Merge branch 'master' of ssh://git.freedesktop.org/git/mesa/mesa into pipe-video

Conflicts: src/gallium/auxiliary/Makefile src/gallium/auxiliary/SConscript src/gallium/auxiliary/util/u_format.csv src/gallium/auxiliary/vl/vl_compositor.c src/gallium/auxiliary/vl/vl_compositor.h src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h src/gallium/drivers/identity/id_objects.c src/gallium/drivers/identity/id_objects.h src/gallium/drivers/identity/id_screen.c src/gallium/drivers/nv40/Makefile src/gallium/drivers/nv40/nv40_screen.c src/gallium/drivers/softpipe/sp_texture.c src/gallium/drivers/softpipe/sp_texture.h src/gallium/drivers/softpipe/sp_video_context.c src/gallium/drivers/softpipe/sp_video_context.h src/gallium/include/pipe/p_format.h src/gallium/include/pipe/p_screen.h src/gallium/include/pipe/p_video_context.h src/gallium/include/pipe/p_video_state.h src/gallium/include/state_tracker/dri1_api.h src/gallium/include/state_tracker/drm_api.h src/gallium/state_trackers/dri/common/dri_context.c src/gallium/state_trackers/xorg/xvmc/attributes.c src/gallium/state_trackers/xorg/xvmc/block.c src/gallium/state_trackers/xorg/xvmc/context.c src/gallium/state_trackers/xorg/xvmc/subpicture.c src/gallium/state_trackers/xorg/xvmc/surface.c src/gallium/state_trackers/xorg/xvmc/tests/.gitignore src/gallium/state_trackers/xorg/xvmc/tests/Makefile src/gallium/state_trackers/xorg/xvmc/xvmc_private.h src/gallium/winsys/drm/radeon/core/radeon_drm.c src/gallium/winsys/g3dvl/vl_winsys.h src/gallium/winsys/g3dvl/xlib/xsp_winsys.c src/gallium/winsys/sw/Makefile
author: Younes Manton <[email protected]> 2010-04-30 20:42:30 -0400
committer: Younes Manton <[email protected]> 2010-04-30 20:42:30 -0400
commit: a8ea1dacc63ac567498049e5756c247b9fec6cd9 (patch)
tree: 4031e2e2b6166bd926b43fa4bbb3aab773a30ee5 /src/gallium/auxiliary/util
parent: 404fb63b4649f58fce443615e49337d42b8ddece (diff)
parent: 35d960cc744c374ccaad48c3d80559b59c74e28a (diff)
62 files changed, 8503 insertions, 2726 deletions
diff --git a/src/gallium/auxiliary/util/.gitignore b/src/gallium/auxiliary/util/.gitignore
index 448d2f304fb..5dd0408effb 100644
--- a/src/gallium/auxiliary/util/.gitignore
+++ b/src/gallium/auxiliary/util/.gitignore
@@ -1,3 +1,3 @@
-u_format_access.c
+u_format_srgb.c
 u_format_table.c
-u_format_pack.h
+u_half.c
diff --git a/src/gallium/auxiliary/util/u_atomic.h b/src/gallium/auxiliary/util/u_atomic.h
index 3c42477ad4f..a1568233906 100644
--- a/src/gallium/auxiliary/util/u_atomic.h
+++ b/src/gallium/auxiliary/util/u_atomic.h
@@ -29,7 +29,7 @@
 #define PIPE_ATOMIC_ASM_MSVC_X86                
 #elif (defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86))
 #define PIPE_ATOMIC_ASM_GCC_X86
-#elif defined(PIPE_CC_GCC)
+#elif defined(PIPE_CC_GCC) && (PIPE_CC_GCC_VERSION >= 401)
 #define PIPE_ATOMIC_GCC_INTRINSIC
 #else
 #error "Unsupported platform"
diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index 0b263a9db5c..e45310b9bb7 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -45,6 +45,7 @@
 #include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
+#include "util/u_sampler.h"
 #include "util/u_simple_shaders.h"
 #include "util/u_surface.h"
 #include "util/u_rect.h"
@@ -63,11 +64,12 @@ struct blit_state
    struct pipe_sampler_state sampler;
    struct pipe_viewport_state viewport;
    struct pipe_clip_state clip;
+   struct pipe_vertex_element velem[2];
 
    void *vs;
    void *fs[TGSI_WRITEMASK_XYZW + 1];
 
-   struct pipe_buffer *vbuf;  /**< quad vertices */
+   struct pipe_resource *vbuf;  /**< quad vertices */
    unsigned vbuf_slot;
 
    float vertices[4][2][4];   /**< vertex/texcoords for quad */
@@ -114,6 +116,15 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
    ctx->sampler.mag_img_filter = 0; /* set later */
    ctx->sampler.normalized_coords = 1;
 
+   /* vertex elements state */
+   memset(&ctx->velem[0], 0, sizeof(ctx->velem[0]) * 2);
+   for (i = 0; i < 2; i++) {
+      ctx->velem[i].src_offset = i * 4 * sizeof(float);
+      ctx->velem[i].instance_divisor = 0;
+      ctx->velem[i].vertex_buffer_index = 0;
+      ctx->velem[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+   }
+
    /* vertex shader - still required to provide the linkage between
     * fragment shader input semantics and vertex_element/buffers.
     */
@@ -156,7 +167,7 @@ util_destroy_blit(struct blit_state *ctx)
       if (ctx->fs[i])
          pipe->delete_fs_state(pipe, ctx->fs[i]);
 
-   pipe_buffer_reference(&ctx->vbuf, NULL);
+   pipe_resource_reference(&ctx->vbuf, NULL);
 
    FREE(ctx);
 }
@@ -175,8 +186,7 @@ get_next_slot( struct blit_state *ctx )
 
    if (!ctx->vbuf) {
       ctx->vbuf = pipe_buffer_create(ctx->pipe->screen,
-                                     32,
-                                     PIPE_BUFFER_USAGE_VERTEX,
+                                     PIPE_BIND_VERTEX_BUFFER,
                                      max_slots * sizeof ctx->vertices);
    }
    
@@ -225,7 +235,7 @@ setup_vertex_data_tex(struct blit_state *ctx,
 
    offset = get_next_slot( ctx );
 
-   pipe_buffer_write_nooverlap(ctx->pipe->screen, ctx->vbuf,
+   pipe_buffer_write_nooverlap(ctx->pipe, ctx->vbuf,
                                offset, sizeof(ctx->vertices), ctx->vertices);
 
    return offset;
@@ -270,6 +280,7 @@ regions_overlap(int srcX0, int srcY0,
 void
 util_blit_pixels_writemask(struct blit_state *ctx,
                            struct pipe_surface *src,
+                           struct pipe_sampler_view *src_sampler_view,
                            int srcX0, int srcY0,
                            int srcX1, int srcY1,
                            struct pipe_surface *dst,
@@ -280,7 +291,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
 {
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
-   struct pipe_texture *tex = NULL;
+   struct pipe_sampler_view *sampler_view = NULL;
    struct pipe_framebuffer_state fb;
    const int srcW = abs(srcX1 - srcX0);
    const int srcH = abs(srcY1 - srcY0);
@@ -292,9 +303,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
           filter == PIPE_TEX_MIPFILTER_LINEAR);
 
    assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
+                                      PIPE_BIND_SAMPLER_VIEW, 0));
    assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
+                                      PIPE_BIND_RENDER_TARGET, 0));
 
    /* do the regions overlap? */
    overlap = util_same_surface(src, dst) &&
@@ -323,7 +334,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    }
    
    assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
+                                      PIPE_BIND_RENDER_TARGET, 0));
 
    /* Create a temporary texture when src and dest alias or when src
     * is anything other than a single-level 2d texture.
@@ -334,7 +345,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
        src->texture->target != PIPE_TEXTURE_2D ||
        src->texture->last_level != 0)
    {
-      struct pipe_texture texTemp;
+      struct pipe_resource texTemp;
+      struct pipe_resource *tex;
+      struct pipe_sampler_view sv_templ;
       struct pipe_surface *texSurf;
       const int srcLeft = MIN2(srcX0, srcX1);
       const int srcTop = MIN2(srcY0, srcY1);
@@ -361,13 +374,22 @@ util_blit_pixels_writemask(struct blit_state *ctx,
       texTemp.width0 = srcW;
       texTemp.height0 = srcH;
       texTemp.depth0 = 1;
+      texTemp.bind = PIPE_BIND_SAMPLER_VIEW;
 
-      tex = screen->texture_create(screen, &texTemp);
+      tex = screen->resource_create(screen, &texTemp);
       if (!tex)
          return;
 
+      u_sampler_view_default_template(&sv_templ, tex, tex->format);
+
+      sampler_view = ctx->pipe->create_sampler_view(ctx->pipe, tex, &sv_templ);
+      if (!sampler_view) {
+         pipe_resource_reference(&tex, NULL);
+         return;
+      }
+
       texSurf = screen->get_tex_surface(screen, tex, 0, 0, 0, 
-                                        PIPE_BUFFER_USAGE_GPU_WRITE);
+                                        PIPE_BIND_BLIT_DESTINATION);
 
       /* load temp texture */
       if (pipe->surface_copy) {
@@ -389,33 +411,38 @@ util_blit_pixels_writemask(struct blit_state *ctx,
       s1 = 1.0f;
       t0 = 0.0f;
       t1 = 1.0f;
+
+      pipe_resource_reference(&tex, NULL);
    }
    else {
-      pipe_texture_reference(&tex, src->texture);
-      s0 = srcX0 / (float)tex->width0;
-      s1 = srcX1 / (float)tex->width0;
-      t0 = srcY0 / (float)tex->height0;
-      t1 = srcY1 / (float)tex->height0;
+      pipe_sampler_view_reference(&sampler_view, src_sampler_view);
+      s0 = srcX0 / (float)src->texture->width0;
+      s1 = srcX1 / (float)src->texture->width0;
+      t0 = srcY0 / (float)src->texture->height0;
+      t1 = srcY1 / (float)src->texture->height0;
    }
 
+   
 
    /* save state (restored below) */
    cso_save_blend(ctx->cso);
    cso_save_depth_stencil_alpha(ctx->cso);
    cso_save_rasterizer(ctx->cso);
    cso_save_samplers(ctx->cso);
-   cso_save_sampler_textures(ctx->cso);
+   cso_save_fragment_sampler_views(ctx->cso);
    cso_save_viewport(ctx->cso);
    cso_save_framebuffer(ctx->cso);
    cso_save_fragment_shader(ctx->cso);
    cso_save_vertex_shader(ctx->cso);
    cso_save_clip(ctx->cso);
+   cso_save_vertex_elements(ctx->cso);
 
    /* set misc state we care about */
    cso_set_blend(ctx->cso, &ctx->blend);
    cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
    cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
    cso_set_clip(ctx->cso, &ctx->clip);
+   cso_set_vertex_elements(ctx->cso, 2, ctx->velem);
 
    /* sampler */
    ctx->sampler.min_img_filter = filter;
@@ -435,7 +462,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    cso_set_viewport(ctx->cso, &ctx->viewport);
 
    /* texture */
-   cso_set_sampler_textures(ctx->cso, 1, &tex);
+   cso_set_fragment_sampler_views(ctx->cso, 1, &sampler_view);
 
    if (ctx->fs[writemask] == NULL)
       ctx->fs[writemask] =
@@ -474,20 +501,22 @@ util_blit_pixels_writemask(struct blit_state *ctx,
    cso_restore_depth_stencil_alpha(ctx->cso);
    cso_restore_rasterizer(ctx->cso);
    cso_restore_samplers(ctx->cso);
-   cso_restore_sampler_textures(ctx->cso);
+   cso_restore_fragment_sampler_views(ctx->cso);
    cso_restore_viewport(ctx->cso);
    cso_restore_framebuffer(ctx->cso);
    cso_restore_fragment_shader(ctx->cso);
    cso_restore_vertex_shader(ctx->cso);
    cso_restore_clip(ctx->cso);
+   cso_restore_vertex_elements(ctx->cso);
 
-   pipe_texture_reference(&tex, NULL);
+   pipe_sampler_view_reference(&sampler_view, NULL);
 }
 
 
 void
 util_blit_pixels(struct blit_state *ctx,
                  struct pipe_surface *src,
+                 struct pipe_sampler_view *src_sampler_view,
                  int srcX0, int srcY0,
                  int srcX1, int srcY1,
                  struct pipe_surface *dst,
@@ -495,7 +524,7 @@ util_blit_pixels(struct blit_state *ctx,
                  int dstX1, int dstY1,
                  float z, uint filter )
 {
-   util_blit_pixels_writemask( ctx, src, 
+   util_blit_pixels_writemask( ctx, src, src_sampler_view,
                                srcX0, srcY0,
                                srcX1, srcY1,
                                dst,
@@ -511,7 +540,7 @@ util_blit_pixels(struct blit_state *ctx,
  */
 void util_blit_flush( struct blit_state *ctx )
 {
-   pipe_buffer_reference(&ctx->vbuf, NULL);
+   pipe_resource_reference(&ctx->vbuf, NULL);
    ctx->vbuf_slot = 0;
 } 
 
@@ -526,21 +555,23 @@ void util_blit_flush( struct blit_state *ctx )
  */
 void
 util_blit_pixels_tex(struct blit_state *ctx,
-                 struct pipe_texture *tex,
-                 int srcX0, int srcY0,
-                 int srcX1, int srcY1,
-                 struct pipe_surface *dst,
-                 int dstX0, int dstY0,
-                 int dstX1, int dstY1,
-                 float z, uint filter)
+                     struct pipe_sampler_view *src_sampler_view,
+                     int srcX0, int srcY0,
+                     int srcX1, int srcY1,
+                     struct pipe_surface *dst,
+                     int dstX0, int dstY0,
+                     int dstX1, int dstY1,
+                     float z, uint filter)
 {
    struct pipe_framebuffer_state fb;
    float s0, t0, s1, t1;
    unsigned offset;
+   struct pipe_resource *tex = src_sampler_view->texture;
 
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
 
+   assert(tex);
    assert(tex->width0 != 0);
    assert(tex->height0 != 0);
 
@@ -551,7 +582,7 @@ util_blit_pixels_tex(struct blit_state *ctx,
 
    assert(ctx->pipe->screen->is_format_supported(ctx->pipe->screen, dst->format,
                                                  PIPE_TEXTURE_2D,
-                                                 PIPE_TEXTURE_USAGE_RENDER_TARGET,
+                                                 PIPE_BIND_RENDER_TARGET,
                                                  0));
 
    /* save state (restored below) */
@@ -559,17 +590,19 @@ util_blit_pixels_tex(struct blit_state *ctx,
    cso_save_depth_stencil_alpha(ctx->cso);
    cso_save_rasterizer(ctx->cso);
    cso_save_samplers(ctx->cso);
-   cso_save_sampler_textures(ctx->cso);
+   cso_save_fragment_sampler_views(ctx->cso);
    cso_save_framebuffer(ctx->cso);
    cso_save_fragment_shader(ctx->cso);
    cso_save_vertex_shader(ctx->cso);
    cso_save_clip(ctx->cso);
+   cso_save_vertex_elements(ctx->cso);
 
    /* set misc state we care about */
    cso_set_blend(ctx->cso, &ctx->blend);
    cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
    cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
    cso_set_clip(ctx->cso, &ctx->clip);
+   cso_set_vertex_elements(ctx->cso, 2, ctx->velem);
 
    /* sampler */
    ctx->sampler.min_img_filter = filter;
@@ -589,7 +622,7 @@ util_blit_pixels_tex(struct blit_state *ctx,
    cso_set_viewport(ctx->cso, &ctx->viewport);
 
    /* texture */
-   cso_set_sampler_textures(ctx->cso, 1, &tex);
+   cso_set_fragment_sampler_views(ctx->cso, 1, &src_sampler_view);
 
    /* shaders */
    cso_set_fragment_shader_handle(ctx->cso, ctx->fs[TGSI_WRITEMASK_XYZW]);
@@ -623,9 +656,10 @@ util_blit_pixels_tex(struct blit_state *ctx,
    cso_restore_depth_stencil_alpha(ctx->cso);
    cso_restore_rasterizer(ctx->cso);
    cso_restore_samplers(ctx->cso);
-   cso_restore_sampler_textures(ctx->cso);
+   cso_restore_fragment_sampler_views(ctx->cso);
    cso_restore_framebuffer(ctx->cso);
    cso_restore_fragment_shader(ctx->cso);
    cso_restore_vertex_shader(ctx->cso);
    cso_restore_clip(ctx->cso);
+   cso_restore_vertex_elements(ctx->cso);
 }
diff --git a/src/gallium/auxiliary/util/u_blit.h b/src/gallium/auxiliary/util/u_blit.h
index a102021529e..464ff9aaced 100644
--- a/src/gallium/auxiliary/util/u_blit.h
+++ b/src/gallium/auxiliary/util/u_blit.h
@@ -37,7 +37,7 @@ extern "C" {
    
 struct pipe_context;
 struct pipe_surface;
-struct pipe_texture;
+struct pipe_resource;
 struct cso_context;
 
 
@@ -53,6 +53,7 @@ util_destroy_blit(struct blit_state *ctx);
 extern void
 util_blit_pixels(struct blit_state *ctx,
                  struct pipe_surface *src,
+                 struct pipe_sampler_view *src_sampler_view,
                  int srcX0, int srcY0,
                  int srcX1, int srcY1,
                  struct pipe_surface *dst,
@@ -63,6 +64,7 @@ util_blit_pixels(struct blit_state *ctx,
 void
 util_blit_pixels_writemask(struct blit_state *ctx,
                            struct pipe_surface *src,
+                           struct pipe_sampler_view *src_sampler_view,
                            int srcX0, int srcY0,
                            int srcX1, int srcY1,
                            struct pipe_surface *dst,
@@ -73,7 +75,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
 
 extern void
 util_blit_pixels_tex(struct blit_state *ctx,
-                     struct pipe_texture *tex,
+                     struct pipe_sampler_view *src_sampler_view,
                      int srcX0, int srcY0,
                      int srcX1, int srcY1,
                      struct pipe_surface *dst,
diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c
index 0ba09d33bfc..956aedc8a15 100644
--- a/src/gallium/auxiliary/util/u_blitter.c
+++ b/src/gallium/auxiliary/util/u_blitter.c
@@ -45,6 +45,7 @@
 #include "util/u_draw_quad.h"
 #include "util/u_pack_color.h"
 #include "util/u_rect.h"
+#include "util/u_sampler.h"
 #include "util/u_simple_shaders.h"
 #include "util/u_texture.h"
 
@@ -55,7 +56,7 @@ struct blitter_context_priv
    struct blitter_context blitter;
 
    struct pipe_context *pipe; /**< pipe context */
-   struct pipe_buffer *vbuf;  /**< quad */
+   struct pipe_resource *vbuf;  /**< quad */
 
    float vertices[4][2][4];   /**< {pos, color} or {pos, texcoord} */
 
@@ -88,12 +89,16 @@ struct blitter_context_priv
    void *dsa_write_depth_keep_stencil;
    void *dsa_keep_depth_stencil;
 
+   void *velem_state;
+
    /* Sampler state for clamping to a miplevel. */
    void *sampler_state[PIPE_MAX_TEXTURE_LEVELS];
 
    /* Rasterizer state. */
    void *rs_state;
 
+   struct pipe_sampler_view *sampler_view;
+
    /* Viewport state. */
    struct pipe_viewport_state viewport;
 
