5 files changed, 3063 insertions, 3060 deletions
diff --git a/mesalib/src/mesa/swrast/s_fragprog.c b/mesalib/src/mesa/swrast/s_fragprog.c
index 2608d5373..b6bfeaed4 100644
--- a/mesalib/src/mesa/swrast/s_fragprog.c
+++ b/mesalib/src/mesa/swrast/s_fragprog.c
@@ -1,277 +1,277 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.0.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "program/prog_instruction.h"
-
-#include "s_context.h"
-#include "s_fragprog.h"
-#include "s_span.h"
-
-
-/**
- * Apply texture object's swizzle (X/Y/Z/W/0/1) to incoming 'texel'
- * and return results in 'colorOut'.
- */
-static INLINE void
-swizzle_texel(const GLfloat texel[4], GLfloat colorOut[4], GLuint swizzle)
-{
-   if (swizzle == SWIZZLE_NOOP) {
-      COPY_4V(colorOut, texel);
-   }
-   else {
-      GLfloat vector[6];
-      vector[SWIZZLE_X] = texel[0];
-      vector[SWIZZLE_Y] = texel[1];
-      vector[SWIZZLE_Z] = texel[2];
-      vector[SWIZZLE_W] = texel[3];
-      vector[SWIZZLE_ZERO] = 0.0F;
-      vector[SWIZZLE_ONE] = 1.0F;
-      colorOut[0] = vector[GET_SWZ(swizzle, 0)];
-      colorOut[1] = vector[GET_SWZ(swizzle, 1)];
-      colorOut[2] = vector[GET_SWZ(swizzle, 2)];
-      colorOut[3] = vector[GET_SWZ(swizzle, 3)];
-   }
-}
-
-
-/**
- * Fetch a texel with given lod.
- * Called via machine->FetchTexelLod()
- */
-static void
-fetch_texel_lod( struct gl_context *ctx, const GLfloat texcoord[4], GLfloat lambda,
-                 GLuint unit, GLfloat color[4] )
-{
-   const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-
-   if (texObj) {
-      SWcontext *swrast = SWRAST_CONTEXT(ctx);
-      GLfloat rgba[4];
-
-      lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
-
-      swrast->TextureSample[unit](ctx, texObj, 1,
-                                  (const GLfloat (*)[4]) texcoord,
-                                  &lambda, &rgba);
-      swizzle_texel(rgba, color, texObj->_Swizzle);
-   }
-   else {
-      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Fetch a texel with the given partial derivatives to compute a level
- * of detail in the mipmap.
- * Called via machine->FetchTexelDeriv()
- * \param lodBias  the lod bias which may be specified by a TXB instruction,
- *                 otherwise zero.
- */
-static void
-fetch_texel_deriv( struct gl_context *ctx, const GLfloat texcoord[4],
-                   const GLfloat texdx[4], const GLfloat texdy[4],
-                   GLfloat lodBias, GLuint unit, GLfloat color[4] )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   const struct gl_texture_object *texObj = texUnit->_Current;
-
-   if (texObj) {
-      const struct gl_texture_image *texImg =
-         texObj->Image[0][texObj->BaseLevel];
-      const GLfloat texW = (GLfloat) texImg->WidthScale;
-      const GLfloat texH = (GLfloat) texImg->HeightScale;
-      GLfloat lambda;
-      GLfloat rgba[4];
-
-      lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
-                                      texdx[1], texdy[1], /* dt/dx, dt/dy */
-                                      texdx[3], texdy[3], /* dq/dx, dq/dy */
-                                      texW, texH,
-                                      texcoord[0], texcoord[1], texcoord[3],
-                                      1.0F / texcoord[3]);
-
-      lambda += lodBias + texUnit->LodBias + texObj->LodBias;
-
-      lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
-
-      swrast->TextureSample[unit](ctx, texObj, 1,
-                                  (const GLfloat (*)[4]) texcoord,
-                                  &lambda, &rgba);
-      swizzle_texel(rgba, color, texObj->_Swizzle);
-   }
-   else {
-      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Initialize the virtual fragment program machine state prior to running
- * fragment program on a fragment.  This involves initializing the input
- * registers, condition codes, etc.
- * \param machine  the virtual machine state to init
- * \param program  the fragment program we're about to run
- * \param span  the span of pixels we'll operate on
- * \param col  which element (column) of the span we'll operate on
- */
-static void
-init_machine(struct gl_context *ctx, struct gl_program_machine *machine,
-             const struct gl_fragment_program *program,
-             const SWspan *span, GLuint col)
-{
-   GLfloat *wpos = span->array->attribs[FRAG_ATTRIB_WPOS][col];
-
-   if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
-      /* Clear temporary registers (undefined for ARB_f_p) */
-      memset(machine->Temporaries, 0, MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
-   }
-
-   /* ARB_fragment_coord_conventions */
-   if (program->OriginUpperLeft)
-      wpos[1] = ctx->DrawBuffer->Height - 1 - wpos[1];
-   if (!program->PixelCenterInteger) {
-      wpos[0] += 0.5F;
-      wpos[1] += 0.5F;
-   }
-
-   /* Setup pointer to input attributes */
-   machine->Attribs = span->array->attribs;
-
-   machine->DerivX = (GLfloat (*)[4]) span->attrStepX;
-   machine->DerivY = (GLfloat (*)[4]) span->attrStepY;
-   machine->NumDeriv = FRAG_ATTRIB_MAX;
-
-   machine->Samplers = program->Base.SamplerUnits;
-
-   /* if running a GLSL program (not ARB_fragment_program) */
-   if (ctx->Shader.CurrentFragmentProgram) {
-      /* Store front/back facing value */
-      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0F - span->facing;
-   }
-
-   machine->CurElement = col;
-
-   /* init condition codes */
-   machine->CondCodes[0] = COND_EQ;
-   machine->CondCodes[1] = COND_EQ;
-   machine->CondCodes[2] = COND_EQ;
-   machine->CondCodes[3] = COND_EQ;
-
-   /* init call stack */
-   machine->StackDepth = 0;
-
-   machine->FetchTexelLod = fetch_texel_lod;
-   machine->FetchTexelDeriv = fetch_texel_deriv;
-}
-
-
-/**
- * Run fragment program on the pixels in span from 'start' to 'end' - 1.
- */
-static void
-run_program(struct gl_context *ctx, SWspan *span, GLuint start, GLuint end)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
-   const GLbitfield64 outputsWritten = program->Base.OutputsWritten;
-   struct gl_program_machine *machine = &swrast->FragProgMachine;
-   GLuint i;
-
-   for (i = start; i < end; i++) {
-      if (span->array->mask[i]) {
-         init_machine(ctx, machine, program, span, i);
-
-         if (_mesa_execute_program(ctx, &program->Base, machine)) {
-
-            /* Store result color */
-	    if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
-               COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
-                       machine->Outputs[FRAG_RESULT_COLOR]);
-            }
-            else {
-               /* Multiple drawbuffers / render targets
-                * Note that colors beyond 0 and 1 will overwrite other
-                * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.
-                */
-               GLuint buf;
-               for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
-                  if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {
-                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
-                             machine->Outputs[FRAG_RESULT_DATA0 + buf]);
-                  }
-               }
-            }
-
-            /* Store result depth/z */
-            if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
-               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
-               if (depth <= 0.0)
-                  span->array->z[i] = 0;
-               else if (depth >= 1.0)
-                  span->array->z[i] = ctx->DrawBuffer->_DepthMax;
-               else
-                  span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
-            }
-         }
-         else {
-            /* killed fragment */
-            span->array->mask[i] = GL_FALSE;
-            span->writeAll = GL_FALSE;
-         }
-      }
-   }
-}
-
-
-/**
- * Execute the current fragment program for all the fragments
- * in the given span.
- */
-void
-_swrast_exec_fragment_program( struct gl_context *ctx, SWspan *span )
-{
-   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
-
-   /* incoming colors should be floats */
-   if (program->Base.InputsRead & FRAG_BIT_COL0) {
-      ASSERT(span->array->ChanType == GL_FLOAT);
-   }
-
-   run_program(ctx, span, 0, span->end);
-
-   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
-      span->interpMask &= ~SPAN_RGBA;
-      span->arrayMask |= SPAN_RGBA;
-   }
-
-   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
-      span->interpMask &= ~SPAN_Z;
-      span->arrayMask |= SPAN_Z;
-   }
-}
-
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.0.3
+ *
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "main/glheader.h"
+#include "main/colormac.h"
+#include "program/prog_instruction.h"
+
+#include "s_context.h"
+#include "s_fragprog.h"
+#include "s_span.h"
+
+
+/**
+ * Apply texture object's swizzle (X/Y/Z/W/0/1) to incoming 'texel'
+ * and return results in 'colorOut'.
+ */
+static INLINE void
+swizzle_texel(const GLfloat texel[4], GLfloat colorOut[4], GLuint swizzle)
+{
+   if (swizzle == SWIZZLE_NOOP) {
+      COPY_4V(colorOut, texel);
+   }
+   else {
+      GLfloat vector[6];
+      vector[SWIZZLE_X] = texel[0];
+      vector[SWIZZLE_Y] = texel[1];
+      vector[SWIZZLE_Z] = texel[2];
+      vector[SWIZZLE_W] = texel[3];
+      vector[SWIZZLE_ZERO] = 0.0F;
+      vector[SWIZZLE_ONE] = 1.0F;
+      colorOut[0] = vector[GET_SWZ(swizzle, 0)];
+      colorOut[1] = vector[GET_SWZ(swizzle, 1)];
+      colorOut[2] = vector[GET_SWZ(swizzle, 2)];
+      colorOut[3] = vector[GET_SWZ(swizzle, 3)];
+   }
+}
+
+
+/**
+ * Fetch a texel with given lod.
+ * Called via machine->FetchTexelLod()
+ */
+static void
+fetch_texel_lod( struct gl_context *ctx, const GLfloat texcoord[4], GLfloat lambda,
+                 GLuint unit, GLfloat color[4] )
+{
+   const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
+
+   if (texObj) {
+      SWcontext *swrast = SWRAST_CONTEXT(ctx);
+      GLfloat rgba[4];
+
+      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);
+
+      swrast->TextureSample[unit](ctx, texObj, 1,
+                                  (const GLfloat (*)[4]) texcoord,
+                                  &lambda, &rgba);
+      swizzle_texel(rgba, color, texObj->_Swizzle);
+   }
+   else {
+      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
+   }
+}
+
+
+/**
+ * Fetch a texel with the given partial derivatives to compute a level
+ * of detail in the mipmap.
+ * Called via machine->FetchTexelDeriv()
+ * \param lodBias  the lod bias which may be specified by a TXB instruction,
+ *                 otherwise zero.
+ */
+static void
+fetch_texel_deriv( struct gl_context *ctx, const GLfloat texcoord[4],
+                   const GLfloat texdx[4], const GLfloat texdy[4],
+                   GLfloat lodBias, GLuint unit, GLfloat color[4] )
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   const struct gl_texture_object *texObj = texUnit->_Current;
+
+   if (texObj) {
+      const struct gl_texture_image *texImg =
+         texObj->Image[0][texObj->BaseLevel];
+      const GLfloat texW = (GLfloat) texImg->WidthScale;
+      const GLfloat texH = (GLfloat) texImg->HeightScale;
+      GLfloat lambda;
+      GLfloat rgba[4];
+
+      lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
+                                      texdx[1], texdy[1], /* dt/dx, dt/dy */
+                                      texdx[3], texdy[3], /* dq/dx, dq/dy */
+                                      texW, texH,
+                                      texcoord[0], texcoord[1], texcoord[3],
+                                      1.0F / texcoord[3]);
+
+      lambda += lodBias + texUnit->LodBias + texObj->Sampler.LodBias;
+
+      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);
+
+      swrast->TextureSample[unit](ctx, texObj, 1,
+                                  (const GLfloat (*)[4]) texcoord,
+                                  &lambda, &rgba);
+      swizzle_texel(rgba, color, texObj->_Swizzle);
+   }
+   else {
+      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
+   }
+}
+
+
+/**
+ * Initialize the virtual fragment program machine state prior to running
+ * fragment program on a fragment.  This involves initializing the input
+ * registers, condition codes, etc.
+ * \param machine  the virtual machine state to init
+ * \param program  the fragment program we're about to run
+ * \param span  the span of pixels we'll operate on
+ * \param col  which element (column) of the span we'll operate on
+ */
+static void
+init_machine(struct gl_context *ctx, struct gl_program_machine *machine,
+             const struct gl_fragment_program *program,
+             const SWspan *span, GLuint col)
+{
+   GLfloat *wpos = span->array->attribs[FRAG_ATTRIB_WPOS][col];
+
+   if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
+      /* Clear temporary registers (undefined for ARB_f_p) */
+      memset(machine->Temporaries, 0, MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
+   }
+
+   /* ARB_fragment_coord_conventions */
+   if (program->OriginUpperLeft)
+      wpos[1] = ctx->DrawBuffer->Height - 1 - wpos[1];
+   if (!program->PixelCenterInteger) {
+      wpos[0] += 0.5F;
+      wpos[1] += 0.5F;
+   }
+
+   /* Setup pointer to input attributes */
+   machine->Attribs = span->array->attribs;
+
+   machine->DerivX = (GLfloat (*)[4]) span->attrStepX;
+   machine->DerivY = (GLfloat (*)[4]) span->attrStepY;
+   machine->NumDeriv = FRAG_ATTRIB_MAX;
+
+   machine->Samplers = program->Base.SamplerUnits;
+
+   /* if running a GLSL program (not ARB_fragment_program) */
+   if (ctx->Shader.CurrentFragmentProgram) {
+      /* Store front/back facing value */
+      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0F - span->facing;
+   }
+
+   machine->CurElement = col;
+
+   /* init condition codes */
+   machine->CondCodes[0] = COND_EQ;
+   machine->CondCodes[1] = COND_EQ;
+   machine->CondCodes[2] = COND_EQ;
+   machine->CondCodes[3] = COND_EQ;
+
+   /* init call stack */
+   machine->StackDepth = 0;
+
+   machine->FetchTexelLod = fetch_texel_lod;
+   machine->FetchTexelDeriv = fetch_texel_deriv;
+}
+
+
+/**
+ * Run fragment program on the pixels in span from 'start' to 'end' - 1.
+ */
+static void
+run_program(struct gl_context *ctx, SWspan *span, GLuint start, GLuint end)
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
+   const GLbitfield64 outputsWritten = program->Base.OutputsWritten;
+   struct gl_program_machine *machine = &swrast->FragProgMachine;
+   GLuint i;
+
+   for (i = start; i < end; i++) {
+      if (span->array->mask[i]) {
+         init_machine(ctx, machine, program, span, i);
+
+         if (_mesa_execute_program(ctx, &program->Base, machine)) {
+
+            /* Store result color */
+	    if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
+               COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
+                       machine->Outputs[FRAG_RESULT_COLOR]);
+            }
+            else {
+               /* Multiple drawbuffers / render targets
+                * Note that colors beyond 0 and 1 will overwrite other
+                * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.
+                */
+               GLuint buf;
+               for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
+                  if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {
+                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
+                             machine->Outputs[FRAG_RESULT_DATA0 + buf]);
+                  }
+               }
+            }
+
+            /* Store result depth/z */
+            if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
+               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
+               if (depth <= 0.0)
+                  span->array->z[i] = 0;
+               else if (depth >= 1.0)
+                  span->array->z[i] = ctx->DrawBuffer->_DepthMax;
+               else
+                  span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
+            }
+         }
+         else {
+            /* killed fragment */
+            span->array->mask[i] = GL_FALSE;
+            span->writeAll = GL_FALSE;
+         }
+      }
+   }
+}
+
+
+/**
+ * Execute the current fragment program for all the fragments
+ * in the given span.
+ */
+void
+_swrast_exec_fragment_program( struct gl_context *ctx, SWspan *span )
+{
+   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
+
+   /* incoming colors should be floats */
+   if (program->Base.InputsRead & FRAG_BIT_COL0) {
+      ASSERT(span->array->ChanType == GL_FLOAT);
+   }
+
+   run_program(ctx, span, 0, span->end);
+
+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
+      span->interpMask &= ~SPAN_RGBA;
+      span->arrayMask |= SPAN_RGBA;
+   }
+
+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
+      span->interpMask &= ~SPAN_Z;
+      span->arrayMask |= SPAN_Z;
+   }
+}
+
diff --git a/mesalib/src/mesa/swrast/s_span.c b/mesalib/src/mesa/swrast/s_span.c
index 75e3c53a1..9cfecc9e3 100644
--- a/mesalib/src/mesa/swrast/s_span.c
+++ b/mesalib/src/mesa/swrast/s_span.c
@@ -1,1503 +1,1503 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.5
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- * Copyright (C) 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, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/**
- * \file swrast/s_span.c
- * \brief Span processing functions used by all rasterization functions.
- * This is where all the per-fragment tests are performed
- * \author Brian Paul
- */
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "main/macros.h"
-#include "main/imports.h"
-#include "main/image.h"
-
-#include "s_atifragshader.h"
-#include "s_alpha.h"
-#include "s_blend.h"
-#include "s_context.h"
-#include "s_depth.h"
-#include "s_fog.h"
-#include "s_logic.h"
-#include "s_masking.h"
-#include "s_fragprog.h"
-#include "s_span.h"
-#include "s_stencil.h"
-#include "s_texcombine.h"
-
-
-/**
- * Set default fragment attributes for the span using the
- * current raster values.  Used prior to glDraw/CopyPixels
- * and glBitmap.
- */
-void
-_swrast_span_default_attribs(struct gl_context *ctx, SWspan *span)
-{
-   GLchan r, g, b, a;
-   /* Z*/
-   {
-      const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;
-      if (ctx->DrawBuffer->Visual.depthBits <= 16)
-         span->z = FloatToFixed(ctx->Current.RasterPos[2] * depthMax + 0.5F);
-      else {
-         GLfloat tmpf = ctx->Current.RasterPos[2] * depthMax; 
-         tmpf = MIN2(tmpf, depthMax);
-         span->z = (GLint)tmpf;
-      }
-      span->zStep = 0;
-      span->interpMask |= SPAN_Z;
-   }
-
-   /* W (for perspective correction) */
-   span->attrStart[FRAG_ATTRIB_WPOS][3] = 1.0;
-   span->attrStepX[FRAG_ATTRIB_WPOS][3] = 0.0;
-   span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0;
-
-   /* primary color, or color index */
-   UNCLAMPED_FLOAT_TO_CHAN(r, ctx->Current.RasterColor[0]);
-   UNCLAMPED_FLOAT_TO_CHAN(g, ctx->Current.RasterColor[1]);
-   UNCLAMPED_FLOAT_TO_CHAN(b, ctx->Current.RasterColor[2]);
-   UNCLAMPED_FLOAT_TO_CHAN(a, ctx->Current.RasterColor[3]);
-#if CHAN_TYPE == GL_FLOAT
-   span->red = r;
-   span->green = g;
-   span->blue = b;
-   span->alpha = a;
-#else
-   span->red   = IntToFixed(r);
-   span->green = IntToFixed(g);
-   span->blue  = IntToFixed(b);
-   span->alpha = IntToFixed(a);
-#endif
-   span->redStep = 0;
-   span->greenStep = 0;
-   span->blueStep = 0;
-   span->alphaStep = 0;
-   span->interpMask |= SPAN_RGBA;
-
-   COPY_4V(span->attrStart[FRAG_ATTRIB_COL0], ctx->Current.RasterColor);
-   ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
-   ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
-
-   /* Secondary color */
-   if (ctx->Light.Enabled || ctx->Fog.ColorSumEnabled)
-   {
-      COPY_4V(span->attrStart[FRAG_ATTRIB_COL1], ctx->Current.RasterSecondaryColor);
-      ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
-      ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
-   }
-
-   /* fog */
-   {
-      const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-      GLfloat fogVal; /* a coord or a blend factor */
-      if (swrast->_PreferPixelFog) {
-         /* fog blend factors will be computed from fog coordinates per pixel */
-         fogVal = ctx->Current.RasterDistance;
-      }
-      else {
-         /* fog blend factor should be computed from fogcoord now */
-         fogVal = _swrast_z_to_fogfactor(ctx, ctx->Current.RasterDistance);
-      }
-      span->attrStart[FRAG_ATTRIB_FOGC][0] = fogVal;
-      span->attrStepX[FRAG_ATTRIB_FOGC][0] = 0.0;
-      span->attrStepY[FRAG_ATTRIB_FOGC][0] = 0.0;
-   }
-
-   /* texcoords */
-   {
-      GLuint i;
-      for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
-         const GLuint attr = FRAG_ATTRIB_TEX0 + i;
-         const GLfloat *tc = ctx->Current.RasterTexCoords[i];
-         if (ctx->FragmentProgram._Current || ctx->ATIFragmentShader._Enabled) {
-            COPY_4V(span->attrStart[attr], tc);
-         }
-         else if (tc[3] > 0.0F) {
-            /* use (s/q, t/q, r/q, 1) */
-            span->attrStart[attr][0] = tc[0] / tc[3];
-            span->attrStart[attr][1] = tc[1] / tc[3];
-            span->attrStart[attr][2] = tc[2] / tc[3];
-            span->attrStart[attr][3] = 1.0;
-         }
-         else {
-            ASSIGN_4V(span->attrStart[attr], 0.0F, 0.0F, 0.0F, 1.0F);
-         }
-         ASSIGN_4V(span->attrStepX[attr], 0.0F, 0.0F, 0.0F, 0.0F);
-         ASSIGN_4V(span->attrStepY[attr], 0.0F, 0.0F, 0.0F, 0.0F);
-      }
-   }
-}
-
-
-/**
- * Interpolate the active attributes (and'd with attrMask) to
- * fill in span->array->attribs[].
- * Perspective correction will be done.  The point/line/triangle function
- * should have computed attrStart/Step values for FRAG_ATTRIB_WPOS[3]!
- */
-static INLINE void
-interpolate_active_attribs(struct gl_context *ctx, SWspan *span, GLbitfield attrMask)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   /*
-    * Don't overwrite existing array values, such as colors that may have
-    * been produced by glDraw/CopyPixels.
-    */
-   attrMask &= ~span->arrayAttribs;
-
-   ATTRIB_LOOP_BEGIN
-      if (attrMask & (1 << attr)) {
-         const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-         GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];
-         const GLfloat dv0dx = span->attrStepX[attr][0];
-         const GLfloat dv1dx = span->attrStepX[attr][1];
-         const GLfloat dv2dx = span->attrStepX[attr][2];
-         const GLfloat dv3dx = span->attrStepX[attr][3];
-         GLfloat v0 = span->attrStart[attr][0] + span->leftClip * dv0dx;
-         GLfloat v1 = span->attrStart[attr][1] + span->leftClip * dv1dx;
-         GLfloat v2 = span->attrStart[attr][2] + span->leftClip * dv2dx;
-         GLfloat v3 = span->attrStart[attr][3] + span->leftClip * dv3dx;
-         GLuint k;
-         for (k = 0; k < span->end; k++) {
-            const GLfloat invW = 1.0f / w;
-            span->array->attribs[attr][k][0] = v0 * invW;
-            span->array->attribs[attr][k][1] = v1 * invW;
-            span->array->attribs[attr][k][2] = v2 * invW;
-            span->array->attribs[attr][k][3] = v3 * invW;
-            v0 += dv0dx;
-            v1 += dv1dx;
-            v2 += dv2dx;
-            v3 += dv3dx;
-            w += dwdx;
-         }
-         ASSERT((span->arrayAttribs & (1 << attr)) == 0);
-         span->arrayAttribs |= (1 << attr);
-      }
-   ATTRIB_LOOP_END
-}
-
-
-/**
- * Interpolate primary colors to fill in the span->array->rgba8 (or rgb16)
- * color array.
- */
-static INLINE void
-interpolate_int_colors(struct gl_context *ctx, SWspan *span)
-{
-   const GLuint n = span->end;
-   GLuint i;
-
-#if CHAN_BITS != 32
-   ASSERT(!(span->arrayMask & SPAN_RGBA));
-#endif
-
-   switch (span->array->ChanType) {
-#if CHAN_BITS != 32
-   case GL_UNSIGNED_BYTE:
-      {
-         GLubyte (*rgba)[4] = span->array->rgba8;
-         if (span->interpMask & SPAN_FLAT) {
-            GLubyte color[4];
-            color[RCOMP] = FixedToInt(span->red);
-            color[GCOMP] = FixedToInt(span->green);
-            color[BCOMP] = FixedToInt(span->blue);
-            color[ACOMP] = FixedToInt(span->alpha);
-            for (i = 0; i < n; i++) {
-               COPY_4UBV(rgba[i], color);
-            }
-         }
-         else {
-            GLfixed r = span->red;
-            GLfixed g = span->green;
-            GLfixed b = span->blue;
-            GLfixed a = span->alpha;
-            GLint dr = span->redStep;
-            GLint dg = span->greenStep;
-            GLint db = span->blueStep;
-            GLint da = span->alphaStep;
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = FixedToChan(r);
-               rgba[i][GCOMP] = FixedToChan(g);
-               rgba[i][BCOMP] = FixedToChan(b);
-               rgba[i][ACOMP] = FixedToChan(a);
-               r += dr;
-               g += dg;
-               b += db;
-               a += da;
-            }
-         }
-      }
-      break;
-   case GL_UNSIGNED_SHORT:
-      {
-         GLushort (*rgba)[4] = span->array->rgba16;
-         if (span->interpMask & SPAN_FLAT) {
-            GLushort color[4];
-            color[RCOMP] = FixedToInt(span->red);
-            color[GCOMP] = FixedToInt(span->green);
-            color[BCOMP] = FixedToInt(span->blue);
-            color[ACOMP] = FixedToInt(span->alpha);
-            for (i = 0; i < n; i++) {
-               COPY_4V(rgba[i], color);
-            }
-         }
-         else {
-            GLushort (*rgba)[4] = span->array->rgba16;
-            GLfixed r, g, b, a;
-            GLint dr, dg, db, da;
-            r = span->red;
-            g = span->green;
-            b = span->blue;
-            a = span->alpha;
-            dr = span->redStep;
-            dg = span->greenStep;
-            db = span->blueStep;
-            da = span->alphaStep;
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = FixedToChan(r);
-               rgba[i][GCOMP] = FixedToChan(g);
-               rgba[i][BCOMP] = FixedToChan(b);
-               rgba[i][ACOMP] = FixedToChan(a);
-               r += dr;
-               g += dg;
-               b += db;
-               a += da;
-            }
-         }
-      }
-      break;
-#endif
-   case GL_FLOAT:
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-      break;
-   default:
-      _mesa_problem(ctx, "bad datatype 0x%x in interpolate_int_colors",
-                    span->array->ChanType);
-   }
-   span->arrayMask |= SPAN_RGBA;
-}
-
-
-/**
- * Populate the FRAG_ATTRIB_COL0 array.
- */
-static INLINE void
-interpolate_float_colors(SWspan *span)
-{
-   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   const GLuint n = span->end;
-   GLuint i;
-
-   assert(!(span->arrayAttribs & FRAG_BIT_COL0));
-
-   if (span->arrayMask & SPAN_RGBA) {
-      /* convert array of int colors */
-      for (i = 0; i < n; i++) {
-         col0[i][0] = UBYTE_TO_FLOAT(span->array->rgba8[i][0]);
-         col0[i][1] = UBYTE_TO_FLOAT(span->array->rgba8[i][1]);
-         col0[i][2] = UBYTE_TO_FLOAT(span->array->rgba8[i][2]);
-         col0[i][3] = UBYTE_TO_FLOAT(span->array->rgba8[i][3]);
-      }
-   }
-   else {
-      /* interpolate red/green/blue/alpha to get float colors */
-      ASSERT(span->interpMask & SPAN_RGBA);
-      if (span->interpMask & SPAN_FLAT) {
-         GLfloat r = FixedToFloat(span->red);
-         GLfloat g = FixedToFloat(span->green);
-         GLfloat b = FixedToFloat(span->blue);
-         GLfloat a = FixedToFloat(span->alpha);
-         for (i = 0; i < n; i++) {
-            ASSIGN_4V(col0[i], r, g, b, a);
-         }
-      }
-      else {
-         GLfloat r = FixedToFloat(span->red);
-         GLfloat g = FixedToFloat(span->green);
-         GLfloat b = FixedToFloat(span->blue);
-         GLfloat a = FixedToFloat(span->alpha);
-         GLfloat dr = FixedToFloat(span->redStep);
-         GLfloat dg = FixedToFloat(span->greenStep);
-         GLfloat db = FixedToFloat(span->blueStep);
-         GLfloat da = FixedToFloat(span->alphaStep);
-         for (i = 0; i < n; i++) {
-            col0[i][0] = r;
-            col0[i][1] = g;
-            col0[i][2] = b;
-            col0[i][3] = a;
-            r += dr;
-            g += dg;
-            b += db;
-            a += da;
-         }
-      }
-   }
-
-   span->arrayAttribs |= FRAG_BIT_COL0;
-   span->array->ChanType = GL_FLOAT;
-}
-
-
-
-/**
- * Fill in the span.zArray array from the span->z, zStep values.
- */
-void
-_swrast_span_interpolate_z( const struct gl_context *ctx, SWspan *span )
-{
-   const GLuint n = span->end;
-   GLuint i;
-
-   ASSERT(!(span->arrayMask & SPAN_Z));
-
-   if (ctx->DrawBuffer->Visual.depthBits <= 16) {
-      GLfixed zval = span->z;
-      GLuint *z = span->array->z; 
-      for (i = 0; i < n; i++) {
-         z[i] = FixedToInt(zval);
-         zval += span->zStep;
-      }
-   }
-   else {
-      /* Deep Z buffer, no fixed->int shift */
-      GLuint zval = span->z;
-      GLuint *z = span->array->z;
-      for (i = 0; i < n; i++) {
-         z[i] = zval;
-         zval += span->zStep;
-      }
-   }
-   span->interpMask &= ~SPAN_Z;
-   span->arrayMask |= SPAN_Z;
-}
-
-
-/**
- * Compute mipmap LOD from partial derivatives.
- * This the ideal solution, as given in the OpenGL spec.
- */
-GLfloat
-_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
-                       GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
-                       GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
-{
-   GLfloat dudx = texW * ((s + dsdx) / (q + dqdx) - s * invQ);
-   GLfloat dvdx = texH * ((t + dtdx) / (q + dqdx) - t * invQ);
-   GLfloat dudy = texW * ((s + dsdy) / (q + dqdy) - s * invQ);
-   GLfloat dvdy = texH * ((t + dtdy) / (q + dqdy) - t * invQ);
-   GLfloat x = SQRTF(dudx * dudx + dvdx * dvdx);
-   GLfloat y = SQRTF(dudy * dudy + dvdy * dvdy);
-   GLfloat rho = MAX2(x, y);
-   GLfloat lambda = LOG2(rho);
-   return lambda;
-}
-
-
-/**
- * Compute mipmap LOD from partial derivatives.
- * This is a faster approximation than above function.
- */
-#if 0
-GLfloat
-_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
-                     GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
-                     GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
-{
-   GLfloat dsdx2 = (s + dsdx) / (q + dqdx) - s * invQ;
-   GLfloat dtdx2 = (t + dtdx) / (q + dqdx) - t * invQ;
-   GLfloat dsdy2 = (s + dsdy) / (q + dqdy) - s * invQ;
-   GLfloat dtdy2 = (t + dtdy) / (q + dqdy) - t * invQ;
-   GLfloat maxU, maxV, rho, lambda;
-   dsdx2 = FABSF(dsdx2);
-   dsdy2 = FABSF(dsdy2);
-   dtdx2 = FABSF(dtdx2);
-   dtdy2 = FABSF(dtdy2);
-   maxU = MAX2(dsdx2, dsdy2) * texW;
-   maxV = MAX2(dtdx2, dtdy2) * texH;
-   rho = MAX2(maxU, maxV);
-   lambda = LOG2(rho);
-   return lambda;
-}
-#endif
-
-
-/**
- * Fill in the span.array->attrib[FRAG_ATTRIB_TEXn] arrays from the
- * using the attrStart/Step values.
- *
- * This function only used during fixed-function fragment processing.
- *
- * Note: in the places where we divide by Q (or mult by invQ) we're
- * really doing two things: perspective correction and texcoord
- * projection.  Remember, for texcoord (s,t,r,q) we need to index
- * texels with (s/q, t/q, r/q).
- */
-static void
-interpolate_texcoords(struct gl_context *ctx, SWspan *span)
-{
-   const GLuint maxUnit
-      = (ctx->Texture._EnabledCoordUnits > 1) ? ctx->Const.MaxTextureUnits : 1;
-   GLuint u;
-
-   /* XXX CoordUnits vs. ImageUnits */
-   for (u = 0; u < maxUnit; u++) {
-      if (ctx->Texture._EnabledCoordUnits & (1 << u)) {
-         const GLuint attr = FRAG_ATTRIB_TEX0 + u;
-         const struct gl_texture_object *obj = ctx->Texture.Unit[u]._Current;
-         GLfloat texW, texH;
-         GLboolean needLambda;
-         GLfloat (*texcoord)[4] = span->array->attribs[attr];
-         GLfloat *lambda = span->array->lambda[u];
-         const GLfloat dsdx = span->attrStepX[attr][0];
-         const GLfloat dsdy = span->attrStepY[attr][0];
-         const GLfloat dtdx = span->attrStepX[attr][1];
-         const GLfloat dtdy = span->attrStepY[attr][1];
-         const GLfloat drdx = span->attrStepX[attr][2];
-         const GLfloat dqdx = span->attrStepX[attr][3];
-         const GLfloat dqdy = span->attrStepY[attr][3];
-         GLfloat s = span->attrStart[attr][0] + span->leftClip * dsdx;
-         GLfloat t = span->attrStart[attr][1] + span->leftClip * dtdx;
-         GLfloat r = span->attrStart[attr][2] + span->leftClip * drdx;
-         GLfloat q = span->attrStart[attr][3] + span->leftClip * dqdx;
-
-         if (obj) {
-            const struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
-            needLambda = (obj->MinFilter != obj->MagFilter)
-               || ctx->FragmentProgram._Current;
-            texW = img->WidthScale;
-            texH = img->HeightScale;
-         }
-         else {
-            /* using a fragment program */
-            texW = 1.0;
-            texH = 1.0;
-            needLambda = GL_FALSE;
-         }
-
-         if (needLambda) {
-            GLuint i;
-            if (ctx->FragmentProgram._Current
-                || ctx->ATIFragmentShader._Enabled) {
-               /* do perspective correction but don't divide s, t, r by q */
-               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invW = 1.0F / w;
-                  texcoord[i][0] = s * invW;
-                  texcoord[i][1] = t * invW;
-                  texcoord[i][2] = r * invW;
-                  texcoord[i][3] = q * invW;
-                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
-                                                     dqdx, dqdy, texW, texH,
-                                                     s, t, q, invW);
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-                  w += dwdx;
-               }
-            }
-            else {
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
-                                                     dqdx, dqdy, texW, texH,
-                                                     s, t, q, invQ);
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-               }
-            }
-            span->arrayMask |= SPAN_LAMBDA;
-         }
-         else {
-            GLuint i;
-            if (ctx->FragmentProgram._Current ||
-                ctx->ATIFragmentShader._Enabled) {
-               /* do perspective correction but don't divide s, t, r by q */
-               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invW = 1.0F / w;
-                  texcoord[i][0] = s * invW;
-                  texcoord[i][1] = t * invW;
-                  texcoord[i][2] = r * invW;
-                  texcoord[i][3] = q * invW;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-                  w += dwdx;
-               }
-            }
-            else if (dqdx == 0.0F) {
-               /* Ortho projection or polygon's parallel to window X axis */
-               const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-               for (i = 0; i < span->end; i++) {
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-               }
-            }
-            else {
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-               }
-            }
-         } /* lambda */
-      } /* if */
-   } /* for */
-}
-
-
-/**
- * Fill in the arrays->attribs[FRAG_ATTRIB_WPOS] array.
- */
-static INLINE void
-interpolate_wpos(struct gl_context *ctx, SWspan *span)
-{
-   GLfloat (*wpos)[4] = span->array->attribs[FRAG_ATTRIB_WPOS];
-   GLuint i;
-   const GLfloat zScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
-   GLfloat w, dw;
-
-   if (span->arrayMask & SPAN_XY) {
-      for (i = 0; i < span->end; i++) {
-         wpos[i][0] = (GLfloat) span->array->x[i];
-         wpos[i][1] = (GLfloat) span->array->y[i];
-      }
-   }
-   else {
-      for (i = 0; i < span->end; i++) {
-         wpos[i][0] = (GLfloat) span->x + i;
-         wpos[i][1] = (GLfloat) span->y;
-      }
-   }
-
-   dw = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-   w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dw;
-   for (i = 0; i < span->end; i++) {
-      wpos[i][2] = (GLfloat) span->array->z[i] * zScale;
-      wpos[i][3] = w;
-      w += dw;
-   }
-}
-
-
-/**
- * Apply the current polygon stipple pattern to a span of pixels.
- */
-static INLINE void
-stipple_polygon_span(struct gl_context *ctx, SWspan *span)
-{
-   GLubyte *mask = span->array->mask;
-
-   ASSERT(ctx->Polygon.StippleFlag);
-
-   if (span->arrayMask & SPAN_XY) {
-      /* arrays of x/y pixel coords */
-      GLuint i;
-      for (i = 0; i < span->end; i++) {
-         const GLint col = span->array->x[i] % 32;
-         const GLint row = span->array->y[i] % 32;
-         const GLuint stipple = ctx->PolygonStipple[row];
-         if (((1 << col) & stipple) == 0) {
