Merge remote-tracking branch 'origin/released'

47 files changed, 5442 insertions, 3663 deletions

diff --git a/mesalib/src/gallium/auxiliary/util/u_math.h b/mesalib/src/gallium/auxiliary/util/u_math.h
index f908341f6..404bd21e1 100644
--- a/mesalib/src/gallium/auxiliary/util/u_math.h
+++ b/mesalib/src/gallium/auxiliary/util/u_math.h
@@ -51,6 +51,10 @@ extern "C" {
 #include <math.h>
 #include <stdarg.h>
 
+#ifdef PIPE_OS_UNIX
+#include <strings.h> /* for ffs */
+#endif
+
 
 #ifndef M_SQRT2
 #define M_SQRT2 1.41421356237309504880
diff --git a/mesalib/src/mesa/SConscript b/mesalib/src/mesa/SConscript
index bcb6a2491..da4d9b185 100644
--- a/mesalib/src/mesa/SConscript
+++ b/mesalib/src/mesa/SConscript
@@ -74,6 +74,7 @@ main_sources = [
     'main/ffvertex_prog.c',
     'main/fog.c',
     'main/formats.c',
+    'main/format_pack.c',
     'main/format_unpack.c',
     'main/framebuffer.c',
     'main/get.c',
@@ -152,7 +153,6 @@ math_sources = [
 swrast_sources = [
     'swrast/s_aaline.c',
     'swrast/s_aatriangle.c',
-    'swrast/s_accum.c',
     'swrast/s_alpha.c',
     'swrast/s_atifragshader.c',
     'swrast/s_bitmap.c',
@@ -170,6 +170,7 @@ swrast_sources = [
     'swrast/s_logic.c',
     'swrast/s_masking.c',
     'swrast/s_points.c',
+    'swrast/s_renderbuffer.c',
     'swrast/s_span.c',
     'swrast/s_stencil.c',
     'swrast/s_texcombine.c',
@@ -240,7 +241,6 @@ statetracker_sources = [
     'state_tracker/st_atom_stipple.c',
     'state_tracker/st_atom_texture.c',
     'state_tracker/st_atom_viewport.c',
-    'state_tracker/st_cb_accum.c',
     'state_tracker/st_cb_bitmap.c',
     'state_tracker/st_cb_blit.c',
     'state_tracker/st_cb_bufferobjects.c',
diff --git a/mesalib/src/mesa/drivers/common/driverfuncs.c b/mesalib/src/mesa/drivers/common/driverfuncs.c
index 5e25d7fda..c6b42a21d 100644
--- a/mesalib/src/mesa/drivers/common/driverfuncs.c
+++ b/mesalib/src/mesa/drivers/common/driverfuncs.c
@@ -25,6 +25,7 @@
 
 #include "main/glheader.h"
 #include "main/imports.h"
+#include "main/accum.h"
 #include "main/arrayobj.h"
 #include "main/context.h"
 #include "main/framebuffer.h"
@@ -49,6 +50,7 @@
 #include "program/program.h"
 #include "tnl/tnl.h"
 #include "swrast/swrast.h"
+#include "swrast/s_renderbuffer.h"
 
 #include "driverfuncs.h"
 #include "meta.h"
@@ -80,7 +82,7 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
 
    /* framebuffer/image functions */
    driver->Clear = _swrast_Clear;
-   driver->Accum = _swrast_Accum;
+   driver->Accum = _mesa_accum;
    driver->RasterPos = _tnl_RasterPos;
    driver->DrawPixels = _swrast_DrawPixels;
    driver->ReadPixels = _mesa_readpixels;
@@ -119,8 +121,6 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->UnmapTextureImage = _swrast_unmap_teximage;
    driver->MapTexture = NULL;
    driver->UnmapTexture = NULL;
-   driver->TextureMemCpy = memcpy;
-   driver->IsTextureResident = NULL;
    driver->DrawTex = _mesa_meta_DrawTex;
 
    /* Vertex/fragment programs */
@@ -180,9 +180,9 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    _mesa_init_sync_object_functions(driver);
 
    driver->NewFramebuffer = _mesa_new_framebuffer;
-   driver->NewRenderbuffer = _mesa_new_soft_renderbuffer;
-   driver->MapRenderbuffer = _mesa_map_soft_renderbuffer;
-   driver->UnmapRenderbuffer = _mesa_unmap_soft_renderbuffer;
+   driver->NewRenderbuffer = _swrast_new_soft_renderbuffer;
+   driver->MapRenderbuffer = _swrast_map_soft_renderbuffer;
+   driver->UnmapRenderbuffer = _swrast_unmap_soft_renderbuffer;
    driver->RenderTexture = _swrast_render_texture;
    driver->FinishRenderTexture = _swrast_finish_render_texture;
    driver->FramebufferRenderbuffer = _mesa_framebuffer_renderbuffer;
diff --git a/mesalib/src/mesa/drivers/dri/swrast/swrast.c b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
index f4c656d01..b31700d10 100644
--- a/mesalib/src/mesa/drivers/dri/swrast/swrast.c
+++ b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
@@ -43,6 +43,7 @@
 #include "main/imports.h"
 #include "main/renderbuffer.h"
 #include "swrast/swrast.h"
+#include "swrast/s_renderbuffer.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
@@ -495,13 +496,13 @@ dri_create_buffer(__DRIscreen * sPriv,
     }
 
     /* add software renderbuffers */
-    _mesa_add_soft_renderbuffers(fb,
-				 GL_FALSE, /* color */
-				 visual->haveDepthBuffer,
-				 visual->haveStencilBuffer,
-				 visual->haveAccumBuffer,
-				 GL_FALSE, /* alpha */
-				 GL_FALSE /* aux bufs */);
+    _swrast_add_soft_renderbuffers(fb,
+                                   GL_FALSE, /* color */
+                                   visual->haveDepthBuffer,
+                                   visual->haveStencilBuffer,
+                                   visual->haveAccumBuffer,
+                                   GL_FALSE, /* alpha */
+                                   GL_FALSE /* aux bufs */);
 
     return GL_TRUE;
 
diff --git a/mesalib/src/mesa/drivers/windows/gdi/wmesa.c b/mesalib/src/mesa/drivers/windows/gdi/wmesa.c
index 14d15ed6c..675ffdaaa 100644
--- a/mesalib/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/mesalib/src/mesa/drivers/windows/gdi/wmesa.c
@@ -16,6 +16,7 @@
 #include "drivers/common/meta.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
+#include "swrast/s_renderbuffer.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
@@ -1610,13 +1611,13 @@ void WMesaMakeCurrent(WMesaContext c, HDC hdc)
         wmesa_set_renderbuffer_funcs(rb, pwfb->pixelformat, pwfb->cColorBits, 0);
 
 	/* Let Mesa own the Depth, Stencil, and Accum buffers */
-        _mesa_add_soft_renderbuffers(&pwfb->Base,
-                                     GL_FALSE, /* color */
-                                     visual->depthBits > 0,
-                                     visual->stencilBits > 0,
-                                     visual->accumRedBits > 0,
-                                     visual->alphaBits >0, 
-                                     GL_FALSE);
+        _swrast_add_soft_renderbuffers(&pwfb->Base,
+                                       GL_FALSE, /* color */
+                                       visual->depthBits > 0,
+                                       visual->stencilBits > 0,
+                                       visual->accumRedBits > 0,
+                                       visual->alphaBits >0, 
+                                       GL_FALSE);
     }
 
     if (c && pwfb)
diff --git a/mesalib/src/mesa/main/accum.c b/mesalib/src/mesa/main/accum.c
index d7ed3a849..eb06bbb6e 100644
--- a/mesalib/src/mesa/main/accum.c
+++ b/mesalib/src/mesa/main/accum.c
@@ -24,7 +24,10 @@
 
 #include "glheader.h"
 #include "accum.h"
+#include "condrender.h"
 #include "context.h"
+#include "format_unpack.h"
+#include "format_pack.h"
 #include "imports.h"
 #include "macros.h"
 #include "mfeatures.h"
@@ -101,7 +104,7 @@ _mesa_Accum( GLenum op, GLfloat value )
       return;
 
    if (ctx->RenderMode == GL_RENDER) {
-      ctx->Driver.Accum(ctx, op, value);
+      _mesa_accum(ctx, op, value);
    }
 }
 
@@ -123,3 +126,384 @@ _mesa_init_accum( struct gl_context *ctx )
    /* Accumulate buffer group */
    ASSIGN_4V( ctx->Accum.ClearColor, 0.0, 0.0, 0.0, 0.0 );
 }
+
+
+
+
+/**
+ * Clear the accumulation buffer by mapping the renderbuffer and
+ * writing the clear color to it.  Called by the driver's implementation
+ * of the glClear function.
+ */
+void
+_mesa_clear_accum_buffer(struct gl_context *ctx)
+{
+   GLuint x, y, width, height;
+   GLubyte *accMap;
+   GLint accRowStride;
+   struct gl_renderbuffer *accRb;
+
+   if (!ctx->DrawBuffer)
+      return;
+
+   accRb = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
+   if (!accRb)
+      return;   /* missing accum buffer, not an error */
+
+   /* bounds, with scissor */
+   x = ctx->DrawBuffer->_Xmin;
+   y = ctx->DrawBuffer->_Ymin;
+   width = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
+   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
+
+   ctx->Driver.MapRenderbuffer(ctx, accRb, x, y, width, height,
+                               GL_MAP_WRITE_BIT, &accMap, &accRowStride);
+
+   if (!accMap) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      return;
+   }
+
+   if (accRb->Format == MESA_FORMAT_SIGNED_RGBA_16) {
+      const GLshort clearR = FLOAT_TO_SHORT(ctx->Accum.ClearColor[0]);
+      const GLshort clearG = FLOAT_TO_SHORT(ctx->Accum.ClearColor[1]);
+      const GLshort clearB = FLOAT_TO_SHORT(ctx->Accum.ClearColor[2]);
+      const GLshort clearA = FLOAT_TO_SHORT(ctx->Accum.ClearColor[3]);
+      GLuint i, j;
+
+      for (j = 0; j < height; j++) {
+         GLshort *row = (GLshort *) accMap;
+         
+         for (i = 0; i < width; i++) {
+            row[i * 4 + 0] = clearR;
+            row[i * 4 + 1] = clearG;
+            row[i * 4 + 2] = clearB;
+            row[i * 4 + 3] = clearA;
+         }
+         accMap += accRowStride;
+      }
+   }
+   else {
+      /* other types someday? */
+      _mesa_warning(ctx, "unexpected accum buffer type");
+   }
+
+   ctx->Driver.UnmapRenderbuffer(ctx, accRb);
+}
+
+
+/**
+ * if (bias)
+ *    Accum += value
+ * else
+ *    Accum *= value
+ */
+static void
+accum_scale_or_bias(struct gl_context *ctx, GLfloat value,
+                    GLint xpos, GLint ypos, GLint width, GLint height,
+                    GLboolean bias)
+{
+   struct gl_renderbuffer *accRb =
+      ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
+   GLubyte *accMap;
+   GLint accRowStride;
+
+   assert(accRb);
+
+   ctx->Driver.MapRenderbuffer(ctx, accRb, xpos, ypos, width, height,
+                               GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
+                               &accMap, &accRowStride);
+
+   if (!accMap) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      return;
+   }
+
+   if (accRb->Format == MESA_FORMAT_SIGNED_RGBA_16) {
+      const GLshort incr = (GLshort) (value * 32767.0f);
+      GLuint i, j;
+      if (bias) {
+         for (j = 0; j < height; j++) {
+            GLshort *acc = (GLshort *) accMap;
+            for (i = 0; i < 4 * width; i++) {
+               acc[i] += incr;
+            }
+            accMap += accRowStride;
+         }
+      }
+      else {
+         /* scale */
+         for (j = 0; j < height; j++) {
+            GLshort *acc = (GLshort *) accMap;
+            for (i = 0; i < 4 * width; i++) {
+               acc[i] = (GLshort) (acc[i] * value);
+            }
+            accMap += accRowStride;
+         }
+      }
+   }
+   else {
+      /* other types someday? */
+   }
+
+   ctx->Driver.UnmapRenderbuffer(ctx, accRb);
+}
+
+
+/**
+ * if (load)
+ *    Accum = ColorBuf * value
+ * else
+ *    Accum += ColorBuf * value
+ */
+static void
+accum_or_load(struct gl_context *ctx, GLfloat value,
+              GLint xpos, GLint ypos, GLint width, GLint height,
+              GLboolean load)
+{
+   struct gl_renderbuffer *accRb =
+      ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
+   struct gl_renderbuffer *colorRb = ctx->ReadBuffer->_ColorReadBuffer;
+   GLubyte *accMap, *colorMap;
+   GLint accRowStride, colorRowStride;
+   GLbitfield mappingFlags;
+
+   if (!colorRb) {
+      /* no read buffer - OK */
+      return;
+   }
+
+   assert(accRb);
+
+   mappingFlags = GL_MAP_WRITE_BIT;
+   if (!load) /* if we're accumulating */
+      mappingFlags |= GL_MAP_READ_BIT;
+
+   /* Map accum buffer */
+   ctx->Driver.MapRenderbuffer(ctx, accRb, xpos, ypos, width, height,
+                               mappingFlags, &accMap, &accRowStride);
+   if (!accMap) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      return;
+   }
+
+   /* Map color buffer */
+   ctx->Driver.MapRenderbuffer(ctx, colorRb, xpos, ypos, width, height,
+                               GL_MAP_READ_BIT,
+                               &colorMap, &colorRowStride);
+   if (!colorMap) {
+      ctx->Driver.UnmapRenderbuffer(ctx, accRb);
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      return;
+   }
+
+   if (accRb->Format == MESA_FORMAT_SIGNED_RGBA_16) {
+      const GLfloat scale = value * 32767.0f;
+      GLuint i, j;
+      GLfloat (*rgba)[4];
+
+      rgba = (GLfloat (*)[4]) malloc(width * 4 * sizeof(GLfloat));
+      if (rgba) {
+         for (j = 0; j < height; j++) {
+            GLshort *acc = (GLshort *) accMap;
+
+            /* read colors from source color buffer */
+            _mesa_unpack_rgba_row(colorRb->Format, width, colorMap, rgba);
+
+            if (load) {
+               for (i = 0; i < width; i++) {
+                  acc[i * 4 + 0] = (GLshort) (rgba[i][RCOMP] * scale);
+                  acc[i * 4 + 1] = (GLshort) (rgba[i][GCOMP] * scale);
+                  acc[i * 4 + 2] = (GLshort) (rgba[i][BCOMP] * scale);
+                  acc[i * 4 + 3] = (GLshort) (rgba[i][ACOMP] * scale);
+               }
+            }
+            else {
+               /* accumulate */
+               for (i = 0; i < width; i++) {
+                  acc[i * 4 + 0] += (GLshort) (rgba[i][RCOMP] * scale);
+                  acc[i * 4 + 1] += (GLshort) (rgba[i][GCOMP] * scale);
+                  acc[i * 4 + 2] += (GLshort) (rgba[i][BCOMP] * scale);
+                  acc[i * 4 + 3] += (GLshort) (rgba[i][ACOMP] * scale);
+               }
+            }
+
+            colorMap += colorRowStride;
+            accMap += accRowStride;
+         }
+
+         free(rgba);
+      }
+      else {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      }
+   }
+   else {
+      /* other types someday? */
+   }
+
+   ctx->Driver.UnmapRenderbuffer(ctx, accRb);
+   ctx->Driver.UnmapRenderbuffer(ctx, colorRb);
+}
+
+
+/**
+ * ColorBuffer = Accum * value
+ */
+static void
+accum_return(struct gl_context *ctx, GLfloat value,
+             GLint xpos, GLint ypos, GLint width, GLint height)
+{
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   struct gl_renderbuffer *accRb = fb->Attachment[BUFFER_ACCUM].Renderbuffer;
+   GLubyte *accMap, *colorMap;
+   GLint accRowStride, colorRowStride;
+   GLuint buffer;
+
+   /* Map accum buffer */
+   ctx->Driver.MapRenderbuffer(ctx, accRb, xpos, ypos, width, height,
+                               GL_MAP_READ_BIT,
+                               &accMap, &accRowStride);
+   if (!accMap) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+      return;
+   }
+
+   /* Loop over destination buffers */
+   for (buffer = 0; buffer < fb->_NumColorDrawBuffers; buffer++) {
+      struct gl_renderbuffer *colorRb = fb->_ColorDrawBuffers[buffer];
+      const GLboolean masking = (!ctx->Color.ColorMask[buffer][RCOMP] ||
+                                 !ctx->Color.ColorMask[buffer][GCOMP] ||
+                                 !ctx->Color.ColorMask[buffer][BCOMP] ||
+                                 !ctx->Color.ColorMask[buffer][ACOMP]);
+      GLbitfield mappingFlags = GL_MAP_WRITE_BIT;
+
+      if (masking)
+         mappingFlags |= GL_MAP_READ_BIT;
+
+      /* Map color buffer */
+      ctx->Driver.MapRenderbuffer(ctx, colorRb, xpos, ypos, width, height,
+                                  mappingFlags, &colorMap, &colorRowStride);
+      if (!colorMap) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+         continue;
+      }
+
+      if (accRb->Format == MESA_FORMAT_SIGNED_RGBA_16) {
+         const GLfloat scale = value / 32767.0f;
+         GLint i, j;
+         GLfloat (*rgba)[4], (*dest)[4];
+
+         rgba = (GLfloat (*)[4]) malloc(width * 4 * sizeof(GLfloat));
+         dest = (GLfloat (*)[4]) malloc(width * 4 * sizeof(GLfloat));
+
+         if (rgba && dest) {
+            for (j = 0; j < height; j++) {
+               GLshort *acc = (GLshort *) accMap;
+
+               for (i = 0; i < width; i++) {
+                  rgba[i][0] = acc[i * 4 + 0] * scale;
+                  rgba[i][1] = acc[i * 4 + 1] * scale;
+                  rgba[i][2] = acc[i * 4 + 2] * scale;
+                  rgba[i][3] = acc[i * 4 + 3] * scale;
+               }
+
+               if (masking) {
+
+                  /* get existing colors from dest buffer */
+                  _mesa_unpack_rgba_row(colorRb->Format, width, colorMap, dest);
+
+                  /* use the dest colors where mask[channel] = 0 */
+                  if (ctx->Color.ColorMask[buffer][RCOMP] == 0) {
+                     for (i = 0; i < width; i++)
+                        rgba[i][RCOMP] = dest[i][RCOMP];
+                  }
+                  if (ctx->Color.ColorMask[buffer][GCOMP] == 0) {
+                     for (i = 0; i < width; i++)
+                        rgba[i][GCOMP] = dest[i][GCOMP];
+                  }
+                  if (ctx->Color.ColorMask[buffer][BCOMP] == 0) {
+                     for (i = 0; i < width; i++)
+                        rgba[i][BCOMP] = dest[i][BCOMP];
+                  }
+                  if (ctx->Color.ColorMask[buffer][ACOMP] == 0) {
+                     for (i = 0; i < width; i++)
+                        rgba[i][ACOMP] = dest[i][ACOMP];
+                  }
+               }
+
+               _mesa_pack_float_rgba_row(colorRb->Format, width,
+                                         (const GLfloat (*)[4]) rgba, colorMap);
+
+               accMap += accRowStride;
+               colorMap += colorRowStride;
+            }
+         }
+         else {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glAccum");
+         }
+         free(rgba);
+         free(dest);
+      }
+      else {
+         /* other types someday? */
+      }
+
+      ctx->Driver.UnmapRenderbuffer(ctx, colorRb);
+   }
+
+   ctx->Driver.UnmapRenderbuffer(ctx, accRb);
+}
+
+
+
+/**
+ * Software fallback for glAccum.  A hardware driver that supports
+ * signed 16-bit color channels could implement hardware accumulation
+ * operations, but no driver does so at this time.
+ */
+void
+_mesa_accum(struct gl_context *ctx, GLenum op, GLfloat value)
+{
+   GLint xpos, ypos, width, height;
+
+   if (!ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer) {
+      _mesa_warning(ctx, "Calling glAccum() without an accumulation buffer");
+      return;
+   }
+
+   if (!_mesa_check_conditional_render(ctx))
+      return;
+
+   xpos = ctx->DrawBuffer->_Xmin;
+   ypos = ctx->DrawBuffer->_Ymin;
+   width =  ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
+   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
+
+   switch (op) {
+   case GL_ADD:
+      if (value != 0.0F) {
+         accum_scale_or_bias(ctx, value, xpos, ypos, width, height, GL_TRUE);
+      }
+      break;
+   case GL_MULT:
+      if (value != 1.0F) {
+         accum_scale_or_bias(ctx, value, xpos, ypos, width, height, GL_FALSE);
+      }
+      break;
+   case GL_ACCUM:
+      if (value != 0.0F) {
+         accum_or_load(ctx, value, xpos, ypos, width, height, GL_FALSE);
+      }
+      break;
+   case GL_LOAD:
+      accum_or_load(ctx, value, xpos, ypos, width, height, GL_TRUE);
+      break;
+   case GL_RETURN:
+      accum_return(ctx, value, xpos, ypos, width, height);
+      break;
+   default:
+      _mesa_problem(ctx, "invalid mode in _mesa_accum()");
+      break;
+   }
+}
diff --git a/mesalib/src/mesa/main/accum.h b/mesalib/src/mesa/main/accum.h
index f16378cc1..5b3f06aa9 100644
--- a/mesalib/src/mesa/main/accum.h
+++ b/mesalib/src/mesa/main/accum.h
@@ -42,6 +42,7 @@
 
 struct _glapi_table;
 struct gl_context;
+struct gl_renderbuffer;
 
 #if FEATURE_accum
 
@@ -51,6 +52,12 @@ _mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha );
 extern void
 _mesa_init_accum_dispatch(struct _glapi_table *disp);
 
+extern void
+_mesa_accum(struct gl_context *ctx, GLenum op, GLfloat value);
+
+extern void
+_mesa_clear_accum_buffer(struct gl_context *ctx);
+
 #else /* FEATURE_accum */
 
 #include "main/compiler.h"
@@ -67,6 +74,17 @@ _mesa_init_accum_dispatch(struct _glapi_table *disp)
 {
 }
 
+static inline void
+_mesa_accum(struct gl_context *ctx, GLenum op, GLfloat value)
+{
+}
+
+static inline void
+_mesa_clear_accum_buffer(struct gl_context *ctx)
+{
+}
+
+
 #endif /* FEATURE_accum */
 
 extern void
diff --git a/mesalib/src/mesa/main/dd.h b/mesalib/src/mesa/main/dd.h
index 9842540da..d6f70d1c4 100644
--- a/mesalib/src/mesa/main/dd.h
+++ b/mesalib/src/mesa/main/dd.h
@@ -532,24 +532,6 @@ struct dd_function_table {
    void (*UnmapRenderbuffer)(struct gl_context *ctx,
 			     struct gl_renderbuffer *rb);
 
-   /**
-    * Note: no context argument.  This function doesn't initially look
-    * like it belongs here, except that the driver is the only entity
-    * that knows for sure how the texture memory is allocated - via
-    * the above callbacks.  There is then an argument that the driver
-    * knows what memcpy paths might be fast.  Typically this is invoked with
-    * 
-    * to -- a pointer into texture memory allocated by NewTextureImage() above.
-    * from -- a pointer into client memory or a mesa temporary.
-    * sz -- nr bytes to copy.
-    */
-   void* (*TextureMemCpy)( void *to, const void *from, size_t sz );
-
-   /**
-    * Called by glAreTextureResident().
-    */
-   GLboolean (*IsTextureResident)( struct gl_context *ctx,
-                                   struct gl_texture_object *t );
    /*@}*/
 
 
diff --git a/mesalib/src/mesa/main/depthstencil.c b/mesalib/src/mesa/main/depthstencil.c
index 40d6c9612..af5c12f67 100644
--- a/mesalib/src/mesa/main/depthstencil.c
+++ b/mesalib/src/mesa/main/depthstencil.c
@@ -957,158 +957,3 @@ _mesa_new_s8_renderbuffer_wrapper(struct gl_context *ctx, struct gl_renderbuffer
 
    return s8rb;
 }
-
-
-
-/**
- ** The following functions are useful for hardware drivers that only
- ** implement combined depth/stencil buffers.
- ** The GL_EXT_framebuffer_object extension allows indepedent depth and
- ** stencil buffers to be used in any combination.
- ** Therefore, we sometimes have to merge separate depth and stencil
- ** renderbuffers into a single depth+stencil renderbuffer.  And sometimes
- ** we have to split combined depth+stencil renderbuffers into separate
- ** renderbuffers.
- **/
-
-
-/**
- * Extract stencil values from the combined depth/stencil renderbuffer, storing
- * the values into a separate stencil renderbuffer.
- * \param dsRb  the source depth/stencil renderbuffer
- * \param stencilRb  the destination stencil renderbuffer
- *                   (either 8-bit or 32-bit)
- */
-void
-_mesa_extract_stencil(struct gl_context *ctx,
-                      struct gl_renderbuffer *dsRb,
-                      struct gl_renderbuffer *stencilRb)
-{
-   GLuint row, width, height;
-
-   ASSERT(dsRb);
-   ASSERT(stencilRb);
-
-   ASSERT(dsRb->Format == MESA_FORMAT_Z24_S8);
-   ASSERT(dsRb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   ASSERT(stencilRb->Format == MESA_FORMAT_Z24_S8 ||
-          stencilRb->Format == MESA_FORMAT_S8);
-   ASSERT(dsRb->Width == stencilRb->Width);
-   ASSERT(dsRb->Height == stencilRb->Height);
-
-   width = dsRb->Width;
-   height = dsRb->Height;
-
-   for (row = 0; row < height; row++) {
-      GLuint depthStencil[MAX_WIDTH];
-      dsRb->GetRow(ctx, dsRb, width, 0, row, depthStencil);
-      if (stencilRb->Format == MESA_FORMAT_S8) {
-         /* 8bpp stencil */
-         GLubyte stencil[MAX_WIDTH];
-         GLuint i;
-         for (i = 0; i < width; i++) {
-            stencil[i] = depthStencil[i] & 0xff;
-         }
-         stencilRb->PutRow(ctx, stencilRb, width, 0, row, stencil, NULL);
-      }
-      else {
-         /* 32bpp stencil */
-         /* the 24 depth bits will be ignored */
-         ASSERT(stencilRb->Format == MESA_FORMAT_Z24_S8);
-         ASSERT(stencilRb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-         stencilRb->PutRow(ctx, stencilRb, width, 0, row, depthStencil, NULL);
-      }
-   }
-}
-
-
-/**
- * Copy stencil values from a stencil renderbuffer into a combined
- * depth/stencil renderbuffer.
- * \param dsRb  the destination depth/stencil renderbuffer
- * \param stencilRb  the source stencil buffer (either 8-bit or 32-bit)
- */
-void
-_mesa_insert_stencil(struct gl_context *ctx,
-                     struct gl_renderbuffer *dsRb,
-                     struct gl_renderbuffer *stencilRb)
-{
-   GLuint row, width, height;
-
-   ASSERT(dsRb);
-   ASSERT(stencilRb);
-
-   ASSERT(dsRb->Format == MESA_FORMAT_Z24_S8);
-   ASSERT(dsRb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   ASSERT(stencilRb->Format == MESA_FORMAT_Z24_S8 ||
-          stencilRb->Format == MESA_FORMAT_S8);
-
-   ASSERT(dsRb->Width == stencilRb->Width);
-   ASSERT(dsRb->Height == stencilRb->Height);
-
-   width = dsRb->Width;
-   height = dsRb->Height;
-
-   for (row = 0; row < height; row++) {
-      GLuint depthStencil[MAX_WIDTH];
-
-      dsRb->GetRow(ctx, dsRb, width, 0, row, depthStencil);
-
-      if (stencilRb->Format == MESA_FORMAT_S8) {
-         /* 8bpp stencil */
-         GLubyte stencil[MAX_WIDTH];
-         GLuint i;
-         stencilRb->GetRow(ctx, stencilRb, width, 0, row, stencil);
-         for (i = 0; i < width; i++) {
-            depthStencil[i] = (depthStencil[i] & 0xffffff00) | stencil[i];
-         }
-      }
-      else {
-         /* 32bpp stencil buffer */
-         GLuint stencil[MAX_WIDTH], i;
-         ASSERT(stencilRb->Format == MESA_FORMAT_Z24_S8);
-         ASSERT(stencilRb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-         stencilRb->GetRow(ctx, stencilRb, width, 0, row, stencil);
-         for (i = 0; i < width; i++) {
-            depthStencil[i]
-               = (depthStencil[i] & 0xffffff00) | (stencil[i] & 0xff);
-         }
-      }
-
-      dsRb->PutRow(ctx, dsRb, width, 0, row, depthStencil, NULL);
-   }
-}
-
-
-/**
- * Convert the stencil buffer from 8bpp to 32bpp depth/stencil.
- * \param stencilRb  the stencil renderbuffer to promote
- */
-void
-_mesa_promote_stencil(struct gl_context *ctx, struct gl_renderbuffer *stencilRb)
-{
-   const GLsizei width = stencilRb->Width;
-   const GLsizei height = stencilRb->Height;
-   GLubyte *data;
-   GLint i, j, k;
-
-   ASSERT(stencilRb->Format == MESA_FORMAT_S8);
-   ASSERT(stencilRb->Data);
-
-   data = (GLubyte *) stencilRb->Data;
-   stencilRb->Data = NULL;
-   stencilRb->AllocStorage(ctx, stencilRb, GL_DEPTH24_STENCIL8_EXT,
-                           width, height);
-
-   ASSERT(stencilRb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-
-   k = 0;
-   for (i = 0; i < height; i++) {
-      GLuint depthStencil[MAX_WIDTH];
-      for (j = 0; j < width; j++) {
-         depthStencil[j] = data[k++];
-      }
-      stencilRb->PutRow(ctx, stencilRb, width, 0, i, depthStencil, NULL);
-   }
-   free(data);
-}
diff --git a/mesalib/src/mesa/main/depthstencil.h b/mesalib/src/mesa/main/depthstencil.h
index b47a2e482..c3871d805 100644
--- a/mesalib/src/mesa/main/depthstencil.h
+++ b/mesalib/src/mesa/main/depthstencil.h
@@ -43,20 +43,4 @@ _mesa_new_s8_renderbuffer_wrapper(struct gl_context *ctx,
                                   struct gl_renderbuffer *dsrb);
 
