i965: Use sample barycentric coordinates with per sample shading
[mesa.git] / src / mesa / drivers / dri / i965 / brw_wm_surface_state.c
index 220af1bf9d5ce0927bc7d82a3f089d66efcb76f1..c7386103c7bd2e60bb5683f950a016e4f15c4b28 100644 (file)
@@ -1,8 +1,8 @@
 /*
  Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
+ Intel funded Tungsten Graphics to
  develop this 3D driver.
+
  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
  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 (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 NONINFRINGEMENT.
  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
  **********************************************************************/
  /*
   * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Keith Whitwell <keithw@vmware.com>
   */
-                   
 
+
+#include "main/context.h"
+#include "main/blend.h"
 #include "main/mtypes.h"
 #include "main/samplerobj.h"
 #include "program/prog_parameter.h"
@@ -49,583 +51,30 @@ GLuint
 translate_tex_target(GLenum target)
 {
    switch (target) {
-   case GL_TEXTURE_1D: 
+   case GL_TEXTURE_1D:
    case GL_TEXTURE_1D_ARRAY_EXT:
       return BRW_SURFACE_1D;
 
-   case GL_TEXTURE_RECTANGLE_NV: 
+   case GL_TEXTURE_RECTANGLE_NV:
       return BRW_SURFACE_2D;
 
-   case GL_TEXTURE_2D: 
+   case GL_TEXTURE_2D:
    case GL_TEXTURE_2D_ARRAY_EXT:
    case GL_TEXTURE_EXTERNAL_OES:
+   case GL_TEXTURE_2D_MULTISAMPLE:
+   case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
       return BRW_SURFACE_2D;
 
-   case GL_TEXTURE_3D: 
+   case GL_TEXTURE_3D:
       return BRW_SURFACE_3D;
 
-   case GL_TEXTURE_CUBE_MAP: 
+   case GL_TEXTURE_CUBE_MAP:
+   case GL_TEXTURE_CUBE_MAP_ARRAY:
       return BRW_SURFACE_CUBE;
 