-            mask[i] = 0;
-         }
-      }
-   }
-   else {
-      /* horizontal span of pixels */
-      const GLuint highBit = 1 << 31;
-      const GLuint stipple = ctx->PolygonStipple[span->y % 32];
-      GLuint i, m = highBit >> (GLuint) (span->x % 32);
-      for (i = 0; i < span->end; i++) {
-         if ((m & stipple) == 0) {
-            mask[i] = 0;
-         }
-         m = m >> 1;
-         if (m == 0) {
-            m = highBit;
-         }
-      }
-   }
-   span->writeAll = GL_FALSE;
-}
-
-
-/**
- * Clip a pixel span to the current buffer/window boundaries:
- * DrawBuffer->_Xmin, _Xmax, _Ymin, _Ymax.  This will accomplish
- * window clipping and scissoring.
- * Return:   GL_TRUE   some pixels still visible
- *           GL_FALSE  nothing visible
- */
-static INLINE GLuint
-clip_span( struct gl_context *ctx, SWspan *span )
-{
-   const GLint xmin = ctx->DrawBuffer->_Xmin;
-   const GLint xmax = ctx->DrawBuffer->_Xmax;
-   const GLint ymin = ctx->DrawBuffer->_Ymin;
-   const GLint ymax = ctx->DrawBuffer->_Ymax;
-
-   span->leftClip = 0;
-
-   if (span->arrayMask & SPAN_XY) {
-      /* arrays of x/y pixel coords */
-      const GLint *x = span->array->x;
-      const GLint *y = span->array->y;
-      const GLint n = span->end;
-      GLubyte *mask = span->array->mask;
-      GLint i;
-      if (span->arrayMask & SPAN_MASK) {
-         /* note: using & intead of && to reduce branches */
-         for (i = 0; i < n; i++) {
-            mask[i] &= (x[i] >= xmin) & (x[i] < xmax)
-                     & (y[i] >= ymin) & (y[i] < ymax);
-         }
-      }
-      else {
-         /* note: using & intead of && to reduce branches */
-         for (i = 0; i < n; i++) {
-            mask[i] = (x[i] >= xmin) & (x[i] < xmax)
-                    & (y[i] >= ymin) & (y[i] < ymax);
-         }
-      }
-      return GL_TRUE;  /* some pixels visible */
-   }
-   else {
-      /* horizontal span of pixels */
-      const GLint x = span->x;
-      const GLint y = span->y;
-      GLint n = span->end;
-
-      /* Trivial rejection tests */
-      if (y < ymin || y >= ymax || x + n <= xmin || x >= xmax) {
-         span->end = 0;
-         return GL_FALSE;  /* all pixels clipped */
-      }
-
-      /* Clip to right */
-      if (x + n > xmax) {
-         ASSERT(x < xmax);
-         n = span->end = xmax - x;
-      }
-
-      /* Clip to the left */
-      if (x < xmin) {
-         const GLint leftClip = xmin - x;
-         GLuint i;
-
-         ASSERT(leftClip > 0);
-         ASSERT(x + n > xmin);
-
-         /* Clip 'leftClip' pixels from the left side.
-          * The span->leftClip field will be applied when we interpolate
-          * fragment attributes.
-          * For arrays of values, shift them left.
-          */
-         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
-            if (span->interpMask & (1 << i)) {
-               GLuint j;
-               for (j = 0; j < 4; j++) {
-                  span->attrStart[i][j] += leftClip * span->attrStepX[i][j];
-               }
-            }
-         }
-
-         span->red += leftClip * span->redStep;
-         span->green += leftClip * span->greenStep;
-         span->blue += leftClip * span->blueStep;
-         span->alpha += leftClip * span->alphaStep;
-         span->index += leftClip * span->indexStep;
-         span->z += leftClip * span->zStep;
-         span->intTex[0] += leftClip * span->intTexStep[0];
-         span->intTex[1] += leftClip * span->intTexStep[1];
-
-#define SHIFT_ARRAY(ARRAY, SHIFT, LEN) \
-         memcpy(ARRAY, ARRAY + (SHIFT), (LEN) * sizeof(ARRAY[0]))
-
-         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
-            if (span->arrayAttribs & (1 << i)) {
-               /* shift array elements left by 'leftClip' */
-               SHIFT_ARRAY(span->array->attribs[i], leftClip, n - leftClip);
-            }
-         }
-
-         SHIFT_ARRAY(span->array->mask, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->rgba8, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->rgba16, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->x, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->y, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->z, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->index, leftClip, n - leftClip);
-         for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
-            SHIFT_ARRAY(span->array->lambda[i], leftClip, n - leftClip);
-         }
-         SHIFT_ARRAY(span->array->coverage, leftClip, n - leftClip);
-
-#undef SHIFT_ARRAY
-
-         span->leftClip = leftClip;
-         span->x = xmin;
-         span->end -= leftClip;
-         span->writeAll = GL_FALSE;
-      }
-
-      ASSERT(span->x >= xmin);
-      ASSERT(span->x + span->end <= xmax);
-      ASSERT(span->y >= ymin);
-      ASSERT(span->y < ymax);
-
-      return GL_TRUE;  /* some pixels visible */
-   }
-}
-
-
-/**
- * Add specular colors to primary colors.
- * Only called during fixed-function operation.
- * Result is float color array (FRAG_ATTRIB_COL0).
- */
-static INLINE void
-add_specular(struct gl_context *ctx, SWspan *span)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const GLubyte *mask = span->array->mask;
-   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   GLfloat (*col1)[4] = span->array->attribs[FRAG_ATTRIB_COL1];
-   GLuint i;
-
-   ASSERT(!ctx->FragmentProgram._Current);
-   ASSERT(span->arrayMask & SPAN_RGBA);
-   ASSERT(swrast->_ActiveAttribMask & FRAG_BIT_COL1);
-   (void) swrast; /* silence warning */
-
-   if (span->array->ChanType == GL_FLOAT) {
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-      }
-   }
-   else {
-      /* need float colors */
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_float_colors(span);
-      }
-   }
-
-   if ((span->arrayAttribs & FRAG_BIT_COL1) == 0) {
-      /* XXX could avoid this and interpolate COL1 in the loop below */
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL1);
-   }
-
-   ASSERT(span->arrayAttribs & FRAG_BIT_COL0);
-   ASSERT(span->arrayAttribs & FRAG_BIT_COL1);
-
-   for (i = 0; i < span->end; i++) {
-      if (mask[i]) {
-         col0[i][0] += col1[i][0];
-         col0[i][1] += col1[i][1];
-         col0[i][2] += col1[i][2];
-      }
-   }
-
-   span->array->ChanType = GL_FLOAT;
-}
-
-
-/**
- * Apply antialiasing coverage value to alpha values.
- */
-static INLINE void
-apply_aa_coverage(SWspan *span)
-{
-   const GLfloat *coverage = span->array->coverage;
-   GLuint i;
-   if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      GLubyte (*rgba)[4] = span->array->rgba8;
-      for (i = 0; i < span->end; i++) {
-         const GLfloat a = rgba[i][ACOMP] * coverage[i];
-         rgba[i][ACOMP] = (GLubyte) CLAMP(a, 0.0, 255.0);
-         ASSERT(coverage[i] >= 0.0);
-         ASSERT(coverage[i] <= 1.0);
-      }
-   }
-   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-      GLushort (*rgba)[4] = span->array->rgba16;
-      for (i = 0; i < span->end; i++) {
-         const GLfloat a = rgba[i][ACOMP] * coverage[i];
-         rgba[i][ACOMP] = (GLushort) CLAMP(a, 0.0, 65535.0);
-      }
-   }
-   else {
-      GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-      for (i = 0; i < span->end; i++) {
-         rgba[i][ACOMP] = rgba[i][ACOMP] * coverage[i];
-         /* clamp later */
-      }
-   }
-}
-
-
-/**
- * Clamp span's float colors to [0,1]
- */
-static INLINE void
-clamp_colors(SWspan *span)
-{
-   GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   GLuint i;
-   ASSERT(span->array->ChanType == GL_FLOAT);
-   for (i = 0; i < span->end; i++) {
-      rgba[i][RCOMP] = CLAMP(rgba[i][RCOMP], 0.0F, 1.0F);
-      rgba[i][GCOMP] = CLAMP(rgba[i][GCOMP], 0.0F, 1.0F);
-      rgba[i][BCOMP] = CLAMP(rgba[i][BCOMP], 0.0F, 1.0F);
-      rgba[i][ACOMP] = CLAMP(rgba[i][ACOMP], 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Convert the span's color arrays to the given type.
- * The only way 'output' can be greater than zero is when we have a fragment
- * program that writes to gl_FragData[1] or higher.
- * \param output  which fragment program color output is being processed
- */
-static INLINE void
-convert_color_type(SWspan *span, GLenum newType, GLuint output)
-{
-   GLvoid *src, *dst;
-
-   if (output > 0 || span->array->ChanType == GL_FLOAT) {
-      src = span->array->attribs[FRAG_ATTRIB_COL0 + output];
-      span->array->ChanType = GL_FLOAT;
-   }
-   else if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      src = span->array->rgba8;
-   }
-   else {
-      ASSERT(span->array->ChanType == GL_UNSIGNED_SHORT);
-      src = span->array->rgba16;
-   }
-
-   if (newType == GL_UNSIGNED_BYTE) {
-      dst = span->array->rgba8;
-   }
-   else if (newType == GL_UNSIGNED_SHORT) {
-      dst = span->array->rgba16;
-   }
-   else {
-      dst = span->array->attribs[FRAG_ATTRIB_COL0];
-   }
-
-   _mesa_convert_colors(span->array->ChanType, src,
-                        newType, dst,
-                        span->end, span->array->mask);
-
-   span->array->ChanType = newType;
-   span->array->rgba = dst;
-}
-
-
-
-/**
- * Apply fragment shader, fragment program or normal texturing to span.
- */
-static INLINE void
-shade_texture_span(struct gl_context *ctx, SWspan *span)
-{
-   GLbitfield inputsRead;
-
-   /* Determine which fragment attributes are actually needed */
-   if (ctx->FragmentProgram._Current) {
-      inputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
-   }
-   else {
-      /* XXX we could be a bit smarter about this */
-      inputsRead = ~0;
-   }
-
-   if (ctx->FragmentProgram._Current ||
-       ctx->ATIFragmentShader._Enabled) {
-      /* programmable shading */
-      if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {
-         convert_color_type(span, GL_FLOAT, 0);
-      }
-      else {
-         span->array->rgba = (void *) span->array->attribs[FRAG_ATTRIB_COL0];
-      }
-
-      if (span->primitive != GL_POINT ||
-	  (span->interpMask & SPAN_RGBA) ||
-	  ctx->Point.PointSprite) {
-         /* for single-pixel points, we populated the arrays already */
-         interpolate_active_attribs(ctx, span, ~0);
-      }
-      span->array->ChanType = GL_FLOAT;
-
-      if (!(span->arrayMask & SPAN_Z))
-         _swrast_span_interpolate_z (ctx, span);
-
-#if 0
-      if (inputsRead & FRAG_BIT_WPOS)
-#else
-      /* XXX always interpolate wpos so that DDX/DDY work */
-#endif
-         interpolate_wpos(ctx, span);
-
-      /* Run fragment program/shader now */
-      if (ctx->FragmentProgram._Current) {
-         _swrast_exec_fragment_program(ctx, span);
-      }
-      else {
-         ASSERT(ctx->ATIFragmentShader._Enabled);
-         _swrast_exec_fragment_shader(ctx, span);
-      }
-   }
-   else if (ctx->Texture._EnabledCoordUnits) {
-      /* conventional texturing */
-
-#if CHAN_BITS == 32
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_int_colors(ctx, span);
-      }
-#else
-      if (!(span->arrayMask & SPAN_RGBA))
-         interpolate_int_colors(ctx, span);
-#endif
-      if ((span->arrayAttribs & FRAG_BITS_TEX_ANY) == 0x0)
-         interpolate_texcoords(ctx, span);
-
-      _swrast_texture_span(ctx, span);
-   }
-}
-
-
-
-/**
- * Apply all the per-fragment operations to a span.
- * This now includes texturing (_swrast_write_texture_span() is history).
- * This function may modify any of the array values in the span.
- * span->interpMask and span->arrayMask may be changed but will be restored
- * to their original values before returning.
- */
-void
-_swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const GLuint *colorMask = (GLuint *) ctx->Color.ColorMask;
-   const GLbitfield origInterpMask = span->interpMask;
-   const GLbitfield origArrayMask = span->arrayMask;
-   const GLbitfield origArrayAttribs = span->arrayAttribs;
-   const GLenum origChanType = span->array->ChanType;
-   void * const origRgba = span->array->rgba;
-   const GLboolean shader = (ctx->FragmentProgram._Current
-                             || ctx->ATIFragmentShader._Enabled);
-   const GLboolean shaderOrTexture = shader || ctx->Texture._EnabledCoordUnits;
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-   /*
-   printf("%s()  interp 0x%x  array 0x%x\n", __FUNCTION__,
-          span->interpMask, span->arrayMask);
-   */
-
-   ASSERT(span->primitive == GL_POINT ||
-          span->primitive == GL_LINE ||
-	  span->primitive == GL_POLYGON ||
-          span->primitive == GL_BITMAP);
-
-   /* Fragment write masks */
-   if (span->arrayMask & SPAN_MASK) {
-      /* mask was initialized by caller, probably glBitmap */
-      span->writeAll = GL_FALSE;
-   }
-   else {
-      memset(span->array->mask, 1, span->end);
-      span->writeAll = GL_TRUE;
-   }
-
-   /* Clip to window/scissor box */
-   if (!clip_span(ctx, span)) {
-      return;
-   }
-
-   ASSERT(span->end <= MAX_WIDTH);
-
-   /* Depth bounds test */
-   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
-      if (!_swrast_depth_bounds_test(ctx, span)) {
-         return;
-      }
-   }
-
-#ifdef DEBUG
-   /* Make sure all fragments are within window bounds */
-   if (span->arrayMask & SPAN_XY) {
-      /* array of pixel locations */
-      GLuint i;
-      for (i = 0; i < span->end; i++) {
-         if (span->array->mask[i]) {
-            assert(span->array->x[i] >= fb->_Xmin);
-            assert(span->array->x[i] < fb->_Xmax);
-            assert(span->array->y[i] >= fb->_Ymin);
-            assert(span->array->y[i] < fb->_Ymax);
-         }
-      }
-   }
-#endif
-
-   /* Polygon Stippling */
-   if (ctx->Polygon.StippleFlag && span->primitive == GL_POLYGON) {
-      stipple_polygon_span(ctx, span);
-   }
-
-   /* This is the normal place to compute the fragment color/Z
-    * from texturing or shading.
-    */
-   if (shaderOrTexture && !swrast->_DeferredTexture) {
-      shade_texture_span(ctx, span);
-   }
-
-   /* Do the alpha test */
-   if (ctx->Color.AlphaEnabled) {
-      if (!_swrast_alpha_test(ctx, span)) {
-         /* all fragments failed test */
-         goto end;
-      }
-   }
-
-   /* Stencil and Z testing */
-   if (ctx->Stencil._Enabled || ctx->Depth.Test) {
-      if (!(span->arrayMask & SPAN_Z))
-         _swrast_span_interpolate_z(ctx, span);
-
-      if (ctx->Transform.DepthClamp)
-	 _swrast_depth_clamp_span(ctx, span);
-
-      if (ctx->Stencil._Enabled) {
-         /* Combined Z/stencil tests */
-         if (!_swrast_stencil_and_ztest_span(ctx, span)) {
-            /* all fragments failed test */
-            goto end;
-         }
-      }
-      else if (fb->Visual.depthBits > 0) {
-         /* Just regular depth testing */
-         ASSERT(ctx->Depth.Test);
-         ASSERT(span->arrayMask & SPAN_Z);
-         if (!_swrast_depth_test_span(ctx, span)) {
-            /* all fragments failed test */
-            goto end;
-         }
-      }
-   }
-
-   if (ctx->Query.CurrentOcclusionObject) {
-      /* update count of 'passed' fragments */
-      struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;
-      GLuint i;
-      for (i = 0; i < span->end; i++)
-         q->Result += span->array->mask[i];
-   }
-
-   /* We had to wait until now to check for glColorMask(0,0,0,0) because of
-    * the occlusion test.
-    */
-   if (fb->_NumColorDrawBuffers == 1 && colorMask[0] == 0x0) {
-      /* no colors to write */
-      goto end;
-   }
-
-   /* If we were able to defer fragment color computation to now, there's
-    * a good chance that many fragments will have already been killed by
-    * Z/stencil testing.
-    */
-   if (shaderOrTexture && swrast->_DeferredTexture) {
-      shade_texture_span(ctx, span);
-   }
-
-#if CHAN_BITS == 32
-   if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-   }
-#else
-   if ((span->arrayMask & SPAN_RGBA) == 0) {
-      interpolate_int_colors(ctx, span);
-   }
-#endif
-
-   ASSERT(span->arrayMask & SPAN_RGBA);
-
-   if (span->primitive == GL_BITMAP || !swrast->SpecularVertexAdd) {
-      /* Add primary and specular (diffuse + specular) colors */
-      if (!shader) {
-         if (ctx->Fog.ColorSumEnabled ||
-             (ctx->Light.Enabled &&
-              ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {
-            add_specular(ctx, span);
-         }
-      }
-   }
-
-   /* Fog */
-   if (swrast->_FogEnabled) {
-      _swrast_fog_rgba_span(ctx, span);
-   }
-
-   /* Antialias coverage application */
-   if (span->arrayMask & SPAN_COVERAGE) {
-      apply_aa_coverage(span);
-   }
-
-   /* Clamp color/alpha values over the range [0.0, 1.0] before storage */
-   if (ctx->Color.ClampFragmentColor == GL_TRUE &&
-       span->array->ChanType == GL_FLOAT) {
-      clamp_colors(span);
-   }
-
-   /*
-    * Write to renderbuffers.
-    * Depending on glDrawBuffer() state and the which color outputs are
-    * written by the fragment shader, we may either replicate one color to
-    * all renderbuffers or write a different color to each renderbuffer.
-    * multiFragOutputs=TRUE for the later case.
-    */
-   {
-      const GLuint numBuffers = fb->_NumColorDrawBuffers;
-      const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
-      const GLboolean multiFragOutputs = 
-         (fp && fp->Base.OutputsWritten >= (1 << FRAG_RESULT_DATA0));
-      GLuint buf;
-
-      for (buf = 0; buf < numBuffers; buf++) {
-         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];
-
-         /* color[fragOutput] will be written to buffer[buf] */
-
-         if (rb) {
-            GLchan rgbaSave[MAX_WIDTH][4];
-            const GLuint fragOutput = multiFragOutputs ? buf : 0;
-
-            /* set span->array->rgba to colors for render buffer's datatype */
-            if (rb->DataType != span->array->ChanType || fragOutput > 0) {
-               convert_color_type(span, rb->DataType, fragOutput);
-            }
-            else {
-               if (rb->DataType == GL_UNSIGNED_BYTE) {
-                  span->array->rgba = span->array->rgba8;
-               }
-               else if (rb->DataType == GL_UNSIGNED_SHORT) {
-                  span->array->rgba = (void *) span->array->rgba16;
-               }
-               else {
-                  span->array->rgba = (void *)
-                     span->array->attribs[FRAG_ATTRIB_COL0];
-               }
-            }
-
-            if (!multiFragOutputs && numBuffers > 1) {
-               /* save colors for second, third renderbuffer writes */
-               memcpy(rgbaSave, span->array->rgba,
-                      4 * span->end * sizeof(GLchan));
-            }
-
-            ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB ||
-		   rb->_BaseFormat == GL_ALPHA);
-
-            if (ctx->Color._LogicOpEnabled) {
-               _swrast_logicop_rgba_span(ctx, rb, span);
-            }
-            else if ((ctx->Color.BlendEnabled >> buf) & 1) {
-               _swrast_blend_span(ctx, rb, span);
-            }
-
-            if (colorMask[buf] != 0xffffffff) {
-               _swrast_mask_rgba_span(ctx, rb, span, buf);
-            }
-
-            if (span->arrayMask & SPAN_XY) {
-               /* array of pixel coords */
-               ASSERT(rb->PutValues);
-               rb->PutValues(ctx, rb, span->end,
-                             span->array->x, span->array->y,
-                             span->array->rgba, span->array->mask);
-            }
-            else {
-               /* horizontal run of pixels */
-               ASSERT(rb->PutRow);
-               rb->PutRow(ctx, rb, span->end, span->x, span->y,
-                          span->array->rgba,
-                          span->writeAll ? NULL: span->array->mask);
-            }
-
-            if (!multiFragOutputs && numBuffers > 1) {
-               /* restore original span values */
-               memcpy(span->array->rgba, rgbaSave,
-                      4 * span->end * sizeof(GLchan));
-            }
-
-         } /* if rb */
-      } /* for buf */
-   }
-
-end:
-   /* restore these values before returning */
-   span->interpMask = origInterpMask;
-   span->arrayMask = origArrayMask;
-   span->arrayAttribs = origArrayAttribs;
-   span->array->ChanType = origChanType;
-   span->array->rgba = origRgba;
-}
-
-
-/**
- * Read RGBA pixels from a renderbuffer.  Clipping will be done to prevent
- * reading ouside the buffer's boundaries.
- * \param dstType  datatype for returned colors
- * \param rgba  the returned colors
- */
-void
-_swrast_read_rgba_span( struct gl_context *ctx, struct gl_renderbuffer *rb,
-                        GLuint n, GLint x, GLint y, GLenum dstType,
-                        GLvoid *rgba)
-{
-   const GLint bufWidth = (GLint) rb->Width;
-   const GLint bufHeight = (GLint) rb->Height;
-
-   if (y < 0 || y >= bufHeight || x + (GLint) n < 0 || x >= bufWidth) {
-      /* completely above, below, or right */
-      /* XXX maybe leave rgba values undefined? */
-      memset(rgba, 0, 4 * n * sizeof(GLchan));
-   }
-   else {
-      GLint skip, length;
-      if (x < 0) {
-         /* left edge clipping */
-         skip = -x;
-         length = (GLint) n - skip;
-         if (length < 0) {
-            /* completely left of window */
-            return;
-         }
-         if (length > bufWidth) {
-            length = bufWidth;
-         }
-      }
-      else if ((GLint) (x + n) > bufWidth) {
-         /* right edge clipping */
-         skip = 0;
-         length = bufWidth - x;
-         if (length < 0) {
-            /* completely to right of window */
-            return;
-         }
-      }
-      else {
-         /* no clipping */
-         skip = 0;
-         length = (GLint) n;
-      }
-
-      ASSERT(rb);
-      ASSERT(rb->GetRow);
-      ASSERT(rb->_BaseFormat == GL_RGB || rb->_BaseFormat == GL_RGBA ||
-	     rb->_BaseFormat == GL_ALPHA);
-
-      if (rb->DataType == dstType) {
-         rb->GetRow(ctx, rb, length, x + skip, y,
-                    (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(rb->DataType));
-      }
-      else {
-         GLuint temp[MAX_WIDTH * 4];
-         rb->GetRow(ctx, rb, length, x + skip, y, temp);
-         _mesa_convert_colors(rb->DataType, temp,
-                   dstType, (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(dstType),
-                   length, NULL);
-      }
-   }
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::GetValues() which does clipping to avoid
- * reading values outside the buffer bounds.
- * We can use this for reading any format/type of renderbuffer.
- * \param valueSize is the size in bytes of each value (pixel) put into the
- *                  values array.
- */
-void
-_swrast_get_values(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                   GLuint count, const GLint x[], const GLint y[],
-                   void *values, GLuint valueSize)
-{
-   GLuint i, inCount = 0, inStart = 0;
-
-   for (i = 0; i < count; i++) {
-      if (x[i] >= 0 && y[i] >= 0 &&
-	  x[i] < (GLint) rb->Width && y[i] < (GLint) rb->Height) {
-         /* inside */
-         if (inCount == 0)
-            inStart = i;
-         inCount++;
-      }
-      else {
-         if (inCount > 0) {
-            /* read [inStart, inStart + inCount) */
-            rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
-                          (GLubyte *) values + inStart * valueSize);
-            inCount = 0;
-         }
-      }
-   }
-   if (inCount > 0) {
-      /* read last values */
-      rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
-                    (GLubyte *) values + inStart * valueSize);
-   }
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::PutRow() which does clipping.
- * \param valueSize  size of each value (pixel) in bytes
- */
-void
-_swrast_put_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                GLuint count, GLint x, GLint y,
-                const GLvoid *values, GLuint valueSize)
-{
-   GLint skip = 0;
-
-   if (y < 0 || y >= (GLint) rb->Height)
-      return; /* above or below */
-
-   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
-      return; /* entirely left or right */
-
-   if ((GLint) (x + count) > (GLint) rb->Width) {
-      /* right clip */
-      GLint clip = x + count - rb->Width;
-      count -= clip;
-   }
-
-   if (x < 0) {
-      /* left clip */
-      skip = -x;
-      x = 0;
-      count -= skip;
-   }
-
-   rb->PutRow(ctx, rb, count, x, y,
-              (const GLubyte *) values + skip * valueSize, NULL);
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::GetRow() which does clipping.
- * \param valueSize  size of each value (pixel) in bytes
- */
-void
-_swrast_get_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                GLuint count, GLint x, GLint y,
-                GLvoid *values, GLuint valueSize)
-{
-   GLint skip = 0;
-
-   if (y < 0 || y >= (GLint) rb->Height)
-      return; /* above or below */
-
-   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
-      return; /* entirely left or right */
-
-   if (x + count > rb->Width) {
-      /* right clip */
-      GLint clip = x + count - rb->Width;
-      count -= clip;
-   }
-
-   if (x < 0) {
-      /* left clip */
-      skip = -x;
-      x = 0;
-      count -= skip;
-   }
-
-   rb->GetRow(ctx, rb, count, x, y, (GLubyte *) values + skip * valueSize);
-}
-
-
-/**
- * Get RGBA pixels from the given renderbuffer.
- * Used by blending, logicop and masking functions.
- * \return pointer to the colors we read.
- */
-void *
-_swrast_get_dest_rgba(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                      SWspan *span)
-{
-   const GLuint pixelSize = RGBA_PIXEL_SIZE(span->array->ChanType);
-   void *rbPixels;
-
-   /* Point rbPixels to a temporary space */
-   rbPixels = span->array->attribs[FRAG_ATTRIB_MAX - 1];
-
-   /* Get destination values from renderbuffer */
-   if (span->arrayMask & SPAN_XY) {
-      _swrast_get_values(ctx, rb, span->end, span->array->x, span->array->y,
-                         rbPixels, pixelSize);
-   }
-   else {
-      _swrast_get_row(ctx, rb, span->end, span->x, span->y,
-                      rbPixels, pixelSize);
-   }
-
-   return rbPixels;
-}
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.5
+ *
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 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, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/**
+ * \file swrast/s_span.c
+ * \brief Span processing functions used by all rasterization functions.
+ * This is where all the per-fragment tests are performed
+ * \author Brian Paul
+ */
+
+#include "main/glheader.h"
+#include "main/colormac.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/image.h"
+
+#include "s_atifragshader.h"
+#include "s_alpha.h"
+#include "s_blend.h"
+#include "s_context.h"
+#include "s_depth.h"
+#include "s_fog.h"
+#include "s_logic.h"
+#include "s_masking.h"
+#include "s_fragprog.h"
+#include "s_span.h"
+#include "s_stencil.h"
+#include "s_texcombine.h"
+
+
+/**
+ * Set default fragment attributes for the span using the
+ * current raster values.  Used prior to glDraw/CopyPixels
+ * and glBitmap.
+ */
+void
+_swrast_span_default_attribs(struct gl_context *ctx, SWspan *span)
+{
+   GLchan r, g, b, a;
+   /* Z*/
+   {
+      const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;
+      if (ctx->DrawBuffer->Visual.depthBits <= 16)
+         span->z = FloatToFixed(ctx->Current.RasterPos[2] * depthMax + 0.5F);
+      else {
+         GLfloat tmpf = ctx->Current.RasterPos[2] * depthMax; 
+         tmpf = MIN2(tmpf, depthMax);
+         span->z = (GLint)tmpf;
+      }
+      span->zStep = 0;
+      span->interpMask |= SPAN_Z;
+   }
+
+   /* W (for perspective correction) */
+   span->attrStart[FRAG_ATTRIB_WPOS][3] = 1.0;
+   span->attrStepX[FRAG_ATTRIB_WPOS][3] = 0.0;
+   span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0;
+
+   /* primary color, or color index */
+   UNCLAMPED_FLOAT_TO_CHAN(r, ctx->Current.RasterColor[0]);
+   UNCLAMPED_FLOAT_TO_CHAN(g, ctx->Current.RasterColor[1]);
+   UNCLAMPED_FLOAT_TO_CHAN(b, ctx->Current.RasterColor[2]);
+   UNCLAMPED_FLOAT_TO_CHAN(a, ctx->Current.RasterColor[3]);
+#if CHAN_TYPE == GL_FLOAT
+   span->red = r;
+   span->green = g;
+   span->blue = b;
+   span->alpha = a;
+#else
+   span->red   = IntToFixed(r);
+   span->green = IntToFixed(g);
+   span->blue  = IntToFixed(b);
+   span->alpha = IntToFixed(a);
+#endif
+   span->redStep = 0;
+   span->greenStep = 0;
+   span->blueStep = 0;
+   span->alphaStep = 0;
+   span->interpMask |= SPAN_RGBA;
+
+   COPY_4V(span->attrStart[FRAG_ATTRIB_COL0], ctx->Current.RasterColor);
+   ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
+   ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
+
+   /* Secondary color */
+   if (ctx->Light.Enabled || ctx->Fog.ColorSumEnabled)
+   {
+      COPY_4V(span->attrStart[FRAG_ATTRIB_COL1], ctx->Current.RasterSecondaryColor);
+      ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
+      ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
+   }
+
+   /* fog */
+   {
+      const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+      GLfloat fogVal; /* a coord or a blend factor */
+      if (swrast->_PreferPixelFog) {
+         /* fog blend factors will be computed from fog coordinates per pixel */
+         fogVal = ctx->Current.RasterDistance;
+      }
+      else {
+         /* fog blend factor should be computed from fogcoord now */
+         fogVal = _swrast_z_to_fogfactor(ctx, ctx->Current.RasterDistance);
+      }
+      span->attrStart[FRAG_ATTRIB_FOGC][0] = fogVal;
+      span->attrStepX[FRAG_ATTRIB_FOGC][0] = 0.0;
+      span->attrStepY[FRAG_ATTRIB_FOGC][0] = 0.0;
+   }
+
+   /* texcoords */
+   {
+      GLuint i;
+      for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
+         const GLuint attr = FRAG_ATTRIB_TEX0 + i;
+         const GLfloat *tc = ctx->Current.RasterTexCoords[i];
+         if (ctx->FragmentProgram._Current || ctx->ATIFragmentShader._Enabled) {
+            COPY_4V(span->attrStart[attr], tc);
+         }
+         else if (tc[3] > 0.0F) {
+            /* use (s/q, t/q, r/q, 1) */
+            span->attrStart[attr][0] = tc[0] / tc[3];
+            span->attrStart[attr][1] = tc[1] / tc[3];
+            span->attrStart[attr][2] = tc[2] / tc[3];
+            span->attrStart[attr][3] = 1.0;
+         }
+         else {
+            ASSIGN_4V(span->attrStart[attr], 0.0F, 0.0F, 0.0F, 1.0F);
+         }
+         ASSIGN_4V(span->attrStepX[attr], 0.0F, 0.0F, 0.0F, 0.0F);
+         ASSIGN_4V(span->attrStepY[attr], 0.0F, 0.0F, 0.0F, 0.0F);
+      }
+   }
+}
+
+
+/**
+ * Interpolate the active attributes (and'd with attrMask) to
+ * fill in span->array->attribs[].
+ * Perspective correction will be done.  The point/line/triangle function
+ * should have computed attrStart/Step values for FRAG_ATTRIB_WPOS[3]!
+ */
+static INLINE void
+interpolate_active_attribs(struct gl_context *ctx, SWspan *span, GLbitfield attrMask)
+{
+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+
+   /*
+    * Don't overwrite existing array values, such as colors that may have
+    * been produced by glDraw/CopyPixels.
+    */
+   attrMask &= ~span->arrayAttribs;
+
+   ATTRIB_LOOP_BEGIN
+      if (attrMask & (1 << attr)) {
+         const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
+         GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];
+         const GLfloat dv0dx = span->attrStepX[attr][0];
+         const GLfloat dv1dx = span->attrStepX[attr][1];
+         const GLfloat dv2dx = span->attrStepX[attr][2];
+         const GLfloat dv3dx = span->attrStepX[attr][3];
+         GLfloat v0 = span->attrStart[attr][0] + span->leftClip * dv0dx;
+         GLfloat v1 = span->attrStart[attr][1] + span->leftClip * dv1dx;
+         GLfloat v2 = span->attrStart[attr][2] + span->leftClip * dv2dx;
+         GLfloat v3 = span->attrStart[attr][3] + span->leftClip * dv3dx;
+         GLuint k;
+         for (k = 0; k < span->end; k++) {
+            const GLfloat invW = 1.0f / w;
+            span->array->attribs[attr][k][0] = v0 * invW;
+            span->array->attribs[attr][k][1] = v1 * invW;
+            span->array->attribs[attr][k][2] = v2 * invW;
+            span->array->attribs[attr][k][3] = v3 * invW;
+            v0 += dv0dx;
+            v1 += dv1dx;
+            v2 += dv2dx;
+            v3 += dv3dx;
+            w += dwdx;
+         }
+         ASSERT((span->arrayAttribs & (1 << attr)) == 0);
+         span->arrayAttribs |= (1 << attr);
+      }
+   ATTRIB_LOOP_END
+}
+
+
+/**
+ * Interpolate primary colors to fill in the span->array->rgba8 (or rgb16)
+ * color array.
+ */
+static INLINE void
+interpolate_int_colors(struct gl_context *ctx, SWspan *span)
+{
+   const GLuint n = span->end;
+   GLuint i;
+
+#if CHAN_BITS != 32
+   ASSERT(!(span->arrayMask & SPAN_RGBA));
+#endif
+
+   switch (span->array->ChanType) {
+#if CHAN_BITS != 32
+   case GL_UNSIGNED_BYTE:
+      {
+         GLubyte (*rgba)[4] = span->array->rgba8;
+         if (span->interpMask & SPAN_FLAT) {
+            GLubyte color[4];
+            color[RCOMP] = FixedToInt(span->red);
+            color[GCOMP] = FixedToInt(span->green);
+            color[BCOMP] = FixedToInt(span->blue);
+            color[ACOMP] = FixedToInt(span->alpha);
+            for (i = 0; i < n; i++) {
+               COPY_4UBV(rgba[i], color);
+            }
+         }
+         else {
+            GLfixed r = span->red;
+            GLfixed g = span->green;
+            GLfixed b = span->blue;
+            GLfixed a = span->alpha;
+            GLint dr = span->redStep;
+            GLint dg = span->greenStep;
+            GLint db = span->blueStep;
+            GLint da = span->alphaStep;
+            for (i = 0; i < n; i++) {
+               rgba[i][RCOMP] = FixedToChan(r);
+               rgba[i][GCOMP] = FixedToChan(g);
+               rgba[i][BCOMP] = FixedToChan(b);
+               rgba[i][ACOMP] = FixedToChan(a);
+               r += dr;
+               g += dg;
+               b += db;
+               a += da;
+            }
+         }
+      }
+      break;
+   case GL_UNSIGNED_SHORT:
+      {
+         GLushort (*rgba)[4] = span->array->rgba16;
+         if (span->interpMask & SPAN_FLAT) {
+            GLushort color[4];
+            color[RCOMP] = FixedToInt(span->red);
+            color[GCOMP] = FixedToInt(span->green);
+            color[BCOMP] = FixedToInt(span->blue);
+            color[ACOMP] = FixedToInt(span->alpha);
+            for (i = 0; i < n; i++) {
+               COPY_4V(rgba[i], color);
+            }
+         }
+         else {
+            GLushort (*rgba)[4] = span->array->rgba16;
+            GLfixed r, g, b, a;
+            GLint dr, dg, db, da;
+            r = span->red;
+            g = span->green;
+            b = span->blue;
+            a = span->alpha;
+            dr = span->redStep;
+            dg = span->greenStep;
+            db = span->blueStep;
+            da = span->alphaStep;
+            for (i = 0; i < n; i++) {
+               rgba[i][RCOMP] = FixedToChan(r);
+               rgba[i][GCOMP] = FixedToChan(g);
+               rgba[i][BCOMP] = FixedToChan(b);
+               rgba[i][ACOMP] = FixedToChan(a);
+               r += dr;
+               g += dg;
+               b += db;
+               a += da;
+            }
+         }
+      }
+      break;
+#endif
+   case GL_FLOAT:
+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
+      break;
+   default:
+      _mesa_problem(ctx, "bad datatype 0x%x in interpolate_int_colors",
+                    span->array->ChanType);
+   }
+   span->arrayMask |= SPAN_RGBA;
+}
+
+
+/**
+ * Populate the FRAG_ATTRIB_COL0 array.
+ */
+static INLINE void
+interpolate_float_colors(SWspan *span)
+{
+   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
+   const GLuint n = span->end;
+   GLuint i;
+
+   assert(!(span->arrayAttribs & FRAG_BIT_COL0));
+
+   if (span->arrayMask & SPAN_RGBA) {
+      /* convert array of int colors */
+      for (i = 0; i < n; i++) {
+         col0[i][0] = UBYTE_TO_FLOAT(span->array->rgba8[i][0]);
+         col0[i][1] = UBYTE_TO_FLOAT(span->array->rgba8[i][1]);
+         col0[i][2] = UBYTE_TO_FLOAT(span->array->rgba8[i][2]);
+         col0[i][3] = UBYTE_TO_FLOAT(span->array->rgba8[i][3]);
+      }
+   }
+   else {
+      /* interpolate red/green/blue/alpha to get float colors */
+      ASSERT(span->interpMask & SPAN_RGBA);
+      if (span->interpMask & SPAN_FLAT) {
+         GLfloat r = FixedToFloat(span->red);
+         GLfloat g = FixedToFloat(span->green);
+         GLfloat b = FixedToFloat(span->blue);
+         GLfloat a = FixedToFloat(span->alpha);
+         for (i = 0; i < n; i++) {
+            ASSIGN_4V(col0[i], r, g, b, a);
+         }
+      }
+      else {
+         GLfloat r = FixedToFloat(span->red);
+         GLfloat g = FixedToFloat(span->green);
+         GLfloat b = FixedToFloat(span->blue);
+         GLfloat a = FixedToFloat(span->alpha);
+         GLfloat dr = FixedToFloat(span->redStep);
+         GLfloat dg = FixedToFloat(span->greenStep);