 
-extern void
-_mesa_extract_stencil(struct gl_context *ctx,
-                      struct gl_renderbuffer *dsRb,
-                      struct gl_renderbuffer *stencilRb);
-
-
-extern void
-_mesa_insert_stencil(struct gl_context *ctx,
-                     struct gl_renderbuffer *dsRb,
-                     struct gl_renderbuffer *stencilRb);
-
-
-extern void
-_mesa_promote_stencil(struct gl_context *ctx, struct gl_renderbuffer *stencilRb);
-
-
 #endif /* DEPTHSTENCIL_H */
diff --git a/mesalib/src/mesa/main/format_pack.c b/mesalib/src/mesa/main/format_pack.c
new file mode 100644
index 000000000..390b494c0
--- /dev/null
+++ b/mesalib/src/mesa/main/format_pack.c
@@ -0,0 +1,2496 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (c) 2011 VMware, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS 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.
+ */
+
+
+/**
+ * Color, depth, stencil packing functions.
+ * Used to pack basic color, depth and stencil formats to specific
+ * hardware formats.
+ *
+ * There are both per-pixel and per-row packing functions:
+ * - The former will be used by swrast to write values to the color, depth,
+ *   stencil buffers when drawing points, lines and masked spans.
+ * - The later will be used for image-oriented functions like glDrawPixels,
+ *   glAccum, and glTexImage.
+ */
+
+
+#include "colormac.h"
+#include "format_pack.h"
+#include "macros.h"
+#include "../../gallium/auxiliary/util/u_format_rgb9e5.h"
+#include "../../gallium/auxiliary/util/u_format_r11g11b10f.h"
+
+
+typedef void (*pack_ubyte_rgba_row_func)(GLuint n,
+                                         const GLubyte src[][4], void *dst);
+
+typedef void (*pack_float_rgba_row_func)(GLuint n,
+                                         const GLfloat src[][4], void *dst);
+
+
+
+static inline GLfloat
+linear_to_srgb(GLfloat cl)
+{
+   if (cl < 0.0f)
+      return 0.0f;
+   else if (cl < 0.0031308f)
+      return 12.92f * cl;
+   else if (cl < 1.0f)
+      return 1.055f * powf(cl, 0.41666f) - 0.055f;
+   else
+      return 1.0f;
+}
+
+
+static inline GLubyte
+linear_float_to_srgb_ubyte(GLfloat cl)
+{
+   GLubyte res = FLOAT_TO_UBYTE(linear_to_srgb(cl));
+   return res;
+}
+
+
+static inline GLubyte
+linear_ubyte_to_srgb_ubyte(GLubyte cl)
+{
+   GLubyte res = FLOAT_TO_UBYTE(linear_to_srgb(cl / 255.0f));
+   return res;
+}
+
+
+
+
+/*
+ * MESA_FORMAT_RGBA8888
+ */
+
+static void
+pack_ubyte_RGBA8888(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(src[RCOMP], src[GCOMP], src[BCOMP], src[ACOMP]);
+}
+
+static void
+pack_float_RGBA8888(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_RGBA8888(v, dst);
+}
+
+static void
+pack_row_ubyte_RGBA8888(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(src[i][RCOMP], src[i][GCOMP],
+                             src[i][BCOMP], src[i][ACOMP]);
+   }
+}
+
+static void
+pack_row_float_RGBA8888(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_RGBA8888(v, d + i);
+   }
+}
+
+
+
+/*
+ * MESA_FORMAT_RGBA8888_REV
+ */
+
+static void
+pack_ubyte_RGBA8888_REV(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(src[ACOMP], src[BCOMP], src[GCOMP], src[RCOMP]);
+}
+
+static void
+pack_float_RGBA8888_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_RGBA8888_REV(v, dst);
+}
+
+static void
+pack_row_ubyte_RGBA8888_REV(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(src[i][ACOMP], src[i][BCOMP],
+                             src[i][GCOMP], src[i][RCOMP]);
+   }
+}
+
+static void
+pack_row_float_RGBA8888_REV(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_RGBA8888_REV(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_ARGB8888
+ */
+
+static void
+pack_ubyte_ARGB8888(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(src[ACOMP], src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_ARGB8888(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB8888(v, dst);
+}
+
+static void
+pack_row_ubyte_ARGB8888(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(src[i][ACOMP], src[i][RCOMP],
+                             src[i][GCOMP], src[i][BCOMP]);
+   }
+}
+
+static void
+pack_row_float_ARGB8888(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_ARGB8888(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_ARGB8888_REV
+ */
+
+static void
+pack_ubyte_ARGB8888_REV(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(src[BCOMP], src[GCOMP], src[RCOMP], src[ACOMP]);
+}
+
+static void
+pack_float_ARGB8888_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB8888_REV(v, dst);
+}
+
+static void
+pack_row_ubyte_ARGB8888_REV(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(src[i][BCOMP], src[i][GCOMP],
+                             src[i][RCOMP], src[i][ACOMP]);
+   }
+}
+
+static void
+pack_row_float_ARGB8888_REV(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_ARGB8888_REV(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_XRGB8888
+ */
+
+static void
+pack_ubyte_XRGB8888(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(0x0, src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_XRGB8888(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_XRGB8888(v, dst);
+}
+
+static void
+pack_row_ubyte_XRGB8888(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(0, src[i][RCOMP], src[i][GCOMP], src[i][BCOMP]);
+   }
+}
+
+static void
+pack_row_float_XRGB8888(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_XRGB8888(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_XRGB8888_REV
+ */
+
+static void
+pack_ubyte_XRGB8888_REV(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = PACK_COLOR_8888(src[BCOMP], src[GCOMP], src[RCOMP], 0);
+}
+
+static void
+pack_float_XRGB8888_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_XRGB8888_REV(v, dst);
+}
+
+static void
+pack_row_ubyte_XRGB8888_REV(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i] = PACK_COLOR_8888(src[i][BCOMP], src[i][GCOMP], src[i][RCOMP], 0);
+   }
+}
+
+static void
+pack_row_float_XRGB8888_REV(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_XRGB8888_REV(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_RGB888
+ */
+
+static void
+pack_ubyte_RGB888(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   d[2] = src[RCOMP];
+   d[1] = src[GCOMP];
+   d[0] = src[BCOMP];
+}
+
+static void
+pack_float_RGB888(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[2], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[1], src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[0], src[BCOMP]);
+}
+
+static void
+pack_row_ubyte_RGB888(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i*3+2] = src[i][RCOMP];
+      d[i*3+1] = src[i][GCOMP];
+      d[i*3+0] = src[i][BCOMP];
+   }
+}
+
+static void
+pack_row_float_RGB888(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      d[i*3+2] = v[RCOMP];
+      d[i*3+1] = v[GCOMP];
+      d[i*3+0] = v[BCOMP];
+   }
+}
+
+
+/*
+ * MESA_FORMAT_BGR888
+ */
+
+static void
+pack_ubyte_BGR888(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   d[2] = src[BCOMP];
+   d[1] = src[GCOMP];
+   d[0] = src[RCOMP];
+}
+
+static void
+pack_float_BGR888(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[2], src[BCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[1], src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[0], src[RCOMP]);
+}
+
+static void
+pack_row_ubyte_BGR888(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      d[i*3+2] = src[i][BCOMP];
+      d[i*3+1] = src[i][GCOMP];
+      d[i*3+0] = src[i][RCOMP];
+   }
+}
+
+static void
+pack_row_float_BGR888(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      d[i*3+2] = v[BCOMP];
+      d[i*3+1] = v[GCOMP];
+      d[i*3+0] = v[RCOMP];
+   }
+}
+
+
+/*
+ * MESA_FORMAT_RGB565
+ */
+
+static void
+pack_ubyte_RGB565(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_565(src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_RGB565(const GLfloat src[4], void *dst)
+{
+   GLubyte v[3];
+   UNCLAMPED_FLOAT_TO_UBYTE(v[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[1], src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[2], src[BCOMP]);
+   pack_ubyte_RGB565(v, dst);
+}
+
+static void
+pack_row_ubyte_RGB565(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      pack_ubyte_RGB565(src[i], d + i);
+   }
+}
+
+static void
+pack_row_float_RGB565(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_RGB565(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_RGB565_REV
+ */
+
+static void
+pack_ubyte_RGB565_REV(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_565_REV(src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_RGB565_REV(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLubyte r, g, b;
+   UNCLAMPED_FLOAT_TO_UBYTE(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(g, src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(b, src[BCOMP]);
+   *d = PACK_COLOR_565_REV(r, g, b);
+}
+
+static void
+pack_row_ubyte_RGB565_REV(GLuint n, const GLubyte src[][4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      pack_ubyte_RGB565_REV(src[i], d + i);
+   }
+}
+
+static void
+pack_row_float_RGB565_REV(GLuint n, const GLfloat src[][4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLuint i;
+   for (i = 0; i < n; i++) {
+      GLubyte v[4];
+      _mesa_unclamped_float_rgba_to_ubyte(v, src[i]);
+      pack_ubyte_RGB565_REV(v, d + i);
+   }
+}
+
+
+/*
+ * MESA_FORMAT_ARGB4444
+ */
+
+static void
+pack_ubyte_ARGB4444(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_4444(src[ACOMP], src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_ARGB4444(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB4444(v, dst);
+}
+
+/* use fallback row packing functions */
+
+
+/*
+ * MESA_FORMAT_ARGB4444_REV
+ */
+
+static void
+pack_ubyte_ARGB4444_REV(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_4444(src[GCOMP], src[BCOMP], src[ACOMP], src[RCOMP]);
+}
+
+static void
+pack_float_ARGB4444_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB4444_REV(v, dst);
+}
+
+/* use fallback row packing functions */
+
+
+/*
+ * MESA_FORMAT_RGBA5551
+ */
+
+static void
+pack_ubyte_RGBA5551(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_5551(src[RCOMP], src[GCOMP], src[BCOMP], src[ACOMP]);
+}
+
+static void
+pack_float_RGBA5551(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_RGBA5551(v, dst);
+}
+
+/* use fallback row packing functions */
+
+
+/*
+ * MESA_FORMAT_ARGB1555
+ */
+
+static void
+pack_ubyte_ARGB1555(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_1555(src[ACOMP], src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_ARGB1555(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB1555(v, dst);
+}
+
+
+/* MESA_FORMAT_ARGB1555_REV */
+
+static void
+pack_ubyte_ARGB1555_REV(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst), tmp;
+   tmp = PACK_COLOR_1555(src[ACOMP], src[RCOMP], src[GCOMP], src[BCOMP]);
+   *d = (tmp >> 8) | (tmp << 8);
+}
+
+static void
+pack_float_ARGB1555_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   _mesa_unclamped_float_rgba_to_ubyte(v, src);
+   pack_ubyte_ARGB1555_REV(v, dst);
+}
+
+
+/* MESA_FORMAT_AL44 */
+
+static void
+pack_ubyte_AL44(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_44(src[ACOMP], src[RCOMP]);
+}
+
+static void
+pack_float_AL44(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   UNCLAMPED_FLOAT_TO_UBYTE(v[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[3], src[ACOMP]);
+   pack_ubyte_AL44(v, dst);
+}
+
+
+/* MESA_FORMAT_AL88 */
+
+static void
+pack_ubyte_AL88(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_88(src[ACOMP], src[RCOMP]);
+}
+
+static void
+pack_float_AL88(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   UNCLAMPED_FLOAT_TO_UBYTE(v[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[3], src[ACOMP]);
+   pack_ubyte_AL88(v, dst);
+}
+
+
+/* MESA_FORMAT_AL88_REV */
+
+static void
+pack_ubyte_AL88_REV(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_88(src[RCOMP], src[ACOMP]);
+}
+
+static void
+pack_float_AL88_REV(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   UNCLAMPED_FLOAT_TO_UBYTE(v[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[3], src[ACOMP]);
+   pack_ubyte_AL88_REV(v, dst);
+}
+
+
+/* MESA_FORMAT_AL1616 */
+
+static void
+pack_ubyte_AL1616(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort l = UBYTE_TO_USHORT(src[RCOMP]);
+   GLushort a = UBYTE_TO_USHORT(src[ACOMP]);
+   *d = PACK_COLOR_1616(a, l);
+}
+
+static void
+pack_float_AL1616(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort l, a;
+   UNCLAMPED_FLOAT_TO_USHORT(l, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(a, src[ACOMP]);
+   *d = PACK_COLOR_1616(a, l);
+}
+
+
+/* MESA_FORMAT_AL1616_REV */
+
+static void
+pack_ubyte_AL1616_REV(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort l = UBYTE_TO_USHORT(src[RCOMP]);
+   GLushort a = UBYTE_TO_USHORT(src[ACOMP]);
+   *d = PACK_COLOR_1616(l, a);
+}
+
+static void
+pack_float_AL1616_REV(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort l, a;
+   UNCLAMPED_FLOAT_TO_USHORT(l, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(a, src[ACOMP]);
+   *d = PACK_COLOR_1616(l, a);
+}
+
+
+/* MESA_FORMAT_RGB332 */
+
+static void
+pack_ubyte_RGB332(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = PACK_COLOR_332(src[RCOMP], src[GCOMP], src[BCOMP]);
+}
+
+static void
+pack_float_RGB332(const GLfloat src[4], void *dst)
+{
+   GLubyte v[4];
+   UNCLAMPED_FLOAT_TO_UBYTE(v[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[1], src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(v[2], src[BCOMP]);
+   pack_ubyte_RGB332(v, dst);
+}
+
+
+/* MESA_FORMAT_A8 */
+
+static void
+pack_ubyte_A8(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = src[ACOMP];
+}
+
+static void
+pack_float_A8(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[0], src[ACOMP]);
+}
+
+
+/* MESA_FORMAT_A16 */
+
+static void
+pack_ubyte_A16(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = UBYTE_TO_USHORT(src[ACOMP]);
+}
+
+static void
+pack_float_A16(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   UNCLAMPED_FLOAT_TO_USHORT(d[0], src[ACOMP]);
+}
+
+
+/* MESA_FORMAT_L8 */
+
+static void
+pack_ubyte_L8(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = src[RCOMP];
+}
+
+static void
+pack_float_L8(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   UNCLAMPED_FLOAT_TO_UBYTE(d[0], src[RCOMP]);
+}
+
+
+/* MESA_FORMAT_L16 */
+
+static void
+pack_ubyte_L16(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = UBYTE_TO_USHORT(src[RCOMP]);
+}
+
+static void
+pack_float_L16(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   UNCLAMPED_FLOAT_TO_USHORT(d[0], src[RCOMP]);
+}
+
+
+/* MESA_FORMAT_YCBCR */
+
+static void
+pack_ubyte_YCBCR(const GLubyte src[4], void *dst)
+{
+   /* todo */
+}
+
+static void
+pack_float_YCBCR(const GLfloat src[4], void *dst)
+{
+   /* todo */
+}
+
+
+/* MESA_FORMAT_YCBCR_REV */
+
+static void
+pack_ubyte_YCBCR_REV(const GLubyte src[4], void *dst)
+{
+   /* todo */
+}
+
+static void
+pack_float_YCBCR_REV(const GLfloat src[4], void *dst)
+{
+   /* todo */
+}
+
+
+/* MESA_FORMAT_R8 */
+
+static void
+pack_ubyte_R8(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = src[RCOMP];
+}
+
+static void
+pack_float_R8(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLubyte r;
+   UNCLAMPED_FLOAT_TO_UBYTE(r, src[RCOMP]);
+   d[0] = r;
+}
+
+
+/* MESA_FORMAT_GR88 */
+
+static void
+pack_ubyte_GR88(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = PACK_COLOR_88(src[GCOMP], src[RCOMP]);
+}
+
+static void
+pack_float_GR88(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLubyte r, g;
+   UNCLAMPED_FLOAT_TO_UBYTE(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(g, src[GCOMP]);
+   *d = PACK_COLOR_88(g, r);
+}
+
+
+/* MESA_FORMAT_RG88 */
+
+static void
+pack_ubyte_RG88(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = PACK_COLOR_88(src[RCOMP], src[GCOMP]);
+}
+
+static void
+pack_float_RG88(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLubyte r, g;
+   UNCLAMPED_FLOAT_TO_UBYTE(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(g, src[GCOMP]);
+   *d = PACK_COLOR_88(r, g);
+}
+
+
+/* MESA_FORMAT_R16 */
+
+static void
+pack_ubyte_R16(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = UBYTE_TO_USHORT(src[RCOMP]);
+}
+
+static void
+pack_float_R16(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   UNCLAMPED_FLOAT_TO_USHORT(d[0], src[RCOMP]);
+}
+
+
+/* MESA_FORMAT_RG1616 */
+
+static void
+pack_ubyte_RG1616(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r = UBYTE_TO_USHORT(src[RCOMP]);
+   GLushort g = UBYTE_TO_USHORT(src[GCOMP]);
+   *d = PACK_COLOR_1616(g, r);
+}
+
+static void
+pack_float_RG1616(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r, g;
+   UNCLAMPED_FLOAT_TO_USHORT(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(g, src[GCOMP]);
+   *d = PACK_COLOR_1616(g, r);
+}
+
+
+/* MESA_FORMAT_RG1616_REV */
+
+static void
+pack_ubyte_RG1616_REV(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r = UBYTE_TO_USHORT(src[RCOMP]);
+   GLushort g = UBYTE_TO_USHORT(src[GCOMP]);
+   *d = PACK_COLOR_1616(r, g);
+}
+
+
+static void
+pack_float_RG1616_REV(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r, g;
+   UNCLAMPED_FLOAT_TO_USHORT(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(g, src[GCOMP]);
+   *d = PACK_COLOR_1616(r, g);
+}
+
+
+/* MESA_FORMAT_ARGB2101010 */
+
+static void
+pack_ubyte_ARGB2101010(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r = UBYTE_TO_USHORT(src[RCOMP]);
+   GLushort g = UBYTE_TO_USHORT(src[GCOMP]);
+   GLushort b = UBYTE_TO_USHORT(src[BCOMP]);
+   GLushort a = UBYTE_TO_USHORT(src[ACOMP]);
+   *d = PACK_COLOR_2101010_US(a, r, g, b);
+}
+
+static void
+pack_float_ARGB2101010(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLushort r, g, b, a;
+   UNCLAMPED_FLOAT_TO_USHORT(r, src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(g, src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(b, src[BCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(a, src[ACOMP]);
+   *d = PACK_COLOR_2101010_US(a, r, g, b);
+}
+
+
+/* MESA_FORMAT_SRGB8 */
+
+static void
+pack_ubyte_SRGB8(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   d[2] = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   d[1] = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   d[0] = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+}
+
+static void
+pack_float_SRGB8(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   d[2] = linear_float_to_srgb_ubyte(src[RCOMP]);
+   d[1] = linear_float_to_srgb_ubyte(src[GCOMP]);
+   d[0] = linear_float_to_srgb_ubyte(src[BCOMP]);
+}
+
+
+/* MESA_FORMAT_SRGBA8 */
+
+static void
+pack_ubyte_SRGBA8(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLubyte r = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   GLubyte g = linear_ubyte_to_srgb_ubyte(src[GCOMP]);
+   GLubyte b = linear_ubyte_to_srgb_ubyte(src[BCOMP]);
+   *d = PACK_COLOR_8888(r, g, b, src[ACOMP]);
+}
+
+static void
+pack_float_SRGBA8(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLubyte r, g, b, a;
+   r = linear_float_to_srgb_ubyte(src[RCOMP]);
+   g = linear_float_to_srgb_ubyte(src[GCOMP]);
+   b = linear_float_to_srgb_ubyte(src[BCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(a, src[ACOMP]);
+   *d = PACK_COLOR_8888(r, g, b, a);
+}
+
+
+/* MESA_FORMAT_SARGB8 */
+
+static void
+pack_ubyte_SARGB8(const GLubyte src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLubyte r = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   GLubyte g = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   GLubyte b = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   *d = PACK_COLOR_8888(src[ACOMP], r, g, b);
+}
+
+static void
+pack_float_SARGB8(const GLfloat src[4], void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   GLubyte r, g, b, a;
+   r = linear_float_to_srgb_ubyte(src[RCOMP]);
+   g = linear_float_to_srgb_ubyte(src[GCOMP]);
+   b = linear_float_to_srgb_ubyte(src[BCOMP]);
+   UNCLAMPED_FLOAT_TO_UBYTE(a, src[ACOMP]);
+   *d = PACK_COLOR_8888(a, r, g, b);
+}
+
+
+/* MESA_FORMAT_SL8 */
+
+static void
+pack_ubyte_SL8(const GLubyte src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   *d = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+}
+
+static void
+pack_float_SL8(const GLfloat src[4], void *dst)
+{
+   GLubyte *d = ((GLubyte *) dst);
+   GLubyte l = linear_float_to_srgb_ubyte(src[RCOMP]);
+   *d = l;
+}
+
+
+/* MESA_FORMAT_SLA8 */
+
+static void
+pack_ubyte_SLA8(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLubyte l = linear_ubyte_to_srgb_ubyte(src[RCOMP]);
+   *d = PACK_COLOR_88(src[ACOMP], l);
+}
+
+static void
+pack_float_SLA8(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLubyte a, l = linear_float_to_srgb_ubyte(src[RCOMP]);
+   CLAMPED_FLOAT_TO_UBYTE(a, src[ACOMP]);
+   *d = PACK_COLOR_88(a, l);
+}
+
+
+/* MESA_FORMAT_RGBA_FLOAT32 */
+
+static void
+pack_ubyte_RGBA_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[0]);
+   d[1] = UBYTE_TO_FLOAT(src[1]);
+   d[2] = UBYTE_TO_FLOAT(src[2]);
+   d[3] = UBYTE_TO_FLOAT(src[3]);
+}
+
+static void
+pack_float_RGBA_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[0];
+   d[1] = src[1];
+   d[2] = src[2];
+   d[3] = src[3];
+}
+
+
+/* MESA_FORMAT_RGBA_FLOAT16 */
+
+static void
+pack_ubyte_RGBA_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[0]));
+   d[1] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[1]));
+   d[2] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[2]));
+   d[3] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[3]));
+}
+
+static void
+pack_float_RGBA_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[0]);
+   d[1] = _mesa_float_to_half(src[1]);
+   d[2] = _mesa_float_to_half(src[2]);
+   d[3] = _mesa_float_to_half(src[3]);
+}
+
+
+/* MESA_FORMAT_RGB_FLOAT32 */
+
+static void
+pack_ubyte_RGB_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[0]);
+   d[1] = UBYTE_TO_FLOAT(src[1]);
+   d[2] = UBYTE_TO_FLOAT(src[2]);
+}
+
+static void
+pack_float_RGB_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[0];
+   d[1] = src[1];
+   d[2] = src[2];
+}
+
+
+/* MESA_FORMAT_RGB_FLOAT16 */
+
+static void
+pack_ubyte_RGB_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[0]));
+   d[1] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[1]));
+   d[2] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[2]));
+}
+
+static void
+pack_float_RGB_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[0]);
+   d[1] = _mesa_float_to_half(src[1]);
+   d[2] = _mesa_float_to_half(src[2]);
+}
+
+
+/* MESA_FORMAT_ALPHA_FLOAT32 */
+
+static void
+pack_ubyte_ALPHA_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[ACOMP]);
+}
+
+static void
+pack_float_ALPHA_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[ACOMP];
+}
+
+
+/* MESA_FORMAT_ALPHA_FLOAT16 */
+
+static void
+pack_ubyte_ALPHA_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[ACOMP]));
+}
+
+static void
+pack_float_ALPHA_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[ACOMP]);
+}
+
+
+/* MESA_FORMAT_LUMINANCE_FLOAT32 (and INTENSITY_FLOAT32, R_FLOAT32) */
+
+static void
+pack_ubyte_LUMINANCE_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[RCOMP]);
+}
+
+static void
+pack_float_LUMINANCE_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[RCOMP];
+}
+
+
+/* MESA_FORMAT_LUMINANCE_FLOAT16 (and INTENSITY_FLOAT16, R_FLOAT32) */
+
+static void
+pack_ubyte_LUMINANCE_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[RCOMP]));
+}
+
+static void
+pack_float_LUMINANCE_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[RCOMP]);
+}
+
+
+/* MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32 */
+
+static void
+pack_ubyte_LUMINANCE_ALPHA_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[RCOMP]);
+   d[1] = UBYTE_TO_FLOAT(src[ACOMP]);
+}
+
+static void
+pack_float_LUMINANCE_ALPHA_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[RCOMP];
+   d[1] = src[ACOMP];
+}
+
+
+/* MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16 */
+
+static void
+pack_ubyte_LUMINANCE_ALPHA_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[RCOMP]));
+   d[1] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[ACOMP]));
+}
+
+static void
+pack_float_LUMINANCE_ALPHA_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[RCOMP]);
+   d[1] = _mesa_float_to_half(src[ACOMP]);
+}
+
+
+/* MESA_FORMAT_RG_FLOAT32 */
+
+static void
+pack_ubyte_RG_FLOAT32(const GLubyte src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = UBYTE_TO_FLOAT(src[RCOMP]);
+   d[1] = UBYTE_TO_FLOAT(src[GCOMP]);
+}
+
+static void
+pack_float_RG_FLOAT32(const GLfloat src[4], void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[0] = src[RCOMP];
+   d[1] = src[GCOMP];
+}
+
+
+/* MESA_FORMAT_RG_FLOAT16 */
+
+static void
+pack_ubyte_RG_FLOAT16(const GLubyte src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[RCOMP]));
+   d[1] = _mesa_float_to_half(UBYTE_TO_FLOAT(src[GCOMP]));
+}
+
+static void
+pack_float_RG_FLOAT16(const GLfloat src[4], void *dst)
+{
+   GLhalfARB *d = ((GLhalfARB *) dst);
+   d[0] = _mesa_float_to_half(src[RCOMP]);
+   d[1] = _mesa_float_to_half(src[GCOMP]);
+}
+
+
+/* MESA_FORMAT_DUDV8 */
+
+static void
+pack_ubyte_DUDV8(const GLubyte src[4], void *dst)
+{
+   /* XXX is this ever used? */
+   GLushort *d = ((GLushort *) dst);
+   *d = PACK_COLOR_88(src[0], src[1]);
+}
+
+static void
+pack_float_DUDV8(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   GLbyte du, dv;
+   du = FLOAT_TO_BYTE(CLAMP(src[0], 0.0F, 1.0F));
+   dv = FLOAT_TO_BYTE(CLAMP(src[1], 0.0F, 1.0F));
+   *d = PACK_COLOR_88(du, dv);
+}
+
+
+/* MESA_FORMAT_RGBA_16 */
+
+static void
+pack_ubyte_RGBA_16(const GLubyte src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   d[0] = UBYTE_TO_USHORT(src[RCOMP]);
+   d[1] = UBYTE_TO_USHORT(src[GCOMP]);
+   d[2] = UBYTE_TO_USHORT(src[BCOMP]);
+   d[3] = UBYTE_TO_USHORT(src[ACOMP]);
+}
+
+static void
+pack_float_RGBA_16(const GLfloat src[4], void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   UNCLAMPED_FLOAT_TO_USHORT(d[0], src[RCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(d[1], src[GCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(d[2], src[BCOMP]);
+   UNCLAMPED_FLOAT_TO_USHORT(d[3], src[ACOMP]);
+}
+
+
+
+/*
+ * MESA_FORMAT_SIGNED_R8
+ */
+
+static void
+pack_float_SIGNED_R8(const GLfloat src[4], void *dst)
+{
+   GLbyte *d = (GLbyte *) dst;
+   *d = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RG88_REV
+ */
+
+static void
+pack_float_SIGNED_RG88_REV(const GLfloat src[4], void *dst)
+{
+   GLushort *d = (GLushort *) dst;
+   GLbyte r = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLbyte g = FLOAT_TO_BYTE(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   *d = (g << 8) | r;
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RGBX8888
+ */
+
+static void
+pack_float_SIGNED_RGBX8888(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLbyte r = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLbyte g = FLOAT_TO_BYTE(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   GLbyte b = FLOAT_TO_BYTE(CLAMP(src[BCOMP], -1.0f, 1.0f));
+   GLbyte a = 127;
+   *d = PACK_COLOR_8888(r, g, b, a);
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RGBA8888
+ */
+
+static void
+pack_float_SIGNED_RGBA8888(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLbyte r = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLbyte g = FLOAT_TO_BYTE(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   GLbyte b = FLOAT_TO_BYTE(CLAMP(src[BCOMP], -1.0f, 1.0f));
+   GLbyte a = FLOAT_TO_BYTE(CLAMP(src[ACOMP], -1.0f, 1.0f));
+   *d = PACK_COLOR_8888(r, g, b, a);
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RGBA8888_REV
+ */
+
+static void
+pack_float_SIGNED_RGBA8888_REV(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLbyte r = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLbyte g = FLOAT_TO_BYTE(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   GLbyte b = FLOAT_TO_BYTE(CLAMP(src[BCOMP], -1.0f, 1.0f));
+   GLbyte a = FLOAT_TO_BYTE(CLAMP(src[ACOMP], -1.0f, 1.0f));
+   *d = PACK_COLOR_8888(a, b, g, r);
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_R16
+ */
+
+static void
+pack_float_SIGNED_R16(const GLfloat src[4], void *dst)
+{
+   GLshort *d = (GLshort *) dst;
+   *d = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_GR1616
+ */
+
+static void
+pack_float_SIGNED_GR1616(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLshort r = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLshort g = FLOAT_TO_SHORT(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   *d = (g << 16) | (r & 0xffff);
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RGB_16
+ */
+
+static void
+pack_float_SIGNED_RGB_16(const GLfloat src[4], void *dst)
+{
+   GLshort *d = (GLshort *) dst;
+   d[0] = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   d[1] = FLOAT_TO_SHORT(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   d[2] = FLOAT_TO_SHORT(CLAMP(src[BCOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_RGBA_16
+ */
+
+static void
+pack_float_SIGNED_RGBA_16(const GLfloat src[4], void *dst)
+{
+   GLshort *d = (GLshort *) dst;
+   d[0] = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   d[1] = FLOAT_TO_SHORT(CLAMP(src[GCOMP], -1.0f, 1.0f));
+   d[2] = FLOAT_TO_SHORT(CLAMP(src[BCOMP], -1.0f, 1.0f));
+   d[3] = FLOAT_TO_SHORT(CLAMP(src[ACOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_A8
+ */
+
+static void
+pack_float_SIGNED_A8(const GLfloat src[4], void *dst)
+{
+   GLbyte *d = (GLbyte *) dst;
+   *d = FLOAT_TO_BYTE(CLAMP(src[ACOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_L8
+ */
+
+static void
+pack_float_SIGNED_L8(const GLfloat src[4], void *dst)
+{
+   GLbyte *d = (GLbyte *) dst;
+   *d = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_AL88
+ */
+
+static void
+pack_float_SIGNED_AL88(const GLfloat src[4], void *dst)
+{
+   GLushort *d = (GLushort *) dst;
+   GLbyte l = FLOAT_TO_BYTE(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLbyte a = FLOAT_TO_BYTE(CLAMP(src[ACOMP], -1.0f, 1.0f));
+   *d = (a << 8) | l;
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_A16
+ */
+
+static void
+pack_float_SIGNED_A16(const GLfloat src[4], void *dst)
+{
+   GLshort *d = (GLshort *) dst;
+   *d = FLOAT_TO_SHORT(CLAMP(src[ACOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_L16
+ */
+
+static void
+pack_float_SIGNED_L16(const GLfloat src[4], void *dst)
+{
+   GLshort *d = (GLshort *) dst;
+   *d = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+}
+
+
+/*
+ * MESA_FORMAT_SIGNED_AL1616
+ */
+
+static void
+pack_float_SIGNED_AL1616(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLshort l = FLOAT_TO_SHORT(CLAMP(src[RCOMP], -1.0f, 1.0f));
+   GLshort a = FLOAT_TO_SHORT(CLAMP(src[ACOMP], -1.0f, 1.0f));
+   *d = PACK_COLOR_1616(a, l);
+}
+
+
+/*
+ * MESA_FORMAT_RGB9_E5_FLOAT;
+ */
+
+static void
+pack_float_RGB9_E5_FLOAT(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   *d = float3_to_rgb9e5(src);
+}
+
+static void
+pack_ubyte_RGB9_E5_FLOAT(const GLubyte src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLfloat rgb[3];
+   rgb[0] = UBYTE_TO_FLOAT(src[RCOMP]);
+   rgb[1] = UBYTE_TO_FLOAT(src[GCOMP]);
+   rgb[2] = UBYTE_TO_FLOAT(src[BCOMP]);
+   *d = float3_to_rgb9e5(rgb);
+}
+
+
+
+/*
+ * MESA_FORMAT_R11_G11_B10_FLOAT;
+ */
+
+static void
+pack_ubyte_R11_G11_B10_FLOAT(const GLubyte src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   GLfloat rgb[3];
+   rgb[0] = UBYTE_TO_FLOAT(src[RCOMP]);
+   rgb[1] = UBYTE_TO_FLOAT(src[GCOMP]);
+   rgb[2] = UBYTE_TO_FLOAT(src[BCOMP]);
+   *d = float3_to_r11g11b10f(rgb);
+}
+
+static void
+pack_float_R11_G11_B10_FLOAT(const GLfloat src[4], void *dst)
+{
+   GLuint *d = (GLuint *) dst;
+   *d = float3_to_r11g11b10f(src);
+}
+
+
+
+/**
+ * Return a function that can pack a GLubyte rgba[4] color.
+ */
+gl_pack_ubyte_rgba_func
+_mesa_get_pack_ubyte_rgba_function(gl_format format)
+{
+   static gl_pack_ubyte_rgba_func table[MESA_FORMAT_COUNT];
+   static GLboolean initialized = GL_FALSE;
+
+   if (!initialized) {
+      memset(table, 0, sizeof(table));
+
+      table[MESA_FORMAT_NONE] = NULL;
+
+      table[MESA_FORMAT_RGBA8888] = pack_ubyte_RGBA8888;
+      table[MESA_FORMAT_RGBA8888_REV] = pack_ubyte_RGBA8888_REV;
+      table[MESA_FORMAT_ARGB8888] = pack_ubyte_ARGB8888;
+      table[MESA_FORMAT_ARGB8888_REV] = pack_ubyte_ARGB8888_REV;
+      table[MESA_FORMAT_RGBX8888] = pack_ubyte_RGBA8888; /* reused */
+      table[MESA_FORMAT_RGBX8888_REV] = pack_ubyte_RGBA8888_REV; /* reused */
+      table[MESA_FORMAT_XRGB8888] = pack_ubyte_XRGB8888;
+      table[MESA_FORMAT_XRGB8888_REV] = pack_ubyte_XRGB8888_REV;
+      table[MESA_FORMAT_RGB888] = pack_ubyte_RGB888;
+      table[MESA_FORMAT_BGR888] = pack_ubyte_BGR888;
+      table[MESA_FORMAT_RGB565] = pack_ubyte_RGB565;
+      table[MESA_FORMAT_RGB565_REV] = pack_ubyte_RGB565_REV;
+      table[MESA_FORMAT_ARGB4444] = pack_ubyte_ARGB4444;
+      table[MESA_FORMAT_ARGB4444_REV] = pack_ubyte_ARGB4444_REV;
+      table[MESA_FORMAT_RGBA5551] = pack_ubyte_RGBA5551;
+      table[MESA_FORMAT_ARGB1555] = pack_ubyte_ARGB1555;
+      table[MESA_FORMAT_ARGB1555_REV] = pack_ubyte_ARGB1555_REV;
+      table[MESA_FORMAT_AL44] = pack_ubyte_AL44;
+      table[MESA_FORMAT_AL88] = pack_ubyte_AL88;
+      table[MESA_FORMAT_AL88_REV] = pack_ubyte_AL88_REV;
+      table[MESA_FORMAT_AL1616] = pack_ubyte_AL1616;
+      table[MESA_FORMAT_AL1616_REV] = pack_ubyte_AL1616_REV;
+      table[MESA_FORMAT_RGB332] = pack_ubyte_RGB332;
+      table[MESA_FORMAT_A8] = pack_ubyte_A8;
+      table[MESA_FORMAT_A16] = pack_ubyte_A16;
+      table[MESA_FORMAT_L8] = pack_ubyte_L8;
+      table[MESA_FORMAT_L16] = pack_ubyte_L16;
+      table[MESA_FORMAT_I8] = pack_ubyte_L8; /* reuse pack_ubyte_L8 */
+      table[MESA_FORMAT_I16] = pack_ubyte_L16; /* reuse pack_ubyte_L16 */
+      table[MESA_FORMAT_YCBCR] = pack_ubyte_YCBCR;
+      table[MESA_FORMAT_YCBCR_REV] = pack_ubyte_YCBCR_REV;
+      table[MESA_FORMAT_R8] = pack_ubyte_R8;
+      table[MESA_FORMAT_GR88] = pack_ubyte_GR88;
+      table[MESA_FORMAT_RG88] = pack_ubyte_RG88;
+      table[MESA_FORMAT_R16] = pack_ubyte_R16;
+      table[MESA_FORMAT_RG1616] = pack_ubyte_RG1616;
+      table[MESA_FORMAT_RG1616_REV] = pack_ubyte_RG1616_REV;
+      table[MESA_FORMAT_ARGB2101010] = pack_ubyte_ARGB2101010;
+
+      /* should never convert RGBA to these formats */
+      table[MESA_FORMAT_Z24_S8] = NULL;
+      table[MESA_FORMAT_S8_Z24] = NULL;
+      table[MESA_FORMAT_Z16] = NULL;
+      table[MESA_FORMAT_X8_Z24] = NULL;
+      table[MESA_FORMAT_Z24_X8] = NULL;
+      table[MESA_FORMAT_Z32] = NULL;
+      table[MESA_FORMAT_S8] = NULL;
+
+      /* sRGB */
+      table[MESA_FORMAT_SRGB8] = pack_ubyte_SRGB8;
+      table[MESA_FORMAT_SRGBA8] = pack_ubyte_SRGBA8;
+      table[MESA_FORMAT_SARGB8] = pack_ubyte_SARGB8;
+      table[MESA_FORMAT_SL8] = pack_ubyte_SL8;
+      table[MESA_FORMAT_SLA8] = pack_ubyte_SLA8;
+
+      /* n/a */
+      table[MESA_FORMAT_SRGB_DXT1] = NULL; /* pack_ubyte_SRGB_DXT1; */
+      table[MESA_FORMAT_SRGBA_DXT1] = NULL; /* pack_ubyte_SRGBA_DXT1; */
+      table[MESA_FORMAT_SRGBA_DXT3] = NULL; /* pack_ubyte_SRGBA_DXT3; */
+      table[MESA_FORMAT_SRGBA_DXT5] = NULL; /* pack_ubyte_SRGBA_DXT5; */
+
+      table[MESA_FORMAT_RGB_FXT1] = NULL; /* pack_ubyte_RGB_FXT1; */
+      table[MESA_FORMAT_RGBA_FXT1] = NULL; /* pack_ubyte_RGBA_FXT1; */
+      table[MESA_FORMAT_RGB_DXT1] = NULL; /* pack_ubyte_RGB_DXT1; */
+      table[MESA_FORMAT_RGBA_DXT1] = NULL; /* pack_ubyte_RGBA_DXT1; */
+      table[MESA_FORMAT_RGBA_DXT3] = NULL; /* pack_ubyte_RGBA_DXT3; */
+      table[MESA_FORMAT_RGBA_DXT5] = NULL; /* pack_ubyte_RGBA_DXT5; */
+
+      table[MESA_FORMAT_RGBA_FLOAT32] = pack_ubyte_RGBA_FLOAT32;
+      table[MESA_FORMAT_RGBA_FLOAT16] = pack_ubyte_RGBA_FLOAT16;
+      table[MESA_FORMAT_RGB_FLOAT32] = pack_ubyte_RGB_FLOAT32;
+      table[MESA_FORMAT_RGB_FLOAT16] = pack_ubyte_RGB_FLOAT16;
+      table[MESA_FORMAT_ALPHA_FLOAT32] = pack_ubyte_ALPHA_FLOAT32;
+      table[MESA_FORMAT_ALPHA_FLOAT16] = pack_ubyte_ALPHA_FLOAT16;
+      table[MESA_FORMAT_LUMINANCE_FLOAT32] = pack_ubyte_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_LUMINANCE_FLOAT16] = pack_ubyte_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32] = pack_ubyte_LUMINANCE_ALPHA_FLOAT32;
+      table[MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16] = pack_ubyte_LUMINANCE_ALPHA_FLOAT16;
+      table[MESA_FORMAT_INTENSITY_FLOAT32] = pack_ubyte_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_INTENSITY_FLOAT16] = pack_ubyte_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_R_FLOAT32] = pack_ubyte_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_R_FLOAT16] = pack_ubyte_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_RG_FLOAT32] = pack_ubyte_RG_FLOAT32;
+      table[MESA_FORMAT_RG_FLOAT16] = pack_ubyte_RG_FLOAT16;
+
+      /* n/a */
+      table[MESA_FORMAT_RGBA_INT8] = NULL; /* pack_ubyte_RGBA_INT8 */
+      table[MESA_FORMAT_RGBA_INT16] = NULL; /* pack_ubyte_RGBA_INT16 */
+      table[MESA_FORMAT_RGBA_INT32] = NULL; /* pack_ubyte_RGBA_INT32 */
+      table[MESA_FORMAT_RGBA_UINT8] = NULL; /* pack_ubyte_RGBA_UINT8 */
+      table[MESA_FORMAT_RGBA_UINT16] = NULL; /* pack_ubyte_RGBA_UINT16 */
+      table[MESA_FORMAT_RGBA_UINT32] = NULL; /* pack_ubyte_RGBA_UINT32 */
+
+      table[MESA_FORMAT_DUDV8] = pack_ubyte_DUDV8;
+
+      table[MESA_FORMAT_RGBA_16] = pack_ubyte_RGBA_16;
+
+      /* n/a */
+      table[MESA_FORMAT_SIGNED_R8] = NULL;
+      table[MESA_FORMAT_SIGNED_RG88_REV] = NULL;
+      table[MESA_FORMAT_SIGNED_RGBX8888] = NULL;
+      table[MESA_FORMAT_SIGNED_RGBA8888] = NULL;
+      table[MESA_FORMAT_SIGNED_RGBA8888_REV] = NULL;
+      table[MESA_FORMAT_SIGNED_R16] = NULL;
+      table[MESA_FORMAT_SIGNED_GR1616] = NULL;
+      table[MESA_FORMAT_SIGNED_RGB_16] = NULL;
+      table[MESA_FORMAT_SIGNED_RGBA_16] = NULL;
+      table[MESA_FORMAT_SIGNED_A8] = NULL;
+      table[MESA_FORMAT_SIGNED_L8] = NULL;
+      table[MESA_FORMAT_SIGNED_AL88] = NULL;
+      table[MESA_FORMAT_SIGNED_I8] = NULL;
+      table[MESA_FORMAT_SIGNED_A16] = NULL;
+      table[MESA_FORMAT_SIGNED_L16] = NULL;
+      table[MESA_FORMAT_SIGNED_AL1616] = NULL;
+      table[MESA_FORMAT_SIGNED_I16] = NULL;
+
+
+      table[MESA_FORMAT_RGBA_16] = pack_ubyte_RGBA_16;
+
+      table[MESA_FORMAT_RGB9_E5_FLOAT] = pack_ubyte_RGB9_E5_FLOAT;
+      table[MESA_FORMAT_R11_G11_B10_FLOAT] = pack_ubyte_R11_G11_B10_FLOAT;
+
+      initialized = GL_TRUE;
+   }
+
+   return table[format];
+}
+
+
+
+/**
+ * Return a function that can pack a GLfloat rgba[4] color.
+ */
+gl_pack_float_rgba_func
+_mesa_get_pack_float_rgba_function(gl_format format)
+{
+   static gl_pack_float_rgba_func table[MESA_FORMAT_COUNT];
+   static GLboolean initialized = GL_FALSE;
+
+   if (!initialized) {
+      memset(table, 0, sizeof(table));
+
+      table[MESA_FORMAT_NONE] = NULL;
+
+      table[MESA_FORMAT_RGBA8888] = pack_float_RGBA8888;
+      table[MESA_FORMAT_RGBA8888_REV] = pack_float_RGBA8888_REV;
+      table[MESA_FORMAT_ARGB8888] = pack_float_ARGB8888;
+      table[MESA_FORMAT_ARGB8888_REV] = pack_float_ARGB8888_REV;
+      table[MESA_FORMAT_RGBX8888] = pack_float_RGBA8888; /* reused */
+      table[MESA_FORMAT_RGBX8888_REV] = pack_float_RGBA8888_REV; /* reused */
+      table[MESA_FORMAT_XRGB8888] = pack_float_XRGB8888;
+      table[MESA_FORMAT_XRGB8888_REV] = pack_float_XRGB8888_REV;
+      table[MESA_FORMAT_RGB888] = pack_float_RGB888;
+      table[MESA_FORMAT_BGR888] = pack_float_BGR888;
+      table[MESA_FORMAT_RGB565] = pack_float_RGB565;
+      table[MESA_FORMAT_RGB565_REV] = pack_float_RGB565_REV;
+      table[MESA_FORMAT_ARGB4444] = pack_float_ARGB4444;
+      table[MESA_FORMAT_ARGB4444_REV] = pack_float_ARGB4444_REV;
+      table[MESA_FORMAT_RGBA5551] = pack_float_RGBA5551;
+      table[MESA_FORMAT_ARGB1555] = pack_float_ARGB1555;
+      table[MESA_FORMAT_ARGB1555_REV] = pack_float_ARGB1555_REV;
+
+      table[MESA_FORMAT_AL44] = pack_float_AL44;
+      table[MESA_FORMAT_AL88] = pack_float_AL88;
+      table[MESA_FORMAT_AL88_REV] = pack_float_AL88_REV;
+      table[MESA_FORMAT_AL1616] = pack_float_AL1616;
+      table[MESA_FORMAT_AL1616_REV] = pack_float_AL1616_REV;
+      table[MESA_FORMAT_RGB332] = pack_float_RGB332;
+      table[MESA_FORMAT_A8] = pack_float_A8;
+      table[MESA_FORMAT_A16] = pack_float_A16;
+      table[MESA_FORMAT_L8] = pack_float_L8;
+      table[MESA_FORMAT_L16] = pack_float_L16;
+      table[MESA_FORMAT_I8] = pack_float_L8; /* reuse pack_float_L8 */
+      table[MESA_FORMAT_I16] = pack_float_L16; /* reuse pack_float_L16 */
+      table[MESA_FORMAT_YCBCR] = pack_float_YCBCR;
+      table[MESA_FORMAT_YCBCR_REV] = pack_float_YCBCR_REV;
+      table[MESA_FORMAT_R8] = pack_float_R8;
+      table[MESA_FORMAT_GR88] = pack_float_GR88;
+      table[MESA_FORMAT_RG88] = pack_float_RG88;
+      table[MESA_FORMAT_R16] = pack_float_R16;
+      table[MESA_FORMAT_RG1616] = pack_float_RG1616;
+      table[MESA_FORMAT_RG1616_REV] = pack_float_RG1616_REV;
+      table[MESA_FORMAT_ARGB2101010] = pack_float_ARGB2101010;
+
+      /* should never convert RGBA to these formats */
+      table[MESA_FORMAT_Z24_S8] = NULL;
+      table[MESA_FORMAT_S8_Z24] = NULL;
+      table[MESA_FORMAT_Z16] = NULL;
+      table[MESA_FORMAT_X8_Z24] = NULL;
+      table[MESA_FORMAT_Z24_X8] = NULL;
+      table[MESA_FORMAT_Z32] = NULL;
+      table[MESA_FORMAT_S8] = NULL;
+
+      table[MESA_FORMAT_SRGB8] = pack_float_SRGB8;
+      table[MESA_FORMAT_SRGBA8] = pack_float_SRGBA8;
+      table[MESA_FORMAT_SARGB8] = pack_float_SARGB8;
+      table[MESA_FORMAT_SL8] = pack_float_SL8;
+      table[MESA_FORMAT_SLA8] = pack_float_SLA8;
+
+      /* n/a */
+      table[MESA_FORMAT_SRGB_DXT1] = NULL;
+      table[MESA_FORMAT_SRGBA_DXT1] = NULL;
+      table[MESA_FORMAT_SRGBA_DXT3] = NULL;
+      table[MESA_FORMAT_SRGBA_DXT5] = NULL;
+
+      table[MESA_FORMAT_RGB_FXT1] = NULL;
+      table[MESA_FORMAT_RGBA_FXT1] = NULL;
+      table[MESA_FORMAT_RGB_DXT1] = NULL;
+      table[MESA_FORMAT_RGBA_DXT1] = NULL;
+      table[MESA_FORMAT_RGBA_DXT3] = NULL;
+      table[MESA_FORMAT_RGBA_DXT5] = NULL;
+
+      table[MESA_FORMAT_RGBA_FLOAT32] = pack_float_RGBA_FLOAT32;
+      table[MESA_FORMAT_RGBA_FLOAT16] = pack_float_RGBA_FLOAT16;
+      table[MESA_FORMAT_RGB_FLOAT32] = pack_float_RGB_FLOAT32;
+      table[MESA_FORMAT_RGB_FLOAT16] = pack_float_RGB_FLOAT16;
+      table[MESA_FORMAT_ALPHA_FLOAT32] = pack_float_ALPHA_FLOAT32;
+      table[MESA_FORMAT_ALPHA_FLOAT16] = pack_float_ALPHA_FLOAT16;
+      table[MESA_FORMAT_LUMINANCE_FLOAT32] = pack_float_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_LUMINANCE_FLOAT16] = pack_float_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32] = pack_float_LUMINANCE_ALPHA_FLOAT32;
+      table[MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16] = pack_float_LUMINANCE_ALPHA_FLOAT16;
+
+      table[MESA_FORMAT_INTENSITY_FLOAT32] = pack_float_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_INTENSITY_FLOAT16] = pack_float_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_R_FLOAT32] = pack_float_LUMINANCE_FLOAT32;
+      table[MESA_FORMAT_R_FLOAT16] = pack_float_LUMINANCE_FLOAT16;
+      table[MESA_FORMAT_RG_FLOAT32] = pack_float_RG_FLOAT32;
+      table[MESA_FORMAT_RG_FLOAT16] = pack_float_RG_FLOAT16;
+
+      /* n/a */
+      table[MESA_FORMAT_RGBA_INT8] = NULL;
+      table[MESA_FORMAT_RGBA_INT16] = NULL;
+      table[MESA_FORMAT_RGBA_INT32] = NULL;
+      table[MESA_FORMAT_RGBA_UINT8] = NULL;
+      table[MESA_FORMAT_RGBA_UINT16] = NULL;
+      table[MESA_FORMAT_RGBA_UINT32] = NULL;
+
+      table[MESA_FORMAT_DUDV8] = pack_float_DUDV8;
+
+      table[MESA_FORMAT_RGBA_16] = pack_float_RGBA_16;
+
+      table[MESA_FORMAT_SIGNED_R8] = pack_float_SIGNED_R8;
+      table[MESA_FORMAT_SIGNED_RG88_REV] = pack_float_SIGNED_RG88_REV;
+      table[MESA_FORMAT_SIGNED_RGBX8888] = pack_float_SIGNED_RGBX8888;
+      table[MESA_FORMAT_SIGNED_RGBA8888] = pack_float_SIGNED_RGBA8888;
+      table[MESA_FORMAT_SIGNED_RGBA8888_REV] = pack_float_SIGNED_RGBA8888_REV;
+      table[MESA_FORMAT_SIGNED_R16] = pack_float_SIGNED_R16;
+      table[MESA_FORMAT_SIGNED_GR1616] = pack_float_SIGNED_GR1616;
+      table[MESA_FORMAT_SIGNED_RGB_16] = pack_float_SIGNED_RGB_16;
+      table[MESA_FORMAT_SIGNED_RGBA_16] = pack_float_SIGNED_RGBA_16;
+      table[MESA_FORMAT_SIGNED_A8] = pack_float_SIGNED_A8;
+      table[MESA_FORMAT_SIGNED_L8] = pack_float_SIGNED_L8;
+      table[MESA_FORMAT_SIGNED_AL88] = pack_float_SIGNED_AL88;
+      table[MESA_FORMAT_SIGNED_I8] = pack_float_SIGNED_L8; /* reused */
+      table[MESA_FORMAT_SIGNED_A16] = pack_float_SIGNED_A16;
+      table[MESA_FORMAT_SIGNED_L16] = pack_float_SIGNED_L16;
+      table[MESA_FORMAT_SIGNED_AL1616] = pack_float_SIGNED_AL1616;
+      table[MESA_FORMAT_SIGNED_I16] = pack_float_SIGNED_L16; /* reused */
+
+      table[MESA_FORMAT_RGB9_E5_FLOAT] = pack_float_RGB9_E5_FLOAT;
+      table[MESA_FORMAT_R11_G11_B10_FLOAT] = pack_float_R11_G11_B10_FLOAT;
+
+      initialized = GL_TRUE;
+   }
+
+   return table[format];
+}
+
+
+
+static pack_float_rgba_row_func
+get_pack_float_rgba_row_function(gl_format format)
+{
+   static pack_float_rgba_row_func table[MESA_FORMAT_COUNT];
+   static GLboolean initialized = GL_FALSE;
+
+   if (!initialized) {
+      /* We don't need a special row packing function for each format.
+       * There's a generic fallback which uses a per-pixel packing function.
+       */
+      memset(table, 0, sizeof(table));
+
+      table[MESA_FORMAT_RGBA8888] = pack_row_float_RGBA8888;
+      table[MESA_FORMAT_RGBA8888_REV] = pack_row_float_RGBA8888_REV;
+      table[MESA_FORMAT_ARGB8888] = pack_row_float_ARGB8888;
+      table[MESA_FORMAT_ARGB8888_REV] = pack_row_float_ARGB8888_REV;
+      table[MESA_FORMAT_RGBX8888] = pack_row_float_RGBA8888; /* reused */
+      table[MESA_FORMAT_RGBX8888_REV] = pack_row_float_RGBA8888_REV; /* reused */
+      table[MESA_FORMAT_XRGB8888] = pack_row_float_XRGB8888;
+      table[MESA_FORMAT_XRGB8888_REV] = pack_row_float_XRGB8888_REV;
+      table[MESA_FORMAT_RGB888] = pack_row_float_RGB888;
+      table[MESA_FORMAT_BGR888] = pack_row_float_BGR888;
+      table[MESA_FORMAT_RGB565] = pack_row_float_RGB565;
+      table[MESA_FORMAT_RGB565_REV] = pack_row_float_RGB565_REV;
+
+      initialized = GL_TRUE;
+   }
+
+   return table[format];
+}
+
+
+
+static pack_ubyte_rgba_row_func
+get_pack_ubyte_rgba_row_function(gl_format format)
+{
+   static pack_ubyte_rgba_row_func table[MESA_FORMAT_COUNT];
+   static GLboolean initialized = GL_FALSE;
+
+   if (!initialized) {
+      /* We don't need a special row packing function for each format.
+       * There's a generic fallback which uses a per-pixel packing function.
+       */
+      memset(table, 0, sizeof(table));
+
+      table[MESA_FORMAT_RGBA8888] = pack_row_ubyte_RGBA8888;
+      table[MESA_FORMAT_RGBA8888_REV] = pack_row_ubyte_RGBA8888_REV;
+      table[MESA_FORMAT_ARGB8888] = pack_row_ubyte_ARGB8888;
+      table[MESA_FORMAT_ARGB8888_REV] = pack_row_ubyte_ARGB8888_REV;
+      table[MESA_FORMAT_RGBX8888] = pack_row_ubyte_RGBA8888; /* reused */
+      table[MESA_FORMAT_RGBX8888_REV] = pack_row_ubyte_RGBA8888_REV; /* reused */
+      table[MESA_FORMAT_XRGB8888] = pack_row_ubyte_XRGB8888;
+      table[MESA_FORMAT_XRGB8888_REV] = pack_row_ubyte_XRGB8888_REV;
+      table[MESA_FORMAT_RGB888] = pack_row_ubyte_RGB888;
+      table[MESA_FORMAT_BGR888] = pack_row_ubyte_BGR888;
+      table[MESA_FORMAT_RGB565] = pack_row_ubyte_RGB565;
+      table[MESA_FORMAT_RGB565_REV] = pack_row_ubyte_RGB565_REV;
+
+      initialized = GL_TRUE;
+   }
+
+   return table[format];
+}
+
+
+
+/**
+ * Pack a row of GLfloat rgba[4] values to the destination.
+ */
+void
+_mesa_pack_float_rgba_row(gl_format format, GLuint n,
+                          const GLfloat src[][4], void *dst)
+{
+   pack_float_rgba_row_func packrow = get_pack_float_rgba_row_function(format);
+   if (packrow) {
+      /* use "fast" function */
+      packrow(n, src, dst);
+   }
+   else {
+      /* slower fallback */
+      gl_pack_float_rgba_func pack = _mesa_get_pack_float_rgba_function(format);
+      GLuint dstStride = _mesa_get_format_bytes(format);
+      GLubyte *dstPtr = (GLubyte *) dst;
+      GLuint i;
+
+      assert(pack);
+      if (!pack)
+         return;
+
+      for (i = 0; i < n; i++) {
+         pack(src[i], dstPtr);
+         dstPtr += dstStride;
+      }
+   }
+}
+
+
+/**
+ * Pack a row of GLubyte rgba[4] values to the destination.
+ */
+void
+_mesa_pack_ubyte_rgba_row(gl_format format, GLuint n,
+                          const GLubyte src[][4], void *dst)
+{
+   pack_ubyte_rgba_row_func packrow = get_pack_ubyte_rgba_row_function(format);
+   if (packrow) {
+      /* use "fast" function */
+      packrow(n, src, dst);
+   }
+   else {
+      /* slower fallback */
+      gl_pack_ubyte_rgba_func pack = _mesa_get_pack_ubyte_rgba_function(format);
+      const GLuint stride = _mesa_get_format_bytes(format);
+      GLubyte *d = ((GLubyte *) dst);
+      GLuint i;
+
+      assert(pack);
+      if (!pack)
+         return;
+
+      for (i = 0; i < n; i++) {
+         pack(src[i], d);
+         d += stride;
+      }
+   }
+}
+
+
+/**
+ ** Pack float Z pixels
+ **/
+
+static void
+pack_float_z_Z24_S8(const GLfloat *src, void *dst)
+{
+   /* don't disturb the stencil values */
+   GLuint *d = ((GLuint *) dst);
+   const GLdouble scale = (GLdouble) 0xffffff;
+   GLuint s = *d & 0xff;
+   GLuint z = (GLuint) (*src * scale);
+   assert(z <= 0xffffff);
+   *d = (z << 8) | s;
+}
+
+static void
+pack_float_z_S8_Z24(const GLfloat *src, void *dst)
+{
+   /* don't disturb the stencil values */
+   GLuint *d = ((GLuint *) dst);
+   const GLdouble scale = (GLdouble) 0xffffff;
+   GLuint s = *d & 0xff000000;
+   GLuint z = (GLuint) (*src * scale);
+   assert(z <= 0xffffff);
+   *d = s | z;
+}
+
+static void
+pack_float_z_Z16(const GLfloat *src, void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   const GLfloat scale = (GLfloat) 0xffff;
+   *d = (GLushort) (*src * scale);
+}
+
+static void
+pack_float_z_Z32(const GLfloat *src, void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   const GLdouble scale = (GLdouble) 0xffffffff;
+   *d = (GLuint) (*src * scale);
+}
+
+static void
+pack_float_z_Z32_FLOAT(const GLfloat *src, void *dst)
+{
+   GLfloat *d = (GLfloat *) dst;
+   *d = *src;
+}
+
+gl_pack_float_z_func
+_mesa_get_pack_float_z_func(gl_format format)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+   case MESA_FORMAT_Z24_X8:
+      return pack_float_z_Z24_S8;
+   case MESA_FORMAT_S8_Z24:
+   case MESA_FORMAT_X8_Z24:
+      return pack_float_z_S8_Z24;
+   case MESA_FORMAT_Z16:
+      return pack_float_z_Z16;
+   case MESA_FORMAT_Z32:
+      return pack_float_z_Z32;
+   case MESA_FORMAT_Z32_FLOAT:
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      return pack_float_z_Z32_FLOAT;
+   default:
+      _mesa_problem(NULL,
+                    "unexpected format in _mesa_get_pack_float_z_func()");
+      return NULL;
+   }
+}
+
+
+
+/**
+ ** Pack uint Z pixels.  The incoming src value is always in
+ ** the range [0, 2^32-1].
+ **/
+
+static void
+pack_uint_z_Z24_S8(const GLuint *src, void *dst)
+{
+   /* don't disturb the stencil values */
+   GLuint *d = ((GLuint *) dst);
+   GLuint s = *d & 0xff;
+   GLuint z = *src & 0xffffff00;
+   *d = z | s;
+}
+
+static void
+pack_uint_z_S8_Z24(const GLuint *src, void *dst)
+{
+   /* don't disturb the stencil values */
+   GLuint *d = ((GLuint *) dst);
+   GLuint s = *d & 0xff000000;
+   GLuint z = *src >> 8;
+   *d = s | z;
+}
+
+static void
+pack_uint_z_Z16(const GLuint *src, void *dst)
+{
+   GLushort *d = ((GLushort *) dst);
+   *d = *src >> 16;
+}
+
+static void
+pack_uint_z_Z32(const GLuint *src, void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   *d = *src;
+}
+
+static void
+pack_uint_z_Z32_FLOAT(const GLuint *src, void *dst)
+{
+   GLuint *d = ((GLuint *) dst);
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
+   *d = *src * scale;
+   assert(*d >= 0.0f);
+   assert(*d <= 1.0f);
+}
+
+static void
+pack_uint_z_Z32_FLOAT_X24S8(const GLuint *src, void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
+   *d = *src * scale;
+   assert(*d >= 0.0f);
+   assert(*d <= 1.0f);
+}
+
+gl_pack_uint_z_func
+_mesa_get_pack_uint_z_func(gl_format format)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+   case MESA_FORMAT_Z24_X8:
+      return pack_uint_z_Z24_S8;
+   case MESA_FORMAT_S8_Z24:
+   case MESA_FORMAT_X8_Z24:
+      return pack_uint_z_S8_Z24;
+   case MESA_FORMAT_Z16:
+      return pack_uint_z_Z16;
+   case MESA_FORMAT_Z32:
+      return pack_uint_z_Z32;
+   case MESA_FORMAT_Z32_FLOAT:
+      return pack_uint_z_Z32_FLOAT;
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      return pack_uint_z_Z32_FLOAT_X24S8;
+   default:
+      _mesa_problem(NULL, "unexpected format in _mesa_get_pack_uint_z_func()");
+      return NULL;
+   }
+}
+
+
+/**
+ ** Pack ubyte stencil pixels
+ **/
+
+static void
+pack_ubyte_stencil_Z24_S8(const GLubyte *src, void *dst)
+{
+   /* don't disturb the Z values */
+   GLuint *d = ((GLuint *) dst);
+   GLuint s = *src;
+   GLuint z = *d & 0xffffff00;
+   *d = z | s;
+}
+
+static void
+pack_ubyte_stencil_S8_Z24(const GLubyte *src, void *dst)
+{
+   /* don't disturb the Z values */
+   GLuint *d = ((GLuint *) dst);
+   GLuint s = *src << 24;
+   GLuint z = *d & 0xffffff;
+   *d = s | z;
+}
+
+static void
+pack_ubyte_stencil_S8(const GLubyte *src, void *dst)
+{
+   GLubyte *d = (GLubyte *) dst;
+   *d = *src;
+}
+
+static void
+pack_ubyte_stencil_Z32_FLOAT_X24S8(const GLubyte *src, void *dst)
+{
+   GLfloat *d = ((GLfloat *) dst);
+   d[1] = *src;
+}
+
+
+gl_pack_ubyte_stencil_func
+_mesa_get_pack_ubyte_stencil_func(gl_format format)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+      return pack_ubyte_stencil_Z24_S8;
+   case MESA_FORMAT_S8_Z24:
+      return pack_ubyte_stencil_S8_Z24;
+   case MESA_FORMAT_S8:
+      return pack_ubyte_stencil_S8;
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      return pack_ubyte_stencil_Z32_FLOAT_X24S8;
+   default:
+      _mesa_problem(NULL,
+                    "unexpected format in _mesa_pack_ubyte_stencil_func()");
+      return NULL;
+   }
+}
+
+
+
+void
+_mesa_pack_float_z_row(gl_format format, GLuint n,
+                       const GLfloat *src, void *dst)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+   case MESA_FORMAT_Z24_X8:
+      {
+         /* don't disturb the stencil values */
+         GLuint *d = ((GLuint *) dst);
+         const GLdouble scale = (GLdouble) 0xffffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = d[i] & 0xff;
+            GLuint z = (GLuint) (src[i] * scale);
+            assert(z <= 0xffffff);
+            d[i] = (z << 8) | s;
+         }
+      }
+      break;
+   case MESA_FORMAT_S8_Z24:
+   case MESA_FORMAT_X8_Z24:
+      {
+         /* don't disturb the stencil values */
+         GLuint *d = ((GLuint *) dst);
+         const GLdouble scale = (GLdouble) 0xffffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = d[i] & 0xff000000;
+            GLuint z = (GLuint) (src[i] * scale);
+            assert(z <= 0xffffff);
+            d[i] = s | z;
+         }
+      }
+      break;
+   case MESA_FORMAT_Z16:
+      {
+         GLushort *d = ((GLushort *) dst);
+         const GLfloat scale = (GLfloat) 0xffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i] = (GLushort) (src[i] * scale);
+         }
+      }
+      break;
+   case MESA_FORMAT_Z32:
+      {
+         GLuint *d = ((GLuint *) dst);
+         const GLdouble scale = (GLdouble) 0xffffffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i] = (GLuint) (src[i] * scale);
+         }
+      }
+      break;
+   case MESA_FORMAT_Z32_FLOAT:
+      memcpy(dst, src, n * sizeof(GLfloat));
+      break;
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      {
+         GLfloat *d = ((GLfloat *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i * 2] = src[i];
+         }
+      }
+      break;
+   default:
+      _mesa_problem(NULL, "unexpected format in _mesa_pack_float_z_row()");
+   }
+}
+
+
+/**
+ * The incoming Z values are always in the range [0, 0xffffffff].
+ */
+void
+_mesa_pack_uint_z_row(gl_format format, GLuint n,
+                      const GLuint *src, void *dst)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+   case MESA_FORMAT_Z24_X8:
+      {
+         /* don't disturb the stencil values */
+         GLuint *d = ((GLuint *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = d[i] & 0xff;
+            GLuint z = src[i] & 0xffffff00;
+            d[i] = z | s;
+         }
+      }
+      break;
+   case MESA_FORMAT_S8_Z24:
+   case MESA_FORMAT_X8_Z24:
+      {
+         /* don't disturb the stencil values */
+         GLuint *d = ((GLuint *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = d[i] & 0xff000000;
+            GLuint z = src[i] >> 8;
+            d[i] = s | z;
+         }
+      }
+      break;
+   case MESA_FORMAT_Z16:
+      {
+         GLushort *d = ((GLushort *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i] = src[i] >> 16;
+         }
+      }
+      break;
+   case MESA_FORMAT_Z32:
+      memcpy(dst, src, n * sizeof(GLfloat));
+      break;
+   case MESA_FORMAT_Z32_FLOAT:
+      {
+         GLuint *d = ((GLuint *) dst);
+         const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i] = src[i] * scale;
+            assert(d[i] >= 0.0f);
+            assert(d[i] <= 1.0f);
+         }
+      }
+      break;
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      {
+         GLfloat *d = ((GLfloat *) dst);
+         const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i * 2] = src[i] * scale;
+            assert(d[i * 2] >= 0.0f);
+            assert(d[i * 2] <= 1.0f);
+         }
+      }
+      break;
+   default:
+      _mesa_problem(NULL, "unexpected format in _mesa_pack_uint_z_row()");
+   }
+}
+
+
+void
+_mesa_pack_ubyte_stencil_row(gl_format format, GLuint n,
+                             const GLubyte *src, void *dst)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+      {
+         /* don't disturb the Z values */
+         GLuint *d = ((GLuint *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = src[i];
+            GLuint z = d[i] & 0xffffff00;
+            d[i] = z | s;
+         }
+      }
+      break;
+   case MESA_FORMAT_S8_Z24:
+      {
+         /* don't disturb the Z values */
+         GLuint *d = ((GLuint *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = src[i] << 24;
+            GLuint z = d[i] & 0xffffff;
+            d[i] = s | z;
+         }
+      }
+      break;
+   case MESA_FORMAT_S8:
+      memcpy(dst, src, n * sizeof(GLubyte));
+      break;
+   case MESA_FORMAT_Z32_FLOAT_X24S8:
+      {
+         GLfloat *d = ((GLfloat *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            d[i * 2 + 1] = src[i];
+         }
+      }
+      break;
+   default:
+      _mesa_problem(NULL, "unexpected format in _mesa_pack_ubyte_stencil_row()");
+   }
+}
+
+
+/**
+ * Incoming Z/stencil values are always in uint_24_8 format.
+ */
+void
+_mesa_pack_uint_24_8_depth_stencil_row(gl_format format, GLuint n,
+                                       const GLuint *src, void *dst)
+{
+   switch (format) {
+   case MESA_FORMAT_Z24_S8:
+      memcpy(dst, src, n * sizeof(GLuint));
+      break;
+   case MESA_FORMAT_S8_Z24:
+      {
+         GLuint *d = ((GLuint *) dst);
+         GLuint i;
+         for (i = 0; i < n; i++) {
+            GLuint s = src[i] << 24;
+            GLuint z = src[i] >> 8;
+            d[i] = s | z;
+         }
+      }
+      break;
+   default:
+      _mesa_problem(NULL, "bad format %s in _mesa_pack_ubyte_s_row",
+                    _mesa_get_format_name(format));
+      return;
+   }
+}
diff --git a/mesalib/src/mesa/main/format_pack.h b/mesalib/src/mesa/main/format_pack.h
new file mode 100644
index 000000000..7df135632
--- /dev/null
+++ b/mesalib/src/mesa/main/format_pack.h
@@ -0,0 +1,98 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (c) 2011 VMware, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#ifndef FORMAT_PACK_H
+#define FORMAT_PACK_H
+
+
+#include "formats.h"
+
+
+/** Pack a GLubyte rgba[4] color to dest address */
+typedef void (*gl_pack_ubyte_rgba_func)(const GLubyte src[4], void *dst);
+
+/** Pack a GLfloat rgba[4] color to dest address */
+typedef void (*gl_pack_float_rgba_func)(const GLfloat src[4], void *dst);
+
+/** Pack a GLfloat Z value to dest address */
+typedef void (*gl_pack_float_z_func)(const GLfloat *src, void *dst);
+
+/** Pack a GLuint Z value to dest address */
+typedef void (*gl_pack_uint_z_func)(const GLuint *src, void *dst);
+
+/** Pack a GLubyte stencil value to dest address */
+typedef void (*gl_pack_ubyte_stencil_func)(const GLubyte *src, void *dst);
+
+
+
+
+extern gl_pack_ubyte_rgba_func
+_mesa_get_pack_ubyte_rgba_function(gl_format format);
+
+
+extern gl_pack_float_rgba_func
+_mesa_get_pack_float_rgba_function(gl_format format);
+
+
+extern gl_pack_float_z_func
+_mesa_get_pack_float_z_func(gl_format format);
+
+
+extern gl_pack_uint_z_func
+_mesa_get_pack_uint_z_func(gl_format format);
+
+
+extern gl_pack_ubyte_stencil_func
+_mesa_get_pack_ubyte_stencil_func(gl_format format);
+
+
+
+extern void
+_mesa_pack_float_rgba_row(gl_format format, GLuint n,
+                          const GLfloat src[][4], void *dst);
+
+extern void
+_mesa_pack_ubyte_rgba_row(gl_format format, GLuint n,
+                          const GLubyte src[][4], void *dst);
+
+
+
+extern void
+_mesa_pack_float_z_row(gl_format format, GLuint n,
+                       const GLfloat *src, void *dst);
+
+extern void
+_mesa_pack_uint_z_row(gl_format format, GLuint n,
+                      const GLuint *src, void *dst);
+
+extern void
+_mesa_pack_ubyte_stencil_row(gl_format format, GLuint n,
+                             const GLubyte *src, void *dst);
+
+extern void
+_mesa_pack_uint_24_8_depth_stencil_row(gl_format format, GLuint n,
+                                       const GLuint *src, void *dst);
+
+
+#endif
diff --git a/mesalib/src/mesa/main/format_unpack.c b/mesalib/src/mesa/main/format_unpack.c
index f821c2be4..4f23f3dec 100644
--- a/mesalib/src/mesa/main/format_unpack.c
+++ b/mesalib/src/mesa/main/format_unpack.c
@@ -587,7 +587,7 @@ unpack_Z24_S8(const void *src, GLfloat dst[][4], GLuint n)
 {
    /* only return Z, not stencil data */
    const GLuint *s = ((const GLuint *) src);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    GLuint i;
    for (i = 0; i < n; i++) {
       dst[i][0] =
@@ -604,7 +604,7 @@ unpack_S8_Z24(const void *src, GLfloat dst[][4], GLuint n)
 {
    /* only return Z, not stencil data */
    const GLuint *s = ((const GLuint *) src);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    GLuint i;
    for (i = 0; i < n; i++) {
       dst[i][0] =
@@ -1761,7 +1761,7 @@ unpack_float_z_Z24_X8(GLuint n, const void *src, GLfloat *dst)
 {
    /* only return Z, not stencil data */
    const GLuint *s = ((const GLuint *) src);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    GLuint i;
    for (i = 0; i < n; i++) {
       dst[i] = (s[i] >> 8) * scale;
@@ -1775,7 +1775,7 @@ unpack_float_z_X8_Z24(GLuint n, const void *src, GLfloat *dst)
 {
    /* only return Z, not stencil data */
    const GLuint *s = ((const GLuint *) src);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    GLuint i;
    for (i = 0; i < n; i++) {
       dst[i] = (s[i] & 0x00ffffff) * scale;
diff --git a/mesalib/src/mesa/main/formats.c b/mesalib/src/mesa/main/formats.c
index 595da773b..1f83a5368 100644
--- a/mesalib/src/mesa/main/formats.c
+++ b/mesalib/src/mesa/main/formats.c
@@ -1553,6 +1553,8 @@ _mesa_get_format_bytes(gl_format format)
 {
    const struct gl_format_info *info = _mesa_get_format_info(format);
    ASSERT(info->BytesPerBlock);
+   ASSERT(info->BytesPerBlock <= MAX_PIXEL_BYTES ||
+          _mesa_is_format_compressed(format));
    return info->BytesPerBlock;
 }
 