-   default: 
-      assert(0); 
-      return 0;
-   }
-}
-
-struct surface_format_info {
-   bool exists;
-   int sampling;
-   int filtering;
-   int shadow_compare;
-   int chroma_key;
-   int render_target;
-   int alpha_blend;
-   int input_vb;
-   int streamed_output_vb;
-   int color_processing;
-};
-
-/* This macro allows us to write the table almost as it appears in the PRM,
- * while restructuring it to turn it into the C code we want.
- */
-#define SF(sampl, filt, shad, ck, rt, ab, vb, so, color, sf) \
-   [sf] = { true, sampl, filt, shad, ck, rt, ab, vb, so, color },
-
-#define Y 0
-#define x 999
-/**
- * This is the table of support for surface (texture, renderbuffer, and vertex
- * buffer, but not depthbuffer) formats across the various hardware generations.
- *
- * The table is formatted to match the documentation, except that the docs have
- * this ridiculous mapping of Y[*+~^#&] for "supported on DevWhatever".  To put
- * it in our table, here's the mapping:
- *
- * Y*: 45
- * Y+: 45 (g45/gm45)
- * Y~: 50 (gen5)
- * Y^: 60 (gen6)
- * Y#: 70 (gen7)
- *
- * See page 88 of the Sandybridge PRM VOL4_Part1 PDF.
- */
-const struct surface_format_info surface_formats[] = {
-/* smpl filt shad CK  RT  AB  VB  SO  color */
-   SF( Y, 50,  x,  x,  Y,  Y,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32A32_FLOAT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32A32_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32A32_UINT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32A32_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32A32_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R64G64_FLOAT)
-   SF( Y, 50,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R32G32B32X32_FLOAT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32A32_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32A32_USCALED)
-   SF( Y, 50,  x,  x,  x,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32_FLOAT)
-   SF( Y,  x,  x,  x,  x,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32_SINT)
-   SF( Y,  x,  x,  x,  x,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32B32_UINT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32B32_USCALED)
-   SF( Y,  Y,  x,  x,  Y, 45,  Y,  x, 60, BRW_SURFACEFORMAT_R16G16B16A16_UNORM)
-   SF( Y,  Y,  x,  x,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_UINT)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_FLOAT)
-   SF( Y, 50,  x,  x,  Y,  Y,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32_FLOAT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32G32_UINT)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R32_FLOAT_X8X24_TYPELESS)
-   SF( Y,  x,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_X32_TYPELESS_G8X24_UINT)
-   SF( Y, 50,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L32A32_FLOAT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R64_FLOAT)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R16G16B16X16_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R16G16B16X16_FLOAT)
-   SF( Y, 50,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A32X32_FLOAT)
-   SF( Y, 50,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L32X32_FLOAT)
-   SF( Y, 50,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I32X32_FLOAT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16A16_USCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32G32_USCALED)
-   SF( Y,  Y,  x,  Y,  Y,  Y,  Y,  x, 60, BRW_SURFACEFORMAT_B8G8R8A8_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB)
-/* smpl filt shad CK  RT  AB  VB  SO  color */
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x, 60, BRW_SURFACEFORMAT_R10G10B10A2_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x, 60, BRW_SURFACEFORMAT_R10G10B10A2_UNORM_SRGB)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R10G10B10A2_UINT)
-   SF( Y,  Y,  x,  x,  x,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R10G10B10_SNORM_A2_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x, 60, BRW_SURFACEFORMAT_R8G8B8A8_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x, 60, BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8A8_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8A8_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8A8_UINT)
-   SF( Y,  Y,  x,  x,  Y, 45,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_UNORM)
-   SF( Y,  Y,  x,  x,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_UINT)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_FLOAT)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x, 60, BRW_SURFACEFORMAT_B10G10R10A2_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x, 60, BRW_SURFACEFORMAT_B10G10R10A2_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R11G11B10_FLOAT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  Y,  x, BRW_SURFACEFORMAT_R32_UINT)
-   SF( Y, 50,  Y,  x,  Y,  Y,  Y,  Y,  x, BRW_SURFACEFORMAT_R32_FLOAT)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R24_UNORM_X8_TYPELESS)
-   SF( Y,  x,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_X24_TYPELESS_G8_UINT)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L16A16_UNORM)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I24X8_UNORM)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L24X8_UNORM)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A24X8_UNORM)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I32_FLOAT)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L32_FLOAT)
-   SF( Y, 50,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A32_FLOAT)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x, 60, BRW_SURFACEFORMAT_B8G8R8X8_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_B8G8R8X8_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R8G8B8X8_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R8G8B8X8_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R9G9B9E5_SHAREDEXP)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_B10G10R10X2_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L16A16_FLOAT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32_SNORM)
-/* smpl filt shad CK  RT  AB  VB  SO  color */
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R10G10B10X2_USCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8A8_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8A8_USCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16_USCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R32_USCALED)
-   SF( Y,  Y,  x,  Y,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G6R5_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G6R5_UNORM_SRGB)
-   SF( Y,  Y,  x,  Y,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G5R5A1_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G5R5A1_UNORM_SRGB)
-   SF( Y,  Y,  x,  Y,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B4G4R4A4_UNORM)
-   SF( Y,  Y,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B4G4R4A4_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_UNORM)
-   SF( Y,  Y,  x,  Y,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_UINT)
-   SF( Y,  Y,  Y,  x,  Y, 45,  Y,  x, 70, BRW_SURFACEFORMAT_R16_UNORM)
-   SF( Y,  Y,  x,  x,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R16_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16_UINT)
-   SF( Y,  Y,  x,  x,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R16_FLOAT)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I16_UNORM)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L16_UNORM)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A16_UNORM)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L8A8_UNORM)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I16_FLOAT)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L16_FLOAT)
-   SF( Y,  Y,  Y,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A16_FLOAT)
-   SF(45, 45,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L8A8_UNORM_SRGB)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R5G5_SNORM_B6_UNORM)
-   SF( x,  x,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G5R5X1_UNORM)
-   SF( x,  x,  x,  x,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_B5G5R5X1_UNORM_SRGB)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8_USCALED)
-/* smpl filt shad CK  RT  AB  VB  SO  color */
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16_USCALED)
-   SF( Y,  Y,  x, 45,  Y,  Y,  Y,  x,  x, BRW_SURFACEFORMAT_R8_UNORM)
-   SF( Y,  Y,  x,  x,  Y, 60,  Y,  x,  x, BRW_SURFACEFORMAT_R8_SNORM)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8_SINT)
-   SF( Y,  x,  x,  x,  Y,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8_UINT)
-   SF( Y,  Y,  x,  Y,  Y,  Y,  x,  x,  x, BRW_SURFACEFORMAT_A8_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_I8_UNORM)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L8_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_P4A4_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_A4P4_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8_USCALED)
-   SF(45, 45,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_L8_UNORM_SRGB)
-   SF(45, 45,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_DXT1_RGB_SRGB)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_R1_UINT)
-   SF( Y,  Y,  x,  Y,  Y,  x,  x,  x, 60, BRW_SURFACEFORMAT_YCRCB_NORMAL)
-   SF( Y,  Y,  x,  Y,  Y,  x,  x,  x, 60, BRW_SURFACEFORMAT_YCRCB_SWAPUVY)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC1_UNORM)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC2_UNORM)
-   SF( Y,  Y,  x,  Y,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC3_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC4_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC5_UNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC1_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC2_UNORM_SRGB)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC3_UNORM_SRGB)
-   SF( Y,  x,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_MONO8)
-   SF( Y,  Y,  x,  x,  Y,  x,  x,  x, 60, BRW_SURFACEFORMAT_YCRCB_SWAPUV)
-   SF( Y,  Y,  x,  x,  Y,  x,  x,  x, 60, BRW_SURFACEFORMAT_YCRCB_SWAPY)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_DXT1_RGB)
-/* smpl filt shad CK  RT  AB  VB  SO  color */
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_FXT1)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R8G8B8_USCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R64G64B64A64_FLOAT)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R64G64B64_FLOAT)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC4_SNORM)
-   SF( Y,  Y,  x,  x,  x,  x,  x,  x,  x, BRW_SURFACEFORMAT_BC5_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16_UNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16_SNORM)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16_SSCALED)
-   SF( x,  x,  x,  x,  x,  x,  Y,  x,  x, BRW_SURFACEFORMAT_R16G16B16_USCALED)
-};
-#undef x
-#undef Y
-
-uint32_t
-brw_format_for_mesa_format(gl_format mesa_format)
-{
-   /* This table is ordered according to the enum ordering in formats.h.  We do
-    * expect that enum to be extended without our explicit initialization
-    * staying in sync, so we initialize to 0 even though
-    * BRW_SURFACEFORMAT_R32G32B32A32_FLOAT happens to also be 0.
-    */
-   static const uint32_t table[MESA_FORMAT_COUNT] =
-   {
-      [MESA_FORMAT_RGBA8888] = 0,
-      [MESA_FORMAT_RGBA8888_REV] = BRW_SURFACEFORMAT_R8G8B8A8_UNORM,
-      [MESA_FORMAT_ARGB8888] = BRW_SURFACEFORMAT_B8G8R8A8_UNORM,
-      [MESA_FORMAT_ARGB8888_REV] = 0,
-      [MESA_FORMAT_RGBX8888] = 0,
-      [MESA_FORMAT_RGBX8888_REV] = BRW_SURFACEFORMAT_R8G8B8X8_UNORM,
-      [MESA_FORMAT_XRGB8888] = BRW_SURFACEFORMAT_B8G8R8X8_UNORM,
-      [MESA_FORMAT_XRGB8888_REV] = 0,
-      [MESA_FORMAT_RGB888] = 0,
-      [MESA_FORMAT_BGR888] = 0,
-      [MESA_FORMAT_RGB565] = BRW_SURFACEFORMAT_B5G6R5_UNORM,
-      [MESA_FORMAT_RGB565_REV] = 0,
-      [MESA_FORMAT_ARGB4444] = BRW_SURFACEFORMAT_B4G4R4A4_UNORM,
-      [MESA_FORMAT_ARGB4444_REV] = 0,
-      [MESA_FORMAT_RGBA5551] = 0,
-      [MESA_FORMAT_ARGB1555] = BRW_SURFACEFORMAT_B5G5R5A1_UNORM,
-      [MESA_FORMAT_ARGB1555_REV] = 0,
-      [MESA_FORMAT_AL44] = 0,
-      [MESA_FORMAT_AL88] = BRW_SURFACEFORMAT_L8A8_UNORM,
-      [MESA_FORMAT_AL88_REV] = 0,
-      [MESA_FORMAT_AL1616] = BRW_SURFACEFORMAT_L16A16_UNORM,
-      [MESA_FORMAT_AL1616_REV] = 0,
-      [MESA_FORMAT_RGB332] = 0,
-      [MESA_FORMAT_A8] = BRW_SURFACEFORMAT_A8_UNORM,
-      [MESA_FORMAT_A16] = BRW_SURFACEFORMAT_A16_UNORM,
-      [MESA_FORMAT_L8] = BRW_SURFACEFORMAT_L8_UNORM,
-      [MESA_FORMAT_L16] = BRW_SURFACEFORMAT_L16_UNORM,
-      [MESA_FORMAT_I8] = BRW_SURFACEFORMAT_I8_UNORM,
-      [MESA_FORMAT_I16] = BRW_SURFACEFORMAT_I16_UNORM,
-      [MESA_FORMAT_YCBCR_REV] = BRW_SURFACEFORMAT_YCRCB_NORMAL,
-      [MESA_FORMAT_YCBCR] = BRW_SURFACEFORMAT_YCRCB_SWAPUVY,
-      [MESA_FORMAT_R8] = BRW_SURFACEFORMAT_R8_UNORM,