+         GLfloat db = FixedToFloat(span->blueStep);
+         GLfloat da = FixedToFloat(span->alphaStep);
+         for (i = 0; i < n; i++) {
+            col0[i][0] = r;
+            col0[i][1] = g;
+            col0[i][2] = b;
+            col0[i][3] = a;
+            r += dr;
+            g += dg;
+            b += db;
+            a += da;
+         }
+      }
+   }
+
+   span->arrayAttribs |= FRAG_BIT_COL0;
+   span->array->ChanType = GL_FLOAT;
+}
+
+
+
+/**
+ * Fill in the span.zArray array from the span->z, zStep values.
+ */
+void
+_swrast_span_interpolate_z( const struct gl_context *ctx, SWspan *span )
+{
+   const GLuint n = span->end;
+   GLuint i;
+
+   ASSERT(!(span->arrayMask & SPAN_Z));
+
+   if (ctx->DrawBuffer->Visual.depthBits <= 16) {
+      GLfixed zval = span->z;
+      GLuint *z = span->array->z; 
+      for (i = 0; i < n; i++) {
+         z[i] = FixedToInt(zval);
+         zval += span->zStep;
+      }
+   }
+   else {
+      /* Deep Z buffer, no fixed->int shift */
+      GLuint zval = span->z;
+      GLuint *z = span->array->z;
+      for (i = 0; i < n; i++) {
+         z[i] = zval;
+         zval += span->zStep;
+      }
+   }
+   span->interpMask &= ~SPAN_Z;
+   span->arrayMask |= SPAN_Z;
+}
+
+
+/**
+ * Compute mipmap LOD from partial derivatives.
+ * This the ideal solution, as given in the OpenGL spec.
+ */
+GLfloat
+_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
+                       GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
+                       GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
+{
+   GLfloat dudx = texW * ((s + dsdx) / (q + dqdx) - s * invQ);
+   GLfloat dvdx = texH * ((t + dtdx) / (q + dqdx) - t * invQ);
+   GLfloat dudy = texW * ((s + dsdy) / (q + dqdy) - s * invQ);
+   GLfloat dvdy = texH * ((t + dtdy) / (q + dqdy) - t * invQ);
+   GLfloat x = SQRTF(dudx * dudx + dvdx * dvdx);
+   GLfloat y = SQRTF(dudy * dudy + dvdy * dvdy);
+   GLfloat rho = MAX2(x, y);
+   GLfloat lambda = LOG2(rho);
+   return lambda;
+}
+
+
+/**
+ * Compute mipmap LOD from partial derivatives.
+ * This is a faster approximation than above function.
+ */
+#if 0
+GLfloat
+_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
+                     GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
+                     GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
+{
+   GLfloat dsdx2 = (s + dsdx) / (q + dqdx) - s * invQ;
+   GLfloat dtdx2 = (t + dtdx) / (q + dqdx) - t * invQ;
+   GLfloat dsdy2 = (s + dsdy) / (q + dqdy) - s * invQ;
+   GLfloat dtdy2 = (t + dtdy) / (q + dqdy) - t * invQ;
+   GLfloat maxU, maxV, rho, lambda;
+   dsdx2 = FABSF(dsdx2);
+   dsdy2 = FABSF(dsdy2);
+   dtdx2 = FABSF(dtdx2);
+   dtdy2 = FABSF(dtdy2);
+   maxU = MAX2(dsdx2, dsdy2) * texW;
+   maxV = MAX2(dtdx2, dtdy2) * texH;
+   rho = MAX2(maxU, maxV);
+   lambda = LOG2(rho);
+   return lambda;
+}
+#endif
+
+
+/**
+ * Fill in the span.array->attrib[FRAG_ATTRIB_TEXn] arrays from the
+ * using the attrStart/Step values.
+ *
+ * This function only used during fixed-function fragment processing.
+ *
+ * Note: in the places where we divide by Q (or mult by invQ) we're
+ * really doing two things: perspective correction and texcoord
+ * projection.  Remember, for texcoord (s,t,r,q) we need to index
+ * texels with (s/q, t/q, r/q).
+ */
+static void
+interpolate_texcoords(struct gl_context *ctx, SWspan *span)
+{
+   const GLuint maxUnit
+      = (ctx->Texture._EnabledCoordUnits > 1) ? ctx->Const.MaxTextureUnits : 1;
+   GLuint u;
+
+   /* XXX CoordUnits vs. ImageUnits */
+   for (u = 0; u < maxUnit; u++) {
+      if (ctx->Texture._EnabledCoordUnits & (1 << u)) {
+         const GLuint attr = FRAG_ATTRIB_TEX0 + u;
+         const struct gl_texture_object *obj = ctx->Texture.Unit[u]._Current;
+         GLfloat texW, texH;
+         GLboolean needLambda;
+         GLfloat (*texcoord)[4] = span->array->attribs[attr];
+         GLfloat *lambda = span->array->lambda[u];
+         const GLfloat dsdx = span->attrStepX[attr][0];
+         const GLfloat dsdy = span->attrStepY[attr][0];
+         const GLfloat dtdx = span->attrStepX[attr][1];
+         const GLfloat dtdy = span->attrStepY[attr][1];
+         const GLfloat drdx = span->attrStepX[attr][2];
+         const GLfloat dqdx = span->attrStepX[attr][3];
+         const GLfloat dqdy = span->attrStepY[attr][3];
+         GLfloat s = span->attrStart[attr][0] + span->leftClip * dsdx;
+         GLfloat t = span->attrStart[attr][1] + span->leftClip * dtdx;
+         GLfloat r = span->attrStart[attr][2] + span->leftClip * drdx;
+         GLfloat q = span->attrStart[attr][3] + span->leftClip * dqdx;
+
+         if (obj) {
+            const struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
+            needLambda = (obj->Sampler.MinFilter != obj->Sampler.MagFilter)
+               || ctx->FragmentProgram._Current;
+            texW = img->WidthScale;
+            texH = img->HeightScale;
+         }
+         else {
+            /* using a fragment program */
+            texW = 1.0;
+            texH = 1.0;
+            needLambda = GL_FALSE;
+         }
+
+         if (needLambda) {
+            GLuint i;
+            if (ctx->FragmentProgram._Current
+                || ctx->ATIFragmentShader._Enabled) {
+               /* do perspective correction but don't divide s, t, r by q */
+               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
+               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
+               for (i = 0; i < span->end; i++) {
+                  const GLfloat invW = 1.0F / w;
+                  texcoord[i][0] = s * invW;
+                  texcoord[i][1] = t * invW;
+                  texcoord[i][2] = r * invW;
+                  texcoord[i][3] = q * invW;
+                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
+                                                     dqdx, dqdy, texW, texH,
+                                                     s, t, q, invW);
+                  s += dsdx;
+                  t += dtdx;
+                  r += drdx;
+                  q += dqdx;
+                  w += dwdx;
+               }
+            }
+            else {
+               for (i = 0; i < span->end; i++) {
+                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
+                  texcoord[i][0] = s * invQ;
+                  texcoord[i][1] = t * invQ;
+                  texcoord[i][2] = r * invQ;
+                  texcoord[i][3] = q;
+                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
+                                                     dqdx, dqdy, texW, texH,
+                                                     s, t, q, invQ);
+                  s += dsdx;
+                  t += dtdx;
+                  r += drdx;
+                  q += dqdx;
+               }
+            }
+            span->arrayMask |= SPAN_LAMBDA;
+         }
+         else {
+            GLuint i;
+            if (ctx->FragmentProgram._Current ||
+                ctx->ATIFragmentShader._Enabled) {
+               /* do perspective correction but don't divide s, t, r by q */
+               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
+               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
+               for (i = 0; i < span->end; i++) {
+                  const GLfloat invW = 1.0F / w;
+                  texcoord[i][0] = s * invW;
+                  texcoord[i][1] = t * invW;
+                  texcoord[i][2] = r * invW;
+                  texcoord[i][3] = q * invW;
+                  lambda[i] = 0.0;
+                  s += dsdx;
+                  t += dtdx;
+                  r += drdx;
+                  q += dqdx;
+                  w += dwdx;
+               }
+            }
+            else if (dqdx == 0.0F) {
+               /* Ortho projection or polygon's parallel to window X axis */
+               const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
+               for (i = 0; i < span->end; i++) {
+                  texcoord[i][0] = s * invQ;
+                  texcoord[i][1] = t * invQ;
+                  texcoord[i][2] = r * invQ;
+                  texcoord[i][3] = q;
+                  lambda[i] = 0.0;
+                  s += dsdx;
+                  t += dtdx;
+                  r += drdx;
+               }
+            }
+            else {
+               for (i = 0; i < span->end; i++) {
+                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
+                  texcoord[i][0] = s * invQ;
+                  texcoord[i][1] = t * invQ;
+                  texcoord[i][2] = r * invQ;
+                  texcoord[i][3] = q;
+                  lambda[i] = 0.0;
+                  s += dsdx;
+                  t += dtdx;
+                  r += drdx;
+                  q += dqdx;
+               }
+            }
+         } /* lambda */
+      } /* if */
+   } /* for */
+}
+
+
+/**
+ * Fill in the arrays->attribs[FRAG_ATTRIB_WPOS] array.
+ */
+static INLINE void
+interpolate_wpos(struct gl_context *ctx, SWspan *span)
+{
+   GLfloat (*wpos)[4] = span->array->attribs[FRAG_ATTRIB_WPOS];
+   GLuint i;
+   const GLfloat zScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
+   GLfloat w, dw;
+
+   if (span->arrayMask & SPAN_XY) {
+      for (i = 0; i < span->end; i++) {
+         wpos[i][0] = (GLfloat) span->array->x[i];
+         wpos[i][1] = (GLfloat) span->array->y[i];
+      }
+   }
+   else {
+      for (i = 0; i < span->end; i++) {
+         wpos[i][0] = (GLfloat) span->x + i;
+         wpos[i][1] = (GLfloat) span->y;
+      }
+   }
+
+   dw = span->attrStepX[FRAG_ATTRIB_WPOS][3];
+   w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dw;
+   for (i = 0; i < span->end; i++) {
+      wpos[i][2] = (GLfloat) span->array->z[i] * zScale;
+      wpos[i][3] = w;
+      w += dw;
+   }
+}
+
+
+/**
+ * Apply the current polygon stipple pattern to a span of pixels.
+ */
+static INLINE void
+stipple_polygon_span(struct gl_context *ctx, SWspan *span)
+{
+   GLubyte *mask = span->array->mask;
+
+   ASSERT(ctx->Polygon.StippleFlag);
+
+   if (span->arrayMask & SPAN_XY) {
+      /* arrays of x/y pixel coords */
+      GLuint i;
+      for (i = 0; i < span->end; i++) {
+         const GLint col = span->array->x[i] % 32;
+         const GLint row = span->array->y[i] % 32;
+         const GLuint stipple = ctx->PolygonStipple[row];
+         if (((1 << col) & stipple) == 0) {
+            mask[i] = 0;
+         }
+      }
+   }
+   else {
+      /* horizontal span of pixels */
+      const GLuint highBit = 1 << 31;
+      const GLuint stipple = ctx->PolygonStipple[span->y % 32];
+      GLuint i, m = highBit >> (GLuint) (span->x % 32);
+      for (i = 0; i < span->end; i++) {
+         if ((m & stipple) == 0) {
+            mask[i] = 0;
+         }
+         m = m >> 1;
+         if (m == 0) {
+            m = highBit;
+         }
+      }
+   }
+   span->writeAll = GL_FALSE;
+}
+
+
+/**
+ * Clip a pixel span to the current buffer/window boundaries:
+ * DrawBuffer->_Xmin, _Xmax, _Ymin, _Ymax.  This will accomplish
+ * window clipping and scissoring.
+ * Return:   GL_TRUE   some pixels still visible
+ *           GL_FALSE  nothing visible
+ */
+static INLINE GLuint
+clip_span( struct gl_context *ctx, SWspan *span )
+{
+   const GLint xmin = ctx->DrawBuffer->_Xmin;
+   const GLint xmax = ctx->DrawBuffer->_Xmax;
+   const GLint ymin = ctx->DrawBuffer->_Ymin;
+   const GLint ymax = ctx->DrawBuffer->_Ymax;
+
+   span->leftClip = 0;
+
+   if (span->arrayMask & SPAN_XY) {
+      /* arrays of x/y pixel coords */
+      const GLint *x = span->array->x;
+      const GLint *y = span->array->y;
+      const GLint n = span->end;
+      GLubyte *mask = span->array->mask;
+      GLint i;
+      if (span->arrayMask & SPAN_MASK) {
+         /* note: using & intead of && to reduce branches */
+         for (i = 0; i < n; i++) {
+            mask[i] &= (x[i] >= xmin) & (x[i] < xmax)
+                     & (y[i] >= ymin) & (y[i] < ymax);
+         }
+      }
+      else {
+         /* note: using & intead of && to reduce branches */
+         for (i = 0; i < n; i++) {
+            mask[i] = (x[i] >= xmin) & (x[i] < xmax)
+                    & (y[i] >= ymin) & (y[i] < ymax);
+         }
+      }
+      return GL_TRUE;  /* some pixels visible */
+   }
+   else {
+      /* horizontal span of pixels */
+      const GLint x = span->x;
+      const GLint y = span->y;
+      GLint n = span->end;
+
+      /* Trivial rejection tests */
+      if (y < ymin || y >= ymax || x + n <= xmin || x >= xmax) {
+         span->end = 0;
+         return GL_FALSE;  /* all pixels clipped */
+      }
+
+      /* Clip to right */
+      if (x + n > xmax) {
+         ASSERT(x < xmax);
+         n = span->end = xmax - x;
+      }
+
+      /* Clip to the left */
+      if (x < xmin) {
+         const GLint leftClip = xmin - x;
+         GLuint i;
+
+         ASSERT(leftClip > 0);
+         ASSERT(x + n > xmin);
+
+         /* Clip 'leftClip' pixels from the left side.
+          * The span->leftClip field will be applied when we interpolate
+          * fragment attributes.
+          * For arrays of values, shift them left.
+          */
+         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
+            if (span->interpMask & (1 << i)) {
+               GLuint j;
+               for (j = 0; j < 4; j++) {
+                  span->attrStart[i][j] += leftClip * span->attrStepX[i][j];
+               }
+            }
+         }
+
+         span->red += leftClip * span->redStep;
+         span->green += leftClip * span->greenStep;
+         span->blue += leftClip * span->blueStep;
+         span->alpha += leftClip * span->alphaStep;
+         span->index += leftClip * span->indexStep;
+         span->z += leftClip * span->zStep;
+         span->intTex[0] += leftClip * span->intTexStep[0];
+         span->intTex[1] += leftClip * span->intTexStep[1];
+
+#define SHIFT_ARRAY(ARRAY, SHIFT, LEN) \
+         memcpy(ARRAY, ARRAY + (SHIFT), (LEN) * sizeof(ARRAY[0]))
+
+         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
+            if (span->arrayAttribs & (1 << i)) {
+               /* shift array elements left by 'leftClip' */
+               SHIFT_ARRAY(span->array->attribs[i], leftClip, n - leftClip);
+            }
+         }
+
+         SHIFT_ARRAY(span->array->mask, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->rgba8, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->rgba16, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->x, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->y, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->z, leftClip, n - leftClip);
+         SHIFT_ARRAY(span->array->index, leftClip, n - leftClip);
+         for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
+            SHIFT_ARRAY(span->array->lambda[i], leftClip, n - leftClip);
+         }
+         SHIFT_ARRAY(span->array->coverage, leftClip, n - leftClip);
+
+#undef SHIFT_ARRAY
+
+         span->leftClip = leftClip;
+         span->x = xmin;
+         span->end -= leftClip;
+         span->writeAll = GL_FALSE;
+      }
+
+      ASSERT(span->x >= xmin);
+      ASSERT(span->x + span->end <= xmax);
+      ASSERT(span->y >= ymin);
+      ASSERT(span->y < ymax);
+
+      return GL_TRUE;  /* some pixels visible */
+   }
+}
+
+
+/**
+ * Add specular colors to primary colors.
+ * Only called during fixed-function operation.
+ * Result is float color array (FRAG_ATTRIB_COL0).
+ */
+static INLINE void
+add_specular(struct gl_context *ctx, SWspan *span)
+{
+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const GLubyte *mask = span->array->mask;
+   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
+   GLfloat (*col1)[4] = span->array->attribs[FRAG_ATTRIB_COL1];
+   GLuint i;
+
+   ASSERT(!ctx->FragmentProgram._Current);
+   ASSERT(span->arrayMask & SPAN_RGBA);
+   ASSERT(swrast->_ActiveAttribMask & FRAG_BIT_COL1);
+   (void) swrast; /* silence warning */
+
+   if (span->array->ChanType == GL_FLOAT) {
+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
+         interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
+      }
+   }
+   else {
+      /* need float colors */
+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
+         interpolate_float_colors(span);
+      }
+   }
+
+   if ((span->arrayAttribs & FRAG_BIT_COL1) == 0) {
+      /* XXX could avoid this and interpolate COL1 in the loop below */
+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL1);
+   }
+
+   ASSERT(span->arrayAttribs & FRAG_BIT_COL0);
+   ASSERT(span->arrayAttribs & FRAG_BIT_COL1);
+
+   for (i = 0; i < span->end; i++) {
+      if (mask[i]) {
+         col0[i][0] += col1[i][0];
+         col0[i][1] += col1[i][1];
+         col0[i][2] += col1[i][2];
+      }
+   }
+
+   span->array->ChanType = GL_FLOAT;
+}
+
+
+/**
+ * Apply antialiasing coverage value to alpha values.
+ */
+static INLINE void
+apply_aa_coverage(SWspan *span)
+{
+   const GLfloat *coverage = span->array->coverage;
+   GLuint i;
+   if (span->array->ChanType == GL_UNSIGNED_BYTE) {
+      GLubyte (*rgba)[4] = span->array->rgba8;
+      for (i = 0; i < span->end; i++) {
+         const GLfloat a = rgba[i][ACOMP] * coverage[i];
+         rgba[i][ACOMP] = (GLubyte) CLAMP(a, 0.0, 255.0);
+         ASSERT(coverage[i] >= 0.0);
+         ASSERT(coverage[i] <= 1.0);
+      }
+   }
+   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
+      GLushort (*rgba)[4] = span->array->rgba16;
+      for (i = 0; i < span->end; i++) {
+         const GLfloat a = rgba[i][ACOMP] * coverage[i];
+         rgba[i][ACOMP] = (GLushort) CLAMP(a, 0.0, 65535.0);
+      }
+   }
+   else {
+      GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
+      for (i = 0; i < span->end; i++) {
+         rgba[i][ACOMP] = rgba[i][ACOMP] * coverage[i];
+         /* clamp later */
+      }
+   }
+}
+
+
+/**
+ * Clamp span's float colors to [0,1]
+ */
+static INLINE void
+clamp_colors(SWspan *span)
+{
+   GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
+   GLuint i;
+   ASSERT(span->array->ChanType == GL_FLOAT);
+   for (i = 0; i < span->end; i++) {
+      rgba[i][RCOMP] = CLAMP(rgba[i][RCOMP], 0.0F, 1.0F);
+      rgba[i][GCOMP] = CLAMP(rgba[i][GCOMP], 0.0F, 1.0F);
+      rgba[i][BCOMP] = CLAMP(rgba[i][BCOMP], 0.0F, 1.0F);
+      rgba[i][ACOMP] = CLAMP(rgba[i][ACOMP], 0.0F, 1.0F);
+   }
+}
+
+
+/**
+ * Convert the span's color arrays to the given type.
+ * The only way 'output' can be greater than zero is when we have a fragment
+ * program that writes to gl_FragData[1] or higher.
+ * \param output  which fragment program color output is being processed
+ */
+static INLINE void
+convert_color_type(SWspan *span, GLenum newType, GLuint output)
+{
+   GLvoid *src, *dst;
+
+   if (output > 0 || span->array->ChanType == GL_FLOAT) {
+      src = span->array->attribs[FRAG_ATTRIB_COL0 + output];
+      span->array->ChanType = GL_FLOAT;
+   }
+   else if (span->array->ChanType == GL_UNSIGNED_BYTE) {
+      src = span->array->rgba8;
+   }
+   else {
+      ASSERT(span->array->ChanType == GL_UNSIGNED_SHORT);
+      src = span->array->rgba16;
+   }
+
+   if (newType == GL_UNSIGNED_BYTE) {
+      dst = span->array->rgba8;
+   }
+   else if (newType == GL_UNSIGNED_SHORT) {
+      dst = span->array->rgba16;
+   }
+   else {
+      dst = span->array->attribs[FRAG_ATTRIB_COL0];
+   }
+
+   _mesa_convert_colors(span->array->ChanType, src,
+                        newType, dst,
+                        span->end, span->array->mask);
+
+   span->array->ChanType = newType;
+   span->array->rgba = dst;
+}
+
+
+
+/**
+ * Apply fragment shader, fragment program or normal texturing to span.
+ */
+static INLINE void
+shade_texture_span(struct gl_context *ctx, SWspan *span)
+{
+   GLbitfield inputsRead;
+
+   /* Determine which fragment attributes are actually needed */
+   if (ctx->FragmentProgram._Current) {
+      inputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
+   }
+   else {
+      /* XXX we could be a bit smarter about this */
+      inputsRead = ~0;
+   }
+
+   if (ctx->FragmentProgram._Current ||
+       ctx->ATIFragmentShader._Enabled) {
+      /* programmable shading */
+      if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {
+         convert_color_type(span, GL_FLOAT, 0);
+      }
+      else {
+         span->array->rgba = (void *) span->array->attribs[FRAG_ATTRIB_COL0];
+      }
+
+      if (span->primitive != GL_POINT ||
+	  (span->interpMask & SPAN_RGBA) ||
+	  ctx->Point.PointSprite) {
+         /* for single-pixel points, we populated the arrays already */
+         interpolate_active_attribs(ctx, span, ~0);
+      }
+      span->array->ChanType = GL_FLOAT;
+
+      if (!(span->arrayMask & SPAN_Z))
+         _swrast_span_interpolate_z (ctx, span);
+
+#if 0
+      if (inputsRead & FRAG_BIT_WPOS)
+#else
+      /* XXX always interpolate wpos so that DDX/DDY work */
+#endif
+         interpolate_wpos(ctx, span);
+
+      /* Run fragment program/shader now */
+      if (ctx->FragmentProgram._Current) {
+         _swrast_exec_fragment_program(ctx, span);
+      }
+      else {
+         ASSERT(ctx->ATIFragmentShader._Enabled);
+         _swrast_exec_fragment_shader(ctx, span);
+      }
+   }
+   else if (ctx->Texture._EnabledCoordUnits) {
+      /* conventional texturing */
+
+#if CHAN_BITS == 32
+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
+         interpolate_int_colors(ctx, span);
+      }
+#else
+      if (!(span->arrayMask & SPAN_RGBA))
+         interpolate_int_colors(ctx, span);
+#endif
+      if ((span->arrayAttribs & FRAG_BITS_TEX_ANY) == 0x0)
+         interpolate_texcoords(ctx, span);
+
+      _swrast_texture_span(ctx, span);
+   }
+}
+
+
+
+/**
+ * Apply all the per-fragment operations to a span.
+ * This now includes texturing (_swrast_write_texture_span() is history).
+ * This function may modify any of the array values in the span.
+ * span->interpMask and span->arrayMask may be changed but will be restored
+ * to their original values before returning.
+ */
+void
+_swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)
+{
+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const GLuint *colorMask = (GLuint *) ctx->Color.ColorMask;
+   const GLbitfield origInterpMask = span->interpMask;
+   const GLbitfield origArrayMask = span->arrayMask;
+   const GLbitfield origArrayAttribs = span->arrayAttribs;
+   const GLenum origChanType = span->array->ChanType;
+   void * const origRgba = span->array->rgba;
+   const GLboolean shader = (ctx->FragmentProgram._Current
+                             || ctx->ATIFragmentShader._Enabled);
+   const GLboolean shaderOrTexture = shader || ctx->Texture._EnabledCoordUnits;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+
+   /*
+   printf("%s()  interp 0x%x  array 0x%x\n", __FUNCTION__,
+          span->interpMask, span->arrayMask);
+   */
+
+   ASSERT(span->primitive == GL_POINT ||
+          span->primitive == GL_LINE ||
+	  span->primitive == GL_POLYGON ||
+          span->primitive == GL_BITMAP);
+
+   /* Fragment write masks */
+   if (span->arrayMask & SPAN_MASK) {
+      /* mask was initialized by caller, probably glBitmap */
+      span->writeAll = GL_FALSE;
+   }
+   else {
+      memset(span->array->mask, 1, span->end);
+      span->writeAll = GL_TRUE;
+   }
+
+   /* Clip to window/scissor box */
+   if (!clip_span(ctx, span)) {
+      return;
+   }
+
+   ASSERT(span->end <= MAX_WIDTH);
+
+   /* Depth bounds test */
+   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
+      if (!_swrast_depth_bounds_test(ctx, span)) {
+         return;
+      }
+   }
+
+#ifdef DEBUG
+   /* Make sure all fragments are within window bounds */
+   if (span->arrayMask & SPAN_XY) {
+      /* array of pixel locations */
+      GLuint i;
+      for (i = 0; i < span->end; i++) {
+         if (span->array->mask[i]) {
+            assert(span->array->x[i] >= fb->_Xmin);
+            assert(span->array->x[i] < fb->_Xmax);
+            assert(span->array->y[i] >= fb->_Ymin);
+            assert(span->array->y[i] < fb->_Ymax);
+         }
+      }
+   }
+#endif
+
+   /* Polygon Stippling */
+   if (ctx->Polygon.StippleFlag && span->primitive == GL_POLYGON) {
+      stipple_polygon_span(ctx, span);
+   }
+
+   /* This is the normal place to compute the fragment color/Z
+    * from texturing or shading.
+    */
+   if (shaderOrTexture && !swrast->_DeferredTexture) {
+      shade_texture_span(ctx, span);
+   }
+
+   /* Do the alpha test */
+   if (ctx->Color.AlphaEnabled) {
+      if (!_swrast_alpha_test(ctx, span)) {
+         /* all fragments failed test */
+         goto end;
+      }
+   }
+
+   /* Stencil and Z testing */
+   if (ctx->Stencil._Enabled || ctx->Depth.Test) {
+      if (!(span->arrayMask & SPAN_Z))
+         _swrast_span_interpolate_z(ctx, span);
+
+      if (ctx->Transform.DepthClamp)
+	 _swrast_depth_clamp_span(ctx, span);
+
+      if (ctx->Stencil._Enabled) {
+         /* Combined Z/stencil tests */
+         if (!_swrast_stencil_and_ztest_span(ctx, span)) {
+            /* all fragments failed test */
+            goto end;
+         }
+      }
+      else if (fb->Visual.depthBits > 0) {
+         /* Just regular depth testing */
+         ASSERT(ctx->Depth.Test);
+         ASSERT(span->arrayMask & SPAN_Z);
+         if (!_swrast_depth_test_span(ctx, span)) {
+            /* all fragments failed test */
+            goto end;
+         }
+      }
+   }
+
+   if (ctx->Query.CurrentOcclusionObject) {
+      /* update count of 'passed' fragments */
+      struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;
+      GLuint i;
+      for (i = 0; i < span->end; i++)
+         q->Result += span->array->mask[i];
+   }
+
+   /* We had to wait until now to check for glColorMask(0,0,0,0) because of
+    * the occlusion test.
+    */
+   if (fb->_NumColorDrawBuffers == 1 && colorMask[0] == 0x0) {
+      /* no colors to write */
+      goto end;
+   }
+
+   /* If we were able to defer fragment color computation to now, there's
+    * a good chance that many fragments will have already been killed by
+    * Z/stencil testing.
+    */
+   if (shaderOrTexture && swrast->_DeferredTexture) {
+      shade_texture_span(ctx, span);
+   }
+
+#if CHAN_BITS == 32
+   if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
+   }
+#else
+   if ((span->arrayMask & SPAN_RGBA) == 0) {
+      interpolate_int_colors(ctx, span);
+   }
+#endif
+
+   ASSERT(span->arrayMask & SPAN_RGBA);
+
+   if (span->primitive == GL_BITMAP || !swrast->SpecularVertexAdd) {
+      /* Add primary and specular (diffuse + specular) colors */
+      if (!shader) {
+         if (ctx->Fog.ColorSumEnabled ||
+             (ctx->Light.Enabled &&
+              ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {
+            add_specular(ctx, span);
+         }
+      }
+   }
+
+   /* Fog */
+   if (swrast->_FogEnabled) {
+      _swrast_fog_rgba_span(ctx, span);
+   }
+
+   /* Antialias coverage application */
+   if (span->arrayMask & SPAN_COVERAGE) {
+      apply_aa_coverage(span);
+   }
+
+   /* Clamp color/alpha values over the range [0.0, 1.0] before storage */
+   if (ctx->Color.ClampFragmentColor == GL_TRUE &&
+       span->array->ChanType == GL_FLOAT) {
+      clamp_colors(span);
+   }
+
+   /*
+    * Write to renderbuffers.
+    * Depending on glDrawBuffer() state and the which color outputs are
+    * written by the fragment shader, we may either replicate one color to
+    * all renderbuffers or write a different color to each renderbuffer.
+    * multiFragOutputs=TRUE for the later case.
+    */
+   {
+      const GLuint numBuffers = fb->_NumColorDrawBuffers;
+      const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
+      const GLboolean multiFragOutputs = 
+         (fp && fp->Base.OutputsWritten >= (1 << FRAG_RESULT_DATA0));
+      GLuint buf;
+
+      for (buf = 0; buf < numBuffers; buf++) {
+         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];
+
+         /* color[fragOutput] will be written to buffer[buf] */
+
+         if (rb) {
+            GLchan rgbaSave[MAX_WIDTH][4];
+            const GLuint fragOutput = multiFragOutputs ? buf : 0;
+
+            /* set span->array->rgba to colors for render buffer's datatype */
+            if (rb->DataType != span->array->ChanType || fragOutput > 0) {
+               convert_color_type(span, rb->DataType, fragOutput);
+            }
+            else {
+               if (rb->DataType == GL_UNSIGNED_BYTE) {
+                  span->array->rgba = span->array->rgba8;
+               }
+               else if (rb->DataType == GL_UNSIGNED_SHORT) {
+                  span->array->rgba = (void *) span->array->rgba16;
+               }
+               else {
+                  span->array->rgba = (void *)
+                     span->array->attribs[FRAG_ATTRIB_COL0];
+               }
+            }
+
+            if (!multiFragOutputs && numBuffers > 1) {
+               /* save colors for second, third renderbuffer writes */
+               memcpy(rgbaSave, span->array->rgba,
+                      4 * span->end * sizeof(GLchan));
+            }
+
+            ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB ||
+		   rb->_BaseFormat == GL_ALPHA);
+
+            if (ctx->Color._LogicOpEnabled) {
+               _swrast_logicop_rgba_span(ctx, rb, span);
+            }
+            else if ((ctx->Color.BlendEnabled >> buf) & 1) {
+               _swrast_blend_span(ctx, rb, span);
+            }
+
+            if (colorMask[buf] != 0xffffffff) {
+               _swrast_mask_rgba_span(ctx, rb, span, buf);
+            }
+
+            if (span->arrayMask & SPAN_XY) {
+               /* array of pixel coords */
+               ASSERT(rb->PutValues);
+               rb->PutValues(ctx, rb, span->end,
+                             span->array->x, span->array->y,
+                             span->array->rgba, span->array->mask);
+            }
+            else {
+               /* horizontal run of pixels */
+               ASSERT(rb->PutRow);
+               rb->PutRow(ctx, rb, span->end, span->x, span->y,
+                          span->array->rgba,
+                          span->writeAll ? NULL: span->array->mask);
+            }
+
+            if (!multiFragOutputs && numBuffers > 1) {
+               /* restore original span values */
+               memcpy(span->array->rgba, rgbaSave,
+                      4 * span->end * sizeof(GLchan));
+            }
+
+         } /* if rb */
+      } /* for buf */
+   }
+
+end:
+   /* restore these values before returning */
+   span->interpMask = origInterpMask;
+   span->arrayMask = origArrayMask;
+   span->arrayAttribs = origArrayAttribs;
+   span->array->ChanType = origChanType;
+   span->array->rgba = origRgba;
+}
+
+
+/**
+ * Read RGBA pixels from a renderbuffer.  Clipping will be done to prevent
+ * reading ouside the buffer's boundaries.
+ * \param dstType  datatype for returned colors
+ * \param rgba  the returned colors
+ */
+void
+_swrast_read_rgba_span( struct gl_context *ctx, struct gl_renderbuffer *rb,
+                        GLuint n, GLint x, GLint y, GLenum dstType,
+                        GLvoid *rgba)
+{
+   const GLint bufWidth = (GLint) rb->Width;
+   const GLint bufHeight = (GLint) rb->Height;
+
+   if (y < 0 || y >= bufHeight || x + (GLint) n < 0 || x >= bufWidth) {
+      /* completely above, below, or right */
+      /* XXX maybe leave rgba values undefined? */
+      memset(rgba, 0, 4 * n * sizeof(GLchan));
+   }
+   else {
+      GLint skip, length;
+      if (x < 0) {
+         /* left edge clipping */
+         skip = -x;
+         length = (GLint) n - skip;
+         if (length < 0) {
+            /* completely left of window */
+            return;
+         }
+         if (length > bufWidth) {
+            length = bufWidth;
+         }
+      }
+      else if ((GLint) (x + n) > bufWidth) {
+         /* right edge clipping */
+         skip = 0;
+         length = bufWidth - x;
+         if (length < 0) {
+            /* completely to right of window */
+            return;
+         }
+      }
+      else {
+         /* no clipping */
+         skip = 0;
+         length = (GLint) n;
+      }
+
+      ASSERT(rb);
+      ASSERT(rb->GetRow);
+      ASSERT(rb->_BaseFormat == GL_RGB || rb->_BaseFormat == GL_RGBA ||
+	     rb->_BaseFormat == GL_ALPHA);
+
+      if (rb->DataType == dstType) {
+         rb->GetRow(ctx, rb, length, x + skip, y,
+                    (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(rb->DataType));
+      }
+      else {
+         GLuint temp[MAX_WIDTH * 4];
+         rb->GetRow(ctx, rb, length, x + skip, y, temp);
+         _mesa_convert_colors(rb->DataType, temp,
+                   dstType, (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(dstType),
+                   length, NULL);
+      }
+   }
+}
+
+
+/**
+ * Wrapper for gl_renderbuffer::GetValues() which does clipping to avoid
+ * reading values outside the buffer bounds.
+ * We can use this for reading any format/type of renderbuffer.
+ * \param valueSize is the size in bytes of each value (pixel) put into the
+ *                  values array.
+ */
+void
+_swrast_get_values(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                   GLuint count, const GLint x[], const GLint y[],
+                   void *values, GLuint valueSize)
+{
+   GLuint i, inCount = 0, inStart = 0;
+
+   for (i = 0; i < count; i++) {
+      if (x[i] >= 0 && y[i] >= 0 &&
+	  x[i] < (GLint) rb->Width && y[i] < (GLint) rb->Height) {
+         /* inside */
+         if (inCount == 0)
+            inStart = i;
+         inCount++;
+      }
+      else {
+         if (inCount > 0) {
+            /* read [inStart, inStart + inCount) */
+            rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
+                          (GLubyte *) values + inStart * valueSize);
+            inCount = 0;
+         }
+      }
+   }
+   if (inCount > 0) {
+      /* read last values */
+      rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
+                    (GLubyte *) values + inStart * valueSize);
+   }
+}
+
+
+/**
+ * Wrapper for gl_renderbuffer::PutRow() which does clipping.
+ * \param valueSize  size of each value (pixel) in bytes
+ */
+void
+_swrast_put_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                GLuint count, GLint x, GLint y,
+                const GLvoid *values, GLuint valueSize)
+{
+   GLint skip = 0;
+
+   if (y < 0 || y >= (GLint) rb->Height)
+      return; /* above or below */
+
+   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
+      return; /* entirely left or right */
+
+   if ((GLint) (x + count) > (GLint) rb->Width) {
+      /* right clip */
+      GLint clip = x + count - rb->Width;
+      count -= clip;
+   }
+
+   if (x < 0) {
+      /* left clip */
+      skip = -x;
+      x = 0;
+      count -= skip;
+   }
+
+   rb->PutRow(ctx, rb, count, x, y,
+              (const GLubyte *) values + skip * valueSize, NULL);
+}
+
+
+/**
+ * Wrapper for gl_renderbuffer::GetRow() which does clipping.
+ * \param valueSize  size of each value (pixel) in bytes
+ */
+void
+_swrast_get_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                GLuint count, GLint x, GLint y,
+                GLvoid *values, GLuint valueSize)
+{
+   GLint skip = 0;
+
+   if (y < 0 || y >= (GLint) rb->Height)
+      return; /* above or below */
+
+   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
+      return; /* entirely left or right */
+
+   if (x + count > rb->Width) {
+      /* right clip */
+      GLint clip = x + count - rb->Width;
+      count -= clip;
+   }
+
+   if (x < 0) {
+      /* left clip */
+      skip = -x;
+      x = 0;
+      count -= skip;
+   }
+
+   rb->GetRow(ctx, rb, count, x, y, (GLubyte *) values + skip * valueSize);
+}
+
+
+/**
+ * Get RGBA pixels from the given renderbuffer.
+ * Used by blending, logicop and masking functions.
+ * \return pointer to the colors we read.
+ */
+void *
+_swrast_get_dest_rgba(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                      SWspan *span)
+{
+   const GLuint pixelSize = RGBA_PIXEL_SIZE(span->array->ChanType);
+   void *rbPixels;
+
+   /* Point rbPixels to a temporary space */
+   rbPixels = span->array->attribs[FRAG_ATTRIB_MAX - 1];
+
+   /* Get destination values from renderbuffer */
+   if (span->arrayMask & SPAN_XY) {
+      _swrast_get_values(ctx, rb, span->end, span->array->x, span->array->y,
+                         rbPixels, pixelSize);
+   }
+   else {
+      _swrast_get_row(ctx, rb, span->end, span->x, span->y,
+                      rbPixels, pixelSize);
+   }
+
+   return rbPixels;
+}
diff --git a/mesalib/src/mesa/swrast/s_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c
index 0c8cc9ff3..7f49b6ffa 100644
--- a/mesalib/src/mesa/swrast/s_texcombine.c
+++ b/mesalib/src/mesa/swrast/s_texcombine.c
@@ -631,9 +631,9 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
 
