11 files changed, 5146 insertions, 5146 deletions
diff --git a/mesalib/src/mesa/state_tracker/st_atom_blend.c b/mesalib/src/mesa/state_tracker/st_atom_blend.c
index fb1c7a4ef..e3b3b9936 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_blend.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_blend.c
@@ -1,302 +1,302 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  *   Brian Paul
-  */
- 
-
-#include "st_context.h"
-#include "st_atom.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "cso_cache/cso_context.h"
-
-#include "main/macros.h"
-
-/**
- * Convert GLenum blend tokens to pipe tokens.
- * Both blend factors and blend funcs are accepted.
- */
-static GLuint
-translate_blend(GLenum blend)
-{
-   switch (blend) {
-   /* blend functions */
-   case GL_FUNC_ADD:
-      return PIPE_BLEND_ADD;
-   case GL_FUNC_SUBTRACT:
-      return PIPE_BLEND_SUBTRACT;
-   case GL_FUNC_REVERSE_SUBTRACT:
-      return PIPE_BLEND_REVERSE_SUBTRACT;
-   case GL_MIN:
-      return PIPE_BLEND_MIN;
-   case GL_MAX:
-      return PIPE_BLEND_MAX;
-
-   /* blend factors */
-   case GL_ONE:
-      return PIPE_BLENDFACTOR_ONE;
-   case GL_SRC_COLOR:
-      return PIPE_BLENDFACTOR_SRC_COLOR;
-   case GL_SRC_ALPHA:
-      return PIPE_BLENDFACTOR_SRC_ALPHA;
-   case GL_DST_ALPHA:
-      return PIPE_BLENDFACTOR_DST_ALPHA;
-   case GL_DST_COLOR:
-      return PIPE_BLENDFACTOR_DST_COLOR;
-   case GL_SRC_ALPHA_SATURATE:
-      return PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE;
-   case GL_CONSTANT_COLOR:
-      return PIPE_BLENDFACTOR_CONST_COLOR;
-   case GL_CONSTANT_ALPHA:
-      return PIPE_BLENDFACTOR_CONST_ALPHA;
-      /*
-      return PIPE_BLENDFACTOR_SRC1_COLOR;
-      return PIPE_BLENDFACTOR_SRC1_ALPHA;
-      */
-   case GL_ZERO:
-      return PIPE_BLENDFACTOR_ZERO;
-   case GL_ONE_MINUS_SRC_COLOR:
-      return PIPE_BLENDFACTOR_INV_SRC_COLOR;
-   case GL_ONE_MINUS_SRC_ALPHA:
-      return PIPE_BLENDFACTOR_INV_SRC_ALPHA;
-   case GL_ONE_MINUS_DST_COLOR:
-      return PIPE_BLENDFACTOR_INV_DST_COLOR;
-   case GL_ONE_MINUS_DST_ALPHA:
-      return PIPE_BLENDFACTOR_INV_DST_ALPHA;
-   case GL_ONE_MINUS_CONSTANT_COLOR:
-      return PIPE_BLENDFACTOR_INV_CONST_COLOR;
-   case GL_ONE_MINUS_CONSTANT_ALPHA:
-      return PIPE_BLENDFACTOR_INV_CONST_ALPHA;
-      /*
-      return PIPE_BLENDFACTOR_INV_SRC1_COLOR;
-      return PIPE_BLENDFACTOR_INV_SRC1_ALPHA;
-      */
-   default:
-      assert("invalid GL token in translate_blend()" == NULL);
-      return 0;
-   }
-}
-
-
-/**
- * Convert GLenum logicop tokens to pipe tokens.
- */
-static GLuint
-translate_logicop(GLenum logicop)
-{
-   switch (logicop) {
-   case GL_CLEAR:
-      return PIPE_LOGICOP_CLEAR;
-   case GL_NOR:
-      return PIPE_LOGICOP_NOR;
-   case GL_AND_INVERTED:
-      return PIPE_LOGICOP_AND_INVERTED;
-   case GL_COPY_INVERTED:
-      return PIPE_LOGICOP_COPY_INVERTED;
-   case GL_AND_REVERSE:
-      return PIPE_LOGICOP_AND_REVERSE;
-   case GL_INVERT:
-      return PIPE_LOGICOP_INVERT;
-   case GL_XOR:
-      return PIPE_LOGICOP_XOR;
-   case GL_NAND:
-      return PIPE_LOGICOP_NAND;
-   case GL_AND:
-      return PIPE_LOGICOP_AND;
-   case GL_EQUIV:
-      return PIPE_LOGICOP_EQUIV;
-   case GL_NOOP:
-      return PIPE_LOGICOP_NOOP;
-   case GL_OR_INVERTED:
-      return PIPE_LOGICOP_OR_INVERTED;
-   case GL_COPY:
-      return PIPE_LOGICOP_COPY;
-   case GL_OR_REVERSE:
-      return PIPE_LOGICOP_OR_REVERSE;
-   case GL_OR:
-      return PIPE_LOGICOP_OR;
-   case GL_SET:
-      return PIPE_LOGICOP_SET;
-   default:
-      assert("invalid GL token in translate_logicop()" == NULL);
-      return 0;
-   }
-}
-
-/**
- * Figure out if colormasks are different per rt.
- */
-static GLboolean
-colormask_per_rt(struct gl_context *ctx)
-{
-   /* a bit suboptimal have to compare lots of values */
-   unsigned i;
-   for (i = 1; i < ctx->Const.MaxDrawBuffers; i++) {
-      if (memcmp(ctx->Color.ColorMask[0], ctx->Color.ColorMask[i], 4)) {
-         return GL_TRUE;
-      }
-   }
-   return GL_FALSE;
-}
-
-/**
- * Figure out if blend enables/state are different per rt.
- */
-static GLboolean
-blend_per_rt(struct gl_context *ctx)
-{
-   if (ctx->Color.BlendEnabled &&
-      (ctx->Color.BlendEnabled != ((1 << ctx->Const.MaxDrawBuffers) - 1))) {
-      /* This can only happen if GL_EXT_draw_buffers2 is enabled */
-      return GL_TRUE;
-   }
-   if (ctx->Color._BlendFuncPerBuffer || ctx->Color._BlendEquationPerBuffer) {
-      /* this can only happen if GL_ARB_draw_buffers_blend is enabled */
-      return GL_TRUE;
-   }
-   return GL_FALSE;
-}
-
-static void 
-update_blend( struct st_context *st )
-{
-   struct pipe_blend_state *blend = &st->state.blend;
-   unsigned num_state = 1;
-   unsigned i, j;
-
-   memset(blend, 0, sizeof(*blend));
-
-   if (blend_per_rt(st->ctx) || colormask_per_rt(st->ctx)) {
-      num_state = st->ctx->Const.MaxDrawBuffers;
-      blend->independent_blend_enable = 1;
-   }
-   /* Note it is impossible to correctly deal with EXT_blend_logic_op and
-      EXT_draw_buffers2/EXT_blend_equation_separate at the same time.
-      These combinations would require support for per-rt logicop enables
-      and separate alpha/rgb logicop/blend support respectively. Neither
-      possible in gallium nor most hardware. Assume these combinations
-      don't happen. */
-   if (st->ctx->Color.ColorLogicOpEnabled ||
-       (st->ctx->Color.BlendEnabled &&
-        st->ctx->Color.Blend[0].EquationRGB == GL_LOGIC_OP)) {
-      /* logicop enabled */
-      blend->logicop_enable = 1;
-      blend->logicop_func = translate_logicop(st->ctx->Color.LogicOp);
-   }
-   else if (st->ctx->Color.BlendEnabled) {
-      /* blending enabled */
-      for (i = 0, j = 0; i < num_state; i++) {
-
-         blend->rt[i].blend_enable = (st->ctx->Color.BlendEnabled >> i) & 0x1;
-
-         if (st->ctx->Extensions.ARB_draw_buffers_blend)
-            j = i;
-
-         blend->rt[i].rgb_func =
-            translate_blend(st->ctx->Color.Blend[j].EquationRGB);
-
-         if (st->ctx->Color.Blend[i].EquationRGB == GL_MIN ||
-             st->ctx->Color.Blend[i].EquationRGB == GL_MAX) {
-            /* Min/max are special */
-            blend->rt[i].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
-            blend->rt[i].rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
-         }
-         else {
-            blend->rt[i].rgb_src_factor =
-               translate_blend(st->ctx->Color.Blend[j].SrcRGB);
-            blend->rt[i].rgb_dst_factor =
-               translate_blend(st->ctx->Color.Blend[j].DstRGB);
-         }
-
-         blend->rt[i].alpha_func =
-            translate_blend(st->ctx->Color.Blend[j].EquationA);
-
-         if (st->ctx->Color.Blend[i].EquationA == GL_MIN ||
-             st->ctx->Color.Blend[i].EquationA == GL_MAX) {
-            /* Min/max are special */
-            blend->rt[i].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
-            blend->rt[i].alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
-         }
-         else {
-            blend->rt[i].alpha_src_factor =
-               translate_blend(st->ctx->Color.Blend[j].SrcA);
-            blend->rt[i].alpha_dst_factor =
-               translate_blend(st->ctx->Color.Blend[j].DstA);
-         }
-      }
-   }
-   else {
-      /* no blending / logicop */
-   }
-
-   /* Colormask - maybe reverse these bits? */
-   for (i = 0; i < num_state; i++) {
-      if (st->ctx->Color.ColorMask[i][0])
-         blend->rt[i].colormask |= PIPE_MASK_R;
-      if (st->ctx->Color.ColorMask[i][1])
-         blend->rt[i].colormask |= PIPE_MASK_G;
-      if (st->ctx->Color.ColorMask[i][2])
-         blend->rt[i].colormask |= PIPE_MASK_B;
-      if (st->ctx->Color.ColorMask[i][3])
-         blend->rt[i].colormask |= PIPE_MASK_A;
-   }
-
-   if (st->ctx->Color.DitherFlag)
-      blend->dither = 1;
-
-   if (st->ctx->Multisample.Enabled) {
-      /* unlike in gallium/d3d10 these operations are only performed
-         if msaa is enabled */
-      if (st->ctx->Multisample.SampleAlphaToCoverage)
-         blend->alpha_to_coverage = 1;
-      if (st->ctx->Multisample.SampleAlphaToOne)
-         blend->alpha_to_one = 1;
-   }
-
-   cso_set_blend(st->cso_context, blend);
-
-   {
-      struct pipe_blend_color bc;
-      COPY_4FV(bc.color, st->ctx->Color.BlendColor);
-      cso_set_blend_color(st->cso_context, &bc);
-   }
-}
-
-
-const struct st_tracked_state st_update_blend = {
-   "st_update_blend",					/* name */
-   {							/* dirty */
-      (_NEW_COLOR | _NEW_MULTISAMPLE),  /* XXX _NEW_BLEND someday? */	/* mesa */
-      0,						/* st */
-   },
-   update_blend,					/* update */
-};
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+ /*
+  * Authors:
+  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Brian Paul
+  */
+ 
+
+#include "st_context.h"
+#include "st_atom.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "cso_cache/cso_context.h"
+
+#include "main/macros.h"
+
+/**
+ * Convert GLenum blend tokens to pipe tokens.
+ * Both blend factors and blend funcs are accepted.
+ */
+static GLuint
+translate_blend(GLenum blend)
+{
+   switch (blend) {
+   /* blend functions */
+   case GL_FUNC_ADD:
+      return PIPE_BLEND_ADD;
+   case GL_FUNC_SUBTRACT:
+      return PIPE_BLEND_SUBTRACT;
+   case GL_FUNC_REVERSE_SUBTRACT:
+      return PIPE_BLEND_REVERSE_SUBTRACT;
+   case GL_MIN:
+      return PIPE_BLEND_MIN;
+   case GL_MAX:
+      return PIPE_BLEND_MAX;
+
+   /* blend factors */
+   case GL_ONE:
+      return PIPE_BLENDFACTOR_ONE;
+   case GL_SRC_COLOR:
+      return PIPE_BLENDFACTOR_SRC_COLOR;
+   case GL_SRC_ALPHA:
+      return PIPE_BLENDFACTOR_SRC_ALPHA;
+   case GL_DST_ALPHA:
+      return PIPE_BLENDFACTOR_DST_ALPHA;
+   case GL_DST_COLOR:
+      return PIPE_BLENDFACTOR_DST_COLOR;
+   case GL_SRC_ALPHA_SATURATE:
+      return PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE;
+   case GL_CONSTANT_COLOR:
+      return PIPE_BLENDFACTOR_CONST_COLOR;
+   case GL_CONSTANT_ALPHA:
+      return PIPE_BLENDFACTOR_CONST_ALPHA;
+      /*
+      return PIPE_BLENDFACTOR_SRC1_COLOR;
+      return PIPE_BLENDFACTOR_SRC1_ALPHA;
+      */
+   case GL_ZERO:
+      return PIPE_BLENDFACTOR_ZERO;
+   case GL_ONE_MINUS_SRC_COLOR:
+      return PIPE_BLENDFACTOR_INV_SRC_COLOR;
+   case GL_ONE_MINUS_SRC_ALPHA:
+      return PIPE_BLENDFACTOR_INV_SRC_ALPHA;
+   case GL_ONE_MINUS_DST_COLOR:
+      return PIPE_BLENDFACTOR_INV_DST_COLOR;
+   case GL_ONE_MINUS_DST_ALPHA:
+      return PIPE_BLENDFACTOR_INV_DST_ALPHA;
+   case GL_ONE_MINUS_CONSTANT_COLOR:
+      return PIPE_BLENDFACTOR_INV_CONST_COLOR;
+   case GL_ONE_MINUS_CONSTANT_ALPHA:
+      return PIPE_BLENDFACTOR_INV_CONST_ALPHA;
+      /*
+      return PIPE_BLENDFACTOR_INV_SRC1_COLOR;
+      return PIPE_BLENDFACTOR_INV_SRC1_ALPHA;
+      */
+   default:
+      assert("invalid GL token in translate_blend()" == NULL);
+      return 0;
+   }
+}
+
+
+/**
+ * Convert GLenum logicop tokens to pipe tokens.
+ */
+static GLuint
+translate_logicop(GLenum logicop)
+{
+   switch (logicop) {
+   case GL_CLEAR:
+      return PIPE_LOGICOP_CLEAR;
+   case GL_NOR:
+      return PIPE_LOGICOP_NOR;
+   case GL_AND_INVERTED:
+      return PIPE_LOGICOP_AND_INVERTED;
+   case GL_COPY_INVERTED:
+      return PIPE_LOGICOP_COPY_INVERTED;
+   case GL_AND_REVERSE:
+      return PIPE_LOGICOP_AND_REVERSE;
+   case GL_INVERT:
+      return PIPE_LOGICOP_INVERT;
+   case GL_XOR:
+      return PIPE_LOGICOP_XOR;
+   case GL_NAND:
+      return PIPE_LOGICOP_NAND;
+   case GL_AND:
+      return PIPE_LOGICOP_AND;
+   case GL_EQUIV:
+      return PIPE_LOGICOP_EQUIV;
+   case GL_NOOP:
+      return PIPE_LOGICOP_NOOP;
+   case GL_OR_INVERTED:
+      return PIPE_LOGICOP_OR_INVERTED;
+   case GL_COPY:
+      return PIPE_LOGICOP_COPY;
+   case GL_OR_REVERSE:
+      return PIPE_LOGICOP_OR_REVERSE;
+   case GL_OR:
+      return PIPE_LOGICOP_OR;
+   case GL_SET:
+      return PIPE_LOGICOP_SET;
+   default:
+      assert("invalid GL token in translate_logicop()" == NULL);
+      return 0;
+   }
+}
+
+/**
+ * Figure out if colormasks are different per rt.
+ */
+static GLboolean
+colormask_per_rt(struct gl_context *ctx)
+{
+   /* a bit suboptimal have to compare lots of values */
+   unsigned i;
+   for (i = 1; i < ctx->Const.MaxDrawBuffers; i++) {
+      if (memcmp(ctx->Color.ColorMask[0], ctx->Color.ColorMask[i], 4)) {
+         return GL_TRUE;
+      }
+   }
+   return GL_FALSE;
+}
+
+/**
+ * Figure out if blend enables/state are different per rt.
+ */
+static GLboolean
+blend_per_rt(struct gl_context *ctx)
+{
+   if (ctx->Color.BlendEnabled &&
+      (ctx->Color.BlendEnabled != ((1 << ctx->Const.MaxDrawBuffers) - 1))) {
+      /* This can only happen if GL_EXT_draw_buffers2 is enabled */
+      return GL_TRUE;
+   }
+   if (ctx->Color._BlendFuncPerBuffer || ctx->Color._BlendEquationPerBuffer) {
+      /* this can only happen if GL_ARB_draw_buffers_blend is enabled */
+      return GL_TRUE;
+   }
+   return GL_FALSE;
+}
+
+static void 
+update_blend( struct st_context *st )
+{
+   struct pipe_blend_state *blend = &st->state.blend;
+   unsigned num_state = 1;
+   unsigned i, j;
+
+   memset(blend, 0, sizeof(*blend));
+
+   if (blend_per_rt(st->ctx) || colormask_per_rt(st->ctx)) {
+      num_state = st->ctx->Const.MaxDrawBuffers;
+      blend->independent_blend_enable = 1;
+   }
+   /* Note it is impossible to correctly deal with EXT_blend_logic_op and
+      EXT_draw_buffers2/EXT_blend_equation_separate at the same time.
+      These combinations would require support for per-rt logicop enables
+      and separate alpha/rgb logicop/blend support respectively. Neither
+      possible in gallium nor most hardware. Assume these combinations
+      don't happen. */
+   if (st->ctx->Color.ColorLogicOpEnabled ||
+       (st->ctx->Color.BlendEnabled &&
+        st->ctx->Color.Blend[0].EquationRGB == GL_LOGIC_OP)) {
+      /* logicop enabled */
+      blend->logicop_enable = 1;
+      blend->logicop_func = translate_logicop(st->ctx->Color.LogicOp);
+   }
+   else if (st->ctx->Color.BlendEnabled) {
+      /* blending enabled */
+      for (i = 0, j = 0; i < num_state; i++) {
+
+         blend->rt[i].blend_enable = (st->ctx->Color.BlendEnabled >> i) & 0x1;
+
+         if (st->ctx->Extensions.ARB_draw_buffers_blend)
+            j = i;
+
+         blend->rt[i].rgb_func =
+            translate_blend(st->ctx->Color.Blend[j].EquationRGB);
+
+         if (st->ctx->Color.Blend[i].EquationRGB == GL_MIN ||
+             st->ctx->Color.Blend[i].EquationRGB == GL_MAX) {
+            /* Min/max are special */
+            blend->rt[i].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+            blend->rt[i].rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
+         }
+         else {
+            blend->rt[i].rgb_src_factor =
+               translate_blend(st->ctx->Color.Blend[j].SrcRGB);
+            blend->rt[i].rgb_dst_factor =
+               translate_blend(st->ctx->Color.Blend[j].DstRGB);
+         }
+
+         blend->rt[i].alpha_func =
+            translate_blend(st->ctx->Color.Blend[j].EquationA);
+
+         if (st->ctx->Color.Blend[i].EquationA == GL_MIN ||
+             st->ctx->Color.Blend[i].EquationA == GL_MAX) {
+            /* Min/max are special */
+            blend->rt[i].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+            blend->rt[i].alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
+         }
+         else {
+            blend->rt[i].alpha_src_factor =
+               translate_blend(st->ctx->Color.Blend[j].SrcA);
+            blend->rt[i].alpha_dst_factor =
+               translate_blend(st->ctx->Color.Blend[j].DstA);
+         }
+      }
+   }
+   else {
+      /* no blending / logicop */
+   }
+
+   /* Colormask - maybe reverse these bits? */
+   for (i = 0; i < num_state; i++) {
+      if (st->ctx->Color.ColorMask[i][0])
+         blend->rt[i].colormask |= PIPE_MASK_R;
+      if (st->ctx->Color.ColorMask[i][1])
+         blend->rt[i].colormask |= PIPE_MASK_G;
+      if (st->ctx->Color.ColorMask[i][2])
+         blend->rt[i].colormask |= PIPE_MASK_B;
+      if (st->ctx->Color.ColorMask[i][3])
+         blend->rt[i].colormask |= PIPE_MASK_A;
+   }
+
+   if (st->ctx->Color.DitherFlag)
+      blend->dither = 1;
+
+   if (st->ctx->Multisample.Enabled) {
+      /* unlike in gallium/d3d10 these operations are only performed
+         if msaa is enabled */
+      if (st->ctx->Multisample.SampleAlphaToCoverage)
+         blend->alpha_to_coverage = 1;
+      if (st->ctx->Multisample.SampleAlphaToOne)
+         blend->alpha_to_one = 1;
+   }
+
+   cso_set_blend(st->cso_context, blend);
+
+   {
+      struct pipe_blend_color bc;
+      COPY_4FV(bc.color, st->ctx->Color.BlendColor);
+      cso_set_blend_color(st->cso_context, &bc);
+   }
+}
+
+
+const struct st_tracked_state st_update_blend = {
+   "st_update_blend",					/* name */
+   {							/* dirty */
+      (_NEW_COLOR | _NEW_MULTISAMPLE),  /* XXX _NEW_BLEND someday? */	/* mesa */
+      0,						/* st */
+   },
+   update_blend,					/* update */
+};
diff --git a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
index 76386fe01..678a270c7 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -1,182 +1,182 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  *   Brian Paul
-  */
- 
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_cb_fbo.h"
-#include "st_texture.h"
-#include "pipe/p_context.h"
-#include "cso_cache/cso_context.h"
-#include "util/u_math.h"
-#include "util/u_inlines.h"
-#include "util/u_format.h"
-
-
-/**
- * When doing GL render to texture, we have to be sure that finalize_texture()
- * didn't yank out the pipe_resource that we earlier created a surface for.
- * Check for that here and create a new surface if needed.
- */
-static void
-update_renderbuffer_surface(struct st_context *st,
-                            struct st_renderbuffer *strb)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_resource *resource = strb->rtt->pt;
-   int rtt_width = strb->Base.Width;
-   int rtt_height = strb->Base.Height;
-   enum pipe_format format = st->ctx->Color.sRGBEnabled ? resource->format : util_format_linear(resource->format);
-
-   if (!strb->surface ||
-       strb->surface->format != format ||
-       strb->surface->texture != resource ||
-       strb->surface->width != rtt_width ||
-       strb->surface->height != rtt_height) {
-      GLuint level;
-      /* find matching mipmap level size */
-      for (level = 0; level <= resource->last_level; level++) {
-         if (u_minify(resource->width0, level) == rtt_width &&
-             u_minify(resource->height0, level) == rtt_height) {
-            struct pipe_surface surf_tmpl;
-            memset(&surf_tmpl, 0, sizeof(surf_tmpl));
-            surf_tmpl.format = format;
-            surf_tmpl.usage = PIPE_BIND_RENDER_TARGET;
-            surf_tmpl.u.tex.level = level;
-            surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice;
-            surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice;
-
-            pipe_surface_reference(&strb->surface, NULL);
-
-            strb->surface = pipe->create_surface(pipe,
-                                                 resource,
-                                                 &surf_tmpl);
-#if 0
-            printf("-- alloc new surface %d x %d into tex %p\n",
-                   strb->surface->width, strb->surface->height,
-                   texture);
-#endif
-            break;
-         }
-      }
-   }
-}
-
-
-/**
- * Update framebuffer state (color, depth, stencil, etc. buffers)
- */
-static void
-update_framebuffer_state( struct st_context *st )
-{
-   struct pipe_framebuffer_state *framebuffer = &st->state.framebuffer;
-   struct gl_framebuffer *fb = st->ctx->DrawBuffer;
-   struct st_renderbuffer *strb;
-   GLuint i;
-
-   framebuffer->width = fb->Width;
-   framebuffer->height = fb->Height;
-
-   /*printf("------ fb size %d x %d\n", fb->Width, fb->Height);*/
-
-   /* Examine Mesa's ctx->DrawBuffer->_ColorDrawBuffers state
-    * to determine which surfaces to draw to
-    */
-   framebuffer->nr_cbufs = 0;
-   for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
-      strb = st_renderbuffer(fb->_ColorDrawBuffers[i]);
-
-      if (strb) {
-         /*printf("--------- framebuffer surface rtt %p\n", strb->rtt);*/
-         if (strb->rtt) {
-            /* rendering to a GL texture, may have to update surface */
-            update_renderbuffer_surface(st, strb);
-         }
-
-         if (strb->surface) {
-            pipe_surface_reference(&framebuffer->cbufs[framebuffer->nr_cbufs],
-                                   strb->surface);
-            framebuffer->nr_cbufs++;
-         }
-         strb->defined = GL_TRUE; /* we'll be drawing something */
-      }
-   }
-   for (i = framebuffer->nr_cbufs; i < PIPE_MAX_COLOR_BUFS; i++) {
-      pipe_surface_reference(&framebuffer->cbufs[i], NULL);
-   }
-
-   /*
-    * Depth/Stencil renderbuffer/surface.
-    */
-   strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
-   if (strb) {
-      strb = st_renderbuffer(strb->Base.Wrapped);
-      if (strb->rtt) {
-         /* rendering to a GL texture, may have to update surface */
-         update_renderbuffer_surface(st, strb);
-      }
-      pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
-   }
-   else {
-      strb = st_renderbuffer(fb->Attachment[BUFFER_STENCIL].Renderbuffer);
-      if (strb) {
-         strb = st_renderbuffer(strb->Base.Wrapped);
-         assert(strb->surface);
-         pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
-      }
-      else
-         pipe_surface_reference(&framebuffer->zsbuf, NULL);
-   }
-
-#ifdef DEBUG
-   /* Make sure the resource binding flags were set properly */
-   for (i = 0; i < framebuffer->nr_cbufs; i++) {
-      assert(framebuffer->cbufs[i]->texture->bind & PIPE_BIND_RENDER_TARGET);
-   }
-   if (framebuffer->zsbuf) {
-      assert(framebuffer->zsbuf->texture->bind & PIPE_BIND_DEPTH_STENCIL);
-   }
-#endif
-
-   cso_set_framebuffer(st->cso_context, framebuffer);
-}
-
-
-const struct st_tracked_state st_update_framebuffer = {
-   "st_update_framebuffer",				/* name */
-   {							/* dirty */
-      _NEW_BUFFERS,					/* mesa */
-      ST_NEW_FRAMEBUFFER,				/* st */
-   },
-   update_framebuffer_state				/* update */
-};
-
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+ /*
+  * Authors:
+  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Brian Paul
+  */
+ 
+#include "st_context.h"
+#include "st_atom.h"
+#include "st_cb_fbo.h"
+#include "st_texture.h"
+#include "pipe/p_context.h"
+#include "cso_cache/cso_context.h"
+#include "util/u_math.h"
+#include "util/u_inlines.h"
+#include "util/u_format.h"
+
+
+/**
+ * When doing GL render to texture, we have to be sure that finalize_texture()
+ * didn't yank out the pipe_resource that we earlier created a surface for.
+ * Check for that here and create a new surface if needed.
+ */
+static void
+update_renderbuffer_surface(struct st_context *st,
+                            struct st_renderbuffer *strb)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_resource *resource = strb->rtt->pt;
+   int rtt_width = strb->Base.Width;
+   int rtt_height = strb->Base.Height;
+   enum pipe_format format = st->ctx->Color.sRGBEnabled ? resource->format : util_format_linear(resource->format);
+
+   if (!strb->surface ||
+       strb->surface->format != format ||
+       strb->surface->texture != resource ||
+       strb->surface->width != rtt_width ||
+       strb->surface->height != rtt_height) {
+      GLuint level;
+      /* find matching mipmap level size */
+      for (level = 0; level <= resource->last_level; level++) {
+         if (u_minify(resource->width0, level) == rtt_width &&
+             u_minify(resource->height0, level) == rtt_height) {
+            struct pipe_surface surf_tmpl;
+            memset(&surf_tmpl, 0, sizeof(surf_tmpl));
+            surf_tmpl.format = format;
+            surf_tmpl.usage = PIPE_BIND_RENDER_TARGET;
+            surf_tmpl.u.tex.level = level;
+            surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice;
+            surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice;
+
+            pipe_surface_reference(&strb->surface, NULL);
+
+            strb->surface = pipe->create_surface(pipe,
+                                                 resource,
+                                                 &surf_tmpl);
+#if 0
+            printf("-- alloc new surface %d x %d into tex %p\n",
+                   strb->surface->width, strb->surface->height,
+                   texture);
+#endif
+            break;
+         }
+      }
+   }
+}
+
+
+/**
+ * Update framebuffer state (color, depth, stencil, etc. buffers)
+ */
+static void
+update_framebuffer_state( struct st_context *st )
+{
+   struct pipe_framebuffer_state *framebuffer = &st->state.framebuffer;
+   struct gl_framebuffer *fb = st->ctx->DrawBuffer;
+   struct st_renderbuffer *strb;
+   GLuint i;
+
+   framebuffer->width = fb->Width;
+   framebuffer->height = fb->Height;
+
+   /*printf("------ fb size %d x %d\n", fb->Width, fb->Height);*/
+
+   /* Examine Mesa's ctx->DrawBuffer->_ColorDrawBuffers state
+    * to determine which surfaces to draw to
+    */
+   framebuffer->nr_cbufs = 0;
+   for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
+      strb = st_renderbuffer(fb->_ColorDrawBuffers[i]);
+
+      if (strb) {
+         /*printf("--------- framebuffer surface rtt %p\n", strb->rtt);*/
+         if (strb->rtt) {
+            /* rendering to a GL texture, may have to update surface */
+            update_renderbuffer_surface(st, strb);
+         }
+
+         if (strb->surface) {
+            pipe_surface_reference(&framebuffer->cbufs[framebuffer->nr_cbufs],
+                                   strb->surface);
+            framebuffer->nr_cbufs++;
+         }
+         strb->defined = GL_TRUE; /* we'll be drawing something */
+      }
+   }
+   for (i = framebuffer->nr_cbufs; i < PIPE_MAX_COLOR_BUFS; i++) {
+      pipe_surface_reference(&framebuffer->cbufs[i], NULL);
+   }
+
+   /*
+    * Depth/Stencil renderbuffer/surface.
+    */
+   strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
+   if (strb) {
+      strb = st_renderbuffer(strb->Base.Wrapped);
+      if (strb->rtt) {
+         /* rendering to a GL texture, may have to update surface */
+         update_renderbuffer_surface(st, strb);
+      }
+      pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
+   }
+   else {
+      strb = st_renderbuffer(fb->Attachment[BUFFER_STENCIL].Renderbuffer);
+      if (strb) {
+         strb = st_renderbuffer(strb->Base.Wrapped);
+         assert(strb->surface);
+         pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
+      }
+      else
+         pipe_surface_reference(&framebuffer->zsbuf, NULL);
+   }
+
+#ifdef DEBUG
+   /* Make sure the resource binding flags were set properly */
+   for (i = 0; i < framebuffer->nr_cbufs; i++) {
+      assert(framebuffer->cbufs[i]->texture->bind & PIPE_BIND_RENDER_TARGET);
+   }
+   if (framebuffer->zsbuf) {
+      assert(framebuffer->zsbuf->texture->bind & PIPE_BIND_DEPTH_STENCIL);
+   }
+#endif
+
+   cso_set_framebuffer(st->cso_context, framebuffer);
+}
+
+
+const struct st_tracked_state st_update_framebuffer = {
+   "st_update_framebuffer",				/* name */
+   {							/* dirty */
+      _NEW_BUFFERS,					/* mesa */
+      ST_NEW_FRAMEBUFFER,				/* st */
+   },
+   update_framebuffer_state				/* update */
+};
+
diff --git a/mesalib/src/mesa/state_tracker/st_cb_bitmap.c b/mesalib/src/mesa/state_tracker/st_cb_bitmap.c
index 0ea567155..51623c179 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_bitmap.c
@@ -1,893 +1,893 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Brian Paul
-  */
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/bufferobj.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-#include "program/program.h"
-#include "program/prog_print.h"
-
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_atom_constbuf.h"
-#include "st_program.h"
-#include "st_cb_bitmap.h"
-#include "st_texture.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_shader_tokens.h"
-#include "util/u_inlines.h"
-#include "util/u_draw_quad.h"
-#include "util/u_simple_shaders.h"
-#include "program/prog_instruction.h"
-#include "cso_cache/cso_context.h"
-
-
-#if FEATURE_drawpix
-
-/**
- * glBitmaps are drawn as textured quads.  The user's bitmap pattern
- * is stored in a texture image.  An alpha8 texture format is used.
- * The fragment shader samples a bit (texel) from the texture, then
- * discards the fragment if the bit is off.
- *
- * Note that we actually store the inverse image of the bitmap to
- * simplify the fragment program.  An "on" bit gets stored as texel=0x0
- * and an "off" bit is stored as texel=0xff.  Then we kill the
- * fragment if the negated texel value is less than zero.
- */
-
-
-/**
- * The bitmap cache attempts to accumulate multiple glBitmap calls in a
- * buffer which is then rendered en mass upon a flush, state change, etc.
- * A wide, short buffer is used to target the common case of a series
- * of glBitmap calls being used to draw text.
- */
-static GLboolean UseBitmapCache = GL_TRUE;
-
-
-#define BITMAP_CACHE_WIDTH  512
-#define BITMAP_CACHE_HEIGHT 32
-
-struct bitmap_cache
-{
-   /** Window pos to render the cached image */
-   GLint xpos, ypos;
-   /** Bounds of region used in window coords */
-   GLint xmin, ymin, xmax, ymax;
-
-   GLfloat color[4];
-
-   /** Bitmap's Z position */
-   GLfloat zpos;
-
-   struct pipe_resource *texture;
-   struct pipe_transfer *trans;
-
-   GLboolean empty;
-
-   /** An I8 texture image: */
-   ubyte *buffer;
-};
-
-
-/** Epsilon for Z comparisons */
-#define Z_EPSILON 1e-06
-
-
-/**
- * Make fragment program for glBitmap:
- *   Sample the texture and kill the fragment if the bit is 0.
- * This program will be combined with the user's fragment program.
- */
-static struct st_fragment_program *
-make_bitmap_fragment_program(struct gl_context *ctx, GLuint samplerIndex)
-{
-   struct st_context *st = st_context(ctx);
-   struct st_fragment_program *stfp;
-   struct gl_program *p;
-   GLuint ic = 0;
-
-   p = ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0);
-   if (!p)
-      return NULL;
-
-   p->NumInstructions = 3;
-
-   p->Instructions = _mesa_alloc_instructions(p->NumInstructions);
-   if (!p->Instructions) {
-      ctx->Driver.DeleteProgram(ctx, p);
-      return NULL;
-   }
-   _mesa_init_instructions(p->Instructions, p->NumInstructions);
-
-   /* TEX tmp0, fragment.texcoord[0], texture[0], 2D; */
-   p->Instructions[ic].Opcode = OPCODE_TEX;
-   p->Instructions[ic].DstReg.File = PROGRAM_TEMPORARY;
-   p->Instructions[ic].DstReg.Index = 0;
-   p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT;
-   p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_TEX0;
-   p->Instructions[ic].TexSrcUnit = samplerIndex;
-   p->Instructions[ic].TexSrcTarget = TEXTURE_2D_INDEX;
-   ic++;
-
-   /* KIL if -tmp0 < 0 # texel=0 -> keep / texel=0 -> discard */
-   p->Instructions[ic].Opcode = OPCODE_KIL;
-   p->Instructions[ic].SrcReg[0].File = PROGRAM_TEMPORARY;
-
-   if (st->bitmap.tex_format == PIPE_FORMAT_L8_UNORM)
-      p->Instructions[ic].SrcReg[0].Swizzle = SWIZZLE_XXXX;
-
-   p->Instructions[ic].SrcReg[0].Index = 0;
-   p->Instructions[ic].SrcReg[0].Negate = NEGATE_XYZW;
-   ic++;
-
-   /* END; */
-   p->Instructions[ic++].Opcode = OPCODE_END;
-
-   assert(ic == p->NumInstructions);
-
-   p->InputsRead = FRAG_BIT_TEX0;
-   p->OutputsWritten = 0x0;
-   p->SamplersUsed = (1 << samplerIndex);
-
-   stfp = (struct st_fragment_program *) p;
-   stfp->Base.UsesKill = GL_TRUE;
-
-   return stfp;
-}
-
-
-static int
-find_free_bit(uint bitfield)
-{
-   int i;
-   for (i = 0; i < 32; i++) {
-      if ((bitfield & (1 << i)) == 0) {
-         return i;
-      }
-   }
-   return -1;
-}
-
-
-/**
- * Combine basic bitmap fragment program with the user-defined program.
- * \param st  current context
- * \param fpIn  the incoming fragment program
- * \param fpOut  the new fragment program which does fragment culling
- * \param bitmap_sampler  sampler number for the bitmap texture
- */
-void
-st_make_bitmap_fragment_program(struct st_context *st,
-                                struct gl_fragment_program *fpIn,
-                                struct gl_fragment_program **fpOut,
-                                GLuint *bitmap_sampler)
-{
-   struct st_fragment_program *bitmap_prog;
-   struct gl_program *newProg;
-   uint sampler;
-
-   /*
-    * Generate new program which is the user-defined program prefixed
-    * with the bitmap sampler/kill instructions.
-    */
-   sampler = find_free_bit(fpIn->Base.SamplersUsed);
-   bitmap_prog = make_bitmap_fragment_program(st->ctx, sampler);
-
-   newProg = _mesa_combine_programs(st->ctx,
-                                    &bitmap_prog->Base.Base,
-                                    &fpIn->Base);
-   /* done with this after combining */
-   st_reference_fragprog(st, &bitmap_prog, NULL);
-
-#if 0
-   {
-      printf("Combined bitmap program:\n");
-      _mesa_print_program(newProg);
-      printf("InputsRead: 0x%x\n", newProg->InputsRead);
-      printf("OutputsWritten: 0x%x\n", newProg->OutputsWritten);
-      _mesa_print_parameter_list(newProg->Parameters);
-   }
-#endif
-
-   /* return results */
-   *fpOut = (struct gl_fragment_program *) newProg;
-   *bitmap_sampler = sampler;
-}
-
-
-/**
- * Copy user-provide bitmap bits into texture buffer, expanding
- * bits into texels.
- * "On" bits will set texels to 0x0.
- * "Off" bits will not modify texels.
- * Note that the image is actually going to be upside down in
- * the texture.  We deal with that with texcoords.
- */
-static void
-unpack_bitmap(struct st_context *st,
-              GLint px, GLint py, GLsizei width, GLsizei height,
-              const struct gl_pixelstore_attrib *unpack,
-              const GLubyte *bitmap,
-              ubyte *destBuffer, uint destStride)
-{
-   destBuffer += py * destStride + px;
-
-   _mesa_expand_bitmap(width, height, unpack, bitmap,
-                       destBuffer, destStride, 0x0);
-}
-
-
-/**
- * Create a texture which represents a bitmap image.
- */
-static struct pipe_resource *
-make_bitmap_texture(struct gl_context *ctx, GLsizei width, GLsizei height,
-                    const struct gl_pixelstore_attrib *unpack,
-                    const GLubyte *bitmap)
-{
-   struct st_context *st = st_context(ctx);
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_transfer *transfer;
-   ubyte *dest;
-   struct pipe_resource *pt;
-
-   /* PBO source... */
-   bitmap = _mesa_map_pbo_source(ctx, unpack, bitmap);
-   if (!bitmap) {
-      return NULL;
-   }
-
-   /**
-    * Create texture to hold bitmap pattern.
-    */
-   pt = st_texture_create(st, st->internal_target, st->bitmap.tex_format,
-                          0, width, height, 1, 1,
-                          PIPE_BIND_SAMPLER_VIEW);
-   if (!pt) {
-      _mesa_unmap_pbo_source(ctx, unpack);
-      return NULL;
-   }
-
-   transfer = pipe_get_transfer(st->pipe, pt, 0, 0,
-                                PIPE_TRANSFER_WRITE,
-                                0, 0, width, height);
-
-   dest = pipe_transfer_map(pipe, transfer);
-
-   /* Put image into texture transfer */
-   memset(dest, 0xff, height * transfer->stride);
-   unpack_bitmap(st, 0, 0, width, height, unpack, bitmap,
-                 dest, transfer->stride);
-
-   _mesa_unmap_pbo_source(ctx, unpack);
-
-   /* Release transfer */
-   pipe_transfer_unmap(pipe, transfer);
-   pipe->transfer_destroy(pipe, transfer);
-
-   return pt;
-}
-
-static GLuint
-setup_bitmap_vertex_data(struct st_context *st, bool normalized,
-                         int x, int y, int width, int height,
-                         float z, const float color[4])
-{
-   struct pipe_context *pipe = st->pipe;
-   const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
-   const GLfloat fb_width = (GLfloat)fb->Width;
-   const GLfloat fb_height = (GLfloat)fb->Height;
-   const GLfloat x0 = (GLfloat)x;
-   const GLfloat x1 = (GLfloat)(x + width);
-   const GLfloat y0 = (GLfloat)y;
-   const GLfloat y1 = (GLfloat)(y + height);
-   GLfloat sLeft = (GLfloat)0.0, sRight = (GLfloat)1.0;
-   GLfloat tTop = (GLfloat)0.0, tBot = (GLfloat)1.0 - tTop;
-   const GLfloat clip_x0 = (GLfloat)(x0 / fb_width * 2.0 - 1.0);
-   const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
-   const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
-   const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
-   const GLuint max_slots = 1; /* 4096 / sizeof(st->bitmap.vertices); */
-   GLuint i;
-
-   if(!normalized)
-   {
-      sRight = width;
-      tBot = height;
-   }
-
-   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
-    * no_flush) updates to buffers where we know there is no conflict
-    * with previous data.  Currently using max_slots > 1 will cause
-    * synchronous rendering if the driver flushes its command buffers
-    * between one bitmap and the next.  Our flush hook below isn't
-    * sufficient to catch this as the driver doesn't tell us when it
-    * flushes its own command buffers.  Until this gets fixed, pay the
-    * price of allocating a new buffer for each bitmap cache-flush to
-    * avoid synchronous rendering.