-      [MESA_FORMAT_GR88] = BRW_SURFACEFORMAT_R8G8_UNORM,
-      [MESA_FORMAT_RG88] = 0,
-      [MESA_FORMAT_R16] = BRW_SURFACEFORMAT_R16_UNORM,
-      [MESA_FORMAT_RG1616] = BRW_SURFACEFORMAT_R16G16_UNORM,
-      [MESA_FORMAT_RG1616_REV] = 0,
-      [MESA_FORMAT_ARGB2101010] = BRW_SURFACEFORMAT_B10G10R10A2_UNORM,
-      [MESA_FORMAT_ABGR2101010_UINT] = BRW_SURFACEFORMAT_R10G10B10A2_UINT,
-      [MESA_FORMAT_Z24_S8] = 0,
-      [MESA_FORMAT_S8_Z24] = 0,
-      [MESA_FORMAT_Z16] = 0,
-      [MESA_FORMAT_X8_Z24] = 0,
-      [MESA_FORMAT_Z24_X8] = 0,
-      [MESA_FORMAT_Z32] = 0,
-      [MESA_FORMAT_S8] = 0,
-
-      [MESA_FORMAT_SRGB8] = 0,
-      [MESA_FORMAT_SRGBA8] = 0,
-      [MESA_FORMAT_SARGB8] = BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB,
-      [MESA_FORMAT_SL8] = BRW_SURFACEFORMAT_L8_UNORM_SRGB,
-      [MESA_FORMAT_SLA8] = BRW_SURFACEFORMAT_L8A8_UNORM_SRGB,
-      [MESA_FORMAT_SRGB_DXT1] = BRW_SURFACEFORMAT_DXT1_RGB_SRGB,
-      [MESA_FORMAT_SRGBA_DXT1] = BRW_SURFACEFORMAT_BC1_UNORM_SRGB,
-      [MESA_FORMAT_SRGBA_DXT3] = BRW_SURFACEFORMAT_BC2_UNORM_SRGB,
-      [MESA_FORMAT_SRGBA_DXT5] = BRW_SURFACEFORMAT_BC3_UNORM_SRGB,
-
-      [MESA_FORMAT_RGB_FXT1] = BRW_SURFACEFORMAT_FXT1,
-      [MESA_FORMAT_RGBA_FXT1] = BRW_SURFACEFORMAT_FXT1,
-      [MESA_FORMAT_RGB_DXT1] = BRW_SURFACEFORMAT_DXT1_RGB,
-      [MESA_FORMAT_RGBA_DXT1] = BRW_SURFACEFORMAT_BC1_UNORM,
-      [MESA_FORMAT_RGBA_DXT3] = BRW_SURFACEFORMAT_BC2_UNORM,
-      [MESA_FORMAT_RGBA_DXT5] = BRW_SURFACEFORMAT_BC3_UNORM,
-
-      [MESA_FORMAT_RGBA_FLOAT32] = BRW_SURFACEFORMAT_R32G32B32A32_FLOAT,
-      [MESA_FORMAT_RGBA_FLOAT16] = BRW_SURFACEFORMAT_R16G16B16A16_FLOAT,
-      [MESA_FORMAT_RGB_FLOAT32] = 0,
-      [MESA_FORMAT_RGB_FLOAT16] = 0,
-      [MESA_FORMAT_ALPHA_FLOAT32] = BRW_SURFACEFORMAT_A32_FLOAT,
-      [MESA_FORMAT_ALPHA_FLOAT16] = BRW_SURFACEFORMAT_A16_FLOAT,
-      [MESA_FORMAT_LUMINANCE_FLOAT32] = BRW_SURFACEFORMAT_L32_FLOAT,
-      [MESA_FORMAT_LUMINANCE_FLOAT16] = BRW_SURFACEFORMAT_L16_FLOAT,
-      [MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32] = BRW_SURFACEFORMAT_L32A32_FLOAT,
-      [MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16] = BRW_SURFACEFORMAT_L16A16_FLOAT,
-      [MESA_FORMAT_INTENSITY_FLOAT32] = BRW_SURFACEFORMAT_I32_FLOAT,
-      [MESA_FORMAT_INTENSITY_FLOAT16] = BRW_SURFACEFORMAT_I16_FLOAT,
-      [MESA_FORMAT_R_FLOAT32] = BRW_SURFACEFORMAT_R32_FLOAT,
-      [MESA_FORMAT_R_FLOAT16] = BRW_SURFACEFORMAT_R16_FLOAT,
-      [MESA_FORMAT_RG_FLOAT32] = BRW_SURFACEFORMAT_R32G32_FLOAT,
-      [MESA_FORMAT_RG_FLOAT16] = BRW_SURFACEFORMAT_R16G16_FLOAT,
-
-      [MESA_FORMAT_ALPHA_UINT8] = 0,
-      [MESA_FORMAT_ALPHA_UINT16] = 0,
-      [MESA_FORMAT_ALPHA_UINT32] = 0,
-      [MESA_FORMAT_ALPHA_INT8] = 0,
-      [MESA_FORMAT_ALPHA_INT16] = 0,
-      [MESA_FORMAT_ALPHA_INT32] = 0,
-
-      [MESA_FORMAT_INTENSITY_UINT8] = 0,
-      [MESA_FORMAT_INTENSITY_UINT16] = 0,
-      [MESA_FORMAT_INTENSITY_UINT32] = 0,
-      [MESA_FORMAT_INTENSITY_INT8] = 0,
-      [MESA_FORMAT_INTENSITY_INT16] = 0,
-      [MESA_FORMAT_INTENSITY_INT32] = 0,
-
-      [MESA_FORMAT_LUMINANCE_UINT8] = 0,
-      [MESA_FORMAT_LUMINANCE_UINT16] = 0,
-      [MESA_FORMAT_LUMINANCE_UINT32] = 0,
-      [MESA_FORMAT_LUMINANCE_INT8] = 0,
-      [MESA_FORMAT_LUMINANCE_INT16] = 0,
-      [MESA_FORMAT_LUMINANCE_INT32] = 0,
-
-      [MESA_FORMAT_LUMINANCE_ALPHA_UINT8] = 0,
-      [MESA_FORMAT_LUMINANCE_ALPHA_UINT16] = 0,
-      [MESA_FORMAT_LUMINANCE_ALPHA_UINT32] = 0,
-      [MESA_FORMAT_LUMINANCE_ALPHA_INT8] = 0,
-      [MESA_FORMAT_LUMINANCE_ALPHA_INT16] = 0,
-      [MESA_FORMAT_LUMINANCE_ALPHA_INT32] = 0,
-
-      [MESA_FORMAT_R_INT8] = BRW_SURFACEFORMAT_R8_SINT,
-      [MESA_FORMAT_RG_INT8] = BRW_SURFACEFORMAT_R8G8_SINT,
-      [MESA_FORMAT_RGB_INT8] = 0,
-      [MESA_FORMAT_RGBA_INT8] = BRW_SURFACEFORMAT_R8G8B8A8_SINT,
-      [MESA_FORMAT_R_INT16] = BRW_SURFACEFORMAT_R16_SINT,
-      [MESA_FORMAT_RG_INT16] = BRW_SURFACEFORMAT_R16G16_SINT,
-      [MESA_FORMAT_RGB_INT16] = 0,
-      [MESA_FORMAT_RGBA_INT16] = BRW_SURFACEFORMAT_R16G16B16A16_SINT,
-      [MESA_FORMAT_R_INT32] = BRW_SURFACEFORMAT_R32_SINT,
-      [MESA_FORMAT_RG_INT32] = BRW_SURFACEFORMAT_R32G32_SINT,
-      [MESA_FORMAT_RGB_INT32] = BRW_SURFACEFORMAT_R32G32B32_SINT,
-      [MESA_FORMAT_RGBA_INT32] = BRW_SURFACEFORMAT_R32G32B32A32_SINT,
-
-      [MESA_FORMAT_R_UINT8] = BRW_SURFACEFORMAT_R8_UINT,
-      [MESA_FORMAT_RG_UINT8] = BRW_SURFACEFORMAT_R8G8_UINT,
-      [MESA_FORMAT_RGB_UINT8] = 0,
-      [MESA_FORMAT_RGBA_UINT8] = BRW_SURFACEFORMAT_R8G8B8A8_UINT,
-      [MESA_FORMAT_R_UINT16] = BRW_SURFACEFORMAT_R16_UINT,
-      [MESA_FORMAT_RG_UINT16] = BRW_SURFACEFORMAT_R16G16_UINT,
-      [MESA_FORMAT_RGB_UINT16] = 0,
-      [MESA_FORMAT_RGBA_UINT16] = BRW_SURFACEFORMAT_R16G16B16A16_UINT,
-      [MESA_FORMAT_R_UINT32] = BRW_SURFACEFORMAT_R32_UINT,
-      [MESA_FORMAT_RG_UINT32] = BRW_SURFACEFORMAT_R32G32_UINT,
-      [MESA_FORMAT_RGB_UINT32] = BRW_SURFACEFORMAT_R32G32B32_UINT,
-      [MESA_FORMAT_RGBA_UINT32] = BRW_SURFACEFORMAT_R32G32B32A32_UINT,
-
-      [MESA_FORMAT_DUDV8] = BRW_SURFACEFORMAT_R8G8_SNORM,
-      [MESA_FORMAT_SIGNED_R8] = BRW_SURFACEFORMAT_R8_SNORM,
-      [MESA_FORMAT_SIGNED_RG88_REV] = BRW_SURFACEFORMAT_R8G8_SNORM,
-      [MESA_FORMAT_SIGNED_RGBX8888] = 0,
-      [MESA_FORMAT_SIGNED_RGBA8888] = 0,
-      [MESA_FORMAT_SIGNED_RGBA8888_REV] = BRW_SURFACEFORMAT_R8G8B8A8_SNORM,
-      [MESA_FORMAT_SIGNED_R16] = BRW_SURFACEFORMAT_R16_SNORM,
-      [MESA_FORMAT_SIGNED_GR1616] = BRW_SURFACEFORMAT_R16G16_SNORM,
-      [MESA_FORMAT_SIGNED_RGB_16] = 0,
-      [MESA_FORMAT_SIGNED_RGBA_16] = BRW_SURFACEFORMAT_R16G16B16A16_SNORM,
-      [MESA_FORMAT_RGBA_16] = BRW_SURFACEFORMAT_R16G16B16A16_UNORM,
-
-      [MESA_FORMAT_RED_RGTC1] = BRW_SURFACEFORMAT_BC4_UNORM,
-      [MESA_FORMAT_SIGNED_RED_RGTC1] = BRW_SURFACEFORMAT_BC4_SNORM,
-      [MESA_FORMAT_RG_RGTC2] = BRW_SURFACEFORMAT_BC5_UNORM,
-      [MESA_FORMAT_SIGNED_RG_RGTC2] = BRW_SURFACEFORMAT_BC5_SNORM,
-
-      [MESA_FORMAT_L_LATC1] = 0,
-      [MESA_FORMAT_SIGNED_L_LATC1] = 0,
-      [MESA_FORMAT_LA_LATC2] = 0,
-      [MESA_FORMAT_SIGNED_LA_LATC2] = 0,
-
-      [MESA_FORMAT_SIGNED_A8] = 0,
-      [MESA_FORMAT_SIGNED_L8] = 0,
-      [MESA_FORMAT_SIGNED_AL88] = 0,
-      [MESA_FORMAT_SIGNED_I8] = 0,
-      [MESA_FORMAT_SIGNED_A16] = 0,
-      [MESA_FORMAT_SIGNED_L16] = 0,
-      [MESA_FORMAT_SIGNED_AL1616] = 0,
-      [MESA_FORMAT_SIGNED_I16] = 0,
-
-      [MESA_FORMAT_RGB9_E5_FLOAT] = BRW_SURFACEFORMAT_R9G9B9E5_SHAREDEXP,
-      [MESA_FORMAT_R11_G11_B10_FLOAT] = BRW_SURFACEFORMAT_R11G11B10_FLOAT,
-
-      [MESA_FORMAT_Z32_FLOAT] = 0,
-      [MESA_FORMAT_Z32_FLOAT_X24S8] = 0,
-   };
-   assert(mesa_format < MESA_FORMAT_COUNT);
-   return table[mesa_format];
-}
-
-void
-brw_init_surface_formats(struct brw_context *brw)
-{
-   struct intel_context *intel = &brw->intel;
-   struct gl_context *ctx = &intel->ctx;
-   int gen;
-   gl_format format;
-
-   gen = intel->gen * 10;
-   if (intel->is_g4x)
-      gen += 5;
-
-   for (format = MESA_FORMAT_NONE + 1; format < MESA_FORMAT_COUNT; format++) {
-      uint32_t texture, render;
-      const struct surface_format_info *rinfo, *tinfo;
-      bool is_integer = _mesa_is_format_integer_color(format);
-
-      render = texture = brw_format_for_mesa_format(format);
-      tinfo = &surface_formats[texture];
-
-      /* The value of BRW_SURFACEFORMAT_R32G32B32A32_FLOAT is 0, so don't skip
-       * it.
-       */
-      if (texture == 0 && format != MESA_FORMAT_RGBA_FLOAT32)
-        continue;
-
-      if (gen >= tinfo->sampling && (gen >= tinfo->filtering || is_integer))
-        ctx->TextureFormatSupported[format] = true;
-
-      /* Re-map some render target formats to make them supported when they
-       * wouldn't be using their format for texturing.
-       */
-      switch (render) {
-        /* For these formats, we just need to read/write the first
-         * channel into R, which is to say that we just treat them as
-         * GL_RED.
-         */
-      case BRW_SURFACEFORMAT_I32_FLOAT:
-      case BRW_SURFACEFORMAT_L32_FLOAT:
-        render = BRW_SURFACEFORMAT_R32_FLOAT;
-        break;
-      case BRW_SURFACEFORMAT_I16_FLOAT:
-      case BRW_SURFACEFORMAT_L16_FLOAT:
-        render = BRW_SURFACEFORMAT_R16_FLOAT;
-        break;
-      case BRW_SURFACEFORMAT_B8G8R8X8_UNORM:
-        /* XRGB is handled as ARGB because the chips in this family
-         * cannot render to XRGB targets.  This means that we have to
-         * mask writes to alpha (ala glColorMask) and reconfigure the
-         * alpha blending hardware to use GL_ONE (or GL_ZERO) for
-         * cases where GL_DST_ALPHA (or GL_ONE_MINUS_DST_ALPHA) is
-         * used.
-         */
-        render = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-        break;
-      }
-
-      rinfo = &surface_formats[render];
-
-      /* Note that GL_EXT_texture_integer says that blending doesn't occur for
-       * integer, so we don't need hardware support for blending on it.  Other
-       * than that, GL in general requires alpha blending for render targets,
-       * even though we don't support it for some formats.
-       */
-      if (gen >= rinfo->render_target &&
-         (gen >= rinfo->alpha_blend || is_integer)) {
-        brw->render_target_format[format] = render;
-        brw->format_supported_as_render_target[format] = true;
-      }
-   }
-
-   /* We will check this table for FBO completeness, but the surface format
-    * table above only covered color rendering.
-    */
-   brw->format_supported_as_render_target[MESA_FORMAT_S8_Z24] = true;
-   brw->format_supported_as_render_target[MESA_FORMAT_X8_Z24] = true;
-   brw->format_supported_as_render_target[MESA_FORMAT_S8] = true;
-   brw->format_supported_as_render_target[MESA_FORMAT_Z16] = true;
-   brw->format_supported_as_render_target[MESA_FORMAT_Z32_FLOAT] = true;
-   brw->format_supported_as_render_target[MESA_FORMAT_Z32_FLOAT_X24S8] = true;
-
-   /* We remap depth formats to a supported texturing format in
-    * translate_tex_format().
-    */
-   ctx->TextureFormatSupported[MESA_FORMAT_S8_Z24] = true;
-   ctx->TextureFormatSupported[MESA_FORMAT_X8_Z24] = true;
-   ctx->TextureFormatSupported[MESA_FORMAT_Z32_FLOAT] = true;
-   ctx->TextureFormatSupported[MESA_FORMAT_Z32_FLOAT_X24S8] = true;
-   ctx->TextureFormatSupported[MESA_FORMAT_Z16] = true;
-
-   /* On hardware that lacks support for ETC1, we map ETC1 to RGBX
-    * during glCompressedTexImage2D(). See intel_mipmap_tree::wraps_etc1.
-    */
-   ctx->TextureFormatSupported[MESA_FORMAT_ETC1_RGB8] = true;
-}
-
-bool
-brw_render_target_supported(struct intel_context *intel,
-                           struct gl_renderbuffer *rb)
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   gl_format format = rb->Format;
-
-   /* Many integer formats are promoted to RGBA (like XRGB8888 is), which means
-    * we would consider them renderable even though we don't have surface
-    * support for their alpha behavior and don't have the blending unit
-    * available to fake it like we do for XRGB8888.  Force them to being
-    * unsupported.
-    */
-   if ((rb->_BaseFormat != GL_RGBA &&
-       rb->_BaseFormat != GL_RG &&
-       rb->_BaseFormat != GL_RED) && _mesa_is_format_integer_color(format))
-      return false;
-
-   /* Under some conditions, MSAA is not supported for formats whose width is
-    * more than 64 bits.
-    */
-   if (rb->NumSamples > 0 && _mesa_get_format_bytes(format) > 8) {
-      /* Gen6: MSAA on >64 bit formats is unsupported. */
-      if (intel->gen <= 6)
-         return false;
-
-      /* Gen7: 8x MSAA on >64 bit formats is unsupported. */
-      if (rb->NumSamples >= 8)
-         return false;
-   }
-
-   return brw->format_supported_as_render_target[format];
-}
-
-GLuint
-translate_tex_format(gl_format mesa_format,
-                    GLenum internal_format,
-                    GLenum depth_mode,
-                    GLenum srgb_decode)
-{
-   if (srgb_decode == GL_SKIP_DECODE_EXT)
-      mesa_format = _mesa_get_srgb_format_linear(mesa_format);
-
-   switch( mesa_format ) {
-
-   case MESA_FORMAT_Z16:
-      return BRW_SURFACEFORMAT_I16_UNORM;
-
-   case MESA_FORMAT_S8_Z24:
-   case MESA_FORMAT_X8_Z24:
-      return BRW_SURFACEFORMAT_I24X8_UNORM;
-
-   case MESA_FORMAT_Z32_FLOAT:
-      return BRW_SURFACEFORMAT_I32_FLOAT;
-
-   case MESA_FORMAT_Z32_FLOAT_X24S8:
-      return BRW_SURFACEFORMAT_R32G32_FLOAT;
-
-   case MESA_FORMAT_RGBA_FLOAT32:
-      /* The value of this BRW_SURFACEFORMAT is 0, which tricks the
-       * assertion below.
-       */
-      return BRW_SURFACEFORMAT_R32G32B32A32_FLOAT;
-
    default:
-      assert(brw_format_for_mesa_format(mesa_format) != 0);
-      return brw_format_for_mesa_format(mesa_format);
+      assert(0);
+      return 0;
    }
 }
 