          /* adjust texture lod (lambda) */
          if (span->arrayMask & SPAN_LAMBDA) {
-            if (texUnit->LodBias + curObj->LodBias != 0.0F) {
+            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {
                /* apply LOD bias, but don't clamp yet */
-               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias,
+               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,
                                           -ctx->Const.MaxTextureLodBias,
                                           ctx->Const.MaxTextureLodBias);
                GLuint i;
@@ -642,10 +642,11 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
                }
             }
 
-            if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) {
+            if (curObj->Sampler.MinLod != -1000.0 ||
+                curObj->Sampler.MaxLod != 1000.0) {
                /* apply LOD clamping to lambda */
-               const GLfloat min = curObj->MinLod;
-               const GLfloat max = curObj->MaxLod;
+               const GLfloat min = curObj->Sampler.MinLod;
+               const GLfloat max = curObj->Sampler.MaxLod;
                GLuint i;
                for (i = 0; i < span->end; i++) {
                   GLfloat l = lambda[i];
@@ -686,9 +687,9 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
 
          /* adjust texture lod (lambda) */
          if (span->arrayMask & SPAN_LAMBDA) {
-            if (texUnit->LodBias + curObj->LodBias != 0.0F) {
+            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {
                /* apply LOD bias, but don't clamp yet */
-               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias,
+               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,
                                           -ctx->Const.MaxTextureLodBias,
                                           ctx->Const.MaxTextureLodBias);
                GLuint i;
@@ -697,10 +698,11 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
                }
             }
 