diff --git a/mesalib/src/mesa/main/formats.h b/mesalib/src/mesa/main/formats.h
index 3960f028e..e6b429d5f 100644
--- a/mesalib/src/mesa/main/formats.h
+++ b/mesalib/src/mesa/main/formats.h
@@ -47,6 +47,14 @@ extern "C" {
 
 
 /**
+ * Max number of bytes for any non-compressed pixel format below, or for
+ * intermediate pixel storage in Mesa.  This should never be less than
+ * 16.  Maybe 32 someday?
+ */
+#define MAX_PIXEL_BYTES 16
+
+
+/**
  * Mesa texture/renderbuffer image formats.
  */
 typedef enum
diff --git a/mesalib/src/mesa/main/framebuffer.c b/mesalib/src/mesa/main/framebuffer.c
index 8d0763db3..e30f31a39 100644
--- a/mesalib/src/mesa/main/framebuffer.c
+++ b/mesalib/src/mesa/main/framebuffer.c
@@ -879,6 +879,7 @@ renderbuffer_exists(struct gl_context *ctx,
    case GL_BGRA:
    case GL_ABGR_EXT:
    case GL_RED_INTEGER_EXT:
+   case GL_RG_INTEGER:
    case GL_GREEN_INTEGER_EXT:
    case GL_BLUE_INTEGER_EXT:
    case GL_ALPHA_INTEGER_EXT:
diff --git a/mesalib/src/mesa/main/image.c b/mesalib/src/mesa/main/image.c
index b266e26c6..b37cd7e7d 100644
--- a/mesalib/src/mesa/main/image.c
+++ b/mesalib/src/mesa/main/image.c
@@ -238,6 +238,7 @@ _mesa_components_in_format( GLenum format )
       case GL_DEPTH_STENCIL_EXT:
       case GL_DUDV_ATI:
       case GL_DU8DV8_ATI:
+      case GL_RG_INTEGER:
 	 return 2;
 
       case GL_RGB:
@@ -527,6 +528,7 @@ _mesa_is_legal_format_and_type(const struct gl_context *ctx,
       case GL_GREEN_INTEGER_EXT:
       case GL_BLUE_INTEGER_EXT:
       case GL_ALPHA_INTEGER_EXT:
+      case GL_RG_INTEGER:
          switch (type) {
             case GL_BYTE:
             case GL_UNSIGNED_BYTE:
@@ -751,6 +753,7 @@ _mesa_is_color_format(GLenum format)
       case GL_RGBA_INTEGER_EXT:
       case GL_BGR_INTEGER_EXT:
       case GL_BGRA_INTEGER_EXT:
+      case GL_RG_INTEGER:
       case GL_LUMINANCE_INTEGER_EXT:
       case GL_LUMINANCE_ALPHA_INTEGER_EXT:
       /* sized integer formats */
@@ -954,6 +957,7 @@ _mesa_is_integer_format(GLenum format)
    case GL_BGRA_INTEGER_EXT:
    case GL_LUMINANCE_INTEGER_EXT:
    case GL_LUMINANCE_ALPHA_INTEGER_EXT:
+   case GL_RG_INTEGER:
    /* specific integer formats */
    case GL_RGBA32UI_EXT:
    case GL_RGB32UI_EXT:
@@ -1160,32 +1164,30 @@ _mesa_base_format_has_channel(GLenum base_format, GLenum pname)
 
