Merge branch 'master' into gallium-texture-transfer

Conflicts:

	src/mesa/state_tracker/st_cb_accum.c
	src/mesa/state_tracker/st_cb_drawpixels.c

17 files changed, 305 insertions, 209 deletions

diff --git a/src/mesa/swrast/s_accum.c b/src/mesa/swrast/s_accum.c
index ff741777e75..c6c7dbf5cf0 100644
--- a/src/mesa/swrast/s_accum.c
+++ b/src/mesa/swrast/s_accum.c
@@ -550,7 +550,7 @@ _swrast_Accum(GLcontext *ctx, GLenum op, GLfloat value)
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLint xpos, ypos, width, height;
 
-   if (SWRAST_CONTEXT(ctx)->NewState)
+   if (swrast->NewState)
       _swrast_validate_derived( ctx );
 
    if (!ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer) {
@@ -558,9 +558,9 @@ _swrast_Accum(GLcontext *ctx, GLenum op, GLfloat value)
       return;
    }
 
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
-   /* Compute region after calling RENDER_START so that we know the
+   /* Compute region after calling swrast_render_start() so that we know the
     * drawbuffer's size/bounds are up to date.
     */
    xpos = ctx->DrawBuffer->_Xmin;
@@ -595,5 +595,5 @@ _swrast_Accum(GLcontext *ctx, GLenum op, GLfloat value)
          break;
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 }
diff --git a/src/mesa/swrast/s_bitmap.c b/src/mesa/swrast/s_bitmap.c
index 35b34e654ff..5e7822cf323 100644
--- a/src/mesa/swrast/s_bitmap.c
+++ b/src/mesa/swrast/s_bitmap.c
@@ -50,7 +50,6 @@ _swrast_Bitmap( GLcontext *ctx, GLint px, GLint py,
 		const struct gl_pixelstore_attrib *unpack,
 		const GLubyte *bitmap )
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLint row, col;
    GLuint count = 0;
    SWspan span;
@@ -61,7 +60,7 @@ _swrast_Bitmap( GLcontext *ctx, GLint px, GLint py,
    if (!bitmap)
       return;
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
 
    if (SWRAST_CONTEXT(ctx)->NewState)
       _swrast_validate_derived( ctx );
@@ -132,7 +131,7 @@ _swrast_Bitmap( GLcontext *ctx, GLint px, GLint py,
       }
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 
    _mesa_unmap_bitmap_pbo(ctx, unpack);
 }
@@ -157,7 +156,7 @@ _swrast_Bitmap( GLcontext *ctx, GLint px, GLint py,
    ASSERT(ctx->RenderMode == GL_RENDER);
    ASSERT(bitmap);
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
 
    if (SWRAST_CONTEXT(ctx)->NewState)
       _swrast_validate_derived( ctx );
@@ -224,6 +223,6 @@ _swrast_Bitmap( GLcontext *ctx, GLint px, GLint py,
       }
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 }
 #endif
diff --git a/src/mesa/swrast/s_blit.c b/src/mesa/swrast/s_blit.c
index bc4b2ac6251..0e32cb8f653 100644
--- a/src/mesa/swrast/s_blit.c
+++ b/src/mesa/swrast/s_blit.c
@@ -737,7 +737,6 @@ _swrast_BlitFramebuffer(GLcontext *ctx,
                         GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
                         GLbitfield mask, GLenum filter)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    static const GLint buffers[3] = {
       GL_COLOR_BUFFER_BIT,
       GL_DEPTH_BUFFER_BIT,
@@ -753,7 +752,7 @@ _swrast_BlitFramebuffer(GLcontext *ctx,
       return;
    }
 
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
    if (srcX1 - srcX0 == dstX1 - dstX0 &&
        srcY1 - srcY0 == dstY1 - dstY0 &&
@@ -789,5 +788,5 @@ _swrast_BlitFramebuffer(GLcontext *ctx,
       }
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 }
diff --git a/src/mesa/swrast/s_buffers.c b/src/mesa/swrast/s_buffers.c
index 9e87d6d4857..af475ad8cb5 100644
--- a/src/mesa/swrast/s_buffers.c
+++ b/src/mesa/swrast/s_buffers.c
@@ -307,8 +307,6 @@ clear_color_buffers(GLcontext *ctx)
 void
 _swrast_Clear(GLcontext *ctx, GLbitfield buffers)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
 #ifdef DEBUG_FOO
    {
       const GLbitfield legalBits =
@@ -327,7 +325,7 @@ _swrast_Clear(GLcontext *ctx, GLbitfield buffers)
    }
 #endif
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
 
    /* do software clearing here */
    if (buffers) {
@@ -347,5 +345,5 @@ _swrast_Clear(GLcontext *ctx, GLbitfield buffers)
       }
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 }
diff --git a/src/mesa/swrast/s_context.h b/src/mesa/swrast/s_context.h
index a511d1c9a17..cdd6fa5048d 100644
--- a/src/mesa/swrast/s_context.h
+++ b/src/mesa/swrast/s_context.h
@@ -238,21 +238,43 @@ extern void
 _swrast_update_texture_samplers(GLcontext *ctx);
 