@@ -653,110 +102,217 @@ brw_get_surface_num_multisamples(unsigned num_samples)
 }
 
 
+/**
+ * Compute the combination of DEPTH_TEXTURE_MODE and EXT_texture_swizzle
+ * swizzling.
+ */
+int
+brw_get_texture_swizzle(const struct gl_context *ctx,
+                        const struct gl_texture_object *t)
+{
+   const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
+
+   int swizzles[SWIZZLE_NIL + 1] = {
+      SWIZZLE_X,
+      SWIZZLE_Y,
+      SWIZZLE_Z,
+      SWIZZLE_W,
+      SWIZZLE_ZERO,
+      SWIZZLE_ONE,
+      SWIZZLE_NIL
+   };
+
+   if (img->_BaseFormat == GL_DEPTH_COMPONENT ||
+       img->_BaseFormat == GL_DEPTH_STENCIL) {
+      GLenum depth_mode = t->DepthMode;
+
+      /* In ES 3.0, DEPTH_TEXTURE_MODE is expected to be GL_RED for textures
+       * with depth component data specified with a sized internal format.
+       * Otherwise, it's left at the old default, GL_LUMINANCE.
+       */
+      if (_mesa_is_gles3(ctx) &&
+          img->InternalFormat != GL_DEPTH_COMPONENT &&
+          img->InternalFormat != GL_DEPTH_STENCIL) {
+         depth_mode = GL_RED;
+      }
+
+      switch (depth_mode) {
+      case GL_ALPHA:
+         swizzles[0] = SWIZZLE_ZERO;
+         swizzles[1] = SWIZZLE_ZERO;
+         swizzles[2] = SWIZZLE_ZERO;
+         swizzles[3] = SWIZZLE_X;
+         break;
+      case GL_LUMINANCE:
+         swizzles[0] = SWIZZLE_X;
+         swizzles[1] = SWIZZLE_X;
+         swizzles[2] = SWIZZLE_X;
+         swizzles[3] = SWIZZLE_ONE;
+         break;
+      case GL_INTENSITY:
+         swizzles[0] = SWIZZLE_X;
+         swizzles[1] = SWIZZLE_X;
+         swizzles[2] = SWIZZLE_X;
+         swizzles[3] = SWIZZLE_X;
+         break;
+      case GL_RED:
+         swizzles[0] = SWIZZLE_X;
+         swizzles[1] = SWIZZLE_ZERO;
+         swizzles[2] = SWIZZLE_ZERO;
+         swizzles[3] = SWIZZLE_ONE;
+         break;
+      }
+   }
+
+   /* If the texture's format is alpha-only, force R, G, and B to
+    * 0.0. Similarly, if the texture's format has no alpha channel,
+    * force the alpha value read to 1.0. This allows for the
+    * implementation to use an RGBA texture for any of these formats
+    * without leaking any unexpected values.
+    */
+   switch (img->_BaseFormat) {
+   case GL_ALPHA:
+      swizzles[0] = SWIZZLE_ZERO;
+      swizzles[1] = SWIZZLE_ZERO;
+      swizzles[2] = SWIZZLE_ZERO;
+      break;
+   case GL_RED:
+   case GL_RG:
+   case GL_RGB:
+      if (_mesa_get_format_bits(img->TexFormat, GL_ALPHA_BITS) > 0)
+         swizzles[3] = SWIZZLE_ONE;
+      break;
+   }
+
+   return MAKE_SWIZZLE4(swizzles[GET_SWZ(t->_Swizzle, 0)],
+                        swizzles[GET_SWZ(t->_Swizzle, 1)],
+                        swizzles[GET_SWZ(t->_Swizzle, 2)],
+                        swizzles[GET_SWZ(t->_Swizzle, 3)]);
+}
+
 static void
-brw_update_buffer_texture_surface(struct gl_context *ctx, GLuint unit)
+gen4_emit_buffer_surface_state(struct brw_context *brw,
+                               uint32_t *out_offset,
+                               drm_intel_bo *bo,
+                               unsigned buffer_offset,
+                               unsigned surface_format,
+                               unsigned buffer_size,
+                               unsigned pitch,
+                               unsigned mocs,
+                               bool rw)
+{
+   uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
+                                    6 * 4, 32, out_offset);
+   memset(surf, 0, 6 * 4);
+
+   surf[0] = BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
+             surface_format << BRW_SURFACE_FORMAT_SHIFT |
+             (brw->gen >= 6 ? BRW_SURFACE_RC_READ_WRITE : 0);
+   surf[1] = (bo ? bo->offset64 : 0) + buffer_offset; /* reloc */
+   surf[2] = (buffer_size & 0x7f) << BRW_SURFACE_WIDTH_SHIFT |
+             ((buffer_size >> 7) & 0x1fff) << BRW_SURFACE_HEIGHT_SHIFT;
+   surf[3] = ((buffer_size >> 20) & 0x7f) << BRW_SURFACE_DEPTH_SHIFT |
+             (pitch - 1) << BRW_SURFACE_PITCH_SHIFT;
+
+   /* Emit relocation to surface contents.  The 965 PRM, Volume 4, section
+    * 5.1.2 "Data Cache" says: "the data cache does not exist as a separate
+    * physical cache.  It is mapped in hardware to the sampler cache."
+    */
+   if (bo) {
+      drm_intel_bo_emit_reloc(brw->batch.bo, *out_offset + 4,
+                              bo, buffer_offset,
+                              I915_GEM_DOMAIN_SAMPLER,
+                              (rw ? I915_GEM_DOMAIN_SAMPLER : 0));
+   }
+}
+
+void
+brw_update_buffer_texture_surface(struct gl_context *ctx,
+                                  unsigned unit,
+                                  uint32_t *surf_offset)
 {
    struct brw_context *brw = brw_context(ctx);
-   struct intel_context *intel = &brw->intel;
    struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
-   const GLuint surf_index = SURF_INDEX_TEXTURE(unit);
-   uint32_t *surf;
    struct intel_buffer_object *intel_obj =
       intel_buffer_object(tObj->BufferObject);
-   drm_intel_bo *bo = intel_obj ? intel_obj->buffer : NULL;
+   uint32_t size = tObj->BufferSize;
+   drm_intel_bo *bo = NULL;
    gl_format format = tObj->_BufferObjectFormat;
    uint32_t brw_format = brw_format_for_mesa_format(format);
    int texel_size = _mesa_get_format_bytes(format);
 
