT_TYPE_VOID) {
-            ++components;
-        }
-    }
-
     /* Add sign. */
-    for (i = 0; i < components; i++) {
+    for (i = 0; i < desc->nr_channels; i++) {
         if (desc->channel[i].type == UTIL_FORMAT_TYPE_SIGNED) {
             result |= sign_bit[i];
         }
     }
 
+    /* RGTC formats. */
+    if (desc->layout == UTIL_FORMAT_LAYOUT_RGTC) {
+        switch (format) {
+            case PIPE_FORMAT_RGTC1_UNORM:
+            case PIPE_FORMAT_RGTC1_SNORM:
+                return R500_TX_FORMAT_ATI1N | result;
+            case PIPE_FORMAT_RGTC2_UNORM:
+            case PIPE_FORMAT_RGTC2_SNORM:
+                return R400_TX_FORMAT_ATI2N | result;
+            default:
+                return ~0; /* Unsupported/unknown. */
+        }
+    }
+
     /* See whether the components are of the same size. */
-    for (i = 1; i < components; i++) {
+    for (i = 1; i < desc->nr_channels; i++) {
         uniform = uniform && desc->channel[0].size == desc->channel[i].size;
     }
 
     /* Non-uniform formats. */
     if (!uniform) {
-        switch (components) {
+        switch (desc->nr_channels) {
             case 3:
                 if (desc->channel[0].size == 5 &&
                     desc->channel[1].size == 6 &&
@@ -240,7 +250,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
 
             switch (desc->channel[0].size) {
                 case 4:
-                    switch (components) {
+                    switch (desc->nr_channels) {
                         case 2:
                             return R300_TX_FORMAT_Y4X4 | result;
                         case 4:
@@ -249,7 +259,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
                     return ~0;
 
                 case 8:
-                    switch (components) {
+                    switch (desc->nr_channels) {
                         case 1:
                             return R300_TX_FORMAT_X8 | result;
                         case 2:
@@ -260,7 +270,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
                     return ~0;
 
                 case 16:
-                    switch (components) {
+                    switch (desc->nr_channels) {
                         case 1:
                             return R300_TX_FORMAT_X16 | result;
                         case 2:
@@ -271,12 +281,11 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
             }
             return ~0;
 
-/* XXX Enable float textures here. */
-#if 0
+#if defined(ENABLE_FLOAT_TEXTURES)
         case UTIL_FORMAT_TYPE_FLOAT:
             switch (desc->channel[0].size) {
                 case 16:
-                    switch (components) {
+                    switch (desc->nr_channels) {
                         case 1:
                             return R300_TX_FORMAT_16F | result;
                         case 2:
@@ -287,7 +296,7 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
                     return ~0;
 
                 case 32:
-                    switch (components) {
+                    switch (desc->nr_channels) {
                         case 1:
                             return R300_TX_FORMAT_32F | result;
                         case 2:
@@ -302,6 +311,17 @@ static uint32_t r300_translate_texformat(enum pipe_format format)
     return ~0; /* Unsupported/unknown. */
 }
 
+static uint32_t r500_tx_format_msb_bit(enum pipe_format format)
+{
+    switch (format) {
+        case PIPE_FORMAT_RGTC1_UNORM:
+        case PIPE_FORMAT_RGTC1_SNORM:
+            return R500_TXFORMAT_MSB;
+        default:
+            return 0;
+    }
+}
+
 /* Buffer formats. */
 
 /* Colorbuffer formats. This is the unswizzled format of the RB3D block's
@@ -342,12 +362,13 @@ static uint32_t r300_translate_colorformat(enum pipe_format format)
         /* 64-bit buffers. */
         case PIPE_FORMAT_R16G16B16A16_UNORM:
         case PIPE_FORMAT_R16G16B16A16_SNORM:
-        //case PIPE_FORMAT_R16G16B16A16_FLOAT: /* not in pipe_format */
+#if defined(ENABLE_FLOAT_TEXTURES)
+        case PIPE_FORMAT_R16G16B16A16_FLOAT:
+#endif
             return R300_COLOR_FORMAT_ARGB16161616;
 
-/* XXX Enable float textures here. */
-#if 0
         /* 128-bit buffers. */
+#if defined(ENABLE_FLOAT_TEXTURES)
         case PIPE_FORMAT_R32G32B32A32_FLOAT:
             return R300_COLOR_FORMAT_ARGB32323232;
 #endif
@@ -372,7 +393,7 @@ static uint32_t r300_translate_zsformat(enum pipe_format format)
         /* 24-bit depth, ignored stencil */
         case PIPE_FORMAT_X8Z24_UNORM:
         /* 24-bit depth, 8-bit stencil */
-        case PIPE_FORMAT_S8Z24_UNORM:
+        case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
             return R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
         default:
             return ~0; /* Unsupported. */
@@ -527,6 +548,7 @@ static void r300_setup_texture_state(struct r300_screen* screen, struct r300_tex
         if (pt->height0 > 2048) {
             state->format2 |= R500_TXHEIGHT_BIT11;
         }
+        state->format2 |= r500_tx_format_msb_bit(pt->format);
     }
 
     SCREEN_DBG(screen, DBG_TEX, "r300: Set texture state (%dx%d, %d levels)\n",
diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
index 17cd5b8207..4ee31969e6 100644
--- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c
+++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
@@ -75,7 +75,7 @@ get_depth_stencil_values( struct depth_data *data,
       }
       break;
    case PIPE_FORMAT_Z24X8_UNORM:
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       for (j = 0; j < QUAD_SIZE; j++) {
          int x = quad->input.x0 % TILE_SIZE + (j & 1);
          int y = quad->input.y0 % TILE_SIZE + (j >> 1);
@@ -84,7 +84,7 @@ get_depth_stencil_values( struct depth_data *data,
       }
    break;
    case PIPE_FORMAT_X8Z24_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       for (j = 0; j < QUAD_SIZE; j++) {
          int x = quad->input.x0 % TILE_SIZE + (j & 1);
          int y = quad->input.y0 % TILE_SIZE + (j >> 1);
@@ -153,7 +153,7 @@ convert_quad_depth( struct depth_data *data,
       }
       break;
    case PIPE_FORMAT_Z24X8_UNORM:
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       {
          float scale = (float) ((1 << 24) - 1);
 