 
-#define SWRAST_CONTEXT(ctx) ((SWcontext *)ctx->swrast_context)
-
-#define RENDER_START(SWctx, GLctx)			\
-   do {							\
-      if ((SWctx)->Driver.SpanRenderStart) {		\
-         (*(SWctx)->Driver.SpanRenderStart)(GLctx);	\
-      }							\
-   } while (0)
-
-#define RENDER_FINISH(SWctx, GLctx)			\
-   do {							\
-      if ((SWctx)->Driver.SpanRenderFinish) {		\
-         (*(SWctx)->Driver.SpanRenderFinish)(GLctx);	\
-      }							\
-   } while (0)
+/** Return SWcontext for the given GLcontext */
+static INLINE SWcontext *
+SWRAST_CONTEXT(GLcontext *ctx)
+{
+   return (SWcontext *) ctx->swrast_context;
+}
+
+/** const version of above */
+static INLINE const SWcontext *
+CONST_SWRAST_CONTEXT(const GLcontext *ctx)
+{
+   return (const SWcontext *) ctx->swrast_context;
+}
+
+
+/**
+ * Called prior to framebuffer reading/writing.
+ * For drivers that rely on swrast for fallback rendering, this is the
+ * driver's opportunity to map renderbuffers and textures.
+ */
+static INLINE void
+swrast_render_start(GLcontext *ctx)
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   if (swrast->Driver.SpanRenderStart)
+      swrast->Driver.SpanRenderStart(ctx);
+}
+
+
+/** Called after framebuffer reading/writing */
+static INLINE void
+swrast_render_finish(GLcontext *ctx)
+{
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   if (swrast->Driver.SpanRenderFinish)
+      swrast->Driver.SpanRenderFinish(ctx);
+}
 
 
 
diff --git a/src/mesa/swrast/s_copypix.c b/src/mesa/swrast/s_copypix.c
index fc5990b261c..5ecfb1e90a9 100644
--- a/src/mesa/swrast/s_copypix.c
+++ b/src/mesa/swrast/s_copypix.c
@@ -899,7 +899,7 @@ _swrast_CopyPixels( GLcontext *ctx,
 		    GLint destx, GLint desty, GLenum type )
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
       
    if (swrast->NewState)
       _swrast_validate_derived( ctx );
@@ -928,5 +928,5 @@ _swrast_CopyPixels( GLcontext *ctx,
       }
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 }
diff --git a/src/mesa/swrast/s_drawpix.c b/src/mesa/swrast/s_drawpix.c
index 7af3e3dad1c..700f76d4bc4 100644
--- a/src/mesa/swrast/s_drawpix.c
+++ b/src/mesa/swrast/s_drawpix.c
@@ -830,7 +830,7 @@ _swrast_DrawPixels( GLcontext *ctx,
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
 
    if (ctx->NewState)
       _mesa_update_state(ctx);
@@ -840,7 +840,7 @@ _swrast_DrawPixels( GLcontext *ctx,
 
     pixels = _mesa_map_drawpix_pbo(ctx, unpack, pixels);
     if (!pixels) {
-       RENDER_FINISH(swrast,ctx);
+       swrast_render_finish(ctx);
        return;
     }
 
@@ -879,9 +879,9 @@ _swrast_DrawPixels( GLcontext *ctx,
       /* don't return yet, clean-up */
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 
-   _mesa_unmap_drapix_pbo(ctx, unpack);
+   _mesa_unmap_drawpix_pbo(ctx, unpack);
 }
 
 
@@ -904,7 +904,7 @@ _swrast_DrawDepthPixelsMESA( GLcontext *ctx,
    if (swrast->NewState)
       _swrast_validate_derived( ctx );
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
 
    switch (colorFormat) {
    case GL_COLOR_INDEX:
@@ -933,6 +933,6 @@ _swrast_DrawDepthPixelsMESA( GLcontext *ctx,
       _mesa_problem(ctx, "unexpected format in glDrawDepthPixelsMESA");
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 }
 #endif
diff --git a/src/mesa/swrast/s_fog.c b/src/mesa/swrast/s_fog.c
index b9ba265db6d..77ed0cfef96 100644
--- a/src/mesa/swrast/s_fog.c
+++ b/src/mesa/swrast/s_fog.c
@@ -164,7 +164,7 @@ else {										\
 void
 _swrast_fog_rgba_span( const GLcontext *ctx, SWspan *span )
 {
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const SWcontext *swrast = CONST_SWRAST_CONTEXT(ctx);
    GLfloat rFog, gFog, bFog;
 
    ASSERT(swrast->_FogEnabled);
@@ -283,7 +283,7 @@ _swrast_fog_rgba_span( const GLcontext *ctx, SWspan *span )
 void
 _swrast_fog_ci_span( const GLcontext *ctx, SWspan *span )
 {
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   const SWcontext *swrast = CONST_SWRAST_CONTEXT(ctx);
    const GLuint fogIndex = (GLuint) ctx->Fog.Index;
    GLuint *index = span->array->index;
 
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 525cf9d7245..c6601f5593d 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -33,6 +33,35 @@
 