-    */
-   if (st->bitmap.vbuf_slot >= max_slots) {
-      pipe_resource_reference(&st->bitmap.vbuf, NULL);
-      st->bitmap.vbuf_slot = 0;
-   }
-
-   if (!st->bitmap.vbuf) {
-      st->bitmap.vbuf = pipe_buffer_create(pipe->screen, 
-                                           PIPE_BIND_VERTEX_BUFFER,
-                                           PIPE_USAGE_STREAM,
-                                           max_slots *
-                                           sizeof(st->bitmap.vertices));
-   }
-
-   /* Positions are in clip coords since we need to do clipping in case
-    * the bitmap quad goes beyond the window bounds.
-    */
-   st->bitmap.vertices[0][0][0] = clip_x0;
-   st->bitmap.vertices[0][0][1] = clip_y0;
-   st->bitmap.vertices[0][2][0] = sLeft;
-   st->bitmap.vertices[0][2][1] = tTop;
-
-   st->bitmap.vertices[1][0][0] = clip_x1;
-   st->bitmap.vertices[1][0][1] = clip_y0;
-   st->bitmap.vertices[1][2][0] = sRight;
-   st->bitmap.vertices[1][2][1] = tTop;
-   
-   st->bitmap.vertices[2][0][0] = clip_x1;
-   st->bitmap.vertices[2][0][1] = clip_y1;
-   st->bitmap.vertices[2][2][0] = sRight;
-   st->bitmap.vertices[2][2][1] = tBot;
-   
-   st->bitmap.vertices[3][0][0] = clip_x0;
-   st->bitmap.vertices[3][0][1] = clip_y1;
-   st->bitmap.vertices[3][2][0] = sLeft;
-   st->bitmap.vertices[3][2][1] = tBot;
-   
-   /* same for all verts: */
-   for (i = 0; i < 4; i++) {
-      st->bitmap.vertices[i][0][2] = z;
-      st->bitmap.vertices[i][0][3] = 1.0;
-      st->bitmap.vertices[i][1][0] = color[0];
-      st->bitmap.vertices[i][1][1] = color[1];
-      st->bitmap.vertices[i][1][2] = color[2];
-      st->bitmap.vertices[i][1][3] = color[3];
-      st->bitmap.vertices[i][2][2] = 0.0; /*R*/
-      st->bitmap.vertices[i][2][3] = 1.0; /*Q*/
-   }
-
-   /* put vertex data into vbuf */
-   pipe_buffer_write_nooverlap(st->pipe,
-                               st->bitmap.vbuf,
-                               st->bitmap.vbuf_slot
-                               * sizeof(st->bitmap.vertices),
-                               sizeof st->bitmap.vertices,
-                               st->bitmap.vertices);
-
-   return st->bitmap.vbuf_slot++ * sizeof st->bitmap.vertices;
-}
-
-
-
-/**
- * Render a glBitmap by drawing a textured quad
- */
-static void
-draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
-                 GLsizei width, GLsizei height,
-                 struct pipe_sampler_view *sv,
-                 const GLfloat *color)
-{
-   struct st_context *st = st_context(ctx);
-   struct pipe_context *pipe = st->pipe;
-   struct cso_context *cso = st->cso_context;
-   struct st_fp_variant *fpv;
-   struct st_fp_variant_key key;
-   GLuint maxSize;
-   GLuint offset;
-
-   memset(&key, 0, sizeof(key));
-   key.st = st;
-   key.bitmap = GL_TRUE;
-
-   fpv = st_get_fp_variant(st, st->fp, &key);
-
-   /* As an optimization, Mesa's fragment programs will sometimes get the
-    * primary color from a statevar/constant rather than a varying variable.
-    * when that's the case, we need to ensure that we use the 'color'
-    * parameter and not the current attribute color (which may have changed
-    * through glRasterPos and state validation.
-    * So, we force the proper color here.  Not elegant, but it works.
-    */
-   {
-      GLfloat colorSave[4];
-      COPY_4V(colorSave, ctx->Current.Attrib[VERT_ATTRIB_COLOR0]);
-      COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], color);
-      st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT);
-      COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], colorSave);
-   }
-
-
-   /* limit checks */
-   /* XXX if the bitmap is larger than the max texture size, break
-    * it up into chunks.
-    */
-   maxSize = 1 << (pipe->screen->get_param(pipe->screen,
-                                    PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1);
-   assert(width <= (GLsizei)maxSize);
-   assert(height <= (GLsizei)maxSize);
-
-   cso_save_rasterizer(cso);
-   cso_save_samplers(cso);
-   cso_save_fragment_sampler_views(cso);
-   cso_save_viewport(cso);
-   cso_save_fragment_shader(cso);
-   cso_save_vertex_shader(cso);
-   cso_save_vertex_elements(cso);
-   cso_save_vertex_buffers(cso);
-
-   /* rasterizer state: just scissor */
-   st->bitmap.rasterizer.scissor = ctx->Scissor.Enabled;
-   cso_set_rasterizer(cso, &st->bitmap.rasterizer);
-
-   /* fragment shader state: TEX lookup program */
-   cso_set_fragment_shader_handle(cso, fpv->driver_shader);
-
-   /* vertex shader state: position + texcoord pass-through */
-   cso_set_vertex_shader_handle(cso, st->bitmap.vs);
-
-   /* user samplers, plus our bitmap sampler */
-   {
-      struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS];
-      uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_samplers);
-      uint i;
-      for (i = 0; i < st->state.num_samplers; i++) {
-         samplers[i] = &st->state.samplers[i];
-      }
-      samplers[fpv->bitmap_sampler] =
-         &st->bitmap.samplers[sv->texture->target != PIPE_TEXTURE_RECT];
-      cso_set_samplers(cso, num, (const struct pipe_sampler_state **) samplers);
-   }
-
-   /* user textures, plus the bitmap texture */
-   {
-      struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS];
-      uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_textures);
-      memcpy(sampler_views, st->state.sampler_views, sizeof(sampler_views));
-      sampler_views[fpv->bitmap_sampler] = sv;
-      cso_set_fragment_sampler_views(cso, num, sampler_views);
-   }
-
-   /* viewport state: viewport matching window dims */
-   {
-      const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
-      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
-      const GLfloat width = (GLfloat)fb->Width;
-      const GLfloat height = (GLfloat)fb->Height;
-      struct pipe_viewport_state vp;
-      vp.scale[0] =  0.5f * width;
-      vp.scale[1] = height * (invert ? -0.5f : 0.5f);
-      vp.scale[2] = 0.5f;
-      vp.scale[3] = 1.0f;
-      vp.translate[0] = 0.5f * width;
-      vp.translate[1] = 0.5f * height;
-      vp.translate[2] = 0.5f;
-      vp.translate[3] = 0.0f;
-      cso_set_viewport(cso, &vp);
-   }
-
-   cso_set_vertex_elements(cso, 3, st->velems_util_draw);
-
-   /* convert Z from [0,1] to [-1,-1] to match viewport Z scale/bias */
-   z = z * 2.0 - 1.0;
-
-   /* draw textured quad */
-   offset = setup_bitmap_vertex_data(st,
-                                     sv->texture->target != PIPE_TEXTURE_RECT,
-                                     x, y, width, height, z, color);
-
-   util_draw_vertex_buffer(pipe, st->cso_context, st->bitmap.vbuf, offset,
-                           PIPE_PRIM_TRIANGLE_FAN,
-                           4,  /* verts */
-                           3); /* attribs/vert */
-
-
-   /* restore state */
-   cso_restore_rasterizer(cso);
-   cso_restore_samplers(cso);
-   cso_restore_fragment_sampler_views(cso);
-   cso_restore_viewport(cso);
-   cso_restore_fragment_shader(cso);
-   cso_restore_vertex_shader(cso);
-   cso_restore_vertex_elements(cso);
-   cso_restore_vertex_buffers(cso);
-}
-
-
-static void
-reset_cache(struct st_context *st)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct bitmap_cache *cache = st->bitmap.cache;
-
-   /*memset(cache->buffer, 0xff, sizeof(cache->buffer));*/
-   cache->empty = GL_TRUE;
-
-   cache->xmin = 1000000;
-   cache->xmax = -1000000;
-   cache->ymin = 1000000;
-   cache->ymax = -1000000;
-
-   if (cache->trans) {
-      pipe->transfer_destroy(pipe, cache->trans);
-      cache->trans = NULL;
-   }
-
-   assert(!cache->texture);
-
-   /* allocate a new texture */
-   cache->texture = st_texture_create(st, PIPE_TEXTURE_2D,
-                                      st->bitmap.tex_format, 0,
-                                      BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT,
-                                      1, 1,
-				      PIPE_BIND_SAMPLER_VIEW);
-}
-
-
-/** Print bitmap image to stdout (debug) */
-static void
-print_cache(const struct bitmap_cache *cache)
-{
-   int i, j, k;
-
-   for (i = 0; i < BITMAP_CACHE_HEIGHT; i++) {
-      k = BITMAP_CACHE_WIDTH * (BITMAP_CACHE_HEIGHT - i - 1);
-      for (j = 0; j < BITMAP_CACHE_WIDTH; j++) {
-         if (cache->buffer[k])
-            printf("X");
-         else
-            printf(" ");
-         k++;
-      }
-      printf("\n");
-   }
-}
-
-
-/**
- * Create gallium pipe_transfer object for the bitmap cache.
- */
-static void
-create_cache_trans(struct st_context *st)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct bitmap_cache *cache = st->bitmap.cache;
-
-   if (cache->trans)
-      return;
-
-   /* Map the texture transfer.
-    * Subsequent glBitmap calls will write into the texture image.
-    */
-   cache->trans = pipe_get_transfer(st->pipe, cache->texture, 0, 0,
-                                    PIPE_TRANSFER_WRITE, 0, 0,
-                                    BITMAP_CACHE_WIDTH,
-                                    BITMAP_CACHE_HEIGHT);
-   cache->buffer = pipe_transfer_map(pipe, cache->trans);
-
-   /* init image to all 0xff */
-   memset(cache->buffer, 0xff, cache->trans->stride * BITMAP_CACHE_HEIGHT);
-}
-
-
-/**
- * If there's anything in the bitmap cache, draw/flush it now.
- */
-void
-st_flush_bitmap_cache(struct st_context *st)
-{
-   if (!st->bitmap.cache->empty) {
-      struct bitmap_cache *cache = st->bitmap.cache;
-
-      if (st->ctx->DrawBuffer) {
-         struct pipe_context *pipe = st->pipe;
-         struct pipe_sampler_view *sv;
-
-         assert(cache->xmin <= cache->xmax);
- 
-/*         printf("flush size %d x %d  at %d, %d\n",
-                cache->xmax - cache->xmin,
-                cache->ymax - cache->ymin,
-                cache->xpos, cache->ypos);
-*/
-
-         /* The texture transfer has been mapped until now.
-          * So unmap and release the texture transfer before drawing.
-          */
-         if (cache->trans) {
-            if (0)
-               print_cache(cache);
-            pipe_transfer_unmap(pipe, cache->trans);
-            cache->buffer = NULL;
-
-            pipe->transfer_destroy(pipe, cache->trans);
-            cache->trans = NULL;
-         }
-
-         sv = st_create_texture_sampler_view(st->pipe, cache->texture);
-         if (sv) {
-            draw_bitmap_quad(st->ctx,
-                             cache->xpos,
-                             cache->ypos,
-                             cache->zpos,
-                             BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT,
-                             sv,
-                             cache->color);
-
-            pipe_sampler_view_reference(&sv, NULL);
-         }
-      }
-
-      /* release/free the texture */
-      pipe_resource_reference(&cache->texture, NULL);
-
-      reset_cache(st);
-   }
-}
-
-
-/**
- * Flush bitmap cache and release vertex buffer.
- */
-void
-st_flush_bitmap( struct st_context *st )
-{
-   st_flush_bitmap_cache(st);
-
-   /* Release vertex buffer to avoid synchronous rendering if we were
-    * to map it in the next frame.
-    */
-   pipe_resource_reference(&st->bitmap.vbuf, NULL);
-   st->bitmap.vbuf_slot = 0;
-}
-
-
-/**
- * Try to accumulate this glBitmap call in the bitmap cache.
- * \return  GL_TRUE for success, GL_FALSE if bitmap is too large, etc.
- */
-static GLboolean
-accum_bitmap(struct st_context *st,
-             GLint x, GLint y, GLsizei width, GLsizei height,
-             const struct gl_pixelstore_attrib *unpack,
-             const GLubyte *bitmap )
-{
-   struct bitmap_cache *cache = st->bitmap.cache;
-   int px = -999, py = -999;
-   const GLfloat z = st->ctx->Current.RasterPos[2];
-
-   if (width > BITMAP_CACHE_WIDTH ||
-       height > BITMAP_CACHE_HEIGHT)
-      return GL_FALSE; /* too big to cache */
-
-   if (!cache->empty) {
-      px = x - cache->xpos;  /* pos in buffer */
-      py = y - cache->ypos;
-      if (px < 0 || px + width > BITMAP_CACHE_WIDTH ||
-          py < 0 || py + height > BITMAP_CACHE_HEIGHT ||
-          !TEST_EQ_4V(st->ctx->Current.RasterColor, cache->color) ||
-          ((fabs(z - cache->zpos) > Z_EPSILON))) {
-         /* This bitmap would extend beyond cache bounds, or the bitmap
-          * color is changing
-          * so flush and continue.
-          */
-         st_flush_bitmap_cache(st);
-      }
-   }
-
-   if (cache->empty) {
-      /* Initialize.  Center bitmap vertically in the buffer. */
-      px = 0;
-      py = (BITMAP_CACHE_HEIGHT - height) / 2;
-      cache->xpos = x;
-      cache->ypos = y - py;
-      cache->zpos = z;
-      cache->empty = GL_FALSE;
-      COPY_4FV(cache->color, st->ctx->Current.RasterColor);
-   }
-
-   assert(px != -999);
-   assert(py != -999);
-
-   if (x < cache->xmin)
-      cache->xmin = x;
-   if (y < cache->ymin)
-      cache->ymin = y;
-   if (x + width > cache->xmax)
-      cache->xmax = x + width;
-   if (y + height > cache->ymax)
-      cache->ymax = y + height;
-
-   /* create the transfer if needed */
-   create_cache_trans(st);
-
-   unpack_bitmap(st, px, py, width, height, unpack, bitmap,
-                 cache->buffer, BITMAP_CACHE_WIDTH);
-
-   return GL_TRUE; /* accumulated */
-}
-
-
-
-/**
- * Called via ctx->Driver.Bitmap()
- */
-static void
-st_Bitmap(struct gl_context *ctx, GLint x, GLint y,
-          GLsizei width, GLsizei height,
-          const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap )
-{
-   struct st_context *st = st_context(ctx);
-   struct pipe_resource *pt;
-
-   if (width == 0 || height == 0)
-      return;
-
-   st_validate_state(st);
-
-   if (!st->bitmap.vs) {
-      /* create pass-through vertex shader now */
-      const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
-                                      TGSI_SEMANTIC_COLOR,
-                                      TGSI_SEMANTIC_GENERIC };
-      const uint semantic_indexes[] = { 0, 0, 0 };
-      st->bitmap.vs = util_make_vertex_passthrough_shader(st->pipe, 3,
-                                                          semantic_names,
-                                                          semantic_indexes);
-   }
-
-   if (UseBitmapCache && accum_bitmap(st, x, y, width, height, unpack, bitmap))
-      return;
-
-   pt = make_bitmap_texture(ctx, width, height, unpack, bitmap);
-   if (pt) {
-      struct pipe_sampler_view *sv =
-         st_create_texture_sampler_view(st->pipe, pt);
-
-      assert(pt->target == PIPE_TEXTURE_2D || pt->target == PIPE_TEXTURE_RECT);
-
-      if (sv) {
-         draw_bitmap_quad(ctx, x, y, ctx->Current.RasterPos[2],
-                          width, height, sv,
-                          st->ctx->Current.RasterColor);
-
-         pipe_sampler_view_reference(&sv, NULL);
-      }
-
-      /* release/free the texture */
-      pipe_resource_reference(&pt, NULL);
-   }
-}
-
-
-/** Per-context init */
-void
-st_init_bitmap_functions(struct dd_function_table *functions)
-{
-   functions->Bitmap = st_Bitmap;
-}
-
-
-/** Per-context init */
-void
-st_init_bitmap(struct st_context *st)
-{
-   struct pipe_sampler_state *sampler = &st->bitmap.samplers[0];
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_screen *screen = pipe->screen;
-
-   /* init sampler state once */
-   memset(sampler, 0, sizeof(*sampler));
-   sampler->wrap_s = PIPE_TEX_WRAP_CLAMP;
-   sampler->wrap_t = PIPE_TEX_WRAP_CLAMP;
-   sampler->wrap_r = PIPE_TEX_WRAP_CLAMP;
-   sampler->min_img_filter = PIPE_TEX_FILTER_NEAREST;
-   sampler->min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
-   sampler->mag_img_filter = PIPE_TEX_FILTER_NEAREST;
-   st->bitmap.samplers[1] = *sampler;
-   st->bitmap.samplers[1].normalized_coords = 1;
-
-   /* init baseline rasterizer state once */
-   memset(&st->bitmap.rasterizer, 0, sizeof(st->bitmap.rasterizer));
-   st->bitmap.rasterizer.gl_rasterization_rules = 1;
-
-   /* find a usable texture format */
-   if (screen->is_format_supported(screen, PIPE_FORMAT_I8_UNORM,
-                                   PIPE_TEXTURE_2D, 0,
-                                   PIPE_BIND_SAMPLER_VIEW, 0)) {
-      st->bitmap.tex_format = PIPE_FORMAT_I8_UNORM;
-   }
-   else if (screen->is_format_supported(screen, PIPE_FORMAT_A8_UNORM,
-                                        PIPE_TEXTURE_2D, 0,
-                                        PIPE_BIND_SAMPLER_VIEW, 0)) {
-      st->bitmap.tex_format = PIPE_FORMAT_A8_UNORM;
-   }
-   else if (screen->is_format_supported(screen, PIPE_FORMAT_L8_UNORM,
-                                        PIPE_TEXTURE_2D, 0,
-                                        PIPE_BIND_SAMPLER_VIEW, 0)) {
-      st->bitmap.tex_format = PIPE_FORMAT_L8_UNORM;
-   }
-   else {
-      /* XXX support more formats */
-      assert(0);
-   }
-
-   /* alloc bitmap cache object */
-   st->bitmap.cache = ST_CALLOC_STRUCT(bitmap_cache);
-
-   reset_cache(st);
-}
-
-
-/** Per-context tear-down */
-void
-st_destroy_bitmap(struct st_context *st)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct bitmap_cache *cache = st->bitmap.cache;
-
-   if (st->bitmap.vs) {
-      cso_delete_vertex_shader(st->cso_context, st->bitmap.vs);
-      st->bitmap.vs = NULL;
-   }
-
-   if (st->bitmap.vbuf) {
-      pipe_resource_reference(&st->bitmap.vbuf, NULL);
-      st->bitmap.vbuf = NULL;
-   }
-
-   if (cache) {
-      if (cache->trans) {
-         pipe_transfer_unmap(pipe, cache->trans);
-         pipe->transfer_destroy(pipe, cache->trans);
-      }
-      pipe_resource_reference(&st->bitmap.cache->texture, NULL);
-      free(st->bitmap.cache);
-      st->bitmap.cache = NULL;
-   }
-}
-
-#endif /* FEATURE_drawpix */
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+ /*
+  * Authors:
+  *   Brian Paul
+  */
+
+#include "main/imports.h"
+#include "main/image.h"
+#include "main/bufferobj.h"
+#include "main/macros.h"
+#include "main/mfeatures.h"
+#include "program/program.h"
+#include "program/prog_print.h"
+
+#include "st_context.h"
+#include "st_atom.h"
+#include "st_atom_constbuf.h"
+#include "st_program.h"
+#include "st_cb_bitmap.h"
+#include "st_texture.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+#include "util/u_inlines.h"
+#include "util/u_draw_quad.h"
+#include "util/u_simple_shaders.h"
+#include "program/prog_instruction.h"
+#include "cso_cache/cso_context.h"
+
+
+#if FEATURE_drawpix
+
+/**
+ * glBitmaps are drawn as textured quads.  The user's bitmap pattern
+ * is stored in a texture image.  An alpha8 texture format is used.
+ * The fragment shader samples a bit (texel) from the texture, then
+ * discards the fragment if the bit is off.
+ *
+ * Note that we actually store the inverse image of the bitmap to
+ * simplify the fragment program.  An "on" bit gets stored as texel=0x0
+ * and an "off" bit is stored as texel=0xff.  Then we kill the
+ * fragment if the negated texel value is less than zero.
+ */
+
+
+/**
+ * The bitmap cache attempts to accumulate multiple glBitmap calls in a
+ * buffer which is then rendered en mass upon a flush, state change, etc.
+ * A wide, short buffer is used to target the common case of a series
+ * of glBitmap calls being used to draw text.
+ */
+static GLboolean UseBitmapCache = GL_TRUE;
+
+
+#define BITMAP_CACHE_WIDTH  512
+#define BITMAP_CACHE_HEIGHT 32
+
+struct bitmap_cache
+{
+   /** Window pos to render the cached image */
+   GLint xpos, ypos;
+   /** Bounds of region used in window coords */
+   GLint xmin, ymin, xmax, ymax;
+
+   GLfloat color[4];
+
+   /** Bitmap's Z position */
+   GLfloat zpos;
+
+   struct pipe_resource *texture;
+   struct pipe_transfer *trans;
+
+   GLboolean empty;
+
+   /** An I8 texture image: */
+   ubyte *buffer;
+};
+
+
+/** Epsilon for Z comparisons */
+#define Z_EPSILON 1e-06
+
+
+/**
+ * Make fragment program for glBitmap:
+ *   Sample the texture and kill the fragment if the bit is 0.
+ * This program will be combined with the user's fragment program.
+ */
+static struct st_fragment_program *
+make_bitmap_fragment_program(struct gl_context *ctx, GLuint samplerIndex)
+{
+   struct st_context *st = st_context(ctx);
+   struct st_fragment_program *stfp;
+   struct gl_program *p;
+   GLuint ic = 0;
+
+   p = ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0);
+   if (!p)
+      return NULL;
+
+   p->NumInstructions = 3;
+
+   p->Instructions = _mesa_alloc_instructions(p->NumInstructions);
+   if (!p->Instructions) {
+      ctx->Driver.DeleteProgram(ctx, p);
+      return NULL;
+   }
+   _mesa_init_instructions(p->Instructions, p->NumInstructions);
+
+   /* TEX tmp0, fragment.texcoord[0], texture[0], 2D; */
+   p->Instructions[ic].Opcode = OPCODE_TEX;
+   p->Instructions[ic].DstReg.File = PROGRAM_TEMPORARY;
+   p->Instructions[ic].DstReg.Index = 0;
+   p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT;
+   p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_TEX0;
+   p->Instructions[ic].TexSrcUnit = samplerIndex;
+   p->Instructions[ic].TexSrcTarget = TEXTURE_2D_INDEX;
+   ic++;
+
+   /* KIL if -tmp0 < 0 # texel=0 -> keep / texel=0 -> discard */
+   p->Instructions[ic].Opcode = OPCODE_KIL;
+   p->Instructions[ic].SrcReg[0].File = PROGRAM_TEMPORARY;
+
+   if (st->bitmap.tex_format == PIPE_FORMAT_L8_UNORM)
+      p->Instructions[ic].SrcReg[0].Swizzle = SWIZZLE_XXXX;
+
+   p->Instructions[ic].SrcReg[0].Index = 0;
+   p->Instructions[ic].SrcReg[0].Negate = NEGATE_XYZW;
+   ic++;
+
+   /* END; */
+   p->Instructions[ic++].Opcode = OPCODE_END;
+
+   assert(ic == p->NumInstructions);
+
+   p->InputsRead = FRAG_BIT_TEX0;
+   p->OutputsWritten = 0x0;
+   p->SamplersUsed = (1 << samplerIndex);
+
+   stfp = (struct st_fragment_program *) p;
+   stfp->Base.UsesKill = GL_TRUE;
+
+   return stfp;
+}
+
+
+static int
+find_free_bit(uint bitfield)
+{
+   int i;
+   for (i = 0; i < 32; i++) {
+      if ((bitfield & (1 << i)) == 0) {
+         return i;
+      }
+   }
+   return -1;
+}
+
+
+/**
+ * Combine basic bitmap fragment program with the user-defined program.
+ * \param st  current context
+ * \param fpIn  the incoming fragment program
+ * \param fpOut  the new fragment program which does fragment culling
+ * \param bitmap_sampler  sampler number for the bitmap texture
+ */
+void
+st_make_bitmap_fragment_program(struct st_context *st,
+                                struct gl_fragment_program *fpIn,
+                                struct gl_fragment_program **fpOut,
+                                GLuint *bitmap_sampler)
+{
+   struct st_fragment_program *bitmap_prog;
+   struct gl_program *newProg;
+   uint sampler;
+
+   /*
+    * Generate new program which is the user-defined program prefixed
+    * with the bitmap sampler/kill instructions.
+    */
+   sampler = find_free_bit(fpIn->Base.SamplersUsed);
+   bitmap_prog = make_bitmap_fragment_program(st->ctx, sampler);
+
+   newProg = _mesa_combine_programs(st->ctx,
+                                    &bitmap_prog->Base.Base,
+                                    &fpIn->Base);
+   /* done with this after combining */
+   st_reference_fragprog(st, &bitmap_prog, NULL);
+
+#if 0
+   {
+      printf("Combined bitmap program:\n");
+      _mesa_print_program(newProg);
+      printf("InputsRead: 0x%x\n", newProg->InputsRead);
+      printf("OutputsWritten: 0x%x\n", newProg->OutputsWritten);
+      _mesa_print_parameter_list(newProg->Parameters);
+   }
+#endif
+
+   /* return results */
+   *fpOut = (struct gl_fragment_program *) newProg;
+   *bitmap_sampler = sampler;
+}
+
+
+/**
+ * Copy user-provide bitmap bits into texture buffer, expanding
+ * bits into texels.
+ * "On" bits will set texels to 0x0.
+ * "Off" bits will not modify texels.
+ * Note that the image is actually going to be upside down in
+ * the texture.  We deal with that with texcoords.
+ */
+static void
+unpack_bitmap(struct st_context *st,
+              GLint px, GLint py, GLsizei width, GLsizei height,
+              const struct gl_pixelstore_attrib *unpack,
+              const GLubyte *bitmap,
+              ubyte *destBuffer, uint destStride)
+{
+   destBuffer += py * destStride + px;
+
+   _mesa_expand_bitmap(width, height, unpack, bitmap,
+                       destBuffer, destStride, 0x0);
+}
+
+
+/**
+ * Create a texture which represents a bitmap image.
+ */
+static struct pipe_resource *
+make_bitmap_texture(struct gl_context *ctx, GLsizei width, GLsizei height,
+                    const struct gl_pixelstore_attrib *unpack,
+                    const GLubyte *bitmap)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_transfer *transfer;
+   ubyte *dest;
+   struct pipe_resource *pt;
+
+   /* PBO source... */
+   bitmap = _mesa_map_pbo_source(ctx, unpack, bitmap);
+   if (!bitmap) {
+      return NULL;
+   }
+
+   /**
+    * Create texture to hold bitmap pattern.
+    */
+   pt = st_texture_create(st, st->internal_target, st->bitmap.tex_format,
+                          0, width, height, 1, 1,
+                          PIPE_BIND_SAMPLER_VIEW);
+   if (!pt) {
+      _mesa_unmap_pbo_source(ctx, unpack);
+      return NULL;
+   }
+
+   transfer = pipe_get_transfer(st->pipe, pt, 0, 0,
+                                PIPE_TRANSFER_WRITE,
+                                0, 0, width, height);
+
+   dest = pipe_transfer_map(pipe, transfer);
+
+   /* Put image into texture transfer */
+   memset(dest, 0xff, height * transfer->stride);
+   unpack_bitmap(st, 0, 0, width, height, unpack, bitmap,
+                 dest, transfer->stride);
+
+   _mesa_unmap_pbo_source(ctx, unpack);
+
+   /* Release transfer */
+   pipe_transfer_unmap(pipe, transfer);
+   pipe->transfer_destroy(pipe, transfer);
+
+   return pt;
+}
+
+static GLuint
+setup_bitmap_vertex_data(struct st_context *st, bool normalized,
+                         int x, int y, int width, int height,
+                         float z, const float color[4])
+{
+   struct pipe_context *pipe = st->pipe;
+   const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
+   const GLfloat fb_width = (GLfloat)fb->Width;
+   const GLfloat fb_height = (GLfloat)fb->Height;
+   const GLfloat x0 = (GLfloat)x;
+   const GLfloat x1 = (GLfloat)(x + width);
+   const GLfloat y0 = (GLfloat)y;
+   const GLfloat y1 = (GLfloat)(y + height);
+   GLfloat sLeft = (GLfloat)0.0, sRight = (GLfloat)1.0;
+   GLfloat tTop = (GLfloat)0.0, tBot = (GLfloat)1.0 - tTop;
+   const GLfloat clip_x0 = (GLfloat)(x0 / fb_width * 2.0 - 1.0);
+   const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
+   const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
+   const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
+   const GLuint max_slots = 1; /* 4096 / sizeof(st->bitmap.vertices); */
+   GLuint i;
+
+   if(!normalized)
+   {
+      sRight = width;
+      tBot = height;
+   }
+
+   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
+    * no_flush) updates to buffers where we know there is no conflict
+    * with previous data.  Currently using max_slots > 1 will cause
+    * synchronous rendering if the driver flushes its command buffers
+    * between one bitmap and the next.  Our flush hook below isn't
+    * sufficient to catch this as the driver doesn't tell us when it
+    * flushes its own command buffers.  Until this gets fixed, pay the
+    * price of allocating a new buffer for each bitmap cache-flush to
+    * avoid synchronous rendering.
+    */
+   if (st->bitmap.vbuf_slot >= max_slots) {
+      pipe_resource_reference(&st->bitmap.vbuf, NULL);
+      st->bitmap.vbuf_slot = 0;
+   }
+
+   if (!st->bitmap.vbuf) {
+      st->bitmap.vbuf = pipe_buffer_create(pipe->screen, 
+                                           PIPE_BIND_VERTEX_BUFFER,
+                                           PIPE_USAGE_STREAM,
+                                           max_slots *
+                                           sizeof(st->bitmap.vertices));
+   }
+
+   /* Positions are in clip coords since we need to do clipping in case
+    * the bitmap quad goes beyond the window bounds.
+    */
+   st->bitmap.vertices[0][0][0] = clip_x0;
+   st->bitmap.vertices[0][0][1] = clip_y0;
+   st->bitmap.vertices[0][2][0] = sLeft;
+   st->bitmap.vertices[0][2][1] = tTop;
+
+   st->bitmap.vertices[1][0][0] = clip_x1;
+   st->bitmap.vertices[1][0][1] = clip_y0;
+   st->bitmap.vertices[1][2][0] = sRight;
+   st->bitmap.vertices[1][2][1] = tTop;
+   
+   st->bitmap.vertices[2][0][0] = clip_x1;
+   st->bitmap.vertices[2][0][1] = clip_y1;
+   st->bitmap.vertices[2][2][0] = sRight;
+   st->bitmap.vertices[2][2][1] = tBot;
+   
+   st->bitmap.vertices[3][0][0] = clip_x0;
+   st->bitmap.vertices[3][0][1] = clip_y1;
+   st->bitmap.vertices[3][2][0] = sLeft;
+   st->bitmap.vertices[3][2][1] = tBot;
+   
+   /* same for all verts: */
+   for (i = 0; i < 4; i++) {
+      st->bitmap.vertices[i][0][2] = z;
+      st->bitmap.vertices[i][0][3] = 1.0;
+      st->bitmap.vertices[i][1][0] = color[0];
+      st->bitmap.vertices[i][1][1] = color[1];
+      st->bitmap.vertices[i][1][2] = color[2];
+      st->bitmap.vertices[i][1][3] = color[3];
+      st->bitmap.vertices[i][2][2] = 0.0; /*R*/
+      st->bitmap.vertices[i][2][3] = 1.0; /*Q*/
+   }
+
+   /* put vertex data into vbuf */
+   pipe_buffer_write_nooverlap(st->pipe,
+                               st->bitmap.vbuf,
+                               st->bitmap.vbuf_slot
+                               * sizeof(st->bitmap.vertices),
+                               sizeof st->bitmap.vertices,
+                               st->bitmap.vertices);
+
+   return st->bitmap.vbuf_slot++ * sizeof st->bitmap.vertices;
+}
+
+
+
+/**
+ * Render a glBitmap by drawing a textured quad
+ */
+static void
+draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
+                 GLsizei width, GLsizei height,
+                 struct pipe_sampler_view *sv,
+                 const GLfloat *color)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct cso_context *cso = st->cso_context;
+   struct st_fp_variant *fpv;
+   struct st_fp_variant_key key;
+   GLuint maxSize;
+   GLuint offset;
+
+   memset(&key, 0, sizeof(key));
+   key.st = st;
+   key.bitmap = GL_TRUE;
+
+   fpv = st_get_fp_variant(st, st->fp, &key);
+
+   /* As an optimization, Mesa's fragment programs will sometimes get the
+    * primary color from a statevar/constant rather than a varying variable.
+    * when that's the case, we need to ensure that we use the 'color'
+    * parameter and not the current attribute color (which may have changed
+    * through glRasterPos and state validation.
+    * So, we force the proper color here.  Not elegant, but it works.
+    */
+   {
+      GLfloat colorSave[4];
+      COPY_4V(colorSave, ctx->Current.Attrib[VERT_ATTRIB_COLOR0]);
+      COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], color);
+      st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT);
+      COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], colorSave);
+   }
+
+
+   /* limit checks */
+   /* XXX if the bitmap is larger than the max texture size, break
+    * it up into chunks.
+    */
+   maxSize = 1 << (pipe->screen->get_param(pipe->screen,
+                                    PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1);
+   assert(width <= (GLsizei)maxSize);
+   assert(height <= (GLsizei)maxSize);
+
+   cso_save_rasterizer(cso);
+   cso_save_samplers(cso);
+   cso_save_fragment_sampler_views(cso);
+   cso_save_viewport(cso);
+   cso_save_fragment_shader(cso);
+   cso_save_vertex_shader(cso);
+   cso_save_vertex_elements(cso);
+   cso_save_vertex_buffers(cso);
+
+   /* rasterizer state: just scissor */
+   st->bitmap.rasterizer.scissor = ctx->Scissor.Enabled;
+   cso_set_rasterizer(cso, &st->bitmap.rasterizer);
+
+   /* fragment shader state: TEX lookup program */
+   cso_set_fragment_shader_handle(cso, fpv->driver_shader);
+
+   /* vertex shader state: position + texcoord pass-through */
+   cso_set_vertex_shader_handle(cso, st->bitmap.vs);
+
+   /* user samplers, plus our bitmap sampler */
+   {
+      struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS];
+      uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_samplers);
+      uint i;
+      for (i = 0; i < st->state.num_samplers; i++) {
+         samplers[i] = &st->state.samplers[i];
+      }
+      samplers[fpv->bitmap_sampler] =
+         &st->bitmap.samplers[sv->texture->target != PIPE_TEXTURE_RECT];
+      cso_set_samplers(cso, num, (const struct pipe_sampler_state **) samplers);
+   }
+
+   /* user textures, plus the bitmap texture */
+   {
+      struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS];
+      uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_textures);
+      memcpy(sampler_views, st->state.sampler_views, sizeof(sampler_views));
+      sampler_views[fpv->bitmap_sampler] = sv;
+      cso_set_fragment_sampler_views(cso, num, sampler_views);
+   }
+
+   /* viewport state: viewport matching window dims */
+   {
+      const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
+      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
+      const GLfloat width = (GLfloat)fb->Width;
+      const GLfloat height = (GLfloat)fb->Height;
+      struct pipe_viewport_state vp;
+      vp.scale[0] =  0.5f * width;
+      vp.scale[1] = height * (invert ? -0.5f : 0.5f);
+      vp.scale[2] = 0.5f;
+      vp.scale[3] = 1.0f;
+      vp.translate[0] = 0.5f * width;
+      vp.translate[1] = 0.5f * height;
+      vp.translate[2] = 0.5f;
+      vp.translate[3] = 0.0f;
+      cso_set_viewport(cso, &vp);
+   }
+
+   cso_set_vertex_elements(cso, 3, st->velems_util_draw);
+
+   /* convert Z from [0,1] to [-1,-1] to match viewport Z scale/bias */
+   z = z * 2.0 - 1.0;
+
+   /* draw textured quad */
+   offset = setup_bitmap_vertex_data(st,
+                                     sv->texture->target != PIPE_TEXTURE_RECT,
+                                     x, y, width, height, z, color);
+
+   util_draw_vertex_buffer(pipe, st->cso_context, st->bitmap.vbuf, offset,
+                           PIPE_PRIM_TRIANGLE_FAN,
+                           4,  /* verts */
+                           3); /* attribs/vert */
+
+
+   /* restore state */
+   cso_restore_rasterizer(cso);
+   cso_restore_samplers(cso);
+   cso_restore_fragment_sampler_views(cso);
+   cso_restore_viewport(cso);
+   cso_restore_fragment_shader(cso);
+   cso_restore_vertex_shader(cso);
+   cso_restore_vertex_elements(cso);
+   cso_restore_vertex_buffers(cso);
+}
+
+
+static void
+reset_cache(struct st_context *st)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct bitmap_cache *cache = st->bitmap.cache;
+
+   /*memset(cache->buffer, 0xff, sizeof(cache->buffer));*/
+   cache->empty = GL_TRUE;
+
+   cache->xmin = 1000000;
+   cache->xmax = -1000000;
+   cache->ymin = 1000000;
+   cache->ymax = -1000000;
+
+   if (cache->trans) {
+      pipe->transfer_destroy(pipe, cache->trans);
+      cache->trans = NULL;
+   }
+
+   assert(!cache->texture);
+
+   /* allocate a new texture */
+   cache->texture = st_texture_create(st, PIPE_TEXTURE_2D,
+                                      st->bitmap.tex_format, 0,
+                                      BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT,
+                                      1, 1,
+				      PIPE_BIND_SAMPLER_VIEW);
+}
+
+
+/** Print bitmap image to stdout (debug) */
+static void
+print_cache(const struct bitmap_cache *cache)
+{
+   int i, j, k;
+
+   for (i = 0; i < BITMAP_CACHE_HEIGHT; i++) {
+      k = BITMAP_CACHE_WIDTH * (BITMAP_CACHE_HEIGHT - i - 1);
+      for (j = 0; j < BITMAP_CACHE_WIDTH; j++) {
+         if (cache->buffer[k])
+            printf("X");
+         else
+            printf(" ");
+         k++;
+      }
+      printf("\n");
+   }
+}
+
+
+/**
+ * Create gallium pipe_transfer object for the bitmap cache.
+ */
+static void
+create_cache_trans(struct st_context *st)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct bitmap_cache *cache = st->bitmap.cache;
+
+   if (cache->trans)
+      return;
+
+   /* Map the texture transfer.
+    * Subsequent glBitmap calls will write into the texture image.
+    */
+   cache->trans = pipe_get_transfer(st->pipe, cache->texture, 0, 0,
+                                    PIPE_TRANSFER_WRITE, 0, 0,
+                                    BITMAP_CACHE_WIDTH,
+                                    BITMAP_CACHE_HEIGHT);
+   cache->buffer = pipe_transfer_map(pipe, cache->trans);
+
+   /* init image to all 0xff */
+   memset(cache->buffer, 0xff, cache->trans->stride * BITMAP_CACHE_HEIGHT);
+}
+
+
+/**
+ * If there's anything in the bitmap cache, draw/flush it now.