+   if (intel_obj) {
+      size = MIN2(size, intel_obj->Base.Size);
+      bo = intel_bufferobj_buffer(brw, intel_obj, tObj->BufferOffset, size);
+   }
+
    if (brw_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
       _mesa_problem(NULL, "bad format %s for texture buffer\n",
                    _mesa_get_format_name(format));
    }
 
-   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
-                         6 * 4, 32, &brw->wm.surf_offset[surf_index]);
-
-   surf[0] = (BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
-             (brw_format_for_mesa_format(format) << BRW_SURFACE_FORMAT_SHIFT));
-
-   if (intel->gen >= 6)
-      surf[0] |= BRW_SURFACE_RC_READ_WRITE;
-
-   if (bo) {
-      surf[1] = bo->offset; /* reloc */
-
-      /* Emit relocation to surface contents. */
-      drm_intel_bo_emit_reloc(brw->intel.batch.bo,
-                             brw->wm.surf_offset[surf_index] + 4,
-                             bo, 0, I915_GEM_DOMAIN_SAMPLER, 0);
-
-      int w = intel_obj->Base.Size / texel_size;
-      surf[2] = ((w & 0x7f) << BRW_SURFACE_WIDTH_SHIFT |
-                ((w >> 7) & 0x1fff) << BRW_SURFACE_HEIGHT_SHIFT);
-      surf[3] = (((w >> 20) & 0x7f) << BRW_SURFACE_DEPTH_SHIFT |
-                (texel_size - 1) << BRW_SURFACE_PITCH_SHIFT);
-   } else {
-      surf[1] = 0;
-      surf[2] = 0;
-      surf[3] = 0;
-   }
-
-   surf[4] = 0;
-   surf[5] = 0;
+   brw->vtbl.emit_buffer_surface_state(brw, surf_offset, bo,
+                                       tObj->BufferOffset,
+                                       brw_format,
+                                       size / texel_size,
+                                       texel_size,
+                                       0, /* mocs */
+                                       false /* rw */);
 }
 
 static void
-brw_update_texture_surface( struct gl_context *ctx, GLuint unit )
+brw_update_texture_surface(struct gl_context *ctx,
+                           unsigned unit,
+                           uint32_t *surf_offset,
+                           bool for_gather)
 {
    struct brw_context *brw = brw_context(ctx);
    struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
    struct intel_texture_object *intelObj = intel_texture_object(tObj);
    struct intel_mipmap_tree *mt = intelObj->mt;
-   struct gl_texture_image *firstImage = tObj->Image[0][tObj->BaseLevel];
    struct gl_sampler_object *sampler = _mesa_get_samplerobj(ctx, unit);
-   const GLuint surf_index = SURF_INDEX_TEXTURE(unit);
    uint32_t *surf;
-   int width, height, depth;
 
+   /* BRW_NEW_UNIFORM_BUFFER */
    if (tObj->Target == GL_TEXTURE_BUFFER) {
-      brw_update_buffer_texture_surface(ctx, unit);
+      brw_update_buffer_texture_surface(ctx, unit, surf_offset);
       return;
    }
 
-   intel_miptree_get_dimensions_for_image(firstImage, &width, &height, &depth);
-
    surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
-                         6 * 4, 32, &brw->wm.surf_offset[surf_index]);
+                         6 * 4, 32, surf_offset);
+
+   (void) for_gather;   /* no w/a to apply for this gen */
 
    surf[0] = (translate_tex_target(tObj->Target) << BRW_SURFACE_TYPE_SHIFT |
              BRW_SURFACE_MIPMAPLAYOUT_BELOW << BRW_SURFACE_MIPLAYOUT_SHIFT |
              BRW_SURFACE_CUBEFACE_ENABLES |
-             (translate_tex_format(mt->format,
-                                   firstImage->InternalFormat,
-                                   tObj->DepthMode,
+             (translate_tex_format(brw,
+                                    mt->format,
                                    sampler->sRGBDecode) <<
               BRW_SURFACE_FORMAT_SHIFT));
 
-   surf[1] = intelObj->mt->region->bo->offset + intelObj->mt->offset; /* reloc */
+   surf[1] = intelObj->mt->region->bo->offset64 + intelObj->mt->offset; /* reloc */
 
    surf[2] = ((intelObj->_MaxLevel - tObj->BaseLevel) << BRW_SURFACE_LOD_SHIFT |
-             (width - 1) << BRW_SURFACE_WIDTH_SHIFT |
-             (height - 1) << BRW_SURFACE_HEIGHT_SHIFT);
+             (mt->logical_width0 - 1) << BRW_SURFACE_WIDTH_SHIFT |
+             (mt->logical_height0 - 1) << BRW_SURFACE_HEIGHT_SHIFT);
 
    surf[3] = (brw_get_surface_tiling_bits(intelObj->mt->region->tiling) |
-             (depth - 1) << BRW_SURFACE_DEPTH_SHIFT |
-             ((intelObj->mt->region->pitch * intelObj->mt->cpp) - 1) <<
+             (mt->logical_depth0 - 1) << BRW_SURFACE_DEPTH_SHIFT |
+             (intelObj->mt->region->pitch - 1) <<
              BRW_SURFACE_PITCH_SHIFT);
 
-   surf[4] = 0;
+   surf[4] = (brw_get_surface_num_multisamples(intelObj->mt->num_samples) |
+              SET_FIELD(tObj->BaseLevel - mt->first_level, BRW_SURFACE_MIN_LOD));
 
-   surf[5] = (mt->align_h == 4) ? BRW_SURFACE_VERTICAL_ALIGN_ENABLE : 0;
+   surf[5] = mt->align_h == 4 ? BRW_SURFACE_VERTICAL_ALIGN_ENABLE : 0;
 
    /* Emit relocation to surface contents */
-   drm_intel_bo_emit_reloc(brw->intel.batch.bo,
-                          brw->wm.surf_offset[surf_index] + 4,
+   drm_intel_bo_emit_reloc(brw->batch.bo,
+                          *surf_offset + 4,
                           intelObj->mt->region->bo,
-                           intelObj->mt->offset,
+                           surf[1] - intelObj->mt->region->bo->offset64,
                           I915_GEM_DOMAIN_SAMPLER, 0);
 }
 
@@ -768,42 +324,16 @@ void
 brw_create_constant_surface(struct brw_context *brw,
                            drm_intel_bo *bo,
                            uint32_t offset,
-                           int width,
-                           uint32_t *out_offset)
+                           uint32_t size,
+                           uint32_t *out_offset,
+                            bool dword_pitch)
 {
-   struct intel_context *intel = &brw->intel;
-   const GLint w = width - 1;
-   uint32_t *surf;
-
-   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
-                         6 * 4, 32, out_offset);
+   uint32_t stride = dword_pitch ? 4 : 16;
+   uint32_t elements = ALIGN(size, stride) / stride;
 
-   surf[0] = (BRW_SURFACE_BUFFER << BRW_SURFACE_TYPE_SHIFT |
-             BRW_SURFACE_MIPMAPLAYOUT_BELOW << BRW_SURFACE_MIPLAYOUT_SHIFT |
-             BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_SURFACE_FORMAT_SHIFT);
-
-   if (intel->gen >= 6)
-      surf[0] |= BRW_SURFACE_RC_READ_WRITE;
-
-   surf[1] = bo->offset + offset; /* reloc */
-
-   surf[2] = ((w & 0x7f) << BRW_SURFACE_WIDTH_SHIFT |
-             ((w >> 7) & 0x1fff) << BRW_SURFACE_HEIGHT_SHIFT);
-
-   surf[3] = (((w >> 20) & 0x7f) << BRW_SURFACE_DEPTH_SHIFT |
-             (16 - 1) << BRW_SURFACE_PITCH_SHIFT); /* ignored */
-
-   surf[4] = 0;
-   surf[5] = 0;
-
-   /* Emit relocation to surface contents.  Section 5.1.1 of the gen4
-    * bspec ("Data Cache") says that the data cache does not exist as
-    * a separate cache and is just the sampler cache.
-    */
-   drm_intel_bo_emit_reloc(brw->intel.batch.bo,
-                          *out_offset + 4,
-                          bo, offset,
-                          I915_GEM_DOMAIN_SAMPLER, 0);
+   brw->vtbl.emit_buffer_surface_state(brw, out_offset, bo, offset,
+                                       BRW_SURFACEFORMAT_R32G32B32A32_FLOAT,
+                                       elements, stride, 0, false);
 }
 