 
 /**
+ * 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 GLchan texel[4], GLfloat colorOut[4], GLuint swizzle)
+{
+   if (swizzle == SWIZZLE_NOOP) {
+      colorOut[0] = CHAN_TO_FLOAT(texel[0]);
+      colorOut[1] = CHAN_TO_FLOAT(texel[1]);
+      colorOut[2] = CHAN_TO_FLOAT(texel[2]);
+      colorOut[3] = CHAN_TO_FLOAT(texel[3]);
+   }
+   else {
+      GLfloat vector[6];
+      vector[SWIZZLE_X] = CHAN_TO_FLOAT(texel[0]);
+      vector[SWIZZLE_Y] = CHAN_TO_FLOAT(texel[1]);
+      vector[SWIZZLE_Z] = CHAN_TO_FLOAT(texel[2]);
+      vector[SWIZZLE_W] = CHAN_TO_FLOAT(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()
  */
@@ -40,20 +69,23 @@ static void
 fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
                  GLuint unit, GLfloat color[4] )
 {
-   GLchan rgba[4];
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
 
-   if (texObj)
+   if (texObj) {
+      SWcontext *swrast = SWRAST_CONTEXT(ctx);
+      GLchan rgba[4];
+
       lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
 
-   /* XXX use a float-valued TextureSample routine here!!! */
-   swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord,
-                               &lambda, &rgba);
-   color[0] = CHAN_TO_FLOAT(rgba[0]);
-   color[1] = CHAN_TO_FLOAT(rgba[1]);
-   color[2] = CHAN_TO_FLOAT(rgba[2]);
-   color[3] = CHAN_TO_FLOAT(rgba[3]);
+      /* XXX use a float-valued TextureSample routine here!!! */
+      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);
+   }
 }
 
 
@@ -69,13 +101,14 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4],
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-   GLfloat lambda;
-   GLchan rgba[4];
 
    if (texObj) {
-      const struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel];
+      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;
+      GLchan rgba[4];
 
       lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
                                       texdx[1], texdy[1], /* dt/dx, dt/dy */
@@ -85,14 +118,16 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4],
                                       1.0F / texcoord[3]) + lodBias;
 
       lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
-   }
 
-   swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord,
-                               &lambda, &rgba);
-   color[0] = CHAN_TO_FLOAT(rgba[0]);
-   color[1] = CHAN_TO_FLOAT(rgba[1]);
-   color[2] = CHAN_TO_FLOAT(rgba[2]);
-   color[3] = CHAN_TO_FLOAT(rgba[3]);
+      /* XXX use a float-valued TextureSample routine here!!! */
+      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);
+   }
 }
 
 
diff --git a/src/mesa/swrast/s_imaging.c b/src/mesa/swrast/s_imaging.c
index 591857c3423..d6be3aa022e 100644
--- a/src/mesa/swrast/s_imaging.c
+++ b/src/mesa/swrast/s_imaging.c
@@ -39,7 +39,6 @@ _swrast_CopyColorTable( GLcontext *ctx,
 			GLenum target, GLenum internalformat,
 			GLint x, GLint y, GLsizei width)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLchan data[MAX_WIDTH][4];
    struct gl_buffer_object *bufferSave;
 
@@ -51,13 +50,13 @@ _swrast_CopyColorTable( GLcontext *ctx,
    if (width > MAX_WIDTH)
       width = MAX_WIDTH;
 
-   RENDER_START( swrast, ctx );
+   swrast_render_start(ctx);
 
    /* read the data from framebuffer */
    _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
                            width, x, y, CHAN_TYPE, data );
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 
    /* save PBO binding */
    bufferSave = ctx->Unpack.BufferObj;
@@ -74,7 +73,6 @@ void
 _swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start,
 			   GLint x, GLint y, GLsizei width)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLchan data[MAX_WIDTH][4];
    struct gl_buffer_object *bufferSave;
 
@@ -86,13 +84,13 @@ _swrast_CopyColorSubTable( GLcontext *ctx,GLenum target, GLsizei start,
    if (width > MAX_WIDTH)
       width = MAX_WIDTH;
 
-   RENDER_START( swrast, ctx );
+   swrast_render_start(ctx);
 
    /* read the data from framebuffer */
    _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
                            width, x, y, CHAN_TYPE, data );
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 
    /* save PBO binding */
    bufferSave = ctx->Unpack.BufferObj;
@@ -110,7 +108,6 @@ _swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target,
 				GLenum internalFormat, 
 				GLint x, GLint y, GLsizei width)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLchan rgba[MAX_CONVOLUTION_WIDTH][4];
    struct gl_buffer_object *bufferSave;
 
@@ -119,13 +116,13 @@ _swrast_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target,
       return;
    }
 
-   RENDER_START( swrast, ctx );
+   swrast_render_start(ctx);
 
    /* read the data from framebuffer */
    _swrast_read_rgba_span( ctx, ctx->ReadBuffer->_ColorReadBuffer,
                            width, x, y, CHAN_TYPE, rgba );
    
-   RENDER_FINISH( swrast, ctx );
+   swrast_render_finish(ctx);
 
    /* save PBO binding */
    bufferSave = ctx->Unpack.BufferObj;
@@ -145,7 +142,6 @@ _swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target,
 				GLenum internalFormat, 
 				GLint x, GLint y, GLsizei width, GLsizei height)
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_pixelstore_attrib packSave;
    GLchan rgba[MAX_CONVOLUTION_HEIGHT][MAX_CONVOLUTION_WIDTH][4];
    GLint i;
@@ -156,7 +152,7 @@ _swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target,
       return;
    }
 
-   RENDER_START(swrast,ctx);
+   swrast_render_start(ctx);
    
    /* read pixels from framebuffer */
    for (i = 0; i < height; i++) {
@@ -164,7 +160,7 @@ _swrast_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target,
                               width, x, y + i, CHAN_TYPE, rgba[i] );
    }
 