 
 /**
- * Return the address of a specific pixel in an image (1D, 2D or 3D).
+ * Return the byte offset of a specific pixel in an image (1D, 2D or 3D).
  *
  * Pixel unpacking/packing parameters are observed according to \p packing.
  *
  * \param dimensions either 1, 2 or 3 to indicate dimensionality of image
- * \param image  starting address of image data
- * \param width  the image width
- * \param height  theimage height
- * \param format  the pixel format
- * \param type  the pixel data type
  * \param packing  the pixelstore attributes
+ * \param width  the image width
+ * \param height  the image height
+ * \param format  the pixel format (must be validated beforehand)
+ * \param type  the pixel data type (must be validated beforehand)
  * \param img  which image in the volume (0 for 1D or 2D images)
  * \param row  row of pixel in the image (0 for 1D images)
  * \param column column of pixel in the image
- * 
- * \return address of pixel on success, or NULL on error.
+ *
+ * \return offset of pixel.
  *
  * \sa gl_pixelstore_attrib.
  */
-GLvoid *
-_mesa_image_address( GLuint dimensions,
-                     const struct gl_pixelstore_attrib *packing,
-                     const GLvoid *image,
-                     GLsizei width, GLsizei height,
-                     GLenum format, GLenum type,
-                     GLint img, GLint row, GLint column )
+GLintptr
+_mesa_image_offset( GLuint dimensions,
+                    const struct gl_pixelstore_attrib *packing,
+                    GLsizei width, GLsizei height,
+                    GLenum format, GLenum type,
+                    GLint img, GLint row, GLint column )
 {
    GLint alignment;        /* 1, 2 or 4 */
    GLint pixels_per_row;
@@ -1193,7 +1195,7 @@ _mesa_image_address( GLuint dimensions,
    GLint skiprows;
    GLint skippixels;
    GLint skipimages;       /* for 3-D volume images */
-   GLubyte *pixel_addr;
+   GLintptr offset;
 
    ASSERT(dimensions >= 1 && dimensions <= 3);
 
@@ -1219,30 +1221,20 @@ _mesa_image_address( GLuint dimensions,
 
    if (type == GL_BITMAP) {
       /* BITMAP data */
-      GLint comp_per_pixel;   /* components per pixel */
-      GLint bytes_per_comp;   /* bytes per component */
       GLint bytes_per_row;
       GLint bytes_per_image;
+      /* components per pixel for color or stencil index: */
+      const GLint comp_per_pixel = 1;
 
-      /* Compute bytes per component */
-      bytes_per_comp = _mesa_sizeof_packed_type( type );
-      if (bytes_per_comp < 0) {
-         return NULL;
-      }
-
-      /* Compute number of components per pixel */
-      comp_per_pixel = _mesa_components_in_format( format );
-      if (comp_per_pixel < 0) {
-         return NULL;
-      }
+      /* The pixel type and format should have been error checked earlier */
+      assert(format == GL_COLOR_INDEX || format == GL_STENCIL_INDEX);
 
       bytes_per_row = alignment
                     * CEILING( comp_per_pixel*pixels_per_row, 8*alignment );
 
       bytes_per_image = bytes_per_row * rows_per_image;
 
-      pixel_addr = (GLubyte *) image
-                 + (skipimages + img) * bytes_per_image
+      offset = (skipimages + img) * bytes_per_image
                  + (skiprows + row) * bytes_per_row
                  + (skippixels + column) / 8;
    }
@@ -1275,14 +1267,50 @@ _mesa_image_address( GLuint dimensions,
       }
 
       /* compute final pixel address */
-      pixel_addr = (GLubyte *) image
-                 + (skipimages + img) * bytes_per_image
+      offset = (skipimages + img) * bytes_per_image
                  + topOfImage
                  + (skiprows + row) * bytes_per_row
                  + (skippixels + column) * bytes_per_pixel;
    }
 
-   return (GLvoid *) pixel_addr;
+   return offset;
+}
+
+
+/**
+ * Return the address of a specific pixel in an image (1D, 2D or 3D).
+ *
+ * Pixel unpacking/packing parameters are observed according to \p packing.
+ *
+ * \param dimensions either 1, 2 or 3 to indicate dimensionality of image
+ * \param packing  the pixelstore attributes
+ * \param image  starting address of image data
+ * \param width  the image width
+ * \param height  the image height
+ * \param format  the pixel format (must be validated beforehand)
+ * \param type  the pixel data type (must be validated beforehand)
+ * \param img  which image in the volume (0 for 1D or 2D images)
+ * \param row  row of pixel in the image (0 for 1D images)
+ * \param column column of pixel in the image
+ *
+ * \return address of pixel.
+ *
+ * \sa gl_pixelstore_attrib.
+ */
+GLvoid *
+_mesa_image_address( GLuint dimensions,
+                     const struct gl_pixelstore_attrib *packing,
+                     const GLvoid *image,
+                     GLsizei width, GLsizei height,
+                     GLenum format, GLenum type,
+                     GLint img, GLint row, GLint column )
+{
+   const GLubyte *addr = (const GLubyte *) image;
+
+   addr += _mesa_image_offset(dimensions, packing, width, height,
+                              format, type, img, row, column);
+
+   return (GLvoid *) addr;
 }
 
 
@@ -1520,9 +1548,13 @@ _mesa_convert_colors(GLenum srcType, const GLvoid *src,
                      GLenum dstType, GLvoid *dst,
                      GLuint count, const GLubyte mask[])
 {
-   GLuint tempBuffer[MAX_WIDTH][4];
+   GLuint *tempBuffer;
    const GLboolean useTemp = (src == dst);
 
+   tempBuffer = malloc(count * MAX_PIXEL_BYTES);
+   if (!tempBuffer)
+      return;
+
    ASSERT(srcType != dstType);
 
    switch (srcType) {
@@ -1624,6 +1656,8 @@ _mesa_convert_colors(GLenum srcType, const GLvoid *src,
    default:
       _mesa_problem(NULL, "Invalid datatype in _mesa_convert_colors");
    }
+
+   free(tempBuffer);
 }
 
 
diff --git a/mesalib/src/mesa/main/image.h b/mesalib/src/mesa/main/image.h
index b606545b7..e4961ed3d 100644
--- a/mesalib/src/mesa/main/image.h
+++ b/mesalib/src/mesa/main/image.h
@@ -87,6 +87,13 @@ _mesa_is_compressed_format(struct gl_context *ctx, GLenum format);
 extern GLboolean
 _mesa_base_format_has_channel(GLenum base_format, GLenum pname);
 
+extern GLintptr
+_mesa_image_offset( GLuint dimensions,
+                    const struct gl_pixelstore_attrib *packing,
+                    GLsizei width, GLsizei height,
+                    GLenum format, GLenum type,
+                    GLint img, GLint row, GLint column );
+
 extern GLvoid *
 _mesa_image_address( GLuint dimensions,
                      const struct gl_pixelstore_attrib *packing,
diff --git a/mesalib/src/mesa/main/pack.c b/mesalib/src/mesa/main/pack.c
index 0bd4ff199..2933ff638 100644
--- a/mesalib/src/mesa/main/pack.c
+++ b/mesalib/src/mesa/main/pack.c
@@ -467,6 +467,7 @@ _mesa_pack_rgba_span_int(struct gl_context *ctx, GLuint n, GLuint rgba[][4],
       case GL_GREEN_INTEGER_EXT:
       case GL_BLUE_INTEGER_EXT:
       case GL_ALPHA_INTEGER_EXT:
+      case GL_RG_INTEGER:
       case GL_RGB_INTEGER_EXT:
       case GL_RGBA_INTEGER_EXT:
       case GL_BGR_INTEGER_EXT:
@@ -490,6 +491,7 @@ _mesa_pack_rgba_span_int(struct gl_context *ctx, GLuint n, GLuint rgba[][4],
       case GL_GREEN_INTEGER_EXT:
       case GL_BLUE_INTEGER_EXT:
       case GL_ALPHA_INTEGER_EXT:
+      case GL_RG_INTEGER:
       case GL_RGB_INTEGER_EXT:
       case GL_RGBA_INTEGER_EXT:
       case GL_BGR_INTEGER_EXT:
@@ -3021,6 +3023,7 @@ extract_uint_rgba(GLuint n, GLuint rgba[][4],
           srcFormat == GL_DU8DV8_ATI ||
           srcFormat == GL_DUDV_ATI ||
           srcFormat == GL_RED_INTEGER_EXT ||
+          srcFormat == GL_RG_INTEGER ||
           srcFormat == GL_GREEN_INTEGER_EXT ||
           srcFormat == GL_BLUE_INTEGER_EXT ||
           srcFormat == GL_ALPHA_INTEGER_EXT ||
@@ -3980,6 +3983,7 @@ _mesa_unpack_color_span_uint(struct gl_context *ctx,
           srcFormat == GL_GREEN_INTEGER_EXT ||
           srcFormat == GL_BLUE_INTEGER_EXT ||
           srcFormat == GL_ALPHA_INTEGER_EXT ||
+          srcFormat == GL_RG_INTEGER ||
           srcFormat == GL_RGB_INTEGER_EXT ||
           srcFormat == GL_RGBA_INTEGER_EXT ||
           srcFormat == GL_BGR_INTEGER_EXT ||
diff --git a/mesalib/src/mesa/main/pbo.c b/mesalib/src/mesa/main/pbo.c
index 41ff2ff44..d8fa9191d 100644
--- a/mesalib/src/mesa/main/pbo.c
+++ b/mesalib/src/mesa/main/pbo.c
@@ -68,8 +68,8 @@ _mesa_validate_pbo_access(GLuint dimensions,
                           GLenum format, GLenum type, GLsizei clientMemSize,
                           const GLvoid *ptr)
 {
-   const GLvoid *start, *end, *offset;
-   const GLubyte *sizeAddr; /* buffer size, cast to a pointer */
+   /* unsigned, to detect overflow/wrap-around */
+   uintptr_t start, end, offset, size;
 
    /* If no PBO is bound, 'ptr' is a pointer to client memory containing
       'clientMemSize' bytes.
@@ -78,10 +78,10 @@ _mesa_validate_pbo_access(GLuint dimensions,
     */
    if (!_mesa_is_bufferobj(pack->BufferObj)) {
       offset = 0;
-      sizeAddr = ((const GLubyte *) 0) + clientMemSize;
+      size = clientMemSize;
    } else {
-      offset = ptr;
-      sizeAddr = ((const GLubyte *) 0) + pack->BufferObj->Size;
+      offset = (uintptr_t)ptr;
+      size = pack->BufferObj->Size;
       /* The ARB_pixel_buffer_object spec says:
        *    "INVALID_OPERATION is generated by ColorTable, ColorSubTable,
        *    ConvolutionFilter2D, ConvolutionFilter1D, SeparableFilter2D,
@@ -93,27 +93,30 @@ _mesa_validate_pbo_access(GLuint dimensions,
        *    parameter."
        */
       if (type != GL_BITMAP &&
-          ((GLintptr)offset % _mesa_sizeof_packed_type(type)))
+          (offset % _mesa_sizeof_packed_type(type)))
          return GL_FALSE;
    }
 
-   if (sizeAddr == 0)
+   if (size == 0)
       /* no buffer! */
       return GL_FALSE;
 
    /* get the offset to the first pixel we'll read/write */
-   start = _mesa_image_address(dimensions, pack, offset, width, height,
-                               format, type, 0, 0, 0);
+   start = _mesa_image_offset(dimensions, pack, width, height,
+                              format, type, 0, 0, 0);
 
    /* get the offset to just past the last pixel we'll read/write */
-   end =  _mesa_image_address(dimensions, pack, offset, width, height,
-                              format, type, depth-1, height-1, width);
+   end =  _mesa_image_offset(dimensions, pack, width, height,
+                             format, type, depth-1, height-1, width);
 
-   if ((const GLubyte *) start > sizeAddr) {
+   start += offset;
+   end += offset;
+
+   if (start > size) {
       /* This will catch negative values / wrap-around */
       return GL_FALSE;
    }
-   if ((const GLubyte *) end > sizeAddr) {
+   if (end > size) {
       /* Image read/write goes beyond end of buffer */
       return GL_FALSE;
    }
diff --git a/mesalib/src/mesa/main/readpix.c b/mesalib/src/mesa/main/readpix.c
index a7b7ed7f2..38b9c64ed 100644
--- a/mesalib/src/mesa/main/readpix.c
+++ b/mesalib/src/mesa/main/readpix.c
@@ -252,10 +252,7 @@ slow_read_rgba_pixels( struct gl_context *ctx,
 {
    struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer;
    const gl_format rbFormat = _mesa_get_srgb_format_linear(rb->Format);
-   union {
-      float f[MAX_WIDTH][4];
-      unsigned int i[MAX_WIDTH][4];
-   } rgba;
+   void *rgba;
    GLubyte *dst, *map;
    int dstStride, stride, j;
 
@@ -270,19 +267,27 @@ slow_read_rgba_pixels( struct gl_context *ctx,
       return;
    }
 
+   rgba = malloc(width * MAX_PIXEL_BYTES);
+   if (!rgba)
+      goto done;
+
    for (j = 0; j < height; j++) {
       if (_mesa_is_integer_format(format)) {
-	 _mesa_unpack_int_rgba_row(rbFormat, width, map, rgba.i);
-	 _mesa_pack_rgba_span_int(ctx, width, rgba.i, format, type, dst);
+	 _mesa_unpack_int_rgba_row(rbFormat, width, map, (GLuint (*)[4]) rgba);
+	 _mesa_pack_rgba_span_int(ctx, width, (GLuint (*)[4]) rgba, format,
+                                  type, dst);
       } else {
-	 _mesa_unpack_rgba_row(rbFormat, width, map, rgba.f);
-	 _mesa_pack_rgba_span_float(ctx, width, rgba.f, format, type, dst,
-				    packing, transferOps);
+	 _mesa_unpack_rgba_row(rbFormat, width, map, (GLfloat (*)[4]) rgba);
+	 _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format,
+                                    type, dst, packing, transferOps);
       }
       dst += dstStride;
       map += stride;
    }
 
+   free(rgba);
+
+done:
    ctx->Driver.UnmapRenderbuffer(ctx, rb);
 }
 
diff --git a/mesalib/src/mesa/main/renderbuffer.c b/mesalib/src/mesa/main/renderbuffer.c
index 24060e695..d82b19ad5 100644
--- a/mesalib/src/mesa/main/renderbuffer.c
+++ b/mesalib/src/mesa/main/renderbuffer.c
@@ -23,18 +23,6 @@
  */
 
 
-/**
- * Functions for allocating/managing renderbuffers.
- * Also, routines for reading/writing software-based renderbuffer data as
- * ubytes, ushorts, uints, etc.
- *
- * Down the road we'll use this for run-time support of 8, 16 and 32-bit
- * color channels.  For example, Mesa may use 32-bit/float color channels
- * internally (swrast) and use wrapper renderbuffers to convert 32-bit
- * values down to 16 or 8-bit values for whatever kind of framebuffer we have.
- */
-
-
 #include "glheader.h"
 #include "imports.h"
 #include "context.h"
@@ -44,1700 +32,6 @@
 #include "renderbuffer.h"
 
 
-/*
- * Routines for get/put values in common buffer formats follow.
- */
-
-/* Returns a bytes per pixel of the DataType in the get/put span
- * functions for at least a subset of the available combinations a
- * renderbuffer can have.
- *
- * It would be nice to see gl_renderbuffer start talking about a
- * gl_format instead of a GLenum DataType.
- */
-static int
-get_datatype_bytes(struct gl_renderbuffer *rb)
-{
-   int component_size;
-
-   switch (rb->DataType) {
-   case GL_FLOAT_32_UNSIGNED_INT_24_8_REV:
-      component_size = 8;
-      break;
-   case GL_FLOAT:
-   case GL_UNSIGNED_INT:
-   case GL_UNSIGNED_INT_24_8_EXT:
-      component_size = 4;
-      break;
-   case GL_UNSIGNED_SHORT:
-      component_size = 2;
-      break;
-   case GL_UNSIGNED_BYTE:
-      component_size = 1;
-      break;
-   default:
-      component_size = 1;
-      assert(0);
-   }
-
-   switch (rb->_BaseFormat) {
-   case GL_DEPTH_COMPONENT:
-   case GL_DEPTH_STENCIL:
-      return component_size;
-   default:
-      return 4 * component_size;
-   }
-}
-
-/* This is commonly used by most of the accessors. */
-static void *
-get_pointer_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		    GLint x, GLint y)
-{
-   if (!rb->Data)
-      return NULL;
-
-   return ((char *) rb->Data +
-	   (y * rb->RowStride + x) * _mesa_get_format_bytes(rb->Format));
-}
-
-/* GetRow() implementation for formats where DataType matches the rb->Format.
- */
-static void
-get_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		GLuint count, GLint x, GLint y, void *values)
-{
-   void *src = rb->GetPointer(ctx, rb, x, y);
-   memcpy(values, src, count * _mesa_get_format_bytes(rb->Format));
-}
-
-/* Only used for float textures currently, but might also be used for
- * RGBA8888, RGBA16, etc.
- */
-static void
-get_values_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		   GLuint count, const GLint x[], const GLint y[], void *values)
-{
-   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const void *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      char *dst = (char *) values + i * format_bytes;
-      memcpy(dst, src, format_bytes);
-   }
-}
-
-/* For the GL_RED/GL_RG/GL_RGB format/DataType combinations (and
- * GL_LUMINANCE/GL_INTENSITY?), the Put functions are a matter of
- * storing those initial components of the value per pixel into the
- * destination.
- */
-static void
-put_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		GLuint count, GLint x, GLint y,
-		const void *values, const GLubyte *mask)
-{
-   void *row = rb->GetPointer(ctx, rb, x, y);
-   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
-   int datatype_bytes = get_datatype_bytes(rb);
-   unsigned int i;
-
-   if (mask) {
-      for (i = 0; i < count; i++) {
-	 char *dst = (char *) row + i * format_bytes;
-	 const char *src = (const char *) values + i * datatype_bytes;
-
-         if (mask[i]) {
-	    memcpy(dst, src, format_bytes);
-         }
-      }
-   }
-   else {
-      for (i = 0; i < count; i++) {
-	 char *dst = (char *) row + i * format_bytes;
-	 const char *src = (const char *) values + i * datatype_bytes;
-	 memcpy(dst, src, format_bytes);
-      }
-   }
-}
-
-static void
-put_mono_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, GLint x, GLint y,
-		     const void *value, const GLubyte *mask)
-{
-   void *row = rb->GetPointer(ctx, rb, x, y);
-   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
-   unsigned int i;
-
-   if (mask) {
-      for (i = 0; i < count; i++) {
-	 char *dst = (char *) row + i * format_bytes;
-         if (mask[i]) {
-	    memcpy(dst, value, format_bytes);
-         }
-      }
-   }
-   else {
-      for (i = 0; i < count; i++) {
-	 char *dst = (char *) row + i * format_bytes;
-	 memcpy(dst, value, format_bytes);
-      }
-   }
-}
-
-
-static void
-put_values_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		   GLuint count, const GLint x[], const GLint y[],
-		   const void *values, const GLubyte *mask)
-{
-   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
-   int datatype_bytes = get_datatype_bytes(rb);
-   unsigned int i;
-
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-	 void *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
-	 const char *src = (const char *) values + i * datatype_bytes;
-	 memcpy(dst, src, format_bytes);
-      }
-   }
-}
-
-
-static void
-put_mono_values_generic(struct gl_context *ctx,
-			struct gl_renderbuffer *rb,
-			GLuint count, const GLint x[], const GLint y[],
-			const void *value, const GLubyte *mask)
-{
-   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
-   unsigned int i;
-
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-	 void *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
-	 memcpy(dst, value, format_bytes);
-      }
-   }
-}
-
-/**********************************************************************
- * Functions for buffers of 1 X GLubyte values.
- * Typically stencil.
- */
-
-static void
-get_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                 const GLint x[], const GLint y[], void *values)
-{
-   GLubyte *dst = (GLubyte *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      const GLubyte *src = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
-      dst[i] = *src;
-   }
-}
-
-
-static void
-put_row_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-              GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   const GLubyte *src = (const GLubyte *) values;
-   GLubyte *dst = (GLubyte *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = src[i];
-         }
-      }
-   }
-   else {
-      memcpy(dst, values, count * sizeof(GLubyte));
-   }
-}
-
-
-static void
-put_mono_row_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                   GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   const GLubyte val = *((const GLubyte *) value);
-   GLubyte *dst = (GLubyte *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = val;
-         }
-      }
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         dst[i] = val;
-      }
-   }
-}
-
-
-static void
-put_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                 const GLint x[], const GLint y[],
-                 const void *values, const GLubyte *mask)
-{
-   const GLubyte *src = (const GLubyte *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLubyte *dst = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = src[i];
-      }
-   }
-}
-
-
-static void
-put_mono_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                      const GLint x[], const GLint y[],
-                      const void *value, const GLubyte *mask)
-{
-   const GLubyte val = *((const GLubyte *) value);
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLubyte *dst = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = val;
-      }
-   }
-}
-
-
-/**********************************************************************
- * Functions for buffers of 1 X GLushort values.
- * Typically depth/Z.
- */
-
-static void
-get_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], void *values)
-{
-   GLushort *dst = (GLushort *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-   for (i = 0; i < count; i++) {
-      const GLushort *src = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
-      dst[i] = *src;
-   }
-}
-
-
-static void
-put_row_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-               GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   const GLushort *src = (const GLushort *) values;
-   GLushort *dst = (GLushort *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = src[i];
-         }
-      }
-   }
-   else {
-      memcpy(dst, src, count * sizeof(GLushort));
-   }
-}
-
-
-static void
-put_mono_row_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                    GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   const GLushort val = *((const GLushort *) value);
-   GLushort *dst = (GLushort *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = val;
-         }
-      }
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         dst[i] = val;
-      }
-   }
-}
-
-
-static void
-put_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], const void *values,
-                  const GLubyte *mask)
-{
-   const GLushort *src = (const GLushort *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = src[i];
-      }
-   }
-}
- 
-
-static void
-put_mono_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                       GLuint count, const GLint x[], const GLint y[],
-                       const void *value, const GLubyte *mask)
-{
-   const GLushort val = *((const GLushort *) value);
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
-            *dst = val;
-         }
-      }
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = val;
-      }
-   }
-}
- 
-
-/**********************************************************************
- * Functions for buffers of 1 X GLuint values.
- * Typically depth/Z or color index.
- */
-
-static void
-get_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                const GLint x[], const GLint y[], void *values)
-{
-   GLuint *dst = (GLuint *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   for (i = 0; i < count; i++) {
-      const GLuint *src = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
-      dst[i] = *src;
-   }
-}
-
-
-static void
-put_row_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-             GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   const GLuint *src = (const GLuint *) values;
-   GLuint *dst = (GLuint *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = src[i];
-         }
-      }
-   }
-   else {
-      memcpy(dst, src, count * sizeof(GLuint));
-   }
-}
-
-
-static void
-put_mono_row_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   const GLuint val = *((const GLuint *) value);
-   GLuint *dst = (GLuint *) rb->Data + y * rb->RowStride + x;
-   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = val;
-         }
-      }
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         dst[i] = val;
-      }
-   }
-}
-
-
-static void
-put_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                const GLint x[], const GLint y[], const void *values,
-                const GLubyte *mask)
-{
-   const GLuint *src = (const GLuint *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLuint *dst = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = src[i];
-      }
-   }
-}
-
-
-static void
-put_mono_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                     const GLint x[], const GLint y[], const void *value,
-                     const GLubyte *mask)
-{
-   const GLuint val = *((const GLuint *) value);
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLuint *dst = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
-         *dst = val;
-      }
-   }
-}
-
-
-/**********************************************************************
- * Functions for buffers of 3 X GLubyte (or GLbyte) values.
- * Typically color buffers.
- * NOTE: the incoming and outgoing colors are RGBA!  We ignore incoming
- * alpha values and return 255 for outgoing alpha values.
- */
-
-static void *
-get_pointer_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                   GLint x, GLint y)
-{
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   /* No direct access since this buffer is RGB but caller will be
-    * treating it as if it were RGBA.
-    */
-   return NULL;
-}
-
-
-static void
-get_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-               GLint x, GLint y, void *values)
-{
-   const GLubyte *src = ((const GLubyte *) rb->Data) +
-					   3 * (y * rb->RowStride + x);
-   GLubyte *dst = (GLubyte *) values;
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + 0] = src[i * 3 + 0];
-      dst[i * 4 + 1] = src[i * 3 + 1];
-      dst[i * 4 + 2] = src[i * 3 + 2];
-      dst[i * 4 + 3] = 255;
-   }
-}
-
-
-static void
-get_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], void *values)
-{
-   GLubyte *dst = (GLubyte *) values;
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      const GLubyte *src
-         = (GLubyte *) rb->Data + 3 * (y[i] * rb->RowStride + x[i]);
-      dst[i * 4 + 0] = src[0];
-      dst[i * 4 + 1] = src[1];
-      dst[i * 4 + 2] = src[2];
-      dst[i * 4 + 3] = 255;
-   }
-}
-
-
-static void
-put_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-               GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   /* note: incoming values are RGB+A! */
-   const GLubyte *src = (const GLubyte *) values;
-   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         dst[i * 3 + 0] = src[i * 4 + 0];
-         dst[i * 3 + 1] = src[i * 4 + 1];
-         dst[i * 3 + 2] = src[i * 4 + 2];
-      }
-   }
-}
-
-
-static void
-put_row_rgb_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                   GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   /* note: incoming values are RGB+A! */
-   const GLubyte *src = (const GLubyte *) values;
-   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         dst[i * 3 + 0] = src[i * 3 + 0];
-         dst[i * 3 + 1] = src[i * 3 + 1];
-         dst[i * 3 + 2] = src[i * 3 + 2];
-      }
-   }
-}
-
-
-static void
-put_mono_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                    GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   /* note: incoming value is RGB+A! */
-   const GLubyte val0 = ((const GLubyte *) value)[0];
-   const GLubyte val1 = ((const GLubyte *) value)[1];
-   const GLubyte val2 = ((const GLubyte *) value)[2];
-   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   if (!mask && val0 == val1 && val1 == val2) {
-      /* optimized case */
-      memset(dst, val0, 3 * count);
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (!mask || mask[i]) {
-            dst[i * 3 + 0] = val0;
-            dst[i * 3 + 1] = val1;
-            dst[i * 3 + 2] = val2;
-         }
-      }
-   }
-}
-
-
-static void
-put_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], const void *values,
-                  const GLubyte *mask)
-{
-   /* note: incoming values are RGB+A! */
-   const GLubyte *src = (const GLubyte *) values;
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLubyte *dst = (GLubyte *) rb->Data + 3 * (y[i] * rb->RowStride + x[i]);
-         dst[0] = src[i * 4 + 0];
-         dst[1] = src[i * 4 + 1];
-         dst[2] = src[i * 4 + 2];
-      }
-   }
-}
-
-
-static void
-put_mono_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                       GLuint count, const GLint x[], const GLint y[],
-                       const void *value, const GLubyte *mask)
-{
-   /* note: incoming value is RGB+A! */
-   const GLubyte val0 = ((const GLubyte *) value)[0];
-   const GLubyte val1 = ((const GLubyte *) value)[1];
-   const GLubyte val2 = ((const GLubyte *) value)[2];
-   GLuint i;
-   ASSERT(rb->Format == MESA_FORMAT_RGB888);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLubyte *dst = ((GLubyte *) rb->Data) +
-				     3 * (y[i] * rb->RowStride + x[i]);
-         dst[0] = val0;
-         dst[1] = val1;
-         dst[2] = val2;
-      }
-   }
-}
-
-
-/**********************************************************************
- * Functions for buffers of 4 X GLubyte (or GLbyte) values.
- * Typically color buffers.
- */
-
-static void
-get_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], void *values)
-{
-   /* treat 4*GLubyte as 1*GLuint */
-   GLuint *dst = (GLuint *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   for (i = 0; i < count; i++) {
-      const GLuint *src = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
-      dst[i] = *src;
-   }
-}
-
-
-static void
-put_row_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-               GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   /* treat 4*GLubyte as 1*GLuint */
-   const GLuint *src = (const GLuint *) values;
-   GLuint *dst = (GLuint *) rb->Data + (y * rb->RowStride + x);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i] = src[i];
-         }
-      }
-   }
-   else {
-      memcpy(dst, src, 4 * count * sizeof(GLubyte));
-   }
-}
-
-
-static void
-put_row_rgb_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                   GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   /* Store RGB values in RGBA buffer */
-   const GLubyte *src = (const GLubyte *) values;
-   GLubyte *dst = (GLubyte *) rb->Data + 4 * (y * rb->RowStride + x);
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         dst[i * 4 + 0] = src[i * 3 + 0];
-         dst[i * 4 + 1] = src[i * 3 + 1];
-         dst[i * 4 + 2] = src[i * 3 + 2];
-         dst[i * 4 + 3] = 0xff;
-      }
-   }
-}
-
-
-static void
-put_mono_row_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                    GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   /* treat 4*GLubyte as 1*GLuint */
-   const GLuint val = *((const GLuint *) value);
-   GLuint *dst = (GLuint *) rb->Data + (y * rb->RowStride + x);
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   if (!mask && val == 0) {
-      /* common case */
-      memset(dst, 0, count * 4 * sizeof(GLubyte));
-   }
-   else {
-      /* general case */
-      if (mask) {
-         GLuint i;
-         for (i = 0; i < count; i++) {
-            if (mask[i]) {
-               dst[i] = val;
-            }
-         }
-      }
-      else {
-         GLuint i;
-         for (i = 0; i < count; i++) {
-            dst[i] = val;
-         }
-      }
-   }
-}
-
-
-static void
-put_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                  const GLint x[], const GLint y[], const void *values,
-                  const GLubyte *mask)
-{
-   /* treat 4*GLubyte as 1*GLuint */
-   const GLuint *src = (const GLuint *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLuint *dst = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
-         *dst = src[i];
-      }
-   }
-}
-
-
-static void
-put_mono_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                       GLuint count, const GLint x[], const GLint y[],
-                       const void *value, const GLubyte *mask)
-{
-   /* treat 4*GLubyte as 1*GLuint */
-   const GLuint val = *((const GLuint *) value);
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
-   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
-          rb->Format == MESA_FORMAT_RGBA8888_REV);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLuint *dst = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
-         *dst = val;
-      }
-   }
-}
-
-
-/**********************************************************************
- * Functions for buffers of 4 X GLushort (or GLshort) values.
- * Typically accum buffer.
- */
-
-static void
-get_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                   const GLint x[], const GLint y[], void *values)
-{
-   GLushort *dst = (GLushort *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   for (i = 0; i < count; i++) {
-      const GLushort *src
-         = (GLushort *) rb->Data + 4 * (y[i] * rb->RowStride + x[i]);
-      dst[i] = *src;
-   }
-}
-
-
-static void
-put_row_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   const GLushort *src = (const GLushort *) values;
-   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i * 4 + 0] = src[i * 4 + 0];
-            dst[i * 4 + 1] = src[i * 4 + 1];
-            dst[i * 4 + 2] = src[i * 4 + 2];
-            dst[i * 4 + 3] = src[i * 4 + 3];
-         }
-      }
-   }
-   else {
-      memcpy(dst, src, 4 * count * sizeof(GLushort));
-   }
-}
-
-
-static void
-put_row_rgb_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                    GLint x, GLint y, const void *values, const GLubyte *mask)
-{
-   /* Put RGB values in RGBA buffer */
-   const GLushort *src = (const GLushort *) values;
-   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   if (mask) {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-            dst[i * 4 + 0] = src[i * 3 + 0];
-            dst[i * 4 + 1] = src[i * 3 + 1];
-            dst[i * 4 + 2] = src[i * 3 + 2];
-            dst[i * 4 + 3] = 0xffff;
-         }
-      }
-   }
-   else {
-      memcpy(dst, src, 4 * count * sizeof(GLushort));
-   }
-}
-
-
-static void
-put_mono_row_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                     GLint x, GLint y, const void *value, const GLubyte *mask)
-{
-   const GLushort val0 = ((const GLushort *) value)[0];
-   const GLushort val1 = ((const GLushort *) value)[1];
-   const GLushort val2 = ((const GLushort *) value)[2];
-   const GLushort val3 = ((const GLushort *) value)[3];
-   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   if (!mask && val0 == 0 && val1 == 0 && val2 == 0 && val3 == 0) {
-      /* common case for clearing accum buffer */
-      memset(dst, 0, count * 4 * sizeof(GLushort));
-   }
-   else {
-      GLuint i;
-      for (i = 0; i < count; i++) {
-         if (!mask || mask[i]) {
-            dst[i * 4 + 0] = val0;
-            dst[i * 4 + 1] = val1;
-            dst[i * 4 + 2] = val2;
-            dst[i * 4 + 3] = val3;
-         }
-      }
-   }
-}
-
-
-static void
-put_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-                   const GLint x[], const GLint y[], const void *values,
-                   const GLubyte *mask)
-{
-   const GLushort *src = (const GLushort *) values;
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLushort *dst =
-            ((GLushort *) rb->Data) + 4 * (y[i] * rb->RowStride + x[i]);
-         dst[0] = src[i * 4 + 0];
-         dst[1] = src[i * 4 + 1];
-         dst[2] = src[i * 4 + 2];
-         dst[3] = src[i * 4 + 3];
-      }
-   }
-}
-
-
-static void
-put_mono_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                        GLuint count, const GLint x[], const GLint y[],
-                        const void *value, const GLubyte *mask)
-{
-   const GLushort val0 = ((const GLushort *) value)[0];
-   const GLushort val1 = ((const GLushort *) value)[1];
-   const GLushort val2 = ((const GLushort *) value)[2];
-   const GLushort val3 = ((const GLushort *) value)[3];
-   GLuint i;
-   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-         GLushort *dst = ((GLushort *) rb->Data) +
-				       4 * (y[i] * rb->RowStride + x[i]);
-         dst[0] = val0;
-         dst[1] = val1;
-         dst[2] = val2;
-         dst[3] = val3;
-      }
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_R8.
- */
-static void
-get_row_r8(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	   GLint x, GLint y, void *values)
-{
-   const GLubyte *src = rb->GetPointer(ctx, rb, x, y);
-   GLuint *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i] = 0xff000000 | src[i];
-   }
-}
-
-static void
-get_values_r8(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	      const GLint x[], const GLint y[], void *values)
-{
-   GLuint *dst = (GLuint *) values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLubyte *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i] = 0xff000000 | *src;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_GR88.
- */
-static void
-get_row_rg88(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	     GLint x, GLint y, void *values)
-{
-   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
-   GLuint *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i] = 0xff000000 | src[i];
-   }
-}
-
-static void
-get_values_rg88(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		GLuint count, const GLint x[], const GLint y[], void *values)
-{
-   GLuint *dst = (GLuint *) values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLshort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i] = 0xff000000 | *src;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_R16.
- */
-static void
-get_row_r16(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	    GLint x, GLint y, void *values)
-{
-   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
-   GLushort *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] = src[i];
-      dst[i * 4 + GCOMP] = 0;
-      dst[i * 4 + BCOMP] = 0;
-      dst[i * 4 + ACOMP] = 0xffff;
-   }
-}
-
-static void
-get_values_r16(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	       const GLint x[], const GLint y[], void *values)
-{
-   GLushort *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLushort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] = *src;
-      dst[i * 4 + GCOMP] = 0;
-      dst[i * 4 + BCOMP] = 0;
-      dst[i * 4 + ACOMP] = 0xffff;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_RG1616.
- */
-static void
-get_row_rg1616(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
-	       GLint x, GLint y, void *values)
-{
-   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
-   GLushort *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] = src[i * 2];
-      dst[i * 4 + GCOMP] = src[i * 2 + 1];
-      dst[i * 4 + BCOMP] = 0;
-      dst[i * 4 + ACOMP] = 0xffff;
-   }
-}
-
-static void
-get_values_rg1616(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, const GLint x[], const GLint y[], void *values)
-{
-   GLushort *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLshort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] = src[0];
-      dst[i * 4 + GCOMP] = src[1];
-      dst[i * 4 + BCOMP] = 0;
-      dst[i * 4 + ACOMP] = 0xffff;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_INTENSITY_FLOAT32.
- */
-static void
-get_row_i_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, GLint x, GLint y, void *values)
-{
-   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] =
-      dst[i * 4 + GCOMP] =
-      dst[i * 4 + BCOMP] =
-      dst[i * 4 + ACOMP] = src[i];
-   }
-}
-
-static void
-get_values_i_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, const GLint x[], const GLint y[],
-		     void *values)
-{
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] =
-      dst[i * 4 + GCOMP] =
-      dst[i * 4 + BCOMP] =
-      dst[i * 4 + ACOMP] = src[0];
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_LUMINANCE_FLOAT32.
- */
-static void
-get_row_l_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, GLint x, GLint y, void *values)
-{
-   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] =
-      dst[i * 4 + GCOMP] =
-      dst[i * 4 + BCOMP] = src[i];
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-static void
-get_values_l_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, const GLint x[], const GLint y[],
-		     void *values)
-{
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] =
-      dst[i * 4 + GCOMP] =
-      dst[i * 4 + BCOMP] = src[0];
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_ALPHA_FLOAT32.
- */
-static void
-get_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, GLint x, GLint y, void *values)
-{
-   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] = 0.0;
-      dst[i * 4 + GCOMP] = 0.0;
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = src[i];
-   }
-}
-
-static void
-get_values_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, const GLint x[], const GLint y[],
-		     void *values)
-{
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] = 0.0;
-      dst[i * 4 + GCOMP] = 0.0;
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = src[0];
-   }
-}
-
-static void
-put_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, GLint x, GLint y,
-		  const void *values, const GLubyte *mask)
-{
-   float *dst = rb->GetPointer(ctx, rb, x, y);
-   const float *src = values;
-   unsigned int i;
-
-   if (mask) {
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-	    dst[i] = src[i * 4 + ACOMP];
-         }
-      }
-   }
-   else {
-      for (i = 0; i < count; i++) {
-	 dst[i] = src[i * 4 + ACOMP];
-      }
-   }
-}
-
-static void
-put_mono_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		       GLuint count, GLint x, GLint y,
-		       const void *value, const GLubyte *mask)
-{
-   float *dst = rb->GetPointer(ctx, rb, x, y);
-   const float *src = value;
-   unsigned int i;
-
-   if (mask) {
-      for (i = 0; i < count; i++) {
-         if (mask[i]) {
-	    dst[i] = src[ACOMP];
-         }
-      }
-   }
-   else {
-      for (i = 0; i < count; i++) {
-	 dst[i] = src[ACOMP];
-      }
-   }
-}
-
-static void
-put_values_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, const GLint x[], const GLint y[],
-		     const void *values, const GLubyte *mask)
-{
-   const float *src = values;
-   unsigned int i;
-
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-	 float *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
-
-	 *dst = src[i * 4 + ACOMP];
-      }
-   }
-}
-
-static void
-put_mono_values_a_float32(struct gl_context *ctx,
-			  struct gl_renderbuffer *rb,
-			  GLuint count, const GLint x[], const GLint y[],
-			  const void *value, const GLubyte *mask)
-{
-   const float *src = value;
-   unsigned int i;
-
-   for (i = 0; i < count; i++) {
-      if (!mask || mask[i]) {
-	 float *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
-	 *dst = src[ACOMP];
-      }
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_R_FLOAT32.
- */
-static void
-get_row_r_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		  GLuint count, GLint x, GLint y, void *values)
-{
-   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] = src[i];
-      dst[i * 4 + GCOMP] = 0.0;
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-static void
-get_values_r_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		     GLuint count, const GLint x[], const GLint y[],
-		     void *values)
-{
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] = src[0];
-      dst[i * 4 + GCOMP] = 0.0;
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-/**********************************************************************
- * Functions for MESA_FORMAT_RG_FLOAT32.
- */
-static void
-get_row_rg_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		   GLuint count, GLint x, GLint y, void *values)
-{
-   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      dst[i * 4 + RCOMP] = src[i * 2 + 0];
-      dst[i * 4 + GCOMP] = src[i * 2 + 1];
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-static void
-get_values_rg_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
-		      GLuint count, const GLint x[], const GLint y[],
-		      void *values)
-{
-   GLfloat *dst = values;
-   GLuint i;
-
-   for (i = 0; i < count; i++) {
-      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
-      dst[i * 4 + RCOMP] = src[0];
-      dst[i * 4 + GCOMP] = src[1];
-      dst[i * 4 + BCOMP] = 0.0;
-      dst[i * 4 + ACOMP] = 1.0;
-   }
-}
-
-/**
- * This is the default software fallback for gl_renderbuffer's span
- * access functions.
- *
- * The assumptions are that rb->Data will be a pointer to (0,0), that pixels
- * are packed in the type of rb->Format, and that subsequent rows appear
- * rb->RowStride pixels later.
- */
-void
-_mesa_set_renderbuffer_accessors(struct gl_renderbuffer *rb)
-{
-   rb->GetPointer = get_pointer_generic;
-   rb->GetRow = get_row_generic;
-
-   switch (rb->Format) {
-   case MESA_FORMAT_RGB888:
-      rb->DataType = GL_UNSIGNED_BYTE;
-      rb->GetPointer = get_pointer_ubyte3;
-      rb->GetRow = get_row_ubyte3;
-      rb->GetValues = get_values_ubyte3;
-      rb->PutRow = put_row_ubyte3;
-      rb->PutRowRGB = put_row_rgb_ubyte3;
-      rb->PutMonoRow = put_mono_row_ubyte3;
-      rb->PutValues = put_values_ubyte3;
-      rb->PutMonoValues = put_mono_values_ubyte3;
-      break;
-
-   case MESA_FORMAT_RGBA8888:
-   case MESA_FORMAT_RGBA8888_REV:
-      rb->DataType = GL_UNSIGNED_BYTE;
-      rb->GetValues = get_values_ubyte4;
-      rb->PutRow = put_row_ubyte4;
-      rb->PutRowRGB = put_row_rgb_ubyte4;
-      rb->PutMonoRow = put_mono_row_ubyte4;
-      rb->PutValues = put_values_ubyte4;
-      rb->PutMonoValues = put_mono_values_ubyte4;
-      break;
-
-   case MESA_FORMAT_R8:
-      rb->DataType = GL_UNSIGNED_BYTE;
-      rb->GetValues = get_values_r8;
-      rb->GetRow = get_row_r8;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = put_row_generic;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_GR88:
-      rb->DataType = GL_UNSIGNED_BYTE;
-      rb->GetValues = get_values_rg88;
-      rb->GetRow = get_row_rg88;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = put_row_generic;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_R16:
-      rb->DataType = GL_UNSIGNED_SHORT;
-      rb->GetValues = get_values_r16;
-      rb->GetRow = get_row_r16;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = put_row_generic;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_RG1616:
-      rb->DataType = GL_UNSIGNED_SHORT;
-      rb->GetValues = get_values_rg1616;
-      rb->GetRow = get_row_rg1616;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = put_row_generic;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_SIGNED_RGBA_16:
-      rb->DataType = GL_SHORT;
-      rb->GetValues = get_values_ushort4;
-      rb->PutRow = put_row_ushort4;
-      rb->PutRowRGB = put_row_rgb_ushort4;
-      rb->PutMonoRow = put_mono_row_ushort4;
-      rb->PutValues = put_values_ushort4;
-      rb->PutMonoValues = put_mono_values_ushort4;
-      break;
-
-   case MESA_FORMAT_S8:
-      rb->DataType = GL_UNSIGNED_BYTE;
-      rb->GetValues = get_values_ubyte;
-      rb->PutRow = put_row_ubyte;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_ubyte;
-      rb->PutValues = put_values_ubyte;
-      rb->PutMonoValues = put_mono_values_ubyte;
-      break;
-
-   case MESA_FORMAT_Z16:
-      rb->DataType = GL_UNSIGNED_SHORT;
-      rb->GetValues = get_values_ushort;
-      rb->PutRow = put_row_ushort;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_ushort;
-      rb->PutValues = put_values_ushort;
-      rb->PutMonoValues = put_mono_values_ushort;
-      break;
-
-   case MESA_FORMAT_Z32:
-   case MESA_FORMAT_X8_Z24:
-   case MESA_FORMAT_Z24_X8:
-      rb->DataType = GL_UNSIGNED_INT;
-      rb->GetValues = get_values_uint;
-      rb->PutRow = put_row_uint;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_uint;
-      rb->PutValues = put_values_uint;
-      rb->PutMonoValues = put_mono_values_uint;
-      break;
-
-   case MESA_FORMAT_Z24_S8:
-   case MESA_FORMAT_S8_Z24:
-      rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->GetValues = get_values_uint;
-      rb->PutRow = put_row_uint;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_uint;
-      rb->PutValues = put_values_uint;
-      rb->PutMonoValues = put_mono_values_uint;
-      break;
-
-   case MESA_FORMAT_RGBA_FLOAT32:
-      rb->GetRow = get_row_generic;
-      rb->GetValues = get_values_generic;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_INTENSITY_FLOAT32:
-      rb->GetRow = get_row_i_float32;
-      rb->GetValues = get_values_i_float32;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_LUMINANCE_FLOAT32:
-      rb->GetRow = get_row_l_float32;
-      rb->GetValues = get_values_l_float32;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_ALPHA_FLOAT32:
-      rb->GetRow = get_row_a_float32;
-      rb->GetValues = get_values_a_float32;
-      rb->PutRow = put_row_a_float32;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_a_float32;
-      rb->PutValues = put_values_a_float32;
-      rb->PutMonoValues = put_mono_values_a_float32;
-      break;
-
-   case MESA_FORMAT_RG_FLOAT32:
-      rb->GetRow = get_row_rg_float32;
-      rb->GetValues = get_values_rg_float32;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   case MESA_FORMAT_R_FLOAT32:
-      rb->GetRow = get_row_r_float32;
-      rb->GetValues = get_values_r_float32;
-      rb->PutRow = put_row_generic;
-      rb->PutRowRGB = NULL;
-      rb->PutMonoRow = put_mono_row_generic;
-      rb->PutValues = put_values_generic;
-      rb->PutMonoValues = put_mono_values_generic;
-      break;
-
-   default:
-      break;
-   }
-}
-
-/**
- * This is a software fallback for the gl_renderbuffer->AllocStorage
- * function.
- * Device drivers will typically override this function for the buffers
- * which it manages (typically color buffers, Z and stencil).
- * Other buffers (like software accumulation and aux buffers) which the driver
- * doesn't manage can be handled with this function.
- *
- * This one multi-purpose function can allocate stencil, depth, accum, color
- * or color-index buffers!
- *
- * This function also plugs in the appropriate GetPointer, Get/PutRow and
- * Get/PutValues functions.
- */
-GLboolean
-_mesa_soft_renderbuffer_storage(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                                GLenum internalFormat,
-                                GLuint width, GLuint height)
-{
-   switch (internalFormat) {
-   case GL_RGB:
-   case GL_R3_G3_B2:
-   case GL_RGB4:
-   case GL_RGB5:
-   case GL_RGB8:
-   case GL_RGB10:
-   case GL_RGB12:
-   case GL_RGB16:
-      rb->Format = MESA_FORMAT_RGB888;
-      break;
-   case GL_RGBA:
-   case GL_RGBA2:
-   case GL_RGBA4:
-   case GL_RGB5_A1:
-   case GL_RGBA8:
-#if 1
-   case GL_RGB10_A2:
-   case GL_RGBA12:
-#endif
-      if (_mesa_little_endian())
-         rb->Format = MESA_FORMAT_RGBA8888_REV;
-      else
-         rb->Format = MESA_FORMAT_RGBA8888;
-      break;
-   case GL_RGBA16:
-   case GL_RGBA16_SNORM:
-      /* for accum buffer */
-      rb->Format = MESA_FORMAT_SIGNED_RGBA_16;
-      break;
-   case GL_STENCIL_INDEX:
-   case GL_STENCIL_INDEX1_EXT:
-   case GL_STENCIL_INDEX4_EXT:
-   case GL_STENCIL_INDEX8_EXT:
-   case GL_STENCIL_INDEX16_EXT:
-      rb->Format = MESA_FORMAT_S8;
-      break;
-   case GL_DEPTH_COMPONENT:
-   case GL_DEPTH_COMPONENT16:
-      rb->Format = MESA_FORMAT_Z16;
-      break;
-   case GL_DEPTH_COMPONENT24:
-      rb->Format = MESA_FORMAT_X8_Z24;
-      break;
-   case GL_DEPTH_COMPONENT32:
-      rb->Format = MESA_FORMAT_Z32;
-      break;
-   case GL_DEPTH_STENCIL_EXT:
-   case GL_DEPTH24_STENCIL8_EXT:
-      rb->Format = MESA_FORMAT_Z24_S8;
-      break;
-   default:
-      /* unsupported format */
-      return GL_FALSE;
-   }
-
-   _mesa_set_renderbuffer_accessors(rb);
-
-   ASSERT(rb->DataType);
-   ASSERT(rb->GetPointer);
-   ASSERT(rb->GetRow);
-   ASSERT(rb->GetValues);
-   ASSERT(rb->PutRow);
-   ASSERT(rb->PutMonoRow);
-   ASSERT(rb->PutValues);
-   ASSERT(rb->PutMonoValues);
-
-   /* free old buffer storage */
-   if (rb->Data) {
-      free(rb->Data);
-      rb->Data = NULL;
-   }
-
-   rb->RowStride = width;
-
-   if (width > 0 && height > 0) {
-      /* allocate new buffer storage */
-      rb->Data = malloc(width * height * _mesa_get_format_bytes(rb->Format));
-
-      if (rb->Data == NULL) {
-         rb->Width = 0;
-         rb->Height = 0;
-	 rb->RowStride = 0;
-         _mesa_error(ctx, GL_OUT_OF_MEMORY,
-                     "software renderbuffer allocation (%d x %d x %d)",
-                     width, height, _mesa_get_format_bytes(rb->Format));
-         return GL_FALSE;
-      }
-   }
-
-   rb->Width = width;
-   rb->Height = height;
-   rb->_BaseFormat = _mesa_base_fbo_format(ctx, internalFormat);
-
-   if (rb->Name == 0 &&
-       internalFormat == GL_RGBA16_SNORM &&
-       rb->_BaseFormat == 0) {
-      /* NOTE: This is a special case just for accumulation buffers.
-       * This is a very limited use case- there's no snorm texturing or
-       * rendering going on.
-       */
-      rb->_BaseFormat = GL_RGBA;
-   }
-   else {
-      /* the internalFormat should have been error checked long ago */
-      ASSERT(rb->_BaseFormat);
-   }
-
-   return GL_TRUE;
-}
-
-
-void
-_mesa_map_soft_renderbuffer(struct gl_context *ctx,
-			    struct gl_renderbuffer *rb,
-			    GLuint x, GLuint y, GLuint w, GLuint h,
-			    GLbitfield mode,
-			    GLubyte **out_map,
-			    GLint *out_stride)
-{
-   GLubyte *map = rb->Data;
-   int cpp = _mesa_get_format_bytes(rb->Format);
-   int stride = rb->RowStride * cpp;
-
-   ASSERT(rb->Data);
-
-   map += y * stride;
-   map += x * cpp;
-
-   *out_map = map;
-   *out_stride = stride;
-}
-
-void
-_mesa_unmap_soft_renderbuffer(struct gl_context *ctx,
-			      struct gl_renderbuffer *rb)
-{
-}
-
-
-
-/**********************************************************************/
-/**********************************************************************/
-/**********************************************************************/
-
-
 /**
  * Default GetPointer routine.  Always return NULL to indicate that
  * direct buffer access is not supported.
@@ -1821,310 +115,6 @@ _mesa_delete_renderbuffer(struct gl_renderbuffer *rb)
 