 /**
@@ -818,14 +348,14 @@ brw_update_sol_surface(struct brw_context *brw,
                        uint32_t *out_offset, unsigned num_vector_components,
                        unsigned stride_dwords, unsigned offset_dwords)
 {
-   struct intel_context *intel = &brw->intel;
    struct intel_buffer_object *intel_bo = intel_buffer_object(buffer_obj);
-   drm_intel_bo *bo =
-      intel_bufferobj_buffer(intel, intel_bo, INTEL_WRITE_PART);
+   uint32_t offset_bytes = 4 * offset_dwords;
+   drm_intel_bo *bo = intel_bufferobj_buffer(brw, intel_bo,
+                                             offset_bytes,
+                                             buffer_obj->Size - offset_bytes);
    uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 6 * 4, 32,
                                     out_offset);
    uint32_t pitch_minus_1 = 4*stride_dwords - 1;
-   uint32_t offset_bytes = 4 * offset_dwords;
    size_t size_dwords = buffer_obj->Size / 4;
    uint32_t buffer_size_minus_1, width, height, depth, surface_format;
 
@@ -878,7 +408,7 @@ brw_update_sol_surface(struct brw_context *brw,
       BRW_SURFACE_MIPMAPLAYOUT_BELOW << BRW_SURFACE_MIPLAYOUT_SHIFT |
       surface_format << BRW_SURFACE_FORMAT_SHIFT |
       BRW_SURFACE_RC_READ_WRITE;
-   surf[1] = bo->offset + offset_bytes; /* reloc */
+   surf[1] = bo->offset64 + offset_bytes; /* reloc */
    surf[2] = (width << BRW_SURFACE_WIDTH_SHIFT |
              height << BRW_SURFACE_HEIGHT_SHIFT);
    surf[3] = (depth << BRW_SURFACE_DEPTH_SHIFT |
@@ -887,7 +417,7 @@ brw_update_sol_surface(struct brw_context *brw,
    surf[5] = 0;
 
    /* Emit relocation to surface contents. */
-   drm_intel_bo_emit_reloc(brw->intel.batch.bo,
+   drm_intel_bo_emit_reloc(brw->batch.bo,
                           *out_offset + 4,
                           bo, offset_bytes,
                           I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER);
@@ -902,14 +432,14 @@ brw_update_sol_surface(struct brw_context *brw,
 static void
 brw_upload_wm_pull_constants(struct brw_context *brw)
 {
-   struct gl_context *ctx = &brw->intel.ctx;
-   struct intel_context *intel = &brw->intel;
+   struct gl_context *ctx = &brw->ctx;
    /* BRW_NEW_FRAGMENT_PROGRAM */
    struct brw_fragment_program *fp =
       (struct brw_fragment_program *) brw->fragment_program;
    struct gl_program_parameter_list *params = fp->program.Base.Parameters;
    const int size = brw->wm.prog_data->nr_pull_params * sizeof(float);
-   const int surf_index = SURF_INDEX_FRAG_CONST_BUFFER;
+   const int surf_index =
+      brw->wm.prog_data->base.binding_table.pull_constants_start;
    float *constants;
    unsigned int i;
 
@@ -917,30 +447,30 @@ brw_upload_wm_pull_constants(struct brw_context *brw)
 
    /* CACHE_NEW_WM_PROG */
    if (brw->wm.prog_data->nr_pull_params == 0) {
-      if (brw->wm.const_bo) {
-        drm_intel_bo_unreference(brw->wm.const_bo);
-        brw->wm.const_bo = NULL;
-        brw->wm.surf_offset[surf_index] = 0;
+      if (brw->wm.base.const_bo) {
+        drm_intel_bo_unreference(brw->wm.base.const_bo);
+        brw->wm.base.const_bo = NULL;
+        brw->wm.base.surf_offset[surf_index] = 0;
         brw->state.dirty.brw |= BRW_NEW_SURFACES;
       }
       return;
    }
 
-   drm_intel_bo_unreference(brw->wm.const_bo);
-   brw->wm.const_bo = drm_intel_bo_alloc(intel->bufmgr, "WM const bo",
+   drm_intel_bo_unreference(brw->wm.base.const_bo);
+   brw->wm.base.const_bo = drm_intel_bo_alloc(brw->bufmgr, "WM const bo",
                                         size, 64);
 
    /* _NEW_PROGRAM_CONSTANTS */
-   drm_intel_gem_bo_map_gtt(brw->wm.const_bo);
-   constants = brw->wm.const_bo->virtual;
+   drm_intel_gem_bo_map_gtt(brw->wm.base.const_bo);
+   constants = brw->wm.base.const_bo->virtual;
    for (i = 0; i < brw->wm.prog_data->nr_pull_params; i++) {
       constants[i] = *brw->wm.prog_data->pull_param[i];
    }
-   drm_intel_gem_bo_unmap_gtt(brw->wm.const_bo);
+   drm_intel_gem_bo_unmap_gtt(brw->wm.base.const_bo);
 
-   intel->vtbl.create_constant_surface(brw, brw->wm.const_bo, 0,
-                                      params->NumParameters,
-                                      &brw->wm.surf_offset[surf_index]);
+   brw_create_constant_surface(brw, brw->wm.base.const_bo, 0, size,
+                               &brw->wm.base.surf_offset[surf_index],
+                               true);
 
    brw->state.dirty.brw |= BRW_NEW_SURFACES;
 }
@@ -975,19 +505,20 @@ brw_update_null_renderbuffer_surface(struct brw_context *brw, unsigned int unit)
     *
     *     - Surface Format must be R8G8B8A8_UNORM.
     */
-   struct intel_context *intel = &brw->intel;
-   struct gl_context *ctx = &intel->ctx;
+   struct gl_context *ctx = &brw->ctx;
    uint32_t *surf;
    unsigned surface_type = BRW_SURFACE_NULL;
    drm_intel_bo *bo = NULL;
    unsigned pitch_minus_1 = 0;
    uint32_t multisampling_state = 0;
+   uint32_t surf_index =
+      brw->wm.prog_data->binding_table.render_target_start + unit;
 
    /* _NEW_BUFFERS */
    const struct gl_framebuffer *fb = ctx->DrawBuffer;
 
-   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
-                         6 * 4, 32, &brw->wm.surf_offset[unit]);
+   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 6 * 4, 32,
+                          &brw->wm.base.surf_offset[surf_index]);
 
    if (fb->Visual.samples > 1) {
       /* On Gen6, null render targets seem to cause GPU hangs when
@@ -1007,7 +538,7 @@ brw_update_null_renderbuffer_surface(struct brw_context *brw, unsigned int unit)
       unsigned width_in_tiles = ALIGN(fb->Width, 16) / 16;
       unsigned height_in_tiles = ALIGN(fb->Height, 16) / 16;
       unsigned size_needed = (width_in_tiles + height_in_tiles - 1) * 4096;
-      brw_get_scratch_bo(intel, &brw->wm.multisampled_null_render_target_bo,
+      brw_get_scratch_bo(brw, &brw->wm.multisampled_null_render_target_bo,
                          size_needed);
       bo = brw->wm.multisampled_null_render_target_bo;
       surface_type = BRW_SURFACE_2D;
@@ -1018,13 +549,13 @@ brw_update_null_renderbuffer_surface(struct brw_context *brw, unsigned int unit)
 
    surf[0] = (surface_type << BRW_SURFACE_TYPE_SHIFT |
              BRW_SURFACEFORMAT_B8G8R8A8_UNORM << BRW_SURFACE_FORMAT_SHIFT);
-   if (intel->gen < 6) {
+   if (brw->gen < 6) {
       surf[0] |= (1 << BRW_SURFACE_WRITEDISABLE_R_SHIFT |
                  1 << BRW_SURFACE_WRITEDISABLE_G_SHIFT |
                  1 << BRW_SURFACE_WRITEDISABLE_B_SHIFT |
                  1 << BRW_SURFACE_WRITEDISABLE_A_SHIFT);
    }
-   surf[1] = bo ? bo->offset : 0;
+   surf[1] = bo ? bo->offset64 : 0;
    surf[2] = ((fb->Width - 1) << BRW_SURFACE_WIDTH_SHIFT |
               (fb->Height - 1) << BRW_SURFACE_HEIGHT_SHIFT);
 