@@ -104,10 +109,11 @@ struct blitter_context_priv
 struct blitter_context *util_blitter_create(struct pipe_context *pipe)
 {
    struct blitter_context_priv *ctx;
-   struct pipe_blend_state blend = { 0 };
-   struct pipe_depth_stencil_alpha_state dsa = { { 0 } };
-   struct pipe_rasterizer_state rs_state = { 0 };
+   struct pipe_blend_state blend;
+   struct pipe_depth_stencil_alpha_state dsa;
+   struct pipe_rasterizer_state rs_state;
    struct pipe_sampler_state *sampler_state;
+   struct pipe_vertex_element velem[2];
    unsigned i;
 
    ctx = CALLOC_STRUCT(blitter_context_priv);
@@ -122,17 +128,21 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
    ctx->blitter.saved_rs_state = INVALID_PTR;
    ctx->blitter.saved_fs = INVALID_PTR;
    ctx->blitter.saved_vs = INVALID_PTR;
+   ctx->blitter.saved_velem_state = INVALID_PTR;
    ctx->blitter.saved_fb_state.nr_cbufs = ~0;
-   ctx->blitter.saved_num_textures = ~0;
+   ctx->blitter.saved_num_sampler_views = ~0;
    ctx->blitter.saved_num_sampler_states = ~0;
+   ctx->blitter.saved_num_vertex_buffers = ~0;
 
    /* blend state objects */
+   memset(&blend, 0, sizeof(blend));
    ctx->blend_keep_color = pipe->create_blend_state(pipe, &blend);
 
    blend.rt[0].colormask = PIPE_MASK_RGBA;
    ctx->blend_write_color = pipe->create_blend_state(pipe, &blend);
 
    /* depth stencil alpha state objects */
+   memset(&dsa, 0, sizeof(dsa));
    ctx->dsa_keep_depth_stencil =
       pipe->create_depth_stencil_alpha_state(pipe, &dsa);
 
@@ -159,6 +169,7 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
    sampler_state->wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
    sampler_state->wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
    sampler_state->wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+   sampler_state->normalized_coords = TRUE;
    /* The sampler state objects which sample from a specified mipmap level
     * are created on-demand. */
 
@@ -170,6 +181,16 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
    rs_state.flatshade = 1;
    ctx->rs_state = pipe->create_rasterizer_state(pipe, &rs_state);
 
+   /* vertex elements state */
+   memset(&velem[0], 0, sizeof(velem[0]) * 2);
+   for (i = 0; i < 2; i++) {
+      velem[i].src_offset = i * 4 * sizeof(float);
+      velem[i].instance_divisor = 0;
+      velem[i].vertex_buffer_index = 0;
+      velem[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+   }
+   ctx->velem_state = pipe->create_vertex_elements_state(pipe, 2, &velem[0]);
+
    /* fragment shaders are created on-demand */
 
    /* vertex shaders */
@@ -196,8 +217,7 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
 
    /* create the vertex buffer */
    ctx->vbuf = pipe_buffer_create(ctx->pipe->screen,
-                                  32,
-                                  PIPE_BUFFER_USAGE_VERTEX,
+                                  PIPE_BIND_VERTEX_BUFFER,
                                   sizeof(ctx->vertices));
 
    return &ctx->blitter;
@@ -219,6 +239,7 @@ void util_blitter_destroy(struct blitter_context *blitter)
    pipe->delete_rasterizer_state(pipe, ctx->rs_state);
    pipe->delete_vs_state(pipe, ctx->vs_col);
    pipe->delete_vs_state(pipe, ctx->vs_tex);
+   pipe->delete_vertex_elements_state(pipe, ctx->velem_state);
 
    for (i = 0; i < PIPE_MAX_TEXTURE_TYPES; i++) {
       if (ctx->fs_texfetch_col[i])
@@ -235,7 +256,11 @@ void util_blitter_destroy(struct blitter_context *blitter)
       if (ctx->sampler_state[i])
          pipe->delete_sampler_state(pipe, ctx->sampler_state[i]);
 
-   pipe_buffer_reference(&ctx->vbuf, NULL);
+   if (ctx->sampler_view) {
+      pipe_sampler_view_reference(&ctx->sampler_view, NULL);
+   }
+
+   pipe_resource_reference(&ctx->vbuf, NULL);
    FREE(ctx);
 }
 
@@ -246,7 +271,8 @@ static void blitter_check_saved_CSOs(struct blitter_context_priv *ctx)
           ctx->blitter.saved_dsa_state != INVALID_PTR &&
           ctx->blitter.saved_rs_state != INVALID_PTR &&
           ctx->blitter.saved_fs != INVALID_PTR &&
-          ctx->blitter.saved_vs != INVALID_PTR);
+          ctx->blitter.saved_vs != INVALID_PTR &&
+          ctx->blitter.saved_velem_state != INVALID_PTR);
 }
 
 static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
@@ -259,12 +285,14 @@ static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
    pipe->bind_rasterizer_state(pipe, ctx->blitter.saved_rs_state);
    pipe->bind_fs_state(pipe, ctx->blitter.saved_fs);
    pipe->bind_vs_state(pipe, ctx->blitter.saved_vs);
+   pipe->bind_vertex_elements_state(pipe, ctx->blitter.saved_velem_state);
 
    ctx->blitter.saved_blend_state = INVALID_PTR;
    ctx->blitter.saved_dsa_state = INVALID_PTR;
    ctx->blitter.saved_rs_state = INVALID_PTR;
    ctx->blitter.saved_fs = INVALID_PTR;
    ctx->blitter.saved_vs = INVALID_PTR;
+   ctx->blitter.saved_velem_state = INVALID_PTR;
 
    pipe->set_stencil_ref(pipe, &ctx->blitter.saved_stencil_ref);
 
@@ -285,11 +313,18 @@ static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
       ctx->blitter.saved_num_sampler_states = ~0;
    }
 
-   if (ctx->blitter.saved_num_textures != ~0) {
-      pipe->set_fragment_sampler_textures(pipe,
-                                          ctx->blitter.saved_num_textures,
-                                          ctx->blitter.saved_textures);
-      ctx->blitter.saved_num_textures = ~0;
+   if (ctx->blitter.saved_num_sampler_views != ~0) {
+      pipe->set_fragment_sampler_views(pipe,
+                                       ctx->blitter.saved_num_sampler_views,
+                                       ctx->blitter.saved_sampler_views);
+      ctx->blitter.saved_num_sampler_views = ~0;
+   }
+
+   if (ctx->blitter.saved_num_vertex_buffers != ~0) {
+      pipe->set_vertex_buffers(pipe,
+                                       ctx->blitter.saved_num_vertex_buffers,
+                                       ctx->blitter.saved_vertex_buffers);
+      ctx->blitter.saved_num_vertex_buffers = ~0;
    }
 }
 
@@ -424,7 +459,7 @@ static void blitter_draw_quad(struct blitter_context_priv *ctx)
    struct pipe_context *pipe = ctx->pipe;
 
    /* write vertices and draw them */
-   pipe_buffer_write(pipe->screen, ctx->vbuf,
+   pipe_buffer_write(pipe, ctx->vbuf,
                      0, sizeof(ctx->vertices), ctx->vertices);
 
    util_draw_vertex_buffer(pipe, ctx->vbuf, 0, PIPE_PRIM_TRIANGLE_FAN,
@@ -569,6 +604,7 @@ void util_blitter_clear(struct blitter_context *blitter,
       pipe->bind_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
 
    pipe->bind_rasterizer_state(pipe, ctx->rs_state);
+   pipe->bind_vertex_elements_state(pipe, ctx->velem_state);
    pipe->bind_fs_state(pipe, blitter_get_fs_col(ctx, num_cbufs));
    pipe->bind_vs_state(pipe, ctx->vs_col);
 
@@ -600,9 +636,10 @@ static void util_blitter_do_copy(struct blitter_context *blitter,
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_framebuffer_state fb_state;
+   struct pipe_sampler_view viewTempl, *view;
 
    assert(blitter->saved_fb_state.nr_cbufs != ~0);
-   assert(blitter->saved_num_textures != ~0);
+   assert(blitter->saved_num_sampler_views != ~0);
    assert(blitter->saved_num_sampler_states != ~0);
    assert(src->texture->target < PIPE_MAX_TEXTURE_TYPES);
 
@@ -630,11 +667,24 @@ static void util_blitter_do_copy(struct blitter_context *blitter,
       fb_state.zsbuf = 0;
    }
 
+   u_sampler_view_default_template(&viewTempl,
+                                   src->texture,
+                                   src->texture->format);
+   view = pipe->create_sampler_view(pipe,
+                                    src->texture,
+                                    &viewTempl);
+
+   if (ctx->sampler_view) {
+      pipe_sampler_view_reference(&ctx->sampler_view, NULL);
+   }
+   ctx->sampler_view = view;
+
    pipe->bind_rasterizer_state(pipe, ctx->rs_state);
    pipe->bind_vs_state(pipe, ctx->vs_tex);
    pipe->bind_fragment_sampler_states(pipe, 1,
       blitter_get_sampler_state(ctx, src->level));
-   pipe->set_fragment_sampler_textures(pipe, 1, &src->texture);
+   pipe->bind_vertex_elements_state(pipe, ctx->velem_state);
+   pipe->set_fragment_sampler_views(pipe, 1, &view);
    pipe->set_framebuffer_state(pipe, &fb_state);
 
    /* set texture coordinates */
@@ -672,8 +722,8 @@ static void util_blitter_overlap_copy(struct blitter_context *blitter,
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
 
-   struct pipe_texture texTemp;
-   struct pipe_texture *texture;
+   struct pipe_resource texTemp;
+   struct pipe_resource *texture;
    struct pipe_surface *tex_surf;
 
    /* check whether the states are properly saved */
@@ -687,13 +737,13 @@ static void util_blitter_overlap_copy(struct blitter_context *blitter,
    texTemp.height0 = height;
    texTemp.depth0 = 1;
 
-   texture = screen->texture_create(screen, &texTemp);
+   texture = screen->resource_create(screen, &texTemp);
    if (!texture)
       return;
 
    tex_surf = screen->get_tex_surface(screen, texture, 0, 0, 0,
-				      PIPE_BUFFER_USAGE_GPU_READ | 
-				      PIPE_BUFFER_USAGE_GPU_WRITE);
+				      PIPE_BIND_BLIT_SOURCE | 
+				      PIPE_BIND_BLIT_DESTINATION);
 
    /* blit from the src to the temp */
    util_blitter_do_copy(blitter, tex_surf, 0, 0,
@@ -705,7 +755,7 @@ static void util_blitter_overlap_copy(struct blitter_context *blitter,
 			width, height,
 			FALSE);
    pipe_surface_reference(&tex_surf, NULL);
-   pipe_texture_reference(&texture, NULL);
+   pipe_resource_reference(&texture, NULL);
    blitter_restore_CSOs(ctx);
 }
 
@@ -739,8 +789,8 @@ void util_blitter_copy(struct blitter_context *blitter,
 		   
    is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
    is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
-   dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
-                                            PIPE_TEXTURE_USAGE_RENDER_TARGET;
+   dst_tex_usage = is_depth || is_stencil ? PIPE_BIND_DEPTH_STENCIL :
+                                            PIPE_BIND_RENDER_TARGET;
 
    /* check if we can sample from and render to the surfaces */
    /* (assuming copying a stencil buffer is not possible) */
@@ -748,7 +798,7 @@ void util_blitter_copy(struct blitter_context *blitter,
        !screen->is_format_supported(screen, dst->format, dst->texture->target,
                                     dst_tex_usage, 0) ||
        !screen->is_format_supported(screen, src->format, src->texture->target,
-                                    PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
+                                    PIPE_BIND_SAMPLER_VIEW, 0)) {
       util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
                         width, height);
       return;
@@ -785,7 +835,7 @@ void util_blitter_fill(struct blitter_context *blitter,
    /* check if we can render to the surface */
    if (util_format_is_depth_or_stencil(dst->format) || /* unlikely, but you never know */
        !screen->is_format_supported(screen, dst->format, dst->texture->target,
-                                    PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) {
+                                    PIPE_BIND_RENDER_TARGET, 0)) {
       util_surface_fill(pipe, dst, dstx, dsty, width, height, value);
       return;
    }
@@ -807,6 +857,7 @@ void util_blitter_fill(struct blitter_context *blitter,
    pipe->bind_rasterizer_state(pipe, ctx->rs_state);
    pipe->bind_fs_state(pipe, blitter_get_fs_col(ctx, 1));
    pipe->bind_vs_state(pipe, ctx->vs_col);
+   pipe->bind_vertex_elements_state(pipe, ctx->velem_state);
 
    /* set a framebuffer state */
    fb_state.width = dst->width;
diff --git a/src/gallium/auxiliary/util/u_blitter.h b/src/gallium/auxiliary/util/u_blitter.h
index 92008fce992..f6e3ce4874e 100644
--- a/src/gallium/auxiliary/util/u_blitter.h
+++ b/src/gallium/auxiliary/util/u_blitter.h
@@ -43,6 +43,7 @@ struct blitter_context
    /* Private members, really. */
    void *saved_blend_state;   /**< blend state */
    void *saved_dsa_state;     /**< depth stencil alpha state */
+   void *saved_velem_state;   /**< vertex elements state */
    void *saved_rs_state;      /**< rasterizer state */
    void *saved_fs, *saved_vs; /**< fragment shader, vertex shader */
 
@@ -52,10 +53,13 @@ struct blitter_context
    struct pipe_clip_state saved_clip;
 
    int saved_num_sampler_states;
-   void *saved_sampler_states[32];
+   void *saved_sampler_states[PIPE_MAX_SAMPLERS];
 
-   int saved_num_textures;
-   struct pipe_texture *saved_textures[32]; /* is 32 enough? */
+   int saved_num_sampler_views;
+   struct pipe_sampler_view *saved_sampler_views[PIPE_MAX_SAMPLERS];
+
+   int saved_num_vertex_buffers;
+   struct pipe_vertex_buffer saved_vertex_buffers[PIPE_MAX_ATTRIBS];
 };
 
 /**
@@ -173,6 +177,13 @@ void util_blitter_save_depth_stencil_alpha(struct blitter_context *blitter,
 }
 
 static INLINE
+void util_blitter_save_vertex_elements(struct blitter_context *blitter,
+                                       void *state)
+{
+   blitter->saved_velem_state = state;
+}
+
+static INLINE
 void util_blitter_save_stencil_ref(struct blitter_context *blitter,
                                    const struct pipe_stencil_ref *state)
 {
@@ -234,17 +245,30 @@ void util_blitter_save_fragment_sampler_states(
           num_sampler_states * sizeof(void *));
 }
 
-static INLINE
-void util_blitter_save_fragment_sampler_textures(
-                  struct blitter_context *blitter,
-                  int num_textures,
-                  struct pipe_texture **textures)
+static INLINE void
+util_blitter_save_fragment_sampler_views(struct blitter_context *blitter,
+                                         int num_views,
+                                         struct pipe_sampler_view **views)
+{
+   assert(num_views <= Elements(blitter->saved_sampler_views));
+
+   blitter->saved_num_sampler_views = num_views;
+   memcpy(blitter->saved_sampler_views,
+          views,
+          num_views * sizeof(struct pipe_sampler_view *));
+}
+
+static INLINE void
+util_blitter_save_vertex_buffers(struct blitter_context *blitter,
+                                         int num_vertex_buffers,
+                                         struct pipe_vertex_buffer *vertex_buffers)
 {
-   assert(num_textures <= Elements(blitter->saved_textures));
+   assert(num_vertex_buffers <= Elements(blitter->saved_vertex_buffers));
 
-   blitter->saved_num_textures = num_textures;
-   memcpy(blitter->saved_textures, textures,
-          num_textures * sizeof(struct pipe_texture *));
+   blitter->saved_num_vertex_buffers = num_vertex_buffers;
+   memcpy(blitter->saved_vertex_buffers,
+          vertex_buffers,
+          num_vertex_buffers * sizeof(struct pipe_vertex_buffer));
 }
 
 #ifdef __cplusplus
diff --git a/src/gallium/auxiliary/util/u_box.h b/src/gallium/auxiliary/util/u_box.h
new file mode 100644
index 00000000000..919967b55a7
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_box.h
@@ -0,0 +1,73 @@
+#ifndef UTIL_BOX_INLINES_H
+#define UTIL_BOX_INLINES_H
+
+#include "pipe/p_state.h"
+
+static INLINE
+void u_box_1d( unsigned x,
+	       unsigned w,
+	       struct pipe_box *box )
+{
+   box->x = x;
+   box->y = 0;
+   box->z = 0;
+   box->width = w;
+   box->height = 1;
+   box->depth = 1;
+}
+
+static INLINE
+void u_box_2d( unsigned x,
+	       unsigned y,
+	       unsigned w,
+	       unsigned h,
+	       struct pipe_box *box )
+{
+   box->x = x;
+   box->y = y;
+   box->z = 0;
+   box->width = w;
+   box->height = h;
+   box->depth = 1;
+}
+
+static INLINE
+void u_box_origin_2d( unsigned w,
+		      unsigned h,
+		      struct pipe_box *box )
+{
+   box->x = 0;
+   box->y = 0;
+   box->z = 0;
+   box->width = w;
+   box->height = h;
+   box->depth = 1;
+}
+
+static INLINE
+void u_box_2d_zslice( unsigned x,
+		      unsigned y,
+		      unsigned z,
+		      unsigned w,
+		      unsigned h,
+		      struct pipe_box *box )
+{
+   box->x = x;
+   box->y = y;
+   box->z = z;
+   box->width = w;
+   box->height = h;
+   box->depth = 1;
+}
+
+static INLINE
+struct pipe_subresource u_subresource( unsigned face,
+				       unsigned level )
+{
+   struct pipe_subresource subresource;
+   subresource.face = face;
+   subresource.level = level;
+   return subresource;
+}
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_cache.c b/src/gallium/auxiliary/util/u_cache.c
index 47c16b1c927..15f4a8831f2 100644
--- a/src/gallium/auxiliary/util/u_cache.c
+++ b/src/gallium/auxiliary/util/u_cache.c
@@ -202,7 +202,7 @@ util_cache_destroy(struct util_cache *cache)
       double stddev = sqrt(mean);
       unsigned i;
       for(i = 0; i < cache->size; ++i) {
-         double z = fabs(cache->count - mean)/stddev;
+         double z = fabs(cache->entries[i].count - mean)/stddev;
          /* This assert should not fail 99.9999998027% of the times, unless 
           * the hash function is a poor one */
          assert(z <= 6.0);
diff --git a/src/gallium/auxiliary/util/u_debug.c b/src/gallium/auxiliary/util/u_debug.c
index 94be682c4b1..0de38e791d6 100644
--- a/src/gallium/auxiliary/util/u_debug.c
+++ b/src/gallium/auxiliary/util/u_debug.c
@@ -74,6 +74,24 @@ void debug_print_blob( const char *name,
 #endif
 