@@ -163,7 +163,7 @@ convert_quad_depth( struct depth_data *data,
       }
       break;
    case PIPE_FORMAT_X8Z24_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       {
          float scale = (float) ((1 << 24) - 1);
 
@@ -206,14 +206,14 @@ write_depth_stencil_values( struct depth_data *data,
          tile->data.depth32[y][x] = data->bzzzz[j];
       }
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       for (j = 0; j < QUAD_SIZE; j++) {
          int x = quad->input.x0 % TILE_SIZE + (j & 1);
          int y = quad->input.y0 % TILE_SIZE + (j >> 1);
          tile->data.depth32[y][x] = (data->stencilVals[j] << 24) | data->bzzzz[j];
       }
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       for (j = 0; j < QUAD_SIZE; j++) {
          int x = quad->input.x0 % TILE_SIZE + (j & 1);
          int y = quad->input.y0 % TILE_SIZE + (j >> 1);
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index 757dc86128..be64048abf 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -27,6 +27,7 @@
 
 
 #include "util/u_memory.h"
+#include "util/u_format_s3tc.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_screen.h"
 
@@ -155,20 +156,20 @@ softpipe_is_format_supported( struct pipe_screen *screen,
           target == PIPE_TEXTURE_CUBE);
 
    switch(format) {
-   case PIPE_FORMAT_L16_UNORM:
    case PIPE_FORMAT_YUYV:
    case PIPE_FORMAT_UYVY:
+      return FALSE;
+
    case PIPE_FORMAT_DXT1_RGB:
    case PIPE_FORMAT_DXT1_RGBA:
    case PIPE_FORMAT_DXT3_RGBA:
    case PIPE_FORMAT_DXT5_RGBA:
+      return util_format_s3tc_enabled;
+
    case PIPE_FORMAT_Z32_FLOAT:
-   case PIPE_FORMAT_R8G8_SNORM:
-   case PIPE_FORMAT_R5SG5SB6U_NORM:
-   case PIPE_FORMAT_R8SG8SB8UX8U_NORM:
-   case PIPE_FORMAT_R8G8B8A8_SNORM:
    case PIPE_FORMAT_NONE:
       return FALSE;
+
    default:
       break;
    }
@@ -240,6 +241,8 @@ softpipe_create_screen(struct sw_winsys *winsys)
    screen->base.context_create = softpipe_create_context;
    screen->base.flush_frontbuffer = softpipe_flush_frontbuffer;
 
+   util_format_s3tc_init();
+
    softpipe_init_screen_texture_funcs(&screen->base);
    softpipe_init_screen_buffer_funcs(&screen->base);
    softpipe_init_screen_fence_funcs(&screen->base);
diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c
index 1c3c2667d7..7551b923fb 100644
--- a/src/gallium/drivers/softpipe/sp_tile_cache.c
+++ b/src/gallium/drivers/softpipe/sp_tile_cache.c
@@ -158,13 +158,13 @@ sp_tile_cache_set_surface(struct softpipe_tile_cache *tc,
                                               PIPE_TRANSFER_READ_WRITE,
                                               0, 0, ps->width, ps->height);
 
-      tc->depth_stencil = (ps->format == PIPE_FORMAT_Z24S8_UNORM ||
+      tc->depth_stencil = (ps->format == PIPE_FORMAT_Z24_UNORM_S8_USCALED ||
                            ps->format == PIPE_FORMAT_Z24X8_UNORM ||
-                           ps->format == PIPE_FORMAT_S8Z24_UNORM ||
+                           ps->format == PIPE_FORMAT_S8_USCALED_Z24_UNORM ||
                            ps->format == PIPE_FORMAT_X8Z24_UNORM ||
                            ps->format == PIPE_FORMAT_Z16_UNORM ||
                            ps->format == PIPE_FORMAT_Z32_UNORM ||
-                           ps->format == PIPE_FORMAT_S8_UNORM);
+                           ps->format == PIPE_FORMAT_S8_USCALED);
    }
 }
 
diff --git a/src/gallium/drivers/svga/svga_pipe_clear.c b/src/gallium/drivers/svga/svga_pipe_clear.c
index 8483a3fad7..bb1664fbed 100644
--- a/src/gallium/drivers/svga/svga_pipe_clear.c
+++ b/src/gallium/drivers/svga/svga_pipe_clear.c
@@ -63,7 +63,7 @@ try_clear(struct svga_context *svga,
    if ((buffers & PIPE_CLEAR_DEPTHSTENCIL) && fb->zsbuf) {
       flags |= SVGA3D_CLEAR_DEPTH;
 