-   RENDER_FINISH(swrast,ctx);
+   swrast_render_finish(ctx);
 
    /*
     * HACK: save & restore context state so we can store this as a
diff --git a/src/mesa/swrast/s_readpix.c b/src/mesa/swrast/s_readpix.c
index f2630451707..e901fc6b5dc 100644
--- a/src/mesa/swrast/s_readpix.c
+++ b/src/mesa/swrast/s_readpix.c
@@ -555,11 +555,11 @@ _swrast_ReadPixels( GLcontext *ctx,
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_pixelstore_attrib clippedPacking = *packing;
 
-   /* Need to do RENDER_START before clipping or anything else since this
-    * is where a driver may grab the hw lock and get an updated window
-    * size.
+   /* Need to do swrast_render_start() before clipping or anything else
+    * since this is where a driver may grab the hw lock and get an updated
+    * window size.
     */
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
    if (ctx->NewState)
       _mesa_update_state(ctx);
@@ -570,7 +570,7 @@ _swrast_ReadPixels( GLcontext *ctx,
    /* Do all needed clipping here, so that we can forget about it later */
    if (!_mesa_clip_readpixels(ctx, &x, &y, &width, &height, &clippedPacking)) {
       /* The ReadPixels region is totally outside the window bounds */
-      RENDER_FINISH(swrast, ctx);
+      swrast_render_finish(ctx);
       return;
    }
 
@@ -614,7 +614,7 @@ _swrast_ReadPixels( GLcontext *ctx,
          /* don't return yet, clean-up */
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 
    _mesa_unmap_readpix_pbo(ctx, &clippedPacking);
 }
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 7f9cc64bb29..ab7b82b19d9 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -1311,6 +1311,7 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
    /* Do the alpha test */
    if (ctx->Color.AlphaEnabled) {
       if (!_swrast_alpha_test(ctx, span)) {
+         /* all fragments failed test */
          goto end;
       }
    }
@@ -1323,6 +1324,7 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
       if (ctx->Stencil.Enabled && fb->Visual.stencilBits > 0) {
          /* Combined Z/stencil tests */
          if (!_swrast_stencil_and_ztest_span(ctx, span)) {
+            /* all fragments failed test */
             goto end;
          }
       }
@@ -1331,6 +1333,7 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
          ASSERT(ctx->Depth.Test);
          ASSERT(span->arrayMask & SPAN_Z);
          if (!_swrast_depth_test_span(ctx, span)) {
+            /* all fragments failed test */
             goto end;
          }
       }
@@ -1350,6 +1353,7 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
     * the occlusion test.
     */
    if (colorMask == 0x0) {
+      /* no colors to write */
       goto end;
    }
 
diff --git a/src/mesa/swrast/s_stencil.c b/src/mesa/swrast/s_stencil.c
index c925922463e..2e84ddec71f 100644
--- a/src/mesa/swrast/s_stencil.c
+++ b/src/mesa/swrast/s_stencil.c
@@ -997,10 +997,12 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
 GLboolean
 _swrast_stencil_and_ztest_span(GLcontext *ctx, SWspan *span)
 {
+   const GLuint face = (span->facing == 0) ? 0 : ctx->Stencil._BackFace;
+
    if (span->arrayMask & SPAN_XY)
-      return stencil_and_ztest_pixels(ctx, span, span->facing);
+      return stencil_and_ztest_pixels(ctx, span, face);
    else
-      return stencil_and_ztest_span(ctx, span, span->facing);
+      return stencil_and_ztest_span(ctx, span, face);
 }
 
 
diff --git a/src/mesa/swrast/s_texcombine.c b/src/mesa/swrast/s_texcombine.c
index 632d650007e..38b5633887f 100644
--- a/src/mesa/swrast/s_texcombine.c
+++ b/src/mesa/swrast/s_texcombine.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.5
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * 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"),
@@ -30,6 +31,7 @@
 #include "main/imports.h"
 #include "main/macros.h"
 #include "main/pixel.h"
+#include "shader/prog_instruction.h"
 
 #include "s_context.h"
 #include "s_texcombine.h"
@@ -66,9 +68,9 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
                  GLchan (*rgba)[4] )
 {
    const struct gl_texture_unit *textureUnit = &(ctx->Texture.Unit[unit]);
-   const GLchan (*argRGB [3])[4];
-   const GLchan (*argA [3])[4];
-   const GLuint RGBshift = textureUnit->_CurrentCombine->ScaleShiftRGB;
+   const GLchan (*argRGB [4])[4];
+   const GLchan (*argA [4])[4];
+   const GLint RGBshift = textureUnit->_CurrentCombine->ScaleShiftRGB;
    const GLuint Ashift   = textureUnit->_CurrentCombine->ScaleShiftA;
 #if CHAN_TYPE == GL_FLOAT
    const GLchan RGBmult = (GLfloat) (1 << RGBshift);
@@ -80,12 +82,12 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
    static const GLchan zero[4] = { 0, 0, 0, 0 };
    const GLuint numColorArgs = textureUnit->_CurrentCombine->_NumArgsRGB;
    const GLuint numAlphaArgs = textureUnit->_CurrentCombine->_NumArgsA;
-   GLchan ccolor[3][MAX_WIDTH][4];
+   GLchan ccolor[4][MAX_WIDTH][4];
    GLuint i, j;
 