 
 /**
- * Allocate a software-based renderbuffer.  This is called via the
- * ctx->Driver.NewRenderbuffer() function when the user creates a new
- * renderbuffer.
- * This would not be used for hardware-based renderbuffers.
- */
-struct gl_renderbuffer *
-_mesa_new_soft_renderbuffer(struct gl_context *ctx, GLuint name)
-{
-   struct gl_renderbuffer *rb = _mesa_new_renderbuffer(ctx, name);
-   if (rb) {
-      rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-      /* Normally, one would setup the PutRow, GetRow, etc functions here.
-       * But we're doing that in the _mesa_soft_renderbuffer_storage() function
-       * instead.
-       */
-   }
-   return rb;
-}
-
-
-/**
- * Add software-based color renderbuffers to the given framebuffer.
- * This is a helper routine for device drivers when creating a
- * window system framebuffer (not a user-created render/framebuffer).
- * Once this function is called, you can basically forget about this
- * renderbuffer; core Mesa will handle all the buffer management and
- * rendering!
- */
-GLboolean
-_mesa_add_color_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
-                              GLuint rgbBits, GLuint alphaBits,
-                              GLboolean frontLeft, GLboolean backLeft,
-                              GLboolean frontRight, GLboolean backRight)
-{
-   gl_buffer_index b;
-
-   if (rgbBits > 16 || alphaBits > 16) {
-      _mesa_problem(ctx,
-                    "Unsupported bit depth in _mesa_add_color_renderbuffers");
-      return GL_FALSE;
-   }
-
-   assert(MAX_COLOR_ATTACHMENTS >= 4);
-
-   for (b = BUFFER_FRONT_LEFT; b <= BUFFER_BACK_RIGHT; b++) {
-      struct gl_renderbuffer *rb;
-
-      if (b == BUFFER_FRONT_LEFT && !frontLeft)
-         continue;
-      else if (b == BUFFER_BACK_LEFT && !backLeft)
-         continue;
-      else if (b == BUFFER_FRONT_RIGHT && !frontRight)
-         continue;
-      else if (b == BUFFER_BACK_RIGHT && !backRight)
-         continue;
-
-      assert(fb->Attachment[b].Renderbuffer == NULL);
-
-      rb = _mesa_new_renderbuffer(ctx, 0);
-      if (!rb) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating color buffer");
-         return GL_FALSE;
-      }
-
-      rb->InternalFormat = GL_RGBA;
-
-      rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-      _mesa_add_renderbuffer(fb, b, rb);
-   }
-
-   return GL_TRUE;
-}
-
-
-/**
- * Add a software-based depth renderbuffer to the given framebuffer.
- * This is a helper routine for device drivers when creating a
- * window system framebuffer (not a user-created render/framebuffer).
- * Once this function is called, you can basically forget about this
- * renderbuffer; core Mesa will handle all the buffer management and
- * rendering!
- */
-GLboolean
-_mesa_add_depth_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                             GLuint depthBits)
-{
-   struct gl_renderbuffer *rb;
-
-   if (depthBits > 32) {
-      _mesa_problem(ctx,
-                    "Unsupported depthBits in _mesa_add_depth_renderbuffer");
-      return GL_FALSE;
-   }
-
-   assert(fb->Attachment[BUFFER_DEPTH].Renderbuffer == NULL);
-
-   rb = _mesa_new_renderbuffer(ctx, 0);
-   if (!rb) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating depth buffer");
-      return GL_FALSE;
-   }
-
-   if (depthBits <= 16) {
-      rb->InternalFormat = GL_DEPTH_COMPONENT16;
-   }
-   else if (depthBits <= 24) {
-      rb->InternalFormat = GL_DEPTH_COMPONENT24;
-   }
-   else {
-      rb->InternalFormat = GL_DEPTH_COMPONENT32;
-   }
-
-   rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-   _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb);
-
-   return GL_TRUE;
-}
-
-
-/**
- * Add a software-based stencil renderbuffer to the given framebuffer.
- * This is a helper routine for device drivers when creating a
- * window system framebuffer (not a user-created render/framebuffer).
- * Once this function is called, you can basically forget about this
- * renderbuffer; core Mesa will handle all the buffer management and
- * rendering!
- */
-GLboolean
-_mesa_add_stencil_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                               GLuint stencilBits)
-{
-   struct gl_renderbuffer *rb;
-
-   if (stencilBits > 16) {
-      _mesa_problem(ctx,
-                  "Unsupported stencilBits in _mesa_add_stencil_renderbuffer");
-      return GL_FALSE;
-   }
-
-   assert(fb->Attachment[BUFFER_STENCIL].Renderbuffer == NULL);
-
-   rb = _mesa_new_renderbuffer(ctx, 0);
-   if (!rb) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating stencil buffer");
-      return GL_FALSE;
-   }
-
-   assert(stencilBits <= 8);
-   rb->InternalFormat = GL_STENCIL_INDEX8;
-
-   rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-   _mesa_add_renderbuffer(fb, BUFFER_STENCIL, rb);
-
-   return GL_TRUE;
-}
-
-
-/**
- * Add a software-based accumulation renderbuffer to the given framebuffer.
- * This is a helper routine for device drivers when creating a
- * window system framebuffer (not a user-created render/framebuffer).
- * Once this function is called, you can basically forget about this
- * renderbuffer; core Mesa will handle all the buffer management and
- * rendering!
- */
-GLboolean
-_mesa_add_accum_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                             GLuint redBits, GLuint greenBits,
-                             GLuint blueBits, GLuint alphaBits)
-{
-   struct gl_renderbuffer *rb;
-
-   if (redBits > 16 || greenBits > 16 || blueBits > 16 || alphaBits > 16) {
-      _mesa_problem(ctx,
-                    "Unsupported accumBits in _mesa_add_accum_renderbuffer");
-      return GL_FALSE;
-   }
-
-   assert(fb->Attachment[BUFFER_ACCUM].Renderbuffer == NULL);
-
-   rb = _mesa_new_renderbuffer(ctx, 0);
-   if (!rb) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating accum buffer");
-      return GL_FALSE;
-   }
-
-   rb->InternalFormat = GL_RGBA16_SNORM;
-   rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-   _mesa_add_renderbuffer(fb, BUFFER_ACCUM, rb);
-
-   return GL_TRUE;
-}
-
-
-
-/**
- * Add a software-based aux renderbuffer to the given framebuffer.
- * This is a helper routine for device drivers when creating a
- * window system framebuffer (not a user-created render/framebuffer).
- * Once this function is called, you can basically forget about this
- * renderbuffer; core Mesa will handle all the buffer management and
- * rendering!
- *
- * NOTE: color-index aux buffers not supported.
- */
-GLboolean
-_mesa_add_aux_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
-                            GLuint colorBits, GLuint numBuffers)
-{
-   GLuint i;
-
-   if (colorBits > 16) {
-      _mesa_problem(ctx,
-                    "Unsupported accumBits in _mesa_add_aux_renderbuffers");
-      return GL_FALSE;
-   }
-
-   assert(numBuffers <= MAX_AUX_BUFFERS);
-
-   for (i = 0; i < numBuffers; i++) {
-      struct gl_renderbuffer *rb = _mesa_new_renderbuffer(ctx, 0);
-
-      assert(fb->Attachment[BUFFER_AUX0 + i].Renderbuffer == NULL);
-
-      if (!rb) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating aux buffer");
-         return GL_FALSE;
-      }
-
-      assert (colorBits <= 8);
-      rb->InternalFormat = GL_RGBA;
-
-      rb->AllocStorage = _mesa_soft_renderbuffer_storage;
-      _mesa_add_renderbuffer(fb, BUFFER_AUX0 + i, rb);
-   }
-   return GL_TRUE;
-}
-
-
-/**
- * Create/attach software-based renderbuffers to the given framebuffer.
- * This is a helper routine for device drivers.  Drivers can just as well
- * call the individual _mesa_add_*_renderbuffer() routines directly.
- */
-void
-_mesa_add_soft_renderbuffers(struct gl_framebuffer *fb,
-                             GLboolean color,
-                             GLboolean depth,
-                             GLboolean stencil,
-                             GLboolean accum,
-                             GLboolean alpha,
-                             GLboolean aux)
-{
-   GLboolean frontLeft = GL_TRUE;
-   GLboolean backLeft = fb->Visual.doubleBufferMode;
-   GLboolean frontRight = fb->Visual.stereoMode;
-   GLboolean backRight = fb->Visual.stereoMode && fb->Visual.doubleBufferMode;
-
-   if (color) {
-      assert(fb->Visual.redBits == fb->Visual.greenBits);
-      assert(fb->Visual.redBits == fb->Visual.blueBits);
-      _mesa_add_color_renderbuffers(NULL, fb,
-				    fb->Visual.redBits,
-				    fb->Visual.alphaBits,
-				    frontLeft, backLeft,
-				    frontRight, backRight);
-   }
-
-   if (depth) {
-      assert(fb->Visual.depthBits > 0);
-      _mesa_add_depth_renderbuffer(NULL, fb, fb->Visual.depthBits);
-   }
-
-   if (stencil) {
-      assert(fb->Visual.stencilBits > 0);
-      _mesa_add_stencil_renderbuffer(NULL, fb, fb->Visual.stencilBits);
-   }
-
-   if (accum) {
-      assert(fb->Visual.accumRedBits > 0);
-      assert(fb->Visual.accumGreenBits > 0);
-      assert(fb->Visual.accumBlueBits > 0);
-      _mesa_add_accum_renderbuffer(NULL, fb,
-                                   fb->Visual.accumRedBits,
-                                   fb->Visual.accumGreenBits,
-                                   fb->Visual.accumBlueBits,
-                                   fb->Visual.accumAlphaBits);
-   }
-
-   if (aux) {
-      assert(fb->Visual.numAuxBuffers > 0);
-      _mesa_add_aux_renderbuffers(NULL, fb, fb->Visual.redBits,
-                                  fb->Visual.numAuxBuffers);
-   }
-
-#if 0
-   if (multisample) {
-      /* maybe someday */
-   }
-#endif
-}
-
-
-/**
  * Attach a renderbuffer to a framebuffer.
  * \param bufferName  one of the BUFFER_x tokens
  */
@@ -2166,17 +156,9 @@ void
 _mesa_remove_renderbuffer(struct gl_framebuffer *fb,
                           gl_buffer_index bufferName)
 {
-   struct gl_renderbuffer *rb;
-
    assert(bufferName < BUFFER_COUNT);
-
-   rb = fb->Attachment[bufferName].Renderbuffer;
-   if (!rb)
-      return;
-
-   _mesa_reference_renderbuffer(&rb, NULL);
-
-   fb->Attachment[bufferName].Renderbuffer = NULL;
+   _mesa_reference_renderbuffer(&fb->Attachment[bufferName].Renderbuffer,
+                                NULL);
 }
 
 
diff --git a/mesalib/src/mesa/main/renderbuffer.h b/mesalib/src/mesa/main/renderbuffer.h
index 3194fc3fe..0934d85df 100644
--- a/mesalib/src/mesa/main/renderbuffer.h
+++ b/mesalib/src/mesa/main/renderbuffer.h
@@ -42,63 +42,6 @@ _mesa_new_renderbuffer(struct gl_context *ctx, GLuint name);
 extern void
 _mesa_delete_renderbuffer(struct gl_renderbuffer *rb);
 
-
-extern struct gl_renderbuffer *
-_mesa_new_soft_renderbuffer(struct gl_context *ctx, GLuint name);
-
-void
-_mesa_map_soft_renderbuffer(struct gl_context *ctx,
-			    struct gl_renderbuffer *rb,
-			    GLuint x, GLuint y, GLuint w, GLuint h,
-			    GLbitfield mode,
-			    GLubyte **out_map,
-			    GLint *out_stride);
-
-void
-_mesa_unmap_soft_renderbuffer(struct gl_context *ctx,
-			      struct gl_renderbuffer *rb);
-
-extern void
-_mesa_set_renderbuffer_accessors(struct gl_renderbuffer *rb);
-
-extern GLboolean
-_mesa_soft_renderbuffer_storage(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                                GLenum internalFormat,
-                                GLuint width, GLuint height);
-
-extern GLboolean
-_mesa_add_color_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
-                              GLuint rgbBits, GLuint alphaBits,
-                              GLboolean frontLeft, GLboolean backLeft,
-                              GLboolean frontRight, GLboolean backRight);
-
-extern GLboolean
-_mesa_add_depth_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                             GLuint depthBits);
-
-extern GLboolean
-_mesa_add_stencil_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                               GLuint stencilBits);
-
-
-extern GLboolean
-_mesa_add_accum_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
-                             GLuint redBits, GLuint greenBits,
-                             GLuint blueBits, GLuint alphaBits);
-
-extern GLboolean
-_mesa_add_aux_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
-                            GLuint bits, GLuint numBuffers);
-
-extern void
-_mesa_add_soft_renderbuffers(struct gl_framebuffer *fb,
-                             GLboolean color,
-                             GLboolean depth,
-                             GLboolean stencil,
-                             GLboolean accum,
-                             GLboolean alpha,
-                             GLboolean aux);
-
 extern void
 _mesa_add_renderbuffer(struct gl_framebuffer *fb,
                        gl_buffer_index bufferName, struct gl_renderbuffer *rb);
diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c
index 17c78ce3e..8e447cabd 100644
--- a/mesalib/src/mesa/main/texobj.c
+++ b/mesalib/src/mesa/main/texobj.c
@@ -1184,10 +1184,7 @@ _mesa_PrioritizeTextures( GLsizei n, const GLuint *texName,
  *
  * \return GL_TRUE if all textures are resident and \p residences is left unchanged, 
  * 
- * \sa glAreTexturesResident().
- *
- * Looks up each texture in the hash and calls
- * dd_function_table::IsTextureResident.
+ * Note: we assume all textures are always resident
  */
 GLboolean GLAPIENTRY
 _mesa_AreTexturesResident(GLsizei n, const GLuint *texName,
@@ -1195,7 +1192,7 @@ _mesa_AreTexturesResident(GLsizei n, const GLuint *texName,
 {
    GET_CURRENT_CONTEXT(ctx);
    GLboolean allResident = GL_TRUE;
-   GLint i, j;
+   GLint i;
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);
 
    if (n < 0) {
@@ -1206,6 +1203,7 @@ _mesa_AreTexturesResident(GLsizei n, const GLuint *texName,
    if (!texName || !residences)
       return GL_FALSE;
 