-            if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) {
+            if (curObj->Sampler.MinLod != -1000.0 ||
+                curObj->Sampler.MaxLod != 1000.0) {
                /* apply LOD clamping to lambda */
-               const GLfloat min = curObj->MinLod;
-               const GLfloat max = curObj->MaxLod;
+               const GLfloat min = curObj->Sampler.MinLod;
+               const GLfloat max = curObj->Sampler.MaxLod;
                GLuint i;
                for (i = 0; i < span->end; i++) {
                   GLfloat l = lambda[i];
diff --git a/mesalib/src/mesa/swrast/s_texfilter.c b/mesalib/src/mesa/swrast/s_texfilter.c
index 42785400c..106f8b75f 100644
--- a/mesalib/src/mesa/swrast/s_texfilter.c
+++ b/mesalib/src/mesa/swrast/s_texfilter.c
@@ -505,28 +505,28 @@ nearest_texcoord(const struct gl_texture_object *texObj,
 
    switch (texObj->Target) {
    case GL_TEXTURE_RECTANGLE_ARB:
-      *i = clamp_rect_coord_nearest(texObj->WrapS, texcoord[0], width);
-      *j = clamp_rect_coord_nearest(texObj->WrapT, texcoord[1], height);
+      *i = clamp_rect_coord_nearest(texObj->Sampler.WrapS, texcoord[0], width);
+      *j = clamp_rect_coord_nearest(texObj->Sampler.WrapT, texcoord[1], height);
       *k = 0;
       break;
    case GL_TEXTURE_1D:
-      *i = nearest_texel_location(texObj->WrapS, img, width, texcoord[0]);
+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
       *j = 0;
       *k = 0;
       break;
    case GL_TEXTURE_2D:
-      *i = nearest_texel_location(texObj->WrapS, img, width, texcoord[0]);
-      *j = nearest_texel_location(texObj->WrapT, img, height, texcoord[1]);
+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
+      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);
       *k = 0;
       break;
    case GL_TEXTURE_1D_ARRAY_EXT:
-      *i = nearest_texel_location(texObj->WrapS, img, width, texcoord[0]);
+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
       *j = tex_array_slice(texcoord[1], height);
       *k = 0;
       break;
    case GL_TEXTURE_2D_ARRAY_EXT:
-      *i = nearest_texel_location(texObj->WrapS, img, width, texcoord[0]);
-      *j = nearest_texel_location(texObj->WrapT, img, height, texcoord[1]);
+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
+      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);
       *k = tex_array_slice(texcoord[2], depth);
       break;
    default:
@@ -553,24 +553,24 @@ linear_texcoord(const struct gl_texture_object *texObj,
 
    switch (texObj->Target) {
    case GL_TEXTURE_RECTANGLE_ARB:
-      clamp_rect_coord_linear(texObj->WrapS, texcoord[0],
+      clamp_rect_coord_linear(texObj->Sampler.WrapS, texcoord[0],
                               width, i0, i1, wi);
-      clamp_rect_coord_linear(texObj->WrapT, texcoord[1],
+      clamp_rect_coord_linear(texObj->Sampler.WrapT, texcoord[1],
                               height, j0, j1, wj);
       *slice = 0;
       break;
 
    case GL_TEXTURE_1D:
    case GL_TEXTURE_2D:
-      linear_texel_locations(texObj->WrapS, img, width,
+      linear_texel_locations(texObj->Sampler.WrapS, img, width,
                              texcoord[0], i0, i1, wi);
-      linear_texel_locations(texObj->WrapT, img, height,
+      linear_texel_locations(texObj->Sampler.WrapT, img, height,
                              texcoord[1], j0, j1, wj);
       *slice = 0;
       break;
 
    case GL_TEXTURE_1D_ARRAY_EXT:
-      linear_texel_locations(texObj->WrapS, img, width,
+      linear_texel_locations(texObj->Sampler.WrapS, img, width,
                              texcoord[0], i0, i1, wi);
       *j0 = tex_array_slice(texcoord[1], height);
       *j1 = *j0;
@@ -578,9 +578,9 @@ linear_texcoord(const struct gl_texture_object *texObj,
       break;
 
    case GL_TEXTURE_2D_ARRAY_EXT:
-      linear_texel_locations(texObj->WrapS, img, width,
+      linear_texel_locations(texObj->Sampler.WrapS, img, width,
                              texcoord[0], i0, i1, wi);
-      linear_texel_locations(texObj->WrapT, img, height,
+      linear_texel_locations(texObj->Sampler.WrapT, img, height,
                              texcoord[1], j0, j1, wj);
       *slice = tex_array_slice(texcoord[2], depth);
       break;
@@ -656,12 +656,12 @@ compute_min_mag_ranges(const struct gl_texture_object *tObj,
    GLfloat minMagThresh;
 
    /* we shouldn't be here if minfilter == magfilter */
-   ASSERT(tObj->MinFilter != tObj->MagFilter);
+   ASSERT(tObj->Sampler.MinFilter != tObj->Sampler.MagFilter);
 
    /* This bit comes from the OpenGL spec: */
-   if (tObj->MagFilter == GL_LINEAR
-       && (tObj->MinFilter == GL_NEAREST_MIPMAP_NEAREST ||
-           tObj->MinFilter == GL_NEAREST_MIPMAP_LINEAR)) {
+   if (tObj->Sampler.MagFilter == GL_LINEAR
+       && (tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_NEAREST ||
+           tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_LINEAR)) {
       minMagThresh = 0.5F;
    }
    else {
@@ -763,28 +763,28 @@ get_border_color(const struct gl_texture_object *tObj,
 {
    switch (img->_BaseFormat) {
    case GL_RGB:
-      rgba[0] = tObj->BorderColor.f[0];
-      rgba[1] = tObj->BorderColor.f[1];
-      rgba[2] = tObj->BorderColor.f[2];
+      rgba[0] = tObj->Sampler.BorderColor.f[0];
+      rgba[1] = tObj->Sampler.BorderColor.f[1];
+      rgba[2] = tObj->Sampler.BorderColor.f[2];
       rgba[3] = 1.0F;
       break;
    case GL_ALPHA:
       rgba[0] = rgba[1] = rgba[2] = 0.0;
-      rgba[3] = tObj->BorderColor.f[3];
+      rgba[3] = tObj->Sampler.BorderColor.f[3];
       break;
    case GL_LUMINANCE:
-      rgba[0] = rgba[1] = rgba[2] = tObj->BorderColor.f[0];
+      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];
       rgba[3] = 1.0;
       break;
    case GL_LUMINANCE_ALPHA:
-      rgba[0] = rgba[1] = rgba[2] = tObj->BorderColor.f[0];
-      rgba[3] = tObj->BorderColor.f[3];
+      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];
+      rgba[3] = tObj->Sampler.BorderColor.f[3];
       break;
    case GL_INTENSITY:
-      rgba[0] = rgba[1] = rgba[2] = rgba[3] = tObj->BorderColor.f[0];
+      rgba[0] = rgba[1] = rgba[2] = rgba[3] = tObj->Sampler.BorderColor.f[0];
       break;
    default:
-      COPY_4V(rgba, tObj->BorderColor.f);
+      COPY_4V(rgba, tObj->Sampler.BorderColor.f);
    }
 }
 
@@ -804,7 +804,7 @@ sample_1d_nearest(struct gl_context *ctx,
 {
    const GLint width = img->Width2;  /* without border, power of two */
    GLint i;
-   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
    /* skip over the border, if any */
    i += img->Border;
    if (i < 0 || i >= (GLint) img->Width) {
@@ -832,7 +832,7 @@ sample_1d_linear(struct gl_context *ctx,
    GLfloat a;
    GLfloat t0[4], t1[4];  /* texels */
 
-   linear_texel_locations(tObj->WrapS, img, width, texcoord[0], &i0, &i1, &a);
+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);
 
    if (img->Border) {
       i0 += img->Border;
@@ -991,7 +991,7 @@ sample_lambda_1d( struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       const GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          for (i = minStart; i < minEnd; i++)
             sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -1026,7 +1026,7 @@ sample_lambda_1d( struct gl_context *ctx,
 
    if (magStart < magEnd) {
       /* do the magnified texels */
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          for (i = magStart; i < magEnd; i++)
             sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -1065,8 +1065,8 @@ sample_2d_nearest(struct gl_context *ctx,
    GLint i, j;
    (void) ctx;
 
-   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
 
    /* skip over the border, if any */
    i += img->Border;
@@ -1100,8 +1100,8 @@ sample_2d_linear(struct gl_context *ctx,
    GLfloat a, b;
    GLfloat t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */
 
-   linear_texel_locations(tObj->WrapS, img, width, texcoord[0],  &i0, &i1, &a);
-   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);
+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
 
    if (img->Border) {
       i0 += img->Border;
@@ -1165,8 +1165,8 @@ sample_2d_linear_repeat(struct gl_context *ctx,
 
    (void) ctx;
 
-   ASSERT(tObj->WrapS == GL_REPEAT);
-   ASSERT(tObj->WrapT == GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);
    ASSERT(img->Border == 0);
    ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
    ASSERT(img->_IsPowerOfTwo);
@@ -1270,8 +1270,8 @@ sample_2d_linear_mipmap_linear_repeat(struct gl_context *ctx,
 {
    GLuint i;
    ASSERT(lambda != NULL);
-   ASSERT(tObj->WrapS == GL_REPEAT);
-   ASSERT(tObj->WrapT == GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);
    for (i = 0; i < n; i++) {
       GLint level = linear_mipmap_level(tObj, lambda[i]);
       if (level >= tObj->_MaxLevel) {
@@ -1317,8 +1317,8 @@ sample_linear_2d(struct gl_context *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   if (tObj->WrapS == GL_REPEAT &&
-       tObj->WrapT == GL_REPEAT &&
+   if (tObj->Sampler.WrapS == GL_REPEAT &&
+       tObj->Sampler.WrapT == GL_REPEAT &&
        image->_IsPowerOfTwo &&
        image->Border == 0) {
       for (i = 0; i < n; i++) {
@@ -1356,8 +1356,8 @@ opt_sample_rgb_2d(struct gl_context *ctx,
    GLuint k;
    (void) ctx;
    (void) lambda;
-   ASSERT(tObj->WrapS==GL_REPEAT);
-   ASSERT(tObj->WrapT==GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);
    ASSERT(img->Border==0);
    ASSERT(img->TexFormat == MESA_FORMAT_RGB888);
    ASSERT(img->_IsPowerOfTwo);
@@ -1398,8 +1398,8 @@ opt_sample_rgba_2d(struct gl_context *ctx,
    GLuint i;
    (void) ctx;
    (void) lambda;
-   ASSERT(tObj->WrapS==GL_REPEAT);
-   ASSERT(tObj->WrapT==GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);
+   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);
    ASSERT(img->Border==0);
    ASSERT(img->TexFormat == MESA_FORMAT_RGBA8888);
    ASSERT(img->_IsPowerOfTwo);
@@ -1428,8 +1428,8 @@ sample_lambda_2d(struct gl_context *ctx,
    GLuint minStart, minEnd;  /* texels with minification */
    GLuint magStart, magEnd;  /* texels with magnification */
 
-   const GLboolean repeatNoBorderPOT = (tObj->WrapS == GL_REPEAT)
-      && (tObj->WrapT == GL_REPEAT)
+   const GLboolean repeatNoBorderPOT = (tObj->Sampler.WrapS == GL_REPEAT)
+      && (tObj->Sampler.WrapT == GL_REPEAT)
       && (tImg->Border == 0 && (tImg->Width == tImg->RowStride))
       && (tImg->_BaseFormat != GL_COLOR_INDEX)
       && tImg->_IsPowerOfTwo;
@@ -1441,7 +1441,7 @@ sample_lambda_2d(struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       const GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          if (repeatNoBorderPOT) {
             switch (tImg->TexFormat) {
@@ -1498,7 +1498,7 @@ sample_lambda_2d(struct gl_context *ctx,
       /* do the magnified texels */
       const GLuint m = magEnd - magStart;
 
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          if (repeatNoBorderPOT) {
             switch (tImg->TexFormat) {
@@ -1552,9 +1552,9 @@ sample_3d_nearest(struct gl_context *ctx,
    GLint i, j, k;
    (void) ctx;
 
-   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
-   k = nearest_texel_location(tObj->WrapR, img, depth, texcoord[2]);
+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
+   k = nearest_texel_location(tObj->Sampler.WrapR, img, depth, texcoord[2]);
 
    if (i < 0 || i >= (GLint) img->Width ||
        j < 0 || j >= (GLint) img->Height ||
@@ -1587,9 +1587,9 @@ sample_3d_linear(struct gl_context *ctx,
    GLfloat t000[4], t010[4], t001[4], t011[4];
    GLfloat t100[4], t110[4], t101[4], t111[4];
 
-   linear_texel_locations(tObj->WrapS, img, width, texcoord[0],  &i0, &i1, &a);
-   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
-   linear_texel_locations(tObj->WrapR, img, depth, texcoord[2],  &k0, &k1, &c);
+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);
+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
+   linear_texel_locations(tObj->Sampler.WrapR, img, depth, texcoord[2],  &k0, &k1, &c);
 
    if (img->Border) {
       i0 += img->Border;
@@ -1794,7 +1794,7 @@ sample_lambda_3d(struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          for (i = minStart; i < minEnd; i++)
             sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -1829,7 +1829,7 @@ sample_lambda_3d(struct gl_context *ctx,
 
    if (magStart < magEnd) {
       /* do the magnified texels */
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          for (i = magStart; i < magEnd; i++)
             sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -2091,7 +2091,7 @@ sample_lambda_cube(struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       const GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          sample_nearest_cube(ctx, tObj, m, texcoords + minStart,
                              lambda + minStart, rgba + minStart);
@@ -2128,7 +2128,7 @@ sample_lambda_cube(struct gl_context *ctx,
    if (magStart < magEnd) {
       /* do the magnified texels */
       const GLuint m = magEnd - magStart;
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          sample_nearest_cube(ctx, tObj, m, texcoords + magStart,
                              lambda + magStart, rgba + magStart);
@@ -2163,18 +2163,18 @@ sample_nearest_rect(struct gl_context *ctx,
    (void) ctx;
    (void) lambda;
 
-   ASSERT(tObj->WrapS == GL_CLAMP ||
-          tObj->WrapS == GL_CLAMP_TO_EDGE ||
-          tObj->WrapS == GL_CLAMP_TO_BORDER);
-   ASSERT(tObj->WrapT == GL_CLAMP ||
-          tObj->WrapT == GL_CLAMP_TO_EDGE ||
-          tObj->WrapT == GL_CLAMP_TO_BORDER);
+   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||
+          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||
+          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);
+   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||
+          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||
+          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);
    ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
 
    for (i = 0; i < n; i++) {
       GLint row, col;
-      col = clamp_rect_coord_nearest(tObj->WrapS, texcoords[i][0], width);
-      row = clamp_rect_coord_nearest(tObj->WrapT, texcoords[i][1], height);
+      col = clamp_rect_coord_nearest(tObj->Sampler.WrapS, texcoords[i][0], width);
+      row = clamp_rect_coord_nearest(tObj->Sampler.WrapT, texcoords[i][1], height);
       if (col < 0 || col >= width || row < 0 || row >= height)
          get_border_color(tObj, img, rgba[i]);
       else
@@ -2197,12 +2197,12 @@ sample_linear_rect(struct gl_context *ctx,
    (void) ctx;
    (void) lambda;
 
-   ASSERT(tObj->WrapS == GL_CLAMP ||
-          tObj->WrapS == GL_CLAMP_TO_EDGE ||
-          tObj->WrapS == GL_CLAMP_TO_BORDER);
-   ASSERT(tObj->WrapT == GL_CLAMP ||
-          tObj->WrapT == GL_CLAMP_TO_EDGE ||
-          tObj->WrapT == GL_CLAMP_TO_BORDER);
+   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||
+          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||
+          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);
+   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||
+          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||
+          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);
    ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
 
    for (i = 0; i < n; i++) {
@@ -2211,9 +2211,9 @@ sample_linear_rect(struct gl_context *ctx,
       GLfloat a, b;
       GLbitfield useBorderColor = 0x0;
 
-      clamp_rect_coord_linear(tObj->WrapS, texcoords[i][0], width,
+      clamp_rect_coord_linear(tObj->Sampler.WrapS, texcoords[i][0], width,
                               &i0, &i1, &a);
-      clamp_rect_coord_linear(tObj->WrapT, texcoords[i][1], height,
+      clamp_rect_coord_linear(tObj->Sampler.WrapT, texcoords[i][1], height,
                               &j0, &j1, &b);
 
       /* compute integer rows/columns */
@@ -2264,7 +2264,7 @@ sample_lambda_rect(struct gl_context *ctx,
                           &minStart, &minEnd, &magStart, &magEnd);
 
    if (minStart < minEnd) {
-      if (tObj->MinFilter == GL_NEAREST) {
+      if (tObj->Sampler.MinFilter == GL_NEAREST) {
          sample_nearest_rect(ctx, tObj, minEnd - minStart,
                              texcoords + minStart, NULL, rgba + minStart);
       }
@@ -2274,7 +2274,7 @@ sample_lambda_rect(struct gl_context *ctx,
       }
    }
    if (magStart < magEnd) {
-      if (tObj->MagFilter == GL_NEAREST) {
+      if (tObj->Sampler.MagFilter == GL_NEAREST) {
          sample_nearest_rect(ctx, tObj, magEnd - magStart,
                              texcoords + magStart, NULL, rgba + magStart);
       }
@@ -2307,8 +2307,8 @@ sample_2d_array_nearest(struct gl_context *ctx,
    GLint array;
    (void) ctx;
 
-   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
    array = tex_array_slice(texcoord[2], depth);
 
    if (i < 0 || i >= (GLint) img->Width ||
@@ -2342,12 +2342,12 @@ sample_2d_array_linear(struct gl_context *ctx,
    GLfloat a, b;
    GLfloat t00[4], t01[4], t10[4], t11[4];
 