    ASSERT(ctx->Extensions.EXT_texture_env_combine ||
           ctx->Extensions.ARB_texture_env_combine);
-   ASSERT(SWRAST_CONTEXT(ctx)->_AnyTextureCombine);
+   ASSERT(CONST_SWRAST_CONTEXT(ctx)->_AnyTextureCombine);
 
    /*
    printf("modeRGB 0x%x  modeA 0x%x  srcRGB1 0x%x  srcA1 0x%x  srcRGB2 0x%x  srcA2 0x%x\n",
@@ -98,7 +100,7 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
    */
 
    /*
-    * Do operand setup for up to 3 operands.  Loop over the terms.
+    * Do operand setup for up to 4 operands.  Loop over the terms.
     */
    for (j = 0; j < numColorArgs; j++) {
       const GLenum srcRGB = textureUnit->_CurrentCombine->SourceRGB[j];
@@ -295,7 +297,36 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
          }
          break;
       case GL_ADD:
-         {
+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
+            /* (a * b) + (c * d) */
+            const GLchan (*arg0)[4] = (const GLchan (*)[4]) argRGB[0];
+            const GLchan (*arg1)[4] = (const GLchan (*)[4]) argRGB[1];
+            const GLchan (*arg2)[4] = (const GLchan (*)[4]) argRGB[2];
+            const GLchan (*arg3)[4] = (const GLchan (*)[4]) argRGB[3];
+            for (i = 0; i < n; i++) {
+#if CHAN_TYPE == GL_FLOAT
+               rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +
+                                 arg2[i][RCOMP] * arg3[i][RCOMP]) * RGBmult;
+               rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +
+                                 arg2[i][GCOMP] * arg3[i][GCOMP]) * RGBmult;
+               rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +
+                                 arg2[i][BCOMP] * arg3[i][BCOMP]) * RGBmult;
+#else
+               const GLint shift = CHAN_BITS - RGBshift;
+               GLint r = (PROD(arg0[i][RCOMP], arg1[i][RCOMP]) >> shift) +
+                         (PROD(arg2[i][RCOMP], arg3[i][RCOMP]) >> shift);
+               GLint g = (PROD(arg0[i][GCOMP], arg1[i][GCOMP]) >> shift) +
+                         (PROD(arg2[i][GCOMP], arg3[i][GCOMP]) >> shift);
+               GLint b = (PROD(arg0[i][BCOMP], arg1[i][BCOMP]) >> shift) +
+                         (PROD(arg2[i][BCOMP], arg3[i][BCOMP]) >> shift);
+               rgba[i][RCOMP] = (GLchan) MIN2(r, CHAN_MAX);
+               rgba[i][GCOMP] = (GLchan) MIN2(g, CHAN_MAX);
+               rgba[i][BCOMP] = (GLchan) MIN2(b, CHAN_MAX);
+#endif
+            }
+         }
+         else {
+            /* 2-term addition */
             const GLchan (*arg0)[4] = (const GLchan (*)[4]) argRGB[0];
             const GLchan (*arg1)[4] = (const GLchan (*)[4]) argRGB[1];
             for (i = 0; i < n; i++) {
@@ -315,7 +346,37 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
          }
          break;
       case GL_ADD_SIGNED:
-         {
+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
+            /* (a * b) + (c * d) - 0.5 */
+            const GLchan (*arg0)[4] = (const GLchan (*)[4]) argRGB[0];
+            const GLchan (*arg1)[4] = (const GLchan (*)[4]) argRGB[1];
+            const GLchan (*arg2)[4] = (const GLchan (*)[4]) argRGB[2];
+            const GLchan (*arg3)[4] = (const GLchan (*)[4]) argRGB[3];
+            for (i = 0; i < n; i++) {
+#if CHAN_TYPE == GL_FLOAT
+               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP] *
+                                 arg2[i][RCOMP] + arg3[i][RCOMP] - 0.5) * RGBmult;
+               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] *
+                                 arg2[i][GCOMP] + arg3[i][GCOMP] - 0.5) * RGBmult;
+               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] *
+                                 arg2[i][BCOMP] + arg3[i][BCOMP] - 0.5) * RGBmult;
+#else
+               GLint r = (((PROD(arg0[i][RCOMP], arg1[i][RCOMP]) +
+                            PROD(arg2[i][RCOMP], arg3[i][RCOMP])) >> CHAN_BITS) - half)
+                          << RGBshift;
+               GLint g = (((PROD(arg0[i][GCOMP], arg1[i][GCOMP]) +
+                            PROD(arg2[i][GCOMP], arg3[i][GCOMP])) >> CHAN_BITS) - half)
+                          << RGBshift;
+               GLint b = (((PROD(arg0[i][BCOMP], arg1[i][BCOMP]) +
+                            PROD(arg2[i][BCOMP], arg3[i][BCOMP])) >> CHAN_BITS) - half)
+                          << RGBshift;
+               rgba[i][RCOMP] = (GLchan) CLAMP(r, 0, CHAN_MAX);
+               rgba[i][GCOMP] = (GLchan) CLAMP(g, 0, CHAN_MAX);
+               rgba[i][BCOMP] = (GLchan) CLAMP(b, 0, CHAN_MAX);
+#endif
+            }
+         }
+         else {
             const GLchan (*arg0)[4] = (const GLchan (*)[4]) argRGB[0];
             const GLchan (*arg1)[4] = (const GLchan (*)[4]) argRGB[1];
             for (i = 0; i < n; i++) {
@@ -324,9 +385,9 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
                rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] - 0.5) * RGBmult;
                rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] - 0.5) * RGBmult;
 #else
-               GLint r = (GLint) arg0[i][RCOMP] + (GLint) arg1[i][RCOMP] -half;
-               GLint g = (GLint) arg0[i][GCOMP] + (GLint) arg1[i][GCOMP] -half;
-               GLint b = (GLint) arg0[i][BCOMP] + (GLint) arg1[i][BCOMP] -half;
+               GLint r = (GLint) arg0[i][RCOMP] + (GLint) arg1[i][RCOMP] - half;
+               GLint g = (GLint) arg0[i][GCOMP] + (GLint) arg1[i][GCOMP] - half;
+               GLint b = (GLint) arg0[i][BCOMP] + (GLint) arg1[i][BCOMP] - half;
                r = (r < 0) ? 0 : r << RGBshift;
                g = (g < 0) ? 0 : g << RGBshift;
                b = (b < 0) ? 0 : b << RGBshift;
@@ -573,9 +634,28 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
          }
          break;
       case GL_ADD:
-         {
+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
+            /* (a * b) + (c * d) */
             const GLchan (*arg0)[4] = (const GLchan (*)[4]) argA[0];
-            const GLchan  (*arg1)[4] = (const GLchan (*)[4]) argA[1];
+            const GLchan (*arg1)[4] = (const GLchan (*)[4]) argA[1];
+            const GLchan (*arg2)[4] = (const GLchan (*)[4]) argA[2];
+            const GLchan (*arg3)[4] = (const GLchan (*)[4]) argA[3];
+            for (i = 0; i < n; i++) {
+#if CHAN_TYPE == GL_FLOAT
+               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +
+                                 arg2[i][ACOMP] * arg3[i][ACOMP]) * Amult;
+#else
+               const GLint shift = CHAN_BITS - Ashift;
+               GLint a = (PROD(arg0[i][ACOMP], arg1[i][ACOMP]) >> shift) +
+                         (PROD(arg2[i][ACOMP], arg3[i][ACOMP]) >> shift);
+               rgba[i][ACOMP] = (GLchan) MIN2(a, CHAN_MAX);
+#endif
+            }
+         }
+         else {
+            /* two-term add */
+            const GLchan (*arg0)[4] = (const GLchan (*)[4]) argA[0];
+            const GLchan (*arg1)[4] = (const GLchan (*)[4]) argA[1];
             for (i = 0; i < n; i++) {
 #if CHAN_TYPE == GL_FLOAT
                rgba[i][ACOMP] = (arg0[i][ACOMP] + arg1[i][ACOMP]) * Amult;
@@ -587,7 +667,27 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
          }
          break;
       case GL_ADD_SIGNED:
-         {
+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
+            /* (a * b) + (c * d) - 0.5 */
+            const GLchan (*arg0)[4] = (const GLchan (*)[4]) argA[0];
+            const GLchan (*arg1)[4] = (const GLchan (*)[4]) argA[1];
+            const GLchan (*arg2)[4] = (const GLchan (*)[4]) argA[2];
+            const GLchan (*arg3)[4] = (const GLchan (*)[4]) argA[3];
+            for (i = 0; i < n; i++) {
+#if CHAN_TYPE == GL_FLOAT
+               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +
+                                 arg2[i][ACOMP] * arg3[i][ACOMP] -
+                                 0.5) * Amult;
+#else
+               GLint a = (((PROD(arg0[i][ACOMP], arg1[i][ACOMP]) +
+                            PROD(arg2[i][ACOMP], arg3[i][ACOMP])) >> CHAN_BITS) - half)
+                          << Ashift;
+               rgba[i][ACOMP] = (GLchan) CLAMP(a, 0, CHAN_MAX);
+#endif
+            }
+         }
+         else {
+            /* a + b - 0.5 */
             const GLchan (*arg0)[4] = (const GLchan (*)[4]) argA[0];
             const GLchan (*arg1)[4] = (const GLchan (*)[4]) argA[1];
             for (i = 0; i < n; i++) {
@@ -596,7 +696,7 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
 #else
                GLint a = (GLint) arg0[i][ACOMP] + (GLint) arg1[i][ACOMP] -half;
                a = (a < 0) ? 0 : a << Ashift;
-               rgba[i][ACOMP] = (GLchan) MIN2(a, CHAN_MAX);
+               rgba[i][ACOMP] = (GLchan) CLAMP(a, 0, CHAN_MAX);
 #endif
             }
          }
@@ -717,6 +817,36 @@ texture_combine( const GLcontext *ctx, GLuint unit, GLuint n,
 