@@ -1039,8 +570,8 @@ brw_update_null_renderbuffer_surface(struct brw_context *brw, unsigned int unit)
    surf[5] = 0;
 
    if (bo) {
-      drm_intel_bo_emit_reloc(brw->intel.batch.bo,
-                              brw->wm.surf_offset[unit] + 4,
+      drm_intel_bo_emit_reloc(brw->batch.bo,
+                              brw->wm.base.surf_offset[surf_index] + 4,
                               bo, 0,
                               I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER);
    }
@@ -1054,20 +585,25 @@ brw_update_null_renderbuffer_surface(struct brw_context *brw, unsigned int unit)
 static void
 brw_update_renderbuffer_surface(struct brw_context *brw,
                                struct gl_renderbuffer *rb,
+                               bool layered,
                                unsigned int unit)
 {
-   struct intel_context *intel = &brw->intel;
-   struct gl_context *ctx = &intel->ctx;
+   struct gl_context *ctx = &brw->ctx;
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);
    struct intel_mipmap_tree *mt = irb->mt;
    struct intel_region *region;
    uint32_t *surf;
    uint32_t tile_x, tile_y;
    uint32_t format = 0;
-   gl_format rb_format = intel_rb_format(irb);
+   /* _NEW_BUFFERS */
+   gl_format rb_format = _mesa_get_render_format(ctx, intel_rb_format(irb));
+   uint32_t surf_index =
+      brw->wm.prog_data->binding_table.render_target_start + unit;
+
+   assert(!layered);
 
-   if (irb->tex_image && !brw->has_surface_tile_offset) {
-      intel_renderbuffer_tile_offsets(irb, &tile_x, &tile_y);
+   if (rb->TexImage && !brw->has_surface_tile_offset) {
+      intel_renderbuffer_get_tile_offsets(irb, &tile_x, &tile_y);
 
       if (tile_x != 0 || tile_y != 0) {
         /* Original gen4 hardware couldn't draw to a non-tile-aligned
@@ -1076,70 +612,36 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
          * select the image.  So, instead, we just make a new single-level
          * miptree and render into that.
          */
-        struct intel_context *intel = intel_context(ctx);
-        struct intel_texture_image *intel_image =
-           intel_texture_image(irb->tex_image);
-        struct intel_mipmap_tree *new_mt;
-        int width, height, depth;
-
-        intel_miptree_get_dimensions_for_image(irb->tex_image, &width, &height, &depth);
-
-        new_mt = intel_miptree_create(intel, irb->tex_image->TexObject->Target,
-                                      intel_image->base.Base.TexFormat,
-                                      intel_image->base.Base.Level,
-                                      intel_image->base.Base.Level,
-                                      width, height, depth,
-                                      true,
-                                       0 /* num_samples */,
-                                       INTEL_MSAA_LAYOUT_NONE);
-
-        intel_miptree_copy_teximage(intel, intel_image, new_mt);
-        intel_miptree_reference(&irb->mt, intel_image->mt);
-        intel_renderbuffer_set_draw_offset(irb);
-        intel_miptree_release(&new_mt);
-
+        intel_renderbuffer_move_to_temp(brw, irb, false);
         mt = irb->mt;
       }
    }
 
+   intel_miptree_used_for_rendering(irb->mt);
+
    region = irb->mt->region;
 
-   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
-                         6 * 4, 32, &brw->wm.surf_offset[unit]);
+   surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 6 * 4, 32,
+                          &brw->wm.base.surf_offset[surf_index]);
 
-   switch (rb_format) {
-   case MESA_FORMAT_SARGB8:
-      /* _NEW_BUFFERS
-       *
-       * Without GL_EXT_framebuffer_sRGB we shouldn't bind sRGB surfaces to the
-       * blend/update as sRGB.
-       */
-      if (ctx->Color.sRGBEnabled)
-        format = brw_format_for_mesa_format(rb_format);
-      else
-        format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-      break;
-   default:
-      format = brw->render_target_format[rb_format];
-      if (unlikely(!brw->format_supported_as_render_target[rb_format])) {
-        _mesa_problem(ctx, "%s: renderbuffer format %s unsupported\n",
-                      __FUNCTION__, _mesa_get_format_name(rb_format));
-      }
-      break;
+   format = brw->render_target_format[rb_format];
+   if (unlikely(!brw->format_supported_as_render_target[rb_format])) {
+      _mesa_problem(ctx, "%s: renderbuffer format %s unsupported\n",
+                    __FUNCTION__, _mesa_get_format_name(rb_format));
    }
 
    surf[0] = (BRW_SURFACE_2D << BRW_SURFACE_TYPE_SHIFT |
              format << BRW_SURFACE_FORMAT_SHIFT);
 
    /* reloc */
-   surf[1] = (intel_renderbuffer_tile_offsets(irb, &tile_x, &tile_y) +
-             region->bo->offset);
+   surf[1] = (intel_renderbuffer_get_tile_offsets(irb, &tile_x, &tile_y) +
+             region->bo->offset64);
 
    surf[2] = ((rb->Width - 1) << BRW_SURFACE_WIDTH_SHIFT |
              (rb->Height - 1) << BRW_SURFACE_HEIGHT_SHIFT);
 
    surf[3] = (brw_get_surface_tiling_bits(region->tiling) |
-             ((region->pitch * region->cpp) - 1) << BRW_SURFACE_PITCH_SHIFT);
+             (region->pitch - 1) << BRW_SURFACE_PITCH_SHIFT);
 
    surf[4] = brw_get_surface_num_multisamples(mt->num_samples);
 
@@ -1153,7 +655,7 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
              (tile_y / 2) << BRW_SURFACE_Y_OFFSET_SHIFT |
              (mt->align_h == 4 ? BRW_SURFACE_VERTICAL_ALIGN_ENABLE : 0));
 
-   if (intel->gen < 6) {
+   if (brw->gen < 6) {
       /* _NEW_COLOR */
       if (!ctx->Color.ColorLogicOpEnabled &&
          (ctx->Color.BlendEnabled & (1 << unit)))
@@ -1175,10 +677,10 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       }
    }
 
-   drm_intel_bo_emit_reloc(brw->intel.batch.bo,
-                          brw->wm.surf_offset[unit] + 4,
+   drm_intel_bo_emit_reloc(brw->batch.bo,
+                          brw->wm.base.surf_offset[surf_index] + 4,
                           region->bo,
-                          surf[1] - region->bo->offset,
+                          surf[1] - region->bo->offset64,
                           I915_GEM_DOMAIN_RENDER,
                           I915_GEM_DOMAIN_RENDER);
 }
@@ -1189,8 +691,7 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
 static void
 brw_update_renderbuffer_surfaces(struct brw_context *brw)
 {
-   struct intel_context *intel = &brw->intel;
-   struct gl_context *ctx = &brw->intel.ctx;
+   struct gl_context *ctx = &brw->ctx;
    GLuint i;
 
    /* _NEW_BUFFERS | _NEW_COLOR */
@@ -1198,13 +699,14 @@ brw_update_renderbuffer_surfaces(struct brw_context *brw)
    if (ctx->DrawBuffer->_NumColorDrawBuffers >= 1) {
       for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
         if (intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[i])) {
-           intel->vtbl.update_renderbuffer_surface(brw, ctx->DrawBuffer->_ColorDrawBuffers[i], i);
+           brw->vtbl.update_renderbuffer_surface(brw, ctx->DrawBuffer->_ColorDrawBuffers[i],
+                                                  ctx->DrawBuffer->MaxNumLayers > 0, i);
         } else {
-           intel->vtbl.update_null_renderbuffer_surface(brw, i);
+           brw->vtbl.update_null_renderbuffer_surface(brw, i);
         }
       }
    } else {
-      intel->vtbl.update_null_renderbuffer_surface(brw, 0);
+      brw->vtbl.update_null_renderbuffer_surface(brw, 0);
    }
    brw->state.dirty.brw |= BRW_NEW_SURFACES;
 }
@@ -1228,29 +730,69 @@ const struct brw_tracked_state gen6_renderbuffer_surfaces = {
    .emit = brw_update_renderbuffer_surfaces,
 };
 
+
+static void
+update_stage_texture_surfaces(struct brw_context *brw,
+                              const struct gl_program *prog,
+                              struct brw_stage_state *stage_state,
+                              bool for_gather)
+{
+   if (!prog)
+      return;
+
+   struct gl_context *ctx = &brw->ctx;
+
+   uint32_t *surf_offset = stage_state->surf_offset;
+   if (for_gather)
+      surf_offset += stage_state->prog_data->binding_table.gather_texture_start;
+   else
+      surf_offset += stage_state->prog_data->binding_table.texture_start;
+
+   unsigned num_samplers = _mesa_fls(prog->SamplersUsed);
+   for (unsigned s = 0; s < num_samplers; s++) {
+      surf_offset[s] = 0;
+
+      if (prog->SamplersUsed & (1 << s)) {
+         const unsigned unit = prog->SamplerUnits[s];
+
+         /* _NEW_TEXTURE */
+         if (ctx->Texture.Unit[unit]._ReallyEnabled) {
+            brw->vtbl.update_texture_surface(ctx, unit, surf_offset + s, for_gather);
+         }
+      }
+   }
+}
+
+
 /**
  * Construct SURFACE_STATE objects for enabled textures.
  */
 static void
 brw_update_texture_surfaces(struct brw_context *brw)
 {
-   struct gl_context *ctx = &brw->intel.ctx;
-
-   for (unsigned i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
-      const GLuint surf = SURF_INDEX_TEXTURE(i);
+   /* BRW_NEW_VERTEX_PROGRAM */
+   struct gl_program *vs = (struct gl_program *) brw->vertex_program;
 
-      /* _NEW_TEXTURE */
-      if (texUnit->_ReallyEnabled) {
-        brw->intel.vtbl.update_texture_surface(ctx, i);
-      } else {
-         brw->wm.surf_offset[surf] = 0;
-      }
+   /* BRW_NEW_GEOMETRY_PROGRAM */
+   struct gl_program *gs = (struct gl_program *) brw->geometry_program;
 