+ */
+void
+st_flush_bitmap_cache(struct st_context *st)
+{
+   if (!st->bitmap.cache->empty) {
+      struct bitmap_cache *cache = st->bitmap.cache;
+
+      if (st->ctx->DrawBuffer) {
+         struct pipe_context *pipe = st->pipe;
+         struct pipe_sampler_view *sv;
+
+         assert(cache->xmin <= cache->xmax);
+ 
+/*         printf("flush size %d x %d  at %d, %d\n",
+                cache->xmax - cache->xmin,
+                cache->ymax - cache->ymin,
+                cache->xpos, cache->ypos);
+*/
+
+         /* The texture transfer has been mapped until now.
+          * So unmap and release the texture transfer before drawing.
+          */
+         if (cache->trans) {
+            if (0)
+               print_cache(cache);
+            pipe_transfer_unmap(pipe, cache->trans);
+            cache->buffer = NULL;
+
+            pipe->transfer_destroy(pipe, cache->trans);
+            cache->trans = NULL;
+         }
+
+         sv = st_create_texture_sampler_view(st->pipe, cache->texture);
+         if (sv) {
+            draw_bitmap_quad(st->ctx,
+                             cache->xpos,
+                             cache->ypos,
+                             cache->zpos,
+                             BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT,
+                             sv,
+                             cache->color);
+
+            pipe_sampler_view_reference(&sv, NULL);
+         }
+      }
+
+      /* release/free the texture */
+      pipe_resource_reference(&cache->texture, NULL);
+
+      reset_cache(st);
+   }
+}
+
+
+/**
+ * Flush bitmap cache and release vertex buffer.
+ */
+void
+st_flush_bitmap( struct st_context *st )
+{
+   st_flush_bitmap_cache(st);
+
+   /* Release vertex buffer to avoid synchronous rendering if we were
+    * to map it in the next frame.
+    */
+   pipe_resource_reference(&st->bitmap.vbuf, NULL);
+   st->bitmap.vbuf_slot = 0;
+}
+
+
+/**
+ * Try to accumulate this glBitmap call in the bitmap cache.
+ * \return  GL_TRUE for success, GL_FALSE if bitmap is too large, etc.
+ */
+static GLboolean
+accum_bitmap(struct st_context *st,
+             GLint x, GLint y, GLsizei width, GLsizei height,
+             const struct gl_pixelstore_attrib *unpack,
+             const GLubyte *bitmap )
+{
+   struct bitmap_cache *cache = st->bitmap.cache;
+   int px = -999, py = -999;
+   const GLfloat z = st->ctx->Current.RasterPos[2];
+
+   if (width > BITMAP_CACHE_WIDTH ||
+       height > BITMAP_CACHE_HEIGHT)
+      return GL_FALSE; /* too big to cache */
+
+   if (!cache->empty) {
+      px = x - cache->xpos;  /* pos in buffer */
+      py = y - cache->ypos;
+      if (px < 0 || px + width > BITMAP_CACHE_WIDTH ||
+          py < 0 || py + height > BITMAP_CACHE_HEIGHT ||
+          !TEST_EQ_4V(st->ctx->Current.RasterColor, cache->color) ||
+          ((fabs(z - cache->zpos) > Z_EPSILON))) {
+         /* This bitmap would extend beyond cache bounds, or the bitmap
+          * color is changing
+          * so flush and continue.
+          */
+         st_flush_bitmap_cache(st);
+      }
+   }
+
+   if (cache->empty) {
+      /* Initialize.  Center bitmap vertically in the buffer. */
+      px = 0;
+      py = (BITMAP_CACHE_HEIGHT - height) / 2;
+      cache->xpos = x;
+      cache->ypos = y - py;
+      cache->zpos = z;
+      cache->empty = GL_FALSE;
+      COPY_4FV(cache->color, st->ctx->Current.RasterColor);
+   }
+
+   assert(px != -999);
+   assert(py != -999);
+
+   if (x < cache->xmin)
+      cache->xmin = x;
+   if (y < cache->ymin)
+      cache->ymin = y;
+   if (x + width > cache->xmax)
+      cache->xmax = x + width;
+   if (y + height > cache->ymax)
+      cache->ymax = y + height;
+
+   /* create the transfer if needed */
+   create_cache_trans(st);
+
+   unpack_bitmap(st, px, py, width, height, unpack, bitmap,
+                 cache->buffer, BITMAP_CACHE_WIDTH);
+
+   return GL_TRUE; /* accumulated */
+}
+
+
+
+/**
+ * Called via ctx->Driver.Bitmap()
+ */
+static void
+st_Bitmap(struct gl_context *ctx, GLint x, GLint y,
+          GLsizei width, GLsizei height,
+          const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap )
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_resource *pt;
+
+   if (width == 0 || height == 0)
+      return;
+
+   st_validate_state(st);
+
+   if (!st->bitmap.vs) {
+      /* create pass-through vertex shader now */
+      const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
+                                      TGSI_SEMANTIC_COLOR,
+                                      TGSI_SEMANTIC_GENERIC };
+      const uint semantic_indexes[] = { 0, 0, 0 };
+      st->bitmap.vs = util_make_vertex_passthrough_shader(st->pipe, 3,
+                                                          semantic_names,
+                                                          semantic_indexes);
+   }
+
+   if (UseBitmapCache && accum_bitmap(st, x, y, width, height, unpack, bitmap))
+      return;
+
+   pt = make_bitmap_texture(ctx, width, height, unpack, bitmap);
+   if (pt) {
+      struct pipe_sampler_view *sv =
+         st_create_texture_sampler_view(st->pipe, pt);
+
+      assert(pt->target == PIPE_TEXTURE_2D || pt->target == PIPE_TEXTURE_RECT);
+
+      if (sv) {
+         draw_bitmap_quad(ctx, x, y, ctx->Current.RasterPos[2],
+                          width, height, sv,
+                          st->ctx->Current.RasterColor);
+
+         pipe_sampler_view_reference(&sv, NULL);
+      }
+
+      /* release/free the texture */
+      pipe_resource_reference(&pt, NULL);
+   }
+}
+
+
+/** Per-context init */
+void
+st_init_bitmap_functions(struct dd_function_table *functions)
+{
+   functions->Bitmap = st_Bitmap;
+}
+
+
+/** Per-context init */
+void
+st_init_bitmap(struct st_context *st)
+{
+   struct pipe_sampler_state *sampler = &st->bitmap.samplers[0];
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_screen *screen = pipe->screen;
+
+   /* init sampler state once */
+   memset(sampler, 0, sizeof(*sampler));
+   sampler->wrap_s = PIPE_TEX_WRAP_CLAMP;
+   sampler->wrap_t = PIPE_TEX_WRAP_CLAMP;
+   sampler->wrap_r = PIPE_TEX_WRAP_CLAMP;
+   sampler->min_img_filter = PIPE_TEX_FILTER_NEAREST;
+   sampler->min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+   sampler->mag_img_filter = PIPE_TEX_FILTER_NEAREST;
+   st->bitmap.samplers[1] = *sampler;
+   st->bitmap.samplers[1].normalized_coords = 1;
+
+   /* init baseline rasterizer state once */
+   memset(&st->bitmap.rasterizer, 0, sizeof(st->bitmap.rasterizer));
+   st->bitmap.rasterizer.gl_rasterization_rules = 1;
+
+   /* find a usable texture format */
+   if (screen->is_format_supported(screen, PIPE_FORMAT_I8_UNORM,
+                                   PIPE_TEXTURE_2D, 0,
+                                   PIPE_BIND_SAMPLER_VIEW, 0)) {
+      st->bitmap.tex_format = PIPE_FORMAT_I8_UNORM;
+   }
+   else if (screen->is_format_supported(screen, PIPE_FORMAT_A8_UNORM,
+                                        PIPE_TEXTURE_2D, 0,
+                                        PIPE_BIND_SAMPLER_VIEW, 0)) {
+      st->bitmap.tex_format = PIPE_FORMAT_A8_UNORM;
+   }
+   else if (screen->is_format_supported(screen, PIPE_FORMAT_L8_UNORM,
+                                        PIPE_TEXTURE_2D, 0,
+                                        PIPE_BIND_SAMPLER_VIEW, 0)) {
+      st->bitmap.tex_format = PIPE_FORMAT_L8_UNORM;
+   }
+   else {
+      /* XXX support more formats */
+      assert(0);
+   }
+
+   /* alloc bitmap cache object */
+   st->bitmap.cache = ST_CALLOC_STRUCT(bitmap_cache);
+
+   reset_cache(st);
+}
+
+
+/** Per-context tear-down */
+void
+st_destroy_bitmap(struct st_context *st)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct bitmap_cache *cache = st->bitmap.cache;
+
+   if (st->bitmap.vs) {
+      cso_delete_vertex_shader(st->cso_context, st->bitmap.vs);
+      st->bitmap.vs = NULL;
+   }
+
+   if (st->bitmap.vbuf) {
+      pipe_resource_reference(&st->bitmap.vbuf, NULL);
+      st->bitmap.vbuf = NULL;
+   }
+
+   if (cache) {
+      if (cache->trans) {
+         pipe_transfer_unmap(pipe, cache->trans);
+         pipe->transfer_destroy(pipe, cache->trans);
+      }
+      pipe_resource_reference(&st->bitmap.cache->texture, NULL);
+      free(st->bitmap.cache);
+      st->bitmap.cache = NULL;
+   }
+}
+
+#endif /* FEATURE_drawpix */
diff --git a/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c b/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
index 12528f49f..9dd1f8a3e 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
@@ -1,468 +1,468 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-/**
- * Functions for pixel buffer objects and vertex/element buffer objects.
- */
-
-
-#include "main/imports.h"
-#include "main/mtypes.h"
-#include "main/arrayobj.h"
-#include "main/bufferobj.h"
-
-#include "st_context.h"
-#include "st_cb_bufferobjects.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-
-
-/**
- * There is some duplication between mesa's bufferobjects and our
- * bufmgr buffers.  Both have an integer handle and a hashtable to
- * lookup an opaque structure.  It would be nice if the handles and
- * internal structure where somehow shared.
- */
-static struct gl_buffer_object *
-st_bufferobj_alloc(struct gl_context *ctx, GLuint name, GLenum target)
-{
-   struct st_buffer_object *st_obj = ST_CALLOC_STRUCT(st_buffer_object);
-
-   if (!st_obj)
-      return NULL;
-
-   _mesa_initialize_buffer_object(&st_obj->Base, name, target);
-
-   return &st_obj->Base;
-}
-
-
-
-/**
- * Deallocate/free a vertex/pixel buffer object.
- * Called via glDeleteBuffersARB().
- */
-static void
-st_bufferobj_free(struct gl_context *ctx, struct gl_buffer_object *obj)
-{
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-
-   assert(obj->RefCount == 0);
-   assert(st_obj->transfer == NULL);
-
-   if (st_obj->buffer) 
-      pipe_resource_reference(&st_obj->buffer, NULL);
-
-   free(st_obj);
-}
-
-
-
-/**
- * Replace data in a subrange of buffer object.  If the data range
- * specified by size + offset extends beyond the end of the buffer or
- * if data is NULL, no copy is performed.
- * Called via glBufferSubDataARB().
- */
-static void
-st_bufferobj_subdata(struct gl_context *ctx,
-		     GLenum target,
-		     GLintptrARB offset,
-		     GLsizeiptrARB size,
-		     const GLvoid * data, struct gl_buffer_object *obj)
-{
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-
-   /* we may be called from VBO code, so double-check params here */
-   ASSERT(offset >= 0);
-   ASSERT(size >= 0);
-   ASSERT(offset + size <= obj->Size);
-
-   if (!size)
-      return;
-
-   /*
-    * According to ARB_vertex_buffer_object specification, if data is null,
-    * then the contents of the buffer object's data store is undefined. We just
-    * ignore, and leave it unchanged.
-    */
-   if (!data)
-      return;
-
-   /* Now that transfers are per-context, we don't have to figure out
-    * flushing here.  Usually drivers won't need to flush in this case
-    * even if the buffer is currently referenced by hardware - they
-    * just queue the upload as dma rather than mapping the underlying
-    * buffer directly.
-    */
-   pipe_buffer_write(st_context(ctx)->pipe,
-		     st_obj->buffer,
-		     offset, size, data);
-}
-
-
-/**
- * Called via glGetBufferSubDataARB().
- */
-static void
-st_bufferobj_get_subdata(struct gl_context *ctx,
-                         GLenum target,
-                         GLintptrARB offset,
-                         GLsizeiptrARB size,
-                         GLvoid * data, struct gl_buffer_object *obj)
-{
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-
-   /* we may be called from VBO code, so double-check params here */
-   ASSERT(offset >= 0);
-   ASSERT(size >= 0);
-   ASSERT(offset + size <= obj->Size);
-
-   if (!size)
-      return;
-
-   pipe_buffer_read(st_context(ctx)->pipe, st_obj->buffer,
-                    offset, size, data);
-}
-
-
-/**
- * Allocate space for and store data in a buffer object.  Any data that was
- * previously stored in the buffer object is lost.  If data is NULL,
- * memory will be allocated, but no copy will occur.
- * Called via ctx->Driver.BufferData().
- * \return GL_TRUE for success, GL_FALSE if out of memory
- */
-static GLboolean
-st_bufferobj_data(struct gl_context *ctx,
-		  GLenum target,
-		  GLsizeiptrARB size,
-		  const GLvoid * data,
-		  GLenum usage, 
-		  struct gl_buffer_object *obj)
-{
-   struct st_context *st = st_context(ctx);
-   struct pipe_context *pipe = st->pipe;
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-   unsigned bind, pipe_usage;
-
-   st_obj->Base.Size = size;
-   st_obj->Base.Usage = usage;
-   
-   switch(target) {
-   case GL_PIXEL_PACK_BUFFER_ARB:
-   case GL_PIXEL_UNPACK_BUFFER_ARB:
-      bind = PIPE_BIND_RENDER_TARGET | PIPE_BIND_SAMPLER_VIEW;
-      break;
-   case GL_ARRAY_BUFFER_ARB:
-      bind = PIPE_BIND_VERTEX_BUFFER;
-      break;
-   case GL_ELEMENT_ARRAY_BUFFER_ARB:
-      bind = PIPE_BIND_INDEX_BUFFER;
-      break;
-   default:
-      bind = 0;
-   }
-
-   switch (usage) {
-   case GL_STATIC_DRAW:
-   case GL_STATIC_READ:
-   case GL_STATIC_COPY:
-      pipe_usage = PIPE_USAGE_STATIC;
-      break;
-   case GL_DYNAMIC_DRAW:
-   case GL_DYNAMIC_READ:
-   case GL_DYNAMIC_COPY:
-      pipe_usage = PIPE_USAGE_DYNAMIC;
-      break;
-   case GL_STREAM_DRAW:
-   case GL_STREAM_READ:
-   case GL_STREAM_COPY:
-      pipe_usage = PIPE_USAGE_STREAM;
-      break;
-   default:
-      pipe_usage = PIPE_USAGE_DEFAULT;
-   }
-
-   pipe_resource_reference( &st_obj->buffer, NULL );
-
-   if (size != 0) {
-      st_obj->buffer = pipe_buffer_create(pipe->screen, bind,
-                                          pipe_usage, size);
-
-      if (!st_obj->buffer) {
-         return GL_FALSE;
-      }
-
-      if (data)
-         pipe_buffer_write(st_context(ctx)->pipe, st_obj->buffer, 0,
-				       size, data);
-      return GL_TRUE;
-   }
-
-   return GL_TRUE;
-}
-
-
-/**
- * Dummy data whose's pointer is used for zero size buffers or ranges.
- */
-static long st_bufferobj_zero_length = 0;
-
-
-
-/**
- * Called via glMapBufferARB().
- */
-static void *
-st_bufferobj_map(struct gl_context *ctx, GLenum target, GLenum access,
-                 struct gl_buffer_object *obj)
-{
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-   uint flags;
-
-   switch (access) {
-   case GL_WRITE_ONLY:
-      flags = PIPE_TRANSFER_WRITE;
-      break;
-   case GL_READ_ONLY:
-      flags = PIPE_TRANSFER_READ;
-      break;
-   case GL_READ_WRITE:
-   default:
-      flags = PIPE_TRANSFER_READ_WRITE;
-      break;      
-   }
-
-   /* Handle zero-size buffers here rather than in drivers */
-   if (obj->Size == 0) {
-      obj->Pointer = &st_bufferobj_zero_length;
-   }
-   else {
-      obj->Pointer = pipe_buffer_map(st_context(ctx)->pipe,
-                                     st_obj->buffer,
-                                     flags,
-                                     &st_obj->transfer);
-   }
-
-   if (obj->Pointer) {
-      obj->Offset = 0;
-      obj->Length = obj->Size;
-   }
-   return obj->Pointer;
-}
-
-
-/**
- * Called via glMapBufferRange().
- */
-static void *
-st_bufferobj_map_range(struct gl_context *ctx, GLenum target, 
-                       GLintptr offset, GLsizeiptr length, GLbitfield access,
-                       struct gl_buffer_object *obj)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-   enum pipe_transfer_usage flags = 0x0;
-
-   if (access & GL_MAP_WRITE_BIT)
-      flags |= PIPE_TRANSFER_WRITE;
-
-   if (access & GL_MAP_READ_BIT)
-      flags |= PIPE_TRANSFER_READ;
-
-   if (access & GL_MAP_FLUSH_EXPLICIT_BIT)
-      flags |= PIPE_TRANSFER_FLUSH_EXPLICIT;
-
-   if (access & GL_MAP_INVALIDATE_RANGE_BIT)
-      flags |= PIPE_TRANSFER_DISCARD;
-
-   if (access & GL_MAP_INVALIDATE_BUFFER_BIT)
-      flags |= PIPE_TRANSFER_DISCARD;
-   
-   if (access & GL_MAP_UNSYNCHRONIZED_BIT)
-      flags |= PIPE_TRANSFER_UNSYNCHRONIZED;
-
-   /* ... other flags ...
-    */
-
-   if (access & MESA_MAP_NOWAIT_BIT)
-      flags |= PIPE_TRANSFER_DONTBLOCK;
-
-   assert(offset >= 0);
-   assert(length >= 0);
-   assert(offset < obj->Size);
-   assert(offset + length <= obj->Size);
-
-   /*
-    * We go out of way here to hide the degenerate yet valid case of zero
-    * length range from the pipe driver.
-    */
-   if (!length) {
-      obj->Pointer = &st_bufferobj_zero_length;
-   }
-   else {
-      obj->Pointer = pipe_buffer_map_range(pipe, 
-                                           st_obj->buffer,
-                                           offset, length,
-                                           flags,
-                                           &st_obj->transfer);
-      if (obj->Pointer) {
-         obj->Pointer = (ubyte *) obj->Pointer + offset;
-      }
-   }
-   
-   if (obj->Pointer) {
-      obj->Offset = offset;
-      obj->Length = length;
-      obj->AccessFlags = access;
-   }
-
-   return obj->Pointer;
-}
-
-
-static void
-st_bufferobj_flush_mapped_range(struct gl_context *ctx, GLenum target, 
-                                GLintptr offset, GLsizeiptr length,
-                                struct gl_buffer_object *obj)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-
-   /* Subrange is relative to mapped range */
-   assert(offset >= 0);
-   assert(length >= 0);
-   assert(offset + length <= obj->Length);
-   assert(obj->Pointer);
-   
-   if (!length)
-      return;
-
-   pipe_buffer_flush_mapped_range(pipe, st_obj->transfer, 
-                                  obj->Offset + offset, length);
-}
-
-
-/**
- * Called via glUnmapBufferARB().
- */
-static GLboolean
-st_bufferobj_unmap(struct gl_context *ctx, GLenum target, struct gl_buffer_object *obj)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   struct st_buffer_object *st_obj = st_buffer_object(obj);
-
-   if (obj->Length)
-      pipe_buffer_unmap(pipe, st_obj->transfer);
-
-   st_obj->transfer = NULL;
-   obj->Pointer = NULL;
-   obj->Offset = 0;
-   obj->Length = 0;
-   return GL_TRUE;
-}
-
-
-/**
- * Called via glCopyBufferSubData().
- */
-static void
-st_copy_buffer_subdata(struct gl_context *ctx,
-                       struct gl_buffer_object *src,
-                       struct gl_buffer_object *dst,
-                       GLintptr readOffset, GLintptr writeOffset,
-                       GLsizeiptr size)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   struct st_buffer_object *srcObj = st_buffer_object(src);
-   struct st_buffer_object *dstObj = st_buffer_object(dst);
-   struct pipe_transfer *src_transfer;
-   struct pipe_transfer *dst_transfer;
-   ubyte *srcPtr, *dstPtr;
-
-   if(!size)
-      return;
-
-   /* buffer should not already be mapped */
-   assert(!src->Pointer);
-   assert(!dst->Pointer);
-
-   srcPtr = (ubyte *) pipe_buffer_map_range(pipe,
-                                            srcObj->buffer,
-                                            readOffset, size,
-                                            PIPE_TRANSFER_READ,
-					    &src_transfer);
-
-   dstPtr = (ubyte *) pipe_buffer_map_range(pipe,
-                                            dstObj->buffer,
-                                            writeOffset, size,
-                                            PIPE_TRANSFER_WRITE,
-					    &dst_transfer);
-
-   if (srcPtr && dstPtr)
-      memcpy(dstPtr + writeOffset, srcPtr + readOffset, size);
-
-   pipe_buffer_unmap(pipe, src_transfer);
-   pipe_buffer_unmap(pipe, dst_transfer);
-}
-
-
-/* TODO: if buffer wasn't created with appropriate usage flags, need
- * to recreate it now and copy contents -- or possibly create a
- * gallium entrypoint to extend the usage flags and let the driver
- * decide if a copy is necessary.
- */
-void
-st_bufferobj_validate_usage(struct st_context *st,
-			    struct st_buffer_object *obj,
-			    unsigned usage)
-{
-}
-
-
-void
-st_init_bufferobject_functions(struct dd_function_table *functions)
-{
-   functions->NewBufferObject = st_bufferobj_alloc;
-   functions->DeleteBuffer = st_bufferobj_free;
-   functions->BufferData = st_bufferobj_data;
-   functions->BufferSubData = st_bufferobj_subdata;
-   functions->GetBufferSubData = st_bufferobj_get_subdata;
-   functions->MapBuffer = st_bufferobj_map;
-   functions->MapBufferRange = st_bufferobj_map_range;
-   functions->FlushMappedBufferRange = st_bufferobj_flush_mapped_range;
-   functions->UnmapBuffer = st_bufferobj_unmap;
-   functions->CopyBufferSubData = st_copy_buffer_subdata;
-
-   /* For GL_APPLE_vertex_array_object */
-   functions->NewArrayObject = _mesa_new_array_object;
-   functions->DeleteArrayObject = _mesa_delete_array_object;
-}
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+/**
+ * Functions for pixel buffer objects and vertex/element buffer objects.
+ */
+
+
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/arrayobj.h"
+#include "main/bufferobj.h"
+
+#include "st_context.h"
+#include "st_cb_bufferobjects.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "util/u_inlines.h"
+
+
+/**
+ * There is some duplication between mesa's bufferobjects and our
+ * bufmgr buffers.  Both have an integer handle and a hashtable to
+ * lookup an opaque structure.  It would be nice if the handles and
+ * internal structure where somehow shared.
+ */
+static struct gl_buffer_object *
+st_bufferobj_alloc(struct gl_context *ctx, GLuint name, GLenum target)
+{
+   struct st_buffer_object *st_obj = ST_CALLOC_STRUCT(st_buffer_object);
+
+   if (!st_obj)
+      return NULL;
+
+   _mesa_initialize_buffer_object(&st_obj->Base, name, target);
+
+   return &st_obj->Base;
+}
+
+
+
+/**
+ * Deallocate/free a vertex/pixel buffer object.
+ * Called via glDeleteBuffersARB().
+ */
+static void
+st_bufferobj_free(struct gl_context *ctx, struct gl_buffer_object *obj)
+{
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+
+   assert(obj->RefCount == 0);
+   assert(st_obj->transfer == NULL);
+
+   if (st_obj->buffer) 
+      pipe_resource_reference(&st_obj->buffer, NULL);
+
+   free(st_obj);
+}
+
+
+
+/**
+ * Replace data in a subrange of buffer object.  If the data range
+ * specified by size + offset extends beyond the end of the buffer or
+ * if data is NULL, no copy is performed.
+ * Called via glBufferSubDataARB().
+ */
+static void
+st_bufferobj_subdata(struct gl_context *ctx,
+		     GLenum target,
+		     GLintptrARB offset,
+		     GLsizeiptrARB size,
+		     const GLvoid * data, struct gl_buffer_object *obj)
+{
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+
+   /* we may be called from VBO code, so double-check params here */
+   ASSERT(offset >= 0);
+   ASSERT(size >= 0);
+   ASSERT(offset + size <= obj->Size);
+
+   if (!size)
+      return;
+
+   /*
+    * According to ARB_vertex_buffer_object specification, if data is null,
+    * then the contents of the buffer object's data store is undefined. We just
+    * ignore, and leave it unchanged.
+    */
+   if (!data)
+      return;
+
+   /* Now that transfers are per-context, we don't have to figure out
+    * flushing here.  Usually drivers won't need to flush in this case
+    * even if the buffer is currently referenced by hardware - they
+    * just queue the upload as dma rather than mapping the underlying
+    * buffer directly.
+    */
+   pipe_buffer_write(st_context(ctx)->pipe,
+		     st_obj->buffer,
+		     offset, size, data);
+}
+
+
+/**
+ * Called via glGetBufferSubDataARB().
+ */
+static void
+st_bufferobj_get_subdata(struct gl_context *ctx,
+                         GLenum target,
+                         GLintptrARB offset,
+                         GLsizeiptrARB size,
+                         GLvoid * data, struct gl_buffer_object *obj)
+{
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+
+   /* we may be called from VBO code, so double-check params here */
+   ASSERT(offset >= 0);
+   ASSERT(size >= 0);
+   ASSERT(offset + size <= obj->Size);
+
+   if (!size)
+      return;
+
+   pipe_buffer_read(st_context(ctx)->pipe, st_obj->buffer,
+                    offset, size, data);
+}
+
+
+/**
+ * Allocate space for and store data in a buffer object.  Any data that was
+ * previously stored in the buffer object is lost.  If data is NULL,
+ * memory will be allocated, but no copy will occur.
+ * Called via ctx->Driver.BufferData().
+ * \return GL_TRUE for success, GL_FALSE if out of memory
+ */
+static GLboolean
+st_bufferobj_data(struct gl_context *ctx,
+		  GLenum target,
+		  GLsizeiptrARB size,
+		  const GLvoid * data,
+		  GLenum usage, 
+		  struct gl_buffer_object *obj)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+   unsigned bind, pipe_usage;
+
+   st_obj->Base.Size = size;
+   st_obj->Base.Usage = usage;
+   
+   switch(target) {
+   case GL_PIXEL_PACK_BUFFER_ARB:
+   case GL_PIXEL_UNPACK_BUFFER_ARB:
+      bind = PIPE_BIND_RENDER_TARGET | PIPE_BIND_SAMPLER_VIEW;
+      break;
+   case GL_ARRAY_BUFFER_ARB:
+      bind = PIPE_BIND_VERTEX_BUFFER;
+      break;
+   case GL_ELEMENT_ARRAY_BUFFER_ARB:
+      bind = PIPE_BIND_INDEX_BUFFER;
+      break;
+   default:
+      bind = 0;
+   }
+
+   switch (usage) {
+   case GL_STATIC_DRAW:
+   case GL_STATIC_READ:
+   case GL_STATIC_COPY:
+      pipe_usage = PIPE_USAGE_STATIC;
+      break;
+   case GL_DYNAMIC_DRAW:
+   case GL_DYNAMIC_READ:
+   case GL_DYNAMIC_COPY:
+      pipe_usage = PIPE_USAGE_DYNAMIC;
+      break;
+   case GL_STREAM_DRAW:
+   case GL_STREAM_READ:
+   case GL_STREAM_COPY:
+      pipe_usage = PIPE_USAGE_STREAM;
+      break;
+   default:
+      pipe_usage = PIPE_USAGE_DEFAULT;
+   }
+
+   pipe_resource_reference( &st_obj->buffer, NULL );
+
+   if (size != 0) {
+      st_obj->buffer = pipe_buffer_create(pipe->screen, bind,
+                                          pipe_usage, size);
+
+      if (!st_obj->buffer) {
+         return GL_FALSE;
+      }
+
+      if (data)
+         pipe_buffer_write(st_context(ctx)->pipe, st_obj->buffer, 0,
+				       size, data);
+      return GL_TRUE;
+   }
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Dummy data whose's pointer is used for zero size buffers or ranges.
+ */
+static long st_bufferobj_zero_length = 0;
+
+
+
+/**
+ * Called via glMapBufferARB().
+ */
+static void *
+st_bufferobj_map(struct gl_context *ctx, GLenum target, GLenum access,
+                 struct gl_buffer_object *obj)
+{
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+   uint flags;
+
+   switch (access) {
+   case GL_WRITE_ONLY:
+      flags = PIPE_TRANSFER_WRITE;
+      break;
+   case GL_READ_ONLY:
+      flags = PIPE_TRANSFER_READ;
+      break;
+   case GL_READ_WRITE:
+   default:
+      flags = PIPE_TRANSFER_READ_WRITE;
+      break;      
+   }
+
+   /* Handle zero-size buffers here rather than in drivers */
+   if (obj->Size == 0) {
+      obj->Pointer = &st_bufferobj_zero_length;
+   }
+   else {
+      obj->Pointer = pipe_buffer_map(st_context(ctx)->pipe,
+                                     st_obj->buffer,
+                                     flags,
+                                     &st_obj->transfer);
+   }
+
+   if (obj->Pointer) {
+      obj->Offset = 0;
+      obj->Length = obj->Size;
+   }
+   return obj->Pointer;
+}
+
+
+/**
+ * Called via glMapBufferRange().
+ */
+static void *
+st_bufferobj_map_range(struct gl_context *ctx, GLenum target, 
+                       GLintptr offset, GLsizeiptr length, GLbitfield access,
+                       struct gl_buffer_object *obj)
+{
+   struct pipe_context *pipe = st_context(ctx)->pipe;
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+   enum pipe_transfer_usage flags = 0x0;
+
+   if (access & GL_MAP_WRITE_BIT)
+      flags |= PIPE_TRANSFER_WRITE;
+
+   if (access & GL_MAP_READ_BIT)
+      flags |= PIPE_TRANSFER_READ;
+
+   if (access & GL_MAP_FLUSH_EXPLICIT_BIT)
+      flags |= PIPE_TRANSFER_FLUSH_EXPLICIT;
+
+   if (access & GL_MAP_INVALIDATE_RANGE_BIT)
+      flags |= PIPE_TRANSFER_DISCARD;
+
+   if (access & GL_MAP_INVALIDATE_BUFFER_BIT)
+      flags |= PIPE_TRANSFER_DISCARD;
+   
+   if (access & GL_MAP_UNSYNCHRONIZED_BIT)
+      flags |= PIPE_TRANSFER_UNSYNCHRONIZED;
+
+   /* ... other flags ...
+    */
+
+   if (access & MESA_MAP_NOWAIT_BIT)
+      flags |= PIPE_TRANSFER_DONTBLOCK;
+
+   assert(offset >= 0);
+   assert(length >= 0);
+   assert(offset < obj->Size);
+   assert(offset + length <= obj->Size);
+
+   /*
+    * We go out of way here to hide the degenerate yet valid case of zero
+    * length range from the pipe driver.
+    */
+   if (!length) {
+      obj->Pointer = &st_bufferobj_zero_length;
+   }
+   else {
+      obj->Pointer = pipe_buffer_map_range(pipe, 
+                                           st_obj->buffer,
+                                           offset, length,
+                                           flags,
+                                           &st_obj->transfer);
+      if (obj->Pointer) {
+         obj->Pointer = (ubyte *) obj->Pointer + offset;
+      }
+   }
+   
+   if (obj->Pointer) {
+      obj->Offset = offset;
+      obj->Length = length;
+      obj->AccessFlags = access;
+   }
+
+   return obj->Pointer;
+}
+
+
+static void
+st_bufferobj_flush_mapped_range(struct gl_context *ctx, GLenum target, 
+                                GLintptr offset, GLsizeiptr length,
+                                struct gl_buffer_object *obj)
+{
+   struct pipe_context *pipe = st_context(ctx)->pipe;
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+
+   /* Subrange is relative to mapped range */
+   assert(offset >= 0);
+   assert(length >= 0);
+   assert(offset + length <= obj->Length);
+   assert(obj->Pointer);
+   
+   if (!length)
+      return;
+
+   pipe_buffer_flush_mapped_range(pipe, st_obj->transfer, 
+                                  obj->Offset + offset, length);
+}
+
+
+/**
+ * Called via glUnmapBufferARB().
+ */
+static GLboolean
+st_bufferobj_unmap(struct gl_context *ctx, GLenum target, struct gl_buffer_object *obj)
+{
+   struct pipe_context *pipe = st_context(ctx)->pipe;
+   struct st_buffer_object *st_obj = st_buffer_object(obj);
+
+   if (obj->Length)
+      pipe_buffer_unmap(pipe, st_obj->transfer);
+
+   st_obj->transfer = NULL;
+   obj->Pointer = NULL;
+   obj->Offset = 0;
+   obj->Length = 0;
+   return GL_TRUE;
+}
+
+
+/**
+ * Called via glCopyBufferSubData().
+ */
+static void
+st_copy_buffer_subdata(struct gl_context *ctx,
+                       struct gl_buffer_object *src,
+                       struct gl_buffer_object *dst,
+                       GLintptr readOffset, GLintptr writeOffset,
+                       GLsizeiptr size)
+{
+   struct pipe_context *pipe = st_context(ctx)->pipe;
+   struct st_buffer_object *srcObj = st_buffer_object(src);
+   struct st_buffer_object *dstObj = st_buffer_object(dst);
+   struct pipe_transfer *src_transfer;
+   struct pipe_transfer *dst_transfer;
+   ubyte *srcPtr, *dstPtr;
+
+   if(!size)
+      return;
+
+   /* buffer should not already be mapped */
+   assert(!src->Pointer);
+   assert(!dst->Pointer);
+
+   srcPtr = (ubyte *) pipe_buffer_map_range(pipe,
+                                            srcObj->buffer,
+                                            readOffset, size,
+                                            PIPE_TRANSFER_READ,
+					    &src_transfer);
+
+   dstPtr = (ubyte *) pipe_buffer_map_range(pipe,
+                                            dstObj->buffer,
+                                            writeOffset, size,
+                                            PIPE_TRANSFER_WRITE,
+					    &dst_transfer);
+
+   if (srcPtr && dstPtr)
+      memcpy(dstPtr + writeOffset, srcPtr + readOffset, size);
+
+   pipe_buffer_unmap(pipe, src_transfer);
+   pipe_buffer_unmap(pipe, dst_transfer);
+}
+
+
+/* TODO: if buffer wasn't created with appropriate usage flags, need
+ * to recreate it now and copy contents -- or possibly create a
+ * gallium entrypoint to extend the usage flags and let the driver
+ * decide if a copy is necessary.
+ */
+void
+st_bufferobj_validate_usage(struct st_context *st,
+			    struct st_buffer_object *obj,
+			    unsigned usage)
+{
+}
+
+
+void
+st_init_bufferobject_functions(struct dd_function_table *functions)
+{
+   functions->NewBufferObject = st_bufferobj_alloc;
+   functions->DeleteBuffer = st_bufferobj_free;
+   functions->BufferData = st_bufferobj_data;
+   functions->BufferSubData = st_bufferobj_subdata;
+   functions->GetBufferSubData = st_bufferobj_get_subdata;
+   functions->MapBuffer = st_bufferobj_map;
+   functions->MapBufferRange = st_bufferobj_map_range;
+   functions->FlushMappedBufferRange = st_bufferobj_flush_mapped_range;
+   functions->UnmapBuffer = st_bufferobj_unmap;
+   functions->CopyBufferSubData = st_copy_buffer_subdata;
+
+   /* For GL_APPLE_vertex_array_object */
+   functions->NewArrayObject = _mesa_new_array_object;
+   functions->DeleteArrayObject = _mesa_delete_array_object;
+}
diff --git a/mesalib/src/mesa/state_tracker/st_cb_clear.c b/mesalib/src/mesa/state_tracker/st_cb_clear.c
index d2e0cd73c..461f4e315 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_clear.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_clear.c
@@ -1,563 +1,563 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 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 TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  *   Brian Paul
-  *   Michel Dänzer
-  */
-
-#include "main/glheader.h"
-#include "main/formats.h"
-#include "main/macros.h"
-#include "program/prog_instruction.h"
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_cb_accum.h"
-#include "st_cb_clear.h"
-#include "st_cb_fbo.h"
-#include "st_program.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_shader_tokens.h"
-#include "pipe/p_state.h"
-#include "pipe/p_defines.h"
-#include "util/u_format.h"
-#include "util/u_inlines.h"
-#include "util/u_simple_shaders.h"
-#include "util/u_draw_quad.h"
-
-#include "cso_cache/cso_context.h"
-
-
-/**
- * Do per-context initialization for glClear.
- */
-void
-st_init_clear(struct st_context *st)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_screen *pscreen = st->pipe->screen;
-
-   memset(&st->clear, 0, sizeof(st->clear));
-
-   st->clear.raster.gl_rasterization_rules = 1;
-   st->clear.enable_ds_separate = pscreen->get_param(pscreen, PIPE_CAP_DEPTHSTENCIL_CLEAR_SEPARATE);
-
-   /* fragment shader state: color pass-through program */
-   st->clear.fs = util_make_fragment_passthrough_shader(pipe);
-
-   /* vertex shader state: color/position pass-through */
-   {
-      const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
-                                      TGSI_SEMANTIC_COLOR };
-      const uint semantic_indexes[] = { 0, 0 };
-      st->clear.vs = util_make_vertex_passthrough_shader(pipe, 2,
-                                                         semantic_names,
-                                                         semantic_indexes);
-   }
-}
-
-
-/**
- * Free per-context state for glClear.
- */
-void
-st_destroy_clear(struct st_context *st)
-{
-   if (st->clear.fs) {
-      cso_delete_fragment_shader(st->cso_context, st->clear.fs);
-      st->clear.fs = NULL;
-   }
-   if (st->clear.vs) {
-      cso_delete_vertex_shader(st->cso_context, st->clear.vs);
-      st->clear.vs = NULL;
-   }
-   if (st->clear.vbuf) {
-      pipe_resource_reference(&st->clear.vbuf, NULL);
-      st->clear.vbuf = NULL;
-   }
-}
-
-
-/**
- * Draw a screen-aligned quadrilateral.
- * Coords are clip coords with y=0=bottom.
- */
-static void
-draw_quad(struct st_context *st,
-          float x0, float y0, float x1, float y1, GLfloat z,
-          const GLfloat color[4])
-{
-   struct pipe_context *pipe = st->pipe;
-
-   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
-    * no_flush) updates to buffers where we know there is no conflict
-    * with previous data.  Currently using max_slots > 1 will cause
-    * synchronous rendering if the driver flushes its command buffers
-    * between one bitmap and the next.  Our flush hook below isn't
-    * sufficient to catch this as the driver doesn't tell us when it
-    * flushes its own command buffers.  Until this gets fixed, pay the
-    * price of allocating a new buffer for each bitmap cache-flush to
-    * avoid synchronous rendering.
-    */
-   const GLuint max_slots = 1; /* 1024 / sizeof(st->clear.vertices); */
-   GLuint i;
-
-   if (st->clear.vbuf_slot >= max_slots) {
-      pipe_resource_reference(&st->clear.vbuf, NULL);
-      st->clear.vbuf_slot = 0;
-   }
-
-   if (!st->clear.vbuf) {
-      st->clear.vbuf = pipe_buffer_create(pipe->screen,
-                                          PIPE_BIND_VERTEX_BUFFER,
-                                          PIPE_USAGE_STREAM,
-                                          max_slots * sizeof(st->clear.vertices));
-   }
-
-   /* positions */
-   st->clear.vertices[0][0][0] = x0;
-   st->clear.vertices[0][0][1] = y0;
-
-   st->clear.vertices[1][0][0] = x1;
-   st->clear.vertices[1][0][1] = y0;
-
-   st->clear.vertices[2][0][0] = x1;
-   st->clear.vertices[2][0][1] = y1;
-
-   st->clear.vertices[3][0][0] = x0;
-   st->clear.vertices[3][0][1] = y1;
-
-   /* same for all verts: */
-   for (i = 0; i < 4; i++) {
-      st->clear.vertices[i][0][2] = z;
-      st->clear.vertices[i][0][3] = 1.0;
-      st->clear.vertices[i][1][0] = color[0];
-      st->clear.vertices[i][1][1] = color[1];
-      st->clear.vertices[i][1][2] = color[2];
-      st->clear.vertices[i][1][3] = color[3];
-   }
-
-   /* put vertex data into vbuf */
-   pipe_buffer_write_nooverlap(st->pipe, st->clear.vbuf,
-                                           st->clear.vbuf_slot
-                                             * sizeof(st->clear.vertices),
-                                           sizeof(st->clear.vertices),
-                                           st->clear.vertices);
-
-   /* draw */
-   util_draw_vertex_buffer(pipe,
-                           st->cso_context,
-                           st->clear.vbuf, 
-                           st->clear.vbuf_slot * sizeof(st->clear.vertices),
-                           PIPE_PRIM_TRIANGLE_FAN,
-                           4,  /* verts */
-                           2); /* attribs/vert */
-
-   /* Increment slot */
-   st->clear.vbuf_slot++;
-}
-
-
-
-/**
- * Do glClear by drawing a quadrilateral.