-   linear_texel_locations(tObj->WrapS, img, width,  texcoord[0], &i0, &i1, &a);
-   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
+   linear_texel_locations(tObj->Sampler.WrapS, img, width,  texcoord[0], &i0, &i1, &a);
+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
    array = tex_array_slice(texcoord[2], depth);
 
    if (array < 0 || array >= depth) {
-      COPY_4V(rgba, tObj->BorderColor.f);
+      COPY_4V(rgba, tObj->Sampler.BorderColor.f);
    }
    else {
       if (img->Border) {
@@ -2532,7 +2532,7 @@ sample_lambda_2d_array(struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          for (i = minStart; i < minEnd; i++)
             sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -2575,7 +2575,7 @@ sample_lambda_2d_array(struct gl_context *ctx,
 
    if (magStart < magEnd) {
       /* do the magnified texels */
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          for (i = magStart; i < magEnd; i++)
             sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -2616,7 +2616,7 @@ sample_1d_array_nearest(struct gl_context *ctx,
    GLint array;
    (void) ctx;
 
-   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
    array = tex_array_slice(texcoord[1], height);
 
    if (i < 0 || i >= (GLint) img->Width ||
@@ -2648,7 +2648,7 @@ sample_1d_array_linear(struct gl_context *ctx,
    GLfloat a;
    GLfloat t0[4], t1[4];
 
-   linear_texel_locations(tObj->WrapS, img, width, texcoord[0], &i0, &i1, &a);
+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);
    array = tex_array_slice(texcoord[1], height);
 
    if (img->Border) {
@@ -2813,7 +2813,7 @@ sample_lambda_1d_array(struct gl_context *ctx,
    if (minStart < minEnd) {
       /* do the minified texels */
       GLuint m = minEnd - minStart;
-      switch (tObj->MinFilter) {
+      switch (tObj->Sampler.MinFilter) {
       case GL_NEAREST:
          for (i = minStart; i < minEnd; i++)
             sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -2852,7 +2852,7 @@ sample_lambda_1d_array(struct gl_context *ctx,
 
    if (magStart < magEnd) {
       /* do the magnified texels */
-      switch (tObj->MagFilter) {
+      switch (tObj->Sampler.MagFilter) {
       case GL_NEAREST:
          for (i = magStart; i < magEnd; i++)
             sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
@@ -2975,13 +2975,13 @@ choose_depth_texture_level(const struct gl_texture_object *tObj, GLfloat lambda)
 {
    GLint level;
 
-   if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) {
+   if (tObj->Sampler.MinFilter == GL_NEAREST || tObj->Sampler.MinFilter == GL_LINEAR) {
       /* no mipmapping - use base level */
       level = tObj->BaseLevel;
    }
    else {
       /* choose mipmap level */
-      lambda = CLAMP(lambda, tObj->MinLod, tObj->MaxLod);
+      lambda = CLAMP(lambda, tObj->Sampler.MinLod, tObj->Sampler.MaxLod);
       level = (GLint) lambda;
       level = CLAMP(level, tObj->BaseLevel, tObj->_MaxLevel);
    }
@@ -3020,14 +3020,14 @@ sample_depth_texture( struct gl_context *ctx,
           tObj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
           tObj->Target == GL_TEXTURE_2D_ARRAY_EXT);
 
-   ambient = tObj->CompareFailValue;
+   ambient = tObj->Sampler.CompareFailValue;
 
-   /* XXXX if tObj->MinFilter != tObj->MagFilter, we're ignoring lambda */
+   /* XXXX if tObj->Sampler.MinFilter != tObj->Sampler.MagFilter, we're ignoring lambda */
 
-   function = (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) ?
-      tObj->CompareFunc : GL_NONE;
+   function = (tObj->Sampler.CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) ?
+      tObj->Sampler.CompareFunc : GL_NONE;
 
-   if (tObj->MagFilter == GL_NEAREST) {
+   if (tObj->Sampler.MagFilter == GL_NEAREST) {
       GLuint i;
       for (i = 0; i < n; i++) {
          GLfloat depthSample, depthRef;
@@ -3040,14 +3040,14 @@ sample_depth_texture( struct gl_context *ctx,
             img->FetchTexelf(img, col, row, slice, &depthSample);
          }
          else {
-            depthSample = tObj->BorderColor.f[0];
+            depthSample = tObj->Sampler.BorderColor.f[0];
          }
 
          depthRef = CLAMP(texcoords[i][compare_coord], 0.0F, 1.0F);
 
          result = shadow_compare(function, depthRef, depthSample, ambient);
 
-         switch (tObj->DepthMode) {
+         switch (tObj->Sampler.DepthMode) {
          case GL_LUMINANCE:
             ASSIGN_4V(texel[i], result, result, result, 1.0F);
             break;
@@ -3067,7 +3067,7 @@ sample_depth_texture( struct gl_context *ctx,
    }
    else {
       GLuint i;
-      ASSERT(tObj->MagFilter == GL_LINEAR);
+      ASSERT(tObj->Sampler.MagFilter == GL_LINEAR);
       for (i = 0; i < n; i++) {
          GLfloat depth00, depth01, depth10, depth11, depthRef;
          GLint i0, i1, j0, j1;
@@ -3095,21 +3095,21 @@ sample_depth_texture( struct gl_context *ctx,
          }
 
          if (slice < 0 || slice >= (GLint) depth) {
-            depth00 = tObj->BorderColor.f[0];
-            depth01 = tObj->BorderColor.f[0];
-            depth10 = tObj->BorderColor.f[0];
-            depth11 = tObj->BorderColor.f[0];
+            depth00 = tObj->Sampler.BorderColor.f[0];
+            depth01 = tObj->Sampler.BorderColor.f[0];
+            depth10 = tObj->Sampler.BorderColor.f[0];
+            depth11 = tObj->Sampler.BorderColor.f[0];
          }
          else {
             /* get four depth samples from the texture */
             if (useBorderTexel & (I0BIT | J0BIT)) {
-               depth00 = tObj->BorderColor.f[0];
+               depth00 = tObj->Sampler.BorderColor.f[0];
             }
             else {
                img->FetchTexelf(img, i0, j0, slice, &depth00);
             }
             if (useBorderTexel & (I1BIT | J0BIT)) {
-               depth10 = tObj->BorderColor.f[0];
+               depth10 = tObj->Sampler.BorderColor.f[0];
             }
             else {
                img->FetchTexelf(img, i1, j0, slice, &depth10);
@@ -3117,13 +3117,13 @@ sample_depth_texture( struct gl_context *ctx,
 
             if (tObj->Target != GL_TEXTURE_1D_ARRAY_EXT) {
                if (useBorderTexel & (I0BIT | J1BIT)) {
-                  depth01 = tObj->BorderColor.f[0];
+                  depth01 = tObj->Sampler.BorderColor.f[0];
                }
                else {
                   img->FetchTexelf(img, i0, j1, slice, &depth01);
                }
                if (useBorderTexel & (I1BIT | J1BIT)) {
-                  depth11 = tObj->BorderColor.f[0];
+                  depth11 = tObj->Sampler.BorderColor.f[0];
                }
                else {
                   img->FetchTexelf(img, i1, j1, slice, &depth11);
@@ -3141,7 +3141,7 @@ sample_depth_texture( struct gl_context *ctx,
                                   depth00, depth01, depth10, depth11,
                                   ambient, wi, wj);
 
-         switch (tObj->DepthMode) {
+         switch (tObj->Sampler.DepthMode) {
          case GL_LUMINANCE:
             ASSIGN_4V(texel[i], result, result, result, 1.0F);
             break;
@@ -3197,7 +3197,8 @@ _swrast_choose_texture_sample_func( struct gl_context *ctx,
       return &null_sample_func;
    }
    else {
-      const GLboolean needLambda = (GLboolean) (t->MinFilter != t->MagFilter);
+      const GLboolean needLambda =
+         (GLboolean) (t->Sampler.MinFilter != t->Sampler.MagFilter);
       const GLenum format = t->Image[0][t->BaseLevel]->_BaseFormat;
 