-      /* For now, just mirror the texture setup to the VS slots. */
-      brw->vs.surf_offset[SURF_INDEX_VS_TEXTURE(i)] =
-        brw->wm.surf_offset[surf];
-   }
+   /* BRW_NEW_FRAGMENT_PROGRAM */
+   struct gl_program *fs = (struct gl_program *) brw->fragment_program;
+
+   /* _NEW_TEXTURE */
+   update_stage_texture_surfaces(brw, vs, &brw->vs.base, false);
+   update_stage_texture_surfaces(brw, gs, &brw->gs.base, false);
+   update_stage_texture_surfaces(brw, fs, &brw->wm.base, false);
+
+   /* emit alternate set of surface state for gather. this
+    * allows the surface format to be overriden for only the
+    * gather4 messages. */
+   if (vs && vs->UsesGather)
+      update_stage_texture_surfaces(brw, vs, &brw->vs.base, true);
+   if (gs && gs->UsesGather)
+      update_stage_texture_surfaces(brw, gs, &brw->gs.base, true);
+   if (fs && fs->UsesGather)
+      update_stage_texture_surfaces(brw, fs, &brw->wm.base, true);
 
    brw->state.dirty.brw |= BRW_NEW_SURFACES;
 }
@@ -1258,7 +800,11 @@ brw_update_texture_surfaces(struct brw_context *brw)
 const struct brw_tracked_state brw_texture_surfaces = {
    .dirty = {
       .mesa = _NEW_TEXTURE,
-      .brw = BRW_NEW_BATCH,
+      .brw = BRW_NEW_BATCH |
+             BRW_NEW_UNIFORM_BUFFER |
+             BRW_NEW_VERTEX_PROGRAM |
+             BRW_NEW_GEOMETRY_PROGRAM |
+             BRW_NEW_FRAGMENT_PROGRAM,
       .cache = 0
    },
    .emit = brw_update_texture_surfaces,
@@ -1267,32 +813,36 @@ const struct brw_tracked_state brw_texture_surfaces = {
 void
 brw_upload_ubo_surfaces(struct brw_context *brw,
                        struct gl_shader *shader,
-                       uint32_t *surf_offsets)
+                        struct brw_stage_state *stage_state,
+                        struct brw_stage_prog_data *prog_data)
 {
-   struct gl_context *ctx = &brw->intel.ctx;
-   struct intel_context *intel = &brw->intel;
+   struct gl_context *ctx = &brw->ctx;
 
    if (!shader)
       return;
 
+   uint32_t *surf_offsets =
+      &stage_state->surf_offset[prog_data->binding_table.ubo_start];
+
    for (int i = 0; i < shader->NumUniformBlocks; i++) {
       struct gl_uniform_buffer_binding *binding;
       struct intel_buffer_object *intel_bo;
 
       binding = &ctx->UniformBufferBindings[shader->UniformBlocks[i].Binding];
       intel_bo = intel_buffer_object(binding->BufferObject);
-      drm_intel_bo *bo = intel_bufferobj_buffer(intel, intel_bo, INTEL_READ);
+      drm_intel_bo *bo =
+         intel_bufferobj_buffer(brw, intel_bo,
+                                binding->Offset,
+                                binding->BufferObject->Size - binding->Offset);
 
       /* Because behavior for referencing outside of the binding's size in the
        * glBindBufferRange case is undefined, we can just bind the whole buffer
        * glBindBufferBase wants and be a correct implementation.
        */
-      int size = bo->size - binding->Offset;
-      size = ALIGN(size, 16) / 16; /* The interface takes a number of vec4s */
-
-      intel->vtbl.create_constant_surface(brw, bo, binding->Offset,
-                                         size,
-                                         &surf_offsets[i]);
+      brw_create_constant_surface(brw, bo, binding->Offset,
+                                  bo->size - binding->Offset,
+                                  &surf_offsets[i],
+                                  shader->Stage == MESA_SHADER_FRAGMENT);
    }
 
    if (shader->NumUniformBlocks)
@@ -1302,70 +852,83 @@ brw_upload_ubo_surfaces(struct brw_context *brw,
 static void
 brw_upload_wm_ubo_surfaces(struct brw_context *brw)
 {
-   struct gl_context *ctx = &brw->intel.ctx;
+   struct gl_context *ctx = &brw->ctx;
    /* _NEW_PROGRAM */
    struct gl_shader_program *prog = ctx->Shader._CurrentFragmentProgram;
 
    if (!prog)
       return;
 
+   /* CACHE_NEW_WM_PROG */
    brw_upload_ubo_surfaces(brw, prog->_LinkedShaders[MESA_SHADER_FRAGMENT],
-                          &brw->wm.surf_offset[SURF_INDEX_WM_UBO(0)]);
+                           &brw->wm.base, &brw->wm.prog_data->base);
 }
 
 const struct brw_tracked_state brw_wm_ubo_surfaces = {
    .dirty = {
-      .mesa = (_NEW_PROGRAM |
-              _NEW_BUFFER_OBJECT),
-      .brw = BRW_NEW_BATCH,
-      .cache = 0,
+      .mesa = _NEW_PROGRAM,
+      .brw = BRW_NEW_BATCH | BRW_NEW_UNIFORM_BUFFER,
+      .cache = CACHE_NEW_WM_PROG,
    },
    .emit = brw_upload_wm_ubo_surfaces,
 };
 
-/**
- * Constructs the binding table for the WM surface state, which maps unit
- * numbers to surface state objects.
- */
-static void
-brw_upload_wm_binding_table(struct brw_context *brw)
+void
+brw_upload_abo_surfaces(struct brw_context *brw,
+                       struct gl_shader_program *prog,
+                        struct brw_stage_state *stage_state,
+                        struct brw_stage_prog_data *prog_data)
 {
-   uint32_t *bind;
-   int i;
+   struct gl_context *ctx = &brw->ctx;
+   uint32_t *surf_offsets =
+      &stage_state->surf_offset[prog_data->binding_table.abo_start];
+
+   for (int i = 0; i < prog->NumAtomicBuffers; i++) {
+      struct gl_atomic_buffer_binding *binding =
+         &ctx->AtomicBufferBindings[prog->AtomicBuffers[i].Binding];
+      struct intel_buffer_object *intel_bo =
+         intel_buffer_object(binding->BufferObject);
+      drm_intel_bo *bo = intel_bufferobj_buffer(
+         brw, intel_bo, binding->Offset, intel_bo->Base.Size - binding->Offset);
+
+      brw->vtbl.create_raw_surface(brw, bo, binding->Offset,
+                                   bo->size - binding->Offset,
+                                   &surf_offsets[i], true);
+   }
 
-   /* Might want to calculate nr_surfaces first, to avoid taking up so much
-    * space for the binding table.
-    */
-   bind = brw_state_batch(brw, AUB_TRACE_BINDING_TABLE,
-                         sizeof(uint32_t) * BRW_MAX_WM_SURFACES,
-                         32, &brw->wm.bind_bo_offset);
+   if (prog->NumUniformBlocks)
+      brw->state.dirty.brw |= BRW_NEW_SURFACES;
+}
 
-   /* BRW_NEW_SURFACES */
-   for (i = 0; i < BRW_MAX_WM_SURFACES; i++) {
-      bind[i] = brw->wm.surf_offset[i];
-   }
+static void
+brw_upload_wm_abo_surfaces(struct brw_context *brw)
+{
+   struct gl_context *ctx = &brw->ctx;
+   /* _NEW_PROGRAM */
+   struct gl_shader_program *prog = ctx->Shader._CurrentFragmentProgram;
 
-   brw->state.dirty.brw |= BRW_NEW_PS_BINDING_TABLE;
+   if (prog) {
+      /* CACHE_NEW_WM_PROG */
+      brw_upload_abo_surfaces(brw, prog, &brw->wm.base,
+                              &brw->wm.prog_data->base);
+   }
 }
 
-const struct brw_tracked_state brw_wm_binding_table = {
+const struct brw_tracked_state brw_wm_abo_surfaces = {
    .dirty = {
-      .mesa = 0,
-      .brw = (BRW_NEW_BATCH |
-             BRW_NEW_SURFACES),
-      .cache = 0
+      .mesa = _NEW_PROGRAM,
+      .brw = BRW_NEW_BATCH | BRW_NEW_ATOMIC_BUFFER,
+      .cache = CACHE_NEW_WM_PROG,
    },
-   .emit = brw_upload_wm_binding_table,
+   .emit = brw_upload_wm_abo_surfaces,
 };
 
 void
 gen4_init_vtable_surface_functions(struct brw_context *brw)
 {
-   struct intel_context *intel = &brw->intel;
-
-   intel->vtbl.update_texture_surface = brw_update_texture_surface;
-   intel->vtbl.update_renderbuffer_surface = brw_update_renderbuffer_surface;
-   intel->vtbl.update_null_renderbuffer_surface =
+   brw->vtbl.update_texture_surface = brw_update_texture_surface;
+   brw->vtbl.update_renderbuffer_surface = brw_update_renderbuffer_surface;
+   brw->vtbl.update_null_renderbuffer_surface =
       brw_update_null_renderbuffer_surface;
-   intel->vtbl.create_constant_surface = brw_create_constant_surface;
+   brw->vtbl.emit_buffer_surface_state = gen4_emit_buffer_surface_state;
 }