- * The vertices of the quad will be computed from the
- * ctx->DrawBuffer->_X/Ymin/max fields.
- */
-static void
-clear_with_quad(struct gl_context *ctx,
-                GLboolean color, GLboolean depth, GLboolean stencil)
-{
-   struct st_context *st = st_context(ctx);
-   const struct gl_framebuffer *fb = ctx->DrawBuffer;
-   const GLfloat fb_width = (GLfloat) fb->Width;
-   const GLfloat fb_height = (GLfloat) fb->Height;
-   const GLfloat x0 = (GLfloat) ctx->DrawBuffer->_Xmin / fb_width * 2.0f - 1.0f;
-   const GLfloat x1 = (GLfloat) ctx->DrawBuffer->_Xmax / fb_width * 2.0f - 1.0f;
-   const GLfloat y0 = (GLfloat) ctx->DrawBuffer->_Ymin / fb_height * 2.0f - 1.0f;
-   const GLfloat y1 = (GLfloat) ctx->DrawBuffer->_Ymax / fb_height * 2.0f - 1.0f;
-
-   /*
-   printf("%s %s%s%s %f,%f %f,%f\n", __FUNCTION__, 
-	  color ? "color, " : "",
-	  depth ? "depth, " : "",
-	  stencil ? "stencil" : "",
-	  x0, y0,
-	  x1, y1);
-   */
-
-   cso_save_blend(st->cso_context);
-   cso_save_stencil_ref(st->cso_context);
-   cso_save_depth_stencil_alpha(st->cso_context);
-   cso_save_rasterizer(st->cso_context);
-   cso_save_viewport(st->cso_context);
-   cso_save_clip(st->cso_context);
-   cso_save_fragment_shader(st->cso_context);
-   cso_save_vertex_shader(st->cso_context);
-   cso_save_vertex_elements(st->cso_context);
-   cso_save_vertex_buffers(st->cso_context);
-
-   /* blend state: RGBA masking */
-   {
-      struct pipe_blend_state blend;
-      memset(&blend, 0, sizeof(blend));
-      blend.rt[0].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
-      blend.rt[0].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
-      blend.rt[0].rgb_dst_factor = PIPE_BLENDFACTOR_ZERO;
-      blend.rt[0].alpha_dst_factor = PIPE_BLENDFACTOR_ZERO;
-      if (color) {
-         if (ctx->Color.ColorMask[0][0])
-            blend.rt[0].colormask |= PIPE_MASK_R;
-         if (ctx->Color.ColorMask[0][1])
-            blend.rt[0].colormask |= PIPE_MASK_G;
-         if (ctx->Color.ColorMask[0][2])
-            blend.rt[0].colormask |= PIPE_MASK_B;
-         if (ctx->Color.ColorMask[0][3])
-            blend.rt[0].colormask |= PIPE_MASK_A;
-         if (st->ctx->Color.DitherFlag)
-            blend.dither = 1;
-      }
-      cso_set_blend(st->cso_context, &blend);
-   }
-
-   /* depth_stencil state: always pass/set to ref value */
-   {
-      struct pipe_depth_stencil_alpha_state depth_stencil;
-      memset(&depth_stencil, 0, sizeof(depth_stencil));
-      if (depth) {
-         depth_stencil.depth.enabled = 1;
-         depth_stencil.depth.writemask = 1;
-         depth_stencil.depth.func = PIPE_FUNC_ALWAYS;
-      }
-
-      if (stencil) {
-         struct pipe_stencil_ref stencil_ref;
-         memset(&stencil_ref, 0, sizeof(stencil_ref));
-         depth_stencil.stencil[0].enabled = 1;
-         depth_stencil.stencil[0].func = PIPE_FUNC_ALWAYS;
-         depth_stencil.stencil[0].fail_op = PIPE_STENCIL_OP_REPLACE;
-         depth_stencil.stencil[0].zpass_op = PIPE_STENCIL_OP_REPLACE;
-         depth_stencil.stencil[0].zfail_op = PIPE_STENCIL_OP_REPLACE;
-         depth_stencil.stencil[0].valuemask = 0xff;
-         depth_stencil.stencil[0].writemask = ctx->Stencil.WriteMask[0] & 0xff;
-         stencil_ref.ref_value[0] = ctx->Stencil.Clear;
-         cso_set_stencil_ref(st->cso_context, &stencil_ref);
-      }
-
-      cso_set_depth_stencil_alpha(st->cso_context, &depth_stencil);
-   }
-
-   cso_set_vertex_elements(st->cso_context, 2, st->velems_util_draw);
-
-   cso_set_rasterizer(st->cso_context, &st->clear.raster);
-
-   /* viewport state: viewport matching window dims */
-   {
-      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
-      struct pipe_viewport_state vp;
-      vp.scale[0] = 0.5f * fb_width;
-      vp.scale[1] = fb_height * (invert ? -0.5f : 0.5f);
-      vp.scale[2] = 1.0f;
-      vp.scale[3] = 1.0f;
-      vp.translate[0] = 0.5f * fb_width;
-      vp.translate[1] = 0.5f * fb_height;
-      vp.translate[2] = 0.0f;
-      vp.translate[3] = 0.0f;
-      cso_set_viewport(st->cso_context, &vp);
-   }
-
-   cso_set_clip(st->cso_context, &st->clear.clip);
-   cso_set_fragment_shader_handle(st->cso_context, st->clear.fs);
-   cso_set_vertex_shader_handle(st->cso_context, st->clear.vs);
-
-   /* draw quad matching scissor rect (XXX verify coord round-off) */
-   draw_quad(st, x0, y0, x1, y1,
-             (GLfloat) ctx->Depth.Clear, ctx->Color.ClearColor);
-
-   /* Restore pipe state */
-   cso_restore_blend(st->cso_context);
-   cso_restore_stencil_ref(st->cso_context);
-   cso_restore_depth_stencil_alpha(st->cso_context);
-   cso_restore_rasterizer(st->cso_context);
-   cso_restore_viewport(st->cso_context);
-   cso_restore_clip(st->cso_context);
-   cso_restore_fragment_shader(st->cso_context);
-   cso_restore_vertex_shader(st->cso_context);
-   cso_restore_vertex_elements(st->cso_context);
-   cso_restore_vertex_buffers(st->cso_context);
-}
-
-
-/**
- * Determine if we need to clear the depth buffer by drawing a quad.
- */
-static INLINE GLboolean
-check_clear_color_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb)
-{
-   if (ctx->Scissor.Enabled &&
-       (ctx->Scissor.X != 0 ||
-        ctx->Scissor.Y != 0 ||
-        ctx->Scissor.Width < rb->Width ||
-        ctx->Scissor.Height < rb->Height))
-      return GL_TRUE;
-
-   if (!ctx->Color.ColorMask[0][0] ||
-       !ctx->Color.ColorMask[0][1] ||
-       !ctx->Color.ColorMask[0][2] ||
-       !ctx->Color.ColorMask[0][3])
-      return GL_TRUE;
-
-   return GL_FALSE;
-}
-
-
-/**
- * Determine if we need to clear the combiend depth/stencil buffer by
- * drawing a quad.
- */
-static INLINE GLboolean
-check_clear_depth_stencil_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb)
-{
-   const GLuint stencilMax = 0xff;
-   GLboolean maskStencil
-      = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax;
-
-   assert(rb->Format == MESA_FORMAT_S8 ||
-          rb->Format == MESA_FORMAT_Z24_S8 ||
-          rb->Format == MESA_FORMAT_S8_Z24);
-
-   if (ctx->Scissor.Enabled &&
-       (ctx->Scissor.X != 0 ||
-        ctx->Scissor.Y != 0 ||
-        ctx->Scissor.Width < rb->Width ||
-        ctx->Scissor.Height < rb->Height))
-      return GL_TRUE;
-
-   if (maskStencil)
-      return GL_TRUE;
-
-   return GL_FALSE;
-}
-
-
-/**
- * Determine if we need to clear the depth buffer by drawing a quad.
- */
-static INLINE GLboolean
-check_clear_depth_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                            boolean ds_separate)
-{
-   const struct st_renderbuffer *strb = st_renderbuffer(rb);
-   const GLboolean isDS = util_format_is_depth_and_stencil(strb->surface->format);
-
-   if (ctx->Scissor.Enabled &&
-       (ctx->Scissor.X != 0 ||
-        ctx->Scissor.Y != 0 ||
-        ctx->Scissor.Width < rb->Width ||
-        ctx->Scissor.Height < rb->Height))
-      return GL_TRUE;
-
-   if (!ds_separate && isDS && ctx->DrawBuffer->Visual.stencilBits > 0)
-      return GL_TRUE;
-
-   return GL_FALSE;
-}
-
-
-/**
- * Determine if we need to clear the stencil buffer by drawing a quad.
- */
-static INLINE GLboolean
-check_clear_stencil_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                              boolean ds_separate)
-{
-   const struct st_renderbuffer *strb = st_renderbuffer(rb);
-   const GLboolean isDS = util_format_is_depth_and_stencil(strb->surface->format);
-   const GLuint stencilMax = 0xff;
-   const GLboolean maskStencil
-      = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax;
-
-   assert(rb->Format == MESA_FORMAT_S8 ||
-          rb->Format == MESA_FORMAT_Z24_S8 ||
-          rb->Format == MESA_FORMAT_S8_Z24);
-
-   if (maskStencil) 
-      return GL_TRUE;
-
-   if (ctx->Scissor.Enabled &&
-       (ctx->Scissor.X != 0 ||
-        ctx->Scissor.Y != 0 ||
-        ctx->Scissor.Width < rb->Width ||
-        ctx->Scissor.Height < rb->Height))
-      return GL_TRUE;
-
-   /* This is correct, but it is necessary to look at the depth clear
-    * value held in the surface when it comes time to issue the clear,
-    * rather than taking depth and stencil clear values from the
-    * current state.
-    */
-   if (!ds_separate && isDS && ctx->DrawBuffer->Visual.depthBits > 0)
-      return GL_TRUE;
-
-   return GL_FALSE;
-}
-
-
-
-/**
- * Called when we need to flush.
- */
-void
-st_flush_clear(struct st_context *st)
-{
-   /* Release vertex buffer to avoid synchronous rendering if we were
-    * to map it in the next frame.
-    */
-   pipe_resource_reference(&st->clear.vbuf, NULL);
-   st->clear.vbuf_slot = 0;
-}
- 
-
-
-/**
- * Called via ctx->Driver.Clear()
- */
-static void
-st_Clear(struct gl_context *ctx, GLbitfield mask)
-{
-   static const GLbitfield BUFFER_BITS_DS
-      = (BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL);
-   struct st_context *st = st_context(ctx);
-   struct gl_renderbuffer *depthRb
-      = ctx->DrawBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
-   struct gl_renderbuffer *stencilRb
-      = ctx->DrawBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
-   GLbitfield quad_buffers = 0x0;
-   GLbitfield clear_buffers = 0x0;
-   GLuint i;
-
-   /* This makes sure the pipe has the latest scissor, etc values */
-   st_validate_state( st );
-
-   if (mask & BUFFER_BITS_COLOR) {
-      for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
-         GLuint b = ctx->DrawBuffer->_ColorDrawBufferIndexes[i];
-
-         if (mask & (1 << b)) {
-            struct gl_renderbuffer *rb
-               = ctx->DrawBuffer->Attachment[b].Renderbuffer;
-            struct st_renderbuffer *strb = st_renderbuffer(rb);
-
-            if (!strb || !strb->surface)
-               continue;
-
-            if (check_clear_color_with_quad( ctx, rb ))
-               quad_buffers |= PIPE_CLEAR_COLOR;
-            else
-               clear_buffers |= PIPE_CLEAR_COLOR;
-         }
-      }
-   }
-
-   if ((mask & BUFFER_BITS_DS) == BUFFER_BITS_DS && depthRb == stencilRb) {
-      /* clearing combined depth + stencil */
-      struct st_renderbuffer *strb = st_renderbuffer(depthRb);
-
-      if (strb->surface) {
-         if (check_clear_depth_stencil_with_quad(ctx, depthRb))
-            quad_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
-         else
-            clear_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
-      }
-   }
-   else {
-      /* separate depth/stencil clears */
-      /* I don't think truly separate buffers are actually possible in gallium or hw? */
-      if (mask & BUFFER_BIT_DEPTH) {
-         struct st_renderbuffer *strb = st_renderbuffer(depthRb);
-
-         if (strb->surface) {
-            if (check_clear_depth_with_quad(ctx, depthRb,
-                                            st->clear.enable_ds_separate))
-               quad_buffers |= PIPE_CLEAR_DEPTH;
-            else
-               clear_buffers |= PIPE_CLEAR_DEPTH;
-         }
-      }
-      if (mask & BUFFER_BIT_STENCIL) {
-         struct st_renderbuffer *strb = st_renderbuffer(stencilRb);
-
-         if (strb->surface) {
-            if (check_clear_stencil_with_quad(ctx, stencilRb,
-                                              st->clear.enable_ds_separate))
-               quad_buffers |= PIPE_CLEAR_STENCIL;
-            else
-               clear_buffers |= PIPE_CLEAR_STENCIL;
-         }
-      }
-   }
-
-   /*
-    * If we're going to use clear_with_quad() for any reason, use it for
-    * everything possible.
-    */
-   if (quad_buffers) {
-      quad_buffers |= clear_buffers;
-      clear_with_quad(ctx,
-                      quad_buffers & PIPE_CLEAR_COLOR,
-                      quad_buffers & PIPE_CLEAR_DEPTH,
-                      quad_buffers & PIPE_CLEAR_STENCIL);
-   } else if (clear_buffers) {
-      /* driver cannot know it can clear everything if the buffer
-       * is a combined depth/stencil buffer but this wasn't actually
-       * required from the visual. Hence fix this up to avoid potential
-       * read-modify-write in the driver.
-       */
-      if ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) &&
-          ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) != PIPE_CLEAR_DEPTHSTENCIL) &&
-          (depthRb == stencilRb) &&
-          (ctx->DrawBuffer->Visual.depthBits == 0 ||
-           ctx->DrawBuffer->Visual.stencilBits == 0))
-         clear_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
-      st->pipe->clear(st->pipe, clear_buffers, ctx->Color.ClearColor,
-                      ctx->Depth.Clear, ctx->Stencil.Clear);
-   }
-   if (mask & BUFFER_BIT_ACCUM)
-      st_clear_accum_buffer(ctx,
-                            ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
-}
-
-
-void
-st_init_clear_functions(struct dd_function_table *functions)
-{
-   functions->Clear = st_Clear;
-}
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 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 TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+ /*
+  * Authors:
+  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Brian Paul
+  *   Michel Dänzer
+  */
+
+#include "main/glheader.h"
+#include "main/formats.h"
+#include "main/macros.h"
+#include "program/prog_instruction.h"
+#include "st_context.h"
+#include "st_atom.h"
+#include "st_cb_accum.h"
+#include "st_cb_clear.h"
+#include "st_cb_fbo.h"
+#include "st_program.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_shader_tokens.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "util/u_format.h"
+#include "util/u_inlines.h"
+#include "util/u_simple_shaders.h"
+#include "util/u_draw_quad.h"
+
+#include "cso_cache/cso_context.h"
+
+
+/**
+ * Do per-context initialization for glClear.
+ */
+void
+st_init_clear(struct st_context *st)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_screen *pscreen = st->pipe->screen;
+
+   memset(&st->clear, 0, sizeof(st->clear));
+
+   st->clear.raster.gl_rasterization_rules = 1;
+   st->clear.enable_ds_separate = pscreen->get_param(pscreen, PIPE_CAP_DEPTHSTENCIL_CLEAR_SEPARATE);
+
+   /* fragment shader state: color pass-through program */
+   st->clear.fs = util_make_fragment_passthrough_shader(pipe);
+
+   /* vertex shader state: color/position pass-through */
+   {
+      const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
+                                      TGSI_SEMANTIC_COLOR };
+      const uint semantic_indexes[] = { 0, 0 };
+      st->clear.vs = util_make_vertex_passthrough_shader(pipe, 2,
+                                                         semantic_names,
+                                                         semantic_indexes);
+   }
+}
+
+
+/**
+ * Free per-context state for glClear.
+ */
+void
+st_destroy_clear(struct st_context *st)
+{
+   if (st->clear.fs) {
+      cso_delete_fragment_shader(st->cso_context, st->clear.fs);
+      st->clear.fs = NULL;
+   }
+   if (st->clear.vs) {
+      cso_delete_vertex_shader(st->cso_context, st->clear.vs);
+      st->clear.vs = NULL;
+   }
+   if (st->clear.vbuf) {
+      pipe_resource_reference(&st->clear.vbuf, NULL);
+      st->clear.vbuf = NULL;
+   }
+}
+
+
+/**
+ * Draw a screen-aligned quadrilateral.
+ * Coords are clip coords with y=0=bottom.
+ */
+static void
+draw_quad(struct st_context *st,
+          float x0, float y0, float x1, float y1, GLfloat z,
+          const GLfloat color[4])
+{
+   struct pipe_context *pipe = st->pipe;
+
+   /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as
+    * no_flush) updates to buffers where we know there is no conflict
+    * with previous data.  Currently using max_slots > 1 will cause
+    * synchronous rendering if the driver flushes its command buffers
+    * between one bitmap and the next.  Our flush hook below isn't
+    * sufficient to catch this as the driver doesn't tell us when it
+    * flushes its own command buffers.  Until this gets fixed, pay the
+    * price of allocating a new buffer for each bitmap cache-flush to
+    * avoid synchronous rendering.
+    */
+   const GLuint max_slots = 1; /* 1024 / sizeof(st->clear.vertices); */
+   GLuint i;
+
+   if (st->clear.vbuf_slot >= max_slots) {
+      pipe_resource_reference(&st->clear.vbuf, NULL);
+      st->clear.vbuf_slot = 0;
+   }
+
+   if (!st->clear.vbuf) {
+      st->clear.vbuf = pipe_buffer_create(pipe->screen,
+                                          PIPE_BIND_VERTEX_BUFFER,
+                                          PIPE_USAGE_STREAM,
+                                          max_slots * sizeof(st->clear.vertices));
+   }
+
+   /* positions */
+   st->clear.vertices[0][0][0] = x0;
+   st->clear.vertices[0][0][1] = y0;
+
+   st->clear.vertices[1][0][0] = x1;
+   st->clear.vertices[1][0][1] = y0;
+
+   st->clear.vertices[2][0][0] = x1;
+   st->clear.vertices[2][0][1] = y1;
+
+   st->clear.vertices[3][0][0] = x0;
+   st->clear.vertices[3][0][1] = y1;
+
+   /* same for all verts: */
+   for (i = 0; i < 4; i++) {
+      st->clear.vertices[i][0][2] = z;
+      st->clear.vertices[i][0][3] = 1.0;
+      st->clear.vertices[i][1][0] = color[0];
+      st->clear.vertices[i][1][1] = color[1];
+      st->clear.vertices[i][1][2] = color[2];
+      st->clear.vertices[i][1][3] = color[3];
+   }
+
+   /* put vertex data into vbuf */
+   pipe_buffer_write_nooverlap(st->pipe, st->clear.vbuf,
+                                           st->clear.vbuf_slot
+                                             * sizeof(st->clear.vertices),
+                                           sizeof(st->clear.vertices),
+                                           st->clear.vertices);
+
+   /* draw */
+   util_draw_vertex_buffer(pipe,
+                           st->cso_context,
+                           st->clear.vbuf, 
+                           st->clear.vbuf_slot * sizeof(st->clear.vertices),
+                           PIPE_PRIM_TRIANGLE_FAN,
+                           4,  /* verts */
+                           2); /* attribs/vert */
+
+   /* Increment slot */
+   st->clear.vbuf_slot++;
+}
+
+
+
+/**
+ * Do glClear by drawing a quadrilateral.
+ * The vertices of the quad will be computed from the
+ * ctx->DrawBuffer->_X/Ymin/max fields.
+ */
+static void
+clear_with_quad(struct gl_context *ctx,
+                GLboolean color, GLboolean depth, GLboolean stencil)
+{
+   struct st_context *st = st_context(ctx);
+   const struct gl_framebuffer *fb = ctx->DrawBuffer;
+   const GLfloat fb_width = (GLfloat) fb->Width;
+   const GLfloat fb_height = (GLfloat) fb->Height;
+   const GLfloat x0 = (GLfloat) ctx->DrawBuffer->_Xmin / fb_width * 2.0f - 1.0f;
+   const GLfloat x1 = (GLfloat) ctx->DrawBuffer->_Xmax / fb_width * 2.0f - 1.0f;
+   const GLfloat y0 = (GLfloat) ctx->DrawBuffer->_Ymin / fb_height * 2.0f - 1.0f;
+   const GLfloat y1 = (GLfloat) ctx->DrawBuffer->_Ymax / fb_height * 2.0f - 1.0f;
+
+   /*
+   printf("%s %s%s%s %f,%f %f,%f\n", __FUNCTION__, 
+	  color ? "color, " : "",
+	  depth ? "depth, " : "",
+	  stencil ? "stencil" : "",
+	  x0, y0,
+	  x1, y1);
+   */
+
+   cso_save_blend(st->cso_context);
+   cso_save_stencil_ref(st->cso_context);
+   cso_save_depth_stencil_alpha(st->cso_context);
+   cso_save_rasterizer(st->cso_context);
+   cso_save_viewport(st->cso_context);
+   cso_save_clip(st->cso_context);
+   cso_save_fragment_shader(st->cso_context);
+   cso_save_vertex_shader(st->cso_context);
+   cso_save_vertex_elements(st->cso_context);
+   cso_save_vertex_buffers(st->cso_context);
+
+   /* blend state: RGBA masking */
+   {
+      struct pipe_blend_state blend;
+      memset(&blend, 0, sizeof(blend));
+      blend.rt[0].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+      blend.rt[0].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+      blend.rt[0].rgb_dst_factor = PIPE_BLENDFACTOR_ZERO;
+      blend.rt[0].alpha_dst_factor = PIPE_BLENDFACTOR_ZERO;
+      if (color) {
+         if (ctx->Color.ColorMask[0][0])
+            blend.rt[0].colormask |= PIPE_MASK_R;
+         if (ctx->Color.ColorMask[0][1])
+            blend.rt[0].colormask |= PIPE_MASK_G;
+         if (ctx->Color.ColorMask[0][2])
+            blend.rt[0].colormask |= PIPE_MASK_B;
+         if (ctx->Color.ColorMask[0][3])
+            blend.rt[0].colormask |= PIPE_MASK_A;
+         if (st->ctx->Color.DitherFlag)
+            blend.dither = 1;
+      }
+      cso_set_blend(st->cso_context, &blend);
+   }
+
+   /* depth_stencil state: always pass/set to ref value */
+   {
+      struct pipe_depth_stencil_alpha_state depth_stencil;
+      memset(&depth_stencil, 0, sizeof(depth_stencil));
+      if (depth) {
+         depth_stencil.depth.enabled = 1;
+         depth_stencil.depth.writemask = 1;
+         depth_stencil.depth.func = PIPE_FUNC_ALWAYS;
+      }
+
+      if (stencil) {
+         struct pipe_stencil_ref stencil_ref;
+         memset(&stencil_ref, 0, sizeof(stencil_ref));
+         depth_stencil.stencil[0].enabled = 1;
+         depth_stencil.stencil[0].func = PIPE_FUNC_ALWAYS;
+         depth_stencil.stencil[0].fail_op = PIPE_STENCIL_OP_REPLACE;
+         depth_stencil.stencil[0].zpass_op = PIPE_STENCIL_OP_REPLACE;
+         depth_stencil.stencil[0].zfail_op = PIPE_STENCIL_OP_REPLACE;
+         depth_stencil.stencil[0].valuemask = 0xff;
+         depth_stencil.stencil[0].writemask = ctx->Stencil.WriteMask[0] & 0xff;
+         stencil_ref.ref_value[0] = ctx->Stencil.Clear;
+         cso_set_stencil_ref(st->cso_context, &stencil_ref);
+      }
+
+      cso_set_depth_stencil_alpha(st->cso_context, &depth_stencil);
+   }
+
+   cso_set_vertex_elements(st->cso_context, 2, st->velems_util_draw);
+
+   cso_set_rasterizer(st->cso_context, &st->clear.raster);
+
+   /* viewport state: viewport matching window dims */
+   {
+      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
+      struct pipe_viewport_state vp;
+      vp.scale[0] = 0.5f * fb_width;
+      vp.scale[1] = fb_height * (invert ? -0.5f : 0.5f);
+      vp.scale[2] = 1.0f;
+      vp.scale[3] = 1.0f;
+      vp.translate[0] = 0.5f * fb_width;
+      vp.translate[1] = 0.5f * fb_height;
+      vp.translate[2] = 0.0f;
+      vp.translate[3] = 0.0f;
+      cso_set_viewport(st->cso_context, &vp);
+   }
+
+   cso_set_clip(st->cso_context, &st->clear.clip);
+   cso_set_fragment_shader_handle(st->cso_context, st->clear.fs);
+   cso_set_vertex_shader_handle(st->cso_context, st->clear.vs);
+
+   /* draw quad matching scissor rect (XXX verify coord round-off) */
+   draw_quad(st, x0, y0, x1, y1,
+             (GLfloat) ctx->Depth.Clear, ctx->Color.ClearColor);
+
+   /* Restore pipe state */
+   cso_restore_blend(st->cso_context);
+   cso_restore_stencil_ref(st->cso_context);
+   cso_restore_depth_stencil_alpha(st->cso_context);
+   cso_restore_rasterizer(st->cso_context);
+   cso_restore_viewport(st->cso_context);
+   cso_restore_clip(st->cso_context);
+   cso_restore_fragment_shader(st->cso_context);
+   cso_restore_vertex_shader(st->cso_context);
+   cso_restore_vertex_elements(st->cso_context);
+   cso_restore_vertex_buffers(st->cso_context);
+}
+
+
+/**
+ * Determine if we need to clear the depth buffer by drawing a quad.
+ */
+static INLINE GLboolean
+check_clear_color_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb)
+{
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
+      return GL_TRUE;
+
+   if (!ctx->Color.ColorMask[0][0] ||
+       !ctx->Color.ColorMask[0][1] ||
+       !ctx->Color.ColorMask[0][2] ||
+       !ctx->Color.ColorMask[0][3])
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+/**
+ * Determine if we need to clear the combiend depth/stencil buffer by
+ * drawing a quad.
+ */
+static INLINE GLboolean
+check_clear_depth_stencil_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb)
+{
+   const GLuint stencilMax = 0xff;
+   GLboolean maskStencil
+      = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax;
+
+   assert(rb->Format == MESA_FORMAT_S8 ||
+          rb->Format == MESA_FORMAT_Z24_S8 ||
+          rb->Format == MESA_FORMAT_S8_Z24);
+
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
+      return GL_TRUE;
+
+   if (maskStencil)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+/**
+ * Determine if we need to clear the depth buffer by drawing a quad.
+ */
+static INLINE GLboolean
+check_clear_depth_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                            boolean ds_separate)
+{
+   const struct st_renderbuffer *strb = st_renderbuffer(rb);
+   const GLboolean isDS = util_format_is_depth_and_stencil(strb->surface->format);
+
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
+      return GL_TRUE;
+
+   if (!ds_separate && isDS && ctx->DrawBuffer->Visual.stencilBits > 0)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+/**
+ * Determine if we need to clear the stencil buffer by drawing a quad.
+ */
+static INLINE GLboolean
+check_clear_stencil_with_quad(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                              boolean ds_separate)
+{
+   const struct st_renderbuffer *strb = st_renderbuffer(rb);
+   const GLboolean isDS = util_format_is_depth_and_stencil(strb->surface->format);
+   const GLuint stencilMax = 0xff;
+   const GLboolean maskStencil
+      = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax;
+
+   assert(rb->Format == MESA_FORMAT_S8 ||
+          rb->Format == MESA_FORMAT_Z24_S8 ||
+          rb->Format == MESA_FORMAT_S8_Z24);
+
+   if (maskStencil) 
+      return GL_TRUE;
+
+   if (ctx->Scissor.Enabled &&
+       (ctx->Scissor.X != 0 ||
+        ctx->Scissor.Y != 0 ||
+        ctx->Scissor.Width < rb->Width ||
+        ctx->Scissor.Height < rb->Height))
+      return GL_TRUE;
+
+   /* This is correct, but it is necessary to look at the depth clear
+    * value held in the surface when it comes time to issue the clear,
+    * rather than taking depth and stencil clear values from the
+    * current state.
+    */
+   if (!ds_separate && isDS && ctx->DrawBuffer->Visual.depthBits > 0)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+
+/**
+ * Called when we need to flush.
+ */
+void
+st_flush_clear(struct st_context *st)
+{
+   /* Release vertex buffer to avoid synchronous rendering if we were
+    * to map it in the next frame.
+    */
+   pipe_resource_reference(&st->clear.vbuf, NULL);
+   st->clear.vbuf_slot = 0;
+}
+ 
+
+
+/**
+ * Called via ctx->Driver.Clear()
+ */
+static void
+st_Clear(struct gl_context *ctx, GLbitfield mask)
+{
+   static const GLbitfield BUFFER_BITS_DS
+      = (BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL);
+   struct st_context *st = st_context(ctx);
+   struct gl_renderbuffer *depthRb
+      = ctx->DrawBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
+   struct gl_renderbuffer *stencilRb
+      = ctx->DrawBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
+   GLbitfield quad_buffers = 0x0;
+   GLbitfield clear_buffers = 0x0;
+   GLuint i;
+
+   /* This makes sure the pipe has the latest scissor, etc values */
+   st_validate_state( st );
+
+   if (mask & BUFFER_BITS_COLOR) {
+      for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
+         GLuint b = ctx->DrawBuffer->_ColorDrawBufferIndexes[i];
+
+         if (mask & (1 << b)) {
+            struct gl_renderbuffer *rb
+               = ctx->DrawBuffer->Attachment[b].Renderbuffer;
+            struct st_renderbuffer *strb = st_renderbuffer(rb);
+
+            if (!strb || !strb->surface)
+               continue;
+
+            if (check_clear_color_with_quad( ctx, rb ))
+               quad_buffers |= PIPE_CLEAR_COLOR;
+            else
+               clear_buffers |= PIPE_CLEAR_COLOR;
+         }
+      }
+   }
+
+   if ((mask & BUFFER_BITS_DS) == BUFFER_BITS_DS && depthRb == stencilRb) {
+      /* clearing combined depth + stencil */
+      struct st_renderbuffer *strb = st_renderbuffer(depthRb);
+
+      if (strb->surface) {
+         if (check_clear_depth_stencil_with_quad(ctx, depthRb))
+            quad_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
+         else
+            clear_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
+      }
+   }
+   else {
+      /* separate depth/stencil clears */
+      /* I don't think truly separate buffers are actually possible in gallium or hw? */
+      if (mask & BUFFER_BIT_DEPTH) {
+         struct st_renderbuffer *strb = st_renderbuffer(depthRb);
+
+         if (strb->surface) {
+            if (check_clear_depth_with_quad(ctx, depthRb,
+                                            st->clear.enable_ds_separate))
+               quad_buffers |= PIPE_CLEAR_DEPTH;
+            else
+               clear_buffers |= PIPE_CLEAR_DEPTH;
+         }
+      }
+      if (mask & BUFFER_BIT_STENCIL) {
+         struct st_renderbuffer *strb = st_renderbuffer(stencilRb);
+
+         if (strb->surface) {
+            if (check_clear_stencil_with_quad(ctx, stencilRb,
+                                              st->clear.enable_ds_separate))
+               quad_buffers |= PIPE_CLEAR_STENCIL;
+            else
+               clear_buffers |= PIPE_CLEAR_STENCIL;
+         }
+      }
+   }
+
+   /*
+    * If we're going to use clear_with_quad() for any reason, use it for
+    * everything possible.
+    */
+   if (quad_buffers) {
+      quad_buffers |= clear_buffers;
+      clear_with_quad(ctx,
+                      quad_buffers & PIPE_CLEAR_COLOR,
+                      quad_buffers & PIPE_CLEAR_DEPTH,
+                      quad_buffers & PIPE_CLEAR_STENCIL);
+   } else if (clear_buffers) {
+      /* driver cannot know it can clear everything if the buffer
+       * is a combined depth/stencil buffer but this wasn't actually
+       * required from the visual. Hence fix this up to avoid potential
+       * read-modify-write in the driver.
+       */
+      if ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) &&
+          ((clear_buffers & PIPE_CLEAR_DEPTHSTENCIL) != PIPE_CLEAR_DEPTHSTENCIL) &&
+          (depthRb == stencilRb) &&
+          (ctx->DrawBuffer->Visual.depthBits == 0 ||
+           ctx->DrawBuffer->Visual.stencilBits == 0))
+         clear_buffers |= PIPE_CLEAR_DEPTHSTENCIL;
+      st->pipe->clear(st->pipe, clear_buffers, ctx->Color.ClearColor,
+                      ctx->Depth.Clear, ctx->Stencil.Clear);
+   }
+   if (mask & BUFFER_BIT_ACCUM)
+      st_clear_accum_buffer(ctx,
+                            ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
+}
+
+
+void
+st_init_clear_functions(struct dd_function_table *functions)
+{
+   functions->Clear = st_Clear;
+}
diff --git a/mesalib/src/mesa/state_tracker/st_cb_drawtex.c b/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
index 86ceb9d78..db299eb8d 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
@@ -1,307 +1,307 @@
-/**************************************************************************
- * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- **************************************************************************/
-
-
-/**
- * Implementation of glDrawTex() for GL_OES_draw_tex
- */
-
-
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-#include "program/program.h"
-#include "program/prog_print.h"
-
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_cb_drawtex.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-#include "pipe/p_shader_tokens.h"
-#include "util/u_draw_quad.h"
-#include "util/u_simple_shaders.h"
-
-#include "cso_cache/cso_context.h"
-
-
-#if FEATURE_OES_draw_texture
-
-
-struct cached_shader
-{
-   void *handle;
-
-   uint num_attribs;
-   uint semantic_names[2 + MAX_TEXTURE_UNITS];
-   uint semantic_indexes[2 + MAX_TEXTURE_UNITS];
-};
-
-#define MAX_SHADERS (2 * MAX_TEXTURE_UNITS)
-
-/**
- * Simple linear list cache.
- * Most of the time there'll only be one cached shader.
- */
-static struct cached_shader CachedShaders[MAX_SHADERS];
-static GLuint NumCachedShaders = 0;
-
-
-static void *
-lookup_shader(struct pipe_context *pipe,
-              uint num_attribs,
-              const uint *semantic_names,
-              const uint *semantic_indexes)
-{
-   GLuint i, j;
-
-   /* look for existing shader with same attributes */
-   for (i = 0; i < NumCachedShaders; i++) {
-      if (CachedShaders[i].num_attribs == num_attribs) {
-         GLboolean match = GL_TRUE;
-         for (j = 0; j < num_attribs; j++) {
-            if (semantic_names[j] != CachedShaders[i].semantic_names[j] ||
-                semantic_indexes[j] != CachedShaders[i].semantic_indexes[j]) {
-               match = GL_FALSE;
-               break;
-            }
-         }
-         if (match)
-            return CachedShaders[i].handle;
-      }
-   }
-
-   /* not found - create new one now */
-   if (NumCachedShaders >= MAX_SHADERS) {
-      return NULL;
-   }
-
-   CachedShaders[i].num_attribs = num_attribs;
-   for (j = 0; j < num_attribs; j++) {
-      CachedShaders[i].semantic_names[j] = semantic_names[j];
-      CachedShaders[i].semantic_indexes[j] = semantic_indexes[j];
-   }
-
-   CachedShaders[i].handle =
-      util_make_vertex_passthrough_shader(pipe,
-                                          num_attribs,
-                                          semantic_names,
-                                          semantic_indexes);
-   NumCachedShaders++;
-
-   return CachedShaders[i].handle;
-}
-
-static void
-st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
-           GLfloat width, GLfloat height)
-{
-   struct st_context *st = ctx->st;
-   struct pipe_context *pipe = st->pipe;
-   struct cso_context *cso = ctx->st->cso_context;
-   struct pipe_resource *vbuffer;
-   struct pipe_transfer *vbuffer_transfer;
-   GLuint i, numTexCoords, numAttribs;
-   GLboolean emitColor;
-   uint semantic_names[2 + MAX_TEXTURE_UNITS];
-   uint semantic_indexes[2 + MAX_TEXTURE_UNITS];
-   struct pipe_vertex_element velements[2 + MAX_TEXTURE_UNITS];
-   GLbitfield inputs = VERT_BIT_POS;
-
-   st_validate_state(st);
-
-   /* determine if we need vertex color */
-   if (ctx->FragmentProgram._Current->Base.InputsRead & FRAG_BIT_COL0)
-      emitColor = GL_TRUE;
-   else
-      emitColor = GL_FALSE;
-
-   /* determine how many enabled sets of texcoords */
-   numTexCoords = 0;
-   for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-      if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
-         inputs |= VERT_BIT_TEX(i);
-         numTexCoords++;
-      }
-   }
-
-   /* total number of attributes per vertex */
-   numAttribs = 1 + emitColor + numTexCoords;
-
-
-   /* create the vertex buffer */
-   vbuffer = pipe_buffer_create(pipe->screen, PIPE_BIND_VERTEX_BUFFER,
-                                PIPE_USAGE_STREAM,
-                                numAttribs * 4 * 4 * sizeof(GLfloat));
-
-   /* load vertex buffer */
-   {
-#define SET_ATTRIB(VERT, ATTR, X, Y, Z, W)                              \
-      do {                                                              \
-         GLuint k = (((VERT) * numAttribs + (ATTR)) * 4);               \
-         assert(k < 4 * 4 * numAttribs);                                \
-         vbuf[k + 0] = X;                                               \
-         vbuf[k + 1] = Y;                                               \
-         vbuf[k + 2] = Z;                                               \
-         vbuf[k + 3] = W;                                               \
-      } while (0)
-
-      const GLfloat x0 = x, y0 = y, x1 = x + width, y1 = y + height;
-      GLfloat *vbuf = (GLfloat *) pipe_buffer_map(pipe, vbuffer,
-                                                  PIPE_TRANSFER_WRITE,
-                                                  &vbuffer_transfer);
-      GLuint attr;
-      
-      z = CLAMP(z, 0.0f, 1.0f);
-
-      /* positions (in clip coords) */
-      {
-         const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
-         const GLfloat fb_width = (GLfloat)fb->Width;
-         const GLfloat fb_height = (GLfloat)fb->Height;
-
-         const GLfloat clip_x0 = (GLfloat)(x0 / fb_width * 2.0 - 1.0);
-         const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
-         const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
-         const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
-
-         SET_ATTRIB(0, 0, clip_x0, clip_y0, z, 1.0f);   /* lower left */
-         SET_ATTRIB(1, 0, clip_x1, clip_y0, z, 1.0f);   /* lower right */
-         SET_ATTRIB(2, 0, clip_x1, clip_y1, z, 1.0f);   /* upper right */
-         SET_ATTRIB(3, 0, clip_x0, clip_y1, z, 1.0f);   /* upper left */
-
-         semantic_names[0] = TGSI_SEMANTIC_POSITION;
-         semantic_indexes[0] = 0;
-      }
-
-      /* colors */
-      if (emitColor) {
-         const GLfloat *c = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
-         SET_ATTRIB(0, 1, c[0], c[1], c[2], c[3]);
-         SET_ATTRIB(1, 1, c[0], c[1], c[2], c[3]);
-         SET_ATTRIB(2, 1, c[0], c[1], c[2], c[3]);
-         SET_ATTRIB(3, 1, c[0], c[1], c[2], c[3]);
-         semantic_names[1] = TGSI_SEMANTIC_COLOR;
-         semantic_indexes[1] = 0;
-         attr = 2;
-      }
-      else {
-         attr = 1;
-      }
-
-      /* texcoords */
-      for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-         if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
-            struct gl_texture_object *obj = ctx->Texture.Unit[i]._Current;
-            struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
-            const GLfloat wt = (GLfloat) img->Width;
-            const GLfloat ht = (GLfloat) img->Height;
-            const GLfloat s0 = obj->CropRect[0] / wt;
-            const GLfloat t0 = obj->CropRect[1] / ht;
-            const GLfloat s1 = (obj->CropRect[0] + obj->CropRect[2]) / wt;
-            const GLfloat t1 = (obj->CropRect[1] + obj->CropRect[3]) / ht;
-
-            /*printf("crop texcoords: %g, %g .. %g, %g\n", s0, t0, s1, t1);*/
-            SET_ATTRIB(0, attr, s0, t0, 0.0f, 1.0f);  /* lower left */
-            SET_ATTRIB(1, attr, s1, t0, 0.0f, 1.0f);  /* lower right */
-            SET_ATTRIB(2, attr, s1, t1, 0.0f, 1.0f);  /* upper right */
-            SET_ATTRIB(3, attr, s0, t1, 0.0f, 1.0f);  /* upper left */
-
-            semantic_names[attr] = TGSI_SEMANTIC_GENERIC;
-            semantic_indexes[attr] = 0;
-
-            attr++;
-         }
-      }
-
-      pipe_buffer_unmap(pipe, vbuffer_transfer);
-
-#undef SET_ATTRIB
-   }
-
-
-   cso_save_viewport(cso);
-   cso_save_vertex_shader(cso);
-   cso_save_vertex_elements(cso);
-   cso_save_vertex_buffers(cso);
-
-   {
-      void *vs = lookup_shader(pipe, numAttribs,
-                               semantic_names, semantic_indexes);
-      cso_set_vertex_shader_handle(cso, vs);
-   }
-
-   for (i = 0; i < numAttribs; i++) {
-      velements[i].src_offset = i * 4 * sizeof(float);
-      velements[i].instance_divisor = 0;
-      velements[i].vertex_buffer_index = 0;
-      velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
-   }
-   cso_set_vertex_elements(cso, numAttribs, velements);
-
-   /* viewport state: viewport matching window dims */
-   {
-      const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
-      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
-      const GLfloat width = (GLfloat)fb->Width;
-      const GLfloat height = (GLfloat)fb->Height;
-      struct pipe_viewport_state vp;
-      vp.scale[0] =  0.5f * width;
-      vp.scale[1] = height * (invert ? -0.5f : 0.5f);
-      vp.scale[2] = 1.0f;
-      vp.scale[3] = 1.0f;
-      vp.translate[0] = 0.5f * width;
-      vp.translate[1] = 0.5f * height;
-      vp.translate[2] = 0.0f;
-      vp.translate[3] = 0.0f;
-      cso_set_viewport(cso, &vp);
-   }
-
-
-   util_draw_vertex_buffer(pipe, cso, vbuffer,
-                           0,  /* offset */
-                           PIPE_PRIM_TRIANGLE_FAN,
-                           4,  /* verts */
-                           numAttribs); /* attribs/vert */
-
-
-   pipe_resource_reference(&vbuffer, NULL);
-
-   /* restore state */
-   cso_restore_viewport(cso);
-   cso_restore_vertex_shader(cso);
-   cso_restore_vertex_elements(cso);
-   cso_restore_vertex_buffers(cso);
-}
-
-
-void
-st_init_drawtex_functions(struct dd_function_table *functions)
-{
-   functions->DrawTex = st_DrawTex;
-}
-
-
-/**
- * Free any cached shaders
- */
-void
-st_destroy_drawtex(struct st_context *st)
-{
-   GLuint i;
-   for (i = 0; i < NumCachedShaders; i++) {
-      cso_delete_vertex_shader(st->cso_context, CachedShaders[i].handle);
-   }
-   NumCachedShaders = 0;
-}
-
-
-#endif /* FEATURE_OES_draw_texture */
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ **************************************************************************/
+
+
+/**
+ * Implementation of glDrawTex() for GL_OES_draw_tex
+ */
+
+
+
+#include "main/imports.h"
+#include "main/image.h"
+#include "main/macros.h"
+#include "main/mfeatures.h"
+#include "program/program.h"
+#include "program/prog_print.h"
+
+#include "st_context.h"
+#include "st_atom.h"
+#include "st_cb_drawtex.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "util/u_inlines.h"
+#include "pipe/p_shader_tokens.h"
+#include "util/u_draw_quad.h"
+#include "util/u_simple_shaders.h"
+
+#include "cso_cache/cso_context.h"
+
+
+#if FEATURE_OES_draw_texture
+
+
+struct cached_shader
+{
+   void *handle;
+
+   uint num_attribs;
+   uint semantic_names[2 + MAX_TEXTURE_UNITS];
+   uint semantic_indexes[2 + MAX_TEXTURE_UNITS];
+};
+
+#define MAX_SHADERS (2 * MAX_TEXTURE_UNITS)
+
+/**
+ * Simple linear list cache.