 
 /**
+ * Apply X/Y/Z/W/0/1 swizzle to an array of colors/texels.
+ * See GL_EXT_texture_swizzle.
+ */
+static void
+swizzle_texels(GLuint swizzle, GLuint count, GLchan (*texels)[4])
+{
+   const GLuint swzR = GET_SWZ(swizzle, 0);
+   const GLuint swzG = GET_SWZ(swizzle, 1);
+   const GLuint swzB = GET_SWZ(swizzle, 2);
+   const GLuint swzA = GET_SWZ(swizzle, 3);
+   GLchan vector[6];
+   GLuint i;
+
+   vector[SWIZZLE_ZERO] = 0;
+   vector[SWIZZLE_ONE] = CHAN_MAX;
+
+   for (i = 0; i < count; i++) {
+      vector[SWIZZLE_X] = texels[i][0];
+      vector[SWIZZLE_Y] = texels[i][1];
+      vector[SWIZZLE_Z] = texels[i][2];
+      vector[SWIZZLE_W] = texels[i][3];
+      texels[i][RCOMP] = vector[swzR];
+      texels[i][GCOMP] = vector[swzG];
+      texels[i][BCOMP] = vector[swzB];
+      texels[i][ACOMP] = vector[swzA];
+   }
+}
+
+
+/**
  * Apply a conventional OpenGL texture env mode (REPLACE, ADD, BLEND,
  * MODULATE, or DECAL) to an array of fragments.
  * Input:  textureUnit - pointer to texture unit to apply
@@ -1142,9 +1272,15 @@ _swrast_texture_span( GLcontext *ctx, SWspan *span )
             _mesa_lookup_rgba_float(&texUnit->ColorTable, span->end, texels);
 #endif
          }
+
+         /* GL_EXT_texture_swizzle */
+         if (curObj->_Swizzle != SWIZZLE_NOOP) {
+            swizzle_texels(curObj->_Swizzle, span->end, texels);
+         }
       }
    }
 