+   /* We only do error checking on the texture names */
    for (i = 0; i < n; i++) {
       struct gl_texture_object *t;
       if (texName[i] == 0) {
@@ -1217,21 +1215,6 @@ _mesa_AreTexturesResident(GLsizei n, const GLuint *texName,
          _mesa_error(ctx, GL_INVALID_VALUE, "glAreTexturesResident");
          return GL_FALSE;
       }
-      if (!ctx->Driver.IsTextureResident ||
-          ctx->Driver.IsTextureResident(ctx, t)) {
-         /* The texture is resident */
-         if (!allResident)
-            residences[i] = GL_TRUE;
-      }
-      else {
-         /* The texture is not resident */
-         if (allResident) {
-            allResident = GL_FALSE;
-            for (j = 0; j < i; j++)
-               residences[j] = GL_TRUE;
-         }
-         residences[i] = GL_FALSE;
-      }
    }
    
    return allResident;
diff --git a/mesalib/src/mesa/main/texparam.c b/mesalib/src/mesa/main/texparam.c
index f4ec63388..0f92a5b98 100644
--- a/mesalib/src/mesa/main/texparam.c
+++ b/mesalib/src/mesa/main/texparam.c
@@ -1114,8 +1114,7 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
          }
          break;
       case GL_TEXTURE_RESIDENT:
-         *params = ctx->Driver.IsTextureResident ?
-            ctx->Driver.IsTextureResident(ctx, obj) : 1.0F;
+         *params = 1.0F;
          break;
       case GL_TEXTURE_PRIORITY:
          *params = obj->Priority;
@@ -1261,8 +1260,7 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
          }
          break;;
       case GL_TEXTURE_RESIDENT:
-         *params = ctx->Driver.IsTextureResident ?
-            ctx->Driver.IsTextureResident(ctx, obj) : 1;
+         *params = 1;
          break;;
       case GL_TEXTURE_PRIORITY:
          *params = FLOAT_TO_INT(obj->Priority);
diff --git a/mesalib/src/mesa/main/texstore.c b/mesalib/src/mesa/main/texstore.c
index 37fea2156..a57a38e12 100644
--- a/mesalib/src/mesa/main/texstore.c
+++ b/mesalib/src/mesa/main/texstore.c
@@ -954,8 +954,7 @@ memcpy_texture(struct gl_context *ctx,
          GLubyte *dstImage = dstSlices[dstZoffset + img]
             + dstYoffset * dstRowStride
             + dstXoffset * texelBytes;
-         ctx->Driver.TextureMemCpy(dstImage, srcImage,
-                                   bytesPerRow * srcHeight);
+         memcpy(dstImage, srcImage, bytesPerRow * srcHeight);
          srcImage += srcImageStride;
       }
    }
@@ -968,7 +967,7 @@ memcpy_texture(struct gl_context *ctx,
             + dstYoffset * dstRowStride
             + dstXoffset * texelBytes;
          for (row = 0; row < srcHeight; row++) {
-            ctx->Driver.TextureMemCpy(dstRow, srcRow, bytesPerRow);
+            memcpy(dstRow, srcRow, bytesPerRow);
             dstRow += dstRowStride;
             srcRow += srcRowStride;
          }
@@ -3569,6 +3568,8 @@ _mesa_texstore_rgba_int8(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_INT8);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
@@ -3639,6 +3640,8 @@ _mesa_texstore_rgba_int16(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_INT16);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
@@ -3709,6 +3712,8 @@ _mesa_texstore_rgba_int32(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_INT32);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
@@ -3779,6 +3784,8 @@ _mesa_texstore_rgba_uint8(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_UINT8);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
@@ -3847,6 +3854,8 @@ _mesa_texstore_rgba_uint16(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_UINT16);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
@@ -3915,6 +3924,8 @@ _mesa_texstore_rgba_uint32(TEXSTORE_PARAMS)
           dstFormat == MESA_FORMAT_LUMINANCE_ALPHA_UINT32);
    ASSERT(baseInternalFormat == GL_RGBA ||
           baseInternalFormat == GL_RGB ||
+          baseInternalFormat == GL_RG ||
+          baseInternalFormat == GL_RED ||
           baseInternalFormat == GL_ALPHA ||
           baseInternalFormat == GL_LUMINANCE ||
           baseInternalFormat == GL_LUMINANCE_ALPHA ||
diff --git a/mesalib/src/mesa/sources.mak b/mesalib/src/mesa/sources.mak
index 1f109fb36..a3ce85535 100644
--- a/mesalib/src/mesa/sources.mak
+++ b/mesalib/src/mesa/sources.mak
@@ -42,6 +42,7 @@ MAIN_SOURCES = \
 	main/ffvertex_prog.c \
 	main/fog.c \
 	main/formats.c \
+	main/format_pack.c \
 	main/format_unpack.c \
 	main/framebuffer.c \
 	main/get.c \
@@ -124,7 +125,6 @@ MATH_XFORM_SOURCES = \
 SWRAST_SOURCES = \
 	swrast/s_aaline.c \
 	swrast/s_aatriangle.c \
-	swrast/s_accum.c \
 	swrast/s_alpha.c \
 	swrast/s_atifragshader.c \
 	swrast/s_bitmap.c \
@@ -142,6 +142,7 @@ SWRAST_SOURCES = \
 	swrast/s_logic.c \
 	swrast/s_masking.c \
 	swrast/s_points.c \
+	swrast/s_renderbuffer.c \
 	swrast/s_span.c \
 	swrast/s_stencil.c \
 	swrast/s_texcombine.c \
@@ -208,7 +209,6 @@ STATETRACKER_SOURCES = \
 	state_tracker/st_atom_stipple.c \
 	state_tracker/st_atom_texture.c \
 	state_tracker/st_atom_viewport.c \
-	state_tracker/st_cb_accum.c \
 	state_tracker/st_cb_bitmap.c \
 	state_tracker/st_cb_blit.c \
 	state_tracker/st_cb_bufferobjects.c \
diff --git a/mesalib/src/mesa/state_tracker/st_cb_accum.c b/mesalib/src/mesa/state_tracker/st_cb_accum.c
deleted file mode 100644
index 3e3659d15..000000000
--- a/mesalib/src/mesa/state_tracker/st_cb_accum.c
+++ /dev/null
@@ -1,347 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Brian Paul
-  */
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/mfeatures.h"
-
-#include "st_debug.h"
-#include "st_context.h"
-#include "st_cb_accum.h"
-#include "st_cb_fbo.h"
-#include "st_texture.h"
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "util/u_format.h"
-#include "util/u_inlines.h"
-#include "util/u_tile.h"
-
-
-#if FEATURE_accum
-
-/**
- * For hardware that supports deep color buffers, we could accelerate
- * most/all the accum operations with blending/texturing.
- * For now, just use the get/put_tile() functions and do things in software.
- */
-
-
-void
-st_clear_accum_buffer(struct gl_context *ctx, struct gl_renderbuffer *rb)
-{
-   struct st_renderbuffer *acc_strb = st_renderbuffer(rb);
-   const GLint xpos = ctx->DrawBuffer->_Xmin;
-   const GLint ypos = ctx->DrawBuffer->_Ymin;
-   const GLint width = ctx->DrawBuffer->_Xmax - xpos;
-   const GLint height = ctx->DrawBuffer->_Ymax - ypos;
-   size_t stride = acc_strb->stride;
-   GLubyte *data = acc_strb->data;
-
-   if(!data)
-      return;
-   
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         GLshort r = FLOAT_TO_SHORT(ctx->Accum.ClearColor[0]);
-         GLshort g = FLOAT_TO_SHORT(ctx->Accum.ClearColor[1]);
-         GLshort b = FLOAT_TO_SHORT(ctx->Accum.ClearColor[2]);
-         GLshort a = FLOAT_TO_SHORT(ctx->Accum.ClearColor[3]);
-         int i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *dst = (GLshort *) (data + (ypos + i) * stride + xpos * 8);
-            for (j = 0; j < width; j++) {
-               dst[0] = r;
-               dst[1] = g;
-               dst[2] = b;
-               dst[3] = a;
-               dst += 4;
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(ctx, "unexpected format in st_clear_accum_buffer()");
-   }
-}
-
-
-/** For ADD/MULT */
-static void
-accum_mad(struct gl_context *ctx, GLfloat scale, GLfloat bias,
-          GLint xpos, GLint ypos, GLint width, GLint height,
-          struct st_renderbuffer *acc_strb)
-{
-   size_t stride = acc_strb->stride;
-   GLubyte *data = acc_strb->data;
-
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         int i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) (data + (ypos + i) * stride + xpos * 8);
-            for (j = 0; j < width * 4; j++) {
-               float val = SHORT_TO_FLOAT(*acc) * scale + bias;
-               *acc++ = FLOAT_TO_SHORT(val);
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()");
-   }
-}
-
-
-static void
-accum_accum(struct st_context *st, GLfloat value,
-            GLint xpos, GLint ypos, GLint width, GLint height,
-            struct st_renderbuffer *acc_strb,
-            struct st_renderbuffer *color_strb)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_transfer *color_trans;
-   size_t stride = acc_strb->stride;
-   GLubyte *data = acc_strb->data;
-   GLfloat *buf;
-
-   if (ST_DEBUG & DEBUG_FALLBACK)
-      debug_printf("%s: fallback processing\n", __FUNCTION__);
-
-   color_trans = pipe_get_transfer(st->pipe,
-                                   color_strb->texture,
-                                   0, 0,
-                                   PIPE_TRANSFER_READ, xpos, ypos,
-                                   width, height);
-
-   buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   pipe_get_tile_rgba_format(pipe, color_trans, 0, 0, width, height,
-                             util_format_linear(color_strb->texture->format),
-                             buf);
-
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         const GLfloat *color = buf;
-         int i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) (data + (ypos + i) * stride + xpos * 8);
-            for (j = 0; j < width * 4; j++) {
-               float val = *color++ * value;
-               *acc++ += FLOAT_TO_SHORT(val);
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()");
-   }
-
-   free(buf);
-   pipe->transfer_destroy(pipe, color_trans);
-}
-
-
-static void
-accum_load(struct st_context *st, GLfloat value,
-           GLint xpos, GLint ypos, GLint width, GLint height,
-           struct st_renderbuffer *acc_strb,
-           struct st_renderbuffer *color_strb)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_transfer *color_trans;
-   size_t stride = acc_strb->stride;
-   GLubyte *data = acc_strb->data;
-   GLfloat *buf;
-
-   if (ST_DEBUG & DEBUG_FALLBACK)
-      debug_printf("%s: fallback processing\n", __FUNCTION__);
-
-   color_trans = pipe_get_transfer(st->pipe, color_strb->texture,
-                                   0, 0,
-                                   PIPE_TRANSFER_READ, xpos, ypos,
-                                   width, height);
-
-   buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   pipe_get_tile_rgba_format(pipe, color_trans, 0, 0, width, height,
-                             util_format_linear(color_strb->texture->format),
-                             buf);
-
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         const GLfloat *color = buf;
-         int i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) (data + (ypos + i) * stride + xpos * 8);
-            for (j = 0; j < width * 4; j++) {
-               float val = *color++ * value;
-               *acc++ = FLOAT_TO_SHORT(val);
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()");
-   }
-
-   free(buf);
-   pipe->transfer_destroy(pipe, color_trans);
-}
-
-
-static void
-accum_return(struct gl_context *ctx, GLfloat value,
-             GLint xpos, GLint ypos, GLint width, GLint height,
-             struct st_renderbuffer *acc_strb,
-             struct st_renderbuffer *color_strb)
-{
-   struct pipe_context *pipe = st_context(ctx)->pipe;
-   const GLubyte *colormask = ctx->Color.ColorMask[0];
-   enum pipe_transfer_usage usage;
-   struct pipe_transfer *color_trans;
-   size_t stride = acc_strb->stride;
-   const GLubyte *data = acc_strb->data;
-   GLfloat *buf;
-   enum pipe_format format = util_format_linear(color_strb->texture->format);
-
-   if (ST_DEBUG & DEBUG_FALLBACK)
-      debug_printf("%s: fallback processing\n", __FUNCTION__);
-
-   buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   if (!colormask[0] || !colormask[1] || !colormask[2] || !colormask[3])
-      usage = PIPE_TRANSFER_READ_WRITE;
-   else
-      usage = PIPE_TRANSFER_WRITE;
-
-   color_trans = pipe_get_transfer(pipe,
-                                   color_strb->texture, 0, 0,
-                                   usage,
-                                   xpos, ypos,
-                                   width, height);
-
-   if (usage & PIPE_TRANSFER_READ)
-      pipe_get_tile_rgba_format(pipe, color_trans, 0, 0, width, height,
-                                format, buf);
-
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         GLfloat *color = buf;
-         int i, j, ch;
-         for (i = 0; i < height; i++) {
-            const GLshort *acc = (const GLshort *) (data + (ypos + i) * stride + xpos * 8);
-            for (j = 0; j < width; j++) {
-               for (ch = 0; ch < 4; ch++) {
-                  if (colormask[ch]) {
-                     GLfloat val = SHORT_TO_FLOAT(*acc * value);
-                     *color = CLAMP(val, 0.0f, 1.0f);
-                  }
-                  else {
-                     /* No change */
-                  }
-                  ++acc;
-                  ++color;
-               }
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()");
-   }
-
-   pipe_put_tile_rgba_format(pipe, color_trans, 0, 0, width, height,
-                             format, buf);
-
-   free(buf);
-   pipe->transfer_destroy(pipe, color_trans);
-}
-
-
-static void
-st_Accum(struct gl_context *ctx, GLenum op, GLfloat value)
-{
-   struct st_context *st = st_context(ctx);
-   struct st_renderbuffer *acc_strb
-     = st_renderbuffer(ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
-   struct st_renderbuffer *color_strb
-      = st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
-
-   const GLint xpos = ctx->DrawBuffer->_Xmin;
-   const GLint ypos = ctx->DrawBuffer->_Ymin;
-   const GLint width = ctx->DrawBuffer->_Xmax - xpos;
-   const GLint height = ctx->DrawBuffer->_Ymax - ypos;
-
-   if(!acc_strb->data)
-      return;
-   
-   switch (op) {
-   case GL_ADD:
-      if (value != 0.0F) {
-         accum_mad(ctx, 1.0, value, xpos, ypos, width, height, acc_strb);
-      }
-      break;
-   case GL_MULT:
-      if (value != 1.0F) {
-         accum_mad(ctx, value, 0.0, xpos, ypos, width, height, acc_strb);
-      }
-      break;
-   case GL_ACCUM:
-      if (value != 0.0F) {
-         accum_accum(st, value, xpos, ypos, width, height, acc_strb, color_strb);
-      }
-      break;
-   case GL_LOAD:
-      accum_load(st, value, xpos, ypos, width, height, acc_strb, color_strb);
-      break;
-   case GL_RETURN:
-      accum_return(ctx, value, xpos, ypos, width, height, acc_strb, color_strb);
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-
-void st_init_accum_functions(struct dd_function_table *functions)
-{
-   functions->Accum = st_Accum;
-}
-
-#endif /* FEATURE_accum */
diff --git a/mesalib/src/mesa/state_tracker/st_cb_accum.h b/mesalib/src/mesa/state_tracker/st_cb_accum.h
deleted file mode 100644
index 90feddd6f..000000000
--- a/mesalib/src/mesa/state_tracker/st_cb_accum.h
+++ /dev/null
@@ -1,63 +0,0 @@
-/**************************************************************************

- * 

- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.

- * All Rights Reserved.

- * 

- * Permission is hereby granted, free of charge, to any person obtaining a

- * copy of this software and associated documentation files (the

- * "Software"), to deal in the Software without restriction, including

- * without limitation the rights to use, copy, modify, merge, publish,

- * distribute, sub license, and/or sell copies of the Software, and to

- * permit persons to whom the Software is furnished to do so, subject to

- * the following conditions:

- * 

- * The above copyright notice and this permission notice (including the

- * next paragraph) shall be included in all copies or substantial portions

- * of the Software.

- * 

- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.

- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR

- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,

- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE

- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

- * 

- **************************************************************************/

-

-

-#ifndef ST_CB_ACCUM_H

-#define ST_CB_ACCUM_H

-

-

-#include "main/mfeatures.h"

-

-struct dd_function_table;

-struct gl_context;

-struct gl_renderbuffer;

-

-#if FEATURE_accum

-

-extern void

-st_clear_accum_buffer(struct gl_context *ctx, struct gl_renderbuffer *rb);

-

-extern void st_init_accum_functions(struct dd_function_table *functions);

-

-#else

-

-#include "main/compiler.h"

-

-static INLINE void

-st_clear_accum_buffer(struct gl_context *ctx, struct gl_renderbuffer *rb)

-{

-   ASSERT_NO_FEATURE();

-}

-

-static INLINE void

-st_init_accum_functions(struct dd_function_table *functions)

-{

-}

-

-#endif /* FEATURE_accum */

-

-#endif /* ST_CB_ACCUM_H */

diff --git a/mesalib/src/mesa/state_tracker/st_cb_clear.c b/mesalib/src/mesa/state_tracker/st_cb_clear.c
index 89273e28e..61d98aeb9 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_clear.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_clear.c
@@ -34,12 +34,12 @@
   */
 
 #include "main/glheader.h"
+#include "main/accum.h"
 #include "main/formats.h"
 #include "main/macros.h"
 #include "program/prog_instruction.h"
 #include "st_context.h"
 #include "st_atom.h"
-#include "st_cb_accum.h"
 #include "st_cb_clear.h"
 #include "st_cb_fbo.h"
 #include "st_format.h"
@@ -599,8 +599,7 @@ st_Clear(struct gl_context *ctx, GLbitfield mask)
                       ctx->Depth.Clear, ctx->Stencil.Clear);
    }
    if (mask & BUFFER_BIT_ACCUM)
-      st_clear_accum_buffer(ctx,
-                            ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
+      _mesa_clear_accum_buffer(ctx);
 }
 
 
diff --git a/mesalib/src/mesa/state_tracker/st_cb_fbo.c b/mesalib/src/mesa/state_tracker/st_cb_fbo.c
index 2a60ed4df..ec40a2b70 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_fbo.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_fbo.c
@@ -667,6 +667,22 @@ st_MapRenderbuffer(struct gl_context *ctx,
    unsigned usage;
    GLuint y2;
 
+   if (strb->software) {
+      /* software-allocated renderbuffer (probably an accum buffer) */
+      GLubyte *map = (GLubyte *) strb->data;
+      if (strb->data) {
+         map += strb->stride * y;
+         map += util_format_get_blocksize(strb->format) * x;
+         *mapOut = map;
+         *rowStrideOut = strb->stride;
+      }
+      else {
+         *mapOut = NULL;
+         *rowStrideOut = 0;
+      }
+      return;
+   }
+
    usage = 0x0;
    if (mode & GL_MAP_READ_BIT)
       usage |= PIPE_TRANSFER_READ;
@@ -716,6 +732,11 @@ st_UnmapRenderbuffer(struct gl_context *ctx,
    struct st_renderbuffer *strb = st_renderbuffer(rb);
    struct pipe_context *pipe = st->pipe;
 
+   if (strb->software) {
+      /* software-allocated renderbuffer (probably an accum buffer) */
+      return;
+   }
+
    pipe_transfer_unmap(pipe, strb->transfer);
    pipe->transfer_destroy(pipe, strb->transfer);
    strb->transfer = NULL;
diff --git a/mesalib/src/mesa/state_tracker/st_cb_texture.c b/mesalib/src/mesa/state_tracker/st_cb_texture.c
index c58a9df56..52f654d7d 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_texture.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_texture.c
@@ -214,56 +214,6 @@ st_UnmapTextureImage(struct gl_context *ctx,
 
 
 /**
- * From linux kernel i386 header files, copes with odd sizes better
- * than COPY_DWORDS would:
- * XXX Put this in src/mesa/main/imports.h ???
- */
-#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
-static INLINE void *
-__memcpy(void *to, const void *from, size_t n)
-{
-   int d0, d1, d2;
-   __asm__ __volatile__("rep ; movsl\n\t"
-                        "testb $2,%b4\n\t"
-                        "je 1f\n\t"
-                        "movsw\n"
-                        "1:\ttestb $1,%b4\n\t"
-                        "je 2f\n\t"
-                        "movsb\n" "2:":"=&c"(d0), "=&D"(d1), "=&S"(d2)
-                        :"0"(n / 4), "q"(n), "1"((long) to), "2"((long) from)
-                        :"memory");
-   return (to);
-}
-#else
-#define __memcpy(a,b,c) memcpy(a,b,c)
-#endif
-
-
-/**
- * The system memcpy (at least on ubuntu 5.10) has problems copying
- * to agp (writecombined) memory from a source which isn't 64-byte
- * aligned - there is a 4x performance falloff.
- *
- * The x86 __memcpy is immune to this but is slightly slower
- * (10%-ish) than the system memcpy.
- *
- * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
- * isn't much faster than x86_memcpy for agp copies.
- * 
- * TODO: switch dynamically.
- */
-static void *
-do_memcpy(void *dest, const void *src, size_t n)
-{
-   if ((((unsigned long) src) & 63) || (((unsigned long) dest) & 63)) {
-      return __memcpy(dest, src, n);
-   }
-   else
-      return memcpy(dest, src, n);
-}
-
-
-/**
  * Return default texture resource binding bitmask for the given format.
  */
 static GLuint
@@ -1420,11 +1370,13 @@ st_copy_texsubimage(struct gl_context *ctx,
    struct pipe_context *pipe = st->pipe;
    struct pipe_screen *screen = pipe->screen;
    enum pipe_format dest_format, src_format;
-   GLboolean use_fallback = GL_TRUE;
    GLboolean matching_base_formats;
    GLuint format_writemask, sample_count;
    struct pipe_surface *dest_surface = NULL;
    GLboolean do_flip = (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP);
+   struct pipe_surface surf_tmpl;
+   unsigned int dst_usage;
+   GLint srcY0, srcY1;
 
    /* make sure finalize_textures has been called? 
     */
@@ -1472,99 +1424,105 @@ st_copy_texsubimage(struct gl_context *ctx,
    matching_base_formats =
       (_mesa_get_format_base_format(strb->Base.Format) ==
        _mesa_get_format_base_format(texImage->TexFormat));
-   format_writemask = compatible_src_dst_formats(ctx, &strb->Base, texImage);
 
-   if (ctx->_ImageTransferState == 0x0) {
+   if (ctx->_ImageTransferState) {
+      goto fallback;
+   }
+
+   if (matching_base_formats &&
+       src_format == dest_format &&
+       !do_flip) {
+      /* use surface_copy() / blit */
+      struct pipe_box src_box;
+      u_box_2d_zslice(srcX, srcY, strb->surface->u.tex.first_layer,
+                      width, height, &src_box);
+
+      /* for resource_copy_region(), y=0=top, always */
+      pipe->resource_copy_region(pipe,
+                                 /* dest */
+                                 stImage->pt,
+                                 stImage->base.Level,
+                                 destX, destY, destZ + stImage->base.Face,
+                                 /* src */
+                                 strb->texture,
+                                 strb->surface->u.tex.level,
+                                 &src_box);
+      return;
+   }
 
-      if (matching_base_formats &&
-          src_format == dest_format &&
-          !do_flip)
-      {
-         /* use surface_copy() / blit */
-         struct pipe_box src_box;
-         u_box_2d_zslice(srcX, srcY, strb->surface->u.tex.first_layer,
-                         width, height, &src_box);
-
-         /* for resource_copy_region(), y=0=top, always */
-         pipe->resource_copy_region(pipe,
-                                    /* dest */
-                                    stImage->pt,
-                                    stImage->base.Level,
-                                    destX, destY, destZ + stImage->base.Face,
-                                    /* src */
-                                    strb->texture,
-                                    strb->surface->u.tex.level,
-                                    &src_box);
-         use_fallback = GL_FALSE;
-      }
-      else if (format_writemask &&
-               texBaseFormat != GL_DEPTH_COMPONENT &&
-               texBaseFormat != GL_DEPTH_STENCIL &&
-               screen->is_format_supported(screen, src_format,
-                                           PIPE_TEXTURE_2D, sample_count,
-                                           PIPE_BIND_SAMPLER_VIEW) &&
-               screen->is_format_supported(screen, dest_format,
-                                           PIPE_TEXTURE_2D, 0,
-                                           PIPE_BIND_RENDER_TARGET)) {
-         /* draw textured quad to do the copy */
-         GLint srcY0, srcY1;
-         struct pipe_surface surf_tmpl;
-         memset(&surf_tmpl, 0, sizeof(surf_tmpl));
-         surf_tmpl.format = util_format_linear(stImage->pt->format);
-         surf_tmpl.usage = PIPE_BIND_RENDER_TARGET;
-         surf_tmpl.u.tex.level = stImage->base.Level;
-         surf_tmpl.u.tex.first_layer = stImage->base.Face + destZ;
-         surf_tmpl.u.tex.last_layer = stImage->base.Face + destZ;
-
-         dest_surface = pipe->create_surface(pipe, stImage->pt,
-                                             &surf_tmpl);
-
-         if (do_flip) {
-            srcY1 = strb->Base.Height - srcY - height;
-            srcY0 = srcY1 + height;
-         }
-         else {
-            srcY0 = srcY;
-            srcY1 = srcY0 + height;
-         }
+   if (texBaseFormat == GL_DEPTH_STENCIL) {
+      goto fallback;
+   }
 
-         /* Disable conditional rendering. */
-         if (st->render_condition) {
-            pipe->render_condition(pipe, NULL, 0);
-         }
+   if (texBaseFormat == GL_DEPTH_COMPONENT) {
+      format_writemask = TGSI_WRITEMASK_XYZW;
+      dst_usage = PIPE_BIND_DEPTH_STENCIL;
+   }
+   else {
+      format_writemask = compatible_src_dst_formats(ctx, &strb->Base, texImage);
+      dst_usage = PIPE_BIND_RENDER_TARGET;
+   }
 
-         util_blit_pixels_writemask(st->blit,
-                                    strb->texture,
-                                    strb->surface->u.tex.level,
-                                    srcX, srcY0,
-                                    srcX + width, srcY1,
-                                    strb->surface->u.tex.first_layer,
-                                    dest_surface,
-                                    destX, destY,
-                                    destX + width, destY + height,
-                                    0.0, PIPE_TEX_MIPFILTER_NEAREST,
-                                    format_writemask);
-
-         /* Restore conditional rendering state. */
-         if (st->render_condition) {
-            pipe->render_condition(pipe, st->render_condition,
-                                   st->condition_mode);
-         }
+   if (!format_writemask ||
+       !screen->is_format_supported(screen, src_format,
+                                    PIPE_TEXTURE_2D, sample_count,
+                                    PIPE_BIND_SAMPLER_VIEW) ||
+       !screen->is_format_supported(screen, dest_format,
+                                    PIPE_TEXTURE_2D, 0,
+                                    dst_usage)) {
+      goto fallback;
+   }
 
-         use_fallback = GL_FALSE;
-      }
+   if (do_flip) {
+      srcY1 = strb->Base.Height - srcY - height;
+      srcY0 = srcY1 + height;
+   }
+   else {
+      srcY0 = srcY;
+      srcY1 = srcY0 + height;
+   }
+
+   /* Disable conditional rendering. */
+   if (st->render_condition) {
+      pipe->render_condition(pipe, NULL, 0);
+   }
+
+   memset(&surf_tmpl, 0, sizeof(surf_tmpl));
+   surf_tmpl.format = util_format_linear(stImage->pt->format);
+   surf_tmpl.usage = dst_usage;
+   surf_tmpl.u.tex.level = stImage->base.Level;
+   surf_tmpl.u.tex.first_layer = stImage->base.Face + destZ;
+   surf_tmpl.u.tex.last_layer = stImage->base.Face + destZ;
+
+   dest_surface = pipe->create_surface(pipe, stImage->pt,
+                                       &surf_tmpl);
+   util_blit_pixels_writemask(st->blit,
+                              strb->texture,
+                              strb->surface->u.tex.level,
+                              srcX, srcY0,
+                              srcX + width, srcY1,
+                              strb->surface->u.tex.first_layer,
+                              dest_surface,
+                              destX, destY,
+                              destX + width, destY + height,
+                              0.0, PIPE_TEX_MIPFILTER_NEAREST,
+                              format_writemask);
+   pipe_surface_reference(&dest_surface, NULL);
 
-      if (dest_surface)
-         pipe_surface_reference(&dest_surface, NULL);
+   /* Restore conditional rendering state. */
+   if (st->render_condition) {
+      pipe->render_condition(pipe, st->render_condition,
+                             st->condition_mode);
    }
 
-   if (use_fallback) {
+   return;
+
+fallback:
       /* software fallback */
       fallback_copy_texsubimage(ctx, target, level,
                                 strb, stImage, texBaseFormat,
                                 destX, destY, destZ,
                                 srcX, srcY, width, height);
-   }
 }
 
 
@@ -1937,8 +1895,6 @@ st_init_texture_functions(struct dd_function_table *functions)
    functions->MapTextureImage = st_MapTextureImage;
    functions->UnmapTextureImage = st_UnmapTextureImage;
 
-   functions->TextureMemCpy = do_memcpy;
-
    /* XXX Temporary until we can query pipe's texture sizes */
    functions->TestProxyTexImage = _mesa_test_proxy_teximage;
 
diff --git a/mesalib/src/mesa/state_tracker/st_context.c b/mesalib/src/mesa/state_tracker/st_context.c
index a1817720d..dc1d33f1d 100644
--- a/mesalib/src/mesa/state_tracker/st_context.c
+++ b/mesalib/src/mesa/state_tracker/st_context.c
@@ -26,6 +26,7 @@
  **************************************************************************/
 
 #include "main/imports.h"
+#include "main/accum.h"
 #include "main/context.h"
 #include "main/samplerobj.h"
 #include "main/shaderobj.h"
@@ -34,7 +35,6 @@
 #include "glapi/glapi.h"
 #include "st_context.h"
 #include "st_debug.h"
-#include "st_cb_accum.h"
 #include "st_cb_bitmap.h"
 #include "st_cb_blit.h"
 #include "st_cb_bufferobjects.h"
@@ -276,7 +276,8 @@ void st_init_driver_functions(struct dd_function_table *functions)
    _mesa_init_shader_object_functions(functions);
    _mesa_init_sampler_object_functions(functions);
 
-   st_init_accum_functions(functions);
+   functions->Accum = _mesa_accum;
+
    st_init_blit_functions(functions);
    st_init_bufferobject_functions(functions);
    st_init_clear_functions(functions);
diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c
index 37fb3e7fd..9e39729e9 100644
--- a/mesalib/src/mesa/state_tracker/st_extensions.c
+++ b/mesalib/src/mesa/state_tracker/st_extensions.c
@@ -210,12 +210,10 @@ void st_init_limits(struct st_context *st)
          options->MaxUnrollIterations = MIN2(screen->get_shader_param(screen, sh, PIPE_SHADER_CAP_MAX_INSTRUCTIONS), 65536);
    }
 
-   /* PIPE_CAP_MAX_FS_INPUTS specifies the number of COLORn + GENERICn inputs
-    * and is set in MaxNativeAttribs. It's always 2 colors + N generic
-    * attributes. The GLSL compiler never uses COLORn for varyings, so we
-    * subtract the 2 colors to get the maximum number of varyings (generic
-    * attributes) supported by a driver. */
-   c->MaxVarying = screen->get_shader_param(screen, PIPE_SHADER_FRAGMENT, PIPE_SHADER_CAP_MAX_INPUTS) - 2;
+   /* PIPE_SHADER_CAP_MAX_INPUTS for the FS specifies the maximum number
+    * of inputs. It's always 2 colors + N generic inputs. */
+   c->MaxVarying = screen->get_shader_param(screen, PIPE_SHADER_FRAGMENT,
+                                            PIPE_SHADER_CAP_MAX_INPUTS);
    c->MaxVarying = MIN2(c->MaxVarying, MAX_VARYING);
 
    c->MinProgramTexelOffset = screen->get_param(screen, PIPE_CAP_MIN_TEXEL_OFFSET);
diff --git a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 59b5ffd4c..6cc655d70 100644
--- a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -3514,25 +3514,23 @@ glsl_to_tgsi_visitor::eliminate_dead_code_advanced(void)
          break;
 
       case TGSI_OPCODE_ENDIF:
-         --level;
-         break;
-
       case TGSI_OPCODE_ELSE:
-         /* Clear all channels written inside the preceding if block from the
-          * write array, but leave those that were not touched.
-          *
-          * FIXME: This destroys opportunities to remove dead code inside of
-          * IF blocks that are followed by an ELSE block.
+         /* Promote the recorded level all channels written inside the preceding
+          * if or else block to the level above the if/else block.
           */
          for (int r = 0; r < this->next_temp; r++) {
             for (int c = 0; c < 4; c++) {
                if (!writes[4 * r + c])
         	         continue;
 
-               if (write_level[4 * r + c] >= level)
-        	         writes[4 * r + c] = NULL;
+               if (write_level[4 * r + c] == level)
+        	         write_level[4 * r + c] = level-1;
             }
          }
+
+         if(inst->op == TGSI_OPCODE_ENDIF)
+            --level;
+         
          break;
 
       case TGSI_OPCODE_IF:
diff --git a/mesalib/src/mesa/state_tracker/st_manager.c b/mesalib/src/mesa/state_tracker/st_manager.c
index d5228d387..55699e721 100644
--- a/mesalib/src/mesa/state_tracker/st_manager.c
+++ b/mesalib/src/mesa/state_tracker/st_manager.c
@@ -404,54 +404,6 @@ st_visual_to_context_mode(const struct st_visual *visual,
 }
 
 /**
- * Determine the default draw or read buffer from a visual.
- */
-static void
-st_visual_to_default_buffer(const struct st_visual *visual,
-                            GLenum *buffer, GLint *index)
-{
-   enum st_attachment_type statt;
-   GLenum buf;
-   gl_buffer_index idx;
-
-   statt = visual->render_buffer;
-   /* do nothing if an invalid render buffer is specified */
-   if (statt == ST_ATTACHMENT_INVALID ||
-       !st_visual_have_buffers(visual, 1 << statt))
-      return;
-
-   switch (statt) {
-   case ST_ATTACHMENT_FRONT_LEFT:
-      buf = GL_FRONT_LEFT;
-      idx = BUFFER_FRONT_LEFT;
-      break;
-   case ST_ATTACHMENT_BACK_LEFT:
-      buf = GL_BACK_LEFT;
-      idx = BUFFER_BACK_LEFT;
-      break;
-   case ST_ATTACHMENT_FRONT_RIGHT:
-      buf = GL_FRONT_RIGHT;
-      idx = BUFFER_FRONT_RIGHT;
-      break;
-   case ST_ATTACHMENT_BACK_RIGHT:
-      buf = GL_BACK_RIGHT;
-      idx = BUFFER_BACK_RIGHT;
-      break;
-   default:
-      buf = GL_NONE;
-      idx = BUFFER_COUNT;
-      break;
-   }
-
-   if (buf != GL_NONE) {
-      if (buffer)
-         *buffer = buf;
-      if (index)
-         *index = idx;
-   }
-}
-
-/**
  * Create a framebuffer from a manager interface.
  */
 static struct st_framebuffer *
@@ -471,12 +423,6 @@ st_framebuffer_create(struct st_framebuffer_iface *stfbi)
    st_visual_to_context_mode(stfbi->visual, &mode);
    _mesa_initialize_window_framebuffer(&stfb->Base, &mode);
 
-   /* modify the draw/read buffers of the fb */
-   st_visual_to_default_buffer(stfbi->visual, &stfb->Base.ColorDrawBuffer[0],
-         &stfb->Base._ColorDrawBufferIndexes[0]);
-   st_visual_to_default_buffer(stfbi->visual, &stfb->Base.ColorReadBuffer,
-         &stfb->Base._ColorReadBufferIndex);
-
    stfb->iface = stfbi;
    stfb->iface_stamp = p_atomic_read(&stfbi->stamp) - 1;
 
@@ -776,16 +722,6 @@ st_api_make_current(struct st_api *stapi, struct st_context_iface *stctxi,
          if (stread != stdraw)
             st_framebuffer_validate(stread, st);
 
-         /* modify the draw/read buffers of the context */
-         if (stdraw->iface) {
-            st_visual_to_default_buffer(stdraw->iface->visual,
-                  &st->ctx->Color.DrawBuffer[0], NULL);
-         }
-         if (stread->iface) {
-            st_visual_to_default_buffer(stread->iface->visual,
-                  &st->ctx->Pixel.ReadBuffer, NULL);
-         }
-
          ret = _mesa_make_current(st->ctx, &stdraw->Base, &stread->Base);
 
          st->draw_stamp = stdraw->stamp - 1;
diff --git a/mesalib/src/mesa/state_tracker/st_program.c b/mesalib/src/mesa/state_tracker/st_program.c
index 768da5114..04d3ef60f 100644
--- a/mesalib/src/mesa/state_tracker/st_program.c
+++ b/mesalib/src/mesa/state_tracker/st_program.c
@@ -649,6 +649,25 @@ st_translate_fragment_program(struct st_context *st,
       if (write_all == GL_TRUE)
          ureg_property_fs_color0_writes_all_cbufs(ureg, 1);
 