+ * Most of the time there'll only be one cached shader.
+ */
+static struct cached_shader CachedShaders[MAX_SHADERS];
+static GLuint NumCachedShaders = 0;
+
+
+static void *
+lookup_shader(struct pipe_context *pipe,
+              uint num_attribs,
+              const uint *semantic_names,
+              const uint *semantic_indexes)
+{
+   GLuint i, j;
+
+   /* look for existing shader with same attributes */
+   for (i = 0; i < NumCachedShaders; i++) {
+      if (CachedShaders[i].num_attribs == num_attribs) {
+         GLboolean match = GL_TRUE;
+         for (j = 0; j < num_attribs; j++) {
+            if (semantic_names[j] != CachedShaders[i].semantic_names[j] ||
+                semantic_indexes[j] != CachedShaders[i].semantic_indexes[j]) {
+               match = GL_FALSE;
+               break;
+            }
+         }
+         if (match)
+            return CachedShaders[i].handle;
+      }
+   }
+
+   /* not found - create new one now */
+   if (NumCachedShaders >= MAX_SHADERS) {
+      return NULL;
+   }
+
+   CachedShaders[i].num_attribs = num_attribs;
+   for (j = 0; j < num_attribs; j++) {
+      CachedShaders[i].semantic_names[j] = semantic_names[j];
+      CachedShaders[i].semantic_indexes[j] = semantic_indexes[j];
+   }
+
+   CachedShaders[i].handle =
+      util_make_vertex_passthrough_shader(pipe,
+                                          num_attribs,
+                                          semantic_names,
+                                          semantic_indexes);
+   NumCachedShaders++;
+
+   return CachedShaders[i].handle;
+}
+
+static void
+st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
+           GLfloat width, GLfloat height)
+{
+   struct st_context *st = ctx->st;
+   struct pipe_context *pipe = st->pipe;
+   struct cso_context *cso = ctx->st->cso_context;
+   struct pipe_resource *vbuffer;
+   struct pipe_transfer *vbuffer_transfer;
+   GLuint i, numTexCoords, numAttribs;
+   GLboolean emitColor;
+   uint semantic_names[2 + MAX_TEXTURE_UNITS];
+   uint semantic_indexes[2 + MAX_TEXTURE_UNITS];
+   struct pipe_vertex_element velements[2 + MAX_TEXTURE_UNITS];
+   GLbitfield inputs = VERT_BIT_POS;
+
+   st_validate_state(st);
+
+   /* determine if we need vertex color */
+   if (ctx->FragmentProgram._Current->Base.InputsRead & FRAG_BIT_COL0)
+      emitColor = GL_TRUE;
+   else
+      emitColor = GL_FALSE;
+
+   /* determine how many enabled sets of texcoords */
+   numTexCoords = 0;
+   for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+      if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
+         inputs |= VERT_BIT_TEX(i);
+         numTexCoords++;
+      }
+   }
+
+   /* total number of attributes per vertex */
+   numAttribs = 1 + emitColor + numTexCoords;
+
+
+   /* create the vertex buffer */
+   vbuffer = pipe_buffer_create(pipe->screen, PIPE_BIND_VERTEX_BUFFER,
+                                PIPE_USAGE_STREAM,
+                                numAttribs * 4 * 4 * sizeof(GLfloat));
+
+   /* load vertex buffer */
+   {
+#define SET_ATTRIB(VERT, ATTR, X, Y, Z, W)                              \
+      do {                                                              \
+         GLuint k = (((VERT) * numAttribs + (ATTR)) * 4);               \
+         assert(k < 4 * 4 * numAttribs);                                \
+         vbuf[k + 0] = X;                                               \
+         vbuf[k + 1] = Y;                                               \
+         vbuf[k + 2] = Z;                                               \
+         vbuf[k + 3] = W;                                               \
+      } while (0)
+
+      const GLfloat x0 = x, y0 = y, x1 = x + width, y1 = y + height;
+      GLfloat *vbuf = (GLfloat *) pipe_buffer_map(pipe, vbuffer,
+                                                  PIPE_TRANSFER_WRITE,
+                                                  &vbuffer_transfer);
+      GLuint attr;
+      
+      z = CLAMP(z, 0.0f, 1.0f);
+
+      /* positions (in clip coords) */
+      {
+         const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
+         const GLfloat fb_width = (GLfloat)fb->Width;
+         const GLfloat fb_height = (GLfloat)fb->Height;
+
+         const GLfloat clip_x0 = (GLfloat)(x0 / fb_width * 2.0 - 1.0);
+         const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0);
+         const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0);
+         const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0);
+
+         SET_ATTRIB(0, 0, clip_x0, clip_y0, z, 1.0f);   /* lower left */
+         SET_ATTRIB(1, 0, clip_x1, clip_y0, z, 1.0f);   /* lower right */
+         SET_ATTRIB(2, 0, clip_x1, clip_y1, z, 1.0f);   /* upper right */
+         SET_ATTRIB(3, 0, clip_x0, clip_y1, z, 1.0f);   /* upper left */
+
+         semantic_names[0] = TGSI_SEMANTIC_POSITION;
+         semantic_indexes[0] = 0;
+      }
+
+      /* colors */
+      if (emitColor) {
+         const GLfloat *c = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
+         SET_ATTRIB(0, 1, c[0], c[1], c[2], c[3]);
+         SET_ATTRIB(1, 1, c[0], c[1], c[2], c[3]);
+         SET_ATTRIB(2, 1, c[0], c[1], c[2], c[3]);
+         SET_ATTRIB(3, 1, c[0], c[1], c[2], c[3]);
+         semantic_names[1] = TGSI_SEMANTIC_COLOR;
+         semantic_indexes[1] = 0;
+         attr = 2;
+      }
+      else {
+         attr = 1;
+      }
+
+      /* texcoords */
+      for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+         if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
+            struct gl_texture_object *obj = ctx->Texture.Unit[i]._Current;
+            struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
+            const GLfloat wt = (GLfloat) img->Width;
+            const GLfloat ht = (GLfloat) img->Height;
+            const GLfloat s0 = obj->CropRect[0] / wt;
+            const GLfloat t0 = obj->CropRect[1] / ht;
+            const GLfloat s1 = (obj->CropRect[0] + obj->CropRect[2]) / wt;
+            const GLfloat t1 = (obj->CropRect[1] + obj->CropRect[3]) / ht;
+
+            /*printf("crop texcoords: %g, %g .. %g, %g\n", s0, t0, s1, t1);*/
+            SET_ATTRIB(0, attr, s0, t0, 0.0f, 1.0f);  /* lower left */
+            SET_ATTRIB(1, attr, s1, t0, 0.0f, 1.0f);  /* lower right */
+            SET_ATTRIB(2, attr, s1, t1, 0.0f, 1.0f);  /* upper right */
+            SET_ATTRIB(3, attr, s0, t1, 0.0f, 1.0f);  /* upper left */
+
+            semantic_names[attr] = TGSI_SEMANTIC_GENERIC;
+            semantic_indexes[attr] = 0;
+
+            attr++;
+         }
+      }
+
+      pipe_buffer_unmap(pipe, vbuffer_transfer);
+
+#undef SET_ATTRIB
+   }
+
+
+   cso_save_viewport(cso);
+   cso_save_vertex_shader(cso);
+   cso_save_vertex_elements(cso);
+   cso_save_vertex_buffers(cso);
+
+   {
+      void *vs = lookup_shader(pipe, numAttribs,
+                               semantic_names, semantic_indexes);
+      cso_set_vertex_shader_handle(cso, vs);
+   }
+
+   for (i = 0; i < numAttribs; i++) {
+      velements[i].src_offset = i * 4 * sizeof(float);
+      velements[i].instance_divisor = 0;
+      velements[i].vertex_buffer_index = 0;
+      velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+   }
+   cso_set_vertex_elements(cso, numAttribs, velements);
+
+   /* viewport state: viewport matching window dims */
+   {
+      const struct gl_framebuffer *fb = st->ctx->DrawBuffer;
+      const GLboolean invert = (st_fb_orientation(fb) == Y_0_TOP);
+      const GLfloat width = (GLfloat)fb->Width;
+      const GLfloat height = (GLfloat)fb->Height;
+      struct pipe_viewport_state vp;
+      vp.scale[0] =  0.5f * width;
+      vp.scale[1] = height * (invert ? -0.5f : 0.5f);
+      vp.scale[2] = 1.0f;
+      vp.scale[3] = 1.0f;
+      vp.translate[0] = 0.5f * width;
+      vp.translate[1] = 0.5f * height;
+      vp.translate[2] = 0.0f;
+      vp.translate[3] = 0.0f;
+      cso_set_viewport(cso, &vp);
+   }
+
+
+   util_draw_vertex_buffer(pipe, cso, vbuffer,
+                           0,  /* offset */
+                           PIPE_PRIM_TRIANGLE_FAN,
+                           4,  /* verts */
+                           numAttribs); /* attribs/vert */
+
+
+   pipe_resource_reference(&vbuffer, NULL);
+
+   /* restore state */
+   cso_restore_viewport(cso);
+   cso_restore_vertex_shader(cso);
+   cso_restore_vertex_elements(cso);
+   cso_restore_vertex_buffers(cso);
+}
+
+
+void
+st_init_drawtex_functions(struct dd_function_table *functions)
+{
+   functions->DrawTex = st_DrawTex;
+}
+
+
+/**
+ * Free any cached shaders
+ */
+void
+st_destroy_drawtex(struct st_context *st)
+{
+   GLuint i;
+   for (i = 0; i < NumCachedShaders; i++) {
+      cso_delete_vertex_shader(st->cso_context, CachedShaders[i].handle);
+   }
+   NumCachedShaders = 0;
+}
+
+
+#endif /* FEATURE_OES_draw_texture */
diff --git a/mesalib/src/mesa/state_tracker/st_context.c b/mesalib/src/mesa/state_tracker/st_context.c
index 7a19f35bb..8f3a7415d 100644
--- a/mesalib/src/mesa/state_tracker/st_context.c
+++ b/mesalib/src/mesa/state_tracker/st_context.c
@@ -1,297 +1,297 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "main/imports.h"
-#include "main/context.h"
-#include "main/shaderobj.h"
-#include "program/prog_cache.h"
-#include "vbo/vbo.h"
-#include "glapi/glapi.h"
-#include "st_context.h"
-#include "st_debug.h"
-#include "st_cb_accum.h"
-#include "st_cb_bitmap.h"
-#include "st_cb_blit.h"
-#include "st_cb_bufferobjects.h"
-#include "st_cb_clear.h"
-#include "st_cb_condrender.h"
-#include "st_cb_drawpixels.h"
-#include "st_cb_rasterpos.h"
-#include "st_cb_drawtex.h"
-#include "st_cb_eglimage.h"
-#include "st_cb_fbo.h"
-#include "st_cb_feedback.h"
-#include "st_cb_program.h"
-#include "st_cb_queryobj.h"
-#include "st_cb_readpixels.h"
-#include "st_cb_texture.h"
-#include "st_cb_xformfb.h"
-#include "st_cb_flush.h"
-#include "st_cb_strings.h"
-#include "st_cb_viewport.h"
-#include "st_atom.h"
-#include "st_draw.h"
-#include "st_extensions.h"
-#include "st_gen_mipmap.h"
-#include "st_program.h"
-#include "pipe/p_context.h"
-#include "util/u_inlines.h"
-#include "cso_cache/cso_context.h"
-
-
-DEBUG_GET_ONCE_BOOL_OPTION(mesa_mvp_dp4, "MESA_MVP_DP4", FALSE)
-
-
-/**
- * Called via ctx->Driver.UpdateState()
- */
-void st_invalidate_state(struct gl_context * ctx, GLuint new_state)
-{
-   struct st_context *st = st_context(ctx);
-
-   st->dirty.mesa |= new_state;
-   st->dirty.st |= ST_NEW_MESA;
-
-   /* This is the only core Mesa module we depend upon.
-    * No longer use swrast, swsetup, tnl.
-    */
-   _vbo_InvalidateState(ctx, new_state);
-}
-
-
-/**
- * Check for multisample env var override.
- */
-int
-st_get_msaa(void)
-{
-   const char *msaa = _mesa_getenv("__GL_FSAA_MODE");
-   if (msaa)
-      return atoi(msaa);
-   return 0;
-}
-
-
-static struct st_context *
-st_create_context_priv( struct gl_context *ctx, struct pipe_context *pipe )
-{
-   uint i;
-   struct st_context *st = ST_CALLOC_STRUCT( st_context );
-   
-   ctx->st = st;
-
-   st->ctx = ctx;
-   st->pipe = pipe;
-
-   /* XXX: this is one-off, per-screen init: */
-   st_debug_init();
-   
-   /* state tracker needs the VBO module */
-   _vbo_CreateContext(ctx);
-
-   st->dirty.mesa = ~0;
-   st->dirty.st = ~0;
-
-   st->cso_context = cso_create_context(pipe);
-
-   st_init_atoms( st );
-   st_init_bitmap(st);
-   st_init_clear(st);
-   st_init_draw( st );
-   st_init_generate_mipmap(st);
-   st_init_blit(st);
-
-   if(pipe->screen->get_param(pipe->screen, PIPE_CAP_NPOT_TEXTURES))
-      st->internal_target = PIPE_TEXTURE_2D;
-   else
-      st->internal_target = PIPE_TEXTURE_RECT;
-
-   for (i = 0; i < PIPE_MAX_SAMPLERS; i++)
-      st->state.sampler_list[i] = &st->state.samplers[i];
-
-   for (i = 0; i < 3; i++) {
-      memset(&st->velems_util_draw[i], 0, sizeof(struct pipe_vertex_element));
-      st->velems_util_draw[i].src_offset = i * 4 * sizeof(float);
-      st->velems_util_draw[i].instance_divisor = 0;
-      st->velems_util_draw[i].vertex_buffer_index = 0;
-      st->velems_util_draw[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
-   }
-
-   /* we want all vertex data to be placed in buffer objects */
-   vbo_use_buffer_objects(ctx);
-
-   /* Need these flags:
-    */
-   st->ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
-
-   st->ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
-
-   st->pixel_xfer.cache = _mesa_new_program_cache();
-
-   st->force_msaa = st_get_msaa();
-
-   /* GL limits and extensions */
-   st_init_limits(st);
-   st_init_extensions(st);
-
-   return st;
-}
-
-
-struct st_context *st_create_context(gl_api api, struct pipe_context *pipe,
-                                     const struct gl_config *visual,
-                                     struct st_context *share)
-{
-   struct gl_context *ctx;
-   struct gl_context *shareCtx = share ? share->ctx : NULL;
-   struct dd_function_table funcs;
-
-   /* Sanity checks */
-   assert(MESA_SHADER_VERTEX == PIPE_SHADER_VERTEX);
-   assert(MESA_SHADER_FRAGMENT == PIPE_SHADER_FRAGMENT);
-   assert(MESA_SHADER_GEOMETRY == PIPE_SHADER_GEOMETRY);
-
-   memset(&funcs, 0, sizeof(funcs));
-   st_init_driver_functions(&funcs);
-
-   ctx = _mesa_create_context(api, visual, shareCtx, &funcs, NULL);
-
-   /* XXX: need a capability bit in gallium to query if the pipe
-    * driver prefers DP4 or MUL/MAD for vertex transformation.
-    */
-   if (debug_get_option_mesa_mvp_dp4())
-      _mesa_set_mvp_with_dp4( ctx, GL_TRUE );
-
-   return st_create_context_priv(ctx, pipe);
-}
-
-
-static void st_destroy_context_priv( struct st_context *st )
-{
-   uint i;
-
-   st_destroy_atoms( st );
-   st_destroy_draw( st );
-   st_destroy_generate_mipmap(st);
-   st_destroy_blit(st);
-   st_destroy_clear(st);
-   st_destroy_bitmap(st);
-   st_destroy_drawpix(st);
-   st_destroy_drawtex(st);
-
-   /* Unreference any user vertex buffers. */
-   for (i = 0; i < st->num_user_vbs; i++) {
-      pipe_resource_reference(&st->user_vb[i], NULL);
-   }
-
-   for (i = 0; i < Elements(st->state.sampler_views); i++) {
-      pipe_sampler_view_reference(&st->state.sampler_views[i], NULL);
-   }
-
-   if (st->default_texture) {
-      st->ctx->Driver.DeleteTexture(st->ctx, st->default_texture);
-      st->default_texture = NULL;
-   }
-
-   free( st );
-}
-
- 
-void st_destroy_context( struct st_context *st )
-{
-   struct pipe_context *pipe = st->pipe;
-   struct cso_context *cso = st->cso_context;
-   struct gl_context *ctx = st->ctx;
-   GLuint i;
-
-   /* need to unbind and destroy CSO objects before anything else */
-   cso_release_all(st->cso_context);
-
-   st_reference_fragprog(st, &st->fp, NULL);
-   st_reference_vertprog(st, &st->vp, NULL);
-
-   /* release framebuffer surfaces */
-   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
-      pipe_surface_reference(&st->state.framebuffer.cbufs[i], NULL);
-   }
-   pipe_surface_reference(&st->state.framebuffer.zsbuf, NULL);
-
-   pipe->set_index_buffer(pipe, NULL);
-
-   for (i = 0; i < PIPE_SHADER_TYPES; i++) {
-      pipe->set_constant_buffer(pipe, i, 0, NULL);
-   }
-
-   _mesa_delete_program_cache(st->ctx, st->pixel_xfer.cache);
-
-   _vbo_DestroyContext(st->ctx);
-
-   st_destroy_program_variants(st);
-
-   _mesa_free_context_data(ctx);
-
-   st_destroy_context_priv(st);
-
-   cso_destroy_context(cso);
-
-   pipe->destroy( pipe );
-
-   free(ctx);
-}
-
-
-void st_init_driver_functions(struct dd_function_table *functions)
-{
-   _mesa_init_shader_object_functions(functions);
-
-   st_init_accum_functions(functions);
-   st_init_blit_functions(functions);
-   st_init_bufferobject_functions(functions);
-   st_init_clear_functions(functions);
-   st_init_bitmap_functions(functions);
-   st_init_drawpixels_functions(functions);
-   st_init_rasterpos_functions(functions);
-
-   st_init_drawtex_functions(functions);
-
-   st_init_eglimage_functions(functions);
-
-   st_init_fbo_functions(functions);
-   st_init_feedback_functions(functions);
-   st_init_program_functions(functions);
-   st_init_query_functions(functions);
-   st_init_cond_render_functions(functions);
-   st_init_readpixels_functions(functions);
-   st_init_texture_functions(functions);
-   st_init_flush_functions(functions);
-   st_init_string_functions(functions);
-   st_init_viewport_functions(functions);
-
-   st_init_xformfb_functions(functions);
-
-   functions->UpdateState = st_invalidate_state;
-}
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/shaderobj.h"
+#include "program/prog_cache.h"
+#include "vbo/vbo.h"
+#include "glapi/glapi.h"
+#include "st_context.h"
+#include "st_debug.h"
+#include "st_cb_accum.h"
+#include "st_cb_bitmap.h"
+#include "st_cb_blit.h"
+#include "st_cb_bufferobjects.h"
+#include "st_cb_clear.h"
+#include "st_cb_condrender.h"
+#include "st_cb_drawpixels.h"
+#include "st_cb_rasterpos.h"
+#include "st_cb_drawtex.h"
+#include "st_cb_eglimage.h"
+#include "st_cb_fbo.h"
+#include "st_cb_feedback.h"
+#include "st_cb_program.h"
+#include "st_cb_queryobj.h"
+#include "st_cb_readpixels.h"
+#include "st_cb_texture.h"
+#include "st_cb_xformfb.h"
+#include "st_cb_flush.h"
+#include "st_cb_strings.h"
+#include "st_cb_viewport.h"
+#include "st_atom.h"
+#include "st_draw.h"
+#include "st_extensions.h"
+#include "st_gen_mipmap.h"
+#include "st_program.h"
+#include "pipe/p_context.h"
+#include "util/u_inlines.h"
+#include "cso_cache/cso_context.h"
+
+
+DEBUG_GET_ONCE_BOOL_OPTION(mesa_mvp_dp4, "MESA_MVP_DP4", FALSE)
+
+
+/**
+ * Called via ctx->Driver.UpdateState()
+ */
+void st_invalidate_state(struct gl_context * ctx, GLuint new_state)
+{
+   struct st_context *st = st_context(ctx);
+
+   st->dirty.mesa |= new_state;
+   st->dirty.st |= ST_NEW_MESA;
+
+   /* This is the only core Mesa module we depend upon.
+    * No longer use swrast, swsetup, tnl.
+    */
+   _vbo_InvalidateState(ctx, new_state);
+}
+
+
+/**
+ * Check for multisample env var override.
+ */
+int
+st_get_msaa(void)
+{
+   const char *msaa = _mesa_getenv("__GL_FSAA_MODE");
+   if (msaa)
+      return atoi(msaa);
+   return 0;
+}
+
+
+static struct st_context *
+st_create_context_priv( struct gl_context *ctx, struct pipe_context *pipe )
+{
+   uint i;
+   struct st_context *st = ST_CALLOC_STRUCT( st_context );
+   
+   ctx->st = st;
+
+   st->ctx = ctx;
+   st->pipe = pipe;
+
+   /* XXX: this is one-off, per-screen init: */
+   st_debug_init();
+   
+   /* state tracker needs the VBO module */
+   _vbo_CreateContext(ctx);
+
+   st->dirty.mesa = ~0;
+   st->dirty.st = ~0;
+
+   st->cso_context = cso_create_context(pipe);
+
+   st_init_atoms( st );
+   st_init_bitmap(st);
+   st_init_clear(st);
+   st_init_draw( st );
+   st_init_generate_mipmap(st);
+   st_init_blit(st);
+
+   if(pipe->screen->get_param(pipe->screen, PIPE_CAP_NPOT_TEXTURES))
+      st->internal_target = PIPE_TEXTURE_2D;
+   else
+      st->internal_target = PIPE_TEXTURE_RECT;
+
+   for (i = 0; i < PIPE_MAX_SAMPLERS; i++)
+      st->state.sampler_list[i] = &st->state.samplers[i];
+
+   for (i = 0; i < 3; i++) {
+      memset(&st->velems_util_draw[i], 0, sizeof(struct pipe_vertex_element));
+      st->velems_util_draw[i].src_offset = i * 4 * sizeof(float);
+      st->velems_util_draw[i].instance_divisor = 0;
+      st->velems_util_draw[i].vertex_buffer_index = 0;
+      st->velems_util_draw[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+   }
+
+   /* we want all vertex data to be placed in buffer objects */
+   vbo_use_buffer_objects(ctx);
+
+   /* Need these flags:
+    */
+   st->ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
+
+   st->ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
+
+   st->pixel_xfer.cache = _mesa_new_program_cache();
+
+   st->force_msaa = st_get_msaa();
+
+   /* GL limits and extensions */
+   st_init_limits(st);
+   st_init_extensions(st);
+
+   return st;
+}
+
+
+struct st_context *st_create_context(gl_api api, struct pipe_context *pipe,
+                                     const struct gl_config *visual,
+                                     struct st_context *share)
+{
+   struct gl_context *ctx;
+   struct gl_context *shareCtx = share ? share->ctx : NULL;
+   struct dd_function_table funcs;
+
+   /* Sanity checks */
+   assert(MESA_SHADER_VERTEX == PIPE_SHADER_VERTEX);
+   assert(MESA_SHADER_FRAGMENT == PIPE_SHADER_FRAGMENT);
+   assert(MESA_SHADER_GEOMETRY == PIPE_SHADER_GEOMETRY);
+
+   memset(&funcs, 0, sizeof(funcs));
+   st_init_driver_functions(&funcs);
+
+   ctx = _mesa_create_context(api, visual, shareCtx, &funcs, NULL);
+
+   /* XXX: need a capability bit in gallium to query if the pipe
+    * driver prefers DP4 or MUL/MAD for vertex transformation.
+    */
+   if (debug_get_option_mesa_mvp_dp4())
+      _mesa_set_mvp_with_dp4( ctx, GL_TRUE );
+
+   return st_create_context_priv(ctx, pipe);
+}
+
+
+static void st_destroy_context_priv( struct st_context *st )
+{
+   uint i;
+
+   st_destroy_atoms( st );
+   st_destroy_draw( st );
+   st_destroy_generate_mipmap(st);
+   st_destroy_blit(st);
+   st_destroy_clear(st);
+   st_destroy_bitmap(st);
+   st_destroy_drawpix(st);
+   st_destroy_drawtex(st);
+
+   /* Unreference any user vertex buffers. */
+   for (i = 0; i < st->num_user_vbs; i++) {
+      pipe_resource_reference(&st->user_vb[i], NULL);
+   }
+
+   for (i = 0; i < Elements(st->state.sampler_views); i++) {
+      pipe_sampler_view_reference(&st->state.sampler_views[i], NULL);
+   }
+
+   if (st->default_texture) {
+      st->ctx->Driver.DeleteTexture(st->ctx, st->default_texture);
+      st->default_texture = NULL;
+   }
+
+   free( st );
+}
+
+ 
+void st_destroy_context( struct st_context *st )
+{
+   struct pipe_context *pipe = st->pipe;
+   struct cso_context *cso = st->cso_context;
+   struct gl_context *ctx = st->ctx;
+   GLuint i;
+
+   /* need to unbind and destroy CSO objects before anything else */
+   cso_release_all(st->cso_context);
+
+   st_reference_fragprog(st, &st->fp, NULL);
+   st_reference_vertprog(st, &st->vp, NULL);
+
+   /* release framebuffer surfaces */
+   for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+      pipe_surface_reference(&st->state.framebuffer.cbufs[i], NULL);
+   }
+   pipe_surface_reference(&st->state.framebuffer.zsbuf, NULL);
+
+   pipe->set_index_buffer(pipe, NULL);
+
+   for (i = 0; i < PIPE_SHADER_TYPES; i++) {
+      pipe->set_constant_buffer(pipe, i, 0, NULL);
+   }
+
+   _mesa_delete_program_cache(st->ctx, st->pixel_xfer.cache);
+
+   _vbo_DestroyContext(st->ctx);
+
+   st_destroy_program_variants(st);
+
+   _mesa_free_context_data(ctx);
+
+   st_destroy_context_priv(st);
+
+   cso_destroy_context(cso);
+
+   pipe->destroy( pipe );
+
+   free(ctx);
+}
+
+
+void st_init_driver_functions(struct dd_function_table *functions)
+{
+   _mesa_init_shader_object_functions(functions);
+
+   st_init_accum_functions(functions);
+   st_init_blit_functions(functions);
+   st_init_bufferobject_functions(functions);
+   st_init_clear_functions(functions);
+   st_init_bitmap_functions(functions);
+   st_init_drawpixels_functions(functions);
+   st_init_rasterpos_functions(functions);
+
+   st_init_drawtex_functions(functions);
+
+   st_init_eglimage_functions(functions);
+
+   st_init_fbo_functions(functions);
+   st_init_feedback_functions(functions);
+   st_init_program_functions(functions);
+   st_init_query_functions(functions);
+   st_init_cond_render_functions(functions);
+   st_init_readpixels_functions(functions);
+   st_init_texture_functions(functions);
+   st_init_flush_functions(functions);
+   st_init_string_functions(functions);
+   st_init_viewport_functions(functions);
+
+   st_init_xformfb_functions(functions);
+
+   functions->UpdateState = st_invalidate_state;
+}
diff --git a/mesalib/src/mesa/state_tracker/st_context.h b/mesalib/src/mesa/state_tracker/st_context.h
index 77765f023..ef54fd7cd 100644
--- a/mesalib/src/mesa/state_tracker/st_context.h
+++ b/mesalib/src/mesa/state_tracker/st_context.h
@@ -1,270 +1,270 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef ST_CONTEXT_H
-#define ST_CONTEXT_H
-
-#include "main/mtypes.h"
-#include "pipe/p_state.h"
-#include "state_tracker/st_api.h"
-
-struct bitmap_cache;
-struct blit_state;
-struct dd_function_table;
-struct draw_context;
-struct draw_stage;
-struct gen_mipmap_state;
-struct st_context;
-struct st_fragment_program;
-
-
-#define ST_NEW_MESA                    0x1 /* Mesa state has changed */
-#define ST_NEW_FRAGMENT_PROGRAM        0x2
-#define ST_NEW_VERTEX_PROGRAM          0x4
-#define ST_NEW_FRAMEBUFFER             0x8
-#define ST_NEW_EDGEFLAGS_DATA          0x10
-#define ST_NEW_GEOMETRY_PROGRAM        0x20
-
-
-struct st_state_flags {
-   GLuint mesa;
-   GLuint st;
-};
-
-struct st_tracked_state {
-   const char *name;
-   struct st_state_flags dirty;
-   void (*update)( struct st_context *st );
-};
-
-
-
-struct st_context
-{
-   struct st_context_iface iface;
-
-   struct gl_context *ctx;
-
-   struct pipe_context *pipe;
-
-   struct draw_context *draw;  /**< For selection/feedback/rastpos only */
-   struct draw_stage *feedback_stage;  /**< For GL_FEEDBACK rendermode */
-   struct draw_stage *selection_stage;  /**< For GL_SELECT rendermode */
-   struct draw_stage *rastpos_stage;  /**< For glRasterPos */
-
-
-   /* On old libGL's for linux we need to invalidate the drawables
-    * on glViewpport calls, this is set via a option.
-    */
-   boolean invalidate_on_gl_viewport;
-
-   /* Some state is contained in constant objects.
-    * Other state is just parameter values.
-    */
-   struct {
-      struct pipe_blend_state               blend;
-      struct pipe_depth_stencil_alpha_state depth_stencil;
-      struct pipe_rasterizer_state          rasterizer;
-      struct pipe_sampler_state             samplers[PIPE_MAX_SAMPLERS];
-      struct pipe_sampler_state             *sampler_list[PIPE_MAX_SAMPLERS];
-      struct pipe_clip_state clip;
-      struct {
-         void *ptr;
-         unsigned size;
-      } constants[PIPE_SHADER_TYPES];
-      struct pipe_framebuffer_state framebuffer;
-      struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS];
-      struct pipe_scissor_state scissor;
-      struct pipe_viewport_state viewport;
-      unsigned sample_mask;
-
-      GLuint num_samplers;
-      GLuint num_textures;
-
-      GLuint poly_stipple[32];  /**< In OpenGL's bottom-to-top order */
-   } state;
-
-   char vendor[100];
-   char renderer[100];
-
-   struct st_state_flags dirty;
-
-   GLboolean missing_textures;
-   GLboolean vertdata_edgeflags;
-
-   /** Mapping from VERT_RESULT_x to post-transformed vertex slot */
-   const GLuint *vertex_result_to_slot;
-
-   struct st_vertex_program *vp;    /**< Currently bound vertex program */
-   struct st_fragment_program *fp;  /**< Currently bound fragment program */
-   struct st_geometry_program *gp;  /**< Currently bound geometry program */
-
-   struct st_vp_variant *vp_variant;
-   struct st_fp_variant *fp_variant;
-   struct st_gp_variant *gp_variant;
-
-   struct gl_texture_object *default_texture;
-
-   struct {
-      struct gl_program_cache *cache;
-      struct st_fragment_program *program;  /**< cur pixel transfer prog */
-      GLuint xfer_prog_sn;  /**< pixel xfer program serial no. */
-      GLuint user_prog_sn;  /**< user fragment program serial no. */
-      struct st_fragment_program *combined_prog;
-      GLuint combined_prog_sn;
-      struct pipe_resource *pixelmap_texture;
-      struct pipe_sampler_view *pixelmap_sampler_view;
-      boolean pixelmap_enabled;  /**< use the pixelmap texture? */
-   } pixel_xfer;
-
-   /** for glBitmap */
-   struct {
-      struct pipe_rasterizer_state rasterizer;
-      struct pipe_sampler_state samplers[2];
-      enum pipe_format tex_format;
-      void *vs;
-      float vertices[4][3][4];  /**< vertex pos + color + texcoord */
-      struct pipe_resource *vbuf;
-      unsigned vbuf_slot;       /* next free slot in vbuf */
-      struct bitmap_cache *cache;
-   } bitmap;
-
-   /** for glDraw/CopyPixels */
-   struct {
-      struct gl_fragment_program *shaders[4];
-      void *vert_shaders[2];   /**< ureg shaders */
-   } drawpix;
-
-   /** for glClear */
-   struct {
-      struct pipe_rasterizer_state raster;
-      struct pipe_viewport_state viewport;
-      struct pipe_clip_state clip;
-      void *vs;
-      void *fs;
-      float vertices[4][2][4];  /**< vertex pos + color */
-      struct pipe_resource *vbuf;
-      unsigned vbuf_slot;
-      boolean enable_ds_separate;
-   } clear;
-
-   /** used for anything using util_draw_vertex_buffer */
-   struct pipe_vertex_element velems_util_draw[3];
-
-   void *passthrough_fs;  /**< simple pass-through frag shader */
-
-   enum pipe_texture_target internal_target;
-   struct gen_mipmap_state *gen_mipmap;
-   struct blit_state *blit;
-
-   struct cso_context *cso_context;
-
-   int force_msaa;
-   void *winsys_drawable_handle;
-
-   /* User vertex buffers. */
-   struct pipe_resource *user_vb[PIPE_MAX_ATTRIBS];
-   unsigned user_vb_stride[PIPE_MAX_ATTRIBS];
-   unsigned num_user_vbs;
-};
-
-
-/* Need this so that we can implement Mesa callbacks in this module.
- */
-static INLINE struct st_context *st_context(struct gl_context *ctx)
-{
-   return ctx->st;
-}
-
-
-/**
- * Wrapper for struct gl_framebuffer.
- * This is an opaque type to the outside world.
- */
-struct st_framebuffer
-{
-   struct gl_framebuffer Base;
-   void *Private;
-
-   struct st_framebuffer_iface *iface;
-   enum st_attachment_type statts[ST_ATTACHMENT_COUNT];
-   unsigned num_statts;
-   int32_t revalidate;
-};
-
-
-extern void st_init_driver_functions(struct dd_function_table *functions);
-
-void st_invalidate_state(struct gl_context * ctx, GLuint new_state);
-
-
-
-#define Y_0_TOP 1
-#define Y_0_BOTTOM 2
-
-static INLINE GLuint
-st_fb_orientation(const struct gl_framebuffer *fb)
-{
-   if (fb && fb->Name == 0) {
-      /* Drawing into a window (on-screen buffer).
-       *
-       * Negate Y scale to flip image vertically.
-       * The NDC Y coords prior to viewport transformation are in the range
-       * [y=-1=bottom, y=1=top]
-       * Hardware window coords are in the range [y=0=top, y=H-1=bottom] where
-       * H is the window height.
-       * Use the viewport transformation to invert Y.
-       */
-      return Y_0_TOP;
-   }
-   else {
-      /* Drawing into user-created FBO (very likely a texture).
-       *
-       * For textures, T=0=Bottom, so by extension Y=0=Bottom for rendering.
-       */
-      return Y_0_BOTTOM;
-   }
-}
-
-
-/** clear-alloc a struct-sized object, with casting */
-#define ST_CALLOC_STRUCT(T)   (struct T *) calloc(1, sizeof(struct T))
-
-
-extern int
-st_get_msaa(void);
-
-extern struct st_context *
-st_create_context(gl_api api, struct pipe_context *pipe,
-                  const struct gl_config *visual,
-                  struct st_context *share);
-
-extern void
-st_destroy_context(struct st_context *st);
-
-
-#endif
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef ST_CONTEXT_H
+#define ST_CONTEXT_H
+
+#include "main/mtypes.h"
+#include "pipe/p_state.h"
+#include "state_tracker/st_api.h"
+
+struct bitmap_cache;
+struct blit_state;
+struct dd_function_table;
+struct draw_context;
+struct draw_stage;
+struct gen_mipmap_state;
+struct st_context;
+struct st_fragment_program;
+
+
+#define ST_NEW_MESA                    0x1 /* Mesa state has changed */
+#define ST_NEW_FRAGMENT_PROGRAM        0x2
+#define ST_NEW_VERTEX_PROGRAM          0x4
+#define ST_NEW_FRAMEBUFFER             0x8
+#define ST_NEW_EDGEFLAGS_DATA          0x10
+#define ST_NEW_GEOMETRY_PROGRAM        0x20
+
+
+struct st_state_flags {
+   GLuint mesa;
+   GLuint st;
+};
+
+struct st_tracked_state {
+   const char *name;
+   struct st_state_flags dirty;
+   void (*update)( struct st_context *st );
+};
+
+
+
+struct st_context
+{
+   struct st_context_iface iface;
+
+   struct gl_context *ctx;
+
+   struct pipe_context *pipe;
+
+   struct draw_context *draw;  /**< For selection/feedback/rastpos only */
+   struct draw_stage *feedback_stage;  /**< For GL_FEEDBACK rendermode */
+   struct draw_stage *selection_stage;  /**< For GL_SELECT rendermode */
+   struct draw_stage *rastpos_stage;  /**< For glRasterPos */
+
+
+   /* On old libGL's for linux we need to invalidate the drawables
+    * on glViewpport calls, this is set via a option.