 
+static boolean
+debug_get_option_should_print(void)
+{
+   static boolean first = TRUE;
+   static boolean value = FALSE;
+
+   if (!first)
+      return value;
+
+   /* Oh hey this will call into this function,
+    * but its cool since we set first to false
+    */
+   first = FALSE;
+   value = debug_get_bool_option("GALLIUM_PRINT_OPTIONS", TRUE);
+   /* XXX should we print this option? Currently it wont */
+   return value;
+}
+
 const char *
 debug_get_option(const char *name, const char *dfault)
 {
@@ -82,8 +100,9 @@ debug_get_option(const char *name, const char *dfault)
    result = os_get_option(name);
    if(!result)
       result = dfault;
-      
-   debug_printf("%s: %s = %s\n", __FUNCTION__, name, result ? result : "(null)");
+
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %s\n", __FUNCTION__, name, result ? result : "(null)");
    
    return result;
 }
@@ -109,7 +128,8 @@ debug_get_bool_option(const char *name, boolean dfault)
    else
       result = TRUE;
 
-   debug_printf("%s: %s = %s\n", __FUNCTION__, name, result ? "TRUE" : "FALSE");
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %s\n", __FUNCTION__, name, result ? "TRUE" : "FALSE");
    
    return result;
 }
@@ -142,8 +162,9 @@ debug_get_num_option(const char *name, long dfault)
       }
       result *= sign;
    }
-   
-   debug_printf("%s: %s = %li\n", __FUNCTION__, name, result);
+
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %li\n", __FUNCTION__, name, result);
 
    return result;
 }
@@ -176,11 +197,12 @@ debug_get_flags_option(const char *name,
       }
    }
 
-   if (str) {
-      debug_printf("%s: %s = 0x%lx (%s)\n", __FUNCTION__, name, result, str);
-   }
-   else {
-      debug_printf("%s: %s = 0x%lx\n", __FUNCTION__, name, result);
+   if (debug_get_option_should_print()) {
+      if (str) {
+         debug_printf("%s: %s = 0x%lx (%s)\n", __FUNCTION__, name, result, str);
+      } else {
+         debug_printf("%s: %s = 0x%lx\n", __FUNCTION__, name, result);
+      }
    }
 
    return result;
@@ -421,45 +443,51 @@ void debug_dump_image(const char *prefix,
 #endif
 }
 
-void debug_dump_surface(const char *prefix,
+void debug_dump_surface(struct pipe_context *pipe,
+			const char *prefix,
                         struct pipe_surface *surface)     
 {
-   struct pipe_texture *texture;
-   struct pipe_screen *screen;
+   struct pipe_resource *texture;
    struct pipe_transfer *transfer;
    void *data;
 
    if (!surface)
       return;
 
+   /* XXX: this doesn't necessarily work, as the driver may be using
+    * temporary storage for the surface which hasn't been propagated
+    * back into the texture.  Need to nail down the semantics of views
+    * and transfers a bit better before we can say if extra work needs
+    * to be done here:
+    */
    texture = surface->texture;
-   screen = texture->screen;
 
-   transfer = screen->get_tex_transfer(screen, texture, surface->face,
-                                       surface->level, surface->zslice,
-                                       PIPE_TRANSFER_READ, 0, 0, surface->width,
-                                       surface->height);
+   transfer = pipe_get_transfer(pipe, texture, surface->face,
+				     surface->level, surface->zslice,
+				     PIPE_TRANSFER_READ, 0, 0, surface->width,
+				     surface->height);
    
-   data = screen->transfer_map(screen, transfer);
+   data = pipe->transfer_map(pipe, transfer);
    if(!data)
       goto error;
    
    debug_dump_image(prefix, 
                     texture->format,
                     util_format_get_blocksize(texture->format), 
-                    util_format_get_nblocksx(texture->format, transfer->width),
-                    util_format_get_nblocksy(texture->format, transfer->height),
+                    util_format_get_nblocksx(texture->format, surface->width),
+                    util_format_get_nblocksy(texture->format, surface->height),
                     transfer->stride,
                     data);
    
-   screen->transfer_unmap(screen, transfer);
+   pipe->transfer_unmap(pipe, transfer);
 error:
-   screen->tex_transfer_destroy(transfer);
+   pipe->transfer_destroy(pipe, transfer);
 }
 
 
-void debug_dump_texture(const char *prefix,
-                        struct pipe_texture *texture)
+void debug_dump_texture(struct pipe_context *pipe,
+                        const char *prefix,
+                        struct pipe_resource *texture)
 {
    struct pipe_surface *surface;
    struct pipe_screen *screen;
@@ -471,9 +499,9 @@ void debug_dump_texture(const char *prefix,
 
    /* XXX for now, just dump image for face=0, level=0 */
    surface = screen->get_tex_surface(screen, texture, 0, 0, 0,
-                                     PIPE_TEXTURE_USAGE_SAMPLER);
+                                     PIPE_BIND_SAMPLER_VIEW);
    if (surface) {
-      debug_dump_surface(prefix, surface);
+      debug_dump_surface(pipe, prefix, surface);
       screen->tex_surface_destroy(surface);
    }
 }
@@ -511,27 +539,28 @@ struct bmp_rgb_quad {
 };
 
 void
-debug_dump_surface_bmp(const char *filename,
+debug_dump_surface_bmp(struct pipe_context *pipe,
+		       const char *filename,
                        struct pipe_surface *surface)
 {
 #ifndef PIPE_SUBSYSTEM_WINDOWS_MINIPORT
    struct pipe_transfer *transfer;
-   struct pipe_texture *texture = surface->texture;
-   struct pipe_screen *screen = texture->screen;
+   struct pipe_resource *texture = surface->texture;
 
-   transfer = screen->get_tex_transfer(screen, texture, surface->face,
-                                       surface->level, surface->zslice,
-                                       PIPE_TRANSFER_READ, 0, 0, surface->width,
-                                       surface->height);
+   transfer = pipe_get_transfer(pipe, texture, surface->face,
+				surface->level, surface->zslice,
+				PIPE_TRANSFER_READ, 0, 0, surface->width,
+				surface->height);
 
-   debug_dump_transfer_bmp(filename, transfer);
+   debug_dump_transfer_bmp(pipe, filename, transfer);
 
-   screen->tex_transfer_destroy(transfer);
+   pipe->transfer_destroy(pipe, transfer);
 #endif
 }
 
 void
-debug_dump_transfer_bmp(const char *filename,
+debug_dump_transfer_bmp(struct pipe_context *pipe,
+                        const char *filename,
                         struct pipe_transfer *transfer)
 {
 #ifndef PIPE_SUBSYSTEM_WINDOWS_MINIPORT
@@ -540,17 +569,20 @@ debug_dump_transfer_bmp(const char *filename,
    if (!transfer)
       goto error1;
 
-   rgba = MALLOC(transfer->width*transfer->height*4*sizeof(float));
+   rgba = MALLOC(transfer->box.width *
+		 transfer->box.height *
+		 transfer->box.depth *
+		 4*sizeof(float));
    if(!rgba)
       goto error1;
 
-   pipe_get_tile_rgba(transfer, 0, 0,
-                      transfer->width, transfer->height,
+   pipe_get_tile_rgba(pipe, transfer, 0, 0,
+                      transfer->box.width, transfer->box.height,
                       rgba);
 
    debug_dump_float_rgba_bmp(filename,
-                             transfer->width, transfer->height,
-                             rgba, transfer->width);
+                             transfer->box.width, transfer->box.height,
+                             rgba, transfer->box.width);
 
    FREE(rgba);
 error1:
diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h
index 0f4768f3444..e8ff2773e69 100644
--- a/src/gallium/auxiliary/util/u_debug.h
+++ b/src/gallium/auxiliary/util/u_debug.h
@@ -303,6 +303,45 @@ debug_get_flags_option(const char *name,
                        const struct debug_named_value *flags,
                        unsigned long dfault);
 
+#define DEBUG_GET_ONCE_BOOL_OPTION(sufix, name, dfault) \
+static boolean \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static boolean value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_bool_option(name, dfault); \
+   } \
+   return value; \
+}
+
+#define DEBUG_GET_ONCE_NUM_OPTION(sufix, name, dfault) \
+static long \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static long value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_num_option(name, dfault); \
+   } \
+   return value; \
+}
+
+#define DEBUG_GET_ONCE_FLAGS_OPTION(sufix, name, flags, dfault) \
+static unsigned long \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static unsigned long value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_flags_option(name, flags, dfault); \
+   } \
+   return value; \
+}
+
 
 unsigned long
 debug_memory_begin(void);
@@ -312,30 +351,35 @@ debug_memory_end(unsigned long beginning);
 
 
 #ifdef DEBUG
+struct pipe_context;
 struct pipe_surface;
 struct pipe_transfer;
-struct pipe_texture;
+struct pipe_resource;
 
 void debug_dump_image(const char *prefix,
                       unsigned format, unsigned cpp,
                       unsigned width, unsigned height,
                       unsigned stride,
                       const void *data);
-void debug_dump_surface(const char *prefix,
+void debug_dump_surface(struct pipe_context *pipe,
+			const char *prefix,
                         struct pipe_surface *surface);   
-void debug_dump_texture(const char *prefix,
-                        struct pipe_texture *texture);
-void debug_dump_surface_bmp(const char *filename,
+void debug_dump_texture(struct pipe_context *pipe,
+			const char *prefix,
+                        struct pipe_resource *texture);
+void debug_dump_surface_bmp(struct pipe_context *pipe,
+                            const char *filename,
                             struct pipe_surface *surface);
-void debug_dump_transfer_bmp(const char *filename,
+void debug_dump_transfer_bmp(struct pipe_context *pipe,
+                             const char *filename,
                              struct pipe_transfer *transfer);
 void debug_dump_float_rgba_bmp(const char *filename,
                                unsigned width, unsigned height,
                                float *rgba, unsigned stride);
 #else
 #define debug_dump_image(prefix, format, cpp, width, height, stride, data) ((void)0)
-#define debug_dump_surface(prefix, surface) ((void)0)
-#define debug_dump_surface_bmp(filename, surface) ((void)0)
+#define debug_dump_surface(pipe, prefix, surface) ((void)0)
+#define debug_dump_surface_bmp(pipe, filename, surface) ((void)0)
 #define debug_dump_transfer_bmp(filename, transfer) ((void)0)
 #define debug_dump_float_rgba_bmp(filename, width, height, rgba, stride) ((void)0)
 #endif
diff --git a/src/gallium/auxiliary/util/u_dirty_surfaces.h b/src/gallium/auxiliary/util/u_dirty_surfaces.h
new file mode 100644
index 00000000000..99f260bf967
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_dirty_surfaces.h
@@ -0,0 +1,88 @@
+#ifndef U_DIRTY_SURFACES_H_
+#define U_DIRTY_SURFACES_H_
+
+#include "util/u_double_list.h"
+#include "util/u_math.h"
+
+typedef void (*util_dirty_surface_flush_t) (struct pipe_context *, struct pipe_surface *);
+
+struct util_dirty_surfaces
+{
+   struct list_head dirty_list;
+};
+
+struct util_dirty_surface
+{
+   struct pipe_surface base;
+   struct list_head dirty_list;
+};
+
+static INLINE void
+util_dirty_surfaces_init(struct util_dirty_surfaces *ds)
+{
+   LIST_INITHEAD(&ds->dirty_list);
+}
+
+static INLINE void
+util_dirty_surfaces_use_for_sampling(struct pipe_context *pipe, struct util_dirty_surfaces *dss, util_dirty_surface_flush_t flush)
+{
+   struct list_head *p, *next;
+   for(p = dss->dirty_list.next; p != &dss->dirty_list; p = next)
+   {
+      struct util_dirty_surface *ds = LIST_ENTRY(struct util_dirty_surface, p, dirty_list);
+      next = p->next;
+
+      flush(pipe, &ds->base);
+   }
+}
+
+static INLINE void
+util_dirty_surfaces_use_levels_for_sampling(struct pipe_context *pipe, struct util_dirty_surfaces *dss, unsigned first, unsigned last, util_dirty_surface_flush_t flush)
+{
+   struct list_head *p, *next;
+   if(first > last)
+      return;
+   for(p = dss->dirty_list.next; p != &dss->dirty_list; p = next)
+   {
+      struct util_dirty_surface *ds = LIST_ENTRY(struct util_dirty_surface, p, dirty_list);
+      next = p->next;
+
+      if(ds->base.level >= first && ds->base.level <= last)
+	 flush(pipe, &ds->base);
+   }
+}
+
+static INLINE void
+util_dirty_surfaces_use_for_sampling_with(struct pipe_context *pipe, struct util_dirty_surfaces *dss, struct pipe_sampler_view *psv, struct pipe_sampler_state *pss, util_dirty_surface_flush_t flush)
+{
+   if(!LIST_IS_EMPTY(&dss->dirty_list))
+      util_dirty_surfaces_use_levels_for_sampling(pipe, dss, (unsigned)pss->min_lod + psv->first_level, MIN2((unsigned)ceilf(pss->max_lod) + psv->first_level, psv->last_level), flush);
+}
+
+static INLINE void
+util_dirty_surface_init(struct util_dirty_surface *ds)
+{
+   LIST_INITHEAD(&ds->dirty_list);
+}
+
+static INLINE boolean
+util_dirty_surface_is_dirty(struct util_dirty_surface *ds)
+{
+   return !LIST_IS_EMPTY(&ds->dirty_list);
+}
+
+static INLINE void
+util_dirty_surface_set_dirty(struct util_dirty_surfaces *dss, struct util_dirty_surface *ds)
+{
+   if(LIST_IS_EMPTY(&ds->dirty_list))
+      LIST_ADDTAIL(&ds->dirty_list, &dss->dirty_list);
+}
+
+static INLINE void
+util_dirty_surface_set_clean(struct util_dirty_surfaces *dss, struct util_dirty_surface *ds)
+{
+   if(!LIST_IS_EMPTY(&ds->dirty_list))
+      LIST_DELINIT(&ds->dirty_list);
+}
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_dl.c b/src/gallium/auxiliary/util/u_dl.c
index 37ed789f955..220860ebf4b 100644
--- a/src/gallium/auxiliary/util/u_dl.c
+++ b/src/gallium/auxiliary/util/u_dl.c
@@ -78,3 +78,16 @@ util_dl_close(struct util_dl_library *library)
    (void)library;
 #endif
 }
+
+
+const char *
+util_dl_error(void)
+{
+#if defined(PIPE_OS_UNIX)
+   return dlerror();
+#elif defined(PIPE_OS_WINDOWS)
+   return "unknown error";
+#else
+   return "unknown error";
+#endif
+}
diff --git a/src/gallium/auxiliary/util/u_dl.h b/src/gallium/auxiliary/util/u_dl.h
index 85296c58af6..2853b447c61 100644
--- a/src/gallium/auxiliary/util/u_dl.h
+++ b/src/gallium/auxiliary/util/u_dl.h
@@ -70,4 +70,11 @@ void
 util_dl_close(struct util_dl_library *library);
 
 
+/**
+ * Return most recent error message.
+ */
+const char *
+util_dl_error(void);
+
+
 #endif /* U_DL_H_ */
diff --git a/src/gallium/auxiliary/util/u_draw_quad.c b/src/gallium/auxiliary/util/u_draw_quad.c
index 14506e84519..b37b48b5aef 100644
--- a/src/gallium/auxiliary/util/u_draw_quad.c
+++ b/src/gallium/auxiliary/util/u_draw_quad.c
@@ -30,6 +30,7 @@
 #include "pipe/p_defines.h"
 #include "util/u_inlines.h"
 #include "util/u_draw_quad.h"
+#include "util/u_memory.h"
 
 
 /**
@@ -38,15 +39,13 @@
  */
 void 
 util_draw_vertex_buffer(struct pipe_context *pipe,
-                        struct pipe_buffer *vbuf,
+                        struct pipe_resource *vbuf,
                         uint offset,
                         uint prim_type,
                         uint num_verts,
                         uint num_attribs)
 {
    struct pipe_vertex_buffer vbuffer;
-   struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS];
-   uint i;
 
    assert(num_attribs <= PIPE_MAX_ATTRIBS);
 
@@ -58,15 +57,7 @@ util_draw_vertex_buffer(struct pipe_context *pipe,
    vbuffer.max_index = num_verts - 1;
    pipe->set_vertex_buffers(pipe, 1, &vbuffer);
 
-   /* tell pipe about the vertex attributes */
-   for (i = 0; i < num_attribs; i++) {
-      velements[i].src_offset = i * 4 * sizeof(float);
-      velements[i].instance_divisor = 0;
-      velements[i].vertex_buffer_index = 0;
-      velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
-      velements[i].nr_components = 4;
-   }
-   pipe->set_vertex_elements(pipe, num_attribs, velements);
+   /* note: vertex elements already set by caller */
 
    /* draw */
    pipe->draw_arrays(pipe, prim_type, 0, num_verts);
@@ -76,60 +67,63 @@ util_draw_vertex_buffer(struct pipe_context *pipe,
 
 /**
  * Draw screen-aligned textured quad.
- * Note: this function allocs/destroys a vertex buffer and isn't especially
- * efficient.
+ * Note: this isn't especially efficient.
  */
 void 
 util_draw_texquad(struct pipe_context *pipe,
                   float x0, float y0, float x1, float y1, float z)
 {
-   struct pipe_buffer *vbuf;
-   uint numAttribs = 2, vertexBytes, i, j;
-
-   vertexBytes = 4 * (4 * numAttribs * sizeof(float));
-
-   /* XXX create one-time */
-   vbuf = pipe_buffer_create(pipe->screen, 32,
-                             PIPE_BUFFER_USAGE_VERTEX, vertexBytes);
-   if (vbuf) {
-      float *v = (float *) pipe_buffer_map(pipe->screen, vbuf,
-                                           PIPE_BUFFER_USAGE_CPU_WRITE);
-      if (v) {
-         /*
-          * Load vertex buffer
-          */
-         for (i = j = 0; i < 4; i++) {
-            v[j + 2] = z;   /* z */
-            v[j + 3] = 1.0; /* w */
-            v[j + 6] = 0.0; /* r */
-            v[j + 7] = 1.0; /* q */
-            j += 8;
-         }
-
-         v[0] = x0;
-         v[1] = y0;
-         v[4] = 0.0; /*s*/
-         v[5] = 0.0; /*t*/
-
-         v[8] = x1;
-         v[9] = y0;
-         v[12] = 1.0;
-         v[13] = 0.0;
-
-         v[16] = x1;
-         v[17] = y1;
-         v[20] = 1.0;
-         v[21] = 1.0;
-
-         v[24] = x0;
-         v[25] = y1;
-         v[28] = 0.0;
-         v[29] = 1.0;
-
-         pipe_buffer_unmap(pipe->screen, vbuf);
-         util_draw_vertex_buffer(pipe, vbuf, 0, PIPE_PRIM_TRIANGLE_FAN, 4, 2);
-      }
-
-      pipe_buffer_reference(&vbuf, NULL);
+   uint numAttribs = 2, i, j;
+   uint vertexBytes = 4 * (4 * numAttribs * sizeof(float));
+   struct pipe_resource *vbuf = NULL;  
+   uint *v = NULL;
+
+   v = MALLOC(vertexBytes);
+   if (v == NULL)
+      goto out;
+
+   /*
+    * Load vertex buffer
+    */
+   for (i = j = 0; i < 4; i++) {
+      v[j + 2] = z;   /* z */
+      v[j + 3] = 1.0; /* w */
+      v[j + 6] = 0.0; /* r */
+      v[j + 7] = 1.0; /* q */
+      j += 8;
    }
+
+   v[0] = x0;
+   v[1] = y0;
+   v[4] = 0.0; /*s*/
+   v[5] = 0.0; /*t*/
+
+   v[8] = x1;
+   v[9] = y0;
+   v[12] = 1.0;
+   v[13] = 0.0;
+
+   v[16] = x1;
+   v[17] = y1;
+   v[20] = 1.0;
+   v[21] = 1.0;
+
+   v[24] = x0;
+   v[25] = y1;
+   v[28] = 0.0;
+   v[29] = 1.0;
+	 
+   vbuf = pipe_user_buffer_create(pipe->screen, v, vertexBytes,
+				  PIPE_BIND_VERTEX_BUFFER);
+   if (!vbuf) 
+      goto out;
+
+   util_draw_vertex_buffer(pipe, vbuf, 0, PIPE_PRIM_TRIANGLE_FAN, 4, 2);
+
+out:
+   if (vbuf)
+      pipe_resource_reference(&vbuf, NULL);
+   
+   if (v)
+      FREE(v);
 }
diff --git a/src/gallium/auxiliary/util/u_draw_quad.h b/src/gallium/auxiliary/util/u_draw_quad.h
index 00d3f5b7158..42eb1844289 100644
--- a/src/gallium/auxiliary/util/u_draw_quad.h
+++ b/src/gallium/auxiliary/util/u_draw_quad.h
@@ -33,11 +33,11 @@
 extern "C" {
 #endif
 