-      if (svga->curr.framebuffer.zsbuf->format == PIPE_FORMAT_S8Z24_UNORM)
+      if (svga->curr.framebuffer.zsbuf->format == PIPE_FORMAT_S8_USCALED_Z24_UNORM)
          flags |= SVGA3D_CLEAR_STENCIL;
 
       rect.w = MAX2(rect.w, fb->zsbuf->width);
diff --git a/src/gallium/drivers/svga/svga_pipe_misc.c b/src/gallium/drivers/svga/svga_pipe_misc.c
index 95bf0e6f91..2c43f9a24f 100644
--- a/src/gallium/drivers/svga/svga_pipe_misc.c
+++ b/src/gallium/drivers/svga/svga_pipe_misc.c
@@ -118,9 +118,9 @@ static void svga_set_framebuffer_state(struct pipe_context *pipe,
       case PIPE_FORMAT_Z16_UNORM:
          svga->curr.depthscale = 1.0f / DEPTH_BIAS_SCALE_FACTOR_D16;
          break;
-      case PIPE_FORMAT_Z24S8_UNORM:
+      case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       case PIPE_FORMAT_Z24X8_UNORM:
-      case PIPE_FORMAT_S8Z24_UNORM:
+      case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       case PIPE_FORMAT_X8Z24_UNORM:
          svga->curr.depthscale = 1.0f / DEPTH_BIAS_SCALE_FACTOR_D24S8;
          break;
diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index 6022c38cfc..9a2e7c6f5b 100644
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -210,7 +210,7 @@ svga_translate_format_cap(enum pipe_format format)
 
    case PIPE_FORMAT_Z16_UNORM:
       return SVGA3D_DEVCAP_SURFACEFMT_Z_D16;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       return SVGA3D_DEVCAP_SURFACEFMT_Z_D24S8;
    case PIPE_FORMAT_X8Z24_UNORM:
       return SVGA3D_DEVCAP_SURFACEFMT_Z_D24X8;
diff --git a/src/gallium/drivers/svga/svga_screen_texture.c b/src/gallium/drivers/svga/svga_screen_texture.c
index 70d3a8a46a..811c746695 100644
--- a/src/gallium/drivers/svga/svga_screen_texture.c
+++ b/src/gallium/drivers/svga/svga_screen_texture.c
@@ -81,7 +81,7 @@ svga_translate_format(enum pipe_format format)
     */
    case PIPE_FORMAT_Z16_UNORM:
       return SVGA3D_Z_D16;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       return SVGA3D_Z_D24S8;
    case PIPE_FORMAT_X8Z24_UNORM:
       return SVGA3D_Z_D24X8;
@@ -114,7 +114,7 @@ svga_translate_format_render(enum pipe_format format)
    case PIPE_FORMAT_B5G5R5A1_UNORM:
    case PIPE_FORMAT_B4G4R4A4_UNORM:
    case PIPE_FORMAT_B5G6R5_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
    case PIPE_FORMAT_Z32_UNORM:
    case PIPE_FORMAT_Z16_UNORM:
diff --git a/src/gallium/drivers/svga/svga_state_framebuffer.c b/src/gallium/drivers/svga/svga_state_framebuffer.c
index b710914acd..bd92f00343 100644
--- a/src/gallium/drivers/svga/svga_state_framebuffer.c
+++ b/src/gallium/drivers/svga/svga_state_framebuffer.c
@@ -70,7 +70,7 @@ static int emit_framebuffer( struct svga_context *svga,
          return ret;
 
       if (curr->zsbuf &&
-          curr->zsbuf->format == PIPE_FORMAT_S8Z24_UNORM) {
+          curr->zsbuf->format == PIPE_FORMAT_S8_USCALED_Z24_UNORM) {
          ret = SVGA3D_SetRenderTarget(svga->swc, SVGA3D_RT_STENCIL, curr->zsbuf);
          if (ret != PIPE_OK)
             return ret;
diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h
index ba2985f449..c7a90a09a1 100644
--- a/src/gallium/include/pipe/p_format.h
+++ b/src/gallium/include/pipe/p_format.h
@@ -44,120 +44,140 @@ extern "C" {
  */
 