       switch (t->Target) {
@@ -3208,11 +3209,11 @@ _swrast_choose_texture_sample_func( struct gl_context *ctx,
          else if (needLambda) {
             return &sample_lambda_1d;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_1d;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_1d;
          }
       case GL_TEXTURE_2D:
@@ -3222,22 +3223,22 @@ _swrast_choose_texture_sample_func( struct gl_context *ctx,
          else if (needLambda) {
             return &sample_lambda_2d;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_2d;
          }
          else {
             /* check for a few optimized cases */
             const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
-            ASSERT(t->MinFilter == GL_NEAREST);
-            if (t->WrapS == GL_REPEAT &&
-                t->WrapT == GL_REPEAT &&
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
+            if (t->Sampler.WrapS == GL_REPEAT &&
+                t->Sampler.WrapT == GL_REPEAT &&
                 img->_IsPowerOfTwo &&
                 img->Border == 0 &&
                 img->TexFormat == MESA_FORMAT_RGB888) {
                return &opt_sample_rgb_2d;
             }
-            else if (t->WrapS == GL_REPEAT &&
-                     t->WrapT == GL_REPEAT &&
+            else if (t->Sampler.WrapS == GL_REPEAT &&
+                     t->Sampler.WrapT == GL_REPEAT &&
                      img->_IsPowerOfTwo &&
                      img->Border == 0 &&
                      img->TexFormat == MESA_FORMAT_RGBA8888) {
@@ -3251,22 +3252,22 @@ _swrast_choose_texture_sample_func( struct gl_context *ctx,
          if (needLambda) {
             return &sample_lambda_3d;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_3d;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_3d;
          }
       case GL_TEXTURE_CUBE_MAP:
          if (needLambda) {
             return &sample_lambda_cube;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_cube;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_cube;
          }
       case GL_TEXTURE_RECTANGLE_NV:
@@ -3276,33 +3277,33 @@ _swrast_choose_texture_sample_func( struct gl_context *ctx,
          else if (needLambda) {
             return &sample_lambda_rect;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_rect;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_rect;
          }
       case GL_TEXTURE_1D_ARRAY_EXT:
          if (needLambda) {
             return &sample_lambda_1d_array;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_1d_array;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_1d_array;
          }
       case GL_TEXTURE_2D_ARRAY_EXT:
          if (needLambda) {
             return &sample_lambda_2d_array;
          }
-         else if (t->MinFilter == GL_LINEAR) {
+         else if (t->Sampler.MinFilter == GL_LINEAR) {
             return &sample_linear_2d_array;
          }
          else {
-            ASSERT(t->MinFilter == GL_NEAREST);
+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
             return &sample_nearest_2d_array;
          }
       default:
diff --git a/mesalib/src/mesa/swrast/s_triangle.c b/mesalib/src/mesa/swrast/s_triangle.c
index c2c1ce221..8a9671aa0 100644
--- a/mesalib/src/mesa/swrast/s_triangle.c
+++ b/mesalib/src/mesa/swrast/s_triangle.c
@@ -1,1143 +1,1143 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/*
- * When the device driver doesn't implement triangle rasterization it
- * can hook in _swrast_Triangle, which eventually calls one of these
- * functions to draw triangles.
- */
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/colormac.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "main/state.h"
-#include "program/prog_instruction.h"
-
-#include "s_aatriangle.h"
-#include "s_context.h"
-#include "s_feedback.h"
-#include "s_span.h"
-#include "s_triangle.h"
-
-
-/**
- * Test if a triangle should be culled.  Used for feedback and selection mode.
- * \return GL_TRUE if the triangle is to be culled, GL_FALSE otherwise.
- */
-GLboolean
-_swrast_culltriangle( struct gl_context *ctx,
-                      const SWvertex *v0,
-                      const SWvertex *v1,
-                      const SWvertex *v2 )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLfloat ex = v1->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
-   GLfloat ey = v1->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
-   GLfloat fx = v2->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
-   GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
-   GLfloat c = ex*fy-ey*fx;
-
-   if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign <= 0.0F)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-
-/*
- * Render a flat-shaded RGBA triangle.
- */
-#define NAME flat_rgba_triangle
-#define INTERP_Z 1
-#define SETUP_CODE				\
-   ASSERT(ctx->Texture._EnabledCoordUnits == 0);\
-   ASSERT(ctx->Light.ShadeModel==GL_FLAT);	\
-   span.interpMask |= SPAN_RGBA;		\
-   span.red = ChanToFixed(v2->color[0]);	\
-   span.green = ChanToFixed(v2->color[1]);	\
-   span.blue = ChanToFixed(v2->color[2]);	\
-   span.alpha = ChanToFixed(v2->color[3]);	\
-   span.redStep = 0;				\
-   span.greenStep = 0;				\
-   span.blueStep = 0;				\
-   span.alphaStep = 0;
-#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-/*
- * Render a smooth-shaded RGBA triangle.
- */
-#define NAME smooth_rgba_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define SETUP_CODE				\
-   {						\
-      /* texturing must be off */		\
-      ASSERT(ctx->Texture._EnabledCoordUnits == 0);	\
-      ASSERT(ctx->Light.ShadeModel==GL_SMOOTH);	\
-   }
-#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-/*
- * Render an RGB, GL_DECAL, textured triangle.
- * Interpolate S,T only w/out mipmapping or perspective correction.
- *
- * No fog.  No depth testing.
- */
-#define NAME simple_textured_triangle
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg =				\
-      obj->Image[0][obj->BaseLevel];					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   const GLint twidth_log2 = texImg->WidthLog2;				\
-   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
-   const GLint smask = texImg->Width - 1;				\
-   const GLint tmask = texImg->Height - 1;				\
-   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
-   if (!rb || !texture) {						\
-      return;								\
-   }
-
-#define RENDER_SPAN( span )						\
-   GLuint i;								\
-   GLubyte rgb[MAX_WIDTH][3];						\
-   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
-   span.intTex[1] -= FIXED_HALF;					\
-   for (i = 0; i < span.end; i++) {					\
-      GLint s = FixedToInt(span.intTex[0]) & smask;			\
-      GLint t = FixedToInt(span.intTex[1]) & tmask;			\
-      GLint pos = (t << twidth_log2) + s;				\
-      pos = pos + pos + pos;  /* multiply by 3 */			\
-      rgb[i][RCOMP] = texture[pos+2];					\
-      rgb[i][GCOMP] = texture[pos+1];					\
-      rgb[i][BCOMP] = texture[pos+0];					\
-      span.intTex[0] += span.intTexStep[0];				\
-      span.intTex[1] += span.intTexStep[1];				\
-   }									\
-   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, NULL);
-
-#include "s_tritemp.h"
-
-
-
-/*
- * Render an RGB, GL_DECAL, textured triangle.
- * Interpolate S,T, GL_LESS depth test, w/out mipmapping or
- * perspective correction.
- * Depth buffer bits must be <= sizeof(DEFAULT_SOFTWARE_DEPTH_TYPE)
- *
- * No fog.
- */
-#define NAME simple_z_textured_triangle
-#define INTERP_Z 1
-#define DEPTH_TYPE DEFAULT_SOFTWARE_DEPTH_TYPE
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-       obj->Image[0][obj->BaseLevel]; 					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   const GLint twidth_log2 = texImg->WidthLog2;				\
-   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
-   const GLint smask = texImg->Width - 1;				\
-   const GLint tmask = texImg->Height - 1;				\
-   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
-   if (!rb || !texture) {						\
-      return;								\
-   }
-
-#define RENDER_SPAN( span )						\
-   GLuint i;				    				\
-   GLubyte rgb[MAX_WIDTH][3];						\
-   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
-   span.intTex[1] -= FIXED_HALF;					\
-   for (i = 0; i < span.end; i++) {					\
-      const GLuint z = FixedToDepth(span.z);				\
-      if (z < zRow[i]) {						\
-         GLint s = FixedToInt(span.intTex[0]) & smask;			\
-         GLint t = FixedToInt(span.intTex[1]) & tmask;			\
-         GLint pos = (t << twidth_log2) + s;				\
-         pos = pos + pos + pos;  /* multiply by 3 */			\
-         rgb[i][RCOMP] = texture[pos+2];				\
-         rgb[i][GCOMP] = texture[pos+1];				\
-         rgb[i][BCOMP] = texture[pos+0];				\
-         zRow[i] = z;							\
-         span.array->mask[i] = 1;					\
-      }									\
-      else {								\
-         span.array->mask[i] = 0;					\
-      }									\
-      span.intTex[0] += span.intTexStep[0];				\
-      span.intTex[1] += span.intTexStep[1];				\
-      span.z += span.zStep;						\
-   }									\
-   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, span.array->mask);
-
-#include "s_tritemp.h"
-
-
-#if CHAN_TYPE != GL_FLOAT
-
-struct affine_info
-{
-   GLenum filter;
-   GLenum format;
-   GLenum envmode;
-   GLint smask, tmask;
-   GLint twidth_log2;
-   const GLchan *texture;
-   GLfixed er, eg, eb, ea;
-   GLint tbytesline, tsize;
-};
-
-
-static INLINE GLint
-ilerp(GLint t, GLint a, GLint b)
-{
-   return a + ((t * (b - a)) >> FIXED_SHIFT);
-}
-
-static INLINE GLint
-ilerp_2d(GLint ia, GLint ib, GLint v00, GLint v10, GLint v01, GLint v11)
-{
-   const GLint temp0 = ilerp(ia, v00, v10);
-   const GLint temp1 = ilerp(ia, v01, v11);
-   return ilerp(ib, temp0, temp1);
-}
-
-
-/* This function can handle GL_NEAREST or GL_LINEAR sampling of 2D RGB or RGBA
- * textures with GL_REPLACE, GL_MODULATE, GL_BLEND, GL_DECAL or GL_ADD
- * texture env modes.
- */
-static INLINE void
-affine_span(struct gl_context *ctx, SWspan *span,
-            struct affine_info *info)
-{
-   GLchan sample[4];  /* the filtered texture sample */
-   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
-
-   /* Instead of defining a function for each mode, a test is done
-    * between the outer and inner loops. This is to reduce code size
-    * and complexity. Observe that an optimizing compiler kills
-    * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
-    */
-
-#define NEAREST_RGB		\
-   sample[RCOMP] = tex00[2];	\
-   sample[GCOMP] = tex00[1];	\
-   sample[BCOMP] = tex00[0];	\
-   sample[ACOMP] = CHAN_MAX;
-
-#define LINEAR_RGB							\
-   sample[RCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
-   sample[GCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
-   sample[BCOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0]);\
-   sample[ACOMP] = CHAN_MAX;
-
-#define NEAREST_RGBA  \
-   sample[RCOMP] = tex00[3];	\
-   sample[GCOMP] = tex00[2];	\
-   sample[BCOMP] = tex00[1];	\
-   sample[ACOMP] = tex00[0];
-
-#define LINEAR_RGBA							\
-   sample[RCOMP] = ilerp_2d(sf, tf, tex00[3], tex01[3], tex10[3], tex11[3]);\
-   sample[GCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
-   sample[BCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
-   sample[ACOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0])
-
-#define MODULATE							  \
-   dest[RCOMP] = span->red   * (sample[RCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[GCOMP] = span->green * (sample[GCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[BCOMP] = span->blue  * (sample[BCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1u) >> (FIXED_SHIFT + 8)
-
-#define DECAL								\
-   dest[RCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->red +		\
-               ((sample[ACOMP] + 1) * sample[RCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[GCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->green +		\
-               ((sample[ACOMP] + 1) * sample[GCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[BCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->blue +		\
-               ((sample[ACOMP] + 1) * sample[BCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[ACOMP] = FixedToInt(span->alpha)
-
-#define BLEND								\
-   dest[RCOMP] = ((CHAN_MAX - sample[RCOMP]) * span->red		\
-               + (sample[RCOMP] + 1) * info->er) >> (FIXED_SHIFT + 8);	\
-   dest[GCOMP] = ((CHAN_MAX - sample[GCOMP]) * span->green		\
-               + (sample[GCOMP] + 1) * info->eg) >> (FIXED_SHIFT + 8);	\
-   dest[BCOMP] = ((CHAN_MAX - sample[BCOMP]) * span->blue		\
-               + (sample[BCOMP] + 1) * info->eb) >> (FIXED_SHIFT + 8);	\
-   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8)
-
-#define REPLACE  COPY_CHAN4(dest, sample)
-
-#define ADD								\
-   {									\
-      GLint rSum = FixedToInt(span->red)   + (GLint) sample[RCOMP];	\
-      GLint gSum = FixedToInt(span->green) + (GLint) sample[GCOMP];	\
-      GLint bSum = FixedToInt(span->blue)  + (GLint) sample[BCOMP];	\
-      dest[RCOMP] = MIN2(rSum, CHAN_MAX);				\
-      dest[GCOMP] = MIN2(gSum, CHAN_MAX);				\
-      dest[BCOMP] = MIN2(bSum, CHAN_MAX);				\
-      dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8); \
-  }
-
-/* shortcuts */
-
-#define NEAREST_RGB_REPLACE		\
-   NEAREST_RGB;				\
-   dest[0] = sample[0];			\
-   dest[1] = sample[1];			\
-   dest[2] = sample[2];			\
-   dest[3] = FixedToInt(span->alpha);
-
-#define NEAREST_RGBA_REPLACE  \
-   dest[RCOMP] = tex00[3]; \
-   dest[GCOMP] = tex00[2]; \
-   dest[BCOMP] = tex00[1]; \
-   dest[ACOMP] = tex00[0]
-
-#define SPAN_NEAREST(DO_TEX, COMPS)					\
-	for (i = 0; i < span->end; i++) {				\
-           /* Isn't it necessary to use FixedFloor below?? */		\
-           GLint s = FixedToInt(span->intTex[0]) & info->smask;		\
-           GLint t = FixedToInt(span->intTex[1]) & info->tmask;		\
-           GLint pos = (t << info->twidth_log2) + s;			\
-           const GLchan *tex00 = info->texture + COMPS * pos;		\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   span->intTex[0] += span->intTexStep[0];			\
-	   span->intTex[1] += span->intTexStep[1];			\
-           dest += 4;							\
-	}
-
-#define SPAN_LINEAR(DO_TEX, COMPS)					\
-	for (i = 0; i < span->end; i++) {				\
-           /* Isn't it necessary to use FixedFloor below?? */		\
-           const GLint s = FixedToInt(span->intTex[0]) & info->smask;	\
-           const GLint t = FixedToInt(span->intTex[1]) & info->tmask;	\
-           const GLfixed sf = span->intTex[0] & FIXED_FRAC_MASK;	\
-           const GLfixed tf = span->intTex[1] & FIXED_FRAC_MASK;	\
-           const GLint pos = (t << info->twidth_log2) + s;		\
-           const GLchan *tex00 = info->texture + COMPS * pos;		\
-           const GLchan *tex10 = tex00 + info->tbytesline;		\
-           const GLchan *tex01 = tex00 + COMPS;				\
-           const GLchan *tex11 = tex10 + COMPS;				\
-           if (t == info->tmask) {					\
-              tex10 -= info->tsize;					\
-              tex11 -= info->tsize;					\
-           }								\
-           if (s == info->smask) {					\
-              tex01 -= info->tbytesline;				\
-              tex11 -= info->tbytesline;				\
-           }								\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   span->intTex[0] += span->intTexStep[0];			\
-	   span->intTex[1] += span->intTexStep[1];			\
-           dest += 4;							\
-	}
-
-
-   GLuint i;
-   GLchan *dest = span->array->rgba[0];
-
-   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */
-   ctx->Texture._EnabledCoordUnits = 0x0;
-
-   span->intTex[0] -= FIXED_HALF;
-   span->intTex[1] -= FIXED_HALF;
-   switch (info->filter) {
-   case GL_NEAREST:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch(info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (2) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-
-   case GL_LINEAR:
-      span->intTex[0] -= FIXED_HALF;
-      span->intTex[1] -= FIXED_HALF;
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (3) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (4) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-   }
-   span->interpMask &= ~SPAN_RGBA;
-   ASSERT(span->arrayMask & SPAN_RGBA);
-
-   _swrast_write_rgba_span(ctx, span);
-
-   /* re-enable texture units */
-   ctx->Texture._EnabledCoordUnits = texEnableSave;
-
-#undef SPAN_NEAREST
-#undef SPAN_LINEAR
-}
-
-
-
-/*
- * Render an RGB/RGBA textured triangle without perspective correction.
- */
-#define NAME affine_textured_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct affine_info info;						\
-   struct gl_texture_unit *unit = ctx->Texture.Unit+0;			\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-      obj->Image[0][obj->BaseLevel]; 					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   info.texture = (const GLchan *) texImg->Data;			\
-   info.twidth_log2 = texImg->WidthLog2;				\
-   info.smask = texImg->Width - 1;					\
-   info.tmask = texImg->Height - 1;					\
-   info.format = texImg->TexFormat;					\
-   info.filter = obj->MinFilter;					\
-   info.envmode = unit->EnvMode;					\
-   info.er = 0;					\
-   info.eg = 0;					\
-   info.eb = 0;					\
-   span.arrayMask |= SPAN_RGBA;						\
-									\
-   if (info.envmode == GL_BLEND) {					\
-      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
-      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
-      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
-      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
-      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
-   }									\
-   if (!info.texture) {							\
-      /* this shouldn't happen */					\
-      return;								\
-   }									\
-									\
-   switch (info.format) {						\
-   case MESA_FORMAT_RGB888:						\
-      info.tbytesline = texImg->Width * 3;				\
-      break;								\
-   case MESA_FORMAT_RGBA8888:						\
-      info.tbytesline = texImg->Width * 4;				\
-      break;								\
-   default:								\
-      _mesa_problem(NULL, "Bad texture format in affine_texture_triangle");\
-      return;								\
-   }									\
-   info.tsize = texImg->Height * info.tbytesline;
-
-#define RENDER_SPAN( span )   affine_span(ctx, &span, &info);
-
-#include "s_tritemp.h"
-
-
-
-struct persp_info
-{
-   GLenum filter;
-   GLenum format;
-   GLenum envmode;
-   GLint smask, tmask;
-   GLint twidth_log2;
-   const GLchan *texture;
-   GLfixed er, eg, eb, ea;   /* texture env color */
-   GLint tbytesline, tsize;
-};
-
-
-static INLINE void
-fast_persp_span(struct gl_context *ctx, SWspan *span,
-		struct persp_info *info)
-{
-   GLchan sample[4];  /* the filtered texture sample */
-
-  /* Instead of defining a function for each mode, a test is done
-   * between the outer and inner loops. This is to reduce code size
-   * and complexity. Observe that an optimizing compiler kills
-   * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
-   */
-#define SPAN_NEAREST(DO_TEX,COMP)					\
-	for (i = 0; i < span->end; i++) {				\
-           GLdouble invQ = tex_coord[2] ?				\
-                                 (1.0 / tex_coord[2]) : 1.0;            \
-           GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);		\
-           GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);		\
-           GLint s = IFLOOR(s_tmp) & info->smask;	        	\
-           GLint t = IFLOOR(t_tmp) & info->tmask;	        	\
-           GLint pos = (t << info->twidth_log2) + s;			\
-           const GLchan *tex00 = info->texture + COMP * pos;		\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   tex_coord[0] += tex_step[0];					\
-	   tex_coord[1] += tex_step[1];					\
-	   tex_coord[2] += tex_step[2];					\
-           dest += 4;							\
-	}
-
-#define SPAN_LINEAR(DO_TEX,COMP)					\
-	for (i = 0; i < span->end; i++) {				\
-           GLdouble invQ = tex_coord[2] ?				\
-                                 (1.0 / tex_coord[2]) : 1.0;            \
-           const GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);	\
-           const GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);	\
-           const GLfixed s_fix = FloatToFixed(s_tmp) - FIXED_HALF;	\
-           const GLfixed t_fix = FloatToFixed(t_tmp) - FIXED_HALF;      \
-           const GLint s = FixedToInt(FixedFloor(s_fix)) & info->smask;	\
-           const GLint t = FixedToInt(FixedFloor(t_fix)) & info->tmask;	\
-           const GLfixed sf = s_fix & FIXED_FRAC_MASK;			\
-           const GLfixed tf = t_fix & FIXED_FRAC_MASK;			\
-           const GLint pos = (t << info->twidth_log2) + s;		\
-           const GLchan *tex00 = info->texture + COMP * pos;		\
-           const GLchan *tex10 = tex00 + info->tbytesline;		\
-           const GLchan *tex01 = tex00 + COMP;				\
-           const GLchan *tex11 = tex10 + COMP;				\
-           if (t == info->tmask) {					\
-              tex10 -= info->tsize;					\
-              tex11 -= info->tsize;					\
-           }								\
-           if (s == info->smask) {					\
-              tex01 -= info->tbytesline;				\
-              tex11 -= info->tbytesline;				\
-           }								\
-           DO_TEX;							\
-           span->red   += span->redStep;				\
-	   span->green += span->greenStep;				\
-           span->blue  += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   tex_coord[0] += tex_step[0];					\
-	   tex_coord[1] += tex_step[1];					\
-	   tex_coord[2] += tex_step[2];					\
-           dest += 4;							\
-	}
-
-   GLuint i;
-   GLfloat tex_coord[3], tex_step[3];
-   GLchan *dest = span->array->rgba[0];
-
-   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
-   ctx->Texture._EnabledCoordUnits = 0;
-
-   tex_coord[0] = span->attrStart[FRAG_ATTRIB_TEX0][0]  * (info->smask + 1);
-   tex_step[0] = span->attrStepX[FRAG_ATTRIB_TEX0][0] * (info->smask + 1);
-   tex_coord[1] = span->attrStart[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
-   tex_step[1] = span->attrStepX[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
-   /* span->attrStart[FRAG_ATTRIB_TEX0][2] only if 3D-texturing, here only 2D */
-   tex_coord[2] = span->attrStart[FRAG_ATTRIB_TEX0][3];
-   tex_step[2] = span->attrStepX[FRAG_ATTRIB_TEX0][3];
-
-   switch (info->filter) {
-   case GL_NEAREST:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (5) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch(info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (6) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-
-   case GL_LINEAR:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (7) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (8) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-   }
-   
-   ASSERT(span->arrayMask & SPAN_RGBA);
-   _swrast_write_rgba_span(ctx, span);
-
-#undef SPAN_NEAREST
-#undef SPAN_LINEAR
-
-   /* restore state */
-   ctx->Texture._EnabledCoordUnits = texEnableSave;
-}
-
-
-/*
- * Render an perspective corrected RGB/RGBA textured triangle.
- * The Q (aka V in Mesa) coordinate must be zero such that the divide
- * by interpolated Q/W comes out right.
- *
- */
-#define NAME persp_textured_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_ATTRIBS 1
-
-#define SETUP_CODE							\
-   struct persp_info info;						\
-   const struct gl_texture_unit *unit = ctx->Texture.Unit+0;		\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-      obj->Image[0][obj->BaseLevel];			 		\
-   info.texture = (const GLchan *) texImg->Data;			\
-   info.twidth_log2 = texImg->WidthLog2;				\
-   info.smask = texImg->Width - 1;					\
-   info.tmask = texImg->Height - 1;					\
-   info.format = texImg->TexFormat;					\
-   info.filter = obj->MinFilter;					\
-   info.envmode = unit->EnvMode;					\
-   info.er = 0;					\
-   info.eg = 0;					\
-   info.eb = 0;					\
-									\
-   if (info.envmode == GL_BLEND) {					\
-      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
-      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
-      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
-      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
-      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
-   }									\
-   if (!info.texture) {							\
-      /* this shouldn't happen */					\
-      return;								\
-   }									\
-									\
-   switch (info.format) {						\
-   case MESA_FORMAT_RGB888:						\
-      info.tbytesline = texImg->Width * 3;				\
-      break;								\
-   case MESA_FORMAT_RGBA8888:						\
-      info.tbytesline = texImg->Width * 4;				\
-      break;								\
-   default:								\
-      _mesa_problem(NULL, "Bad texture format in persp_textured_triangle");\
-      return;								\
-   }									\
-   info.tsize = texImg->Height * info.tbytesline;
-
-#define RENDER_SPAN( span )			\
-   span.interpMask &= ~SPAN_RGBA;		\
-   span.arrayMask |= SPAN_RGBA;			\
-   fast_persp_span(ctx, &span, &info);
-
-#include "s_tritemp.h"
-
-#endif /*CHAN_TYPE != GL_FLOAT*/
-
-
-
-/*
- * Render an RGBA triangle with arbitrary attributes.
- */
-#define NAME general_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_ATTRIBS 1
-#define RENDER_SPAN( span )   _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-
-/*
- * Special tri function for occlusion testing
- */
-#define NAME occlusion_zless_triangle
-#define INTERP_Z 1
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_DepthBuffer;		\
-   struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;	\
-   ASSERT(ctx->Depth.Test);						\
-   ASSERT(!ctx->Depth.Mask);						\
-   ASSERT(ctx->Depth.Func == GL_LESS);					\
-   if (!q) {								\
-      return;								\
-   }
-#define RENDER_SPAN( span )						\
-   if (rb->Format == MESA_FORMAT_Z16) {					\
-      GLuint i;								\
-      const GLushort *zRow = (const GLushort *)				\
-         rb->GetPointer(ctx, rb, span.x, span.y);			\
-      for (i = 0; i < span.end; i++) {					\
-         GLuint z = FixedToDepth(span.z);				\
-         if (z < zRow[i]) {						\
-            q->Result++;						\
-         }								\
-         span.z += span.zStep;						\
-      }									\
-   }									\
-   else {								\
-      GLuint i;								\
-      const GLuint *zRow = (const GLuint *)				\
-         rb->GetPointer(ctx, rb, span.x, span.y);			\
-      for (i = 0; i < span.end; i++) {					\
-         if ((GLuint)span.z < zRow[i]) {				\
-            q->Result++;						\
-         }								\
-         span.z += span.zStep;						\
-      }									\
-   }
-#include "s_tritemp.h"
-
-
-
-static void
-nodraw_triangle( struct gl_context *ctx,
-                 const SWvertex *v0,
-                 const SWvertex *v1,
-                 const SWvertex *v2 )
-{
-   (void) (ctx && v0 && v1 && v2);
-}
-
-
-/*
- * This is used when separate specular color is enabled, but not
- * texturing.  We add the specular color to the primary color,
- * draw the triangle, then restore the original primary color.
- * Inefficient, but seldom needed.
- */
-void
-_swrast_add_spec_terms_triangle(struct gl_context *ctx, const SWvertex *v0,
-                                const SWvertex *v1, const SWvertex *v2)
-{
-   SWvertex *ncv0 = (SWvertex *)v0; /* drop const qualifier */
-   SWvertex *ncv1 = (SWvertex *)v1;
-   SWvertex *ncv2 = (SWvertex *)v2;
-   GLfloat rSum, gSum, bSum;
-   GLchan cSave[3][4];
-
-   /* save original colors */
-   COPY_CHAN4( cSave[0], ncv0->color );
-   COPY_CHAN4( cSave[1], ncv1->color );
-   COPY_CHAN4( cSave[2], ncv2->color );
-   /* sum v0 */
-   rSum = CHAN_TO_FLOAT(ncv0->color[0]) + ncv0->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv0->color[1]) + ncv0->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv0->color[2]) + ncv0->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[2], bSum);
-   /* sum v1 */
-   rSum = CHAN_TO_FLOAT(ncv1->color[0]) + ncv1->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv1->color[1]) + ncv1->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv1->color[2]) + ncv1->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[2], bSum);
-   /* sum v2 */
-   rSum = CHAN_TO_FLOAT(ncv2->color[0]) + ncv2->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv2->color[1]) + ncv2->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv2->color[2]) + ncv2->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[2], bSum);
-   /* draw */
-   SWRAST_CONTEXT(ctx)->SpecTriangle( ctx, ncv0, ncv1, ncv2 );
-   /* restore original colors */
-   COPY_CHAN4( ncv0->color, cSave[0] );
-   COPY_CHAN4( ncv1->color, cSave[1] );
-   COPY_CHAN4( ncv2->color, cSave[2] );
-}
-
-
-
-#ifdef DEBUG
-
-/* record the current triangle function name */
-const char *_mesa_triFuncName = NULL;
-
-#define USE(triFunc)				\
-do {						\
-    _mesa_triFuncName = #triFunc;		\
-    /*printf("%s\n", _mesa_triFuncName);*/	\
-    swrast->Triangle = triFunc;			\
-} while (0)
-
-#else
-
-#define USE(triFunc)  swrast->Triangle = triFunc;
-
-#endif
-
-
-
-
-/*
- * Determine which triangle rendering function to use given the current
- * rendering context.
- *
- * Please update the summary flag _SWRAST_NEW_TRIANGLE if you add or
- * remove tests to this code.
- */
-void
-_swrast_choose_triangle( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   if (ctx->Polygon.CullFlag &&
-       ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) {
-      USE(nodraw_triangle);
-      return;
-   }
-
-   if (ctx->RenderMode==GL_RENDER) {
-
-      if (ctx->Polygon.SmoothFlag) {
-         _swrast_set_aa_triangle_function(ctx);
-         ASSERT(swrast->Triangle);
-         return;
-      }
-
-      /* special case for occlusion testing */
-      if (ctx->Query.CurrentOcclusionObject &&
-          ctx->Depth.Test &&
-          ctx->Depth.Mask == GL_FALSE &&
-          ctx->Depth.Func == GL_LESS &&
-          !ctx->Stencil._Enabled) {
-         if (ctx->Color.ColorMask[0][0] == 0 &&
-	     ctx->Color.ColorMask[0][1] == 0 &&
-	     ctx->Color.ColorMask[0][2] == 0 &&
-	     ctx->Color.ColorMask[0][3] == 0) {
-            USE(occlusion_zless_triangle);
-            return;
-         }
-      }
-
-      /*
-       * XXX should examine swrast->_ActiveAttribMask to determine what
-       * needs to be interpolated.
-       */
-      if (ctx->Texture._EnabledCoordUnits ||
-          ctx->FragmentProgram._Current ||
-          ctx->ATIFragmentShader._Enabled ||
-          _mesa_need_secondary_color(ctx) ||
-          swrast->_FogEnabled) {
-         /* Ugh, we do a _lot_ of tests to pick the best textured tri func */
-         const struct gl_texture_object *texObj2D;
-         const struct gl_texture_image *texImg;
-         GLenum minFilter, magFilter, envMode;
-         gl_format format;
-         texObj2D = ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];
-
-         texImg = texObj2D ? texObj2D->Image[0][texObj2D->BaseLevel] : NULL;
-         format = texImg ? texImg->TexFormat : MESA_FORMAT_NONE;
-         minFilter = texObj2D ? texObj2D->MinFilter : GL_NONE;
-         magFilter = texObj2D ? texObj2D->MagFilter : GL_NONE;
-         envMode = ctx->Texture.Unit[0].EnvMode;
-
-         /* First see if we can use an optimized 2-D texture function */
-         if (ctx->Texture._EnabledCoordUnits == 0x1
-             && !ctx->FragmentProgram._Current
-             && !ctx->ATIFragmentShader._Enabled
-             && ctx->Texture._EnabledUnits == 0x1
-             && ctx->Texture.Unit[0]._ReallyEnabled == TEXTURE_2D_BIT
-             && texObj2D->WrapS == GL_REPEAT
-             && texObj2D->WrapT == GL_REPEAT
-             && texObj2D->_Swizzle == SWIZZLE_NOOP
-             && texImg->_IsPowerOfTwo
-             && texImg->Border == 0
-             && texImg->Width == texImg->RowStride
-             && (format == MESA_FORMAT_RGB888 || format == MESA_FORMAT_RGBA8888)
-             && minFilter == magFilter
-             && ctx->Light.Model.ColorControl == GL_SINGLE_COLOR
-             && !swrast->_FogEnabled
-             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE_EXT
-             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE4_NV) {
-	    if (ctx->Hint.PerspectiveCorrection==GL_FASTEST) {
-	       if (minFilter == GL_NEAREST
-		   && format == MESA_FORMAT_RGB888
-		   && (envMode == GL_REPLACE || envMode == GL_DECAL)
-		   && ((swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)
-			&& ctx->Depth.Func == GL_LESS
-			&& ctx->Depth.Mask == GL_TRUE)
-		       || swrast->_RasterMask == TEXTURE_BIT)
-		   && ctx->Polygon.StippleFlag == GL_FALSE
-                   && ctx->DrawBuffer->Visual.depthBits <= 16) {
-		  if (swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)) {
-		     USE(simple_z_textured_triangle);
-		  }
-		  else {
-		     USE(simple_textured_triangle);
-		  }
-	       }
-	       else {
-#if CHAN_BITS != 8
-                  USE(general_triangle);
-#else
-                  if (format == MESA_FORMAT_RGBA8888 && !_mesa_little_endian()) {
-                     /* We only handle RGBA8888 correctly on little endian
-                      * in the optimized code above.
-                      */
-                     USE(general_triangle);
-                  }
-                  else {
-                     USE(affine_textured_triangle);
-                 }
-#endif
-	       }
-	    }
-	    else {
-#if CHAN_BITS != 8
-               USE(general_triangle);
-#else
-               USE(persp_textured_triangle);
-#endif
-	    }
-	 }
-         else {
-            /* general case textured triangles */
-            USE(general_triangle);
-         }
-      }
-      else {
-         ASSERT(!swrast->_FogEnabled);
-         ASSERT(!_mesa_need_secondary_color(ctx));
-	 if (ctx->Light.ShadeModel==GL_SMOOTH) {
-	    /* smooth shaded, no texturing, stippled or some raster ops */
-#if CHAN_BITS != 8
-               USE(general_triangle);
-#else
-               USE(smooth_rgba_triangle);
-#endif
-	 }
-	 else {
-	    /* flat shaded, no texturing, stippled or some raster ops */
-#if CHAN_BITS != 8
-            USE(general_triangle);
-#else
-            USE(flat_rgba_triangle);
-#endif
-	 }
-      }
-   }
-   else if (ctx->RenderMode==GL_FEEDBACK) {
-      USE(_swrast_feedback_triangle);
-   }
-   else {
-      /* GL_SELECT mode */
-      USE(_swrast_select_triangle);
-   }
-}
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.3
+ *
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/*
+ * When the device driver doesn't implement triangle rasterization it
+ * can hook in _swrast_Triangle, which eventually calls one of these
+ * functions to draw triangles.
+ */
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/colormac.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/state.h"
+#include "program/prog_instruction.h"
+
+#include "s_aatriangle.h"
+#include "s_context.h"
+#include "s_feedback.h"
+#include "s_span.h"
+#include "s_triangle.h"
+
+
+/**
+ * Test if a triangle should be culled.  Used for feedback and selection mode.
+ * \return GL_TRUE if the triangle is to be culled, GL_FALSE otherwise.
+ */
+GLboolean
+_swrast_culltriangle( struct gl_context *ctx,
+                      const SWvertex *v0,
+                      const SWvertex *v1,
+                      const SWvertex *v2 )
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   GLfloat ex = v1->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
+   GLfloat ey = v1->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
+   GLfloat fx = v2->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
+   GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
+   GLfloat c = ex*fy-ey*fx;
+
+   if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign <= 0.0F)
+      return GL_FALSE;
+
+   return GL_TRUE;
+}
+
+
+
+/*
+ * Render a flat-shaded RGBA triangle.
+ */
+#define NAME flat_rgba_triangle
+#define INTERP_Z 1
+#define SETUP_CODE				\
+   ASSERT(ctx->Texture._EnabledCoordUnits == 0);\
+   ASSERT(ctx->Light.ShadeModel==GL_FLAT);	\
+   span.interpMask |= SPAN_RGBA;		\