-struct pipe_buffer;
+struct pipe_resource;
 
 extern void 
 util_draw_vertex_buffer(struct pipe_context *pipe,
-                        struct pipe_buffer *vbuf, uint offset,
+                        struct pipe_resource *vbuf, uint offset,
                         uint num_attribs, uint num_verts, uint prim_type);
 
 
diff --git a/src/gallium/auxiliary/util/u_dump.h b/src/gallium/auxiliary/util/u_dump.h
index 379f18ef38b..bdc73ac47d2 100644
--- a/src/gallium/auxiliary/util/u_dump.h
+++ b/src/gallium/auxiliary/util/u_dump.h
@@ -92,7 +92,7 @@ util_dump_tex_filter(unsigned value, boolean shortened);
 
 void
 util_dump_template(struct os_stream *stream,
-                   const struct pipe_texture *templat);
+                   const struct pipe_resource *templat);
 
 void
 util_dump_rasterizer_state(struct os_stream *stream,
diff --git a/src/gallium/auxiliary/util/u_dump_state.c b/src/gallium/auxiliary/util/u_dump_state.c
index ae7afd7311e..2ce643e90cd 100644
--- a/src/gallium/auxiliary/util/u_dump_state.c
+++ b/src/gallium/auxiliary/util/u_dump_state.c
@@ -255,14 +255,14 @@ util_dump_enum_func(struct os_stream *stream, unsigned value)
 
 
 void
-util_dump_template(struct os_stream *stream, const struct pipe_texture *templat)
+util_dump_template(struct os_stream *stream, const struct pipe_resource *templat)
 {
    if(!templat) {
       util_dump_null(stream);
       return;
    }
 
-   util_dump_struct_begin(stream, "pipe_texture");
+   util_dump_struct_begin(stream, "pipe_resource");
 
    util_dump_member(stream, int, templat, target);
    util_dump_member(stream, format, templat, format);
@@ -280,7 +280,9 @@ util_dump_template(struct os_stream *stream, const struct pipe_texture *templat)
    util_dump_member_end(stream);
 
    util_dump_member(stream, uint, templat, last_level);
-   util_dump_member(stream, uint, templat, tex_usage);
+   util_dump_member(stream, uint, templat, usage);
+   util_dump_member(stream, uint, templat, bind);
+   util_dump_member(stream, uint, templat, flags);
 
    util_dump_struct_end(stream);
 }
@@ -653,16 +655,13 @@ util_dump_transfer(struct os_stream *stream, const struct pipe_transfer *state)
 
    util_dump_struct_begin(stream, "pipe_transfer");
 
-   util_dump_member(stream, uint, state, width);
-   util_dump_member(stream, uint, state, height);
+   util_dump_member(stream, ptr, state, resource);
+   /*util_dump_member(stream, uint, state, box);*/
 
    util_dump_member(stream, uint, state, stride);
-   util_dump_member(stream, uint, state, usage);
+   util_dump_member(stream, uint, state, slice_stride);
 
-   util_dump_member(stream, ptr, state, texture);
-   util_dump_member(stream, uint, state, face);
-   util_dump_member(stream, uint, state, level);
-   util_dump_member(stream, uint, state, zslice);
+   /*util_dump_member(stream, ptr, state, data);*/
 
    util_dump_struct_end(stream);
 }
@@ -700,7 +699,6 @@ util_dump_vertex_element(struct os_stream *stream, const struct pipe_vertex_elem
    util_dump_member(stream, uint, state, src_offset);
 
    util_dump_member(stream, uint, state, vertex_buffer_index);
-   util_dump_member(stream, uint, state, nr_components);
 
    util_dump_member(stream, format, state, src_format);
 
diff --git a/src/gallium/auxiliary/util/u_dynarray.h b/src/gallium/auxiliary/util/u_dynarray.h
new file mode 100644
index 00000000000..9d1c1713a7c
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_dynarray.h
@@ -0,0 +1,111 @@
+/**************************************************************************
+ *
+ * Copyright 2010 Luca Barbieri
+ *
+ * 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 (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.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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 U_DYNARRAY_H
+#define U_DYNARRAY_H
+
+#include "pipe/p_compiler.h"
+#include "util/u_memory.h"
+
+/* A zero-initialized version of this is guaranteed to represent an
+ * empty array.
+ *
+ * Also, size <= capacity and data != 0 if and only if capacity != 0
+ * capacity will always be the allocation size of data
+ */
+struct util_dynarray
+{
+   void *data;
+   unsigned size;
+   unsigned capacity;
+};
+
+static INLINE void
+util_dynarray_init(struct util_dynarray *buf)
+{
+   memset(buf, 0, sizeof(*buf));
+}
+
+static INLINE void
+util_dynarray_fini(struct util_dynarray *buf)
+{
+   if(buf->data)
+   {
+      FREE(buf->data);
+      util_dynarray_init(buf);
+   }
+}
+
+/* use util_dynarray_trim to reduce the allocated storage */
+static INLINE void *
+util_dynarray_resize(struct util_dynarray *buf, unsigned newsize)
+{
+   char *p;
+   if(newsize > buf->capacity)
+   {
+      unsigned newcap = buf->capacity << 1;
+      if(newsize > newcap)
+	      newcap = newsize;
+      buf->data = REALLOC(buf->data, buf->capacity, newcap);
+      buf->capacity = newcap;
+   }
+
+   p = (char *)buf->data + buf->size;
+   buf->size = newsize;
+   return p;
+}
+
+static INLINE void *
+util_dynarray_grow(struct util_dynarray *buf, int diff)
+{
+   return util_dynarray_resize(buf, buf->size + diff);
+}
+
+static INLINE void
+util_dynarray_trim(struct util_dynarray *buf)
+{
+   if (buf->size != buf->capacity) {
+      if (buf->size) {
+         buf->data = REALLOC(buf->data, buf->capacity, buf->size);
+         buf->capacity = buf->size;
+      }
+      else {
+         FREE(buf->data);
+         buf->data = 0;
+         buf->capacity = 0;
+      }
+   }
+}
+
+#define util_dynarray_append(buf, type, v) do {type __v = (v); memcpy(util_dynarray_grow((buf), sizeof(type)), &__v, sizeof(type));} while(0)
+#define util_dynarray_top_ptr(buf, type) (type*)((char*)(buf)->data + (buf)->size - sizeof(type))
+#define util_dynarray_top(buf, type) *util_dynarray_top_ptr(buf, type)
+#define util_dynarray_pop_ptr(buf, type) (type*)((char*)(buf)->data + ((buf)->size -= sizeof(type)))
+#define util_dynarray_pop(buf, type) *util_dynarray_pop_ptr(buf, type)
+#define util_dynarray_contains(buf, type) ((buf)->size >= sizeof(type))
+
+#endif /* U_DYNARRAY_H */
+
diff --git a/src/gallium/auxiliary/util/u_format.c b/src/gallium/auxiliary/util/u_format.c
new file mode 100644
index 00000000000..c50c807eb89
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format.c
@@ -0,0 +1,286 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE 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.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Pixel format accessor functions.
+ *
+ * @author Jose Fonseca <[email protected]>
+ */
+
+#include "u_math.h"
+#include "u_memory.h"
+#include "u_rect.h"
+#include "u_format.h"
+
+
+void
+util_format_read_4f(enum pipe_format format,
+                    float *dst, unsigned dst_stride,
+                    const void *src, unsigned src_stride,
+                    unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   const uint8_t *src_row;
+   float *dst_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   src_row = (const uint8_t *)src + y*src_stride + x*(format_desc->block.bits/8);
+   dst_row = dst;
+
+   format_desc->unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_write_4f(enum pipe_format format,
+                     const float *src, unsigned src_stride,
+                     void *dst, unsigned dst_stride,
+                     unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   uint8_t *dst_row;
+   const float *src_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   dst_row = (uint8_t *)dst + y*dst_stride + x*(format_desc->block.bits/8);
+   src_row = src;
+
+   format_desc->pack_rgba_float(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_read_4ub(enum pipe_format format, uint8_t *dst, unsigned dst_stride, const void *src, unsigned src_stride, unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   const uint8_t *src_row;
+   uint8_t *dst_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   src_row = (const uint8_t *)src + y*src_stride + x*(format_desc->block.bits/8);
+   dst_row = dst;
+
+   format_desc->unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+void
+util_format_write_4ub(enum pipe_format format, const uint8_t *src, unsigned src_stride, void *dst, unsigned dst_stride, unsigned x, unsigned y, unsigned w, unsigned h)
+{
+   const struct util_format_description *format_desc;
+   uint8_t *dst_row;
+   const uint8_t *src_row;
+
+   format_desc = util_format_description(format);
+
+   assert(x % format_desc->block.width == 0);
+   assert(y % format_desc->block.height == 0);
+
+   dst_row = (uint8_t *)dst + y*dst_stride + x*(format_desc->block.bits/8);
+   src_row = src;
+
+   format_desc->pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, w, h);
+}
+
+
+static INLINE boolean
+util_format_fits_8unorm(const struct util_format_description *format_desc)
+{
+   unsigned chan;
+
+   switch (format_desc->layout) {
+
+   case UTIL_FORMAT_LAYOUT_S3TC:
+   case UTIL_FORMAT_LAYOUT_RGTC:
+      /*
+       * These are straight forward.
+       */
+
+      return TRUE;
+
+   case UTIL_FORMAT_LAYOUT_PLAIN:
+      /*
+       * For these we can find a generic rule.
+       */
+
+      for (chan = 0; chan < format_desc->nr_channels; ++chan) {
+         switch (format_desc->channel[chan].type) {
+         case UTIL_FORMAT_TYPE_VOID:
+            break;
+         case UTIL_FORMAT_TYPE_UNSIGNED:
+            if (!format_desc->channel[chan].normalized ||
+                format_desc->channel[chan].size > 8) {
+               return FALSE;
+            }
+            break;
+         default:
+            return FALSE;
+         }
+      }
+      return TRUE;
+
+   default:
+      /*
+       * Handle all others on a case by case basis.
+       */
+
+      switch (format_desc->format) {
+      case PIPE_FORMAT_R1_UNORM:
+      case PIPE_FORMAT_UYVY:
+      case PIPE_FORMAT_YUYV:
+      case PIPE_FORMAT_R8G8_B8G8_UNORM:
+      case PIPE_FORMAT_G8R8_G8B8_UNORM:
+         return TRUE;
+
+      default:
+         return FALSE;
+      }
+   }
+}
+
+
+void
+util_format_translate(enum pipe_format dst_format,
+                      void *dst, unsigned dst_stride,
+                      unsigned dst_x, unsigned dst_y,
+                      enum pipe_format src_format,
+                      const void *src, unsigned src_stride,
+                      unsigned src_x, unsigned src_y,
+                      unsigned width, unsigned height)
+{
+   const struct util_format_description *dst_format_desc;
+   const struct util_format_description *src_format_desc;
+   uint8_t *dst_row;
+   const uint8_t *src_row;
+   unsigned y_step;
+   unsigned dst_step;
+   unsigned src_step;
+
+   if (dst_format == src_format) {
+      /*
+       * Trivial case.
+       */
+
+      util_copy_rect(dst, dst_format, dst_stride,  dst_x, dst_y,
+                     width, height, src, (int)src_stride,
+                     src_x, src_y);
+      return;
+   }
+
+   dst_format_desc = util_format_description(dst_format);
+   src_format_desc = util_format_description(src_format);
+
+   assert(dst_x % dst_format_desc->block.width == 0);
+   assert(dst_y % dst_format_desc->block.height == 0);
+   assert(src_x % src_format_desc->block.width == 0);
+   assert(src_y % src_format_desc->block.height == 0);
+
+   dst_row = (uint8_t *)dst + dst_y*dst_stride + dst_x*(dst_format_desc->block.bits/8);
+   src_row = (const uint8_t *)src + src_y*src_stride + src_x*(src_format_desc->block.bits/8);
+
+   /*
+    * This works because all pixel formats have pixel blocks with power of two
+    * sizes.
+    */
+
+   y_step = MAX2(dst_format_desc->block.height, src_format_desc->block.height);
+   assert(y_step % dst_format_desc->block.height == 0);
+   assert(y_step % src_format_desc->block.height == 0);
+
+   dst_step = y_step / dst_format_desc->block.height * dst_stride;
+   src_step = y_step / src_format_desc->block.height * src_stride;
+
+   /*
+    * TODO: double formats will loose precision
+    * TODO: Add a special case for formats that are mere swizzles of each other
+    */
+
+   if (util_format_fits_8unorm(src_format_desc) ||
+       util_format_fits_8unorm(dst_format_desc)) {
+      unsigned tmp_stride;
+      uint8_t *tmp_row;
+
+      tmp_stride = width * 4 * sizeof *tmp_row;
+      tmp_row = MALLOC(y_step * tmp_stride);
+      if (!tmp_row)
+         return;
+
+      while (height >= y_step) {
+         src_format_desc->unpack_rgba_8unorm(tmp_row, tmp_stride, src_row, src_stride, width, y_step);
+         dst_format_desc->pack_rgba_8unorm(dst_row, dst_stride, tmp_row, tmp_stride, width, y_step);
+
+         dst_row += dst_step;
+         src_row += src_step;
+         height -= y_step;
+      }
+
+      if (height) {
+         src_format_desc->unpack_rgba_8unorm(tmp_row, tmp_stride, src_row, src_stride, width, height);
+         dst_format_desc->pack_rgba_8unorm(dst_row, dst_stride, tmp_row, tmp_stride, width, height);
+      }
+
+      FREE(tmp_row);
+   }
+   else {
+      unsigned tmp_stride;
+      float *tmp_row;
+
+      tmp_stride = width * 4 * sizeof *tmp_row;
+      tmp_row = MALLOC(y_step * tmp_stride);
+      if (!tmp_row)
+         return;
+
+      while (height >= y_step) {
+         src_format_desc->unpack_rgba_float(tmp_row, tmp_stride, src_row, src_stride, width, y_step);
+         dst_format_desc->pack_rgba_float(dst_row, dst_stride, tmp_row, tmp_stride, width, y_step);
+
+         dst_row += dst_step;
+         src_row += src_step;
+         height -= y_step;
+      }
+
+      if (height) {
+         src_format_desc->unpack_rgba_float(tmp_row, tmp_stride, src_row, src_stride, width, height);
+         dst_format_desc->pack_rgba_float(dst_row, dst_stride, tmp_row, tmp_stride, width, height);
+      }
+
+      FREE(tmp_row);
+   }
+}
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index 9a7e4b9cb07..0811280b97b 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -47,31 +47,41 @@
 # - color space: rgb, yub, sz
 #
 # See also:
-# - http://msdn.microsoft.com/en-us/library/ee416489.aspx (D3D9)
-# - http://msdn.microsoft.com/en-us/library/ee415668.aspx (D3D9 -> D3D10)
-# - http://msdn.microsoft.com/en-us/library/ee418116.aspx (D3D10)
+# - http://msdn.microsoft.com/en-us/library/bb172558.aspx (D3D9)
+# - http://msdn.microsoft.com/en-us/library/bb205073.aspx#mapping_texture_formats (D3D9 -> D3D10)
+# - http://msdn.microsoft.com/en-us/library/bb173059.aspx (D3D10)
 #
 # Note that GL doesn't really specify the layout of internal formats. See
 # OpenGL 2.1 specification, Table 3.16, on the "Correspondence of sized
 # internal formats to base in- ternal formats, and desired component
 # resolutions for each sized internal format."
 