 enum pipe_format {
-   PIPE_FORMAT_NONE                  = 0,
-   PIPE_FORMAT_B8G8R8A8_UNORM        = 1,
-   PIPE_FORMAT_B8G8R8X8_UNORM        = 2,
-   PIPE_FORMAT_A8R8G8B8_UNORM        = 3,
-   PIPE_FORMAT_X8R8G8B8_UNORM        = 4,
-   PIPE_FORMAT_B5G5R5A1_UNORM        = 5,
-   PIPE_FORMAT_B4G4R4A4_UNORM        = 6,
-   PIPE_FORMAT_B5G6R5_UNORM          = 7,
-   PIPE_FORMAT_R10G10B10A2_UNORM     = 8,
-   PIPE_FORMAT_L8_UNORM              = 9,    /**< ubyte luminance */
-   PIPE_FORMAT_A8_UNORM              = 10,   /**< ubyte alpha */
-   PIPE_FORMAT_I8_UNORM              = 11,   /**< ubyte intensity */
-   PIPE_FORMAT_L8A8_UNORM            = 12,   /**< ubyte alpha, luminance */
-   PIPE_FORMAT_L16_UNORM             = 13,   /**< ushort luminance */
-   PIPE_FORMAT_UYVY                  = 14,
-   PIPE_FORMAT_YUYV                  = 15,
-   PIPE_FORMAT_Z16_UNORM             = 16,
-   PIPE_FORMAT_Z32_UNORM             = 17,
-   PIPE_FORMAT_Z32_FLOAT             = 18,
-   PIPE_FORMAT_Z24S8_UNORM           = 19,
-   PIPE_FORMAT_S8Z24_UNORM           = 20,
-   PIPE_FORMAT_Z24X8_UNORM           = 21,
-   PIPE_FORMAT_X8Z24_UNORM           = 22,
-   PIPE_FORMAT_S8_UNORM              = 23,   /**< ubyte stencil */
-   PIPE_FORMAT_R64_FLOAT             = 24,
-   PIPE_FORMAT_R64G64_FLOAT          = 25,
-   PIPE_FORMAT_R64G64B64_FLOAT       = 26,
-   PIPE_FORMAT_R64G64B64A64_FLOAT    = 27,
-   PIPE_FORMAT_R32_FLOAT             = 28,
-   PIPE_FORMAT_R32G32_FLOAT          = 29,
-   PIPE_FORMAT_R32G32B32_FLOAT       = 30,
-   PIPE_FORMAT_R32G32B32A32_FLOAT    = 31,
-   PIPE_FORMAT_R32_UNORM             = 32,
-   PIPE_FORMAT_R32G32_UNORM          = 33,
-   PIPE_FORMAT_R32G32B32_UNORM       = 34,
-   PIPE_FORMAT_R32G32B32A32_UNORM    = 35,
-   PIPE_FORMAT_R32_USCALED           = 36,
-   PIPE_FORMAT_R32G32_USCALED        = 37,
-   PIPE_FORMAT_R32G32B32_USCALED     = 38,
-   PIPE_FORMAT_R32G32B32A32_USCALED  = 39,
-   PIPE_FORMAT_R32_SNORM             = 40,
-   PIPE_FORMAT_R32G32_SNORM          = 41,
-   PIPE_FORMAT_R32G32B32_SNORM       = 42,
-   PIPE_FORMAT_R32G32B32A32_SNORM    = 43,
-   PIPE_FORMAT_R32_SSCALED           = 44,
-   PIPE_FORMAT_R32G32_SSCALED        = 45,
-   PIPE_FORMAT_R32G32B32_SSCALED     = 46,
-   PIPE_FORMAT_R32G32B32A32_SSCALED  = 47,
-   PIPE_FORMAT_R16_UNORM             = 48,
-   PIPE_FORMAT_R16G16_UNORM          = 49,
-   PIPE_FORMAT_R16G16B16_UNORM       = 50,
-   PIPE_FORMAT_R16G16B16A16_UNORM    = 51,
-   PIPE_FORMAT_R16_USCALED           = 52,
-   PIPE_FORMAT_R16G16_USCALED        = 53,
-   PIPE_FORMAT_R16G16B16_USCALED     = 54,
-   PIPE_FORMAT_R16G16B16A16_USCALED  = 55,
-   PIPE_FORMAT_R16_SNORM             = 56,
-   PIPE_FORMAT_R16G16_SNORM          = 57,
-   PIPE_FORMAT_R16G16B16_SNORM       = 58,
-   PIPE_FORMAT_R16G16B16A16_SNORM    = 59,
-   PIPE_FORMAT_R16_SSCALED           = 60,
-   PIPE_FORMAT_R16G16_SSCALED        = 61,
-   PIPE_FORMAT_R16G16B16_SSCALED     = 62,
-   PIPE_FORMAT_R16G16B16A16_SSCALED  = 63,
-   PIPE_FORMAT_R8_UNORM              = 64,
-   PIPE_FORMAT_R8G8_UNORM            = 65,
-   PIPE_FORMAT_R8G8B8_UNORM          = 66,
-   PIPE_FORMAT_R8G8B8A8_UNORM        = 67,
-   PIPE_FORMAT_X8B8G8R8_UNORM        = 68,
-   PIPE_FORMAT_R8_USCALED            = 69,
-   PIPE_FORMAT_R8G8_USCALED          = 70,
-   PIPE_FORMAT_R8G8B8_USCALED        = 71,
-   PIPE_FORMAT_R8G8B8A8_USCALED      = 72,
-   PIPE_FORMAT_R8_SNORM              = 74,
-   PIPE_FORMAT_R8G8_SNORM            = 75,
-   PIPE_FORMAT_R8G8B8_SNORM          = 76,
-   PIPE_FORMAT_R8G8B8A8_SNORM        = 77,
-   PIPE_FORMAT_R8_SSCALED            = 82,
-   PIPE_FORMAT_R8G8_SSCALED          = 83,
-   PIPE_FORMAT_R8G8B8_SSCALED        = 84,
-   PIPE_FORMAT_R8G8B8A8_SSCALED      = 85,
-   PIPE_FORMAT_R32_FIXED             = 87,
-   PIPE_FORMAT_R32G32_FIXED          = 88,
-   PIPE_FORMAT_R32G32B32_FIXED       = 89,
-   PIPE_FORMAT_R32G32B32A32_FIXED    = 90,
-   /* sRGB formats */
-   PIPE_FORMAT_L8_SRGB               = 91,
-   PIPE_FORMAT_L8A8_SRGB             = 92,
-   PIPE_FORMAT_R8G8B8_SRGB           = 93,
-   PIPE_FORMAT_A8B8G8R8_SRGB         = 94,
-   PIPE_FORMAT_X8B8G8R8_SRGB         = 95,
-   PIPE_FORMAT_B8G8R8A8_SRGB         = 96,
-   PIPE_FORMAT_B8G8R8X8_SRGB         = 97,
-   PIPE_FORMAT_A8R8G8B8_SRGB         = 98,