+    */
+   boolean invalidate_on_gl_viewport;
+
+   /* Some state is contained in constant objects.
+    * Other state is just parameter values.
+    */
+   struct {
+      struct pipe_blend_state               blend;
+      struct pipe_depth_stencil_alpha_state depth_stencil;
+      struct pipe_rasterizer_state          rasterizer;
+      struct pipe_sampler_state             samplers[PIPE_MAX_SAMPLERS];
+      struct pipe_sampler_state             *sampler_list[PIPE_MAX_SAMPLERS];
+      struct pipe_clip_state clip;
+      struct {
+         void *ptr;
+         unsigned size;
+      } constants[PIPE_SHADER_TYPES];
+      struct pipe_framebuffer_state framebuffer;
+      struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS];
+      struct pipe_scissor_state scissor;
+      struct pipe_viewport_state viewport;
+      unsigned sample_mask;
+
+      GLuint num_samplers;
+      GLuint num_textures;
+
+      GLuint poly_stipple[32];  /**< In OpenGL's bottom-to-top order */
+   } state;
+
+   char vendor[100];
+   char renderer[100];
+
+   struct st_state_flags dirty;
+
+   GLboolean missing_textures;
+   GLboolean vertdata_edgeflags;
+
+   /** Mapping from VERT_RESULT_x to post-transformed vertex slot */
+   const GLuint *vertex_result_to_slot;
+
+   struct st_vertex_program *vp;    /**< Currently bound vertex program */
+   struct st_fragment_program *fp;  /**< Currently bound fragment program */
+   struct st_geometry_program *gp;  /**< Currently bound geometry program */
+
+   struct st_vp_variant *vp_variant;
+   struct st_fp_variant *fp_variant;
+   struct st_gp_variant *gp_variant;
+
+   struct gl_texture_object *default_texture;
+
+   struct {
+      struct gl_program_cache *cache;
+      struct st_fragment_program *program;  /**< cur pixel transfer prog */
+      GLuint xfer_prog_sn;  /**< pixel xfer program serial no. */
+      GLuint user_prog_sn;  /**< user fragment program serial no. */
+      struct st_fragment_program *combined_prog;
+      GLuint combined_prog_sn;
+      struct pipe_resource *pixelmap_texture;
+      struct pipe_sampler_view *pixelmap_sampler_view;
+      boolean pixelmap_enabled;  /**< use the pixelmap texture? */
+   } pixel_xfer;
+
+   /** for glBitmap */
+   struct {
+      struct pipe_rasterizer_state rasterizer;
+      struct pipe_sampler_state samplers[2];
+      enum pipe_format tex_format;
+      void *vs;
+      float vertices[4][3][4];  /**< vertex pos + color + texcoord */
+      struct pipe_resource *vbuf;
+      unsigned vbuf_slot;       /* next free slot in vbuf */
+      struct bitmap_cache *cache;
+   } bitmap;
+
+   /** for glDraw/CopyPixels */
+   struct {
+      struct gl_fragment_program *shaders[4];
+      void *vert_shaders[2];   /**< ureg shaders */
+   } drawpix;
+
+   /** for glClear */
+   struct {
+      struct pipe_rasterizer_state raster;
+      struct pipe_viewport_state viewport;
+      struct pipe_clip_state clip;
+      void *vs;
+      void *fs;
+      float vertices[4][2][4];  /**< vertex pos + color */
+      struct pipe_resource *vbuf;
+      unsigned vbuf_slot;
+      boolean enable_ds_separate;
+   } clear;
+
+   /** used for anything using util_draw_vertex_buffer */
+   struct pipe_vertex_element velems_util_draw[3];
+
+   void *passthrough_fs;  /**< simple pass-through frag shader */
+
+   enum pipe_texture_target internal_target;
+   struct gen_mipmap_state *gen_mipmap;
+   struct blit_state *blit;
+
+   struct cso_context *cso_context;
+
+   int force_msaa;
+   void *winsys_drawable_handle;
+
+   /* User vertex buffers. */
+   struct pipe_resource *user_vb[PIPE_MAX_ATTRIBS];
+   unsigned user_vb_stride[PIPE_MAX_ATTRIBS];
+   unsigned num_user_vbs;
+};
+
+
+/* Need this so that we can implement Mesa callbacks in this module.
+ */
+static INLINE struct st_context *st_context(struct gl_context *ctx)
+{
+   return ctx->st;
+}
+
+
+/**
+ * Wrapper for struct gl_framebuffer.
+ * This is an opaque type to the outside world.
+ */
+struct st_framebuffer
+{
+   struct gl_framebuffer Base;
+   void *Private;
+
+   struct st_framebuffer_iface *iface;
+   enum st_attachment_type statts[ST_ATTACHMENT_COUNT];
+   unsigned num_statts;
+   int32_t revalidate;
+};
+
+
+extern void st_init_driver_functions(struct dd_function_table *functions);
+
+void st_invalidate_state(struct gl_context * ctx, GLuint new_state);
+
+
+
+#define Y_0_TOP 1
+#define Y_0_BOTTOM 2
+
+static INLINE GLuint
+st_fb_orientation(const struct gl_framebuffer *fb)
+{
+   if (fb && fb->Name == 0) {
+      /* Drawing into a window (on-screen buffer).
+       *
+       * Negate Y scale to flip image vertically.
+       * The NDC Y coords prior to viewport transformation are in the range
+       * [y=-1=bottom, y=1=top]
+       * Hardware window coords are in the range [y=0=top, y=H-1=bottom] where
+       * H is the window height.
+       * Use the viewport transformation to invert Y.
+       */
+      return Y_0_TOP;
+   }
+   else {
+      /* Drawing into user-created FBO (very likely a texture).
+       *
+       * For textures, T=0=Bottom, so by extension Y=0=Bottom for rendering.
+       */
+      return Y_0_BOTTOM;
+   }
+}
+
+
+/** clear-alloc a struct-sized object, with casting */
+#define ST_CALLOC_STRUCT(T)   (struct T *) calloc(1, sizeof(struct T))
+
+
+extern int
+st_get_msaa(void);
+
+extern struct st_context *
+st_create_context(gl_api api, struct pipe_context *pipe,
+                  const struct gl_config *visual,
+                  struct st_context *share);
+
+extern void
+st_destroy_context(struct st_context *st);
+
+
+#endif
diff --git a/mesalib/src/mesa/state_tracker/st_draw_feedback.c b/mesalib/src/mesa/state_tracker/st_draw_feedback.c
index 1e1220bfe..f0734eae0 100644
--- a/mesalib/src/mesa/state_tracker/st_draw_feedback.c
+++ b/mesalib/src/mesa/state_tracker/st_draw_feedback.c
@@ -1,277 +1,277 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-
-#include "vbo/vbo.h"
-
-#include "st_context.h"
-#include "st_atom.h"
-#include "st_cb_bufferobjects.h"
-#include "st_draw.h"
-#include "st_program.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-
-#include "draw/draw_private.h"
-#include "draw/draw_context.h"
-
-
-#if FEATURE_feedback || FEATURE_rastpos
-
-/**
- * Set the (private) draw module's post-transformed vertex format when in
- * GL_SELECT or GL_FEEDBACK mode or for glRasterPos.
- */
-static void
-set_feedback_vertex_format(struct gl_context *ctx)
-{
-#if 0
-   struct st_context *st = st_context(ctx);
-   struct vertex_info vinfo;
-   GLuint i;
-
-   memset(&vinfo, 0, sizeof(vinfo));
-
-   if (ctx->RenderMode == GL_SELECT) {
-      assert(ctx->RenderMode == GL_SELECT);
-      vinfo.num_attribs = 1;
-      vinfo.format[0] = FORMAT_4F;
-      vinfo.interp_mode[0] = INTERP_LINEAR;
-   }
-   else {
-      /* GL_FEEDBACK, or glRasterPos */
-      /* emit all attribs (pos, color, texcoord) as GLfloat[4] */
-      vinfo.num_attribs = st->state.vs->cso->state.num_outputs;
-      for (i = 0; i < vinfo.num_attribs; i++) {
-         vinfo.format[i] = FORMAT_4F;
-         vinfo.interp_mode[i] = INTERP_LINEAR;
-      }
-   }
-
-   draw_set_vertex_info(st->draw, &vinfo);
-#endif
-}
-
-
-/**
- * Called by VBO to draw arrays when in selection or feedback mode and
- * to implement glRasterPos.
- * This is very much like the normal draw_vbo() function above.
- * Look at code refactoring some day.
- * Might move this into the failover module some day.
- */
-void
-st_feedback_draw_vbo(struct gl_context *ctx,
-                     const struct gl_client_array **arrays,
-                     const struct _mesa_prim *prims,
-                     GLuint nr_prims,
-                     const struct _mesa_index_buffer *ib,
-		     GLboolean index_bounds_valid,
-                     GLuint min_index,
-                     GLuint max_index)
-{
-   struct st_context *st = st_context(ctx);
-   struct pipe_context *pipe = st->pipe;
-   struct draw_context *draw = st->draw;
-   const struct st_vertex_program *vp;
-   const struct pipe_shader_state *vs;
-   struct pipe_vertex_buffer vbuffers[PIPE_MAX_SHADER_INPUTS];
-   struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS];
-   struct pipe_index_buffer ibuffer;
-   struct pipe_transfer *vb_transfer[PIPE_MAX_ATTRIBS];
-   struct pipe_transfer *ib_transfer = NULL;
-   GLuint attr, i;
-   const void *mapped_indices = NULL;
-
-   assert(draw);
-
-   st_validate_state(st);
-
-   if (!index_bounds_valid)
-      vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index);
-
-   /* must get these after state validation! */
-   vp = st->vp;
-   vs = &st->vp_variant->tgsi;
-
-   if (!st->vp_variant->draw_shader) {
-      st->vp_variant->draw_shader = draw_create_vertex_shader(draw, vs);
-   }
-
-   /*
-    * Set up the draw module's state.
-    *
-    * We'd like to do this less frequently, but the normal state-update
-    * code sends state updates to the pipe, not to our private draw module.
-    */
-   assert(draw);
-   draw_set_viewport_state(draw, &st->state.viewport);
-   draw_set_clip_state(draw, &st->state.clip);
-   draw_set_rasterizer_state(draw, &st->state.rasterizer, NULL);
-   draw_bind_vertex_shader(draw, st->vp_variant->draw_shader);
-   set_feedback_vertex_format(ctx);
-
-   /* loop over TGSI shader inputs to determine vertex buffer
-    * and attribute info
-    */
-   for (attr = 0; attr < vp->num_inputs; attr++) {
-      const GLuint mesaAttr = vp->index_to_input[attr];
-      struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj;
-      void *map;
-
-      if (bufobj && bufobj->Name) {
-         /* Attribute data is in a VBO.
-          * Recall that for VBOs, the gl_client_array->Ptr field is
-          * really an offset from the start of the VBO, not a pointer.
-          */
-         struct st_buffer_object *stobj = st_buffer_object(bufobj);
-         assert(stobj->buffer);
-
-         vbuffers[attr].buffer = NULL;
-         pipe_resource_reference(&vbuffers[attr].buffer, stobj->buffer);
-         vbuffers[attr].buffer_offset = pointer_to_offset(arrays[0]->Ptr);
-         velements[attr].src_offset = arrays[mesaAttr]->Ptr - arrays[0]->Ptr;
-      }
-      else {
-         /* attribute data is in user-space memory, not a VBO */
-         uint bytes = (arrays[mesaAttr]->Size
-                       * _mesa_sizeof_type(arrays[mesaAttr]->Type)
-                       * (max_index + 1));
-
-         /* wrap user data */
-         vbuffers[attr].buffer
-            = pipe_user_buffer_create(pipe->screen, (void *) arrays[mesaAttr]->Ptr,
-                                      bytes,
-				      PIPE_BIND_VERTEX_BUFFER);
-         vbuffers[attr].buffer_offset = 0;
-         velements[attr].src_offset = 0;
-      }
-
-      /* common-case setup */
-      vbuffers[attr].stride = arrays[mesaAttr]->StrideB; /* in bytes */
-      velements[attr].instance_divisor = 0;
-      velements[attr].vertex_buffer_index = attr;
-      velements[attr].src_format = 
-         st_pipe_vertex_format(arrays[mesaAttr]->Type,
-                               arrays[mesaAttr]->Size,
-                               arrays[mesaAttr]->Format,
-                               arrays[mesaAttr]->Normalized);
-      assert(velements[attr].src_format);
-
-      /* tell draw about this attribute */
-#if 0
-      draw_set_vertex_buffer(draw, attr, &vbuffer[attr]);
-#endif
-
-      /* map the attrib buffer */
-      map = pipe_buffer_map(pipe, vbuffers[attr].buffer,
-                            PIPE_TRANSFER_READ,
-			    &vb_transfer[attr]);
-      draw_set_mapped_vertex_buffer(draw, attr, map);
-   }
-
-   draw_set_vertex_buffers(draw, vp->num_inputs, vbuffers);
-   draw_set_vertex_elements(draw, vp->num_inputs, velements);
-
-   memset(&ibuffer, 0, sizeof(ibuffer));
-   if (ib) {
-      struct gl_buffer_object *bufobj = ib->obj;
-
-      switch (ib->type) {
-      case GL_UNSIGNED_INT:
-         ibuffer.index_size = 4;
-         break;
-      case GL_UNSIGNED_SHORT:
-         ibuffer.index_size = 2;
-         break;
-      case GL_UNSIGNED_BYTE:
-         ibuffer.index_size = 1;
-         break;
-      default:
-         assert(0);
-	 goto out_unref_vertex;
-      }
-
-      if (bufobj && bufobj->Name) {
-         struct st_buffer_object *stobj = st_buffer_object(bufobj);
-
-         pipe_resource_reference(&ibuffer.buffer, stobj->buffer);
-         ibuffer.offset = pointer_to_offset(ib->ptr);
-
-         mapped_indices = pipe_buffer_map(pipe, stobj->buffer,
-                                          PIPE_TRANSFER_READ, &ib_transfer);
-      }
-      else {
-         /* skip setting ibuffer.buffer as the draw module does not use it */
-         mapped_indices = ib->ptr;
-      }
-
-      draw_set_index_buffer(draw, &ibuffer);
-      draw_set_mapped_index_buffer(draw, mapped_indices);
-   }
-
-   /* set the constant buffer */
-   draw_set_mapped_constant_buffer(st->draw, PIPE_SHADER_VERTEX, 0,
-                                   st->state.constants[PIPE_SHADER_VERTEX].ptr,
-                                   st->state.constants[PIPE_SHADER_VERTEX].size);
-
-
-   /* draw here */
-   for (i = 0; i < nr_prims; i++) {
-      draw_arrays(draw, prims[i].mode, prims[i].start, prims[i].count);
-   }
-
-
-   /*
-    * unmap vertex/index buffers
-    */
-   if (ib) {
-      draw_set_mapped_index_buffer(draw, NULL);
-      draw_set_index_buffer(draw, NULL);
-
-      if (ib_transfer)
-         pipe_buffer_unmap(pipe, ib_transfer);
-      pipe_resource_reference(&ibuffer.buffer, NULL);
-   }
-
- out_unref_vertex:
-   for (attr = 0; attr < vp->num_inputs; attr++) {
-      pipe_buffer_unmap(pipe, vb_transfer[attr]);
-      draw_set_mapped_vertex_buffer(draw, attr, NULL);
-      pipe_resource_reference(&vbuffers[attr].buffer, NULL);
-   }
-   draw_set_vertex_buffers(draw, 0, NULL);
-}
-
-#endif /* FEATURE_feedback || FEATURE_rastpos */
-
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/imports.h"
+#include "main/image.h"
+#include "main/macros.h"
+#include "main/mfeatures.h"
+
+#include "vbo/vbo.h"
+
+#include "st_context.h"
+#include "st_atom.h"
+#include "st_cb_bufferobjects.h"
+#include "st_draw.h"
+#include "st_program.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "util/u_inlines.h"
+
+#include "draw/draw_private.h"
+#include "draw/draw_context.h"
+
+
+#if FEATURE_feedback || FEATURE_rastpos
+
+/**
+ * Set the (private) draw module's post-transformed vertex format when in
+ * GL_SELECT or GL_FEEDBACK mode or for glRasterPos.
+ */
+static void
+set_feedback_vertex_format(struct gl_context *ctx)
+{
+#if 0
+   struct st_context *st = st_context(ctx);
+   struct vertex_info vinfo;
+   GLuint i;
+
+   memset(&vinfo, 0, sizeof(vinfo));
+
+   if (ctx->RenderMode == GL_SELECT) {
+      assert(ctx->RenderMode == GL_SELECT);
+      vinfo.num_attribs = 1;
+      vinfo.format[0] = FORMAT_4F;
+      vinfo.interp_mode[0] = INTERP_LINEAR;
+   }
+   else {
+      /* GL_FEEDBACK, or glRasterPos */
+      /* emit all attribs (pos, color, texcoord) as GLfloat[4] */
+      vinfo.num_attribs = st->state.vs->cso->state.num_outputs;
+      for (i = 0; i < vinfo.num_attribs; i++) {
+         vinfo.format[i] = FORMAT_4F;
+         vinfo.interp_mode[i] = INTERP_LINEAR;
+      }
+   }
+
+   draw_set_vertex_info(st->draw, &vinfo);
+#endif
+}
+
+
+/**
+ * Called by VBO to draw arrays when in selection or feedback mode and
+ * to implement glRasterPos.
+ * This is very much like the normal draw_vbo() function above.
+ * Look at code refactoring some day.
+ * Might move this into the failover module some day.
+ */
+void
+st_feedback_draw_vbo(struct gl_context *ctx,
+                     const struct gl_client_array **arrays,
+                     const struct _mesa_prim *prims,
+                     GLuint nr_prims,
+                     const struct _mesa_index_buffer *ib,
+		     GLboolean index_bounds_valid,
+                     GLuint min_index,
+                     GLuint max_index)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct draw_context *draw = st->draw;
+   const struct st_vertex_program *vp;
+   const struct pipe_shader_state *vs;
+   struct pipe_vertex_buffer vbuffers[PIPE_MAX_SHADER_INPUTS];
+   struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS];
+   struct pipe_index_buffer ibuffer;
+   struct pipe_transfer *vb_transfer[PIPE_MAX_ATTRIBS];
+   struct pipe_transfer *ib_transfer = NULL;
+   GLuint attr, i;
+   const void *mapped_indices = NULL;
+
+   assert(draw);
+
+   st_validate_state(st);
+
+   if (!index_bounds_valid)
+      vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index);
+
+   /* must get these after state validation! */
+   vp = st->vp;
+   vs = &st->vp_variant->tgsi;
+
+   if (!st->vp_variant->draw_shader) {
+      st->vp_variant->draw_shader = draw_create_vertex_shader(draw, vs);
+   }
+
+   /*
+    * Set up the draw module's state.
+    *
+    * We'd like to do this less frequently, but the normal state-update
+    * code sends state updates to the pipe, not to our private draw module.
+    */
+   assert(draw);
+   draw_set_viewport_state(draw, &st->state.viewport);
+   draw_set_clip_state(draw, &st->state.clip);
+   draw_set_rasterizer_state(draw, &st->state.rasterizer, NULL);
+   draw_bind_vertex_shader(draw, st->vp_variant->draw_shader);
+   set_feedback_vertex_format(ctx);
+
+   /* loop over TGSI shader inputs to determine vertex buffer
+    * and attribute info
+    */
+   for (attr = 0; attr < vp->num_inputs; attr++) {
+      const GLuint mesaAttr = vp->index_to_input[attr];
+      struct gl_buffer_object *bufobj = arrays[mesaAttr]->BufferObj;
+      void *map;
+
+      if (bufobj && bufobj->Name) {
+         /* Attribute data is in a VBO.
+          * Recall that for VBOs, the gl_client_array->Ptr field is
+          * really an offset from the start of the VBO, not a pointer.
+          */
+         struct st_buffer_object *stobj = st_buffer_object(bufobj);
+         assert(stobj->buffer);
+
+         vbuffers[attr].buffer = NULL;
+         pipe_resource_reference(&vbuffers[attr].buffer, stobj->buffer);
+         vbuffers[attr].buffer_offset = pointer_to_offset(arrays[0]->Ptr);
+         velements[attr].src_offset = arrays[mesaAttr]->Ptr - arrays[0]->Ptr;
+      }
+      else {
+         /* attribute data is in user-space memory, not a VBO */
+         uint bytes = (arrays[mesaAttr]->Size
+                       * _mesa_sizeof_type(arrays[mesaAttr]->Type)
+                       * (max_index + 1));
+
+         /* wrap user data */
+         vbuffers[attr].buffer
+            = pipe_user_buffer_create(pipe->screen, (void *) arrays[mesaAttr]->Ptr,
+                                      bytes,
+				      PIPE_BIND_VERTEX_BUFFER);
+         vbuffers[attr].buffer_offset = 0;
+         velements[attr].src_offset = 0;
+      }
+
+      /* common-case setup */
+      vbuffers[attr].stride = arrays[mesaAttr]->StrideB; /* in bytes */
+      velements[attr].instance_divisor = 0;
+      velements[attr].vertex_buffer_index = attr;
+      velements[attr].src_format = 
+         st_pipe_vertex_format(arrays[mesaAttr]->Type,
+                               arrays[mesaAttr]->Size,
+                               arrays[mesaAttr]->Format,
+                               arrays[mesaAttr]->Normalized);
+      assert(velements[attr].src_format);
+
+      /* tell draw about this attribute */
+#if 0
+      draw_set_vertex_buffer(draw, attr, &vbuffer[attr]);
+#endif
+
+      /* map the attrib buffer */
+      map = pipe_buffer_map(pipe, vbuffers[attr].buffer,
+                            PIPE_TRANSFER_READ,
+			    &vb_transfer[attr]);
+      draw_set_mapped_vertex_buffer(draw, attr, map);
+   }
+
+   draw_set_vertex_buffers(draw, vp->num_inputs, vbuffers);
+   draw_set_vertex_elements(draw, vp->num_inputs, velements);
+
+   memset(&ibuffer, 0, sizeof(ibuffer));
+   if (ib) {
+      struct gl_buffer_object *bufobj = ib->obj;
+
+      switch (ib->type) {
+      case GL_UNSIGNED_INT:
+         ibuffer.index_size = 4;
+         break;
+      case GL_UNSIGNED_SHORT:
+         ibuffer.index_size = 2;
+         break;
+      case GL_UNSIGNED_BYTE:
+         ibuffer.index_size = 1;
+         break;
+      default:
+         assert(0);
+	 goto out_unref_vertex;
+      }
+
+      if (bufobj && bufobj->Name) {
+         struct st_buffer_object *stobj = st_buffer_object(bufobj);
+
+         pipe_resource_reference(&ibuffer.buffer, stobj->buffer);
+         ibuffer.offset = pointer_to_offset(ib->ptr);
+
+         mapped_indices = pipe_buffer_map(pipe, stobj->buffer,
+                                          PIPE_TRANSFER_READ, &ib_transfer);
+      }
+      else {
+         /* skip setting ibuffer.buffer as the draw module does not use it */
+         mapped_indices = ib->ptr;
+      }
+
+      draw_set_index_buffer(draw, &ibuffer);
+      draw_set_mapped_index_buffer(draw, mapped_indices);
+   }
+
+   /* set the constant buffer */
+   draw_set_mapped_constant_buffer(st->draw, PIPE_SHADER_VERTEX, 0,
+                                   st->state.constants[PIPE_SHADER_VERTEX].ptr,
+                                   st->state.constants[PIPE_SHADER_VERTEX].size);
+
+
+   /* draw here */
+   for (i = 0; i < nr_prims; i++) {
+      draw_arrays(draw, prims[i].mode, prims[i].start, prims[i].count);
+   }
+
+
+   /*
+    * unmap vertex/index buffers
+    */
+   if (ib) {
+      draw_set_mapped_index_buffer(draw, NULL);
+      draw_set_index_buffer(draw, NULL);
+
+      if (ib_transfer)
+         pipe_buffer_unmap(pipe, ib_transfer);
+      pipe_resource_reference(&ibuffer.buffer, NULL);
+   }
+
+ out_unref_vertex:
+   for (attr = 0; attr < vp->num_inputs; attr++) {
+      pipe_buffer_unmap(pipe, vb_transfer[attr]);
+      draw_set_mapped_vertex_buffer(draw, attr, NULL);
+      pipe_resource_reference(&vbuffers[attr].buffer, NULL);
+   }
+   draw_set_vertex_buffers(draw, 0, NULL);
+}
+
+#endif /* FEATURE_feedback || FEATURE_rastpos */
+
diff --git a/mesalib/src/mesa/state_tracker/st_gen_mipmap.c b/mesalib/src/mesa/state_tracker/st_gen_mipmap.c
index 18eb3be68..5fcd806e0 100644
--- a/mesalib/src/mesa/state_tracker/st_gen_mipmap.c
+++ b/mesalib/src/mesa/state_tracker/st_gen_mipmap.c
@@ -1,422 +1,422 @@
-/**************************************************************************
- * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "main/imports.h"
-#include "main/mipmap.h"
-#include "main/teximage.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_inlines.h"
-#include "util/u_format.h"
-#include "util/u_gen_mipmap.h"
-
-#include "st_debug.h"
-#include "st_context.h"
-#include "st_texture.h"
-#include "st_gen_mipmap.h"
-#include "st_cb_texture.h"
-
-
-/**
- * one-time init for generate mipmap
- * XXX Note: there may be other times we need no-op/simple state like this.
- * In that case, some code refactoring would be good.
- */
-void
-st_init_generate_mipmap(struct st_context *st)
-{
-   st->gen_mipmap = util_create_gen_mipmap(st->pipe, st->cso_context);
-}
-
-
-void
-st_destroy_generate_mipmap(struct st_context *st)
-{
-   util_destroy_gen_mipmap(st->gen_mipmap);
-   st->gen_mipmap = NULL;
-}
-
-
-/**
- * Generate mipmap levels using hardware rendering.
- * \return TRUE if successful, FALSE if not possible
- */
-static boolean
-st_render_mipmap(struct st_context *st,
-                 GLenum target,
-                 struct st_texture_object *stObj,
-                 uint baseLevel, uint lastLevel)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_screen *screen = pipe->screen;
-   struct pipe_sampler_view *psv = st_get_texture_sampler_view(stObj, pipe);
-   const uint face = _mesa_tex_target_to_face(target);
-
-   assert(psv->texture == stObj->pt);
-#if 0
-   assert(target != GL_TEXTURE_3D); /* implemented but untested */
-#endif
-
-   /* check if we can render in the texture's format */
-   /* XXX should probably kill this and always use util_gen_mipmap
-      since this implements a sw fallback as well */
-   if (!screen->is_format_supported(screen, psv->format, psv->texture->target,
-                                    0, PIPE_BIND_RENDER_TARGET, 0)) {
-      return FALSE;
-   }
-
-   util_gen_mipmap(st->gen_mipmap, psv, face, baseLevel, lastLevel,
-                   PIPE_TEX_FILTER_LINEAR);
-
-   return TRUE;
-}
-
-
-/**
- * Helper function to decompress an image.  The result is a 32-bpp RGBA
- * image with stride==width.
- */
-static void
-decompress_image(enum pipe_format format,
-                 const uint8_t *src, uint8_t *dst,
-                 unsigned width, unsigned height, unsigned src_stride)
-{
-   const struct util_format_description *desc = util_format_description(format);
-   const uint bw = util_format_get_blockwidth(format);
-   const uint bh = util_format_get_blockheight(format);
-   const uint dst_stride = 4 * MAX2(width, bw);
-
-   desc->unpack_rgba_8unorm(dst, dst_stride, src, src_stride, width, height);
-
-   if (width < bw || height < bh) {
-      /* We're decompressing an image smaller than the compression
-       * block size.  We don't want garbage pixel values in the region
-       * outside (width x height) so replicate pixels from the (width
-       * x height) region to fill out the (bw x bh) block size.
-       */
-      uint x, y;
-      for (y = 0; y < bh; y++) {
-         for (x = 0; x < bw; x++) {
-            if (x >= width || y >= height) {
-               uint p = (y * bw + x) * 4;
-               dst[p + 0] = dst[0];
-               dst[p + 1] = dst[1];
-               dst[p + 2] = dst[2];
-               dst[p + 3] = dst[3];
-            }
-         }
-      }
-   }
-}
-
-
-/**
- * Helper function to compress an image.  The source is a 32-bpp RGBA image
- * with stride==width.
- */
-static void
-compress_image(enum pipe_format format,
-               const uint8_t *src, uint8_t *dst,
-               unsigned width, unsigned height, unsigned dst_stride)
-{
-   const struct util_format_description *desc = util_format_description(format);
-   const uint src_stride = 4 * width;
-
-   desc->pack_rgba_8unorm(dst, dst_stride, src, src_stride, width, height);
-}
-
-
-/**
- * Software fallback for generate mipmap levels.
- */
-static void
-fallback_generate_mipmap(struct gl_context *ctx, GLenum target,
-                         struct gl_texture_object *texObj)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   struct pipe_resource *pt = st_get_texobj_resource(texObj);
-   const uint baseLevel = texObj->BaseLevel;
-   const uint lastLevel = pt->last_level;
-   const uint face = _mesa_tex_target_to_face(target);
-   uint dstLevel;
-   GLenum datatype;
-   GLuint comps;
-   GLboolean compressed;
-
-   if (ST_DEBUG & DEBUG_FALLBACK)
-      debug_printf("%s: fallback processing\n", __FUNCTION__);
-
-   assert(target != GL_TEXTURE_3D); /* not done yet */
-
-   compressed =
-      _mesa_is_format_compressed(texObj->Image[face][baseLevel]->TexFormat);
-
-   if (compressed) {
-      datatype = GL_UNSIGNED_BYTE;
-      comps = 4;
-   }
-   else {
-      _mesa_format_to_type_and_comps(texObj->Image[face][baseLevel]->TexFormat,
-                                     &datatype, &comps);
-      assert(comps > 0 && "bad texture format in fallback_generate_mipmap()");
-   }
-
-   for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
-      const uint srcLevel = dstLevel - 1;
-      const uint srcWidth = u_minify(pt->width0, srcLevel);
-      const uint srcHeight = u_minify(pt->height0, srcLevel);
-      const uint srcDepth = u_minify(pt->depth0, srcLevel);
-      const uint dstWidth = u_minify(pt->width0, dstLevel);
-      const uint dstHeight = u_minify(pt->height0, dstLevel);
-      const uint dstDepth = u_minify(pt->depth0, dstLevel);
-      struct pipe_transfer *srcTrans, *dstTrans;
-      const ubyte *srcData;
-      ubyte *dstData;
-      int srcStride, dstStride;
-
-      srcTrans = pipe_get_transfer(st_context(ctx)->pipe, pt, srcLevel,
-                                   face,
-                                   PIPE_TRANSFER_READ, 0, 0,
-                                   srcWidth, srcHeight);
-
-      dstTrans = pipe_get_transfer(st_context(ctx)->pipe, pt, dstLevel,
-                                   face,
-                                   PIPE_TRANSFER_WRITE, 0, 0,
-                                   dstWidth, dstHeight);
-
-      srcData = (ubyte *) pipe_transfer_map(pipe, srcTrans);
-      dstData = (ubyte *) pipe_transfer_map(pipe, dstTrans);
-
-      srcStride = srcTrans->stride / util_format_get_blocksize(srcTrans->resource->format);
-      dstStride = dstTrans->stride / util_format_get_blocksize(dstTrans->resource->format);
-
-     /* this cannot work correctly for 3d since it does
-        not respect layerStride. */
-      if (compressed) {
-         const enum pipe_format format = pt->format;
-         const uint bw = util_format_get_blockwidth(format);
-         const uint bh = util_format_get_blockheight(format);
-         const uint srcWidth2 = align(srcWidth, bw);
-         const uint srcHeight2 = align(srcHeight, bh);
-         const uint dstWidth2 = align(dstWidth, bw);
-         const uint dstHeight2 = align(dstHeight, bh);
-         uint8_t *srcTemp, *dstTemp;
-
-         assert(comps == 4);
-
-         srcTemp = malloc(srcWidth2 * srcHeight2 * comps + 000);
-         dstTemp = malloc(dstWidth2 * dstHeight2 * comps + 000);
-
-         /* decompress the src image: srcData -> srcTemp */
-         decompress_image(format, srcData, srcTemp, srcWidth, srcHeight, srcTrans->stride);
-
-         _mesa_generate_mipmap_level(target, datatype, comps,
-                                     0 /*border*/,
-                                     srcWidth2, srcHeight2, srcDepth,
-                                     srcTemp,
-                                     srcWidth2, /* stride in texels */
-                                     dstWidth2, dstHeight2, dstDepth,
-                                     dstTemp,
-                                     dstWidth2); /* stride in texels */
-
-         /* compress the new image: dstTemp -> dstData */
-         compress_image(format, dstTemp, dstData, dstWidth, dstHeight, dstTrans->stride);
-
-         free(srcTemp);
-         free(dstTemp);
-      }
-      else {
-         _mesa_generate_mipmap_level(target, datatype, comps,
-                                     0 /*border*/,
-                                     srcWidth, srcHeight, srcDepth,
-                                     srcData,
-                                     srcStride, /* stride in texels */
-                                     dstWidth, dstHeight, dstDepth,
-                                     dstData,
-                                     dstStride); /* stride in texels */
-      }
-
-      pipe_transfer_unmap(pipe, srcTrans);
-      pipe_transfer_unmap(pipe, dstTrans);
-
-      pipe->transfer_destroy(pipe, srcTrans);
-      pipe->transfer_destroy(pipe, dstTrans);
-   }
-}
-
-
-/**
- * Compute the expected number of mipmap levels in the texture given
- * the width/height/depth of the base image and the GL_TEXTURE_BASE_LEVEL/
- * GL_TEXTURE_MAX_LEVEL settings.  This will tell us how many mipmap
- * levels should be generated.
- */
-static GLuint
-compute_num_levels(struct gl_context *ctx,
-                   struct gl_texture_object *texObj,
-                   GLenum target)
-{
-   if (target == GL_TEXTURE_RECTANGLE_ARB) {
-      return 1;
-   }
-   else {
-      const struct gl_texture_image *baseImage = 
-         _mesa_get_tex_image(ctx, texObj, target, texObj->BaseLevel);
-      GLuint size, numLevels;
-
-      size = MAX2(baseImage->Width2, baseImage->Height2);
-      size = MAX2(size, baseImage->Depth2);
-
-      numLevels = texObj->BaseLevel;
-
-      while (size > 0) {
-         numLevels++;
-         size >>= 1;
-      }
-
-      numLevels = MIN2(numLevels, texObj->MaxLevel + 1);
-
-      assert(numLevels >= 1);
-
-      return numLevels;
-   }
-}
-
-
-/**
- * Called via ctx->Driver.GenerateMipmap().
- */
-void
-st_generate_mipmap(struct gl_context *ctx, GLenum target,
-                   struct gl_texture_object *texObj)
-{
-   struct st_context *st = st_context(ctx);
-   struct st_texture_object *stObj = st_texture_object(texObj);
-   struct pipe_resource *pt = st_get_texobj_resource(texObj);
-   const uint baseLevel = texObj->BaseLevel;
-   uint lastLevel;
-   uint dstLevel;
-
-   if (!pt)
-      return;
-
-   /* not sure if this ultimately actually should work,
-      but we're not supporting multisampled textures yet. */
-   assert(pt->nr_samples < 2);
-
-   /* find expected last mipmap level to generate*/
-   lastLevel = compute_num_levels(ctx, texObj, target) - 1;
-
-   if (lastLevel == 0)
-      return;
-
-   if (pt->last_level < lastLevel) {
-      /* The current gallium texture doesn't have space for all the
-       * mipmap levels we need to generate.  So allocate a new texture.
-       */
-      struct pipe_resource *oldTex = stObj->pt;
-
-      /* create new texture with space for more levels */
-      stObj->pt = st_texture_create(st,
-                                    oldTex->target,
-                                    oldTex->format,
-                                    lastLevel,
-                                    oldTex->width0,
-                                    oldTex->height0,
-                                    oldTex->depth0,
-                                    oldTex->array_size,
-                                    oldTex->bind);
-
-      /* The texture isn't in a "complete" state yet so set the expected
-       * lastLevel here, since it won't get done in st_finalize_texture().
-       */
-      stObj->lastLevel = lastLevel;
-
-      /* This will copy the old texture's base image into the new texture
-       * which we just allocated.
-       */
-      st_finalize_texture(ctx, st->pipe, texObj);
-
-      /* release the old tex (will likely be freed too) */
-      pipe_resource_reference(&oldTex, NULL);
-      pipe_sampler_view_reference(&stObj->sampler_view, NULL);
-
-      pt = stObj->pt;
-   }
-   else {
-      /* Make sure that the base texture image data is present in the
-       * texture buffer.
-       */
-      st_finalize_texture(ctx, st->pipe, texObj);
-   }
-
-   assert(pt->last_level >= lastLevel);
-
-   /* Try to generate the mipmap by rendering/texturing.  If that fails,
-    * use the software fallback.
-    */
-   if (!st_render_mipmap(st, target, stObj, baseLevel, lastLevel)) {
-      /* since the util code actually also has a fallback, should
-         probably make it never fail and kill this */
-      fallback_generate_mipmap(ctx, target, texObj);
-   }
-
-   /* Fill in the Mesa gl_texture_image fields */
-   for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
-      const uint srcLevel = dstLevel - 1;
-      const struct gl_texture_image *srcImage
-         = _mesa_get_tex_image(ctx, texObj, target, srcLevel);
-      struct gl_texture_image *dstImage;
-      struct st_texture_image *stImage;
-      uint dstWidth = u_minify(pt->width0, dstLevel);
-      uint dstHeight = u_minify(pt->height0, dstLevel);
-      uint dstDepth = u_minify(pt->depth0, dstLevel); 
-      uint border = srcImage->Border;
-
-      dstImage = _mesa_get_tex_image(ctx, texObj, target, dstLevel);
-      if (!dstImage) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "generating mipmaps");
-         return;
-      }
-
-      /* Free old image data */
-      if (dstImage->Data)
-         ctx->Driver.FreeTexImageData(ctx, dstImage);
-
-      /* initialize new image */
-      _mesa_init_teximage_fields(ctx, target, dstImage, dstWidth, dstHeight,
-                                 dstDepth, border, srcImage->InternalFormat,
-                                 srcImage->TexFormat);
-
-      stImage = st_texture_image(dstImage);
-      stImage->level = dstLevel;
-
-      pipe_resource_reference(&stImage->pt, pt);
-   }
-}
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "main/imports.h"
+#include "main/mipmap.h"
+#include "main/teximage.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "util/u_inlines.h"
+#include "util/u_format.h"
+#include "util/u_gen_mipmap.h"
+
+#include "st_debug.h"
+#include "st_context.h"
+#include "st_texture.h"
+#include "st_gen_mipmap.h"
+#include "st_cb_texture.h"
+
+
+/**
+ * one-time init for generate mipmap
+ * XXX Note: there may be other times we need no-op/simple state like this.
+ * In that case, some code refactoring would be good.
+ */
+void
+st_init_generate_mipmap(struct st_context *st)
+{
+   st->gen_mipmap = util_create_gen_mipmap(st->pipe, st->cso_context);
+}
+
+
+void
+st_destroy_generate_mipmap(struct st_context *st)
+{
+   util_destroy_gen_mipmap(st->gen_mipmap);
+   st->gen_mipmap = NULL;
+}
+
+
+/**
+ * Generate mipmap levels using hardware rendering.
+ * \return TRUE if successful, FALSE if not possible
+ */
+static boolean
+st_render_mipmap(struct st_context *st,
+                 GLenum target,
+                 struct st_texture_object *stObj,
+                 uint baseLevel, uint lastLevel)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_screen *screen = pipe->screen;
+   struct pipe_sampler_view *psv = st_get_texture_sampler_view(stObj, pipe);
+   const uint face = _mesa_tex_target_to_face(target);
+
+   assert(psv->texture == stObj->pt);
+#if 0
+   assert(target != GL_TEXTURE_3D); /* implemented but untested */
+#endif
+
+   /* check if we can render in the texture's format */
+   /* XXX should probably kill this and always use util_gen_mipmap
+      since this implements a sw fallback as well */
+   if (!screen->is_format_supported(screen, psv->format, psv->texture->target,
+                                    0, PIPE_BIND_RENDER_TARGET, 0)) {
+      return FALSE;
+   }
+
+   util_gen_mipmap(st->gen_mipmap, psv, face, baseLevel, lastLevel,
+                   PIPE_TEX_FILTER_LINEAR);
+
+   return TRUE;
+}
+
+
+/**
+ * Helper function to decompress an image.  The result is a 32-bpp RGBA
+ * image with stride==width.