+      if (stfp->Base.FragDepthLayout != FRAG_DEPTH_LAYOUT_NONE) {
+         switch (stfp->Base.FragDepthLayout) {
+         case FRAG_DEPTH_LAYOUT_ANY:
+            ureg_property_fs_depth_layout(ureg, TGSI_FS_DEPTH_LAYOUT_ANY);
+            break;
+         case FRAG_DEPTH_LAYOUT_GREATER:
+            ureg_property_fs_depth_layout(ureg, TGSI_FS_DEPTH_LAYOUT_GREATER);
+            break;
+         case FRAG_DEPTH_LAYOUT_LESS:
+            ureg_property_fs_depth_layout(ureg, TGSI_FS_DEPTH_LAYOUT_LESS);
+            break;
+         case FRAG_DEPTH_LAYOUT_UNCHANGED:
+            ureg_property_fs_depth_layout(ureg, TGSI_FS_DEPTH_LAYOUT_UNCHANGED);
+            break;
+         default:
+            assert(0);
+         }
+      }
+
       if (stfp->glsl_to_tgsi)
          st_translate_program(st->ctx,
                               TGSI_PROCESSOR_FRAGMENT,
diff --git a/mesalib/src/mesa/swrast/s_accum.c b/mesalib/src/mesa/swrast/s_accum.c
deleted file mode 100644
index 0ec907d79..000000000
--- a/mesalib/src/mesa/swrast/s_accum.c
+++ /dev/null
@@ -1,598 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.2
- *
- * Copyright (C) 1999-2006  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/condrender.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/imports.h"
-
-#include "s_accum.h"
-#include "s_context.h"
-#include "s_masking.h"
-#include "s_span.h"
-
-
-/* XXX this would have to change for accum buffers with more or less
- * than 16 bits per color channel.
- */
-#define ACCUM_SCALE16 32767.0F
-
-
-/*
- * Accumulation buffer notes
- *
- * Normally, accumulation buffer values are GLshorts with values in
- * [-32767, 32767] which represent floating point colors in [-1, 1],
- * as defined by the OpenGL specification.
- *
- * We optimize for the common case used for full-scene antialiasing:
- *    // start with accum buffer cleared to zero
- *    glAccum(GL_LOAD, w);   // or GL_ACCUM the first image
- *    glAccum(GL_ACCUM, w);
- *    ...
- *    glAccum(GL_ACCUM, w);
- *    glAccum(GL_RETURN, 1.0);
- * That is, we start with an empty accumulation buffer and accumulate
- * n images, each with weight w = 1/n.
- * In this scenario, we can simply store unscaled integer values in
- * the accum buffer instead of scaled integers.  We'll also keep track
- * of the w value so when we do GL_RETURN we simply divide the accumulated
- * values by n (n=1/w).
- * This lets us avoid _many_ int->float->int conversions.
- */
-
-
-#if CHAN_BITS == 8
-/* enable the optimization */
-#define USE_OPTIMIZED_ACCUM  1
-#else
-#define USE_OPTIMIZED_ACCUM  0
-#endif
-
-
-/**
- * This is called when we fall out of optimized/unscaled accum buffer mode.
- * That is, we convert each unscaled accum buffer value into a scaled value
- * representing the range[-1, 1].
- */
-static void
-rescale_accum( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLfloat s = swrast->_IntegerAccumScaler * (32767.0F / CHAN_MAXF);
-
-   assert(rb);
-   assert(rb->_BaseFormat == GL_RGBA);
-   /* add other types in future? */
-   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);
-   assert(swrast->_IntegerAccumMode);
-
-   if (rb->GetPointer(ctx, rb, 0, 0)) {
-      /* directly-addressable memory */
-      GLuint y;
-      for (y = 0; y < rb->Height; y++) {
-         GLuint i;
-         GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, 0, y);
-         for (i = 0; i < 4 * rb->Width; i++) {
-            acc[i] = (GLshort) (acc[i] * s);
-         }
-      }
-   }
-   else {
-      /* use get/put row funcs */
-      GLuint y;
-      for (y = 0; y < rb->Height; y++) {
-         GLshort accRow[MAX_WIDTH * 4];
-         GLuint i;
-         rb->GetRow(ctx, rb, rb->Width, 0, y, accRow);
-         for (i = 0; i < 4 * rb->Width; i++) {
-            accRow[i] = (GLshort) (accRow[i] * s);
-         }
-         rb->PutRow(ctx, rb, rb->Width, 0, y, accRow, NULL);
-      }
-   }
-
-   swrast->_IntegerAccumMode = GL_FALSE;
-}
-
-
-
-/**
- * Clear the accumulation Buffer.
- */
-void
-_swrast_clear_accum_buffer( struct gl_context *ctx, struct gl_renderbuffer *rb )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint x, y, width, height;
-
-   /* No accumulation buffer! Not an error. */
-   if (!rb || !rb->Data)
-      return;
-
-   assert(rb->_BaseFormat == GL_RGBA);
-   /* add other types in future? */
-   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);
-
-   /* bounds, with scissor */
-   x = ctx->DrawBuffer->_Xmin;
-   y = ctx->DrawBuffer->_Ymin;
-   width = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat accScale = 32767.0;
-      GLshort clearVal[4];
-      GLuint i;
-
-      clearVal[0] = (GLshort) (ctx->Accum.ClearColor[0] * accScale);
-      clearVal[1] = (GLshort) (ctx->Accum.ClearColor[1] * accScale);
-      clearVal[2] = (GLshort) (ctx->Accum.ClearColor[2] * accScale);
-      clearVal[3] = (GLshort) (ctx->Accum.ClearColor[3] * accScale);
-
-      for (i = 0; i < height; i++) {
-         rb->PutMonoRow(ctx, rb, width, x, y + i, clearVal, NULL);
-      }
-   }
-   else {
-      /* someday support other sizes */
-   }
-
-   /* update optimized accum state vars */
-   if (ctx->Accum.ClearColor[0] == 0.0 && ctx->Accum.ClearColor[1] == 0.0 &&
-       ctx->Accum.ClearColor[2] == 0.0 && ctx->Accum.ClearColor[3] == 0.0) {
-#if USE_OPTIMIZED_ACCUM
-      swrast->_IntegerAccumMode = GL_TRUE;
-#else
-      swrast->_IntegerAccumMode = GL_FALSE;
-#endif
-      swrast->_IntegerAccumScaler = 0.0;  /* denotes empty accum buffer */
-   }
-   else {
-      swrast->_IntegerAccumMode = GL_FALSE;
-   }
-}
-
-
-static void
-accum_add(struct gl_context *ctx, GLfloat value,
-          GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-
-   assert(rb);
-
-   /* Leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLshort incr = (GLshort) (value * ACCUM_SCALE16);
-      if (rb->GetPointer(ctx, rb, 0, 0)) {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-            for (j = 0; j < 4 * width; j++) {
-               acc[j] += incr;
-            }
-         }
-      }
-      else {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort accRow[4 * MAX_WIDTH];
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);
-            for (j = 0; j < 4 * width; j++) {
-               accRow[j] += incr;
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-static void
-accum_mult(struct gl_context *ctx, GLfloat mult,
-           GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-
-   assert(rb);
-
-   /* Leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      if (rb->GetPointer(ctx, rb, 0, 0)) {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-            for (j = 0; j < 4 * width; j++) {
-               acc[j] = (GLshort) (acc[j] * mult);
-            }
-         }
-      }
-      else {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort accRow[4 * MAX_WIDTH];
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);
-            for (j = 0; j < 4 * width; j++) {
-               accRow[j] = (GLshort) (accRow[j] * mult);
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-static void
-accum_accum(struct gl_context *ctx, GLfloat value,
-            GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);
-
-   assert(rb);
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no read buffer - OK */
-      return;
-   }
-
-   /* May have to leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumScaler == 0.0 && value > 0.0 && value <= 1.0)
-      swrast->_IntegerAccumScaler = value;
-   if (swrast->_IntegerAccumMode && value != swrast->_IntegerAccumScaler)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;
-      GLshort accumRow[4 * MAX_WIDTH];
-      GLchan rgba[MAX_WIDTH][4];
-      GLint i;
-
-      for (i = 0; i < height; i++) {
-         GLshort *acc;
-         if (directAccess) {
-            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-         }
-         else {
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* read colors from color buffer */
-         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,
-                                xpos, ypos + i, CHAN_TYPE, rgba);
-
-         /* do accumulation */
-         if (swrast->_IntegerAccumMode) {
-            /* simply add integer color values into accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] += rgba[j][RCOMP];
-               acc[j * 4 + 1] += rgba[j][GCOMP];
-               acc[j * 4 + 2] += rgba[j][BCOMP];
-               acc[j * 4 + 3] += rgba[j][ACOMP];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] += (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);
-               acc[j * 4 + 1] += (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);
-               acc[j * 4 + 2] += (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);
-               acc[j * 4 + 3] += (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);
-            }
-         }
-
-         if (!directAccess) {
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-static void
-accum_load(struct gl_context *ctx, GLfloat value,
-           GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);
-
-   assert(rb);
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no read buffer - OK */
-      return;
-   }
-
-   /* This is a change to go into optimized accum buffer mode */
-   if (value > 0.0 && value <= 1.0) {
-#if USE_OPTIMIZED_ACCUM
-      swrast->_IntegerAccumMode = GL_TRUE;
-#else
-      swrast->_IntegerAccumMode = GL_FALSE;
-#endif
-      swrast->_IntegerAccumScaler = value;
-   }
-   else {
-      swrast->_IntegerAccumMode = GL_FALSE;
-      swrast->_IntegerAccumScaler = 0.0;
-   }
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;
-      GLshort accumRow[4 * MAX_WIDTH];
-      GLchan rgba[MAX_WIDTH][4];
-      GLint i;
-
-      for (i = 0; i < height; i++) {
-         GLshort *acc;
-         if (directAccess) {
-            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-         }
-         else {
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* read colors from color buffer */
-         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,
-                                xpos, ypos + i, CHAN_TYPE, rgba);
-
-         /* do load */
-         if (swrast->_IntegerAccumMode) {
-            /* just copy values in */
-            GLint j;
-            assert(swrast->_IntegerAccumScaler > 0.0);
-            assert(swrast->_IntegerAccumScaler <= 1.0);
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] = rgba[j][RCOMP];
-               acc[j * 4 + 1] = rgba[j][GCOMP];
-               acc[j * 4 + 2] = rgba[j][BCOMP];
-               acc[j * 4 + 3] = rgba[j][ACOMP];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] = (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);
-               acc[j * 4 + 1] = (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);
-               acc[j * 4 + 2] = (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);
-               acc[j * 4 + 3] = (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);
-            }
-         }
-
-         if (!directAccess) {
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);
-         }
-      }
-   }
-}
-
-
-static void
-accum_return(struct gl_context *ctx, GLfloat value,
-             GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_renderbuffer *accumRb = fb->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess
-      = (accumRb->GetPointer(ctx, accumRb, 0, 0) != NULL);
-
-   static GLchan multTable[32768];
-   static GLfloat prevMult = 0.0;
-   const GLfloat mult = swrast->_IntegerAccumScaler;
-   const GLint max = MIN2((GLint) (256 / mult), 32767);
-
-   /* May have to leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode && value != 1.0)
-      rescale_accum(ctx);
-
-   if (swrast->_IntegerAccumMode && swrast->_IntegerAccumScaler > 0) {
-      /* build lookup table to avoid many floating point multiplies */
-      GLint j;
-      assert(swrast->_IntegerAccumScaler <= 1.0);
-      if (mult != prevMult) {
-         for (j = 0; j < max; j++)
-            multTable[j] = IROUND((GLfloat) j * mult);
-         prevMult = mult;
-      }
-   }
-
-   if (accumRb->DataType == GL_SHORT ||
-       accumRb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * CHAN_MAXF / ACCUM_SCALE16;
-      GLuint buffer;
-      GLint i;
-
-      /* XXX maybe transpose the 'i' and 'buffer' loops??? */
-      for (i = 0; i < height; i++) {
-         GLshort accumRow[4 * MAX_WIDTH];
-         GLshort *acc;
-         SWspan span;
-
-         /* init color span */
-         INIT_SPAN(span, GL_BITMAP);
-         span.end = width;
-         span.arrayMask = SPAN_RGBA;
-         span.x = xpos;
-         span.y = ypos + i;
-
-         if (directAccess) {
-            acc = (GLshort *) accumRb->GetPointer(ctx, accumRb, xpos, ypos +i);
-         }
-         else {
-            accumRb->GetRow(ctx, accumRb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* get the colors to return */
-         if (swrast->_IntegerAccumMode) {
-            GLint j;
-            for (j = 0; j < width; j++) {
-               ASSERT(acc[j * 4 + 0] < max);
-               ASSERT(acc[j * 4 + 1] < max);
-               ASSERT(acc[j * 4 + 2] < max);
-               ASSERT(acc[j * 4 + 3] < max);
-               span.array->rgba[j][RCOMP] = multTable[acc[j * 4 + 0]];
-               span.array->rgba[j][GCOMP] = multTable[acc[j * 4 + 1]];
-               span.array->rgba[j][BCOMP] = multTable[acc[j * 4 + 2]];
-               span.array->rgba[j][ACOMP] = multTable[acc[j * 4 + 3]];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-#if CHAN_BITS==32
-               GLchan r = acc[j * 4 + 0] * scale;
-               GLchan g = acc[j * 4 + 1] * scale;
-               GLchan b = acc[j * 4 + 2] * scale;
-               GLchan a = acc[j * 4 + 3] * scale;
-#else
-               GLint r = IROUND( (GLfloat) (acc[j * 4 + 0]) * scale );
-               GLint g = IROUND( (GLfloat) (acc[j * 4 + 1]) * scale );
-               GLint b = IROUND( (GLfloat) (acc[j * 4 + 2]) * scale );
-               GLint a = IROUND( (GLfloat) (acc[j * 4 + 3]) * scale );
-#endif
-               span.array->rgba[j][RCOMP] = CLAMP( r, 0, CHAN_MAX );
-               span.array->rgba[j][GCOMP] = CLAMP( g, 0, CHAN_MAX );
-               span.array->rgba[j][BCOMP] = CLAMP( b, 0, CHAN_MAX );
-               span.array->rgba[j][ACOMP] = CLAMP( a, 0, CHAN_MAX );
-            }
-         }
-
-         /* store colors */
-         for (buffer = 0; buffer < fb->_NumColorDrawBuffers; buffer++) {
-            struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buffer];
-            const GLboolean masking = (!ctx->Color.ColorMask[buffer][RCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][GCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][BCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][ACOMP]);
-            if (masking) {
-               _swrast_mask_rgba_span(ctx, rb, &span, buffer);
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, span.array->rgba, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-/**
- * Software fallback for glAccum.
- */
-void
-_swrast_Accum(struct gl_context *ctx, GLenum op, GLfloat value)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLint xpos, ypos, width, height;
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   if (!ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer) {
-      _mesa_warning(ctx, "Calling glAccum() without an accumulation buffer");
-      return;
-   }
-
-   if (!_mesa_check_conditional_render(ctx))
-      return;
-
-   swrast_render_start(ctx);
-
-   /* Compute region after calling swrast_render_start() so that we know the
-    * drawbuffer's size/bounds are up to date.
-    */
-   xpos = ctx->DrawBuffer->_Xmin;
-   ypos = ctx->DrawBuffer->_Ymin;
-   width =  ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-
-   switch (op) {
-      case GL_ADD:
-         if (value != 0.0F) {
-            accum_add(ctx, value, xpos, ypos, width, height);
-	 }
-	 break;
-      case GL_MULT:
-         if (value != 1.0F) {
-            accum_mult(ctx, value, xpos, ypos, width, height);
-	 }
-	 break;
-      case GL_ACCUM:
-         if (value != 0.0F) {
-            accum_accum(ctx, value, xpos, ypos, width, height);
-         }
-	 break;
-      case GL_LOAD:
-         accum_load(ctx, value, xpos, ypos, width, height);
-	 break;
-      case GL_RETURN:
-         accum_return(ctx, value, xpos, ypos, width, height);
-	 break;
-      default:
-         _mesa_problem(ctx, "invalid mode in _swrast_Accum()");
-         break;
-   }
-
-   swrast_render_finish(ctx);
-}
diff --git a/mesalib/src/mesa/swrast/s_clear.c b/mesalib/src/mesa/swrast/s_clear.c
index 980d29bba..615747d69 100644
--- a/mesalib/src/mesa/swrast/s_clear.c
+++ b/mesalib/src/mesa/swrast/s_clear.c
@@ -23,13 +23,13 @@
  */
 
 #include "main/glheader.h"
+#include "main/accum.h"
 #include "main/colormac.h"
 #include "main/condrender.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
 
-#include "s_accum.h"
 #include "s_context.h"
 #include "s_depth.h"
 #include "s_masking.h"
@@ -225,8 +225,7 @@ _swrast_Clear(struct gl_context *ctx, GLbitfield buffers)
          _swrast_clear_depth_buffer(ctx, ctx->DrawBuffer->_DepthBuffer);
       }
       if (buffers & BUFFER_BIT_ACCUM) {
-         _swrast_clear_accum_buffer(ctx,
-                       ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
+         _mesa_clear_accum_buffer(ctx);
       }
       if (buffers & BUFFER_BIT_STENCIL) {
          _swrast_clear_stencil_buffer(ctx, ctx->DrawBuffer->_StencilBuffer);
diff --git a/mesalib/src/mesa/swrast/s_context.c b/mesalib/src/mesa/swrast/s_context.c
index 98702f068..924612578 100644
--- a/mesalib/src/mesa/swrast/s_context.c
+++ b/mesalib/src/mesa/swrast/s_context.c
@@ -597,7 +597,7 @@ _swrast_validate_derived( struct gl_context *ctx )
 
 #define SWRAST_DEBUG 0
 
-/* Public entrypoints:  See also s_accum.c, s_bitmap.c, etc.
+/* Public entrypoints:  See also s_bitmap.c, etc.
  */
 void
 _swrast_Quad( struct gl_context *ctx,
@@ -755,10 +755,6 @@ _swrast_CreateContext( struct gl_context *ctx )
    ctx->Driver.MapTexture = _swrast_map_texture;
    ctx->Driver.UnmapTexture = _swrast_unmap_texture;
 
-   /* Optimized Accum buffer */
-   swrast->_IntegerAccumMode = GL_FALSE;
-   swrast->_IntegerAccumScaler = 0.0;
-
    for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++)
       swrast->TextureSample[i] = NULL;
 
diff --git a/mesalib/src/mesa/swrast/s_context.h b/mesalib/src/mesa/swrast/s_context.h
index d3ba37819..446b990d9 100644
--- a/mesalib/src/mesa/swrast/s_context.h
+++ b/mesalib/src/mesa/swrast/s_context.h
@@ -203,11 +203,6 @@ typedef struct
    /** Indicates how each attrib is to be interpolated (lines/tris) */
    GLenum _InterpMode[FRAG_ATTRIB_MAX]; /* GL_FLAT or GL_SMOOTH (for now) */
 
-   /* Accum buffer temporaries.
-    */
-   GLboolean _IntegerAccumMode;	/**< Storing unscaled integers? */
-   GLfloat _IntegerAccumScaler;	/**< Implicit scale factor */
-
    /* Working values:
     */
    GLuint StippleCounter;    /**< Line stipple counter */
diff --git a/mesalib/src/mesa/swrast/s_copypix.c b/mesalib/src/mesa/swrast/s_copypix.c
index 3ba31f22c..3bdf48b25 100644
--- a/mesalib/src/mesa/swrast/s_copypix.c
+++ b/mesalib/src/mesa/swrast/s_copypix.c
@@ -610,6 +610,7 @@ fast_copy_pixels(struct gl_context *ctx,
    struct gl_framebuffer *dstFb = ctx->DrawBuffer;
    struct gl_renderbuffer *srcRb, *dstRb;
    GLint row, yStep;
+   void *temp;
 
    if (SWRAST_CONTEXT(ctx)->_RasterMask != 0x0 ||
        ctx->Pixel.ZoomX != 1.0F ||
@@ -667,14 +668,21 @@ fast_copy_pixels(struct gl_context *ctx,
       yStep = 1;
    }
 
+   temp = malloc(width * MAX_PIXEL_BYTES);
+   if (!temp) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels");
+      return GL_FALSE;
+   }
+
    for (row = 0; row < height; row++) {
-      GLuint temp[MAX_WIDTH][4];
       srcRb->GetRow(ctx, srcRb, width, srcX, srcY, temp);
       dstRb->PutRow(ctx, dstRb, width, dstX, dstY, temp, NULL);
       srcY += yStep;
       dstY += yStep;
    }
 