-   PIPE_FORMAT_X8R8G8B8_SRGB         = 99,
+   PIPE_FORMAT_NONE                    = 0,
+   PIPE_FORMAT_B8G8R8A8_UNORM          = 1,
+   PIPE_FORMAT_B8G8R8X8_UNORM          = 2,
+   PIPE_FORMAT_A8R8G8B8_UNORM          = 3,
+   PIPE_FORMAT_X8R8G8B8_UNORM          = 4,
+   PIPE_FORMAT_B5G5R5A1_UNORM          = 5,
+   PIPE_FORMAT_B4G4R4A4_UNORM          = 6,
+   PIPE_FORMAT_B5G6R5_UNORM            = 7,
+   PIPE_FORMAT_R10G10B10A2_UNORM       = 8,
+   PIPE_FORMAT_L8_UNORM                = 9,    /**< ubyte luminance */
+   PIPE_FORMAT_A8_UNORM                = 10,   /**< ubyte alpha */
+   PIPE_FORMAT_I8_UNORM                = 11,   /**< ubyte intensity */
+   PIPE_FORMAT_L8A8_UNORM              = 12,   /**< ubyte alpha, luminance */
+   PIPE_FORMAT_L16_UNORM               = 13,   /**< ushort luminance */
+   PIPE_FORMAT_UYVY                    = 14,
+   PIPE_FORMAT_YUYV                    = 15,
+   PIPE_FORMAT_Z16_UNORM               = 16,
+   PIPE_FORMAT_Z32_UNORM               = 17,
+   PIPE_FORMAT_Z32_FLOAT               = 18,
+   PIPE_FORMAT_Z24_UNORM_S8_USCALED    = 19,
+   PIPE_FORMAT_S8_USCALED_Z24_UNORM    = 20,
+   PIPE_FORMAT_Z24X8_UNORM             = 21,
+   PIPE_FORMAT_X8Z24_UNORM             = 22,
+   PIPE_FORMAT_S8_USCALED              = 23,   /**< ubyte stencil */
+   PIPE_FORMAT_R64_FLOAT               = 24,
+   PIPE_FORMAT_R64G64_FLOAT            = 25,
+   PIPE_FORMAT_R64G64B64_FLOAT         = 26,
+   PIPE_FORMAT_R64G64B64A64_FLOAT      = 27,
+   PIPE_FORMAT_R32_FLOAT               = 28,
+   PIPE_FORMAT_R32G32_FLOAT            = 29,
+   PIPE_FORMAT_R32G32B32_FLOAT         = 30,
+   PIPE_FORMAT_R32G32B32A32_FLOAT      = 31,
+   PIPE_FORMAT_R32_UNORM               = 32,
+   PIPE_FORMAT_R32G32_UNORM            = 33,
+   PIPE_FORMAT_R32G32B32_UNORM         = 34,
+   PIPE_FORMAT_R32G32B32A32_UNORM      = 35,
+   PIPE_FORMAT_R32_USCALED             = 36,
+   PIPE_FORMAT_R32G32_USCALED          = 37,
+   PIPE_FORMAT_R32G32B32_USCALED       = 38,
+   PIPE_FORMAT_R32G32B32A32_USCALED    = 39,
+   PIPE_FORMAT_R32_SNORM               = 40,
+   PIPE_FORMAT_R32G32_SNORM            = 41,
+   PIPE_FORMAT_R32G32B32_SNORM         = 42,
+   PIPE_FORMAT_R32G32B32A32_SNORM      = 43,
+   PIPE_FORMAT_R32_SSCALED             = 44,
+   PIPE_FORMAT_R32G32_SSCALED          = 45,
+   PIPE_FORMAT_R32G32B32_SSCALED       = 46,
+   PIPE_FORMAT_R32G32B32A32_SSCALED    = 47,
+   PIPE_FORMAT_R16_UNORM               = 48,
+   PIPE_FORMAT_R16G16_UNORM            = 49,
+   PIPE_FORMAT_R16G16B16_UNORM         = 50,
+   PIPE_FORMAT_R16G16B16A16_UNORM      = 51,
+   PIPE_FORMAT_R16_USCALED             = 52,
+   PIPE_FORMAT_R16G16_USCALED          = 53,
+   PIPE_FORMAT_R16G16B16_USCALED       = 54,
+   PIPE_FORMAT_R16G16B16A16_USCALED    = 55,
+   PIPE_FORMAT_R16_SNORM               = 56,
+   PIPE_FORMAT_R16G16_SNORM            = 57,
+   PIPE_FORMAT_R16G16B16_SNORM         = 58,
+   PIPE_FORMAT_R16G16B16A16_SNORM      = 59,
+   PIPE_FORMAT_R16_SSCALED             = 60,
+   PIPE_FORMAT_R16G16_SSCALED          = 61,
+   PIPE_FORMAT_R16G16B16_SSCALED       = 62,
+   PIPE_FORMAT_R16G16B16A16_SSCALED    = 63,
+   PIPE_FORMAT_R8_UNORM                = 64,
+   PIPE_FORMAT_R8G8_UNORM              = 65,
+   PIPE_FORMAT_R8G8B8_UNORM            = 66,
+   PIPE_FORMAT_R8G8B8A8_UNORM          = 67,
+   PIPE_FORMAT_X8B8G8R8_UNORM          = 68,
+   PIPE_FORMAT_R8_USCALED              = 69,
+   PIPE_FORMAT_R8G8_USCALED            = 70,
+   PIPE_FORMAT_R8G8B8_USCALED          = 71,
+   PIPE_FORMAT_R8G8B8A8_USCALED        = 72,
+   PIPE_FORMAT_R8_SNORM                = 74,
+   PIPE_FORMAT_R8G8_SNORM              = 75,
+   PIPE_FORMAT_R8G8B8_SNORM            = 76,
+   PIPE_FORMAT_R8G8B8A8_SNORM          = 77,
+   PIPE_FORMAT_R8_SSCALED              = 82,
+   PIPE_FORMAT_R8G8_SSCALED            = 83,
+   PIPE_FORMAT_R8G8B8_SSCALED          = 84,
+   PIPE_FORMAT_R8G8B8A8_SSCALED        = 85,
+   PIPE_FORMAT_R32_FIXED               = 87,
+   PIPE_FORMAT_R32G32_FIXED            = 88,
+   PIPE_FORMAT_R32G32B32_FIXED         = 89,
+   PIPE_FORMAT_R32G32B32A32_FIXED      = 90,
+   PIPE_FORMAT_R16_FLOAT               = 91,
+   PIPE_FORMAT_R16G16_FLOAT            = 92,
+   PIPE_FORMAT_R16G16B16_FLOAT         = 93,
+   PIPE_FORMAT_R16G16B16A16_FLOAT      = 94,
 