+ */
+static void
+decompress_image(enum pipe_format format,
+                 const uint8_t *src, uint8_t *dst,
+                 unsigned width, unsigned height, unsigned src_stride)
+{
+   const struct util_format_description *desc = util_format_description(format);
+   const uint bw = util_format_get_blockwidth(format);
+   const uint bh = util_format_get_blockheight(format);
+   const uint dst_stride = 4 * MAX2(width, bw);
+
+   desc->unpack_rgba_8unorm(dst, dst_stride, src, src_stride, width, height);
+
+   if (width < bw || height < bh) {
+      /* We're decompressing an image smaller than the compression
+       * block size.  We don't want garbage pixel values in the region
+       * outside (width x height) so replicate pixels from the (width
+       * x height) region to fill out the (bw x bh) block size.
+       */
+      uint x, y;
+      for (y = 0; y < bh; y++) {
+         for (x = 0; x < bw; x++) {
+            if (x >= width || y >= height) {
+               uint p = (y * bw + x) * 4;
+               dst[p + 0] = dst[0];
+               dst[p + 1] = dst[1];
+               dst[p + 2] = dst[2];
+               dst[p + 3] = dst[3];
+            }
+         }
+      }
+   }
+}
+
+
+/**
+ * Helper function to compress an image.  The source is a 32-bpp RGBA image
+ * with stride==width.
+ */
+static void
+compress_image(enum pipe_format format,
+               const uint8_t *src, uint8_t *dst,
+               unsigned width, unsigned height, unsigned dst_stride)
+{
+   const struct util_format_description *desc = util_format_description(format);
+   const uint src_stride = 4 * width;
+
+   desc->pack_rgba_8unorm(dst, dst_stride, src, src_stride, width, height);
+}
+
+
+/**
+ * Software fallback for generate mipmap levels.
+ */
+static void
+fallback_generate_mipmap(struct gl_context *ctx, GLenum target,
+                         struct gl_texture_object *texObj)
+{
+   struct pipe_context *pipe = st_context(ctx)->pipe;
+   struct pipe_resource *pt = st_get_texobj_resource(texObj);
+   const uint baseLevel = texObj->BaseLevel;
+   const uint lastLevel = pt->last_level;
+   const uint face = _mesa_tex_target_to_face(target);
+   uint dstLevel;
+   GLenum datatype;
+   GLuint comps;
+   GLboolean compressed;
+
+   if (ST_DEBUG & DEBUG_FALLBACK)
+      debug_printf("%s: fallback processing\n", __FUNCTION__);
+
+   assert(target != GL_TEXTURE_3D); /* not done yet */
+
+   compressed =
+      _mesa_is_format_compressed(texObj->Image[face][baseLevel]->TexFormat);
+
+   if (compressed) {
+      datatype = GL_UNSIGNED_BYTE;
+      comps = 4;
+   }
+   else {
+      _mesa_format_to_type_and_comps(texObj->Image[face][baseLevel]->TexFormat,
+                                     &datatype, &comps);
+      assert(comps > 0 && "bad texture format in fallback_generate_mipmap()");
+   }
+
+   for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
+      const uint srcLevel = dstLevel - 1;
+      const uint srcWidth = u_minify(pt->width0, srcLevel);
+      const uint srcHeight = u_minify(pt->height0, srcLevel);
+      const uint srcDepth = u_minify(pt->depth0, srcLevel);
+      const uint dstWidth = u_minify(pt->width0, dstLevel);
+      const uint dstHeight = u_minify(pt->height0, dstLevel);
+      const uint dstDepth = u_minify(pt->depth0, dstLevel);
+      struct pipe_transfer *srcTrans, *dstTrans;
+      const ubyte *srcData;
+      ubyte *dstData;
+      int srcStride, dstStride;
+
+      srcTrans = pipe_get_transfer(st_context(ctx)->pipe, pt, srcLevel,
+                                   face,
+                                   PIPE_TRANSFER_READ, 0, 0,
+                                   srcWidth, srcHeight);
+
+      dstTrans = pipe_get_transfer(st_context(ctx)->pipe, pt, dstLevel,
+                                   face,
+                                   PIPE_TRANSFER_WRITE, 0, 0,
+                                   dstWidth, dstHeight);
+
+      srcData = (ubyte *) pipe_transfer_map(pipe, srcTrans);
+      dstData = (ubyte *) pipe_transfer_map(pipe, dstTrans);
+
+      srcStride = srcTrans->stride / util_format_get_blocksize(srcTrans->resource->format);
+      dstStride = dstTrans->stride / util_format_get_blocksize(dstTrans->resource->format);
+
+     /* this cannot work correctly for 3d since it does
+        not respect layerStride. */
+      if (compressed) {
+         const enum pipe_format format = pt->format;
+         const uint bw = util_format_get_blockwidth(format);
+         const uint bh = util_format_get_blockheight(format);
+         const uint srcWidth2 = align(srcWidth, bw);
+         const uint srcHeight2 = align(srcHeight, bh);
+         const uint dstWidth2 = align(dstWidth, bw);
+         const uint dstHeight2 = align(dstHeight, bh);
+         uint8_t *srcTemp, *dstTemp;
+
+         assert(comps == 4);
+
+         srcTemp = malloc(srcWidth2 * srcHeight2 * comps + 000);
+         dstTemp = malloc(dstWidth2 * dstHeight2 * comps + 000);
+
+         /* decompress the src image: srcData -> srcTemp */
+         decompress_image(format, srcData, srcTemp, srcWidth, srcHeight, srcTrans->stride);
+
+         _mesa_generate_mipmap_level(target, datatype, comps,
+                                     0 /*border*/,
+                                     srcWidth2, srcHeight2, srcDepth,
+                                     srcTemp,
+                                     srcWidth2, /* stride in texels */
+                                     dstWidth2, dstHeight2, dstDepth,
+                                     dstTemp,
+                                     dstWidth2); /* stride in texels */
+
+         /* compress the new image: dstTemp -> dstData */
+         compress_image(format, dstTemp, dstData, dstWidth, dstHeight, dstTrans->stride);
+
+         free(srcTemp);
+         free(dstTemp);
+      }
+      else {
+         _mesa_generate_mipmap_level(target, datatype, comps,
+                                     0 /*border*/,
+                                     srcWidth, srcHeight, srcDepth,
+                                     srcData,
+                                     srcStride, /* stride in texels */
+                                     dstWidth, dstHeight, dstDepth,
+                                     dstData,
+                                     dstStride); /* stride in texels */
+      }
+
+      pipe_transfer_unmap(pipe, srcTrans);
+      pipe_transfer_unmap(pipe, dstTrans);
+
+      pipe->transfer_destroy(pipe, srcTrans);
+      pipe->transfer_destroy(pipe, dstTrans);
+   }
+}
+
+
+/**
+ * Compute the expected number of mipmap levels in the texture given
+ * the width/height/depth of the base image and the GL_TEXTURE_BASE_LEVEL/
+ * GL_TEXTURE_MAX_LEVEL settings.  This will tell us how many mipmap
+ * levels should be generated.
+ */
+static GLuint
+compute_num_levels(struct gl_context *ctx,
+                   struct gl_texture_object *texObj,
+                   GLenum target)
+{
+   if (target == GL_TEXTURE_RECTANGLE_ARB) {
+      return 1;
+   }
+   else {
+      const struct gl_texture_image *baseImage = 
+         _mesa_get_tex_image(ctx, texObj, target, texObj->BaseLevel);
+      GLuint size, numLevels;
+
+      size = MAX2(baseImage->Width2, baseImage->Height2);
+      size = MAX2(size, baseImage->Depth2);
+
+      numLevels = texObj->BaseLevel;
+
+      while (size > 0) {
+         numLevels++;
+         size >>= 1;
+      }
+
+      numLevels = MIN2(numLevels, texObj->MaxLevel + 1);
+
+      assert(numLevels >= 1);
+
+      return numLevels;
+   }
+}
+
+
+/**
+ * Called via ctx->Driver.GenerateMipmap().
+ */
+void
+st_generate_mipmap(struct gl_context *ctx, GLenum target,
+                   struct gl_texture_object *texObj)
+{
+   struct st_context *st = st_context(ctx);
+   struct st_texture_object *stObj = st_texture_object(texObj);
+   struct pipe_resource *pt = st_get_texobj_resource(texObj);
+   const uint baseLevel = texObj->BaseLevel;
+   uint lastLevel;
+   uint dstLevel;
+
+   if (!pt)
+      return;
+
+   /* not sure if this ultimately actually should work,
+      but we're not supporting multisampled textures yet. */
+   assert(pt->nr_samples < 2);
+
+   /* find expected last mipmap level to generate*/
+   lastLevel = compute_num_levels(ctx, texObj, target) - 1;
+
+   if (lastLevel == 0)
+      return;
+
+   if (pt->last_level < lastLevel) {
+      /* The current gallium texture doesn't have space for all the
+       * mipmap levels we need to generate.  So allocate a new texture.
+       */
+      struct pipe_resource *oldTex = stObj->pt;
+
+      /* create new texture with space for more levels */
+      stObj->pt = st_texture_create(st,
+                                    oldTex->target,
+                                    oldTex->format,
+                                    lastLevel,
+                                    oldTex->width0,
+                                    oldTex->height0,
+                                    oldTex->depth0,
+                                    oldTex->array_size,
+                                    oldTex->bind);
+
+      /* The texture isn't in a "complete" state yet so set the expected
+       * lastLevel here, since it won't get done in st_finalize_texture().
+       */
+      stObj->lastLevel = lastLevel;
+
+      /* This will copy the old texture's base image into the new texture
+       * which we just allocated.
+       */
+      st_finalize_texture(ctx, st->pipe, texObj);
+
+      /* release the old tex (will likely be freed too) */
+      pipe_resource_reference(&oldTex, NULL);
+      pipe_sampler_view_reference(&stObj->sampler_view, NULL);
+
+      pt = stObj->pt;
+   }
+   else {
+      /* Make sure that the base texture image data is present in the
+       * texture buffer.
+       */
+      st_finalize_texture(ctx, st->pipe, texObj);
+   }
+
+   assert(pt->last_level >= lastLevel);
+
+   /* Try to generate the mipmap by rendering/texturing.  If that fails,
+    * use the software fallback.
+    */
+   if (!st_render_mipmap(st, target, stObj, baseLevel, lastLevel)) {
+      /* since the util code actually also has a fallback, should
+         probably make it never fail and kill this */
+      fallback_generate_mipmap(ctx, target, texObj);
+   }
+
+   /* Fill in the Mesa gl_texture_image fields */
+   for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
+      const uint srcLevel = dstLevel - 1;
+      const struct gl_texture_image *srcImage
+         = _mesa_get_tex_image(ctx, texObj, target, srcLevel);
+      struct gl_texture_image *dstImage;
+      struct st_texture_image *stImage;
+      uint dstWidth = u_minify(pt->width0, dstLevel);
+      uint dstHeight = u_minify(pt->height0, dstLevel);
+      uint dstDepth = u_minify(pt->depth0, dstLevel); 
+      uint border = srcImage->Border;
+
+      dstImage = _mesa_get_tex_image(ctx, texObj, target, dstLevel);
+      if (!dstImage) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "generating mipmaps");
+         return;
+      }
+
+      /* Free old image data */
+      if (dstImage->Data)
+         ctx->Driver.FreeTexImageData(ctx, dstImage);
+
+      /* initialize new image */
+      _mesa_init_teximage_fields(ctx, target, dstImage, dstWidth, dstHeight,
+                                 dstDepth, border, srcImage->InternalFormat,
+                                 srcImage->TexFormat);
+
+      stImage = st_texture_image(dstImage);
+      stImage->level = dstLevel;
+
+      pipe_resource_reference(&stImage->pt, pt);
+   }
+}
diff --git a/mesalib/src/mesa/state_tracker/st_program.c b/mesalib/src/mesa/state_tracker/st_program.c
index fc1dfb3ef..e5c26c563 100644
--- a/mesalib/src/mesa/state_tracker/st_program.c
+++ b/mesalib/src/mesa/state_tracker/st_program.c
@@ -1,1165 +1,1165 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  *   Brian Paul
-  */
-
-
-#include "main/imports.h"
-#include "main/hash.h"
-#include "main/mfeatures.h"
-#include "main/mtypes.h"
-#include "program/prog_parameter.h"
-#include "program/prog_print.h"
-#include "program/programopt.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_shader_tokens.h"
-#include "draw/draw_context.h"
-#include "tgsi/tgsi_dump.h"
-#include "tgsi/tgsi_ureg.h"
-
-#include "st_debug.h"
-#include "st_cb_bitmap.h"
-#include "st_cb_drawpixels.h"
-#include "st_context.h"
-#include "st_program.h"
-#include "st_mesa_to_tgsi.h"
-#include "cso_cache/cso_context.h"
-
-
-
-/**
- * Delete a vertex program variant.  Note the caller must unlink
- * the variant from the linked list.
- */
-static void
-delete_vp_variant(struct st_context *st, struct st_vp_variant *vpv)
-{
-   if (vpv->driver_shader) 
-      cso_delete_vertex_shader(st->cso_context, vpv->driver_shader);
-      
-#if FEATURE_feedback || FEATURE_rastpos
-   if (vpv->draw_shader)
-      draw_delete_vertex_shader( st->draw, vpv->draw_shader );
-#endif
-      
-   if (vpv->tgsi.tokens)
-      st_free_tokens(vpv->tgsi.tokens);
-      
-   FREE( vpv );
-}
-
-
-
-/**
- * Clean out any old compilations:
- */
-void
-st_release_vp_variants( struct st_context *st,
-                        struct st_vertex_program *stvp )
-{
-   struct st_vp_variant *vpv;
-
-   for (vpv = stvp->variants; vpv; ) {
-      struct st_vp_variant *next = vpv->next;
-      delete_vp_variant(st, vpv);
-      vpv = next;
-   }
-
-   stvp->variants = NULL;
-}
-
-
-
-/**
- * Delete a fragment program variant.  Note the caller must unlink
- * the variant from the linked list.
- */
-static void
-delete_fp_variant(struct st_context *st, struct st_fp_variant *fpv)
-{
-   if (fpv->driver_shader) 
-      cso_delete_fragment_shader(st->cso_context, fpv->driver_shader);
-   if (fpv->parameters)
-      _mesa_free_parameter_list(fpv->parameters);
-      
-   FREE(fpv);
-}
-
-
-/**
- * Free all variants of a fragment program.
- */
-void
-st_release_fp_variants(struct st_context *st, struct st_fragment_program *stfp)
-{
-   struct st_fp_variant *fpv;
-
-   for (fpv = stfp->variants; fpv; ) {
-      struct st_fp_variant *next = fpv->next;
-      delete_fp_variant(st, fpv);
-      fpv = next;
-   }
-
-   stfp->variants = NULL;
-}
-
-
-/**
- * Delete a geometry program variant.  Note the caller must unlink
- * the variant from the linked list.
- */
-static void
-delete_gp_variant(struct st_context *st, struct st_gp_variant *gpv)
-{
-   if (gpv->driver_shader) 
-      cso_delete_geometry_shader(st->cso_context, gpv->driver_shader);
-      
-   FREE(gpv);
-}
-
-
-/**
- * Free all variants of a geometry program.
- */
-void
-st_release_gp_variants(struct st_context *st, struct st_geometry_program *stgp)
-{
-   struct st_gp_variant *gpv;
-
-   for (gpv = stgp->variants; gpv; ) {
-      struct st_gp_variant *next = gpv->next;
-      delete_gp_variant(st, gpv);
-      gpv = next;
-   }
-
-   stgp->variants = NULL;
-}
-
-
-
-
-/**
- * Translate a Mesa vertex shader into a TGSI shader.
- * \param outputMapping  to map vertex program output registers (VERT_RESULT_x)
- *       to TGSI output slots
- * \param tokensOut  destination for TGSI tokens
- * \return  pointer to cached pipe_shader object.
- */
-static void
-st_prepare_vertex_program(struct st_context *st,
-                            struct st_vertex_program *stvp)
-{
-   GLuint attr;
-
-   stvp->num_inputs = 0;
-   stvp->num_outputs = 0;
-
-   if (stvp->Base.IsPositionInvariant)
-      _mesa_insert_mvp_code(st->ctx, &stvp->Base);
-
-   assert(stvp->Base.Base.NumInstructions > 1);
-
-   /*
-    * Determine number of inputs, the mappings between VERT_ATTRIB_x
-    * and TGSI generic input indexes, plus input attrib semantic info.
-    */
-   for (attr = 0; attr < VERT_ATTRIB_MAX; attr++) {
-      if (stvp->Base.Base.InputsRead & (1 << attr)) {
-         stvp->input_to_index[attr] = stvp->num_inputs;
-         stvp->index_to_input[stvp->num_inputs] = attr;
-         stvp->num_inputs++;
-      }
-   }
-   /* bit of a hack, presetup potentially unused edgeflag input */
-   stvp->input_to_index[VERT_ATTRIB_EDGEFLAG] = stvp->num_inputs;
-   stvp->index_to_input[stvp->num_inputs] = VERT_ATTRIB_EDGEFLAG;
-
-   /* Compute mapping of vertex program outputs to slots.
-    */
-   for (attr = 0; attr < VERT_RESULT_MAX; attr++) {
-      if ((stvp->Base.Base.OutputsWritten & BITFIELD64_BIT(attr)) == 0) {
-         stvp->result_to_output[attr] = ~0;
-      }
-      else {
-         unsigned slot = stvp->num_outputs++;
-
-         stvp->result_to_output[attr] = slot;
-
-         switch (attr) {
-         case VERT_RESULT_HPOS:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-            stvp->output_semantic_index[slot] = 0;
-            break;
-         case VERT_RESULT_COL0:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            stvp->output_semantic_index[slot] = 0;
-            break;
-         case VERT_RESULT_COL1:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            stvp->output_semantic_index[slot] = 1;
-            break;
-         case VERT_RESULT_BFC0:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            stvp->output_semantic_index[slot] = 0;
-            break;
-         case VERT_RESULT_BFC1:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            stvp->output_semantic_index[slot] = 1;
-            break;
-         case VERT_RESULT_FOGC:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-            stvp->output_semantic_index[slot] = 0;
-            break;
-         case VERT_RESULT_PSIZ:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
-            stvp->output_semantic_index[slot] = 0;
-            break;
-         case VERT_RESULT_EDGE:
-            assert(0);
-            break;
-
-         case VERT_RESULT_TEX0:
-         case VERT_RESULT_TEX1:
-         case VERT_RESULT_TEX2:
-         case VERT_RESULT_TEX3:
-         case VERT_RESULT_TEX4:
-         case VERT_RESULT_TEX5:
-         case VERT_RESULT_TEX6:
-         case VERT_RESULT_TEX7:
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stvp->output_semantic_index[slot] = attr - VERT_RESULT_TEX0;
-            break;
-
-         case VERT_RESULT_VAR0:
-         default:
-            assert(attr < VERT_RESULT_MAX);
-            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stvp->output_semantic_index[slot] = (FRAG_ATTRIB_VAR0 - 
-                                                FRAG_ATTRIB_TEX0 +
-                                                attr - 
-                                                VERT_RESULT_VAR0);
-            break;
-         }
-      }
-   }
-   /* similar hack to above, presetup potentially unused edgeflag output */
-   stvp->result_to_output[VERT_RESULT_EDGE] = stvp->num_outputs;
-   stvp->output_semantic_name[stvp->num_outputs] = TGSI_SEMANTIC_EDGEFLAG;
-   stvp->output_semantic_index[stvp->num_outputs] = 0;
-}
-
-
-/**
- * Translate a vertex program to create a new variant.
- */
-static struct st_vp_variant *
-st_translate_vertex_program(struct st_context *st,
-                            struct st_vertex_program *stvp,
-                            const struct st_vp_variant_key *key)
-{
-   struct st_vp_variant *vpv = CALLOC_STRUCT(st_vp_variant);
-   struct pipe_context *pipe = st->pipe;
-   struct ureg_program *ureg;
-   enum pipe_error error;
-   unsigned num_outputs;
-
-   st_prepare_vertex_program( st, stvp );
-
-   _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_OUTPUT);
-   _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_VARYING);
-
-   ureg = ureg_create( TGSI_PROCESSOR_VERTEX );
-   if (ureg == NULL) {
-      FREE(vpv);
-      return NULL;
-   }
-
-   vpv->key = *key;
-
-   vpv->num_inputs = stvp->num_inputs;
-   num_outputs = stvp->num_outputs;
-   if (key->passthrough_edgeflags) {
-      vpv->num_inputs++;
-      num_outputs++;
-   }
-
-   if (ST_DEBUG & DEBUG_MESA) {
-      _mesa_print_program(&stvp->Base.Base);
-      _mesa_print_program_parameters(st->ctx, &stvp->Base.Base);
-      debug_printf("\n");
-   }
-
-   error = st_translate_mesa_program(st->ctx,
-                                     TGSI_PROCESSOR_VERTEX,
-                                     ureg,
-                                     &stvp->Base.Base,
-                                     /* inputs */
-                                     vpv->num_inputs,
-                                     stvp->input_to_index,
-                                     NULL, /* input semantic name */
-                                     NULL, /* input semantic index */
-                                     NULL,
-                                     /* outputs */
-                                     num_outputs,
-                                     stvp->result_to_output,
-                                     stvp->output_semantic_name,
-                                     stvp->output_semantic_index,
-                                     key->passthrough_edgeflags );
-
-   if (error)
-      goto fail;
-
-   vpv->tgsi.tokens = ureg_get_tokens( ureg, NULL );
-   if (!vpv->tgsi.tokens)
-      goto fail;
-
-   ureg_destroy( ureg );
-
-   vpv->driver_shader = pipe->create_vs_state(pipe, &vpv->tgsi);
-
-   if (ST_DEBUG & DEBUG_TGSI) {
-      tgsi_dump( vpv->tgsi.tokens, 0 );
-      debug_printf("\n");
-   }
-
-   return vpv;
-
-fail:
-   debug_printf("%s: failed to translate Mesa program:\n", __FUNCTION__);
-   _mesa_print_program(&stvp->Base.Base);
-   debug_assert(0);
-
-   ureg_destroy( ureg );
-   return NULL;
-}
-
-
-/**
- * Find/create a vertex program variant.
- */
-struct st_vp_variant *
-st_get_vp_variant(struct st_context *st,
-                  struct st_vertex_program *stvp,
-                  const struct st_vp_variant_key *key)
-{
-   struct st_vp_variant *vpv;
-
-   /* Search for existing variant */
-   for (vpv = stvp->variants; vpv; vpv = vpv->next) {
-      if (memcmp(&vpv->key, key, sizeof(*key)) == 0) {
-         break;
-      }
-   }
-
-   if (!vpv) {
-      /* create now */
-      vpv = st_translate_vertex_program(st, stvp, key);
-      if (vpv) {
-         /* insert into list */
-         vpv->next = stvp->variants;
-         stvp->variants = vpv;
-      }
-   }
-
-   return vpv;
-}
-
-
-/**
- * Translate a Mesa fragment shader into a TGSI shader using extra info in
- * the key.
- * \return  new fragment program variant
- */
-static struct st_fp_variant *
-st_translate_fragment_program(struct st_context *st,
-                              struct st_fragment_program *stfp,
-                              const struct st_fp_variant_key *key)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct st_fp_variant *variant = CALLOC_STRUCT(st_fp_variant);
-   GLboolean deleteFP = GL_FALSE;
-
-   if (!variant)
-      return NULL;
-
-   assert(!(key->bitmap && key->drawpixels));
-
-#if FEATURE_drawpix
-   if (key->bitmap) {
-      /* glBitmap drawing */
-      struct gl_fragment_program *fp; /* we free this temp program below */
-
-      st_make_bitmap_fragment_program(st, &stfp->Base,
-                                      &fp, &variant->bitmap_sampler);
-
-      variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters);
-      stfp = st_fragment_program(fp);
-      deleteFP = GL_TRUE;
-   }
-   else if (key->drawpixels) {
-      /* glDrawPixels drawing */
-      struct gl_fragment_program *fp; /* we free this temp program below */
-
-      if (key->drawpixels_z || key->drawpixels_stencil) {
-         fp = st_make_drawpix_z_stencil_program(st, key->drawpixels_z,
-                                                key->drawpixels_stencil);
-      }
-      else {
-         /* RGBA */
-         st_make_drawpix_fragment_program(st, &stfp->Base, &fp);
-         variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters);
-         deleteFP = GL_TRUE;
-      }
-      stfp = st_fragment_program(fp);
-   }
-#endif
-
-   if (!stfp->tgsi.tokens) {
-      /* need to translate Mesa instructions to TGSI now */
-      GLuint outputMapping[FRAG_RESULT_MAX];
-      GLuint inputMapping[FRAG_ATTRIB_MAX];
-      GLuint interpMode[PIPE_MAX_SHADER_INPUTS];  /* XXX size? */
-      GLuint attr;
-      enum pipe_error error;
-      const GLbitfield inputsRead = stfp->Base.Base.InputsRead;
-      struct ureg_program *ureg;
-      GLboolean write_all = GL_FALSE;
-
-      ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS];
-      ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS];
-      uint fs_num_inputs = 0;
-
-      ubyte fs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS];
-      ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
-      uint fs_num_outputs = 0;
-
-
-      _mesa_remove_output_reads(&stfp->Base.Base, PROGRAM_OUTPUT);
-
-      /*
-       * Convert Mesa program inputs to TGSI input register semantics.
-       */
-      for (attr = 0; attr < FRAG_ATTRIB_MAX; attr++) {
-         if (inputsRead & (1 << attr)) {
-            const GLuint slot = fs_num_inputs++;
-
-            inputMapping[attr] = slot;
-
-            switch (attr) {
-            case FRAG_ATTRIB_WPOS:
-               input_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-               input_semantic_index[slot] = 0;
-               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
-               break;
-            case FRAG_ATTRIB_COL0:
-               input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-               input_semantic_index[slot] = 0;
-               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
-               break;
-            case FRAG_ATTRIB_COL1:
-               input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-               input_semantic_index[slot] = 1;
-               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
-               break;
-            case FRAG_ATTRIB_FOGC:
-               input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-               input_semantic_index[slot] = 0;
-               interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
-               break;
-            case FRAG_ATTRIB_FACE:
-               input_semantic_name[slot] = TGSI_SEMANTIC_FACE;
-               input_semantic_index[slot] = 0;
-               interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
-               break;
-               /* In most cases, there is nothing special about these
-                * inputs, so adopt a convention to use the generic
-                * semantic name and the mesa FRAG_ATTRIB_ number as the
-                * index. 
-                * 
-                * All that is required is that the vertex shader labels
-                * its own outputs similarly, and that the vertex shader
-                * generates at least every output required by the
-                * fragment shader plus fixed-function hardware (such as
-                * BFC).
-                * 
-                * There is no requirement that semantic indexes start at
-                * zero or be restricted to a particular range -- nobody
-                * should be building tables based on semantic index.
-                */
-            case FRAG_ATTRIB_PNTC:
-            case FRAG_ATTRIB_TEX0:
-            case FRAG_ATTRIB_TEX1:
-            case FRAG_ATTRIB_TEX2:
-            case FRAG_ATTRIB_TEX3:
-            case FRAG_ATTRIB_TEX4:
-            case FRAG_ATTRIB_TEX5:
-            case FRAG_ATTRIB_TEX6:
-            case FRAG_ATTRIB_TEX7:
-            case FRAG_ATTRIB_VAR0:
-            default:
-               /* Actually, let's try and zero-base this just for
-                * readability of the generated TGSI.
-                */
-               assert(attr >= FRAG_ATTRIB_TEX0);
-               input_semantic_index[slot] = (attr - FRAG_ATTRIB_TEX0);
-               input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-               if (attr == FRAG_ATTRIB_PNTC)
-                  interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
-               else
-                  interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
-               break;
-            }
-         }
-         else {
-            inputMapping[attr] = -1;
-         }
-      }
-
-      /*
-       * Semantics and mapping for outputs
-       */
-      {
-         uint numColors = 0;
-         GLbitfield64 outputsWritten = stfp->Base.Base.OutputsWritten;
-
-         /* if z is written, emit that first */
-         if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
-            fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_POSITION;
-            fs_output_semantic_index[fs_num_outputs] = 0;
-            outputMapping[FRAG_RESULT_DEPTH] = fs_num_outputs;
-            fs_num_outputs++;
-            outputsWritten &= ~(1 << FRAG_RESULT_DEPTH);
-         }
-
-         if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) {
-            fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_STENCIL;
-            fs_output_semantic_index[fs_num_outputs] = 0;
-            outputMapping[FRAG_RESULT_STENCIL] = fs_num_outputs;
-            fs_num_outputs++;
-            outputsWritten &= ~(1 << FRAG_RESULT_STENCIL);
-         }
-
-         /* handle remaning outputs (color) */
-         for (attr = 0; attr < FRAG_RESULT_MAX; attr++) {
-            if (outputsWritten & BITFIELD64_BIT(attr)) {
-               switch (attr) {
-               case FRAG_RESULT_DEPTH:
-               case FRAG_RESULT_STENCIL:
-                  /* handled above */
-                  assert(0);
-                  break;
-               case FRAG_RESULT_COLOR:
-                  write_all = GL_TRUE; /* fallthrough */
-               default:
-                  assert(attr == FRAG_RESULT_COLOR ||
-                         (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX));
-                  fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR;
-                  fs_output_semantic_index[fs_num_outputs] = numColors;
-                  outputMapping[attr] = fs_num_outputs;
-                  numColors++;
-                  break;
-               }
-
-               fs_num_outputs++;
-            }
-         }
-      }
-
-      ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
-      if (ureg == NULL)
-         return NULL;
-
-      if (ST_DEBUG & DEBUG_MESA) {
-         _mesa_print_program(&stfp->Base.Base);
-         _mesa_print_program_parameters(st->ctx, &stfp->Base.Base);
-         debug_printf("\n");
-      }
-      if (write_all == GL_TRUE)
-         ureg_property_fs_color0_writes_all_cbufs(ureg, 1);
-
-      error = st_translate_mesa_program(st->ctx,
-                                        TGSI_PROCESSOR_FRAGMENT,
-                                        ureg,
-                                        &stfp->Base.Base,
-                                        /* inputs */
-                                        fs_num_inputs,
-                                        inputMapping,
-                                        input_semantic_name,
-                                        input_semantic_index,
-                                        interpMode,
-                                        /* outputs */
-                                        fs_num_outputs,
-                                        outputMapping,
-                                        fs_output_semantic_name,
-                                        fs_output_semantic_index, FALSE );
-
-      stfp->tgsi.tokens = ureg_get_tokens( ureg, NULL );
-      ureg_destroy( ureg );
-   }
-
-   /* fill in variant */
-   variant->driver_shader = pipe->create_fs_state(pipe, &stfp->tgsi);
-   variant->key = *key;
-
-   if (ST_DEBUG & DEBUG_TGSI) {
-      tgsi_dump( stfp->tgsi.tokens, 0/*TGSI_DUMP_VERBOSE*/ );
-      debug_printf("\n");
-   }
-
-   if (deleteFP) {
-      /* Free the temporary program made above */
-      struct gl_fragment_program *fp = &stfp->Base;
-      _mesa_reference_fragprog(st->ctx, &fp, NULL);
-   }
-
-   return variant;
-}
-
-
-/**
- * Translate fragment program if needed.
- */
-struct st_fp_variant *
-st_get_fp_variant(struct st_context *st,
-                  struct st_fragment_program *stfp,
-                  const struct st_fp_variant_key *key)
-{
-   struct st_fp_variant *fpv;
-
-   /* Search for existing variant */
-   for (fpv = stfp->variants; fpv; fpv = fpv->next) {
-      if (memcmp(&fpv->key, key, sizeof(*key)) == 0) {
-         break;
-      }
-   }
-
-   if (!fpv) {
-      /* create new */
-      fpv = st_translate_fragment_program(st, stfp, key);
-      if (fpv) {
-         /* insert into list */
-         fpv->next = stfp->variants;
-         stfp->variants = fpv;
-      }
-   }
-
-   return fpv;
-}
-
-
-/**
- * Translate a geometry program to create a new variant.
- */
-static struct st_gp_variant *
-st_translate_geometry_program(struct st_context *st,
-                              struct st_geometry_program *stgp,
-                              const struct st_gp_variant_key *key)
-{
-   GLuint inputMapping[GEOM_ATTRIB_MAX];
-   GLuint outputMapping[GEOM_RESULT_MAX];
-   struct pipe_context *pipe = st->pipe;
-   enum pipe_error error;
-   GLuint attr;
-   const GLbitfield inputsRead = stgp->Base.Base.InputsRead;
-   GLuint vslot = 0;
-   GLuint num_generic = 0;
-
-   uint gs_num_inputs = 0;
-   uint gs_builtin_inputs = 0;
-   uint gs_array_offset = 0;
-
-   ubyte gs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS];
-   ubyte gs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
-   uint gs_num_outputs = 0;
-
-   GLint i;
-   GLuint maxSlot = 0;
-   struct ureg_program *ureg;
-
-   struct st_gp_variant *gpv;
-
-   gpv = CALLOC_STRUCT(st_gp_variant);
-   if (!gpv)
-      return NULL;
-
-   _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_OUTPUT);
-   _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_VARYING);
-
-   ureg = ureg_create( TGSI_PROCESSOR_GEOMETRY );
-   if (ureg == NULL) {
-      FREE(gpv);
-      return NULL;
-   }
-
-   /* which vertex output goes to the first geometry input */
-   vslot = 0;
-
-   memset(inputMapping, 0, sizeof(inputMapping));
-   memset(outputMapping, 0, sizeof(outputMapping));
-
-   /*
-    * Convert Mesa program inputs to TGSI input register semantics.
-    */
-   for (attr = 0; attr < GEOM_ATTRIB_MAX; attr++) {
-      if (inputsRead & (1 << attr)) {
-         const GLuint slot = gs_num_inputs;
-
-         gs_num_inputs++;
-
-         inputMapping[attr] = slot;
-
-         stgp->input_map[slot + gs_array_offset] = vslot - gs_builtin_inputs;
-         stgp->input_to_index[attr] = vslot;
-         stgp->index_to_input[vslot] = attr;
-         ++vslot;
-
-         if (attr != GEOM_ATTRIB_PRIMITIVE_ID) {
-            gs_array_offset += 2;
-         } else
-            ++gs_builtin_inputs;
-
-#if 0
-         debug_printf("input map at %d = %d\n",
-                      slot + gs_array_offset, stgp->input_map[slot + gs_array_offset]);
-#endif
-
-         switch (attr) {
-         case GEOM_ATTRIB_PRIMITIVE_ID:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_PRIMID;
-            stgp->input_semantic_index[slot] = 0;
-            break;
-         case GEOM_ATTRIB_POSITION:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-            stgp->input_semantic_index[slot] = 0;
-            break;
-         case GEOM_ATTRIB_COLOR0:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            stgp->input_semantic_index[slot] = 0;
-            break;
-         case GEOM_ATTRIB_COLOR1:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            stgp->input_semantic_index[slot] = 1;
-            break;
-         case GEOM_ATTRIB_FOG_FRAG_COORD:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-            stgp->input_semantic_index[slot] = 0;
-            break;
-         case GEOM_ATTRIB_TEX_COORD:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stgp->input_semantic_index[slot] = num_generic++;
-            break;
-         case GEOM_ATTRIB_VAR0:
-            /* fall-through */
-         default:
-            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            stgp->input_semantic_index[slot] = num_generic++;
-         }
-      }
-   }
-
-   /* initialize output semantics to defaults */
-   for (i = 0; i < PIPE_MAX_SHADER_OUTPUTS; i++) {
-      gs_output_semantic_name[i] = TGSI_SEMANTIC_GENERIC;
-      gs_output_semantic_index[i] = 0;
-   }
-
-   num_generic = 0;
-   /*
-    * Determine number of outputs, the (default) output register
-    * mapping and the semantic information for each output.
-    */
-   for (attr = 0; attr < GEOM_RESULT_MAX; attr++) {
-      if (stgp->Base.Base.OutputsWritten & (1 << attr)) {
-         GLuint slot;
-
-         slot = gs_num_outputs;
-         gs_num_outputs++;
-         outputMapping[attr] = slot;
-
-         switch (attr) {
-         case GEOM_RESULT_POS:
-            assert(slot == 0);
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
-            gs_output_semantic_index[slot] = 0;
-            break;
-         case GEOM_RESULT_COL0:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            gs_output_semantic_index[slot] = 0;
-            break;
-         case GEOM_RESULT_COL1:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
-            gs_output_semantic_index[slot] = 1;
-            break;
-         case GEOM_RESULT_SCOL0:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            gs_output_semantic_index[slot] = 0;
-            break;
-         case GEOM_RESULT_SCOL1:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
-            gs_output_semantic_index[slot] = 1;
-            break;
-         case GEOM_RESULT_FOGC:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
-            gs_output_semantic_index[slot] = 0;
-            break;
-         case GEOM_RESULT_PSIZ:
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
-            gs_output_semantic_index[slot] = 0;
-            break;
-         case GEOM_RESULT_TEX0:
-         case GEOM_RESULT_TEX1:
-         case GEOM_RESULT_TEX2:
-         case GEOM_RESULT_TEX3:
-         case GEOM_RESULT_TEX4:
-         case GEOM_RESULT_TEX5:
-         case GEOM_RESULT_TEX6:
-         case GEOM_RESULT_TEX7:
-            /* fall-through */
-         case GEOM_RESULT_VAR0:
-            /* fall-through */
-         default:
-            assert(slot < Elements(gs_output_semantic_name));
-            /* use default semantic info */
-            gs_output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
-            gs_output_semantic_index[slot] = num_generic++;
-         }
-      }
-   }
-
-   assert(gs_output_semantic_name[0] == TGSI_SEMANTIC_POSITION);
-
-   /* find max output slot referenced to compute gs_num_outputs */
-   for (attr = 0; attr < GEOM_RESULT_MAX; attr++) {
-      if (outputMapping[attr] != ~0 && outputMapping[attr] > maxSlot)
-         maxSlot = outputMapping[attr];
-   }
-   gs_num_outputs = maxSlot + 1;
-
-#if 0 /* debug */
-   {
-      GLuint i;
-      printf("outputMapping? %d\n", outputMapping ? 1 : 0);
-      if (outputMapping) {
-         printf("attr -> slot\n");
-         for (i = 0; i < 16;  i++) {
-            printf(" %2d       %3d\n", i, outputMapping[i]);
-         }
-      }
-      printf("slot    sem_name  sem_index\n");
-      for (i = 0; i < gs_num_outputs; i++) {
-         printf(" %2d         %d         %d\n",
-                i,
-                gs_output_semantic_name[i],
-                gs_output_semantic_index[i]);
-      }
-   }
-#endif
-
-   /* free old shader state, if any */
-   if (stgp->tgsi.tokens) {
-      st_free_tokens(stgp->tgsi.tokens);
-      stgp->tgsi.tokens = NULL;
-   }
-
-   ureg_property_gs_input_prim(ureg, stgp->Base.InputType);
-   ureg_property_gs_output_prim(ureg, stgp->Base.OutputType);
-   ureg_property_gs_max_vertices(ureg, stgp->Base.VerticesOut);
-
-   error = st_translate_mesa_program(st->ctx,
-                                     TGSI_PROCESSOR_GEOMETRY,
-                                     ureg,
-                                     &stgp->Base.Base,
-                                     /* inputs */
-                                     gs_num_inputs,
-                                     inputMapping,
-                                     stgp->input_semantic_name,
-                                     stgp->input_semantic_index,
-                                     NULL,
-                                     /* outputs */
-                                     gs_num_outputs,
-                                     outputMapping,
-                                     gs_output_semantic_name,
-                                     gs_output_semantic_index,
-                                     FALSE);
-
-   stgp->num_inputs = gs_num_inputs;
-   stgp->tgsi.tokens = ureg_get_tokens( ureg, NULL );
-   ureg_destroy( ureg );
-
-   /* fill in new variant */
-   gpv->driver_shader = pipe->create_gs_state(pipe, &stgp->tgsi);
-   gpv->key = *key;
-
-   if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) {
-      _mesa_print_program(&stgp->Base.Base);
-      debug_printf("\n");
-   }
-
-   if (ST_DEBUG & DEBUG_TGSI) {
-      tgsi_dump(stgp->tgsi.tokens, 0);
-      debug_printf("\n");
-   }
-
-   return gpv;
-}
-
-
-/**
- * Get/create geometry program variant.