+# None
+# Described as regular uint_8 bytes, i.e. PIPE_FORMAT_R8_USCALED
+PIPE_FORMAT_NONE                  , plain, 1, 1, u8  ,     ,     ,     , x001, rgb
+
 # Typical rendertarget formats
 PIPE_FORMAT_B8G8R8A8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , zyxw, rgb
-PIPE_FORMAT_B8G8R8X8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , zyx1, rgb
+PIPE_FORMAT_B8G8R8X8_UNORM        , plain, 1, 1, un8 , un8 , un8 , x8  , zyx1, rgb
 PIPE_FORMAT_A8R8G8B8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , yzwx, rgb
-PIPE_FORMAT_X8R8G8B8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , yzw1, rgb
+PIPE_FORMAT_X8R8G8B8_UNORM        , plain, 1, 1, x8  , un8 , un8 , un8 , yzw1, rgb
 PIPE_FORMAT_A8B8G8R8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , wzyx, rgb
-PIPE_FORMAT_X8B8G8R8_UNORM        , plain, 1, 1, un8 , un8 , un8 , un8 , wzy1, rgb
+PIPE_FORMAT_X8B8G8R8_UNORM        , plain, 1, 1, x8  , un8 , un8 , un8 , wzy1, rgb
+# PIPE_FORMAT_R8G8B8A8_UNORM is below
+PIPE_FORMAT_R8G8B8X8_UNORM        , plain, 1, 1, un8 , un8 , un8 , x8  , xyz1, rgb
+PIPE_FORMAT_B5G5R5X1_UNORM        , plain, 1, 1, un5 , un5 , un5 , x1  , zyx1, rgb
 PIPE_FORMAT_B5G5R5A1_UNORM        , plain, 1, 1, un5 , un5 , un5 , un1 , zyxw, rgb
 PIPE_FORMAT_B4G4R4A4_UNORM        , plain, 1, 1, un4 , un4 , un4 , un4 , zyxw, rgb
+PIPE_FORMAT_B4G4R4X4_UNORM        , plain, 1, 1, un4 , un4 , un4 , x4  , zyx1, rgb
 PIPE_FORMAT_B5G6R5_UNORM          , plain, 1, 1, un5 , un6 , un5 ,     , zyx1, rgb
 PIPE_FORMAT_R10G10B10A2_UNORM     , plain, 1, 1, un10, un10, un10, un2 , xyzw, rgb
+PIPE_FORMAT_B10G10R10A2_UNORM     , plain, 1, 1, un10, un10, un10, un2 , zyxw, rgb
 
 # Luminance/Intensity/Alpha formats
 PIPE_FORMAT_L8_UNORM              , plain, 1, 1, un8 ,     ,     ,     , xxx1, rgb
 PIPE_FORMAT_A8_UNORM              , plain, 1, 1, un8 ,     ,     ,     , 000x, rgb
 PIPE_FORMAT_I8_UNORM              , plain, 1, 1, un8 ,     ,     ,     , xxxx, rgb
+PIPE_FORMAT_L4A4_UNORM            , plain, 1, 1, un4 , un4 ,     ,     , xxxy, rgb
 PIPE_FORMAT_L8A8_UNORM            , plain, 1, 1, un8 , un8 ,     ,     , xxxy, rgb
 PIPE_FORMAT_L16_UNORM             , plain, 1, 1, un16,     ,     ,     , xxx1, rgb
 
@@ -79,44 +89,65 @@ PIPE_FORMAT_L16_UNORM             , plain, 1, 1, un16,     ,     ,     , xxx1, r
 PIPE_FORMAT_L8_SRGB               , plain, 1, 1, un8 ,     ,     ,     , xxx1, srgb 
 PIPE_FORMAT_L8A8_SRGB             , plain, 1, 1, un8 , un8 ,     ,     , xxxy, srgb 
 PIPE_FORMAT_R8G8B8_SRGB           , plain, 1, 1, un8 , un8 , un8 ,     , xyz1, srgb 
+PIPE_FORMAT_R8G8B8A8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , xyzw, srgb 
 PIPE_FORMAT_A8B8G8R8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , wzyx, srgb
-PIPE_FORMAT_X8B8G8R8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , wzy1, srgb
+PIPE_FORMAT_X8B8G8R8_SRGB         , plain, 1, 1, x8  , un8 , un8 , un8 , wzy1, srgb
 PIPE_FORMAT_B8G8R8A8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , zyxw, srgb
-PIPE_FORMAT_B8G8R8X8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , zyx1, srgb
+PIPE_FORMAT_B8G8R8X8_SRGB         , plain, 1, 1, un8 , un8 , un8 , x8  , zyx1, srgb
 PIPE_FORMAT_A8R8G8B8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , yzwx, srgb
-PIPE_FORMAT_X8R8G8B8_SRGB         , plain, 1, 1, un8 , un8 , un8 , un8 , yzw1, srgb
+PIPE_FORMAT_X8R8G8B8_SRGB         , plain, 1, 1, x8  , un8 , un8 , un8 , yzw1, srgb
 
 # Mixed-sign formats (typically used for bump map textures)
 PIPE_FORMAT_R8SG8SB8UX8U_NORM     , plain, 1, 1, sn8 , sn8 , un8 , x8  , xyz1, rgb
+PIPE_FORMAT_R10SG10SB10SA2U_NORM  , plain, 1, 1, sn10, sn10, sn10, un2 , xyzw, rgb
 PIPE_FORMAT_R5SG5SB6U_NORM        , plain, 1, 1, sn5 , sn5 , un6 ,     , xyz1, rgb
 
 # Depth-stencil formats
-PIPE_FORMAT_S8_UNORM              , plain, 1, 1, un8 ,     ,     ,     , _x__, zs 
-PIPE_FORMAT_Z16_UNORM             , plain, 1, 1, un16,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z32_UNORM             , plain, 1, 1, un32,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z32_FLOAT             , plain, 1, 1, f32 ,     ,     ,     , x___, zs 
-PIPE_FORMAT_Z24S8_UNORM           , plain, 1, 1, un24, un8 ,     ,     , xy__, zs 
-PIPE_FORMAT_S8Z24_UNORM           , plain, 1, 1, un8 , un24,     ,     , yx__, zs 
-PIPE_FORMAT_Z24X8_UNORM           , plain, 1, 1, un24, un8 ,     ,     , x___, zs 
-PIPE_FORMAT_X8Z24_UNORM           , plain, 1, 1, un8 , un24,     ,     , y___, zs 
+PIPE_FORMAT_S8_USCALED              , plain, 1, 1, u8  ,     ,     ,     , _x__, zs
+PIPE_FORMAT_Z16_UNORM               , plain, 1, 1, un16,     ,     ,     , x___, zs
+PIPE_FORMAT_Z32_UNORM               , plain, 1, 1, un32,     ,     ,     , x___, zs
+PIPE_FORMAT_Z32_FLOAT               , plain, 1, 1, f32 ,     ,     ,     , x___, zs
+PIPE_FORMAT_Z24_UNORM_S8_USCALED    , plain, 1, 1, un24, u8  ,     ,     , xy__, zs
+PIPE_FORMAT_S8_USCALED_Z24_UNORM    , plain, 1, 1, u8 ,  un24,     ,     , yx__, zs
+PIPE_FORMAT_Z24X8_UNORM             , plain, 1, 1, un24, x8  ,     ,     , x___, zs
+PIPE_FORMAT_X8Z24_UNORM             , plain, 1, 1, x8  , un24,     ,     , y___, zs
+PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED , plain, 1, 1, f32,  u8  , x24 ,     , xy__, zs
 
 # YUV formats
 # http://www.fourcc.org/yuv.php#UYVY
 PIPE_FORMAT_UYVY                 , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, yuv
 # http://www.fourcc.org/yuv.php#YUYV (a.k.a http://www.fourcc.org/yuv.php#YUY2)
-# XXX: u_tile.c's ycbcr_get_tile_rgba actually interprets it as VYUY but the 
-# intent should be to match D3DFMT_YUY2
 PIPE_FORMAT_YUYV                 , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, yuv
+# same subsampling but with rgb channels
+PIPE_FORMAT_R8G8_B8G8_UNORM      , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_G8R8_G8B8_UNORM      , subsampled, 2, 1, x32 ,     ,     ,     , xyz1, rgb
+
+# some special formats not fitting anywhere else
+PIPE_FORMAT_R10G10B10A2_USCALED   , plain,      1,  1, u10 , u10 , u10 , u2  , xyzw, rgb
+PIPE_FORMAT_R11G11B10_FLOAT       , plain,      1,  1, f11 , f11 , f10 ,     , xyz1, rgb
+PIPE_FORMAT_R9G9B9E5_FLOAT        , other,      1,  1, x32 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_R1_UNORM              , other,      8,  1, x8  ,     ,     ,     , x001, rgb
+# A.k.a. D3DFMT_CxV8U8
+PIPE_FORMAT_R8G8Bx_SNORM          , other,      1,  1, sn8 , sn8 ,     ,     , xyz1, rgb
 
 # Compressed formats
-PIPE_FORMAT_DXT1_RGB              , compressed, 4, 4, x64 ,     ,     ,     , xyz1, rgb
-PIPE_FORMAT_DXT1_RGBA             , compressed, 4, 4, x64 ,     ,     ,     , xyzw, rgb
-PIPE_FORMAT_DXT3_RGBA             , compressed, 4, 4, x128,     ,     ,     , xyzw, rgb
-PIPE_FORMAT_DXT5_RGBA             , compressed, 4, 4, x128,     ,     ,     , xyzw, rgb
-PIPE_FORMAT_DXT1_SRGB             , compressed, 4, 4, x64 ,     ,     ,     , xyz1, srgb
-PIPE_FORMAT_DXT1_SRGBA            , compressed, 4, 4, x64 ,     ,     ,     , xyzw, srgb
-PIPE_FORMAT_DXT3_SRGBA            , compressed, 4, 4, x128,     ,     ,     , xyzw, srgb
-PIPE_FORMAT_DXT5_SRGBA            , compressed, 4, 4, x128,     ,     ,     , xyzw, srgb
+# - http://en.wikipedia.org/wiki/S3_Texture_Compression
+# - http://www.opengl.org/registry/specs/EXT/texture_compression_s3tc.txt
+# - http://www.opengl.org/registry/specs/ARB/texture_compression_rgtc.txt
+# - http://msdn.microsoft.com/en-us/library/bb694531.aspx
+PIPE_FORMAT_DXT1_RGB              , s3tc, 4, 4, x64 ,     ,     ,     , xyz1, rgb
+PIPE_FORMAT_DXT1_RGBA             , s3tc, 4, 4, x64 ,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT3_RGBA             , s3tc, 4, 4, x128,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT5_RGBA             , s3tc, 4, 4, x128,     ,     ,     , xyzw, rgb
+PIPE_FORMAT_DXT1_SRGB             , s3tc, 4, 4, x64 ,     ,     ,     , xyz1, srgb
+PIPE_FORMAT_DXT1_SRGBA            , s3tc, 4, 4, x64 ,     ,     ,     , xyzw, srgb
+PIPE_FORMAT_DXT3_SRGBA            , s3tc, 4, 4, x128,     ,     ,     , xyzw, srgb
+PIPE_FORMAT_DXT5_SRGBA            , s3tc, 4, 4, x128,     ,     ,     , xyzw, srgb
+
+PIPE_FORMAT_RGTC1_UNORM           , rgtc, 4, 4, x64,      ,     ,     , x001, rgb
+PIPE_FORMAT_RGTC1_SNORM           , rgtc, 4, 4, x64,      ,     ,     , x001, rgb
+PIPE_FORMAT_RGTC2_UNORM           , rgtc, 4, 4, x128,     ,     ,     , xy01, rgb
+PIPE_FORMAT_RGTC2_SNORM           , rgtc, 4, 4, x128,     ,     ,     , xy01, rgb
 
 # Straightforward D3D10-like formats (also used for 
 # vertex buffer element description)
@@ -148,10 +179,10 @@ PIPE_FORMAT_R32_SSCALED           , plain, 1, 1, s32 ,     ,     ,     , x001, r
 PIPE_FORMAT_R32G32_SSCALED        , plain, 1, 1, s32 , s32 ,     ,     , xy01, rgb
 PIPE_FORMAT_R32G32B32_SSCALED     , plain, 1, 1, s32 , s32 , s32 ,     , xyz1, rgb
 PIPE_FORMAT_R32G32B32A32_SSCALED  , plain, 1, 1, s32 , s32 , s32 , s32 , xyzw, rgb
-PIPE_FORMAT_R32_FIXED             , plain, 1, 1, h32 ,     ,     ,     , x001, rgb
-PIPE_FORMAT_R32G32_FIXED          , plain, 1, 1, h32 , h32 ,     ,     , xy01, rgb
-PIPE_FORMAT_R32G32B32_FIXED       , plain, 1, 1, h32 , h32 , h32 ,     , xyz1, rgb
-PIPE_FORMAT_R32G32B32A32_FIXED    , plain, 1, 1, h32 , h32 , h32 , h32 , xyzw, rgb
+PIPE_FORMAT_R16_FLOAT             , plain, 1, 1, f16 ,     ,     ,     , x001, rgb
+PIPE_FORMAT_R16G16_FLOAT          , plain, 1, 1, f16 , f16 ,     ,     , xy01, rgb
+PIPE_FORMAT_R16G16B16_FLOAT       , plain, 1, 1, f16 , f16 , f16 ,     , xyz1, rgb
+PIPE_FORMAT_R16G16B16A16_FLOAT    , plain, 1, 1, f16 , f16 , f16 , f16 , xyzw, rgb
 PIPE_FORMAT_R16_UNORM             , plain, 1, 1, un16,     ,     ,     , x001, rgb
 PIPE_FORMAT_R16G16_UNORM          , plain, 1, 1, un16, un16,     ,     , xy01, rgb
 PIPE_FORMAT_R16G16B16_UNORM       , plain, 1, 1, un16, un16, un16,     , xyz1, rgb
@@ -184,6 +215,22 @@ PIPE_FORMAT_R8_SSCALED            , plain, 1, 1, s8  ,     ,     ,     , x001, r
 PIPE_FORMAT_R8G8_SSCALED          , plain, 1, 1, s8  , s8  ,     ,     , xy01, rgb
 PIPE_FORMAT_R8G8B8_SSCALED        , plain, 1, 1, s8  , s8  , s8  ,     , xyz1, rgb
 PIPE_FORMAT_R8G8B8A8_SSCALED      , plain, 1, 1, s8  , s8  , s8  , s8  , xyzw, rgb
+
+# GL-specific vertex buffer element formats
+# A.k.a. GL_FIXED
+PIPE_FORMAT_R32_FIXED             , plain, 1, 1, h32 ,     ,     ,     , x001, rgb
+PIPE_FORMAT_R32G32_FIXED          , plain, 1, 1, h32 , h32 ,     ,     , xy01, rgb
+PIPE_FORMAT_R32G32B32_FIXED       , plain, 1, 1, h32 , h32 , h32 ,     , xyz1, rgb
+PIPE_FORMAT_R32G32B32A32_FIXED    , plain, 1, 1, h32 , h32 , h32 , h32 , xyzw, rgb
+
+# D3D9-specific vertex buffer element formats
+# See also:
+# - http://msdn.microsoft.com/en-us/library/bb172533.aspx
+# A.k.a. D3DDECLTYPE_UDEC3
+PIPE_FORMAT_R10G10B10X2_USCALED   , plain, 1, 1, u10 , u10 , u10  , x2 , xyz1, rgb
+# A.k.a. D3DDECLTYPE_DEC3N
+PIPE_FORMAT_R10G10B10X2_SNORM     , plain, 1, 1, sn10, sn10, sn10 , x2 , xyz1, rgb
+
 PIPE_FORMAT_YV12                  , subsampled, 1, 1, x8  , x8  , x8  , x8  , xyzw, yuv
 PIPE_FORMAT_YV16                  , subsampled, 1, 1, x8  , x8  , x8  , x8  , xyzw, yuv
 PIPE_FORMAT_IYUV                  , subsampled, 1, 1, x8  , x8  , x8  , x8  , xyzw, yuv
diff --git a/src/gallium/auxiliary/util/u_format.h b/src/gallium/auxiliary/util/u_format.h
index e8fa0022b5b..605b13bd114 100644
--- a/src/gallium/auxiliary/util/u_format.h
+++ b/src/gallium/auxiliary/util/u_format.h
@@ -56,15 +56,23 @@ enum util_format_layout {
     *
     * This is for formats like YV12 where there is less than one sample per
     * pixel.
-    *
-    * XXX: This could actually b
     */
    UTIL_FORMAT_LAYOUT_SUBSAMPLED = 3,
 
    /**
-    * An unspecified compression algorithm.
+    * S3 Texture Compression formats.
+    */
+   UTIL_FORMAT_LAYOUT_S3TC = 4,
+
+   /**
+    * Red-Green Texture Compression formats.
+    */
+   UTIL_FORMAT_LAYOUT_RGTC = 5,
+
+   /**
+    * Everything else that doesn't fit in any of the above layouts.
     */
-   UTIL_FORMAT_LAYOUT_COMPRESSED = 4
+   UTIL_FORMAT_LAYOUT_OTHER = 6
 };
 
 
@@ -120,9 +128,15 @@ struct util_format_channel_description
 struct util_format_description
 {
    enum pipe_format format;
+
    const char *name;
 
    /**
+    * Short name, striped of the prefix, lower case.
+    */
+   const char *short_name;
+
+   /**
     * Pixel block dimensions.
     */
    struct util_format_block block;
@@ -140,6 +154,15 @@ struct util_format_description
    unsigned is_array:1;
 
    /**
+    * Whether the pixel format can be described as a bitfield structure.
+    *
+    * In particular:
+    * - pixel depth must be 8, 16, or 32 bits;
+    * - all channels must be unsigned, signed, or void
+    */
+   unsigned is_bitmask:1;
+
+   /**
     * Whether channels have mixed types (ignoring UTIL_FORMAT_TYPE_VOID).
     */
    unsigned is_mixed:1;
@@ -166,6 +189,127 @@ struct util_format_description
     * Colorspace transformation.
     */
    enum util_format_colorspace colorspace;
+
+   /**
+    * Unpack pixel blocks to R8G8B8A8_UNORM.
+    * Note: strides are in bytes.
+    *
+    * Only defined for non-depth-stencil formats.
+    */
+   void
+   (*unpack_rgba_8unorm)(uint8_t *dst, unsigned dst_stride,
+                         const uint8_t *src, unsigned src_stride,
+                         unsigned width, unsigned height);
+
+   /**
+    * Pack pixel blocks from R8G8B8A8_UNORM.
+    * Note: strides are in bytes.
+    *
+    * Only defined for non-depth-stencil formats.
+    */
+   void
+   (*pack_rgba_8unorm)(uint8_t *dst, unsigned dst_stride,
+                       const uint8_t *src, unsigned src_stride,
+                       unsigned width, unsigned height);
+
+   /**
+    * Unpack pixel blocks to R32G32B32A32_FLOAT.
+    * Note: strides are in bytes.
+    *
+    * Only defined for non-depth-stencil formats.
+    */
+   void
+   (*unpack_rgba_float)(float *dst, unsigned dst_stride,
+                        const uint8_t *src, unsigned src_stride,
+                        unsigned width, unsigned height);
+
+   /**
+    * Pack pixel blocks from R32G32B32A32_FLOAT.
+    * Note: strides are in bytes.
+    *
+    * Only defined for non-depth-stencil formats.
+    */
+   void
+   (*pack_rgba_float)(uint8_t *dst, unsigned dst_stride,
+                      const float *src, unsigned src_stride,
+                      unsigned width, unsigned height);
+
+   /**
+    * Fetch a single pixel (i, j) from a block.
+    *
+    * Only defined for non-depth-stencil formats.
+    */
+   void
+   (*fetch_rgba_float)(float *dst,
+                       const uint8_t *src,
+                       unsigned i, unsigned j);
+
+   /**
+    * Unpack pixels to Z32_UNORM.
+    * Note: strides are in bytes.
+    *
+    * Only defined for depth formats.
+    */
+   void
+   (*unpack_z_32unorm)(uint32_t *dst, unsigned dst_stride,
+                       const uint8_t *src, unsigned src_stride,
+                       unsigned width, unsigned height);
+
+   /**
+    * Pack pixels from Z32_FLOAT.
+    * Note: strides are in bytes.
+    *
+    * Only defined for depth formats.
+    */
+   void
+   (*pack_z_32unorm)(uint8_t *dst, unsigned dst_stride,
+                     const uint32_t *src, unsigned src_stride,
+                     unsigned width, unsigned height);
+
+   /**
+    * Unpack pixels to Z32_FLOAT.
+    * Note: strides are in bytes.
+    *
+    * Only defined for depth formats.
+    */
+   void
+   (*unpack_z_float)(float *dst, unsigned dst_stride,
+                     const uint8_t *src, unsigned src_stride,
+                     unsigned width, unsigned height);
+
+   /**
+    * Pack pixels from Z32_FLOAT.
+    * Note: strides are in bytes.
+    *
+    * Only defined for depth formats.
+    */
+   void
+   (*pack_z_float)(uint8_t *dst, unsigned dst_stride,
+                   const float *src, unsigned src_stride,
+                   unsigned width, unsigned height);
+
+   /**
+    * Unpack pixels to S8_USCALED.
+    * Note: strides are in bytes.
+    *
+    * Only defined for stencil formats.
+    */
+   void
+   (*unpack_s_8uscaled)(uint8_t *dst, unsigned dst_stride,
+                        const uint8_t *src, unsigned src_stride,
+                        unsigned width, unsigned height);
+
+   /**
+    * Pack pixels from S8_USCALED.
+    * Note: strides are in bytes.
+    *
+    * Only defined for stencil formats.
+    */
+   void
+   (*pack_s_8uscaled)(uint8_t *dst, unsigned dst_stride,
+                      const uint8_t *src, unsigned src_stride,
+                      unsigned width, unsigned height);
+
 };
 
 
@@ -186,8 +330,8 @@ util_format_name(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return "???";
    }
 