-   /* mixed formats */
-   PIPE_FORMAT_R8SG8SB8UX8U_NORM     = 100,
-   PIPE_FORMAT_R5SG5SB6U_NORM        = 101,
+   /* sRGB formats */
+   PIPE_FORMAT_L8_SRGB                 = 95,
+   PIPE_FORMAT_L8A8_SRGB               = 96,
+   PIPE_FORMAT_R8G8B8_SRGB             = 97,
+   PIPE_FORMAT_A8B8G8R8_SRGB           = 98,
+   PIPE_FORMAT_X8B8G8R8_SRGB           = 99,
+   PIPE_FORMAT_B8G8R8A8_SRGB           = 100,
+   PIPE_FORMAT_B8G8R8X8_SRGB           = 101,
+   PIPE_FORMAT_A8R8G8B8_SRGB           = 102,
+   PIPE_FORMAT_X8R8G8B8_SRGB           = 103,
+   PIPE_FORMAT_R8G8B8A8_SRGB           = 104,
 
    /* compressed formats */
-   PIPE_FORMAT_DXT1_RGB              = 102,
-   PIPE_FORMAT_DXT1_RGBA             = 103,
-   PIPE_FORMAT_DXT3_RGBA             = 104,
-   PIPE_FORMAT_DXT5_RGBA             = 105,
+   PIPE_FORMAT_DXT1_RGB                = 105,
+   PIPE_FORMAT_DXT1_RGBA               = 106,
+   PIPE_FORMAT_DXT3_RGBA               = 107,
+   PIPE_FORMAT_DXT5_RGBA               = 108,
 
    /* sRGB, compressed */
-   PIPE_FORMAT_DXT1_SRGB             = 106,
-   PIPE_FORMAT_DXT1_SRGBA            = 107,
-   PIPE_FORMAT_DXT3_SRGBA            = 108,
-   PIPE_FORMAT_DXT5_SRGBA            = 109,
+   PIPE_FORMAT_DXT1_SRGB               = 109,
+   PIPE_FORMAT_DXT1_SRGBA              = 110,
+   PIPE_FORMAT_DXT3_SRGBA              = 111,
+   PIPE_FORMAT_DXT5_SRGBA              = 112,
 
-   PIPE_FORMAT_A8B8G8R8_UNORM        = 110,
-   PIPE_FORMAT_B5G5R5X1_UNORM        = 111,
+   /* rgtc compressed */
+   PIPE_FORMAT_RGTC1_UNORM             = 113,
+   PIPE_FORMAT_RGTC1_SNORM             = 114,
+   PIPE_FORMAT_RGTC2_UNORM             = 115,
+   PIPE_FORMAT_RGTC2_SNORM             = 116,
+
+   PIPE_FORMAT_R8G8_B8G8_UNORM         = 117,
+   PIPE_FORMAT_G8R8_G8B8_UNORM         = 118,
+
+   /* mixed formats */
+   PIPE_FORMAT_R8SG8SB8UX8U_NORM       = 119,
+   PIPE_FORMAT_R5SG5SB6U_NORM          = 120,
+
+   PIPE_FORMAT_A8B8G8R8_UNORM          = 121,
+   PIPE_FORMAT_B5G5R5X1_UNORM          = 122,
+   PIPE_FORMAT_R10G10B10A2_USCALED     = 123,
+   PIPE_FORMAT_R11G11B10_FLOAT         = 124,
+   PIPE_FORMAT_R9G9B9E5_FLOAT          = 125,
+   PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED = 126,
+   PIPE_FORMAT_R1_UNORM                = 127,
 
    PIPE_FORMAT_COUNT
 };
diff --git a/src/gallium/state_trackers/dri/common/dri_drawable.c b/src/gallium/state_trackers/dri/common/dri_drawable.c
index 75b4cc5628..88c17e81bf 100644
--- a/src/gallium/state_trackers/dri/common/dri_drawable.c
+++ b/src/gallium/state_trackers/dri/common/dri_drawable.c
@@ -38,7 +38,6 @@
 #include "pipe/p_screen.h"
 #include "util/u_format.h"
 #include "util/u_memory.h"
-#include "util/u_inlines.h"
  
 /**
  * This is called when we need to set up GL rendering to a new X window.
diff --git a/src/gallium/state_trackers/dri/common/dri_screen.c b/src/gallium/state_trackers/dri/common/dri_screen.c
index 821c1dea62..6fc7487071 100644
--- a/src/gallium/state_trackers/dri/common/dri_screen.c
+++ b/src/gallium/state_trackers/dri/common/dri_screen.c
@@ -97,10 +97,10 @@ dri_fill_in_modes(struct dri_screen *screen,
    pf_z24x8 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_X8Z24_UNORM,
 					    PIPE_TEXTURE_2D,
 					    PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0);
-   pf_s8z24 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_Z24S8_UNORM,
+   pf_s8z24 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_Z24_UNORM_S8_USCALED,
 					    PIPE_TEXTURE_2D,
 					    PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0);
-   pf_z24s8 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_S8Z24_UNORM,
+   pf_z24s8 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_S8_USCALED_Z24_UNORM,
 					    PIPE_TEXTURE_2D,
 					    PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0);
    pf_a8r8g8b8 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_B8G8R8A8_UNORM,
@@ -241,8 +241,8 @@ dri_fill_st_visual(struct st_visual *stvis, struct dri_screen *screen,
                                           PIPE_FORMAT_X8Z24_UNORM;
       } else {
 	 stvis->depth_stencil_format = (screen->sd_depth_bits_last) ?
-                                          PIPE_FORMAT_Z24S8_UNORM:
-                                          PIPE_FORMAT_S8Z24_UNORM;
+                                          PIPE_FORMAT_Z24_UNORM_S8_USCALED:
+                                          PIPE_FORMAT_S8_USCALED_Z24_UNORM;
       }
       break;
    case 32:
diff --git a/src/gallium/state_trackers/dri/common/dri_st_api.c b/src/gallium/state_trackers/dri/common/dri_st_api.c
index 7ba7ec383d..1a5259c68b 100644
--- a/src/gallium/state_trackers/dri/common/dri_st_api.c
+++ b/src/gallium/state_trackers/dri/common/dri_st_api.c
@@ -35,7 +35,6 @@
 #include "dri_context.h"
 #include "dri_drawable.h"
 #include "dri_st_api.h"
-#include "dri1_helper.h"
 #ifndef __NOT_HAVE_DRM_H
 #include "dri1.h"
 #include "dri2.h"
diff --git a/src/gallium/state_trackers/egl/kms/native_kms.c b/src/gallium/state_trackers/egl/kms/native_kms.c
index 7322240856..3174774490 100644
--- a/src/gallium/state_trackers/egl/kms/native_kms.c
+++ b/src/gallium/state_trackers/egl/kms/native_kms.c
@@ -631,9 +631,9 @@ kms_display_get_configs(struct native_display *ndpy, int *num_configs)
       nconf->mode.alphaBits = 8;
       nconf->mode.rgbBits = 32;
 