+
    /*
     * OK, now apply the texture (aka texture combine/blend).
     * We modify the span->color.rgba values.
@@ -1163,6 +1299,8 @@ _swrast_texture_span( GLcontext *ctx, SWspan *span )
             const GLchan (*texels)[4] = (const GLchan (*)[4])
                (swrast->TexelBuffer + unit *
                 (span->end * 4 * sizeof(GLchan)));
+
+
             texture_apply( ctx, texUnit, span->end,
                            (CONST GLchan (*)[4]) primary_rgba, texels,
                            span->array->rgba );
diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c
index a095b255ab5..8d72018cf4f 100644
--- a/src/mesa/swrast/s_texfilter.c
+++ b/src/mesa/swrast/s_texfilter.c
@@ -2826,7 +2826,7 @@ sample_depth_texture( GLcontext *ctx,
           tObj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
           tObj->Target == GL_TEXTURE_2D_ARRAY_EXT);
 
-   UNCLAMPED_FLOAT_TO_CHAN(ambient, tObj->ShadowAmbient);
+   UNCLAMPED_FLOAT_TO_CHAN(ambient, tObj->CompareFailValue);
 
    /* XXXX if tObj->MinFilter != tObj->MagFilter, we're ignoring lambda */
 
@@ -3156,103 +3156,6 @@ sample_depth_texture( GLcontext *ctx,
 }
 
 
-#if 0
-/*
- * Experimental depth texture sampling function.
- */
-static void
-sample_depth_texture2(const GLcontext *ctx,
-                     const struct gl_texture_unit *texUnit,
-                     GLuint n, const GLfloat texcoords[][4],
-                     GLchan texel[][4])
-{
-   const struct gl_texture_object *texObj = texUnit->_Current;
-   const GLint baseLevel = texObj->BaseLevel;
-   const struct gl_texture_image *texImage = texObj->Image[0][baseLevel];
-   const GLuint width = texImage->Width;
-   const GLuint height = texImage->Height;
-   GLchan ambient;
-   GLboolean lequal, gequal;
-
-   if (texObj->Target != GL_TEXTURE_2D) {
-      _mesa_problem(ctx, "only 2-D depth textures supported at this time");
-      return;
-   }
-
-   if (texObj->MinFilter != texObj->MagFilter) {
-      _mesa_problem(ctx, "mipmapped depth textures not supported at this time");
-      return;
-   }
-
-   /* XXX the GL_SGIX_shadow extension spec doesn't say what to do if
-    * GL_TEXTURE_COMPARE_SGIX == GL_TRUE but the current texture object
-    * isn't a depth texture.
-    */
-   if (texImage->TexFormat->BaseFormat != GL_DEPTH_COMPONENT) {
-      _mesa_problem(ctx,"GL_TEXTURE_COMPARE_SGIX enabled with non-depth texture");
-      return;
-   }
-
-   UNCLAMPED_FLOAT_TO_CHAN(ambient, tObj->ShadowAmbient);
-
-   if (texObj->CompareOperator == GL_TEXTURE_LEQUAL_R_SGIX) {
-      lequal = GL_TRUE;
-      gequal = GL_FALSE;
-   }
-   else {
-      lequal = GL_FALSE;
-      gequal = GL_TRUE;
-   }
-
-   {
-      GLuint i;
-      for (i = 0; i < n; i++) {
-         const GLint K = 3;
-         GLint col, row, ii, jj, imin, imax, jmin, jmax, samples, count;
-         GLfloat w;
-         GLchan lum;
-         col = nearest_texel_location(texObj->WrapS, img, width,
-                                      texcoords[i][0]);
-         row = nearest_texel_location(texObj->WrapT, img, height,
-                                      texcoords[i][1]);
-
-         imin = col - K;
-         imax = col + K;
-         jmin = row - K;
-         jmax = row + K;
-
-         if (imin < 0)  imin = 0;
-         if (imax >= width)  imax = width - 1;
-         if (jmin < 0)  jmin = 0;
-         if (jmax >= height) jmax = height - 1;
-
-         samples = (imax - imin + 1) * (jmax - jmin + 1);
-         count = 0;
-         for (jj = jmin; jj <= jmax; jj++) {
-            for (ii = imin; ii <= imax; ii++) {
-               GLfloat depthSample;
-               texImage->FetchTexelf(texImage, ii, jj, 0, &depthSample);
-               if ((depthSample <= r[i] && lequal) ||
-                   (depthSample >= r[i] && gequal)) {
-                  count++;
-               }
-            }
-         }
-
-         w = (GLfloat) count / (GLfloat) samples;
-         w = CHAN_MAXF - w * (CHAN_MAXF - (GLfloat) ambient);
-         lum = (GLint) w;
-
-         texel[i][RCOMP] = lum;
-         texel[i][GCOMP] = lum;
-         texel[i][BCOMP] = lum;
-         texel[i][ACOMP] = CHAN_MAX;
-      }
-   }
-}
-#endif
-
-
 /**
  * We use this function when a texture object is in an "incomplete" state.
  * When a fragment program attempts to sample an incomplete texture we
diff --git a/src/mesa/swrast/s_texstore.c b/src/mesa/swrast/s_texstore.c
index 16b00b9fa1c..f9ff9ad6a42 100644
--- a/src/mesa/swrast/s_texstore.c
+++ b/src/mesa/swrast/s_texstore.c
@@ -67,7 +67,6 @@ static GLvoid *
 read_color_image( GLcontext *ctx, GLint x, GLint y, GLenum type,
                   GLsizei width, GLsizei height )
 {
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer;
    const GLint pixelSize = _mesa_bytes_per_pixel(GL_RGBA, type);
    const GLint stride = width * pixelSize;
@@ -78,7 +77,7 @@ read_color_image( GLcontext *ctx, GLint x, GLint y, GLenum type,
    if (!image)
       return NULL;
 