@@ -195,16 +339,16 @@ util_format_name(enum pipe_format format)
 }
 
 static INLINE boolean 
-util_format_is_compressed(enum pipe_format format)
+util_format_is_s3tc(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
-   return desc->layout == UTIL_FORMAT_LAYOUT_COMPRESSED ? TRUE : FALSE;
+   return desc->layout == UTIL_FORMAT_LAYOUT_S3TC ? TRUE : FALSE;
 }
 
 static INLINE boolean 
@@ -212,8 +356,8 @@ util_format_is_depth_or_stencil(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
@@ -225,8 +369,8 @@ util_format_is_depth_and_stencil(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return FALSE;
    }
 
@@ -238,6 +382,34 @@ util_format_is_depth_and_stencil(enum pipe_format format)
            desc->swizzle[1] != UTIL_FORMAT_SWIZZLE_NONE) ? TRUE : FALSE;
 }
 
+/**
+ * Whether this format is a rgab8 variant.
+ *
+ * That is, any format that matches the
+ *
+ *   PIPE_FORMAT_?8?8?8?8_UNORM
+ */
+static INLINE boolean
+util_format_is_rgba8_variant(const struct util_format_description *desc)
+{
+   unsigned chan;
+
+   if(desc->block.width != 1 ||
+      desc->block.height != 1 ||
+      desc->block.bits != 32)
+      return FALSE;
+
+   for(chan = 0; chan < 4; ++chan) {
+      if(desc->channel[chan].type != UTIL_FORMAT_TYPE_UNSIGNED &&
+         desc->channel[chan].type != UTIL_FORMAT_TYPE_VOID)
+         return FALSE;
+      if(desc->channel[chan].size != 8)
+         return FALSE;
+   }
+
+   return TRUE;
+}
+
 
 /**
  * Return total bits needed for the pixel format per block.
@@ -247,8 +419,8 @@ util_format_get_blocksizebits(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 0;
    }
 
@@ -273,8 +445,8 @@ util_format_get_blockwidth(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 1;
    }
 
@@ -286,8 +458,8 @@ util_format_get_blockheight(enum pipe_format format)
 {
    const struct util_format_description *desc = util_format_description(format);
 
-   assert(format);
-   if (!format) {
+   assert(desc);
+   if (!desc) {
       return 1;
    }
 
@@ -400,6 +572,16 @@ util_format_has_alpha(enum pipe_format format)
    }
 }
 
+/**
+ * Return the number of components stored.
+ * Formats with block size != 1x1 will always have 1 component (the block).
+ */
+static INLINE unsigned
+util_format_get_nr_components(enum pipe_format format)
+{
+   const struct util_format_description *desc = util_format_description(format);
+   return desc->nr_channels;
+}
 
 /*
  * Format access functions.
@@ -429,6 +611,19 @@ util_format_write_4ub(enum pipe_format format,
                       void *dst, unsigned dst_stride, 
                       unsigned x, unsigned y, unsigned w, unsigned h);
 
+/*
+ * Generic format conversion;
+ */
+
+void
+util_format_translate(enum pipe_format dst_format,
+                      void *dst, unsigned dst_stride,
+                      unsigned dst_x, unsigned dst_y,
+                      enum pipe_format src_format,
+                      const void *src, unsigned src_stride,
+                      unsigned src_x, unsigned src_y,
+                      unsigned width, unsigned height);
+
 #ifdef __cplusplus
 } // extern "C" {
 #endif
diff --git a/src/gallium/auxiliary/util/u_format_access.py b/src/gallium/auxiliary/util/u_format_access.py
deleted file mode 100644
index 00424779d28..00000000000
--- a/src/gallium/auxiliary/util/u_format_access.py
+++ /dev/null
@@ -1,341 +0,0 @@
-#!/usr/bin/env python
-
-'''
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL VMWARE 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.
- *
- **************************************************************************/
-
-/**
- * @file
- * Pixel format accessor functions.
- *
- * @author Jose Fonseca <[email protected]>
- */
-'''
-
-
-import math
-import sys
-
-from u_format_pack import *
-
-
-def is_format_supported(format):
-    '''Determines whether we actually have the plumbing necessary to generate the 
-    to read/write to/from this format.'''
-
-    # FIXME: Ideally we would support any format combination here.
-
-    # XXX: It should be straightforward to support srgb
-    if format.colorspace not in ('rgb', 'zs'):
-        return False
-
-    if format.layout != PLAIN:
-        return False
-
-    for i in range(4):
-        channel = format.channels[i]
-        if channel.type not in (VOID, UNSIGNED, FLOAT):
-            return False
-
-    # We can only read a color from a depth/stencil format if the depth channel is present
-    if format.colorspace == 'zs' and format.swizzles[0] == SWIZZLE_NONE:
-        return False
-
-    return True
-
-
-def native_type(format):
-    '''Get the native appropriate for a format.'''
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            # For arithmetic pixel formats return the integer type that matches the whole pixel
-            return 'uint%u_t' % format.block_size()
-        else:
-            # For array pixel formats return the integer type that matches the color channel
-            channel = format.channels[0]
-            if channel.type == UNSIGNED:
-                return 'uint%u_t' % channel.size
-            elif channel.type == SIGNED:
-                return 'int%u_t' % channel.size
-            elif channel.type == FLOAT:
-                if channel.size == 32:
-                    return 'float'
-                elif channel.size == 64:
-                    return 'double'
-                else:
-                    assert False
-            else:
-                assert False
-    else:
-        assert False
-
-
-def generate_srgb_tables():
-    print 'static ubyte srgb_to_linear[256] = {'
-    for i in range(256):
-        print '   %s,' % (int(math.pow((i / 255.0 + 0.055) / 1.055, 2.4) * 255))
-    print '};'
-    print
-    print 'static ubyte linear_to_srgb[256] = {'
-    print '   0,'
-    for i in range(1, 256):
-        print '   %s,' % (int((1.055 * math.pow(i / 255.0, 0.41666) - 0.055) * 255))
-    print '};'
-    print
-
-
-def generate_format_read(format, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the function to read pixels from a particular format'''
-
-    name = format.short_name()
-
-    src_native_type = native_type(format)
-
-    print 'static void'
-    print 'util_format_%s_read_%s(%s *dst, unsigned dst_stride, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, dst_suffix, dst_native_type)
-    print '{'
-    print '   unsigned x, y;'
-    print '   const uint8_t *src_row = src + y0*src_stride;'
-    print '   %s *dst_row = dst;' % dst_native_type
-    print '   for (y = 0; y < h; ++y) {'
-    print '      const %s *src_pixel = (const %s *)(src_row + x0*%u);' % (src_native_type, src_native_type, format.stride())
-    print '      %s *dst_pixel = dst_row;' %dst_native_type
-    print '      for (x = 0; x < w; ++x) {'
-
-    names = ['']*4
-    if format.colorspace == 'rgb':
-        for i in range(4):
-            swizzle = format.swizzles[i]
-            if swizzle < 4:
-                names[swizzle] += 'rgba'[i]
-    elif format.colorspace == 'zs':
-        swizzle = format.swizzles[0]
-        if swizzle < 4:
-            names[swizzle] = 'z'
-        else:
-            assert False
-    else:
-        assert False
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            print '         %s pixel = *src_pixel++;' % src_native_type
-            shift = 0;
-            for i in range(4):
-                src_channel = format.channels[i]
-                width = src_channel.size
-                if names[i]:
-                    value = 'pixel'
-                    mask = (1 << width) - 1
-                    if shift:
-                        value = '(%s >> %u)' % (value, shift)
-                    if shift + width < format.block_size():
-                        value = '(%s & 0x%x)' % (value, mask)
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         %s %s = %s;' % (dst_native_type, names[i], value)
-                shift += width
-        else:
-            for i in range(4):
-                src_channel = format.channels[i]
-                if names[i]:
-                    value = 'src_pixel[%u]' % i
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         %s %s = %s;' % (dst_native_type, names[i], value)
-            print '         src_pixel += %u;' % (format.nr_channels())
-    else:
-        assert False
-
-    for i in range(4):
-        if format.colorspace == 'rgb':
-            swizzle = format.swizzles[i]
-            if swizzle < 4:
-                value = names[swizzle]
-            elif swizzle == SWIZZLE_0:
-                value = '0'
-            elif swizzle == SWIZZLE_1:
-                value = get_one(dst_channel)
-            else:
-                assert False
-        elif format.colorspace == 'zs':
-            if i < 3:
-                value = 'z'
-            else:
-                value = get_one(dst_channel)
-        else:
-            assert False
-        print '         *dst_pixel++ = %s; /* %s */' % (value, 'rgba'[i])
-
-    print '      }'
-    print '      src_row += src_stride;'
-    print '      dst_row += dst_stride/sizeof(*dst_row);'
-    print '   }'
-    print '}'
-    print
-    
-
-def generate_format_write(format, src_channel, src_native_type, src_suffix):
-    '''Generate the function to write pixels to a particular format'''
-
-    name = format.short_name()
-
-    dst_native_type = native_type(format)
-
-    print 'static void'
-    print 'util_format_%s_write_%s(const %s *src, unsigned src_stride, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h)' % (name, src_suffix, src_native_type)
-    print '{'
-    print '   unsigned x, y;'
-    print '   uint8_t *dst_row = dst + y0*dst_stride;'
-    print '   const %s *src_row = src;' % src_native_type
-    print '   for (y = 0; y < h; ++y) {'
-    print '      %s *dst_pixel = (%s *)(dst_row + x0*%u);' % (dst_native_type, dst_native_type, format.stride())
-    print '      const %s *src_pixel = src_row;' %src_native_type
-    print '      for (x = 0; x < w; ++x) {'
-
-    inv_swizzle = format.inv_swizzles()
-
-    if format.layout == PLAIN:
-        if not format.is_array():
-            print '         %s pixel = 0;' % dst_native_type
-            shift = 0;
-            for i in range(4):
-                dst_channel = format.channels[i]
-                width = dst_channel.size
-                if inv_swizzle[i] is not None:
-                    value = 'src_pixel[%u]' % inv_swizzle[i]
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    if shift:
-                        value = '(%s << %u)' % (value, shift)
-                    print '         pixel |= %s;' % value
-                shift += width
-            print '         *dst_pixel++ = pixel;'
-        else:
-            for i in range(4):
-                dst_channel = format.channels[i]
-                if inv_swizzle[i] is not None:
-                    value = 'src_pixel[%u]' % inv_swizzle[i]
-                    value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-                    print '         *dst_pixel++ = %s;' % value
-    else:
-        assert False
-    print '         src_pixel += 4;'
-
-    print '      }'
-    print '      dst_row += dst_stride;'
-    print '      src_row += src_stride/sizeof(*src_row);'
-    print '   }'
-    print '}'
-    print
-    
-
-def generate_read(formats, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the dispatch function to read pixels from any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_read(format, dst_channel, dst_native_type, dst_suffix)
-
-    print 'void'
-    print 'util_format_read_%s(enum pipe_format format, %s *dst, unsigned dst_stride, const void *src, unsigned src_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (dst_suffix, dst_native_type)
-    print '{'
-    print '   void (*func)(%s *dst, unsigned dst_stride, const uint8_t *src, unsigned src_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % dst_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_read_%s;' % (format.short_name(), dst_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(dst, dst_stride, (const uint8_t *)src, src_stride, x, y, w, h);'
-    print '}'
-    print
-
-
-def generate_write(formats, src_channel, src_native_type, src_suffix):
-    '''Generate the dispatch function to write pixels to any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_write(format, src_channel, src_native_type, src_suffix)
-
-    print 'void'
-    print 'util_format_write_%s(enum pipe_format format, const %s *src, unsigned src_stride, void *dst, unsigned dst_stride, unsigned x, unsigned y, unsigned w, unsigned h)' % (src_suffix, src_native_type)
-    
-    print '{'
-    print '   void (*func)(const %s *src, unsigned src_stride, uint8_t *dst, unsigned dst_stride, unsigned x0, unsigned y0, unsigned w, unsigned h);' % src_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_write_%s;' % (format.short_name(), src_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(src, src_stride, (uint8_t *)dst, dst_stride, x, y, w, h);'
-    print '}'
-    print
-
-
-def main():
-    formats = []
-    for arg in sys.argv[1:]:
-        formats.extend(parse(arg))
-
-    print '/* This file is autogenerated by u_format_access.py from u_format.csv. Do not edit directly. */'
-    print
-    # This will print the copyright message on the top of this file
-    print __doc__.strip()
-    print
-    print '#include "pipe/p_compiler.h"'
-    print '#include "u_math.h"'
-    print '#include "u_format_pack.h"'
-    print
-
-    generate_srgb_tables()
-
-    type = Channel(FLOAT, False, 32)
-    native_type = 'float'
-    suffix = '4f'
-
-    generate_read(formats, type, native_type, suffix)
-    generate_write(formats, type, native_type, suffix)
-
-    type = Channel(UNSIGNED, True, 8)
-    native_type = 'uint8_t'
-    suffix = '4ub'
-
-    generate_read(formats, type, native_type, suffix)
-    generate_write(formats, type, native_type, suffix)
-
-
-if __name__ == '__main__':
-    main()
diff --git a/src/gallium/auxiliary/util/u_format_other.c b/src/gallium/auxiliary/util/u_format_other.c
new file mode 100644
index 00000000000..723fa8c3bf9
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_other.c
@@ -0,0 +1,267 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#include "u_math.h"
+#include "u_format_other.h"
+
+
+void
+util_format_r9g9b9e5_float_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+
+}
+
+void
+util_format_r9g9b9e5_float_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src_row, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+
+}
+
+void
+util_format_r9g9b9e5_float_fetch_rgba_float(float *dst, const uint8_t *src,
+                                       unsigned i, unsigned j)
+{
+
+}
+
+
+void
+util_format_r9g9b9e5_float_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+
+}
+
+
+void
+util_format_r9g9b9e5_float_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+
+}
+
+
+void
+util_format_r1_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                  const uint8_t *src_row, unsigned src_stride,
+                                  unsigned width, unsigned height)
+{
+
+}
+
+
+void
+util_format_r1_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                const float *src_row, unsigned src_stride,
+                                unsigned width, unsigned height)
+{
+
+}
+
+
+void
+util_format_r1_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                 unsigned i, unsigned j)
+{
+
+}
+
+
+void
+util_format_r1_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                   const uint8_t *src_row, unsigned src_stride,
+                                   unsigned width, unsigned height)
+{
+
+}
+
+
+void
+util_format_r1_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                 const uint8_t *src_row, unsigned src_stride,
+                                 unsigned width, unsigned height)
+{
+}
+
+
+/*
+ * PIPE_FORMAT_R8G8Bx_SNORM
+ *
+ * A.k.a. D3DFMT_CxV8U8
+ */
+
+
+void
+util_format_r8g8bx_snorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                      const uint8_t *src_row, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for(y = 0; y < height; y += 1) {
+      float *dst = dst_row;
+      const uint16_t *src = (const uint16_t *)src_row;
+      for(x = 0; x < width; x += 1) {
+         uint16_t value = *src++;
+         int16_t r, g;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r = ((int16_t)(value << 8)) >> 8;
+         g = ((int16_t)(value << 0)) >> 8;
+
+         dst[0] = (float)(r * (1.0f/0x7f)); /* r */
+         dst[1] = (float)(g * (1.0f/0x7f)); /* g */
+         dst[2] = sqrtf(1.0f - dst[0] * dst[0] - dst[1] * dst[1]); /* b */
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+      }
+      src_row += src_stride;
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_r8g8bx_snorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; y += 1) {
+      uint8_t *dst = dst_row;
+      const uint16_t *src = (const uint16_t *)src_row;
+      for(x = 0; x < width; x += 1) {
+         uint16_t value = *src++;
+         int16_t r, g;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r = ((int16_t)(value << 8)) >> 8;
+         g = ((int16_t)(value << 0)) >> 8;
+
+         dst[0] = (uint8_t)(((uint16_t)MAX2(r, 0)) * 0xff / 0x7f); /* r */
+         dst[1] = (uint8_t)(((uint16_t)MAX2(g, 0)) * 0xff / 0x7f); /* g */
+         dst[2] = (uint8_t)sqrtf(0x7f*0x7f - r * r - g * g) * 0xff / 0x7f; /* b */
+         dst[3] = 255; /* a */
+         dst += 4;
+      }
+      src_row += src_stride;
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_r8g8bx_snorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                    const float *src_row, unsigned src_stride,
+                                    unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; y += 1) {
+      const float *src = src_row;
+      uint16_t *dst = (uint16_t *)dst_row;
+      for(x = 0; x < width; x += 1) {
+         uint16_t value = 0;
+
+         value |= (uint16_t)(((int8_t)(CLAMP(src[0], -1, 1) * 0x7f)) & 0xff) ;
+         value |= (uint16_t)((((int8_t)(CLAMP(src[1], -1, 1) * 0x7f)) & 0xff) << 8) ;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 4;
+      }
+      dst_row += dst_stride;
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_r8g8bx_snorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                     const uint8_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for(y = 0; y < height; y += 1) {
+      const uint8_t *src = src_row;
+      uint16_t *dst = (uint16_t *)dst_row;
+      for(x = 0; x < width; x += 1) {
+         uint16_t value = 0;
+
+         value |= src[0] >> 1;
+         value |= (src[1] >> 1) << 8;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 4;
+      }
+      dst_row += dst_stride;
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_r8g8bx_snorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                     unsigned i, unsigned j)
+{
+   uint16_t value = *(const uint16_t *)src;
+   int16_t r, g;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+   value = util_bswap32(value);
+#endif
+
+   r = ((int16_t)(value << 8)) >> 8;
+   g = ((int16_t)(value << 0)) >> 8;
+
+   dst[0] = r * (1.0f/0x7f); /* r */
+   dst[1] = g * (1.0f/0x7f); /* g */
+   dst[2] = sqrtf(1.0f - dst[0] * dst[0] - dst[1] * dst[1]); /* b */
+   dst[3] = 1.0f; /* a */
+}
diff --git a/src/gallium/auxiliary/util/u_format_other.h b/src/gallium/auxiliary/util/u_format_other.h
new file mode 100644
index 00000000000..98c97be33fa
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_other.h
@@ -0,0 +1,108 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#ifndef U_FORMAT_OTHER_H_
+#define U_FORMAT_OTHER_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+void
+util_format_r9g9b9e5_float_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height);
+
+void
+util_format_r9g9b9e5_float_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src_row, unsigned src_stride,
+                                      unsigned width, unsigned height);
+
+void
+util_format_r9g9b9e5_float_fetch_rgba_float(float *dst, const uint8_t *src,
+                                       unsigned i, unsigned j);
+
+void
+util_format_r9g9b9e5_float_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height);
+
+void
+util_format_r9g9b9e5_float_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height);
+
+void
+util_format_r1_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                  const uint8_t *src_row, unsigned src_stride,
+                                  unsigned width, unsigned height);
+
+void
+util_format_r1_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                const float *src_row, unsigned src_stride,
+                                unsigned width, unsigned height);
+
+void
+util_format_r1_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                 unsigned i, unsigned j);
+
+void
+util_format_r1_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                   const uint8_t *src_row, unsigned src_stride,
+                                   unsigned width, unsigned height);
+
+void
+util_format_r1_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                 const uint8_t *src_row, unsigned src_stride,
+                                 unsigned width, unsigned height);
+
+void
+util_format_r8g8bx_snorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                      const uint8_t *src_row, unsigned src_stride,
+                                      unsigned width, unsigned height);
+
+void
+util_format_r8g8bx_snorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                    const float *src_row, unsigned src_stride,
+                                    unsigned width, unsigned height);
+
+void
+util_format_r8g8bx_snorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                     unsigned i, unsigned j);
+
+void
+util_format_r8g8bx_snorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height);
+
+void
+util_format_r8g8bx_snorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                     const uint8_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height);
+
+#endif /* U_FORMAT_OTHER_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_pack.py b/src/gallium/auxiliary/util/u_format_pack.py
index 3f33f7cc021..0c1bbc84c17 100644
--- a/src/gallium/auxiliary/util/u_format_pack.py
+++ b/src/gallium/auxiliary/util/u_format_pack.py
@@ -3,7 +3,7 @@
 '''
 /**************************************************************************
  *
- * Copyright 2009 VMware, Inc.
+ * Copyright 2009-2010 VMware, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -38,6 +38,7 @@
 
 
 import sys
+import math
 
 from u_format_parse import *
 
@@ -45,19 +46,35 @@ from u_format_parse import *
 def generate_format_type(format):
     '''Generate a structure that describes the format.'''
 