+   span.red = ChanToFixed(v2->color[0]);	\
+   span.green = ChanToFixed(v2->color[1]);	\
+   span.blue = ChanToFixed(v2->color[2]);	\
+   span.alpha = ChanToFixed(v2->color[3]);	\
+   span.redStep = 0;				\
+   span.greenStep = 0;				\
+   span.blueStep = 0;				\
+   span.alphaStep = 0;
+#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
+#include "s_tritemp.h"
+
+
+
+/*
+ * Render a smooth-shaded RGBA triangle.
+ */
+#define NAME smooth_rgba_triangle
+#define INTERP_Z 1
+#define INTERP_RGB 1
+#define INTERP_ALPHA 1
+#define SETUP_CODE				\
+   {						\
+      /* texturing must be off */		\
+      ASSERT(ctx->Texture._EnabledCoordUnits == 0);	\
+      ASSERT(ctx->Light.ShadeModel==GL_SMOOTH);	\
+   }
+#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
+#include "s_tritemp.h"
+
+
+
+/*
+ * Render an RGB, GL_DECAL, textured triangle.
+ * Interpolate S,T only w/out mipmapping or perspective correction.
+ *
+ * No fog.  No depth testing.
+ */
+#define NAME simple_textured_triangle
+#define INTERP_INT_TEX 1
+#define S_SCALE twidth
+#define T_SCALE theight
+
+#define SETUP_CODE							\
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
+   const struct gl_texture_object *obj = 				\
+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
+   const struct gl_texture_image *texImg =				\
+      obj->Image[0][obj->BaseLevel];					\
+   const GLfloat twidth = (GLfloat) texImg->Width;			\
+   const GLfloat theight = (GLfloat) texImg->Height;			\
+   const GLint twidth_log2 = texImg->WidthLog2;				\
+   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
+   const GLint smask = texImg->Width - 1;				\
+   const GLint tmask = texImg->Height - 1;				\
+   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
+   if (!rb || !texture) {						\
+      return;								\
+   }
+
+#define RENDER_SPAN( span )						\
+   GLuint i;								\
+   GLubyte rgb[MAX_WIDTH][3];						\
+   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
+   span.intTex[1] -= FIXED_HALF;					\
+   for (i = 0; i < span.end; i++) {					\
+      GLint s = FixedToInt(span.intTex[0]) & smask;			\
+      GLint t = FixedToInt(span.intTex[1]) & tmask;			\
+      GLint pos = (t << twidth_log2) + s;				\
+      pos = pos + pos + pos;  /* multiply by 3 */			\
+      rgb[i][RCOMP] = texture[pos+2];					\
+      rgb[i][GCOMP] = texture[pos+1];					\
+      rgb[i][BCOMP] = texture[pos+0];					\
+      span.intTex[0] += span.intTexStep[0];				\
+      span.intTex[1] += span.intTexStep[1];				\
+   }									\
+   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, NULL);
+
+#include "s_tritemp.h"
+
+
+
+/*
+ * Render an RGB, GL_DECAL, textured triangle.
+ * Interpolate S,T, GL_LESS depth test, w/out mipmapping or
+ * perspective correction.
+ * Depth buffer bits must be <= sizeof(DEFAULT_SOFTWARE_DEPTH_TYPE)
+ *
+ * No fog.
+ */
+#define NAME simple_z_textured_triangle
+#define INTERP_Z 1
+#define DEPTH_TYPE DEFAULT_SOFTWARE_DEPTH_TYPE
+#define INTERP_INT_TEX 1
+#define S_SCALE twidth
+#define T_SCALE theight
+
+#define SETUP_CODE							\
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
+   const struct gl_texture_object *obj = 				\
+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
+   const struct gl_texture_image *texImg = 				\
+       obj->Image[0][obj->BaseLevel]; 					\
+   const GLfloat twidth = (GLfloat) texImg->Width;			\
+   const GLfloat theight = (GLfloat) texImg->Height;			\
+   const GLint twidth_log2 = texImg->WidthLog2;				\
+   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
+   const GLint smask = texImg->Width - 1;				\
+   const GLint tmask = texImg->Height - 1;				\
+   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
+   if (!rb || !texture) {						\
+      return;								\
+   }
+
+#define RENDER_SPAN( span )						\
+   GLuint i;				    				\
+   GLubyte rgb[MAX_WIDTH][3];						\
+   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
+   span.intTex[1] -= FIXED_HALF;					\
+   for (i = 0; i < span.end; i++) {					\
+      const GLuint z = FixedToDepth(span.z);				\
+      if (z < zRow[i]) {						\
+         GLint s = FixedToInt(span.intTex[0]) & smask;			\
+         GLint t = FixedToInt(span.intTex[1]) & tmask;			\
+         GLint pos = (t << twidth_log2) + s;				\
+         pos = pos + pos + pos;  /* multiply by 3 */			\
+         rgb[i][RCOMP] = texture[pos+2];				\
+         rgb[i][GCOMP] = texture[pos+1];				\
+         rgb[i][BCOMP] = texture[pos+0];				\
+         zRow[i] = z;							\
+         span.array->mask[i] = 1;					\
+      }									\
+      else {								\
+         span.array->mask[i] = 0;					\
+      }									\
+      span.intTex[0] += span.intTexStep[0];				\
+      span.intTex[1] += span.intTexStep[1];				\
+      span.z += span.zStep;						\
+   }									\
+   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, span.array->mask);
+
+#include "s_tritemp.h"
+
+
+#if CHAN_TYPE != GL_FLOAT
+
+struct affine_info
+{
+   GLenum filter;
+   GLenum format;
+   GLenum envmode;
+   GLint smask, tmask;
+   GLint twidth_log2;
+   const GLchan *texture;
+   GLfixed er, eg, eb, ea;
+   GLint tbytesline, tsize;
+};
+
+
+static INLINE GLint
+ilerp(GLint t, GLint a, GLint b)
+{
+   return a + ((t * (b - a)) >> FIXED_SHIFT);
+}
+
+static INLINE GLint
+ilerp_2d(GLint ia, GLint ib, GLint v00, GLint v10, GLint v01, GLint v11)
+{
+   const GLint temp0 = ilerp(ia, v00, v10);
+   const GLint temp1 = ilerp(ia, v01, v11);
+   return ilerp(ib, temp0, temp1);
+}
+
+
+/* This function can handle GL_NEAREST or GL_LINEAR sampling of 2D RGB or RGBA
+ * textures with GL_REPLACE, GL_MODULATE, GL_BLEND, GL_DECAL or GL_ADD
+ * texture env modes.
+ */
+static INLINE void
+affine_span(struct gl_context *ctx, SWspan *span,
+            struct affine_info *info)
+{
+   GLchan sample[4];  /* the filtered texture sample */
+   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
+
+   /* Instead of defining a function for each mode, a test is done
+    * between the outer and inner loops. This is to reduce code size
+    * and complexity. Observe that an optimizing compiler kills
+    * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
+    */
+
+#define NEAREST_RGB		\
+   sample[RCOMP] = tex00[2];	\
+   sample[GCOMP] = tex00[1];	\
+   sample[BCOMP] = tex00[0];	\
+   sample[ACOMP] = CHAN_MAX;
+
+#define LINEAR_RGB							\
+   sample[RCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
+   sample[GCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
+   sample[BCOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0]);\
+   sample[ACOMP] = CHAN_MAX;
+
+#define NEAREST_RGBA  \
+   sample[RCOMP] = tex00[3];	\
+   sample[GCOMP] = tex00[2];	\
+   sample[BCOMP] = tex00[1];	\
+   sample[ACOMP] = tex00[0];
+
+#define LINEAR_RGBA							\
+   sample[RCOMP] = ilerp_2d(sf, tf, tex00[3], tex01[3], tex10[3], tex11[3]);\
+   sample[GCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
+   sample[BCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
+   sample[ACOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0])
+
+#define MODULATE							  \
+   dest[RCOMP] = span->red   * (sample[RCOMP] + 1u) >> (FIXED_SHIFT + 8); \
+   dest[GCOMP] = span->green * (sample[GCOMP] + 1u) >> (FIXED_SHIFT + 8); \
+   dest[BCOMP] = span->blue  * (sample[BCOMP] + 1u) >> (FIXED_SHIFT + 8); \
+   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1u) >> (FIXED_SHIFT + 8)
+
+#define DECAL								\
+   dest[RCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->red +		\
+               ((sample[ACOMP] + 1) * sample[RCOMP] << FIXED_SHIFT))	\
+               >> (FIXED_SHIFT + 8);					\
+   dest[GCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->green +		\
+               ((sample[ACOMP] + 1) * sample[GCOMP] << FIXED_SHIFT))	\
+               >> (FIXED_SHIFT + 8);					\
+   dest[BCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->blue +		\
+               ((sample[ACOMP] + 1) * sample[BCOMP] << FIXED_SHIFT))	\
+               >> (FIXED_SHIFT + 8);					\
+   dest[ACOMP] = FixedToInt(span->alpha)
+
+#define BLEND								\
+   dest[RCOMP] = ((CHAN_MAX - sample[RCOMP]) * span->red		\
+               + (sample[RCOMP] + 1) * info->er) >> (FIXED_SHIFT + 8);	\
+   dest[GCOMP] = ((CHAN_MAX - sample[GCOMP]) * span->green		\
+               + (sample[GCOMP] + 1) * info->eg) >> (FIXED_SHIFT + 8);	\
+   dest[BCOMP] = ((CHAN_MAX - sample[BCOMP]) * span->blue		\
+               + (sample[BCOMP] + 1) * info->eb) >> (FIXED_SHIFT + 8);	\
+   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8)
+
+#define REPLACE  COPY_CHAN4(dest, sample)
+
+#define ADD								\
+   {									\
+      GLint rSum = FixedToInt(span->red)   + (GLint) sample[RCOMP];	\
+      GLint gSum = FixedToInt(span->green) + (GLint) sample[GCOMP];	\
+      GLint bSum = FixedToInt(span->blue)  + (GLint) sample[BCOMP];	\
+      dest[RCOMP] = MIN2(rSum, CHAN_MAX);				\
+      dest[GCOMP] = MIN2(gSum, CHAN_MAX);				\
+      dest[BCOMP] = MIN2(bSum, CHAN_MAX);				\
+      dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8); \
+  }
+
+/* shortcuts */
+
+#define NEAREST_RGB_REPLACE		\
+   NEAREST_RGB;				\
+   dest[0] = sample[0];			\
+   dest[1] = sample[1];			\
+   dest[2] = sample[2];			\
+   dest[3] = FixedToInt(span->alpha);
+
+#define NEAREST_RGBA_REPLACE  \
+   dest[RCOMP] = tex00[3]; \
+   dest[GCOMP] = tex00[2]; \
+   dest[BCOMP] = tex00[1]; \
+   dest[ACOMP] = tex00[0]
+
+#define SPAN_NEAREST(DO_TEX, COMPS)					\
+	for (i = 0; i < span->end; i++) {				\
+           /* Isn't it necessary to use FixedFloor below?? */		\
+           GLint s = FixedToInt(span->intTex[0]) & info->smask;		\
+           GLint t = FixedToInt(span->intTex[1]) & info->tmask;		\
+           GLint pos = (t << info->twidth_log2) + s;			\
+           const GLchan *tex00 = info->texture + COMPS * pos;		\
+           DO_TEX;							\
+           span->red += span->redStep;					\
+	   span->green += span->greenStep;				\
+           span->blue += span->blueStep;				\
+	   span->alpha += span->alphaStep;				\
+	   span->intTex[0] += span->intTexStep[0];			\
+	   span->intTex[1] += span->intTexStep[1];			\
+           dest += 4;							\
+	}
+
+#define SPAN_LINEAR(DO_TEX, COMPS)					\
+	for (i = 0; i < span->end; i++) {				\
+           /* Isn't it necessary to use FixedFloor below?? */		\
+           const GLint s = FixedToInt(span->intTex[0]) & info->smask;	\
+           const GLint t = FixedToInt(span->intTex[1]) & info->tmask;	\
+           const GLfixed sf = span->intTex[0] & FIXED_FRAC_MASK;	\
+           const GLfixed tf = span->intTex[1] & FIXED_FRAC_MASK;	\
+           const GLint pos = (t << info->twidth_log2) + s;		\
+           const GLchan *tex00 = info->texture + COMPS * pos;		\
+           const GLchan *tex10 = tex00 + info->tbytesline;		\
+           const GLchan *tex01 = tex00 + COMPS;				\
+           const GLchan *tex11 = tex10 + COMPS;				\
+           if (t == info->tmask) {					\
+              tex10 -= info->tsize;					\
+              tex11 -= info->tsize;					\
+           }								\
+           if (s == info->smask) {					\
+              tex01 -= info->tbytesline;				\
+              tex11 -= info->tbytesline;				\
+           }								\
+           DO_TEX;							\
+           span->red += span->redStep;					\
+	   span->green += span->greenStep;				\
+           span->blue += span->blueStep;				\
+	   span->alpha += span->alphaStep;				\
+	   span->intTex[0] += span->intTexStep[0];			\
+	   span->intTex[1] += span->intTexStep[1];			\
+           dest += 4;							\
+	}
+
+
+   GLuint i;
+   GLchan *dest = span->array->rgba[0];
+
+   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */
+   ctx->Texture._EnabledCoordUnits = 0x0;
+
+   span->intTex[0] -= FIXED_HALF;
+   span->intTex[1] -= FIXED_HALF;
+   switch (info->filter) {
+   case GL_NEAREST:
+      switch (info->format) {
+      case MESA_FORMAT_RGB888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
+            break;
+         case GL_DECAL:
+         case GL_REPLACE:
+            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
+            break;
+         case GL_BLEND:
+            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
+            break;
+         case GL_ADD:
+            SPAN_NEAREST(NEAREST_RGB;ADD,3);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode in SPAN_LINEAR");
+            return;
+         }
+         break;
+      case MESA_FORMAT_RGBA8888:
+         switch(info->envmode) {
+         case GL_MODULATE:
+            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
+            break;
+         case GL_DECAL:
+            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
+            break;
+         case GL_BLEND:
+            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
+            break;
+         case GL_ADD:
+            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
+            break;
+         case GL_REPLACE:
+            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (2) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      }
+      break;
+
+   case GL_LINEAR:
+      span->intTex[0] -= FIXED_HALF;
+      span->intTex[1] -= FIXED_HALF;
+      switch (info->format) {
+      case MESA_FORMAT_RGB888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
+            break;
+         case GL_DECAL:
+         case GL_REPLACE:
+            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
+            break;
+         case GL_BLEND:
+            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
+            break;
+         case GL_ADD:
+            SPAN_LINEAR(LINEAR_RGB;ADD,3);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (3) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      case MESA_FORMAT_RGBA8888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
+            break;
+         case GL_DECAL:
+            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
+            break;
+         case GL_BLEND:
+            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
+            break;
+         case GL_ADD:
+            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
+            break;
+         case GL_REPLACE:
+            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (4) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      }
+      break;
+   }
+   span->interpMask &= ~SPAN_RGBA;
+   ASSERT(span->arrayMask & SPAN_RGBA);
+
+   _swrast_write_rgba_span(ctx, span);
+
+   /* re-enable texture units */
+   ctx->Texture._EnabledCoordUnits = texEnableSave;
+
+#undef SPAN_NEAREST
+#undef SPAN_LINEAR
+}
+
+
+
+/*
+ * Render an RGB/RGBA textured triangle without perspective correction.
+ */
+#define NAME affine_textured_triangle
+#define INTERP_Z 1
+#define INTERP_RGB 1
+#define INTERP_ALPHA 1
+#define INTERP_INT_TEX 1
+#define S_SCALE twidth
+#define T_SCALE theight
+
+#define SETUP_CODE							\
+   struct affine_info info;						\
+   struct gl_texture_unit *unit = ctx->Texture.Unit+0;			\
+   const struct gl_texture_object *obj = 				\
+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
+   const struct gl_texture_image *texImg = 				\
+      obj->Image[0][obj->BaseLevel]; 					\
+   const GLfloat twidth = (GLfloat) texImg->Width;			\
+   const GLfloat theight = (GLfloat) texImg->Height;			\
+   info.texture = (const GLchan *) texImg->Data;			\
+   info.twidth_log2 = texImg->WidthLog2;				\
+   info.smask = texImg->Width - 1;					\
+   info.tmask = texImg->Height - 1;					\
+   info.format = texImg->TexFormat;					\
+   info.filter = obj->Sampler.MinFilter;				\
+   info.envmode = unit->EnvMode;					\
+   info.er = 0;					\
+   info.eg = 0;					\
+   info.eb = 0;					\
+   span.arrayMask |= SPAN_RGBA;						\
+									\
+   if (info.envmode == GL_BLEND) {					\
+      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
+      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
+      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
+      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
+      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
+   }									\
+   if (!info.texture) {							\
+      /* this shouldn't happen */					\
+      return;								\
+   }									\
+									\
+   switch (info.format) {						\
+   case MESA_FORMAT_RGB888:						\
+      info.tbytesline = texImg->Width * 3;				\
+      break;								\
+   case MESA_FORMAT_RGBA8888:						\
+      info.tbytesline = texImg->Width * 4;				\
+      break;								\
+   default:								\
+      _mesa_problem(NULL, "Bad texture format in affine_texture_triangle");\
+      return;								\
+   }									\
+   info.tsize = texImg->Height * info.tbytesline;
+
+#define RENDER_SPAN( span )   affine_span(ctx, &span, &info);
+
+#include "s_tritemp.h"
+
+
+
+struct persp_info
+{
+   GLenum filter;
+   GLenum format;
+   GLenum envmode;
+   GLint smask, tmask;
+   GLint twidth_log2;
+   const GLchan *texture;
+   GLfixed er, eg, eb, ea;   /* texture env color */
+   GLint tbytesline, tsize;
+};
+
+
+static INLINE void
+fast_persp_span(struct gl_context *ctx, SWspan *span,
+		struct persp_info *info)
+{
+   GLchan sample[4];  /* the filtered texture sample */
+
+  /* Instead of defining a function for each mode, a test is done
+   * between the outer and inner loops. This is to reduce code size
+   * and complexity. Observe that an optimizing compiler kills
+   * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
+   */
+#define SPAN_NEAREST(DO_TEX,COMP)					\
+	for (i = 0; i < span->end; i++) {				\
+           GLdouble invQ = tex_coord[2] ?				\
+                                 (1.0 / tex_coord[2]) : 1.0;            \
+           GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);		\
+           GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);		\
+           GLint s = IFLOOR(s_tmp) & info->smask;	        	\
+           GLint t = IFLOOR(t_tmp) & info->tmask;	        	\
+           GLint pos = (t << info->twidth_log2) + s;			\
+           const GLchan *tex00 = info->texture + COMP * pos;		\
+           DO_TEX;							\
+           span->red += span->redStep;					\
+	   span->green += span->greenStep;				\
+           span->blue += span->blueStep;				\
+	   span->alpha += span->alphaStep;				\
+	   tex_coord[0] += tex_step[0];					\
+	   tex_coord[1] += tex_step[1];					\
+	   tex_coord[2] += tex_step[2];					\
+           dest += 4;							\
+	}
+
+#define SPAN_LINEAR(DO_TEX,COMP)					\
+	for (i = 0; i < span->end; i++) {				\
+           GLdouble invQ = tex_coord[2] ?				\
+                                 (1.0 / tex_coord[2]) : 1.0;            \
+           const GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);	\
+           const GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);	\
+           const GLfixed s_fix = FloatToFixed(s_tmp) - FIXED_HALF;	\
+           const GLfixed t_fix = FloatToFixed(t_tmp) - FIXED_HALF;      \
+           const GLint s = FixedToInt(FixedFloor(s_fix)) & info->smask;	\
+           const GLint t = FixedToInt(FixedFloor(t_fix)) & info->tmask;	\
+           const GLfixed sf = s_fix & FIXED_FRAC_MASK;			\
+           const GLfixed tf = t_fix & FIXED_FRAC_MASK;			\
+           const GLint pos = (t << info->twidth_log2) + s;		\
+           const GLchan *tex00 = info->texture + COMP * pos;		\
+           const GLchan *tex10 = tex00 + info->tbytesline;		\
+           const GLchan *tex01 = tex00 + COMP;				\
+           const GLchan *tex11 = tex10 + COMP;				\
+           if (t == info->tmask) {					\
+              tex10 -= info->tsize;					\
+              tex11 -= info->tsize;					\
+           }								\
+           if (s == info->smask) {					\
+              tex01 -= info->tbytesline;				\
+              tex11 -= info->tbytesline;				\
+           }								\
+           DO_TEX;							\
+           span->red   += span->redStep;				\
+	   span->green += span->greenStep;				\
+           span->blue  += span->blueStep;				\
+	   span->alpha += span->alphaStep;				\
+	   tex_coord[0] += tex_step[0];					\
+	   tex_coord[1] += tex_step[1];					\
+	   tex_coord[2] += tex_step[2];					\
+           dest += 4;							\
+	}
+
+   GLuint i;
+   GLfloat tex_coord[3], tex_step[3];
+   GLchan *dest = span->array->rgba[0];
+
+   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
+   ctx->Texture._EnabledCoordUnits = 0;
+
+   tex_coord[0] = span->attrStart[FRAG_ATTRIB_TEX0][0]  * (info->smask + 1);
+   tex_step[0] = span->attrStepX[FRAG_ATTRIB_TEX0][0] * (info->smask + 1);
+   tex_coord[1] = span->attrStart[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
+   tex_step[1] = span->attrStepX[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
+   /* span->attrStart[FRAG_ATTRIB_TEX0][2] only if 3D-texturing, here only 2D */
+   tex_coord[2] = span->attrStart[FRAG_ATTRIB_TEX0][3];
+   tex_step[2] = span->attrStepX[FRAG_ATTRIB_TEX0][3];
+
+   switch (info->filter) {
+   case GL_NEAREST:
+      switch (info->format) {
+      case MESA_FORMAT_RGB888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
+            break;
+         case GL_DECAL:
+         case GL_REPLACE:
+            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
+            break;
+         case GL_BLEND:
+            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
+            break;
+         case GL_ADD:
+            SPAN_NEAREST(NEAREST_RGB;ADD,3);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (5) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      case MESA_FORMAT_RGBA8888:
+         switch(info->envmode) {
+         case GL_MODULATE:
+            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
+            break;
+         case GL_DECAL:
+            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
+            break;
+         case GL_BLEND:
+            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
+            break;
+         case GL_ADD:
+            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
+            break;
+         case GL_REPLACE:
+            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (6) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      }
+      break;
+
+   case GL_LINEAR:
+      switch (info->format) {
+      case MESA_FORMAT_RGB888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
+            break;
+         case GL_DECAL:
+         case GL_REPLACE:
+            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
+            break;
+         case GL_BLEND:
+            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
+            break;
+         case GL_ADD:
+            SPAN_LINEAR(LINEAR_RGB;ADD,3);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (7) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      case MESA_FORMAT_RGBA8888:
+         switch (info->envmode) {
+         case GL_MODULATE:
+            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
+            break;
+         case GL_DECAL:
+            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
+            break;
+         case GL_BLEND:
+            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
+            break;
+         case GL_ADD:
+            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
+            break;
+         case GL_REPLACE:
+            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
+            break;
+         default:
+            _mesa_problem(ctx, "bad tex env mode (8) in SPAN_LINEAR");
+            return;
+         }
+         break;
+      }
+      break;
+   }
+   
+   ASSERT(span->arrayMask & SPAN_RGBA);
+   _swrast_write_rgba_span(ctx, span);
+
+#undef SPAN_NEAREST
+#undef SPAN_LINEAR
+
+   /* restore state */
+   ctx->Texture._EnabledCoordUnits = texEnableSave;
+}
+
+
+/*
+ * Render an perspective corrected RGB/RGBA textured triangle.
+ * The Q (aka V in Mesa) coordinate must be zero such that the divide
+ * by interpolated Q/W comes out right.
+ *
+ */
+#define NAME persp_textured_triangle
+#define INTERP_Z 1
+#define INTERP_RGB 1
+#define INTERP_ALPHA 1
+#define INTERP_ATTRIBS 1
+
+#define SETUP_CODE							\
+   struct persp_info info;						\
+   const struct gl_texture_unit *unit = ctx->Texture.Unit+0;		\
+   const struct gl_texture_object *obj = 				\
+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
+   const struct gl_texture_image *texImg = 				\
+      obj->Image[0][obj->BaseLevel];			 		\
+   info.texture = (const GLchan *) texImg->Data;			\
+   info.twidth_log2 = texImg->WidthLog2;				\
+   info.smask = texImg->Width - 1;					\
+   info.tmask = texImg->Height - 1;					\
+   info.format = texImg->TexFormat;					\
+   info.filter = obj->Sampler.MinFilter;				\
+   info.envmode = unit->EnvMode;					\
+   info.er = 0;					\
+   info.eg = 0;					\
+   info.eb = 0;					\
+									\
+   if (info.envmode == GL_BLEND) {					\
+      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
+      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
+      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
+      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
+      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
+   }									\
+   if (!info.texture) {							\
+      /* this shouldn't happen */					\
+      return;								\
+   }									\
+									\
+   switch (info.format) {						\
+   case MESA_FORMAT_RGB888:						\
+      info.tbytesline = texImg->Width * 3;				\
+      break;								\
+   case MESA_FORMAT_RGBA8888:						\
+      info.tbytesline = texImg->Width * 4;				\
+      break;								\
+   default:								\
+      _mesa_problem(NULL, "Bad texture format in persp_textured_triangle");\
+      return;								\
+   }									\
+   info.tsize = texImg->Height * info.tbytesline;
+
+#define RENDER_SPAN( span )			\
+   span.interpMask &= ~SPAN_RGBA;		\
+   span.arrayMask |= SPAN_RGBA;			\
+   fast_persp_span(ctx, &span, &info);
+
+#include "s_tritemp.h"
+
+#endif /*CHAN_TYPE != GL_FLOAT*/
+
+
+
+/*
+ * Render an RGBA triangle with arbitrary attributes.
+ */
+#define NAME general_triangle
+#define INTERP_Z 1
+#define INTERP_RGB 1
+#define INTERP_ALPHA 1
+#define INTERP_ATTRIBS 1
+#define RENDER_SPAN( span )   _swrast_write_rgba_span(ctx, &span);
+#include "s_tritemp.h"
+
+
+
+
+/*
+ * Special tri function for occlusion testing
+ */
+#define NAME occlusion_zless_triangle
+#define INTERP_Z 1
+#define SETUP_CODE							\
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_DepthBuffer;		\
+   struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;	\
+   ASSERT(ctx->Depth.Test);						\
+   ASSERT(!ctx->Depth.Mask);						\
+   ASSERT(ctx->Depth.Func == GL_LESS);					\
+   if (!q) {								\
+      return;								\
+   }
+#define RENDER_SPAN( span )						\
+   if (rb->Format == MESA_FORMAT_Z16) {					\
+      GLuint i;								\
+      const GLushort *zRow = (const GLushort *)				\
+         rb->GetPointer(ctx, rb, span.x, span.y);			\
+      for (i = 0; i < span.end; i++) {					\
+         GLuint z = FixedToDepth(span.z);				\
+         if (z < zRow[i]) {						\
+            q->Result++;						\
+         }								\
+         span.z += span.zStep;						\
+      }									\
+   }									\
+   else {								\
+      GLuint i;								\
+      const GLuint *zRow = (const GLuint *)				\
+         rb->GetPointer(ctx, rb, span.x, span.y);			\
+      for (i = 0; i < span.end; i++) {					\
+         if ((GLuint)span.z < zRow[i]) {				\
+            q->Result++;						\
+         }								\
+         span.z += span.zStep;						\
+      }									\
+   }
+#include "s_tritemp.h"
+
+
+
+static void
+nodraw_triangle( struct gl_context *ctx,
+                 const SWvertex *v0,
+                 const SWvertex *v1,
+                 const SWvertex *v2 )
+{
+   (void) (ctx && v0 && v1 && v2);
+}
+
+
+/*
+ * This is used when separate specular color is enabled, but not
+ * texturing.  We add the specular color to the primary color,
+ * draw the triangle, then restore the original primary color.
+ * Inefficient, but seldom needed.
+ */
+void
+_swrast_add_spec_terms_triangle(struct gl_context *ctx, const SWvertex *v0,
+                                const SWvertex *v1, const SWvertex *v2)
+{
+   SWvertex *ncv0 = (SWvertex *)v0; /* drop const qualifier */
+   SWvertex *ncv1 = (SWvertex *)v1;
+   SWvertex *ncv2 = (SWvertex *)v2;
+   GLfloat rSum, gSum, bSum;
+   GLchan cSave[3][4];
+
+   /* save original colors */
+   COPY_CHAN4( cSave[0], ncv0->color );
+   COPY_CHAN4( cSave[1], ncv1->color );
+   COPY_CHAN4( cSave[2], ncv2->color );
+   /* sum v0 */
+   rSum = CHAN_TO_FLOAT(ncv0->color[0]) + ncv0->attrib[FRAG_ATTRIB_COL1][0];
+   gSum = CHAN_TO_FLOAT(ncv0->color[1]) + ncv0->attrib[FRAG_ATTRIB_COL1][1];
+   bSum = CHAN_TO_FLOAT(ncv0->color[2]) + ncv0->attrib[FRAG_ATTRIB_COL1][2];
+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[0], rSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[1], gSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[2], bSum);
+   /* sum v1 */
+   rSum = CHAN_TO_FLOAT(ncv1->color[0]) + ncv1->attrib[FRAG_ATTRIB_COL1][0];
+   gSum = CHAN_TO_FLOAT(ncv1->color[1]) + ncv1->attrib[FRAG_ATTRIB_COL1][1];
+   bSum = CHAN_TO_FLOAT(ncv1->color[2]) + ncv1->attrib[FRAG_ATTRIB_COL1][2];
+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[0], rSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[1], gSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[2], bSum);
+   /* sum v2 */
+   rSum = CHAN_TO_FLOAT(ncv2->color[0]) + ncv2->attrib[FRAG_ATTRIB_COL1][0];
+   gSum = CHAN_TO_FLOAT(ncv2->color[1]) + ncv2->attrib[FRAG_ATTRIB_COL1][1];
+   bSum = CHAN_TO_FLOAT(ncv2->color[2]) + ncv2->attrib[FRAG_ATTRIB_COL1][2];
+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[0], rSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[1], gSum);
+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[2], bSum);
+   /* draw */
+   SWRAST_CONTEXT(ctx)->SpecTriangle( ctx, ncv0, ncv1, ncv2 );
+   /* restore original colors */
+   COPY_CHAN4( ncv0->color, cSave[0] );
+   COPY_CHAN4( ncv1->color, cSave[1] );
+   COPY_CHAN4( ncv2->color, cSave[2] );
+}
+
+
+
+#ifdef DEBUG
+
+/* record the current triangle function name */
+const char *_mesa_triFuncName = NULL;
+
+#define USE(triFunc)				\
+do {						\
+    _mesa_triFuncName = #triFunc;		\
+    /*printf("%s\n", _mesa_triFuncName);*/	\
+    swrast->Triangle = triFunc;			\
+} while (0)
+
+#else
+
+#define USE(triFunc)  swrast->Triangle = triFunc;
+
+#endif
+
+
+
+
+/*
+ * Determine which triangle rendering function to use given the current
+ * rendering context.
+ *
+ * Please update the summary flag _SWRAST_NEW_TRIANGLE if you add or
+ * remove tests to this code.
+ */
+void
+_swrast_choose_triangle( struct gl_context *ctx )
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+
+   if (ctx->Polygon.CullFlag &&
+       ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) {
+      USE(nodraw_triangle);
+      return;
+   }
+
+   if (ctx->RenderMode==GL_RENDER) {
+
+      if (ctx->Polygon.SmoothFlag) {
+         _swrast_set_aa_triangle_function(ctx);
+         ASSERT(swrast->Triangle);
+         return;
+      }
+
+      /* special case for occlusion testing */
+      if (ctx->Query.CurrentOcclusionObject &&
+          ctx->Depth.Test &&
+          ctx->Depth.Mask == GL_FALSE &&
+          ctx->Depth.Func == GL_LESS &&
+          !ctx->Stencil._Enabled) {
+         if (ctx->Color.ColorMask[0][0] == 0 &&
+	     ctx->Color.ColorMask[0][1] == 0 &&
+	     ctx->Color.ColorMask[0][2] == 0 &&
+	     ctx->Color.ColorMask[0][3] == 0) {
+            USE(occlusion_zless_triangle);
+            return;
+         }
+      }
+
+      /*
+       * XXX should examine swrast->_ActiveAttribMask to determine what
+       * needs to be interpolated.
+       */
+      if (ctx->Texture._EnabledCoordUnits ||
+          ctx->FragmentProgram._Current ||
+          ctx->ATIFragmentShader._Enabled ||
+          _mesa_need_secondary_color(ctx) ||
+          swrast->_FogEnabled) {
+         /* Ugh, we do a _lot_ of tests to pick the best textured tri func */
+         const struct gl_texture_object *texObj2D;
+         const struct gl_texture_image *texImg;
+         GLenum minFilter, magFilter, envMode;
+         gl_format format;
+         texObj2D = ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];
+
+         texImg = texObj2D ? texObj2D->Image[0][texObj2D->BaseLevel] : NULL;
+         format = texImg ? texImg->TexFormat : MESA_FORMAT_NONE;
+         minFilter = texObj2D ? texObj2D->Sampler.MinFilter : GL_NONE;
+         magFilter = texObj2D ? texObj2D->Sampler.MagFilter : GL_NONE;
+         envMode = ctx->Texture.Unit[0].EnvMode;
+
+         /* First see if we can use an optimized 2-D texture function */
+         if (ctx->Texture._EnabledCoordUnits == 0x1
+             && !ctx->FragmentProgram._Current
+             && !ctx->ATIFragmentShader._Enabled
+             && ctx->Texture._EnabledUnits == 0x1
+             && ctx->Texture.Unit[0]._ReallyEnabled == TEXTURE_2D_BIT
+             && texObj2D->Sampler.WrapS == GL_REPEAT
+             && texObj2D->Sampler.WrapT == GL_REPEAT
+             && texObj2D->_Swizzle == SWIZZLE_NOOP
+             && texImg->_IsPowerOfTwo
+             && texImg->Border == 0
+             && texImg->Width == texImg->RowStride
+             && (format == MESA_FORMAT_RGB888 || format == MESA_FORMAT_RGBA8888)
+             && minFilter == magFilter
+             && ctx->Light.Model.ColorControl == GL_SINGLE_COLOR
+             && !swrast->_FogEnabled
+             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE_EXT
+             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE4_NV) {
+	    if (ctx->Hint.PerspectiveCorrection==GL_FASTEST) {
+	       if (minFilter == GL_NEAREST
+		   && format == MESA_FORMAT_RGB888
+		   && (envMode == GL_REPLACE || envMode == GL_DECAL)
+		   && ((swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)
+			&& ctx->Depth.Func == GL_LESS
+			&& ctx->Depth.Mask == GL_TRUE)
+		       || swrast->_RasterMask == TEXTURE_BIT)
+		   && ctx->Polygon.StippleFlag == GL_FALSE
+                   && ctx->DrawBuffer->Visual.depthBits <= 16) {
+		  if (swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)) {
+		     USE(simple_z_textured_triangle);
+		  }
+		  else {
+		     USE(simple_textured_triangle);
+		  }
+	       }
+	       else {
+#if CHAN_BITS != 8
+                  USE(general_triangle);
+#else
+                  if (format == MESA_FORMAT_RGBA8888 && !_mesa_little_endian()) {
+                     /* We only handle RGBA8888 correctly on little endian
+                      * in the optimized code above.
+                      */
+                     USE(general_triangle);
+                  }
+                  else {
+                     USE(affine_textured_triangle);
+                 }
+#endif
+	       }
+	    }
+	    else {
+#if CHAN_BITS != 8
+               USE(general_triangle);
+#else
+               USE(persp_textured_triangle);
+#endif
+	    }
+	 }
+         else {
+            /* general case textured triangles */
+            USE(general_triangle);
+         }
+      }
+      else {
+         ASSERT(!swrast->_FogEnabled);
+         ASSERT(!_mesa_need_secondary_color(ctx));
+	 if (ctx->Light.ShadeModel==GL_SMOOTH) {
+	    /* smooth shaded, no texturing, stippled or some raster ops */
+#if CHAN_BITS != 8
+               USE(general_triangle);
+#else
+               USE(smooth_rgba_triangle);
+#endif
+	 }
+	 else {
+	    /* flat shaded, no texturing, stippled or some raster ops */
+#if CHAN_BITS != 8
+            USE(general_triangle);
+#else
+            USE(flat_rgba_triangle);
+#endif
+	 }
+      }
+   }
+   else if (ctx->RenderMode==GL_FEEDBACK) {
+      USE(_swrast_feedback_triangle);
+   }
+   else {
+      /* GL_SELECT mode */
+      USE(_swrast_select_triangle);
+   }
+}