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
    dst = image;
    for (row = 0; row < height; row++) {
@@ -86,7 +85,7 @@ read_color_image( GLcontext *ctx, GLint x, GLint y, GLenum type,
       dst += stride;
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 
    return image;
 }
@@ -101,7 +100,6 @@ read_depth_image( GLcontext *ctx, GLint x, GLint y,
                   GLsizei width, GLsizei height )
 {
    struct gl_renderbuffer *rb = ctx->ReadBuffer->_DepthBuffer;
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLuint *image, *dst;
    GLint i;
 
@@ -109,7 +107,7 @@ read_depth_image( GLcontext *ctx, GLint x, GLint y,
    if (!image)
       return NULL;
 
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
    dst = image;
    for (i = 0; i < height; i++) {
@@ -117,7 +115,7 @@ read_depth_image( GLcontext *ctx, GLint x, GLint y,
       dst += width;
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 
    return image;
 }
@@ -132,7 +130,6 @@ read_depth_stencil_image(GLcontext *ctx, GLint x, GLint y,
 {
    struct gl_renderbuffer *depthRb = ctx->ReadBuffer->_DepthBuffer;
    struct gl_renderbuffer *stencilRb = ctx->ReadBuffer->_StencilBuffer;
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    GLuint *image, *dst;
    GLint i;
 
@@ -143,7 +140,7 @@ read_depth_stencil_image(GLcontext *ctx, GLint x, GLint y,
    if (!image)
       return NULL;
 
-   RENDER_START(swrast, ctx);
+   swrast_render_start(ctx);
 
    /* read from depth buffer */
    dst = image;
@@ -205,7 +202,7 @@ read_depth_stencil_image(GLcontext *ctx, GLint x, GLint y,
       dst += width;
    }
 
-   RENDER_FINISH(swrast, ctx);
+   swrast_render_finish(ctx);
 
    return image;
 }
diff --git a/src/mesa/swrast/s_triangle.c b/src/mesa/swrast/s_triangle.c
index a2e8433e271..0598052f50c 100644
--- a/src/mesa/swrast/s_triangle.c
+++ b/src/mesa/swrast/s_triangle.c
@@ -35,6 +35,7 @@
 #include "main/imports.h"
 #include "main/macros.h"
 #include "main/texformat.h"
+#include "shader/prog_instruction.h"
 
 #include "s_aatriangle.h"
 #include "s_context.h"
@@ -265,9 +266,6 @@ affine_span(GLcontext *ctx, SWspan *span,
    GLchan sample[4];  /* the filtered texture sample */
    const GLuint texEnableSave = ctx->Texture._EnabledUnits;
 
-   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */
-   ctx->Texture._EnabledUnits = 0x0;
-
    /* 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
@@ -396,6 +394,9 @@ affine_span(GLcontext *ctx, SWspan *span,
    GLuint i;
    GLchan *dest = span->array->rgba[0];
 
+   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */
+   ctx->Texture._EnabledUnits = 0x0;
+
    span->intTex[0] -= FIXED_HALF;
    span->intTex[1] -= FIXED_HALF;
    switch (info->filter) {
@@ -1063,6 +1064,7 @@ _swrast_choose_triangle( GLcontext *ctx )
              && 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
@@ -1070,7 +1072,8 @@ _swrast_choose_triangle( GLcontext *ctx )
              && 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_COMBINE_EXT
+             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE4_NV) {
 	    if (ctx->Hint.PerspectiveCorrection==GL_FASTEST) {
 	       if (minFilter == GL_NEAREST
 		   && format == MESA_FORMAT_RGB