-      format = PIPE_FORMAT_Z24S8_UNORM;
+      format = PIPE_FORMAT_Z24_UNORM_S8_USCALED;
       if (!kms_display_is_format_supported(&kdpy->base, format, FALSE)) {
-         format = PIPE_FORMAT_S8Z24_UNORM;
+         format = PIPE_FORMAT_S8_USCALED_Z24_UNORM;
          if (!kms_display_is_format_supported(&kdpy->base, format, FALSE))
             format = PIPE_FORMAT_NONE;
       }
diff --git a/src/gallium/state_trackers/egl/x11/native_dri2.c b/src/gallium/state_trackers/egl/x11/native_dri2.c
index 9839979231..8c0553ce0d 100644
--- a/src/gallium/state_trackers/egl/x11/native_dri2.c
+++ b/src/gallium/state_trackers/egl/x11/native_dri2.c
@@ -558,8 +558,8 @@ choose_depth_stencil_format(const __GLcontextModes *mode,
       break;
    case 24:
       if (mode->stencilBits) {
-         formats[count++] = PIPE_FORMAT_Z24S8_UNORM;
-         formats[count++] = PIPE_FORMAT_S8Z24_UNORM;
+         formats[count++] = PIPE_FORMAT_Z24_UNORM_S8_USCALED;
+         formats[count++] = PIPE_FORMAT_S8_USCALED_Z24_UNORM;
       }
       else {
          formats[count++] = PIPE_FORMAT_Z24X8_UNORM;
diff --git a/src/gallium/state_trackers/egl/x11/native_ximage.c b/src/gallium/state_trackers/egl/x11/native_ximage.c
index c6b16354f9..d69011ac5e 100644
--- a/src/gallium/state_trackers/egl/x11/native_ximage.c
+++ b/src/gallium/state_trackers/egl/x11/native_ximage.c
@@ -526,8 +526,8 @@ ximage_display_get_configs(struct native_display *ndpy, int *num_configs)
             xconf->base.stencil_format = PIPE_FORMAT_NONE;
             /* create the second config with depth/stencil buffer */
             if (j == 1) {
-               xconf->base.depth_format = PIPE_FORMAT_Z24S8_UNORM;
-               xconf->base.stencil_format = PIPE_FORMAT_Z24S8_UNORM;
+               xconf->base.depth_format = PIPE_FORMAT_Z24_UNORM_S8_USCALED;
+               xconf->base.stencil_format = PIPE_FORMAT_Z24_UNORM_S8_USCALED;
                mode->depthBits = 24;
                mode->stencilBits = 8;
                mode->haveDepthBuffer = TRUE;
diff --git a/src/gallium/state_trackers/glx/xlib/xm_api.c b/src/gallium/state_trackers/glx/xlib/xm_api.c
index fd03d3c46a..c394430b94 100644
--- a/src/gallium/state_trackers/glx/xlib/xm_api.c
+++ b/src/gallium/state_trackers/glx/xlib/xm_api.c
@@ -355,8 +355,8 @@ choose_depth_stencil_format(XMesaDisplay xmdpy, int depth, int stencil)
       formats[count++] = PIPE_FORMAT_Z24X8_UNORM;
    }
    if (depth <= 24 && stencil <= 8) {
-      formats[count++] = PIPE_FORMAT_S8Z24_UNORM;
-      formats[count++] = PIPE_FORMAT_Z24S8_UNORM;
+      formats[count++] = PIPE_FORMAT_S8_USCALED_Z24_UNORM;
+      formats[count++] = PIPE_FORMAT_Z24_UNORM_S8_USCALED;
    }
    if (depth <= 32 && stencil == 0) {
       formats[count++] = PIPE_FORMAT_Z32_UNORM;
diff --git a/src/gallium/state_trackers/python/gallium.i b/src/gallium/state_trackers/python/gallium.i
index 632d71ccbe..aeed1cd118 100644
--- a/src/gallium/state_trackers/python/gallium.i
+++ b/src/gallium/state_trackers/python/gallium.i
@@ -95,7 +95,7 @@
 
 
 %include "p_compiler.i"
-%include "p_defines.h";
+%include "p_defines.h"
 %include "p_format.h"
 
 %include "p_device.i"
@@ -103,3 +103,4 @@
 %include "p_texture.i"
 %include "p_state.i"
 