+    assert format.layout == PLAIN
+    
     print 'union util_format_%s {' % format.short_name()
-    if format.is_bitmask():
+    
+    if format.block_size() in (8, 16, 32, 64):
         print '   uint%u_t value;' % (format.block_size(),)
+
+    use_bitfields = False
+    for channel in format.channels:
+        if channel.size % 8 or not is_pot(channel.size):
+            use_bitfields = True
+
     print '   struct {'
     for channel in format.channels:
-        if format.is_bitmask() and not format.is_array():
+        if use_bitfields:
             if channel.type == VOID:
                 if channel.size:
                     print '      unsigned %s:%u;' % (channel.name, channel.size)
             elif channel.type == UNSIGNED:
                 print '      unsigned %s:%u;' % (channel.name, channel.size)
-            elif channel.type == SIGNED:
+            elif channel.type in (SIGNED, FIXED):
                 print '      int %s:%u;' % (channel.name, channel.size)
+            elif channel.type == FLOAT:
+                if channel.size == 64:
+                    print '      double %s;' % (channel.name)
+                elif channel.size == 32:
+                    print '      float %s;' % (channel.name)
+                else:
+                    print '      unsigned %s:%u;' % (channel.name, channel.size)
             else:
                 assert 0
         else:
@@ -88,7 +105,7 @@ def generate_format_type(format):
 def bswap_format(format):
     '''Generate a structure that describes the format.'''
 
-    if format.is_bitmask() and not format.is_array():
+    if format.is_bitmask() and not format.is_array() and format.block_size() > 8:
         print '#ifdef PIPE_ARCH_BIG_ENDIAN'
         print '   pixel.value = util_bswap%u(pixel.value);' % format.block_size()
         print '#endif'
@@ -105,12 +122,10 @@ def is_format_supported(format):
 
     for i in range(4):
         channel = format.channels[i]
-        if channel.type not in (VOID, UNSIGNED, SIGNED, FLOAT):
+        if channel.type not in (VOID, UNSIGNED, SIGNED, FLOAT, FIXED):
+            return False
+        if channel.type == FLOAT and channel.size not in (16, 32, 64):
             return False
-
-    # We can only read a color from a depth/stencil format if the depth channel is present
-    if format.colorspace == 'zs' and format.swizzles[0] == SWIZZLE_NONE:
-        return False
 
     return True
 
@@ -124,15 +139,17 @@ def native_type(format):
             return 'uint%u_t' % format.block_size()
         else:
             # For array pixel formats return the integer type that matches the color channel
-            type = format.channels[0]
-            if type.type == UNSIGNED:
-                return 'uint%u_t' % type.size
-            elif type.type == SIGNED:
-                return 'int%u_t' % type.size
-            elif type.type == FLOAT:
-                if type.size == 32:
+            channel = format.channels[0]
+            if channel.type in (UNSIGNED, VOID):
+                return 'uint%u_t' % channel.size
+            elif channel.type in (SIGNED, FIXED):
+                return 'int%u_t' % channel.size
+            elif channel.type == FLOAT:
+                if channel.size == 16:
+                    return 'uint16_t'
+                elif channel.size == 32:
                     return 'float'
-                elif type.size == 64:
+                elif channel.size == 64:
                     return 'double'
                 else:
                     assert False
@@ -171,37 +188,35 @@ def get_one_shift(type):
     assert False
 
 
-def get_one(type):
+def value_to_native(type, value):
     '''Get the value of unity for this type.'''
-    if type.type == 'FLOAT' or not type.norm:
-        return 1
+    if type.type == FLOAT:
+        return value
+    if type.type == FIXED:
+        return int(value * (1 << (type.size/2)))
+    if not type.norm:
+        return int(value)
+    if type.type == UNSIGNED:
+        return int(value * ((1 << type.size) - 1))
+    if type.type == SIGNED:
+        return int(value * ((1 << (type.size - 1)) - 1))
+    assert False
+
+
+def native_to_constant(type, value):
+    '''Get the value of unity for this type.'''
+    if type.type == FLOAT:
+        if type.size <= 32:
+            return "%ff" % value 
+        else:
+            return "%ff" % value 
     else:
-        return (1 << get_one_shift(type)) - 1
-
-
-def generate_clamp():
-    '''Code generate the clamping functions for each type.
-
-    We don't use a macro so that arguments with side effects, 
-    like *src_pixel++ are correctly handled.
-    '''
-
-    for suffix, native_type in [
-        ('', 'double'),
-        ('f', 'float'),
-        ('ui', 'unsigned int'),
-        ('si', 'int'),
-    ]:
-        print 'static INLINE %s' % native_type
-        print 'clamp%s(%s value, %s lbound, %s ubound)' % (suffix, native_type, native_type, native_type)
-        print '{'
-        print '   if(value < lbound)'
-        print '      return lbound;'
-        print '   if(value > ubound)'
-        print '      return ubound;'
-        print '   return value;'
-        print '}'
-        print
+        return str(int(value))
+
+
+def get_one(type):
+    '''Get the value of unity for this type.'''
+    return value_to_native(type, 1)
 
 
 def clamp_expr(src_channel, dst_channel, dst_native_type, value):
@@ -211,274 +226,448 @@ def clamp_expr(src_channel, dst_channel, dst_native_type, value):
     if src_channel == dst_channel:
         return value
 
-    # Pick the approriate clamp function
-    if src_channel.type == FLOAT:
-        if src_channel.size == 32:
-            func = 'clampf'
-        elif src_channel.size == 64:
-            func = 'clamp'
-        else:
-            assert False
-    elif src_channel.type == UNSIGNED:
-        func = 'clampui'
-    elif src_channel.type == SIGNED:
-        func = 'clampsi'
-    else:
-        assert False
-
     src_min = src_channel.min()
     src_max = src_channel.max()
     dst_min = dst_channel.min()
     dst_max = dst_channel.max()
+    
+    # Translate the destination range to the src native value
+    dst_min_native = value_to_native(src_channel, dst_min)
+    dst_max_native = value_to_native(src_channel, dst_max)
 
     if src_min < dst_min and src_max > dst_max:
-        return 'CLAMP(%s, %s, %s)' % (value, dst_min, dst_max)
+        return 'CLAMP(%s, %s, %s)' % (value, dst_min_native, dst_max_native)
 
     if src_max > dst_max:
-        return 'MIN2(%s, %s)' % (value, dst_max)
+        return 'MIN2(%s, %s)' % (value, dst_max_native)
         
     if src_min < dst_min:
-        return 'MAX2(%s, %s)' % (value, dst_min)
+        return 'MAX2(%s, %s)' % (value, dst_min_native)
 
     return value
 
 
-def conversion_expr(src_channel, dst_channel, dst_native_type, value, clamp=True):
+def conversion_expr(src_channel, 
+                    dst_channel, dst_native_type, 
+                    value, 
+                    clamp=True, 
+                    src_colorspace = RGB, 
+                    dst_colorspace = RGB):
     '''Generate the expression to convert a value between two types.'''
 
+    if src_colorspace != dst_colorspace:
+        if src_colorspace == SRGB:
+            assert src_channel.type == UNSIGNED
+            assert src_channel.norm
+            assert src_channel.size == 8
+            assert dst_colorspace == RGB
+            if dst_channel.type == FLOAT:
+                return 'util_format_srgb_8unorm_to_linear_float(%s)' % value
+            else:
+                assert dst_channel.type == UNSIGNED
+                assert dst_channel.norm
+                assert dst_channel.size == 8
+                return 'util_format_srgb_to_linear_8unorm(%s)' % value
+        elif dst_colorspace == SRGB:
+            assert dst_channel.type == UNSIGNED
+            assert dst_channel.norm
+            assert dst_channel.size == 8
+            assert src_colorspace == RGB
+            if src_channel.type == FLOAT:
+                return 'util_format_linear_float_to_srgb_8unorm(%s)' % value
+            else:
+                assert src_channel.type == UNSIGNED
+                assert src_channel.norm
+                assert src_channel.size == 8
+                return 'util_format_linear_to_srgb_8unorm(%s)' % value
+        elif src_colorspace == ZS:
+            pass
+        elif dst_colorspace == ZS:
+            pass
+        else:
+            assert 0
+
     if src_channel == dst_channel:
         return value
 
-    if src_channel.type == FLOAT and dst_channel.type == FLOAT:
-        return '(%s)%s' % (dst_native_type, value)
-    
-    if not src_channel.norm and not dst_channel.norm:
-        return '(%s)%s' % (dst_native_type, value)
+    src_type = src_channel.type
+    src_size = src_channel.size
+    src_norm = src_channel.norm
 
-    if clamp:
-        value = clamp_expr(src_channel, dst_channel, dst_native_type, value)
+    # Promote half to float
+    if src_type == FLOAT and src_size == 16:
+        value = 'util_half_to_float(%s)' % value
+        src_size = 32
 
-    if dst_channel.type == FLOAT:
-        if src_channel.norm:
-            one = get_one(src_channel)
-            if src_channel.size <= 23:
-                scale = '(1.0f/0x%x)' % one
-            else:
-                # bigger than single precision mantissa, use double
-                scale = '(1.0/0x%x)' % one
-            value = '(%s * %s)' % (value, scale)
-        return '(%s)%s' % (dst_native_type, value)
+    # Special case for float <-> ubytes for more accurate results
+    # Done before clamping since these functions already take care of that
+    if src_type == UNSIGNED and src_norm and src_size == 8 and dst_channel.type == FLOAT and dst_channel.size == 32:
+        return 'ubyte_to_float(%s)' % value
+    if src_type == FLOAT and src_size == 32 and dst_channel.type == UNSIGNED and dst_channel.norm and dst_channel.size == 8:
+        return 'float_to_ubyte(%s)' % value
 
-    if src_channel.type == FLOAT:
-        if dst_channel.norm:
-            dst_one = get_one(dst_channel)
-            if dst_channel.size <= 23:
-                scale = '0x%x' % dst_one
-            else:
-                # bigger than single precision mantissa, use double
-                scale = '(double)0x%x' % dst_one
-            value = '(%s * %s)' % (value, scale)
-        return '(%s)%s' % (dst_native_type, value)
+    if clamp:
+        if dst_channel.type != FLOAT or src_type != FLOAT:
+            value = clamp_expr(src_channel, dst_channel, dst_native_type, value)
 
-    if not src_channel.norm and not dst_channel.norm:
-        # neither is normalized -- just cast
-        return '(%s)%s' % (dst_native_type, value)
+    if src_type in (SIGNED, UNSIGNED) and dst_channel.type in (SIGNED, UNSIGNED):
+        if not src_norm and not dst_channel.norm:
+            # neither is normalized -- just cast
+            return '(%s)%s' % (dst_native_type, value)
 
-    if src_channel.type in (SIGNED, UNSIGNED) and dst_channel.type in (SIGNED, UNSIGNED):
         src_one = get_one(src_channel)
         dst_one = get_one(dst_channel)
 
-        if src_one > dst_one and src_channel.norm:
+        if src_one > dst_one and src_norm and dst_channel.norm:
             # We can just bitshift
             src_shift = get_one_shift(src_channel)
             dst_shift = get_one_shift(dst_channel)
             value = '(%s >> %s)' % (value, src_shift - dst_shift)
         else:
             # We need to rescale using an intermediate type big enough to hold the multiplication of both
-            tmp_native_type = intermediate_native_type(src_channel.size + dst_channel.size, src_channel.sign and dst_channel.sign)
-            value = '(%s)%s' % (tmp_native_type, value)
+            tmp_native_type = intermediate_native_type(src_size + dst_channel.size, src_channel.sign and dst_channel.sign)
+            value = '((%s)%s)' % (tmp_native_type, value)
             value = '(%s * 0x%x / 0x%x)' % (value, dst_one, src_one)
         value = '(%s)%s' % (dst_native_type, value)
         return value
 
-    assert False
+    # Promote to either float or double
+    if src_type != FLOAT:
+        if src_norm or src_type == FIXED:
+            one = get_one(src_channel)
+            if src_size <= 23:
+                value = '(%s * (1.0f/0x%x))' % (value, one)
+                if dst_channel.size <= 32:
+                    value = '(float)%s' % value
+                src_size = 32
+            else:
+                # bigger than single precision mantissa, use double
+                value = '(%s * (1.0/0x%x))' % (value, one)
+                src_size = 64
+            src_norm = False
+        else:
+            if src_size <= 23 or dst_channel.size <= 32:
+                value = '(float)%s' % value
+                src_size = 32
+            else:
+                # bigger than single precision mantissa, use double
+                value = '(double)%s' % value
+                src_size = 64
+        src_type = FLOAT
 
+    # Convert double or float to non-float
+    if dst_channel.type != FLOAT:
+        if dst_channel.norm or dst_channel.type == FIXED:
+            dst_one = get_one(dst_channel)
+            if dst_channel.size <= 23:
+                value = '(%s * 0x%x)' % (value, dst_one)
+            else:
+                # bigger than single precision mantissa, use double
+                value = '(%s * (double)0x%x)' % (value, dst_one)
+        value = '(%s)%s' % (dst_native_type, value)
+    else:
+        # Cast double to float when converting to either half or float
+        if dst_channel.size <= 32 and src_size > 32:
+            value = '(float)%s' % value
+            src_size = 32
 
-def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the function to unpack pixels from a particular format'''
+        if dst_channel.size == 16:
+            value = 'util_float_to_half(%s)' % value
+        elif dst_channel.size == 64 and src_size < 64:
+            value = '(double)%s' % value
 
-    name = format.short_name()
+    return value
 
-    src_native_type = native_type(format)
 
-    print 'static INLINE void'
-    print 'util_format_%s_unpack_%s(%s *dst, const void *src)' % (name, dst_suffix, dst_native_type)
-    print '{'
-    print '   union util_format_%s pixel;' % format.short_name()
-    print '   memcpy(&pixel, src, sizeof pixel);'
-    bswap_format(format)
+def generate_unpack_kernel(format, dst_channel, dst_native_type):
 
+    if not is_format_supported(format):
+        return
+    
     assert format.layout == PLAIN
 
-    for i in range(4):
-        swizzle = format.swizzles[i]
-        if swizzle < 4:
-            src_channel = format.channels[swizzle]
-            value = 'pixel.chan.%s' % src_channel.name 
-            value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-        elif swizzle == SWIZZLE_0:
-            value = '0'
-        elif swizzle == SWIZZLE_1:
-            value = get_one(dst_channel)
-        elif swizzle == SWIZZLE_NONE:
-            value = '0'
-        else:
-            assert False
-        if format.colorspace == ZS:
-            if i == 3:
-                value = get_one(dst_channel)
-            elif i >= 1:
-                value = 'dst[0]'
-        print '   dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
+    src_native_type = native_type(format)
 
-    print '}'
-    print
+    if format.is_bitmask():
+        depth = format.block_size()
+        print '         uint%u_t value = *(const uint%u_t *)src;' % (depth, depth) 
+
+        # Declare the intermediate variables
+        for i in range(format.nr_channels()):
+            src_channel = format.channels[i]
+            if src_channel.type == UNSIGNED:
+                print '         uint%u_t %s;' % (depth, src_channel.name)
+            elif src_channel.type == SIGNED:
+                print '         int%u_t %s;' % (depth, src_channel.name)
+
+        if depth > 8:
+            print '#ifdef PIPE_ARCH_BIG_ENDIAN'
+            print '         value = util_bswap%u(value);' % depth
+            print '#endif'
+
+        # Compute the intermediate unshifted values 
+        shift = 0
+        for i in range(format.nr_channels()):
+            src_channel = format.channels[i]
+            value = 'value'
+            if src_channel.type == UNSIGNED:
+                if shift:
+                    value = '%s >> %u' % (value, shift)
+                if shift + src_channel.size < depth:
+                    value = '(%s) & 0x%x' % (value, (1 << src_channel.size) - 1)
+            elif src_channel.type == SIGNED:
+                if shift + src_channel.size < depth:
+                    # Align the sign bit
+                    lshift = depth - (shift + src_channel.size)
+                    value = '%s << %u' % (value, lshift)
+                # Cast to signed
+                value = '(int%u_t)(%s) ' % (depth, value)
+                if src_channel.size < depth:
+                    # Align the LSB bit
+                    rshift = depth - src_channel.size
+                    value = '(%s) >> %u' % (value, rshift)
+            else:
+                value = None
+                
+            if value is not None:
+                print '         %s = %s;' % (src_channel.name, value)
+                
+            shift += src_channel.size
+
+        # Convert, swizzle, and store final values
+        for i in range(4):
+            swizzle = format.swizzles[i]
+            if swizzle < 4:
+                src_channel = format.channels[swizzle]
+                src_colorspace = format.colorspace
+                if src_colorspace == SRGB and i == 3:
+                    # Alpha channel is linear
+                    src_colorspace = RGB
+                value = src_channel.name 
+                value = conversion_expr(src_channel, 
+                                        dst_channel, dst_native_type, 
+                                        value,
+                                        src_colorspace = src_colorspace)
+            elif swizzle == SWIZZLE_0:
+                value = '0'
+            elif swizzle == SWIZZLE_1:
+                value = get_one(dst_channel)
+            elif swizzle == SWIZZLE_NONE:
+                value = '0'
+            else:
+                assert False
+            print '         dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
+        
+    else:
+        print '         union util_format_%s pixel;' % format.short_name()
+        print '         memcpy(&pixel, src, sizeof pixel);'
+        bswap_format(format)
+    
+        for i in range(4):
+            swizzle = format.swizzles[i]
+            if swizzle < 4:
+                src_channel = format.channels[swizzle]
+                src_colorspace = format.colorspace
+                if src_colorspace == SRGB and i == 3:
+                    # Alpha channel is linear
+                    src_colorspace = RGB
+                value = 'pixel.chan.%s' % src_channel.name 
+                value = conversion_expr(src_channel, 
+                                        dst_channel, dst_native_type, 
+                                        value,
+                                        src_colorspace = src_colorspace)
+            elif swizzle == SWIZZLE_0:
+                value = '0'
+            elif swizzle == SWIZZLE_1:
+                value = get_one(dst_channel)
+            elif swizzle == SWIZZLE_NONE:
+                value = '0'
+            else:
+                assert False
+            print '         dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
     