+   free(temp);
+
    return GL_TRUE;
 }
 
diff --git a/mesalib/src/mesa/swrast/s_renderbuffer.c b/mesalib/src/mesa/swrast/s_renderbuffer.c
new file mode 100644
index 000000000..7622e38fd
--- /dev/null
+++ b/mesalib/src/mesa/swrast/s_renderbuffer.c
@@ -0,0 +1,2034 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  6.5
+ *
+ * Copyright (C) 1999-2006  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.
+ */
+
+
+/**
+ * Functions for allocating/managing software-based renderbuffers.
+ * Also, routines for reading/writing software-based renderbuffer data as
+ * ubytes, ushorts, uints, etc.
+ */
+
+
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/fbobject.h"
+#include "main/formats.h"
+#include "main/mtypes.h"
+#include "main/renderbuffer.h"
+#include "swrast/s_renderbuffer.h"
+
+
+/*
+ * Routines for get/put values in common buffer formats follow.
+ */
+
+/* Returns a bytes per pixel of the DataType in the get/put span
+ * functions for at least a subset of the available combinations a
+ * renderbuffer can have.
+ *
+ * It would be nice to see gl_renderbuffer start talking about a
+ * gl_format instead of a GLenum DataType.
+ */
+static int
+get_datatype_bytes(struct gl_renderbuffer *rb)
+{
+   int component_size;
+
+   switch (rb->DataType) {
+   case GL_FLOAT_32_UNSIGNED_INT_24_8_REV:
+      component_size = 8;
+      break;
+   case GL_FLOAT:
+   case GL_UNSIGNED_INT:
+   case GL_UNSIGNED_INT_24_8_EXT:
+      component_size = 4;
+      break;
+   case GL_UNSIGNED_SHORT:
+      component_size = 2;
+      break;
+   case GL_UNSIGNED_BYTE:
+      component_size = 1;
+      break;
+   default:
+      component_size = 1;
+      assert(0);
+   }
+
+   switch (rb->_BaseFormat) {
+   case GL_DEPTH_COMPONENT:
+   case GL_DEPTH_STENCIL:
+      return component_size;
+   default:
+      return 4 * component_size;
+   }
+}
+
+/* This is commonly used by most of the accessors. */
+static void *
+get_pointer_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		    GLint x, GLint y)
+{
+   if (!rb->Data)
+      return NULL;
+
+   return ((char *) rb->Data +
+	   (y * rb->RowStride + x) * _mesa_get_format_bytes(rb->Format));
+}
+
+/* GetRow() implementation for formats where DataType matches the rb->Format.
+ */
+static void
+get_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		GLuint count, GLint x, GLint y, void *values)
+{
+   void *src = rb->GetPointer(ctx, rb, x, y);
+   memcpy(values, src, count * _mesa_get_format_bytes(rb->Format));
+}
+
+/* Only used for float textures currently, but might also be used for
+ * RGBA8888, RGBA16, etc.
+ */
+static void
+get_values_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		   GLuint count, const GLint x[], const GLint y[], void *values)
+{
+   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const void *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      char *dst = (char *) values + i * format_bytes;
+      memcpy(dst, src, format_bytes);
+   }
+}
+
+/* For the GL_RED/GL_RG/GL_RGB format/DataType combinations (and
+ * GL_LUMINANCE/GL_INTENSITY?), the Put functions are a matter of
+ * storing those initial components of the value per pixel into the
+ * destination.
+ */
+static void
+put_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		GLuint count, GLint x, GLint y,
+		const void *values, const GLubyte *mask)
+{
+   void *row = rb->GetPointer(ctx, rb, x, y);
+   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
+   int datatype_bytes = get_datatype_bytes(rb);
+   unsigned int i;
+
+   if (mask) {
+      for (i = 0; i < count; i++) {
+	 char *dst = (char *) row + i * format_bytes;
+	 const char *src = (const char *) values + i * datatype_bytes;
+
+         if (mask[i]) {
+	    memcpy(dst, src, format_bytes);
+         }
+      }
+   }
+   else {
+      for (i = 0; i < count; i++) {
+	 char *dst = (char *) row + i * format_bytes;
+	 const char *src = (const char *) values + i * datatype_bytes;
+	 memcpy(dst, src, format_bytes);
+      }
+   }
+}
+
+static void
+put_mono_row_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, GLint x, GLint y,
+		     const void *value, const GLubyte *mask)
+{
+   void *row = rb->GetPointer(ctx, rb, x, y);
+   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
+   unsigned int i;
+
+   if (mask) {
+      for (i = 0; i < count; i++) {
+	 char *dst = (char *) row + i * format_bytes;
+         if (mask[i]) {
+	    memcpy(dst, value, format_bytes);
+         }
+      }
+   }
+   else {
+      for (i = 0; i < count; i++) {
+	 char *dst = (char *) row + i * format_bytes;
+	 memcpy(dst, value, format_bytes);
+      }
+   }
+}
+
+
+static void
+put_values_generic(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		   GLuint count, const GLint x[], const GLint y[],
+		   const void *values, const GLubyte *mask)
+{
+   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
+   int datatype_bytes = get_datatype_bytes(rb);
+   unsigned int i;
+
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+	 void *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
+	 const char *src = (const char *) values + i * datatype_bytes;
+	 memcpy(dst, src, format_bytes);
+      }
+   }
+}
+
+
+static void
+put_mono_values_generic(struct gl_context *ctx,
+			struct gl_renderbuffer *rb,
+			GLuint count, const GLint x[], const GLint y[],
+			const void *value, const GLubyte *mask)
+{
+   int format_bytes = _mesa_get_format_bytes(rb->Format) / sizeof(GLfloat);
+   unsigned int i;
+
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+	 void *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
+	 memcpy(dst, value, format_bytes);
+      }
+   }
+}
+
+/**********************************************************************
+ * Functions for buffers of 1 X GLubyte values.
+ * Typically stencil.
+ */
+
+static void
+get_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                 const GLint x[], const GLint y[], void *values)
+{
+   GLubyte *dst = (GLubyte *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      const GLubyte *src = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
+      dst[i] = *src;
+   }
+}
+
+
+static void
+put_row_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+              GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   const GLubyte *src = (const GLubyte *) values;
+   GLubyte *dst = (GLubyte *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = src[i];
+         }
+      }
+   }
+   else {
+      memcpy(dst, values, count * sizeof(GLubyte));
+   }
+}
+
+
+static void
+put_mono_row_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                   GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   const GLubyte val = *((const GLubyte *) value);
+   GLubyte *dst = (GLubyte *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = val;
+         }
+      }
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         dst[i] = val;
+      }
+   }
+}
+
+
+static void
+put_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                 const GLint x[], const GLint y[],
+                 const void *values, const GLubyte *mask)
+{
+   const GLubyte *src = (const GLubyte *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLubyte *dst = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = src[i];
+      }
+   }
+}
+
+
+static void
+put_mono_values_ubyte(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                      const GLint x[], const GLint y[],
+                      const void *value, const GLubyte *mask)
+{
+   const GLubyte val = *((const GLubyte *) value);
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLubyte *dst = (GLubyte *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = val;
+      }
+   }
+}
+
+
+/**********************************************************************
+ * Functions for buffers of 1 X GLushort values.
+ * Typically depth/Z.
+ */
+
+static void
+get_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], void *values)
+{
+   GLushort *dst = (GLushort *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
+   for (i = 0; i < count; i++) {
+      const GLushort *src = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
+      dst[i] = *src;
+   }
+}
+
+
+static void
+put_row_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+               GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   const GLushort *src = (const GLushort *) values;
+   GLushort *dst = (GLushort *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = src[i];
+         }
+      }
+   }
+   else {
+      memcpy(dst, src, count * sizeof(GLushort));
+   }
+}
+
+
+static void
+put_mono_row_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                    GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   const GLushort val = *((const GLushort *) value);
+   GLushort *dst = (GLushort *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = val;
+         }
+      }
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         dst[i] = val;
+      }
+   }
+}
+
+
+static void
+put_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], const void *values,
+                  const GLubyte *mask)
+{
+   const GLushort *src = (const GLushort *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = src[i];
+      }
+   }
+}
+ 
+
+static void
+put_mono_values_ushort(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                       GLuint count, const GLint x[], const GLint y[],
+                       const void *value, const GLubyte *mask)
+{
+   const GLushort val = *((const GLushort *) value);
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
+            *dst = val;
+         }
+      }
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         GLushort *dst = (GLushort *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = val;
+      }
+   }
+}
+ 
+
+/**********************************************************************
+ * Functions for buffers of 1 X GLuint values.
+ * Typically depth/Z or color index.
+ */
+
+static void
+get_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                const GLint x[], const GLint y[], void *values)
+{
+   GLuint *dst = (GLuint *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
+          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
+   for (i = 0; i < count; i++) {
+      const GLuint *src = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
+      dst[i] = *src;
+   }
+}
+
+
+static void
+put_row_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+             GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   const GLuint *src = (const GLuint *) values;
+   GLuint *dst = (GLuint *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
+          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = src[i];
+         }
+      }
+   }
+   else {
+      memcpy(dst, src, count * sizeof(GLuint));
+   }
+}
+
+
+static void
+put_mono_row_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   const GLuint val = *((const GLuint *) value);
+   GLuint *dst = (GLuint *) rb->Data + y * rb->RowStride + x;
+   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
+          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = val;
+         }
+      }
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         dst[i] = val;
+      }
+   }
+}
+
+
+static void
+put_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                const GLint x[], const GLint y[], const void *values,
+                const GLubyte *mask)
+{
+   const GLuint *src = (const GLuint *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
+          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLuint *dst = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = src[i];
+      }
+   }
+}
+
+
+static void
+put_mono_values_uint(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                     const GLint x[], const GLint y[], const void *value,
+                     const GLubyte *mask)
+{
+   const GLuint val = *((const GLuint *) value);
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_INT ||
+          rb->DataType == GL_UNSIGNED_INT_24_8_EXT);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLuint *dst = (GLuint *) rb->Data + y[i] * rb->RowStride + x[i];
+         *dst = val;
+      }
+   }
+}
+
+
+/**********************************************************************
+ * Functions for buffers of 3 X GLubyte (or GLbyte) values.
+ * Typically color buffers.
+ * NOTE: the incoming and outgoing colors are RGBA!  We ignore incoming
+ * alpha values and return 255 for outgoing alpha values.
+ */
+
+static void *
+get_pointer_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                   GLint x, GLint y)
+{
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   /* No direct access since this buffer is RGB but caller will be
+    * treating it as if it were RGBA.
+    */
+   return NULL;
+}
+
+
+static void
+get_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+               GLint x, GLint y, void *values)
+{
+   const GLubyte *src = ((const GLubyte *) rb->Data) +
+					   3 * (y * rb->RowStride + x);
+   GLubyte *dst = (GLubyte *) values;
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + 0] = src[i * 3 + 0];
+      dst[i * 4 + 1] = src[i * 3 + 1];
+      dst[i * 4 + 2] = src[i * 3 + 2];
+      dst[i * 4 + 3] = 255;
+   }
+}
+
+
+static void
+get_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], void *values)
+{
+   GLubyte *dst = (GLubyte *) values;
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      const GLubyte *src
+         = (GLubyte *) rb->Data + 3 * (y[i] * rb->RowStride + x[i]);
+      dst[i * 4 + 0] = src[0];
+      dst[i * 4 + 1] = src[1];
+      dst[i * 4 + 2] = src[2];
+      dst[i * 4 + 3] = 255;
+   }
+}
+
+
+static void
+put_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+               GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   /* note: incoming values are RGB+A! */
+   const GLubyte *src = (const GLubyte *) values;
+   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         dst[i * 3 + 0] = src[i * 4 + 0];
+         dst[i * 3 + 1] = src[i * 4 + 1];
+         dst[i * 3 + 2] = src[i * 4 + 2];
+      }
+   }
+}
+
+
+static void
+put_row_rgb_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                   GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   /* note: incoming values are RGB+A! */
+   const GLubyte *src = (const GLubyte *) values;
+   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         dst[i * 3 + 0] = src[i * 3 + 0];
+         dst[i * 3 + 1] = src[i * 3 + 1];
+         dst[i * 3 + 2] = src[i * 3 + 2];
+      }
+   }
+}
+
+
+static void
+put_mono_row_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                    GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   /* note: incoming value is RGB+A! */
+   const GLubyte val0 = ((const GLubyte *) value)[0];
+   const GLubyte val1 = ((const GLubyte *) value)[1];
+   const GLubyte val2 = ((const GLubyte *) value)[2];
+   GLubyte *dst = (GLubyte *) rb->Data + 3 * (y * rb->RowStride + x);
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   if (!mask && val0 == val1 && val1 == val2) {
+      /* optimized case */
+      memset(dst, val0, 3 * count);
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            dst[i * 3 + 0] = val0;
+            dst[i * 3 + 1] = val1;
+            dst[i * 3 + 2] = val2;
+         }
+      }
+   }
+}
+
+
+static void
+put_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], const void *values,
+                  const GLubyte *mask)
+{
+   /* note: incoming values are RGB+A! */
+   const GLubyte *src = (const GLubyte *) values;
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLubyte *dst = (GLubyte *) rb->Data + 3 * (y[i] * rb->RowStride + x[i]);
+         dst[0] = src[i * 4 + 0];
+         dst[1] = src[i * 4 + 1];
+         dst[2] = src[i * 4 + 2];
+      }
+   }
+}
+
+
+static void
+put_mono_values_ubyte3(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                       GLuint count, const GLint x[], const GLint y[],
+                       const void *value, const GLubyte *mask)
+{
+   /* note: incoming value is RGB+A! */
+   const GLubyte val0 = ((const GLubyte *) value)[0];
+   const GLubyte val1 = ((const GLubyte *) value)[1];
+   const GLubyte val2 = ((const GLubyte *) value)[2];
+   GLuint i;
+   ASSERT(rb->Format == MESA_FORMAT_RGB888);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLubyte *dst = ((GLubyte *) rb->Data) +
+				     3 * (y[i] * rb->RowStride + x[i]);
+         dst[0] = val0;
+         dst[1] = val1;
+         dst[2] = val2;
+      }
+   }
+}
+
+
+/**********************************************************************
+ * Functions for buffers of 4 X GLubyte (or GLbyte) values.
+ * Typically color buffers.
+ */
+
+static void
+get_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], void *values)
+{
+   /* treat 4*GLubyte as 1*GLuint */
+   GLuint *dst = (GLuint *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   for (i = 0; i < count; i++) {
+      const GLuint *src = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
+      dst[i] = *src;
+   }
+}
+
+
+static void
+put_row_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+               GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   /* treat 4*GLubyte as 1*GLuint */
+   const GLuint *src = (const GLuint *) values;
+   GLuint *dst = (GLuint *) rb->Data + (y * rb->RowStride + x);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i] = src[i];
+         }
+      }
+   }
+   else {
+      memcpy(dst, src, 4 * count * sizeof(GLubyte));
+   }
+}
+
+
+static void
+put_row_rgb_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                   GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   /* Store RGB values in RGBA buffer */
+   const GLubyte *src = (const GLubyte *) values;
+   GLubyte *dst = (GLubyte *) rb->Data + 4 * (y * rb->RowStride + x);
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         dst[i * 4 + 0] = src[i * 3 + 0];
+         dst[i * 4 + 1] = src[i * 3 + 1];
+         dst[i * 4 + 2] = src[i * 3 + 2];
+         dst[i * 4 + 3] = 0xff;
+      }
+   }
+}
+
+
+static void
+put_mono_row_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                    GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   /* treat 4*GLubyte as 1*GLuint */
+   const GLuint val = *((const GLuint *) value);
+   GLuint *dst = (GLuint *) rb->Data + (y * rb->RowStride + x);
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   if (!mask && val == 0) {
+      /* common case */
+      memset(dst, 0, count * 4 * sizeof(GLubyte));
+   }
+   else {
+      /* general case */
+      if (mask) {
+         GLuint i;
+         for (i = 0; i < count; i++) {
+            if (mask[i]) {
+               dst[i] = val;
+            }
+         }
+      }
+      else {
+         GLuint i;
+         for (i = 0; i < count; i++) {
+            dst[i] = val;
+         }
+      }
+   }
+}
+
+
+static void
+put_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                  const GLint x[], const GLint y[], const void *values,
+                  const GLubyte *mask)
+{
+   /* treat 4*GLubyte as 1*GLuint */
+   const GLuint *src = (const GLuint *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLuint *dst = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
+         *dst = src[i];
+      }
+   }
+}
+
+
+static void
+put_mono_values_ubyte4(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                       GLuint count, const GLint x[], const GLint y[],
+                       const void *value, const GLubyte *mask)
+{
+   /* treat 4*GLubyte as 1*GLuint */
+   const GLuint val = *((const GLuint *) value);
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
+   ASSERT(rb->Format == MESA_FORMAT_RGBA8888 ||
+          rb->Format == MESA_FORMAT_RGBA8888_REV);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLuint *dst = (GLuint *) rb->Data + (y[i] * rb->RowStride + x[i]);
+         *dst = val;
+      }
+   }
+}
+
+
+/**********************************************************************
+ * Functions for buffers of 4 X GLushort (or GLshort) values.
+ * Typically accum buffer.
+ */
+
+static void
+get_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                   const GLint x[], const GLint y[], void *values)
+{
+   GLushort *dst = (GLushort *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   for (i = 0; i < count; i++) {
+      const GLushort *src
+         = (GLushort *) rb->Data + 4 * (y[i] * rb->RowStride + x[i]);
+      dst[i] = *src;
+   }
+}
+
+
+static void
+put_row_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   const GLushort *src = (const GLushort *) values;
+   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i * 4 + 0] = src[i * 4 + 0];
+            dst[i * 4 + 1] = src[i * 4 + 1];
+            dst[i * 4 + 2] = src[i * 4 + 2];
+            dst[i * 4 + 3] = src[i * 4 + 3];
+         }
+      }
+   }
+   else {
+      memcpy(dst, src, 4 * count * sizeof(GLushort));
+   }
+}
+
+
+static void
+put_row_rgb_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                    GLint x, GLint y, const void *values, const GLubyte *mask)
+{
+   /* Put RGB values in RGBA buffer */
+   const GLushort *src = (const GLushort *) values;
+   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   if (mask) {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            dst[i * 4 + 0] = src[i * 3 + 0];
+            dst[i * 4 + 1] = src[i * 3 + 1];
+            dst[i * 4 + 2] = src[i * 3 + 2];
+            dst[i * 4 + 3] = 0xffff;
+         }
+      }
+   }
+   else {
+      memcpy(dst, src, 4 * count * sizeof(GLushort));
+   }
+}
+
+
+static void
+put_mono_row_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                     GLint x, GLint y, const void *value, const GLubyte *mask)
+{
+   const GLushort val0 = ((const GLushort *) value)[0];
+   const GLushort val1 = ((const GLushort *) value)[1];
+   const GLushort val2 = ((const GLushort *) value)[2];
+   const GLushort val3 = ((const GLushort *) value)[3];
+   GLushort *dst = (GLushort *) rb->Data + 4 * (y * rb->RowStride + x);
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   if (!mask && val0 == 0 && val1 == 0 && val2 == 0 && val3 == 0) {
+      /* common case for clearing accum buffer */
+      memset(dst, 0, count * 4 * sizeof(GLushort));
+   }
+   else {
+      GLuint i;
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            dst[i * 4 + 0] = val0;
+            dst[i * 4 + 1] = val1;
+            dst[i * 4 + 2] = val2;
+            dst[i * 4 + 3] = val3;
+         }
+      }
+   }
+}
+
+
+static void
+put_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+                   const GLint x[], const GLint y[], const void *values,
+                   const GLubyte *mask)
+{
+   const GLushort *src = (const GLushort *) values;
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLushort *dst =
+            ((GLushort *) rb->Data) + 4 * (y[i] * rb->RowStride + x[i]);
+         dst[0] = src[i * 4 + 0];
+         dst[1] = src[i * 4 + 1];
+         dst[2] = src[i * 4 + 2];
+         dst[3] = src[i * 4 + 3];
+      }
+   }
+}
+
+
+static void
+put_mono_values_ushort4(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                        GLuint count, const GLint x[], const GLint y[],
+                        const void *value, const GLubyte *mask)
+{
+   const GLushort val0 = ((const GLushort *) value)[0];
+   const GLushort val1 = ((const GLushort *) value)[1];
+   const GLushort val2 = ((const GLushort *) value)[2];
+   const GLushort val3 = ((const GLushort *) value)[3];
+   GLuint i;
+   ASSERT(rb->DataType == GL_UNSIGNED_SHORT || rb->DataType == GL_SHORT);
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+         GLushort *dst = ((GLushort *) rb->Data) +
+				       4 * (y[i] * rb->RowStride + x[i]);
+         dst[0] = val0;
+         dst[1] = val1;
+         dst[2] = val2;
+         dst[3] = val3;
+      }
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_R8.
+ */
+static void
+get_row_r8(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	   GLint x, GLint y, void *values)
+{
+   const GLubyte *src = rb->GetPointer(ctx, rb, x, y);
+   GLuint *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i] = 0xff000000 | src[i];
+   }
+}
+
+static void
+get_values_r8(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	      const GLint x[], const GLint y[], void *values)
+{
+   GLuint *dst = (GLuint *) values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLubyte *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i] = 0xff000000 | *src;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_GR88.
+ */
+static void
+get_row_rg88(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	     GLint x, GLint y, void *values)
+{
+   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
+   GLuint *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i] = 0xff000000 | src[i];
+   }
+}
+
+static void
+get_values_rg88(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		GLuint count, const GLint x[], const GLint y[], void *values)
+{
+   GLuint *dst = (GLuint *) values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLshort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i] = 0xff000000 | *src;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_R16.
+ */
+static void
+get_row_r16(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	    GLint x, GLint y, void *values)
+{
+   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
+   GLushort *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] = src[i];
+      dst[i * 4 + GCOMP] = 0;
+      dst[i * 4 + BCOMP] = 0;
+      dst[i * 4 + ACOMP] = 0xffff;
+   }
+}
+
+static void
+get_values_r16(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	       const GLint x[], const GLint y[], void *values)
+{
+   GLushort *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLushort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] = *src;
+      dst[i * 4 + GCOMP] = 0;
+      dst[i * 4 + BCOMP] = 0;
+      dst[i * 4 + ACOMP] = 0xffff;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_RG1616.
+ */
+static void
+get_row_rg1616(struct gl_context *ctx, struct gl_renderbuffer *rb, GLuint count,
+	       GLint x, GLint y, void *values)
+{
+   const GLushort *src = rb->GetPointer(ctx, rb, x, y);
+   GLushort *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] = src[i * 2];
+      dst[i * 4 + GCOMP] = src[i * 2 + 1];
+      dst[i * 4 + BCOMP] = 0;
+      dst[i * 4 + ACOMP] = 0xffff;
+   }
+}
+
+static void
+get_values_rg1616(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, const GLint x[], const GLint y[], void *values)
+{
+   GLushort *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLshort *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] = src[0];
+      dst[i * 4 + GCOMP] = src[1];
+      dst[i * 4 + BCOMP] = 0;
+      dst[i * 4 + ACOMP] = 0xffff;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_INTENSITY_FLOAT32.
+ */
+static void
+get_row_i_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, GLint x, GLint y, void *values)
+{
+   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] =
+      dst[i * 4 + GCOMP] =
+      dst[i * 4 + BCOMP] =
+      dst[i * 4 + ACOMP] = src[i];
+   }
+}
+
+static void
+get_values_i_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, const GLint x[], const GLint y[],
+		     void *values)
+{
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] =
+      dst[i * 4 + GCOMP] =
+      dst[i * 4 + BCOMP] =
+      dst[i * 4 + ACOMP] = src[0];
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_LUMINANCE_FLOAT32.
+ */
+static void
+get_row_l_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, GLint x, GLint y, void *values)
+{
+   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] =
+      dst[i * 4 + GCOMP] =
+      dst[i * 4 + BCOMP] = src[i];
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+static void
+get_values_l_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, const GLint x[], const GLint y[],
+		     void *values)
+{
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] =
+      dst[i * 4 + GCOMP] =
+      dst[i * 4 + BCOMP] = src[0];
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_ALPHA_FLOAT32.
+ */
+static void
+get_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, GLint x, GLint y, void *values)
+{
+   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] = 0.0;
+      dst[i * 4 + GCOMP] = 0.0;
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = src[i];
+   }
+}
+
+static void
+get_values_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, const GLint x[], const GLint y[],
+		     void *values)
+{
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] = 0.0;
+      dst[i * 4 + GCOMP] = 0.0;
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = src[0];
+   }
+}
+
+static void
+put_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, GLint x, GLint y,
+		  const void *values, const GLubyte *mask)
+{
+   float *dst = rb->GetPointer(ctx, rb, x, y);
+   const float *src = values;
+   unsigned int i;
+
+   if (mask) {
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+	    dst[i] = src[i * 4 + ACOMP];
+         }
+      }
+   }
+   else {
+      for (i = 0; i < count; i++) {
+	 dst[i] = src[i * 4 + ACOMP];
+      }
+   }
+}
+
+static void
+put_mono_row_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		       GLuint count, GLint x, GLint y,
+		       const void *value, const GLubyte *mask)
+{
+   float *dst = rb->GetPointer(ctx, rb, x, y);
+   const float *src = value;
+   unsigned int i;
+
+   if (mask) {
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+	    dst[i] = src[ACOMP];
+         }
+      }
+   }
+   else {
+      for (i = 0; i < count; i++) {
+	 dst[i] = src[ACOMP];
+      }
+   }
+}
+
+static void
+put_values_a_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, const GLint x[], const GLint y[],
+		     const void *values, const GLubyte *mask)
+{
+   const float *src = values;
+   unsigned int i;
+
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+	 float *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
+
+	 *dst = src[i * 4 + ACOMP];
+      }
+   }
+}
+
+static void
+put_mono_values_a_float32(struct gl_context *ctx,
+			  struct gl_renderbuffer *rb,
+			  GLuint count, const GLint x[], const GLint y[],
+			  const void *value, const GLubyte *mask)
+{
+   const float *src = value;
+   unsigned int i;
+
+   for (i = 0; i < count; i++) {
+      if (!mask || mask[i]) {
+	 float *dst = rb->GetPointer(ctx, rb, x[i], y[i]);
+	 *dst = src[ACOMP];
+      }
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_R_FLOAT32.
+ */
+static void
+get_row_r_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		  GLuint count, GLint x, GLint y, void *values)
+{
+   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] = src[i];
+      dst[i * 4 + GCOMP] = 0.0;
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+static void
+get_values_r_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		     GLuint count, const GLint x[], const GLint y[],
+		     void *values)
+{
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] = src[0];
+      dst[i * 4 + GCOMP] = 0.0;
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+/**********************************************************************
+ * Functions for MESA_FORMAT_RG_FLOAT32.
+ */
+static void
+get_row_rg_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		   GLuint count, GLint x, GLint y, void *values)
+{
+   const GLfloat *src = rb->GetPointer(ctx, rb, x, y);
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      dst[i * 4 + RCOMP] = src[i * 2 + 0];
+      dst[i * 4 + GCOMP] = src[i * 2 + 1];
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+static void
+get_values_rg_float32(struct gl_context *ctx, struct gl_renderbuffer *rb,
+		      GLuint count, const GLint x[], const GLint y[],
+		      void *values)
+{
+   GLfloat *dst = values;
+   GLuint i;
+
+   for (i = 0; i < count; i++) {
+      const GLfloat *src = rb->GetPointer(ctx, rb, x[i], y[i]);
+      dst[i * 4 + RCOMP] = src[0];
+      dst[i * 4 + GCOMP] = src[1];
+      dst[i * 4 + BCOMP] = 0.0;
+      dst[i * 4 + ACOMP] = 1.0;
+   }
+}
+
+/**
+ * This is the default software fallback for gl_renderbuffer's span
+ * access functions.
+ *
+ * The assumptions are that rb->Data will be a pointer to (0,0), that pixels
+ * are packed in the type of rb->Format, and that subsequent rows appear
+ * rb->RowStride pixels later.
+ */
+void
+_swrast_set_renderbuffer_accessors(struct gl_renderbuffer *rb)
+{
+   rb->GetPointer = get_pointer_generic;
+   rb->GetRow = get_row_generic;
+
+   switch (rb->Format) {
+   case MESA_FORMAT_RGB888:
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->GetPointer = get_pointer_ubyte3;
+      rb->GetRow = get_row_ubyte3;
+      rb->GetValues = get_values_ubyte3;
+      rb->PutRow = put_row_ubyte3;
+      rb->PutRowRGB = put_row_rgb_ubyte3;
+      rb->PutMonoRow = put_mono_row_ubyte3;
+      rb->PutValues = put_values_ubyte3;
+      rb->PutMonoValues = put_mono_values_ubyte3;
+      break;
+
+   case MESA_FORMAT_RGBA8888:
+   case MESA_FORMAT_RGBA8888_REV:
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->GetValues = get_values_ubyte4;
+      rb->PutRow = put_row_ubyte4;
+      rb->PutRowRGB = put_row_rgb_ubyte4;
+      rb->PutMonoRow = put_mono_row_ubyte4;
+      rb->PutValues = put_values_ubyte4;
+      rb->PutMonoValues = put_mono_values_ubyte4;
+      break;
+
+   case MESA_FORMAT_R8:
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->GetValues = get_values_r8;
+      rb->GetRow = get_row_r8;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = put_row_generic;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_GR88:
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->GetValues = get_values_rg88;
+      rb->GetRow = get_row_rg88;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = put_row_generic;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_R16:
+      rb->DataType = GL_UNSIGNED_SHORT;
+      rb->GetValues = get_values_r16;
+      rb->GetRow = get_row_r16;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = put_row_generic;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_RG1616:
+      rb->DataType = GL_UNSIGNED_SHORT;
+      rb->GetValues = get_values_rg1616;
+      rb->GetRow = get_row_rg1616;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = put_row_generic;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_SIGNED_RGBA_16:
+      rb->DataType = GL_SHORT;
+      rb->GetValues = get_values_ushort4;
+      rb->PutRow = put_row_ushort4;
+      rb->PutRowRGB = put_row_rgb_ushort4;
+      rb->PutMonoRow = put_mono_row_ushort4;
+      rb->PutValues = put_values_ushort4;
+      rb->PutMonoValues = put_mono_values_ushort4;
+      break;
+
+   case MESA_FORMAT_S8:
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->GetValues = get_values_ubyte;
+      rb->PutRow = put_row_ubyte;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_ubyte;
+      rb->PutValues = put_values_ubyte;
+      rb->PutMonoValues = put_mono_values_ubyte;
+      break;
+
+   case MESA_FORMAT_Z16:
+      rb->DataType = GL_UNSIGNED_SHORT;
+      rb->GetValues = get_values_ushort;
+      rb->PutRow = put_row_ushort;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_ushort;
+      rb->PutValues = put_values_ushort;
+      rb->PutMonoValues = put_mono_values_ushort;
+      break;
+
+   case MESA_FORMAT_Z32:
+   case MESA_FORMAT_X8_Z24:
+   case MESA_FORMAT_Z24_X8:
+      rb->DataType = GL_UNSIGNED_INT;
+      rb->GetValues = get_values_uint;
+      rb->PutRow = put_row_uint;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_uint;
+      rb->PutValues = put_values_uint;
+      rb->PutMonoValues = put_mono_values_uint;
+      break;
+
+   case MESA_FORMAT_Z24_S8:
+   case MESA_FORMAT_S8_Z24:
+      rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
+      rb->GetValues = get_values_uint;
+      rb->PutRow = put_row_uint;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_uint;
+      rb->PutValues = put_values_uint;
+      rb->PutMonoValues = put_mono_values_uint;
+      break;
+
+   case MESA_FORMAT_RGBA_FLOAT32:
+      rb->GetRow = get_row_generic;
+      rb->GetValues = get_values_generic;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_INTENSITY_FLOAT32:
+      rb->GetRow = get_row_i_float32;
+      rb->GetValues = get_values_i_float32;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_LUMINANCE_FLOAT32:
+      rb->GetRow = get_row_l_float32;
+      rb->GetValues = get_values_l_float32;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_ALPHA_FLOAT32:
+      rb->GetRow = get_row_a_float32;
+      rb->GetValues = get_values_a_float32;
+      rb->PutRow = put_row_a_float32;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_a_float32;
+      rb->PutValues = put_values_a_float32;
+      rb->PutMonoValues = put_mono_values_a_float32;
+      break;
+
+   case MESA_FORMAT_RG_FLOAT32:
+      rb->GetRow = get_row_rg_float32;
+      rb->GetValues = get_values_rg_float32;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   case MESA_FORMAT_R_FLOAT32:
+      rb->GetRow = get_row_r_float32;
+      rb->GetValues = get_values_r_float32;
+      rb->PutRow = put_row_generic;
+      rb->PutRowRGB = NULL;
+      rb->PutMonoRow = put_mono_row_generic;
+      rb->PutValues = put_values_generic;
+      rb->PutMonoValues = put_mono_values_generic;
+      break;
+
+   default:
+      break;
+   }
+}
+
+/**
+ * This is a software fallback for the gl_renderbuffer->AllocStorage
+ * function.
+ * Device drivers will typically override this function for the buffers
+ * which it manages (typically color buffers, Z and stencil).
+ * Other buffers (like software accumulation and aux buffers) which the driver
+ * doesn't manage can be handled with this function.
+ *
+ * This one multi-purpose function can allocate stencil, depth, accum, color
+ * or color-index buffers!
+ *
+ * This function also plugs in the appropriate GetPointer, Get/PutRow and
+ * Get/PutValues functions.
+ */
+static GLboolean
+soft_renderbuffer_storage(struct gl_context *ctx, struct gl_renderbuffer *rb,
+                          GLenum internalFormat,
+                          GLuint width, GLuint height)
+{
+   switch (internalFormat) {
+   case GL_RGB:
+   case GL_R3_G3_B2:
+   case GL_RGB4:
+   case GL_RGB5:
+   case GL_RGB8:
+   case GL_RGB10:
+   case GL_RGB12:
+   case GL_RGB16:
+      rb->Format = MESA_FORMAT_RGB888;
+      break;
+   case GL_RGBA:
+   case GL_RGBA2:
+   case GL_RGBA4:
+   case GL_RGB5_A1:
+   case GL_RGBA8:
+#if 1
+   case GL_RGB10_A2:
+   case GL_RGBA12:
+#endif
+      if (_mesa_little_endian())
+         rb->Format = MESA_FORMAT_RGBA8888_REV;
+      else
+         rb->Format = MESA_FORMAT_RGBA8888;
+      break;
+   case GL_RGBA16:
+   case GL_RGBA16_SNORM:
+      /* for accum buffer */
+      rb->Format = MESA_FORMAT_SIGNED_RGBA_16;
+      break;
+   case GL_STENCIL_INDEX:
+   case GL_STENCIL_INDEX1_EXT:
+   case GL_STENCIL_INDEX4_EXT:
+   case GL_STENCIL_INDEX8_EXT:
+   case GL_STENCIL_INDEX16_EXT:
+      rb->Format = MESA_FORMAT_S8;
+      break;
+   case GL_DEPTH_COMPONENT:
+   case GL_DEPTH_COMPONENT16:
+      rb->Format = MESA_FORMAT_Z16;
+      break;
+   case GL_DEPTH_COMPONENT24:
+      rb->Format = MESA_FORMAT_X8_Z24;
+      break;
+   case GL_DEPTH_COMPONENT32:
+      rb->Format = MESA_FORMAT_Z32;
+      break;
+   case GL_DEPTH_STENCIL_EXT:
+   case GL_DEPTH24_STENCIL8_EXT:
+      rb->Format = MESA_FORMAT_Z24_S8;
+      break;
+   default:
+      /* unsupported format */
+      return GL_FALSE;
+   }
+
+   _swrast_set_renderbuffer_accessors(rb);
+
+   ASSERT(rb->DataType);
+   ASSERT(rb->GetPointer);
+   ASSERT(rb->GetRow);
+   ASSERT(rb->GetValues);
+   ASSERT(rb->PutRow);
+   ASSERT(rb->PutMonoRow);
+   ASSERT(rb->PutValues);
+   ASSERT(rb->PutMonoValues);
+
+   /* free old buffer storage */
+   if (rb->Data) {
+      free(rb->Data);
+      rb->Data = NULL;
+   }
+
+   rb->RowStride = width;
+
+   if (width > 0 && height > 0) {
+      /* allocate new buffer storage */
+      rb->Data = malloc(width * height * _mesa_get_format_bytes(rb->Format));
+
+      if (rb->Data == NULL) {
+         rb->Width = 0;
+         rb->Height = 0;
+	 rb->RowStride = 0;
+         _mesa_error(ctx, GL_OUT_OF_MEMORY,
+                     "software renderbuffer allocation (%d x %d x %d)",
+                     width, height, _mesa_get_format_bytes(rb->Format));
+         return GL_FALSE;
+      }
+   }
+
+   rb->Width = width;
+   rb->Height = height;
+   rb->_BaseFormat = _mesa_base_fbo_format(ctx, internalFormat);
+
+   if (rb->Name == 0 &&
+       internalFormat == GL_RGBA16_SNORM &&
+       rb->_BaseFormat == 0) {
+      /* NOTE: This is a special case just for accumulation buffers.
+       * This is a very limited use case- there's no snorm texturing or
+       * rendering going on.
+       */
+      rb->_BaseFormat = GL_RGBA;
+   }
+   else {
+      /* the internalFormat should have been error checked long ago */
+      ASSERT(rb->_BaseFormat);
+   }
+
+   return GL_TRUE;
+}
+
+
+void
+_swrast_map_soft_renderbuffer(struct gl_context *ctx,
+                              struct gl_renderbuffer *rb,
+                              GLuint x, GLuint y, GLuint w, GLuint h,
+                              GLbitfield mode,
+                              GLubyte **out_map,
+                              GLint *out_stride)
+{
+   GLubyte *map = rb->Data;
+   int cpp = _mesa_get_format_bytes(rb->Format);
+   int stride = rb->RowStride * cpp;
+
+   ASSERT(rb->Data);
+
+   map += y * stride;
+   map += x * cpp;
+
+   *out_map = map;
+   *out_stride = stride;
+}
+
+
+void
+_swrast_unmap_soft_renderbuffer(struct gl_context *ctx,
+                                struct gl_renderbuffer *rb)
+{
+}
+
+
+
+/**
+ * Allocate a software-based renderbuffer.  This is called via the
+ * ctx->Driver.NewRenderbuffer() function when the user creates a new
+ * renderbuffer.
+ * This would not be used for hardware-based renderbuffers.
+ */
+struct gl_renderbuffer *
+_swrast_new_soft_renderbuffer(struct gl_context *ctx, GLuint name)
+{
+   struct gl_renderbuffer *rb = _mesa_new_renderbuffer(ctx, name);
+   if (rb) {
+      rb->AllocStorage = soft_renderbuffer_storage;
+      /* Normally, one would setup the PutRow, GetRow, etc functions here.
+       * But we're doing that in the soft_renderbuffer_storage() function
+       * instead.
+       */
+   }
+   return rb;
+}
+
+
+/**
+ * Add software-based color renderbuffers to the given framebuffer.
+ * This is a helper routine for device drivers when creating a
+ * window system framebuffer (not a user-created render/framebuffer).
+ * Once this function is called, you can basically forget about this
+ * renderbuffer; core Mesa will handle all the buffer management and
+ * rendering!
+ */
+static GLboolean
+add_color_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
+                        GLuint rgbBits, GLuint alphaBits,
+                        GLboolean frontLeft, GLboolean backLeft,
+                        GLboolean frontRight, GLboolean backRight)
+{
+   gl_buffer_index b;
+
+   if (rgbBits > 16 || alphaBits > 16) {
+      _mesa_problem(ctx,
+                    "Unsupported bit depth in add_color_renderbuffers");
+      return GL_FALSE;
+   }
+
+   assert(MAX_COLOR_ATTACHMENTS >= 4);
+
+   for (b = BUFFER_FRONT_LEFT; b <= BUFFER_BACK_RIGHT; b++) {
+      struct gl_renderbuffer *rb;
+
+      if (b == BUFFER_FRONT_LEFT && !frontLeft)
+         continue;
+      else if (b == BUFFER_BACK_LEFT && !backLeft)
+         continue;
+      else if (b == BUFFER_FRONT_RIGHT && !frontRight)
+         continue;
+      else if (b == BUFFER_BACK_RIGHT && !backRight)
+         continue;
+
+      assert(fb->Attachment[b].Renderbuffer == NULL);
+
+      rb = _mesa_new_renderbuffer(ctx, 0);
+      if (!rb) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating color buffer");
+         return GL_FALSE;
+      }
+
+      rb->InternalFormat = GL_RGBA;
+
+      rb->AllocStorage = soft_renderbuffer_storage;
+      _mesa_add_renderbuffer(fb, b, rb);
+   }
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Add a software-based depth renderbuffer to the given framebuffer.
+ * This is a helper routine for device drivers when creating a
+ * window system framebuffer (not a user-created render/framebuffer).
+ * Once this function is called, you can basically forget about this
+ * renderbuffer; core Mesa will handle all the buffer management and
+ * rendering!
+ */
+static GLboolean
+add_depth_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
+                       GLuint depthBits)
+{
+   struct gl_renderbuffer *rb;
+
+   if (depthBits > 32) {
+      _mesa_problem(ctx,
+                    "Unsupported depthBits in add_depth_renderbuffer");
+      return GL_FALSE;
+   }
+
+   assert(fb->Attachment[BUFFER_DEPTH].Renderbuffer == NULL);
+
+   rb = _mesa_new_renderbuffer(ctx, 0);
+   if (!rb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating depth buffer");
+      return GL_FALSE;
+   }
+
+   if (depthBits <= 16) {
+      rb->InternalFormat = GL_DEPTH_COMPONENT16;
+   }
+   else if (depthBits <= 24) {
+      rb->InternalFormat = GL_DEPTH_COMPONENT24;
+   }
+   else {
+      rb->InternalFormat = GL_DEPTH_COMPONENT32;
+   }
+
+   rb->AllocStorage = soft_renderbuffer_storage;
+   _mesa_add_renderbuffer(fb, BUFFER_DEPTH, rb);
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Add a software-based stencil renderbuffer to the given framebuffer.
+ * This is a helper routine for device drivers when creating a
+ * window system framebuffer (not a user-created render/framebuffer).
+ * Once this function is called, you can basically forget about this
+ * renderbuffer; core Mesa will handle all the buffer management and
+ * rendering!
+ */
+static GLboolean
+add_stencil_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
+                         GLuint stencilBits)
+{
+   struct gl_renderbuffer *rb;
+
+   if (stencilBits > 16) {
+      _mesa_problem(ctx,
+                  "Unsupported stencilBits in add_stencil_renderbuffer");
+      return GL_FALSE;
+   }
+
+   assert(fb->Attachment[BUFFER_STENCIL].Renderbuffer == NULL);
+
+   rb = _mesa_new_renderbuffer(ctx, 0);
+   if (!rb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating stencil buffer");
+      return GL_FALSE;
+   }
+
+   assert(stencilBits <= 8);
+   rb->InternalFormat = GL_STENCIL_INDEX8;
+
+   rb->AllocStorage = soft_renderbuffer_storage;
+   _mesa_add_renderbuffer(fb, BUFFER_STENCIL, rb);
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Add a software-based accumulation renderbuffer to the given framebuffer.
+ * This is a helper routine for device drivers when creating a
+ * window system framebuffer (not a user-created render/framebuffer).
+ * Once this function is called, you can basically forget about this
+ * renderbuffer; core Mesa will handle all the buffer management and
+ * rendering!
+ */
+static GLboolean
+add_accum_renderbuffer(struct gl_context *ctx, struct gl_framebuffer *fb,
+                       GLuint redBits, GLuint greenBits,
+                       GLuint blueBits, GLuint alphaBits)
+{
+   struct gl_renderbuffer *rb;
+
+   if (redBits > 16 || greenBits > 16 || blueBits > 16 || alphaBits > 16) {
+      _mesa_problem(ctx,
+                    "Unsupported accumBits in add_accum_renderbuffer");
+      return GL_FALSE;
+   }
+
+   assert(fb->Attachment[BUFFER_ACCUM].Renderbuffer == NULL);
+
+   rb = _mesa_new_renderbuffer(ctx, 0);
+   if (!rb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating accum buffer");
+      return GL_FALSE;
+   }
+
+   rb->InternalFormat = GL_RGBA16_SNORM;
+   rb->AllocStorage = soft_renderbuffer_storage;
+   _mesa_add_renderbuffer(fb, BUFFER_ACCUM, rb);
+
+   return GL_TRUE;
+}
+
+
+
+/**
+ * Add a software-based aux renderbuffer to the given framebuffer.
+ * This is a helper routine for device drivers when creating a
+ * window system framebuffer (not a user-created render/framebuffer).
+ * Once this function is called, you can basically forget about this
+ * renderbuffer; core Mesa will handle all the buffer management and
+ * rendering!
+ *
+ * NOTE: color-index aux buffers not supported.
+ */
+static GLboolean
+add_aux_renderbuffers(struct gl_context *ctx, struct gl_framebuffer *fb,
+                      GLuint colorBits, GLuint numBuffers)
+{
+   GLuint i;
+
+   if (colorBits > 16) {
+      _mesa_problem(ctx,
+                    "Unsupported colorBits in add_aux_renderbuffers");
+      return GL_FALSE;
+   }
+
+   assert(numBuffers <= MAX_AUX_BUFFERS);
+
+   for (i = 0; i < numBuffers; i++) {
+      struct gl_renderbuffer *rb = _mesa_new_renderbuffer(ctx, 0);
+
+      assert(fb->Attachment[BUFFER_AUX0 + i].Renderbuffer == NULL);
+
+      if (!rb) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "Allocating aux buffer");
+         return GL_FALSE;
+      }
+
+      assert (colorBits <= 8);
+      rb->InternalFormat = GL_RGBA;
+
+      rb->AllocStorage = soft_renderbuffer_storage;
+      _mesa_add_renderbuffer(fb, BUFFER_AUX0 + i, rb);
+   }
+   return GL_TRUE;
+}
+
+
+/**
+ * Create/attach software-based renderbuffers to the given framebuffer.
+ * This is a helper routine for device drivers.  Drivers can just as well
+ * call the individual _mesa_add_*_renderbuffer() routines directly.
+ */
+void
+_swrast_add_soft_renderbuffers(struct gl_framebuffer *fb,
+                               GLboolean color,
+                               GLboolean depth,
+                               GLboolean stencil,
+                               GLboolean accum,
+                               GLboolean alpha,
+                               GLboolean aux)
+{
+   GLboolean frontLeft = GL_TRUE;
+   GLboolean backLeft = fb->Visual.doubleBufferMode;
+   GLboolean frontRight = fb->Visual.stereoMode;
+   GLboolean backRight = fb->Visual.stereoMode && fb->Visual.doubleBufferMode;
+
+   if (color) {
+      assert(fb->Visual.redBits == fb->Visual.greenBits);
+      assert(fb->Visual.redBits == fb->Visual.blueBits);
+      add_color_renderbuffers(NULL, fb,
+                              fb->Visual.redBits,
+                              fb->Visual.alphaBits,
+                              frontLeft, backLeft,
+                              frontRight, backRight);
+   }
+
+   if (depth) {
+      assert(fb->Visual.depthBits > 0);
+      add_depth_renderbuffer(NULL, fb, fb->Visual.depthBits);
+   }
+
+   if (stencil) {
+      assert(fb->Visual.stencilBits > 0);
+      add_stencil_renderbuffer(NULL, fb, fb->Visual.stencilBits);
+   }
+
+   if (accum) {
+      assert(fb->Visual.accumRedBits > 0);
+      assert(fb->Visual.accumGreenBits > 0);
+      assert(fb->Visual.accumBlueBits > 0);
+      add_accum_renderbuffer(NULL, fb,
+                             fb->Visual.accumRedBits,
+                             fb->Visual.accumGreenBits,
+                             fb->Visual.accumBlueBits,
+                             fb->Visual.accumAlphaBits);
+   }
+
+   if (aux) {
+      assert(fb->Visual.numAuxBuffers > 0);
+      add_aux_renderbuffers(NULL, fb, fb->Visual.redBits,
+                            fb->Visual.numAuxBuffers);
+   }
+
+#if 0
+   if (multisample) {
+      /* maybe someday */
+   }
+#endif
+}
diff --git a/mesalib/src/mesa/swrast/s_accum.h b/mesalib/src/mesa/swrast/s_renderbuffer.h
index bdd6ac326..266ebd934 100644
--- a/mesalib/src/mesa/swrast/s_accum.h
+++ b/mesalib/src/mesa/swrast/s_renderbuffer.h
@@ -1,38 +1,66 @@
-/*

- * Mesa 3-D graphics library

- * Version:  6.3

- *

- * Copyright (C) 1999-2005  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.

- */

-

-

-#ifndef S_ACCUM_H

-#define S_ACCUM_H

-

-

-struct gl_context;

-struct gl_renderbuffer;

-

-

-extern void

-_swrast_clear_accum_buffer(struct gl_context *ctx, struct gl_renderbuffer *rb);

-

-

-#endif

+/*
+ * Mesa 3-D graphics library
+ * Version:  6.5
+ *
+ * Copyright (C) 1999-2005  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.
+ */
+
+
+#ifndef S_RENDERBUFFER_H
+#define S_RENDERBUFFER_H
+
+#include "main/glheader.h"
+
+
+struct gl_context;
+struct gl_framebuffer;
+struct gl_renderbuffer;
+
+
+extern struct gl_renderbuffer *
+_swrast_new_soft_renderbuffer(struct gl_context *ctx, GLuint name);
+
+extern void
+_swrast_map_soft_renderbuffer(struct gl_context *ctx,
+                              struct gl_renderbuffer *rb,
+                              GLuint x, GLuint y, GLuint w, GLuint h,
+                              GLbitfield mode,
+                              GLubyte **out_map,
+                              GLint *out_stride);
+
+extern void
+_swrast_unmap_soft_renderbuffer(struct gl_context *ctx,
+                                struct gl_renderbuffer *rb);
+
+extern void
+_swrast_set_renderbuffer_accessors(struct gl_renderbuffer *rb);
+
+
+extern void
+_swrast_add_soft_renderbuffers(struct gl_framebuffer *fb,
+                               GLboolean color,
+                               GLboolean depth,
+                               GLboolean stencil,
+                               GLboolean accum,
+                               GLboolean alpha,
+                               GLboolean aux);
+
+
+#endif /* S_RENDERBUFFER_H */
diff --git a/mesalib/src/mesa/swrast/s_texfetch_tmp.h b/mesalib/src/mesa/swrast/s_texfetch_tmp.h
index e9512b561..877c29c9b 100644
--- a/mesalib/src/mesa/swrast/s_texfetch_tmp.h
+++ b/mesalib/src/mesa/swrast/s_texfetch_tmp.h
@@ -2270,7 +2270,7 @@ static void FETCH(f_z24_s8)( const struct swrast_texture_image *texImage,
 {
    /* only return Z, not stencil data */
    const GLuint *src = TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    texel[0] = ((*src) >> 8) * scale;
    ASSERT(texImage->Base.TexFormat == MESA_FORMAT_Z24_S8 ||
 	  texImage->Base.TexFormat == MESA_FORMAT_Z24_X8);
@@ -2298,7 +2298,7 @@ static void FETCH(f_s8_z24)( const struct swrast_texture_image *texImage,
 {
    /* only return Z, not stencil data */
    const GLuint *src = TEXEL_ADDR(GLuint, texImage, i, j, k, 1);
-   const GLfloat scale = 1.0F / (GLfloat) 0xffffff;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
    texel[0] = ((*src) & 0x00ffffff) * scale;
    ASSERT(texImage->Base.TexFormat == MESA_FORMAT_S8_Z24 ||
 	  texImage->Base.TexFormat == MESA_FORMAT_X8_Z24);
diff --git a/mesalib/src/mesa/swrast/s_zoom.c b/mesalib/src/mesa/swrast/s_zoom.c
index 16bb997f3..f407fdcf4 100644
--- a/mesalib/src/mesa/swrast/s_zoom.c
+++ b/mesalib/src/mesa/swrast/s_zoom.c
@@ -299,11 +299,17 @@ zoom_span( struct gl_context *ctx, GLint imgX, GLint imgY, const SWspan *span,
        * Also, clipping may change the span end value, so store it as well.
        */
       const GLint end = zoomed.end; /* save */
-      GLuint rgbaSave[MAX_WIDTH][4];
+      void *rgbaSave;
       const GLint pixelSize =
          (zoomed.array->ChanType == GL_UNSIGNED_BYTE) ? 4 * sizeof(GLubyte) :
          ((zoomed.array->ChanType == GL_UNSIGNED_SHORT) ? 4 * sizeof(GLushort)
           : 4 * sizeof(GLfloat));
+
+      rgbaSave = malloc(zoomed.end * pixelSize);
+      if (!rgbaSave) {
+         return;
+      }
+
       if (y1 - y0 > 1) {
          memcpy(rgbaSave, zoomed.array->rgba, zoomed.end * pixelSize);
       }
@@ -315,6 +321,8 @@ zoom_span( struct gl_context *ctx, GLint imgX, GLint imgY, const SWspan *span,
             memcpy(zoomed.array->rgba, rgbaSave, zoomed.end * pixelSize);
          }
       }
+
+      free(rgbaSave);
    }
 }
 
diff --git a/mesalib/src/mesa/swrast/swrast.h b/mesalib/src/mesa/swrast/swrast.h
index 17b66c664..dd4055159 100644
--- a/mesalib/src/mesa/swrast/swrast.h
+++ b/mesalib/src/mesa/swrast/swrast.h
@@ -127,9 +127,6 @@ _swrast_BlitFramebuffer(struct gl_context *ctx,
 extern void
 _swrast_Clear(struct gl_context *ctx, GLbitfield buffers);
 
-extern void
-_swrast_Accum(struct gl_context *ctx, GLenum op, GLfloat value);
-
 
 
 /* Reset the stipple counter