+%include "u_format.i"
diff --git a/src/gallium/state_trackers/python/u_format.i b/src/gallium/state_trackers/python/u_format.i
new file mode 100644
index 0000000000..2184b42084
--- /dev/null
+++ b/src/gallium/state_trackers/python/u_format.i
@@ -0,0 +1,88 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+static INLINE const char *
+util_format_name(enum pipe_format format);
+
+static INLINE boolean
+util_format_is_s3tc(enum pipe_format format);
+
+static INLINE boolean
+util_format_is_depth_or_stencil(enum pipe_format format);
+
+static INLINE boolean
+util_format_is_depth_and_stencil(enum pipe_format format);
+
+
+uint
+util_format_get_blocksizebits(enum pipe_format format);
+
+uint
+util_format_get_blocksize(enum pipe_format format);
+
+uint
+util_format_get_blockwidth(enum pipe_format format);
+
+uint
+util_format_get_blockheight(enum pipe_format format);
+
+unsigned
+util_format_get_nblocksx(enum pipe_format format,
+                         unsigned x);
+
+unsigned
+util_format_get_nblocksy(enum pipe_format format,
+                         unsigned y);
+
+unsigned
+util_format_get_nblocks(enum pipe_format format,
+                        unsigned width,
+                        unsigned height);
+
+size_t
+util_format_get_stride(enum pipe_format format,
+                       unsigned width);
+
+size_t
+util_format_get_2d_size(enum pipe_format format,
+                        size_t stride,
+                        unsigned height);
+
+uint
+util_format_get_component_bits(enum pipe_format format,
+                               enum util_format_colorspace colorspace,
+                               uint component);
+
+boolean
+util_format_has_alpha(enum pipe_format format);
+
+
+unsigned
+util_format_get_nr_components(enum pipe_format format);
+
+
diff --git a/src/gallium/state_trackers/vega/vg_tracker.c b/src/gallium/state_trackers/vega/vg_tracker.c
index f438e34087..108dd6a975 100644
--- a/src/gallium/state_trackers/vega/vg_tracker.c
+++ b/src/gallium/state_trackers/vega/vg_tracker.c
@@ -208,7 +208,7 @@ struct st_framebuffer * st_create_framebuffer(const void *visual,
       if (stencilFormat == depthFormat)
          stfb->dsrb = st_new_renderbuffer_fb(stencilFormat);
       else
-         stfb->dsrb = st_new_renderbuffer_fb(PIPE_FORMAT_Z24S8_UNORM);
+         stfb->dsrb = st_new_renderbuffer_fb(PIPE_FORMAT_Z24_UNORM_S8_USCALED);
 
       /*### currently we always allocate it but it's possible it's
         not necessary if EGL_ALPHA_MASK_SIZE was 0
diff --git a/src/gallium/state_trackers/wgl/stw_pixelformat.c b/src/gallium/state_trackers/wgl/stw_pixelformat.c
index bc28f31ed1..02ff89a388 100644
--- a/src/gallium/state_trackers/wgl/stw_pixelformat.c
+++ b/src/gallium/state_trackers/wgl/stw_pixelformat.c
@@ -96,8 +96,8 @@ stw_pf_depth_stencil[] = {
    { PIPE_FORMAT_Z24X8_UNORM, {24, 0} },
    { PIPE_FORMAT_Z16_UNORM,   {16, 0} },
    /* combined depth-stencil */
-   { PIPE_FORMAT_Z24S8_UNORM, {24, 8} },
-   { PIPE_FORMAT_S8Z24_UNORM, {24, 8} }
+   { PIPE_FORMAT_Z24_UNORM_S8_USCALED, {24, 8} },
+   { PIPE_FORMAT_S8_USCALED_Z24_UNORM, {24, 8} }
 };
 
 
diff --git a/src/gallium/state_trackers/xorg/xorg_dri2.c b/src/gallium/state_trackers/xorg/xorg_dri2.c
index f23e4c6cc7..91c251bb4f 100644
--- a/src/gallium/state_trackers/xorg/xorg_dri2.c
+++ b/src/gallium/state_trackers/xorg/xorg_dri2.c
@@ -122,7 +122,7 @@ dri2_do_create_buffer(DrawablePtr pDraw, DRI2BufferPtr buffer, unsigned int form
                }
             } else {
                template.format = ms->ds_depth_bits_last ?
-                                 PIPE_FORMAT_Z24S8_UNORM : PIPE_FORMAT_S8Z24_UNORM;
+                                 PIPE_FORMAT_Z24_UNORM_S8_USCALED : PIPE_FORMAT_S8_USCALED_Z24_UNORM;
             }
 	    template.width0 = pDraw->width;
 	    template.height0 = pDraw->height;
@@ -439,7 +439,7 @@ xorg_dri2_init(ScreenPtr pScreen)
 					 PIPE_TEXTURE_2D,
 					 PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0);
     ms->ds_depth_bits_last =
-	 ms->screen->is_format_supported(ms->screen, PIPE_FORMAT_Z24S8_UNORM,
+	 ms->screen->is_format_supported(ms->screen, PIPE_FORMAT_Z24_UNORM_S8_USCALED,
 					 PIPE_TEXTURE_2D,
 					 PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0);
 
diff --git a/src/gallium/targets/libgl-xlib/SConscript b/src/gallium/targets/libgl-xlib/SConscript
index efa7e797d1..c1614d4a9f 100644
--- a/src/gallium/targets/libgl-xlib/SConscript
+++ b/src/gallium/targets/libgl-xlib/SConscript
@@ -43,6 +43,8 @@ sources = [
     'xlib.c',
 ]
 
+env.Tool('x11')
+
 if 'softpipe' in env['drivers']:
     env.Append(CPPDEFINES = 'GALLIUM_SOFTPIPE')
     env.Prepend(LIBS = [softpipe])