- */
-struct st_gp_variant *
-st_get_gp_variant(struct st_context *st,
-                  struct st_geometry_program *stgp,
-                  const struct st_gp_variant_key *key)
-{
-   struct st_gp_variant *gpv;
-
-   /* Search for existing variant */
-   for (gpv = stgp->variants; gpv; gpv = gpv->next) {
-      if (memcmp(&gpv->key, key, sizeof(*key)) == 0) {
-         break;
-      }
-   }
-
-   if (!gpv) {
-      /* create new */
-      gpv = st_translate_geometry_program(st, stgp, key);
-      if (gpv) {
-         /* insert into list */
-         gpv->next = stgp->variants;
-         stgp->variants = gpv;
-      }
-   }
-
-   return gpv;
-}
-
-
-
-
-/**
- * Debug- print current shader text
- */
-void
-st_print_shaders(struct gl_context *ctx)
-{
-   struct gl_shader_program *shProg[3] = {
-      ctx->Shader.CurrentVertexProgram,
-      ctx->Shader.CurrentGeometryProgram,
-      ctx->Shader.CurrentFragmentProgram,
-   };
-   unsigned j;
-
-   for (j = 0; j < 3; j++) {
-      unsigned i;
-
-      if (shProg[j] == NULL)
-	 continue;
-
-      for (i = 0; i < shProg[j]->NumShaders; i++) {
-	 struct gl_shader *sh;
-
-	 switch (shProg[j]->Shaders[i]->Type) {
-	 case GL_VERTEX_SHADER:
-	    sh = (i != 0) ? NULL : shProg[j]->Shaders[i];
-	    break;
-	 case GL_GEOMETRY_SHADER_ARB:
-	    sh = (i != 1) ? NULL : shProg[j]->Shaders[i];
-	    break;
-	 case GL_FRAGMENT_SHADER:
-	    sh = (i != 2) ? NULL : shProg[j]->Shaders[i];
-	    break;
-	 default:
-	    assert(0);
-	    sh = NULL;
-	    break;
-	 }
-
-	 if (sh != NULL) {
-	    printf("GLSL shader %u of %u:\n", i, shProg[j]->NumShaders);
-	    printf("%s\n", sh->Source);
-	 }
-      }
-   }
-}
-
-
-/**
- * Vert/Geom/Frag programs have per-context variants.  Free all the
- * variants attached to the given program which match the given context.
- */
-static void
-destroy_program_variants(struct st_context *st, struct gl_program *program)
-{
-   if (!program)
-      return;
-
-   switch (program->Target) {
-   case GL_VERTEX_PROGRAM_ARB:
-      {
-         struct st_vertex_program *stvp = (struct st_vertex_program *) program;
-         struct st_vp_variant *vpv, **prevPtr = &stvp->variants;
-
-         for (vpv = stvp->variants; vpv; ) {
-            struct st_vp_variant *next = vpv->next;
-            if (vpv->key.st == st) {
-               /* unlink from list */
-               *prevPtr = next;
-               /* destroy this variant */
-               delete_vp_variant(st, vpv);
-            }
-            else {
-               prevPtr = &vpv->next;
-            }
-            vpv = next;
-         }
-      }
-      break;
-   case GL_FRAGMENT_PROGRAM_ARB:
-      {
-         struct st_fragment_program *stfp =
-            (struct st_fragment_program *) program;
-         struct st_fp_variant *fpv, **prevPtr = &stfp->variants;
-
-         for (fpv = stfp->variants; fpv; ) {
-            struct st_fp_variant *next = fpv->next;
-            if (fpv->key.st == st) {
-               /* unlink from list */
-               *prevPtr = next;
-               /* destroy this variant */
-               delete_fp_variant(st, fpv);
-            }
-            else {
-               prevPtr = &fpv->next;
-            }
-            fpv = next;
-         }
-      }
-      break;
-   case MESA_GEOMETRY_PROGRAM:
-      {
-         struct st_geometry_program *stgp =
-            (struct st_geometry_program *) program;
-         struct st_gp_variant *gpv, **prevPtr = &stgp->variants;
-
-         for (gpv = stgp->variants; gpv; ) {
-            struct st_gp_variant *next = gpv->next;
-            if (gpv->key.st == st) {
-               /* unlink from list */
-               *prevPtr = next;
-               /* destroy this variant */
-               delete_gp_variant(st, gpv);
-            }
-            else {
-               prevPtr = &gpv->next;
-            }
-            gpv = next;
-         }
-      }
-      break;
-   default:
-      _mesa_problem(NULL, "Unexpected program target in "
-                    "destroy_program_variants_cb()");
-   }
-}
-
-
-/**
- * Callback for _mesa_HashWalk.  Free all the shader's program variants
- * which match the given context.
- */
-static void
-destroy_shader_program_variants_cb(GLuint key, void *data, void *userData)
-{
-   struct st_context *st = (struct st_context *) userData;
-   struct gl_shader *shader = (struct gl_shader *) data;
-
-   switch (shader->Type) {
-   case GL_SHADER_PROGRAM_MESA:
-      {
-         struct gl_shader_program *shProg = (struct gl_shader_program *) data;
-         GLuint i;
-
-         for (i = 0; i < shProg->NumShaders; i++) {
-            destroy_program_variants(st, shProg->Shaders[i]->Program);
-         }
-
-         destroy_program_variants(st, (struct gl_program *)
-                                  shProg->VertexProgram);
-         destroy_program_variants(st, (struct gl_program *)
-                                  shProg->FragmentProgram);
-         destroy_program_variants(st, (struct gl_program *)
-                                  shProg->GeometryProgram);
-      }
-      break;
-   case GL_VERTEX_SHADER:
-   case GL_FRAGMENT_SHADER:
-   case GL_GEOMETRY_SHADER:
-      {
-         destroy_program_variants(st, shader->Program);
-      }
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-/**
- * Callback for _mesa_HashWalk.  Free all the program variants which match
- * the given context.
- */
-static void
-destroy_program_variants_cb(GLuint key, void *data, void *userData)
-{
-   struct st_context *st = (struct st_context *) userData;
-   struct gl_program *program = (struct gl_program *) data;
-   destroy_program_variants(st, program);
-}
-
-
-/**
- * Walk over all shaders and programs to delete any variants which
- * belong to the given context.
- * This is called during context tear-down.
- */
-void
-st_destroy_program_variants(struct st_context *st)
-{
-   /* ARB vert/frag program */
-   _mesa_HashWalk(st->ctx->Shared->Programs,
-                  destroy_program_variants_cb, st);
-
-   /* GLSL vert/frag/geom shaders */
-   _mesa_HashWalk(st->ctx->Shared->ShaderObjects,
-                  destroy_shader_program_variants_cb, st);
-}
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+ /*
+  * Authors:
+  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Brian Paul
+  */
+
+
+#include "main/imports.h"
+#include "main/hash.h"
+#include "main/mfeatures.h"
+#include "main/mtypes.h"
+#include "program/prog_parameter.h"
+#include "program/prog_print.h"
+#include "program/programopt.h"
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+#include "draw/draw_context.h"
+#include "tgsi/tgsi_dump.h"
+#include "tgsi/tgsi_ureg.h"
+
+#include "st_debug.h"
+#include "st_cb_bitmap.h"
+#include "st_cb_drawpixels.h"
+#include "st_context.h"
+#include "st_program.h"
+#include "st_mesa_to_tgsi.h"
+#include "cso_cache/cso_context.h"
+
+
+
+/**
+ * Delete a vertex program variant.  Note the caller must unlink
+ * the variant from the linked list.
+ */
+static void
+delete_vp_variant(struct st_context *st, struct st_vp_variant *vpv)
+{
+   if (vpv->driver_shader) 
+      cso_delete_vertex_shader(st->cso_context, vpv->driver_shader);
+      
+#if FEATURE_feedback || FEATURE_rastpos
+   if (vpv->draw_shader)
+      draw_delete_vertex_shader( st->draw, vpv->draw_shader );
+#endif
+      
+   if (vpv->tgsi.tokens)
+      st_free_tokens(vpv->tgsi.tokens);
+      
+   FREE( vpv );
+}
+
+
+
+/**
+ * Clean out any old compilations:
+ */
+void
+st_release_vp_variants( struct st_context *st,
+                        struct st_vertex_program *stvp )
+{
+   struct st_vp_variant *vpv;
+
+   for (vpv = stvp->variants; vpv; ) {
+      struct st_vp_variant *next = vpv->next;
+      delete_vp_variant(st, vpv);
+      vpv = next;
+   }
+
+   stvp->variants = NULL;
+}
+
+
+
+/**
+ * Delete a fragment program variant.  Note the caller must unlink
+ * the variant from the linked list.
+ */
+static void
+delete_fp_variant(struct st_context *st, struct st_fp_variant *fpv)
+{
+   if (fpv->driver_shader) 
+      cso_delete_fragment_shader(st->cso_context, fpv->driver_shader);
+   if (fpv->parameters)
+      _mesa_free_parameter_list(fpv->parameters);
+      
+   FREE(fpv);
+}
+
+
+/**
+ * Free all variants of a fragment program.
+ */
+void
+st_release_fp_variants(struct st_context *st, struct st_fragment_program *stfp)
+{
+   struct st_fp_variant *fpv;
+
+   for (fpv = stfp->variants; fpv; ) {
+      struct st_fp_variant *next = fpv->next;
+      delete_fp_variant(st, fpv);
+      fpv = next;
+   }
+
+   stfp->variants = NULL;
+}
+
+
+/**
+ * Delete a geometry program variant.  Note the caller must unlink
+ * the variant from the linked list.
+ */
+static void
+delete_gp_variant(struct st_context *st, struct st_gp_variant *gpv)
+{
+   if (gpv->driver_shader) 
+      cso_delete_geometry_shader(st->cso_context, gpv->driver_shader);
+      
+   FREE(gpv);
+}
+
+
+/**
+ * Free all variants of a geometry program.
+ */
+void
+st_release_gp_variants(struct st_context *st, struct st_geometry_program *stgp)
+{
+   struct st_gp_variant *gpv;
+
+   for (gpv = stgp->variants; gpv; ) {
+      struct st_gp_variant *next = gpv->next;
+      delete_gp_variant(st, gpv);
+      gpv = next;
+   }
+
+   stgp->variants = NULL;
+}
+
+
+
+
+/**
+ * Translate a Mesa vertex shader into a TGSI shader.
+ * \param outputMapping  to map vertex program output registers (VERT_RESULT_x)
+ *       to TGSI output slots
+ * \param tokensOut  destination for TGSI tokens
+ * \return  pointer to cached pipe_shader object.
+ */
+static void
+st_prepare_vertex_program(struct st_context *st,
+                            struct st_vertex_program *stvp)
+{
+   GLuint attr;
+
+   stvp->num_inputs = 0;
+   stvp->num_outputs = 0;
+
+   if (stvp->Base.IsPositionInvariant)
+      _mesa_insert_mvp_code(st->ctx, &stvp->Base);
+
+   assert(stvp->Base.Base.NumInstructions > 1);
+
+   /*
+    * Determine number of inputs, the mappings between VERT_ATTRIB_x
+    * and TGSI generic input indexes, plus input attrib semantic info.
+    */
+   for (attr = 0; attr < VERT_ATTRIB_MAX; attr++) {
+      if (stvp->Base.Base.InputsRead & (1 << attr)) {
+         stvp->input_to_index[attr] = stvp->num_inputs;
+         stvp->index_to_input[stvp->num_inputs] = attr;
+         stvp->num_inputs++;
+      }
+   }
+   /* bit of a hack, presetup potentially unused edgeflag input */
+   stvp->input_to_index[VERT_ATTRIB_EDGEFLAG] = stvp->num_inputs;
+   stvp->index_to_input[stvp->num_inputs] = VERT_ATTRIB_EDGEFLAG;
+
+   /* Compute mapping of vertex program outputs to slots.
+    */
+   for (attr = 0; attr < VERT_RESULT_MAX; attr++) {
+      if ((stvp->Base.Base.OutputsWritten & BITFIELD64_BIT(attr)) == 0) {
+         stvp->result_to_output[attr] = ~0;
+      }
+      else {
+         unsigned slot = stvp->num_outputs++;
+
+         stvp->result_to_output[attr] = slot;
+
+         switch (attr) {
+         case VERT_RESULT_HPOS:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+            stvp->output_semantic_index[slot] = 0;
+            break;
+         case VERT_RESULT_COL0:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stvp->output_semantic_index[slot] = 0;
+            break;
+         case VERT_RESULT_COL1:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stvp->output_semantic_index[slot] = 1;
+            break;
+         case VERT_RESULT_BFC0:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stvp->output_semantic_index[slot] = 0;
+            break;
+         case VERT_RESULT_BFC1:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            stvp->output_semantic_index[slot] = 1;
+            break;
+         case VERT_RESULT_FOGC:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            stvp->output_semantic_index[slot] = 0;
+            break;
+         case VERT_RESULT_PSIZ:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
+            stvp->output_semantic_index[slot] = 0;
+            break;
+         case VERT_RESULT_EDGE:
+            assert(0);
+            break;
+
+         case VERT_RESULT_TEX0:
+         case VERT_RESULT_TEX1:
+         case VERT_RESULT_TEX2:
+         case VERT_RESULT_TEX3:
+         case VERT_RESULT_TEX4:
+         case VERT_RESULT_TEX5:
+         case VERT_RESULT_TEX6:
+         case VERT_RESULT_TEX7:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stvp->output_semantic_index[slot] = attr - VERT_RESULT_TEX0;
+            break;
+
+         case VERT_RESULT_VAR0:
+         default:
+            assert(attr < VERT_RESULT_MAX);
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stvp->output_semantic_index[slot] = (FRAG_ATTRIB_VAR0 - 
+                                                FRAG_ATTRIB_TEX0 +
+                                                attr - 
+                                                VERT_RESULT_VAR0);
+            break;
+         }
+      }
+   }
+   /* similar hack to above, presetup potentially unused edgeflag output */
+   stvp->result_to_output[VERT_RESULT_EDGE] = stvp->num_outputs;
+   stvp->output_semantic_name[stvp->num_outputs] = TGSI_SEMANTIC_EDGEFLAG;
+   stvp->output_semantic_index[stvp->num_outputs] = 0;
+}
+
+
+/**
+ * Translate a vertex program to create a new variant.
+ */
+static struct st_vp_variant *
+st_translate_vertex_program(struct st_context *st,
+                            struct st_vertex_program *stvp,
+                            const struct st_vp_variant_key *key)
+{
+   struct st_vp_variant *vpv = CALLOC_STRUCT(st_vp_variant);
+   struct pipe_context *pipe = st->pipe;
+   struct ureg_program *ureg;
+   enum pipe_error error;
+   unsigned num_outputs;
+
+   st_prepare_vertex_program( st, stvp );
+
+   _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_OUTPUT);
+   _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_VARYING);
+
+   ureg = ureg_create( TGSI_PROCESSOR_VERTEX );
+   if (ureg == NULL) {
+      FREE(vpv);
+      return NULL;
+   }
+
+   vpv->key = *key;
+
+   vpv->num_inputs = stvp->num_inputs;
+   num_outputs = stvp->num_outputs;
+   if (key->passthrough_edgeflags) {
+      vpv->num_inputs++;
+      num_outputs++;
+   }
+
+   if (ST_DEBUG & DEBUG_MESA) {
+      _mesa_print_program(&stvp->Base.Base);
+      _mesa_print_program_parameters(st->ctx, &stvp->Base.Base);
+      debug_printf("\n");
+   }
+
+   error = st_translate_mesa_program(st->ctx,
+                                     TGSI_PROCESSOR_VERTEX,
+                                     ureg,
+                                     &stvp->Base.Base,
+                                     /* inputs */
+                                     vpv->num_inputs,
+                                     stvp->input_to_index,
+                                     NULL, /* input semantic name */
+                                     NULL, /* input semantic index */
+                                     NULL,
+                                     /* outputs */
+                                     num_outputs,
+                                     stvp->result_to_output,
+                                     stvp->output_semantic_name,
+                                     stvp->output_semantic_index,
+                                     key->passthrough_edgeflags );
+
+   if (error)
+      goto fail;
+
+   vpv->tgsi.tokens = ureg_get_tokens( ureg, NULL );
+   if (!vpv->tgsi.tokens)
+      goto fail;
+
+   ureg_destroy( ureg );
+
+   vpv->driver_shader = pipe->create_vs_state(pipe, &vpv->tgsi);
+
+   if (ST_DEBUG & DEBUG_TGSI) {
+      tgsi_dump( vpv->tgsi.tokens, 0 );
+      debug_printf("\n");
+   }
+
+   return vpv;
+
+fail:
+   debug_printf("%s: failed to translate Mesa program:\n", __FUNCTION__);
+   _mesa_print_program(&stvp->Base.Base);
+   debug_assert(0);
+
+   ureg_destroy( ureg );
+   return NULL;
+}
+
+
+/**
+ * Find/create a vertex program variant.
+ */
+struct st_vp_variant *
+st_get_vp_variant(struct st_context *st,
+                  struct st_vertex_program *stvp,
+                  const struct st_vp_variant_key *key)
+{
+   struct st_vp_variant *vpv;
+
+   /* Search for existing variant */
+   for (vpv = stvp->variants; vpv; vpv = vpv->next) {
+      if (memcmp(&vpv->key, key, sizeof(*key)) == 0) {
+         break;
+      }
+   }
+
+   if (!vpv) {
+      /* create now */
+      vpv = st_translate_vertex_program(st, stvp, key);
+      if (vpv) {
+         /* insert into list */
+         vpv->next = stvp->variants;
+         stvp->variants = vpv;
+      }
+   }
+
+   return vpv;
+}
+
+
+/**
+ * Translate a Mesa fragment shader into a TGSI shader using extra info in
+ * the key.
+ * \return  new fragment program variant
+ */
+static struct st_fp_variant *
+st_translate_fragment_program(struct st_context *st,
+                              struct st_fragment_program *stfp,
+                              const struct st_fp_variant_key *key)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct st_fp_variant *variant = CALLOC_STRUCT(st_fp_variant);
+   GLboolean deleteFP = GL_FALSE;
+
+   if (!variant)
+      return NULL;
+
+   assert(!(key->bitmap && key->drawpixels));
+
+#if FEATURE_drawpix
+   if (key->bitmap) {
+      /* glBitmap drawing */
+      struct gl_fragment_program *fp; /* we free this temp program below */
+
+      st_make_bitmap_fragment_program(st, &stfp->Base,
+                                      &fp, &variant->bitmap_sampler);
+
+      variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters);
+      stfp = st_fragment_program(fp);
+      deleteFP = GL_TRUE;
+   }
+   else if (key->drawpixels) {
+      /* glDrawPixels drawing */
+      struct gl_fragment_program *fp; /* we free this temp program below */
+
+      if (key->drawpixels_z || key->drawpixels_stencil) {
+         fp = st_make_drawpix_z_stencil_program(st, key->drawpixels_z,
+                                                key->drawpixels_stencil);
+      }
+      else {
+         /* RGBA */
+         st_make_drawpix_fragment_program(st, &stfp->Base, &fp);
+         variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters);
+         deleteFP = GL_TRUE;
+      }
+      stfp = st_fragment_program(fp);
+   }
+#endif
+
+   if (!stfp->tgsi.tokens) {
+      /* need to translate Mesa instructions to TGSI now */
+      GLuint outputMapping[FRAG_RESULT_MAX];
+      GLuint inputMapping[FRAG_ATTRIB_MAX];
+      GLuint interpMode[PIPE_MAX_SHADER_INPUTS];  /* XXX size? */
+      GLuint attr;
+      enum pipe_error error;
+      const GLbitfield inputsRead = stfp->Base.Base.InputsRead;
+      struct ureg_program *ureg;
+      GLboolean write_all = GL_FALSE;
+
+      ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS];
+      ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS];
+      uint fs_num_inputs = 0;
+
+      ubyte fs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS];
+      ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
+      uint fs_num_outputs = 0;
+
+
+      _mesa_remove_output_reads(&stfp->Base.Base, PROGRAM_OUTPUT);
+
+      /*
+       * Convert Mesa program inputs to TGSI input register semantics.
+       */
+      for (attr = 0; attr < FRAG_ATTRIB_MAX; attr++) {
+         if (inputsRead & (1 << attr)) {
+            const GLuint slot = fs_num_inputs++;
+
+            inputMapping[attr] = slot;
+
+            switch (attr) {
+            case FRAG_ATTRIB_WPOS:
+               input_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+               input_semantic_index[slot] = 0;
+               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
+               break;
+            case FRAG_ATTRIB_COL0:
+               input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+               input_semantic_index[slot] = 0;
+               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
+               break;
+            case FRAG_ATTRIB_COL1:
+               input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+               input_semantic_index[slot] = 1;
+               interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
+               break;
+            case FRAG_ATTRIB_FOGC:
+               input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+               input_semantic_index[slot] = 0;
+               interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
+               break;
+            case FRAG_ATTRIB_FACE:
+               input_semantic_name[slot] = TGSI_SEMANTIC_FACE;
+               input_semantic_index[slot] = 0;
+               interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
+               break;
+               /* In most cases, there is nothing special about these
+                * inputs, so adopt a convention to use the generic
+                * semantic name and the mesa FRAG_ATTRIB_ number as the
+                * index. 
+                * 
+                * All that is required is that the vertex shader labels
+                * its own outputs similarly, and that the vertex shader
+                * generates at least every output required by the
+                * fragment shader plus fixed-function hardware (such as
+                * BFC).
+                * 
+                * There is no requirement that semantic indexes start at
+                * zero or be restricted to a particular range -- nobody
+                * should be building tables based on semantic index.
+                */
+            case FRAG_ATTRIB_PNTC:
+            case FRAG_ATTRIB_TEX0:
+            case FRAG_ATTRIB_TEX1:
+            case FRAG_ATTRIB_TEX2:
+            case FRAG_ATTRIB_TEX3:
+            case FRAG_ATTRIB_TEX4:
+            case FRAG_ATTRIB_TEX5:
+            case FRAG_ATTRIB_TEX6:
+            case FRAG_ATTRIB_TEX7:
+            case FRAG_ATTRIB_VAR0:
+            default:
+               /* Actually, let's try and zero-base this just for
+                * readability of the generated TGSI.
+                */
+               assert(attr >= FRAG_ATTRIB_TEX0);
+               input_semantic_index[slot] = (attr - FRAG_ATTRIB_TEX0);
+               input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+               if (attr == FRAG_ATTRIB_PNTC)
+                  interpMode[slot] = TGSI_INTERPOLATE_LINEAR;
+               else
+                  interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
+               break;
+            }
+         }
+         else {
+            inputMapping[attr] = -1;
+         }
+      }
+
+      /*
+       * Semantics and mapping for outputs
+       */
+      {
+         uint numColors = 0;
+         GLbitfield64 outputsWritten = stfp->Base.Base.OutputsWritten;
+
+         /* if z is written, emit that first */
+         if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
+            fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_POSITION;
+            fs_output_semantic_index[fs_num_outputs] = 0;
+            outputMapping[FRAG_RESULT_DEPTH] = fs_num_outputs;
+            fs_num_outputs++;
+            outputsWritten &= ~(1 << FRAG_RESULT_DEPTH);
+         }
+
+         if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) {
+            fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_STENCIL;
+            fs_output_semantic_index[fs_num_outputs] = 0;
+            outputMapping[FRAG_RESULT_STENCIL] = fs_num_outputs;
+            fs_num_outputs++;
+            outputsWritten &= ~(1 << FRAG_RESULT_STENCIL);
+         }
+
+         /* handle remaning outputs (color) */
+         for (attr = 0; attr < FRAG_RESULT_MAX; attr++) {
+            if (outputsWritten & BITFIELD64_BIT(attr)) {
+               switch (attr) {
+               case FRAG_RESULT_DEPTH:
+               case FRAG_RESULT_STENCIL:
+                  /* handled above */
+                  assert(0);
+                  break;
+               case FRAG_RESULT_COLOR:
+                  write_all = GL_TRUE; /* fallthrough */
+               default:
+                  assert(attr == FRAG_RESULT_COLOR ||
+                         (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX));
+                  fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR;
+                  fs_output_semantic_index[fs_num_outputs] = numColors;
+                  outputMapping[attr] = fs_num_outputs;
+                  numColors++;
+                  break;
+               }
+
+               fs_num_outputs++;
+            }
+         }
+      }
+
+      ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
+      if (ureg == NULL)
+         return NULL;
+
+      if (ST_DEBUG & DEBUG_MESA) {
+         _mesa_print_program(&stfp->Base.Base);
+         _mesa_print_program_parameters(st->ctx, &stfp->Base.Base);
+         debug_printf("\n");
+      }
+      if (write_all == GL_TRUE)
+         ureg_property_fs_color0_writes_all_cbufs(ureg, 1);
+
+      error = st_translate_mesa_program(st->ctx,
+                                        TGSI_PROCESSOR_FRAGMENT,
+                                        ureg,
+                                        &stfp->Base.Base,
+                                        /* inputs */
+                                        fs_num_inputs,
+                                        inputMapping,
+                                        input_semantic_name,
+                                        input_semantic_index,
+                                        interpMode,
+                                        /* outputs */
+                                        fs_num_outputs,
+                                        outputMapping,
+                                        fs_output_semantic_name,
+                                        fs_output_semantic_index, FALSE );
+
+      stfp->tgsi.tokens = ureg_get_tokens( ureg, NULL );
+      ureg_destroy( ureg );
+   }
+
+   /* fill in variant */
+   variant->driver_shader = pipe->create_fs_state(pipe, &stfp->tgsi);
+   variant->key = *key;
+
+   if (ST_DEBUG & DEBUG_TGSI) {
+      tgsi_dump( stfp->tgsi.tokens, 0/*TGSI_DUMP_VERBOSE*/ );
+      debug_printf("\n");
+   }
+
+   if (deleteFP) {
+      /* Free the temporary program made above */
+      struct gl_fragment_program *fp = &stfp->Base;
+      _mesa_reference_fragprog(st->ctx, &fp, NULL);
+   }
+
+   return variant;
+}
+
+
+/**
+ * Translate fragment program if needed.
+ */
+struct st_fp_variant *
+st_get_fp_variant(struct st_context *st,
+                  struct st_fragment_program *stfp,
+                  const struct st_fp_variant_key *key)
+{
+   struct st_fp_variant *fpv;
+
+   /* Search for existing variant */
+   for (fpv = stfp->variants; fpv; fpv = fpv->next) {
+      if (memcmp(&fpv->key, key, sizeof(*key)) == 0) {
+         break;
+      }
+   }
+
+   if (!fpv) {
+      /* create new */
+      fpv = st_translate_fragment_program(st, stfp, key);
+      if (fpv) {
+         /* insert into list */
+         fpv->next = stfp->variants;
+         stfp->variants = fpv;
+      }
+   }
+
+   return fpv;
+}
+
+
+/**
+ * Translate a geometry program to create a new variant.
+ */
+static struct st_gp_variant *
+st_translate_geometry_program(struct st_context *st,
+                              struct st_geometry_program *stgp,
+                              const struct st_gp_variant_key *key)
+{
+   GLuint inputMapping[GEOM_ATTRIB_MAX];
+   GLuint outputMapping[GEOM_RESULT_MAX];
+   struct pipe_context *pipe = st->pipe;
+   enum pipe_error error;
+   GLuint attr;
+   const GLbitfield inputsRead = stgp->Base.Base.InputsRead;
+   GLuint vslot = 0;
+   GLuint num_generic = 0;
+
+   uint gs_num_inputs = 0;
+   uint gs_builtin_inputs = 0;
+   uint gs_array_offset = 0;
+
+   ubyte gs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS];
+   ubyte gs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
+   uint gs_num_outputs = 0;
+
+   GLint i;
+   GLuint maxSlot = 0;
+   struct ureg_program *ureg;
+
+   struct st_gp_variant *gpv;
+
+   gpv = CALLOC_STRUCT(st_gp_variant);
+   if (!gpv)
+      return NULL;
+
+   _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_OUTPUT);
+   _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_VARYING);
+
+   ureg = ureg_create( TGSI_PROCESSOR_GEOMETRY );
+   if (ureg == NULL) {
+      FREE(gpv);
+      return NULL;
+   }
+
+   /* which vertex output goes to the first geometry input */
+   vslot = 0;
+
+   memset(inputMapping, 0, sizeof(inputMapping));
+   memset(outputMapping, 0, sizeof(outputMapping));
+
+   /*
+    * Convert Mesa program inputs to TGSI input register semantics.
+    */
+   for (attr = 0; attr < GEOM_ATTRIB_MAX; attr++) {
+      if (inputsRead & (1 << attr)) {
+         const GLuint slot = gs_num_inputs;
+
+         gs_num_inputs++;
+
+         inputMapping[attr] = slot;
+
+         stgp->input_map[slot + gs_array_offset] = vslot - gs_builtin_inputs;
+         stgp->input_to_index[attr] = vslot;
+         stgp->index_to_input[vslot] = attr;
+         ++vslot;
+
+         if (attr != GEOM_ATTRIB_PRIMITIVE_ID) {
+            gs_array_offset += 2;
+         } else
+            ++gs_builtin_inputs;
+
+#if 0
+         debug_printf("input map at %d = %d\n",
+                      slot + gs_array_offset, stgp->input_map[slot + gs_array_offset]);
+#endif
+
+         switch (attr) {
+         case GEOM_ATTRIB_PRIMITIVE_ID:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_PRIMID;
+            stgp->input_semantic_index[slot] = 0;
+            break;
+         case GEOM_ATTRIB_POSITION:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+            stgp->input_semantic_index[slot] = 0;
+            break;
+         case GEOM_ATTRIB_COLOR0:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stgp->input_semantic_index[slot] = 0;
+            break;
+         case GEOM_ATTRIB_COLOR1:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            stgp->input_semantic_index[slot] = 1;
+            break;
+         case GEOM_ATTRIB_FOG_FRAG_COORD:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            stgp->input_semantic_index[slot] = 0;
+            break;
+         case GEOM_ATTRIB_TEX_COORD:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stgp->input_semantic_index[slot] = num_generic++;
+            break;
+         case GEOM_ATTRIB_VAR0:
+            /* fall-through */
+         default:
+            stgp->input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            stgp->input_semantic_index[slot] = num_generic++;
+         }
+      }
+   }
+
+   /* initialize output semantics to defaults */
+   for (i = 0; i < PIPE_MAX_SHADER_OUTPUTS; i++) {
+      gs_output_semantic_name[i] = TGSI_SEMANTIC_GENERIC;
+      gs_output_semantic_index[i] = 0;
+   }
+
+   num_generic = 0;
+   /*
+    * Determine number of outputs, the (default) output register
+    * mapping and the semantic information for each output.
+    */
+   for (attr = 0; attr < GEOM_RESULT_MAX; attr++) {
+      if (stgp->Base.Base.OutputsWritten & (1 << attr)) {
+         GLuint slot;
+
+         slot = gs_num_outputs;
+         gs_num_outputs++;
+         outputMapping[attr] = slot;
+
+         switch (attr) {
+         case GEOM_RESULT_POS:
+            assert(slot == 0);
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_POSITION;
+            gs_output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_COL0:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            gs_output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_COL1:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_COLOR;
+            gs_output_semantic_index[slot] = 1;
+            break;
+         case GEOM_RESULT_SCOL0:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            gs_output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_SCOL1:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_BCOLOR;
+            gs_output_semantic_index[slot] = 1;
+            break;
+         case GEOM_RESULT_FOGC:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_FOG;
+            gs_output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_PSIZ:
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
+            gs_output_semantic_index[slot] = 0;
+            break;
+         case GEOM_RESULT_TEX0:
+         case GEOM_RESULT_TEX1:
+         case GEOM_RESULT_TEX2:
+         case GEOM_RESULT_TEX3:
+         case GEOM_RESULT_TEX4:
+         case GEOM_RESULT_TEX5:
+         case GEOM_RESULT_TEX6:
+         case GEOM_RESULT_TEX7:
+            /* fall-through */
+         case GEOM_RESULT_VAR0:
+            /* fall-through */
+         default:
+            assert(slot < Elements(gs_output_semantic_name));
+            /* use default semantic info */
+            gs_output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
+            gs_output_semantic_index[slot] = num_generic++;
+         }
+      }
+   }
+
+   assert(gs_output_semantic_name[0] == TGSI_SEMANTIC_POSITION);
+
+   /* find max output slot referenced to compute gs_num_outputs */
+   for (attr = 0; attr < GEOM_RESULT_MAX; attr++) {
+      if (outputMapping[attr] != ~0 && outputMapping[attr] > maxSlot)
+         maxSlot = outputMapping[attr];
+   }
+   gs_num_outputs = maxSlot + 1;
+
+#if 0 /* debug */
+   {
+      GLuint i;
+      printf("outputMapping? %d\n", outputMapping ? 1 : 0);
+      if (outputMapping) {
+         printf("attr -> slot\n");
+         for (i = 0; i < 16;  i++) {
+            printf(" %2d       %3d\n", i, outputMapping[i]);
+         }
+      }
+      printf("slot    sem_name  sem_index\n");
+      for (i = 0; i < gs_num_outputs; i++) {
+         printf(" %2d         %d         %d\n",
+                i,
+                gs_output_semantic_name[i],
+                gs_output_semantic_index[i]);
+      }
+   }
+#endif
+
+   /* free old shader state, if any */
+   if (stgp->tgsi.tokens) {
+      st_free_tokens(stgp->tgsi.tokens);
+      stgp->tgsi.tokens = NULL;
+   }
+
+   ureg_property_gs_input_prim(ureg, stgp->Base.InputType);
+   ureg_property_gs_output_prim(ureg, stgp->Base.OutputType);
+   ureg_property_gs_max_vertices(ureg, stgp->Base.VerticesOut);
+
+   error = st_translate_mesa_program(st->ctx,
+                                     TGSI_PROCESSOR_GEOMETRY,
+                                     ureg,
+                                     &stgp->Base.Base,
+                                     /* inputs */
+                                     gs_num_inputs,
+                                     inputMapping,
+                                     stgp->input_semantic_name,
+                                     stgp->input_semantic_index,
+                                     NULL,
+                                     /* outputs */
+                                     gs_num_outputs,
+                                     outputMapping,
+                                     gs_output_semantic_name,
+                                     gs_output_semantic_index,
+                                     FALSE);
+
+   stgp->num_inputs = gs_num_inputs;
+   stgp->tgsi.tokens = ureg_get_tokens( ureg, NULL );
+   ureg_destroy( ureg );
+
+   /* fill in new variant */
+   gpv->driver_shader = pipe->create_gs_state(pipe, &stgp->tgsi);
+   gpv->key = *key;
+
+   if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) {
+      _mesa_print_program(&stgp->Base.Base);
+      debug_printf("\n");
+   }
+
+   if (ST_DEBUG & DEBUG_TGSI) {
+      tgsi_dump(stgp->tgsi.tokens, 0);
+      debug_printf("\n");
+   }
+
+   return gpv;
+}
+
+
+/**
+ * Get/create geometry program variant.
+ */
+struct st_gp_variant *
+st_get_gp_variant(struct st_context *st,
+                  struct st_geometry_program *stgp,
+                  const struct st_gp_variant_key *key)
+{
+   struct st_gp_variant *gpv;
+
+   /* Search for existing variant */
+   for (gpv = stgp->variants; gpv; gpv = gpv->next) {
+      if (memcmp(&gpv->key, key, sizeof(*key)) == 0) {
+         break;
+      }
+   }
+
+   if (!gpv) {
+      /* create new */
+      gpv = st_translate_geometry_program(st, stgp, key);
+      if (gpv) {
+         /* insert into list */
+         gpv->next = stgp->variants;
+         stgp->variants = gpv;
+      }
+   }
+
+   return gpv;
+}
+
+
+
+
+/**
+ * Debug- print current shader text
+ */
+void
+st_print_shaders(struct gl_context *ctx)
+{
+   struct gl_shader_program *shProg[3] = {
+      ctx->Shader.CurrentVertexProgram,
+      ctx->Shader.CurrentGeometryProgram,
+      ctx->Shader.CurrentFragmentProgram,
+   };
+   unsigned j;
+
+   for (j = 0; j < 3; j++) {
+      unsigned i;
+
+      if (shProg[j] == NULL)
+	 continue;
+
+      for (i = 0; i < shProg[j]->NumShaders; i++) {
+	 struct gl_shader *sh;
+
+	 switch (shProg[j]->Shaders[i]->Type) {
+	 case GL_VERTEX_SHADER:
+	    sh = (i != 0) ? NULL : shProg[j]->Shaders[i];
+	    break;
+	 case GL_GEOMETRY_SHADER_ARB:
+	    sh = (i != 1) ? NULL : shProg[j]->Shaders[i];
+	    break;
+	 case GL_FRAGMENT_SHADER:
+	    sh = (i != 2) ? NULL : shProg[j]->Shaders[i];
+	    break;
+	 default:
+	    assert(0);
+	    sh = NULL;
+	    break;
+	 }
+
+	 if (sh != NULL) {
+	    printf("GLSL shader %u of %u:\n", i, shProg[j]->NumShaders);
+	    printf("%s\n", sh->Source);
+	 }
+      }
+   }
+}
+
+
+/**
+ * Vert/Geom/Frag programs have per-context variants.  Free all the
+ * variants attached to the given program which match the given context.
+ */
+static void
+destroy_program_variants(struct st_context *st, struct gl_program *program)
+{
+   if (!program)
+      return;
+
+   switch (program->Target) {
+   case GL_VERTEX_PROGRAM_ARB:
+      {
+         struct st_vertex_program *stvp = (struct st_vertex_program *) program;
+         struct st_vp_variant *vpv, **prevPtr = &stvp->variants;
+
+         for (vpv = stvp->variants; vpv; ) {
+            struct st_vp_variant *next = vpv->next;
+            if (vpv->key.st == st) {
+               /* unlink from list */
+               *prevPtr = next;
+               /* destroy this variant */
+               delete_vp_variant(st, vpv);
+            }
+            else {
+               prevPtr = &vpv->next;
+            }
+            vpv = next;
+         }
+      }
+      break;
+   case GL_FRAGMENT_PROGRAM_ARB:
+      {
+         struct st_fragment_program *stfp =
+            (struct st_fragment_program *) program;
+         struct st_fp_variant *fpv, **prevPtr = &stfp->variants;
+
+         for (fpv = stfp->variants; fpv; ) {
+            struct st_fp_variant *next = fpv->next;
+            if (fpv->key.st == st) {
+               /* unlink from list */
+               *prevPtr = next;
+               /* destroy this variant */
+               delete_fp_variant(st, fpv);
+            }
+            else {
+               prevPtr = &fpv->next;
+            }
+            fpv = next;
+         }
+      }
+      break;
+   case MESA_GEOMETRY_PROGRAM:
+      {
+         struct st_geometry_program *stgp =
+            (struct st_geometry_program *) program;
+         struct st_gp_variant *gpv, **prevPtr = &stgp->variants;
+
+         for (gpv = stgp->variants; gpv; ) {
+            struct st_gp_variant *next = gpv->next;
+            if (gpv->key.st == st) {
+               /* unlink from list */
+               *prevPtr = next;
+               /* destroy this variant */
+               delete_gp_variant(st, gpv);
+            }
+            else {
+               prevPtr = &gpv->next;
+            }
+            gpv = next;
+         }
+      }
+      break;
+   default:
+      _mesa_problem(NULL, "Unexpected program target in "
+                    "destroy_program_variants_cb()");
+   }
+}
+
+
+/**
+ * Callback for _mesa_HashWalk.  Free all the shader's program variants
+ * which match the given context.
+ */
+static void
+destroy_shader_program_variants_cb(GLuint key, void *data, void *userData)
+{
+   struct st_context *st = (struct st_context *) userData;
+   struct gl_shader *shader = (struct gl_shader *) data;
+
+   switch (shader->Type) {
+   case GL_SHADER_PROGRAM_MESA:
+      {
+         struct gl_shader_program *shProg = (struct gl_shader_program *) data;
+         GLuint i;
+
+         for (i = 0; i < shProg->NumShaders; i++) {
+            destroy_program_variants(st, shProg->Shaders[i]->Program);
+         }
+
+         destroy_program_variants(st, (struct gl_program *)
+                                  shProg->VertexProgram);
+         destroy_program_variants(st, (struct gl_program *)
+                                  shProg->FragmentProgram);
+         destroy_program_variants(st, (struct gl_program *)
+                                  shProg->GeometryProgram);
+      }
+      break;
+   case GL_VERTEX_SHADER:
+   case GL_FRAGMENT_SHADER:
+   case GL_GEOMETRY_SHADER:
+      {
+         destroy_program_variants(st, shader->Program);
+      }
+      break;
+   default:
+      assert(0);
+   }
+}
+
+
+/**
+ * Callback for _mesa_HashWalk.  Free all the program variants which match
+ * the given context.
+ */
+static void
+destroy_program_variants_cb(GLuint key, void *data, void *userData)
+{
+   struct st_context *st = (struct st_context *) userData;
+   struct gl_program *program = (struct gl_program *) data;
+   destroy_program_variants(st, program);
+}
+
+
+/**
+ * Walk over all shaders and programs to delete any variants which
+ * belong to the given context.
+ * This is called during context tear-down.
+ */
+void
+st_destroy_program_variants(struct st_context *st)
+{
+   /* ARB vert/frag program */
+   _mesa_HashWalk(st->ctx->Shared->Programs,
+                  destroy_program_variants_cb, st);
+
+   /* GLSL vert/frag/geom shaders */
+   _mesa_HashWalk(st->ctx->Shared->ShaderObjects,
+                  destroy_shader_program_variants_cb, st);
+}