 
-def generate_format_pack(format, src_channel, src_native_type, src_suffix):
-    '''Generate the function to pack pixels to a particular format'''
-
-    name = format.short_name()
+def generate_pack_kernel(format, src_channel, src_native_type):
 
+    if not is_format_supported(format):
+        return
+    
     dst_native_type = native_type(format)
 
-    print 'static INLINE void'
-    print 'util_format_%s_pack_%s(void *dst, %s r, %s g, %s b, %s a)' % (name, src_suffix, src_native_type, src_native_type, src_native_type, src_native_type)
-    print '{'
-    print '   union util_format_%s pixel;' % format.short_name()
-
     assert format.layout == PLAIN
 
     inv_swizzle = format.inv_swizzles()
 
-    for i in range(4):
-        dst_channel = format.channels[i]
-        width = dst_channel.size
-        if inv_swizzle[i] is None:
-            continue
-        value = 'rgba'[inv_swizzle[i]]
-        value = conversion_expr(src_channel, dst_channel, dst_native_type, value)
-        if format.colorspace == ZS:
-            if i == 3:
-                value = get_one(dst_channel)
-            elif i >= 1:
-                value = '0'
-        print '   pixel.chan.%s = %s;' % (dst_channel.name, value)
+    if format.is_bitmask():
+        depth = format.block_size()
+        print '         uint%u_t value = 0;' % depth 
+
+        shift = 0
+        for i in range(4):
+            dst_channel = format.channels[i]
+            if inv_swizzle[i] is not None:
+                value ='src[%u]' % inv_swizzle[i]
+                dst_colorspace = format.colorspace
+                if dst_colorspace == SRGB and inv_swizzle[i] == 3:
+                    # Alpha channel is linear
+                    dst_colorspace = RGB
+                value = conversion_expr(src_channel, 
+                                        dst_channel, dst_native_type, 
+                                        value,
+                                        dst_colorspace = dst_colorspace)
+                if dst_channel.type in (UNSIGNED, SIGNED):
+                    if shift + dst_channel.size < depth:
+                        value = '(%s) & 0x%x' % (value, (1 << dst_channel.size) - 1)
+                    if shift:
+                        value = '(%s) << %u' % (value, shift)
+                    if dst_channel.type == SIGNED:
+                        # Cast to unsigned
+                        value = '(uint%u_t)(%s) ' % (depth, value)
+                else:
+                    value = None
+                if value is not None:
+                    print '         value |= %s;' % (value)
+                
+            shift += dst_channel.size
+
+        if depth > 8:
+            print '#ifdef PIPE_ARCH_BIG_ENDIAN'
+            print '         value = util_bswap%u(value);' % depth
+            print '#endif'
+        
+        print '         *(uint%u_t *)dst = value;' % depth 
+
+    else:
+        print '         union util_format_%s pixel;' % format.short_name()
+    
+        for i in range(4):
+            dst_channel = format.channels[i]
+            width = dst_channel.size
+            if inv_swizzle[i] is None:
+                continue
+            dst_colorspace = format.colorspace
+            if dst_colorspace == SRGB and inv_swizzle[i] == 3:
+                # Alpha channel is linear
+                dst_colorspace = RGB
+            value ='src[%u]' % inv_swizzle[i]
+            value = conversion_expr(src_channel, 
+                                    dst_channel, dst_native_type, 
+                                    value, 
+                                    dst_colorspace = dst_colorspace)
+            print '         pixel.chan.%s = %s;' % (dst_channel.name, value)
+    
+        bswap_format(format)
+        print '         memcpy(dst, &pixel, sizeof pixel);'
+    
+
+def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
+    '''Generate the function to unpack pixels from a particular format'''
+
+    name = format.short_name()
+
+    print 'static INLINE void'
+    print 'util_format_%s_unpack_%s(%s *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, dst_suffix, dst_native_type)
+    print '{'
+
+    if is_format_supported(format):
+        print '   unsigned x, y;'
+        print '   for(y = 0; y < height; y += %u) {' % (format.block_height,)
+        print '      %s *dst = dst_row;' % (dst_native_type)
+        print '      const uint8_t *src = src_row;'
+        print '      for(x = 0; x < width; x += %u) {' % (format.block_width,)
+        
+        generate_unpack_kernel(format, dst_channel, dst_native_type)
+    
+        print '         src += %u;' % (format.block_size() / 8,)
+        print '         dst += 4;'
+        print '      }'
+        print '      src_row += src_stride;'
+        print '      dst_row += dst_stride/sizeof(*dst_row);'
+        print '   }'
 
-    bswap_format(format)
-    print '   memcpy(dst, &pixel, sizeof pixel);'
     print '}'
     print
     
 
-def generate_unpack(formats, dst_channel, dst_native_type, dst_suffix):
-    '''Generate the dispatch function to unpack pixels from any format'''
+def generate_format_pack(format, src_channel, src_native_type, src_suffix):
+    '''Generate the function to pack pixels to a particular format'''
 
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix)
+    name = format.short_name()
 
     print 'static INLINE void'
-    print 'util_format_unpack_%s(enum pipe_format format, %s *dst, const void *src)' % (dst_suffix, dst_native_type)
+    print 'util_format_%s_pack_%s(uint8_t *dst_row, unsigned dst_stride, const %s *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, src_suffix, src_native_type)
     print '{'
-    print '   void (*func)(%s *dst, const void *src);' % dst_native_type
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_unpack_%s;' % (format.short_name(), dst_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("unsupported format\\n");'
-    print '      return;'
-    print '   }'
-    print '   func(dst, src);'
+    
+    if is_format_supported(format):
+        print '   unsigned x, y;'
+        print '   for(y = 0; y < height; y += %u) {' % (format.block_height,)
+        print '      const %s *src = src_row;' % (src_native_type)
+        print '      uint8_t *dst = dst_row;'
+        print '      for(x = 0; x < width; x += %u) {' % (format.block_width,)
+    
+        generate_pack_kernel(format, src_channel, src_native_type)
+            
+        print '         src += 4;'
+        print '         dst += %u;' % (format.block_size() / 8,)
+        print '      }'
+        print '      dst_row += dst_stride;'
+        print '      src_row += src_stride/sizeof(*src_row);'
+        print '   }'
+        
     print '}'
     print
+    
 
+def generate_format_fetch(format, dst_channel, dst_native_type, dst_suffix):
+    '''Generate the function to unpack pixels from a particular format'''
 
-def generate_pack(formats, src_channel, src_native_type, src_suffix):
-    '''Generate the dispatch function to pack pixels to any format'''
-
-    for format in formats:
-        if is_format_supported(format):
-            generate_format_pack(format, src_channel, src_native_type, src_suffix)
+    name = format.short_name()
 
     print 'static INLINE void'
-    print 'util_format_pack_%s(enum pipe_format format, void *dst, %s r, %s g, %s b, %s a)' % (src_suffix, src_native_type, src_native_type, src_native_type, src_native_type)
+    print 'util_format_%s_fetch_%s(%s *dst, const uint8_t *src, unsigned i, unsigned j)' % (name, dst_suffix, dst_native_type)
     print '{'
-    print '   void (*func)(void *dst, %s r, %s g, %s b, %s a);' % (src_native_type, src_native_type, src_native_type, src_native_type)
-    print '   switch(format) {'
-    for format in formats:
-        if is_format_supported(format):
-            print '   case %s:' % format.name
-            print '      func = &util_format_%s_pack_%s;' % (format.short_name(), src_suffix)
-            print '      break;'
-    print '   default:'
-    print '      debug_printf("%s: unsupported format\\n", __FUNCTION__);'
-    print '      return;'
-    print '   }'
-    print '   func(dst, r, g, b, a);'
+
+    if is_format_supported(format):
+        generate_unpack_kernel(format, dst_channel, dst_native_type)
+
     print '}'
     print
 
 
-def main():
-    formats = []
-    for arg in sys.argv[1:]:
-        formats.extend(parse(arg))
+def is_format_hand_written(format):
+    return format.layout in ('s3tc', 'subsampled', 'other') or format.colorspace == ZS
 
-    print '/* This file is autogenerated by u_format_pack.py from u_format.csv. Do not edit directly. */'
-    print
-    # This will print the copyright message on the top of this file
-    print __doc__.strip()
 
-    print
-    print '#ifndef U_FORMAT_PACK_H'
-    print '#define U_FORMAT_PACK_H'
+def generate(formats):
     print
     print '#include "pipe/p_compiler.h"'
     print '#include "u_math.h"'
+    print '#include "u_half.h"'
     print '#include "u_format.h"'
+    print '#include "u_format_other.h"'
+    print '#include "u_format_srgb.h"'
+    print '#include "u_format_yuv.h"'
+    print '#include "u_format_zs.h"'
     print
 
-    generate_clamp()
-
     for format in formats:
-        if format.layout == PLAIN:
-            generate_format_type(format)
-
-    channel = Channel(FLOAT, False, 32)
-    native_type = 'float'
-    suffix = '4f'
+        if not is_format_hand_written(format):
+            
+            if is_format_supported(format):
+                generate_format_type(format)
 
-    generate_unpack(formats, channel, native_type, suffix)
-    generate_pack(formats, channel, native_type, suffix)
+            channel = Channel(FLOAT, False, 32)
+            native_type = 'float'
+            suffix = 'rgba_float'
 
-    channel = Channel(UNSIGNED, True, 8)
-    native_type = 'uint8_t'
-    suffix = '4ub'
+            generate_format_unpack(format, channel, native_type, suffix)
+            generate_format_pack(format, channel, native_type, suffix)
+            generate_format_fetch(format, channel, native_type, suffix)
 
-    generate_unpack(formats, channel, native_type, suffix)
-    generate_pack(formats, channel, native_type, suffix)
-
-    print
-    print '#ifdef __cplusplus'
-    print '}'
-    print '#endif'
-    print
-    print '#endif /* ! U_FORMAT_PACK_H */'
+            channel = Channel(UNSIGNED, True, 8)
+            native_type = 'uint8_t'
+            suffix = 'rgba_8unorm'
 
+            generate_format_unpack(format, channel, native_type, suffix)
+            generate_format_pack(format, channel, native_type, suffix)
 
-if __name__ == '__main__':
-    main()
diff --git a/src/gallium/auxiliary/util/u_format_parse.py b/src/gallium/auxiliary/util/u_format_parse.py
index 250926418ec..7076c676aaf 100755
--- a/src/gallium/auxiliary/util/u_format_parse.py
+++ b/src/gallium/auxiliary/util/u_format_parse.py
@@ -73,18 +73,22 @@ class Channel:
         '''Maximum representable number.'''
         if self.type == FLOAT:
             return VERY_LARGE
+        if self.type == FIXED:
+            return (1 << (self.size/2)) - 1
         if self.norm:
             return 1
         if self.type == UNSIGNED:
             return (1 << self.size) - 1
         if self.type == SIGNED:
-            return self.size - 1
+            return (1 << (self.size - 1)) - 1
         assert False
     
     def min(self):
         '''Minimum representable number.'''
         if self.type == FLOAT:
             return -VERY_LARGE
+        if self.type == FIXED:
+            return -(1 << (self.size/2))
         if self.type == UNSIGNED:
             return 0
         if self.norm:
@@ -134,6 +138,8 @@ class Format:
         return nr_channels
 
     def is_array(self):
+        if self.layout != PLAIN:
+            return False
         ref_channel = self.channels[0]
         for channel in self.channels[1:]:
             if channel.size and (channel.size != ref_channel.size or channel.size % 8):
@@ -141,7 +147,11 @@ class Format:
         return True
 
     def is_mixed(self):
+        if self.layout != PLAIN:
+            return False
         ref_channel = self.channels[0]
+        if ref_channel.type == VOID:
+           ref_channel = self.channels[1]
         for channel in self.channels[1:]:
             if channel.type != VOID:
                 if channel.type != ref_channel.type:
@@ -154,29 +164,29 @@ class Format:
         return is_pot(self.block_size())
 
     def is_int(self):
+        if self.layout != PLAIN:
+            return False
         for channel in self.channels:
             if channel.type not in (VOID, UNSIGNED, SIGNED):
                 return False
         return True
 
     def is_float(self):
+        if self.layout != PLAIN:
+            return False
         for channel in self.channels:
             if channel.type not in (VOID, FLOAT):
                 return False
         return True
 
     def is_bitmask(self):
-        if self.block_size() > 32:
+        if self.layout != PLAIN:
             return False
-        if not self.is_pot():
+        if self.block_size() not in (8, 16, 32):
             return False
         for channel in self.channels:
-            if not is_pot(channel.size):
-                return True
             if channel.type not in (VOID, UNSIGNED, SIGNED):
                 return False
-            if channel.size >= 32:
-                return False
         return True
 
     def inv_swizzles(self):
diff --git a/src/gallium/auxiliary/util/u_format_s3tc.c b/src/gallium/auxiliary/util/u_format_s3tc.c
new file mode 100644
index 00000000000..5b279b8fe26
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_s3tc.c
@@ -0,0 +1,728 @@
+/**************************************************************************
+ *
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 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
+ * 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 "u_dl.h"
+#include "u_math.h"
+#include "u_format.h"
+#include "u_format_s3tc.h"
+
+
+#if defined(_WIN32) || defined(WIN32)
+#define DXTN_LIBNAME "dxtn.dll"
+#elif defined(__APPLE__)
+#define DXTN_LIBNAME "libtxc_dxtn.dylib"
+#else
+#define DXTN_LIBNAME "libtxc_dxtn.so"
+#endif
+
+
+static void
+util_format_dxt1_rgb_fetch_stub(int src_stride,
+                                const uint8_t *src,
+                                int col, int row,
+                                uint8_t *dst)
+{
+   assert(0);
+}
+
+
+static void
+util_format_dxt1_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
+{
+   assert(0);
+}
+
+
+static void
+util_format_dxt3_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
+{
+   assert(0);
+}
+
+
+static void
+util_format_dxt5_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
+{
+   assert(0);
+}
+
+
+static void
+util_format_dxtn_pack_stub(int src_comps,
+                           int width, int height,
+                           const uint8_t *src,
+                           enum util_format_dxtn dst_format,
+                           uint8_t *dst,
+                           int dst_stride)
+{
+   assert(0);
+}
+
+
+boolean util_format_s3tc_enabled = FALSE;
+
+util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch = util_format_dxt1_rgb_fetch_stub;
+util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch = util_format_dxt1_rgba_fetch_stub;
+util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch = util_format_dxt3_rgba_fetch_stub;
+util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch = util_format_dxt5_rgba_fetch_stub;
+
+util_format_dxtn_pack_t util_format_dxtn_pack = util_format_dxtn_pack_stub;
+
+
+void
+util_format_s3tc_init(void)
+{
+   static boolean first_time = TRUE;
+   struct util_dl_library *library = NULL;
+   util_dl_proc fetch_2d_texel_rgb_dxt1;
+   util_dl_proc fetch_2d_texel_rgba_dxt1;
+   util_dl_proc fetch_2d_texel_rgba_dxt3;
+   util_dl_proc fetch_2d_texel_rgba_dxt5;
+   util_dl_proc tx_compress_dxtn;
+
+   if (!first_time)
+      return;
+   first_time = FALSE;
+
+   if (util_format_s3tc_enabled)
+      return;
+
+   library = util_dl_open(DXTN_LIBNAME);
+   if (!library) {
+      debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
+         "compression/decompression unavailable");
+      return;
+   }
+
+   fetch_2d_texel_rgb_dxt1 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgb_dxt1");
+   fetch_2d_texel_rgba_dxt1 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt1");
+   fetch_2d_texel_rgba_dxt3 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt3");
+   fetch_2d_texel_rgba_dxt5 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt5");
+   tx_compress_dxtn =
+         util_dl_get_proc_address(library, "tx_compress_dxtn");
+
+   if (!util_format_dxt1_rgb_fetch ||
+       !util_format_dxt1_rgba_fetch ||
+       !util_format_dxt3_rgba_fetch ||
+       !util_format_dxt5_rgba_fetch ||
+       !util_format_dxtn_pack) {
+      debug_printf("couldn't reference all symbols in " DXTN_LIBNAME
+                   ", software DXTn compression/decompression "
+                   "unavailable");
+      util_dl_close(library);
+      return;
+   }
+
+   util_format_dxt1_rgb_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgb_dxt1;
+   util_format_dxt1_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt1;
+   util_format_dxt3_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt3;
+   util_format_dxt5_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt5;
+   util_format_dxtn_pack = (util_format_dxtn_pack_t)tx_compress_dxtn;
+   util_format_s3tc_enabled = TRUE;
+}
+
+
+/*
+ * Pixel fetch.
+ */
+
+void
+util_format_dxt1_rgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgb_fetch(0, src, i, j, dst);
+}
+
+void
+util_format_dxt1_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgba_fetch(0, src, i, j, dst);
+}
+
+void
+util_format_dxt3_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt3_rgba_fetch(0, src, i, j, dst);
+}
+
+void
+util_format_dxt5_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt5_rgba_fetch(0, src, i, j, dst);
+}
+
+void
+util_format_dxt1_rgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   uint8_t tmp[4];
+   util_format_dxt1_rgb_fetch(0, src, i, j, tmp);
+   dst[0] = ubyte_to_float(tmp[0]);
+   dst[1] = ubyte_to_float(tmp[1]);
+   dst[2] = ubyte_to_float(tmp[2]);
+   dst[3] = 1.0;
+}
+
+void
+util_format_dxt1_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   uint8_t tmp[4];
+   util_format_dxt1_rgba_fetch(0, src, i, j, tmp);
+   dst[0] = ubyte_to_float(tmp[0]);
+   dst[1] = ubyte_to_float(tmp[1]);
+   dst[2] = ubyte_to_float(tmp[2]);
+   dst[3] = ubyte_to_float(tmp[3]);
+}
+
+void
+util_format_dxt3_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   uint8_t tmp[4];
+   util_format_dxt3_rgba_fetch(0, src, i, j, tmp);
+   dst[0] = ubyte_to_float(tmp[0]);
+   dst[1] = ubyte_to_float(tmp[1]);
+   dst[2] = ubyte_to_float(tmp[2]);
+   dst[3] = ubyte_to_float(tmp[3]);
+}
+
+void
+util_format_dxt5_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   uint8_t tmp[4];
+   util_format_dxt5_rgba_fetch(0, src, i, j, tmp);
+   dst[0] = ubyte_to_float(tmp[0]);
+   dst[1] = ubyte_to_float(tmp[1]);
+   dst[2] = ubyte_to_float(tmp[2]);
+   dst[3] = ubyte_to_float(tmp[3]);
+}
+
+
+/*
+ * Block decompression.
+ */
+
+static INLINE void
+util_format_dxtn_rgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height,
+                                        util_format_dxtn_fetch_t fetch,
+                                        unsigned block_size)
+{
+   const unsigned bw = 4, bh = 4, comps = 4;
+   unsigned x, y, i, j;
+   for(y = 0; y < height; y += bh) {
+      const uint8_t *src = src_row;
+      for(x = 0; x < width; x += bw) {
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*comps;
+               fetch(0, src, i, j, dst);
+            }
+         }
+         src += block_size;
+      }
+      src_row += src_stride;
+   }
+}
+
+void
+util_format_dxt1_rgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt1_rgb_fetch, 8);
+}
+
+void
+util_format_dxt1_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt1_rgba_fetch, 8);
+}
+
+void
+util_format_dxt3_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt3_rgba_fetch, 16);
+}
+
+void
+util_format_dxt5_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt5_rgba_fetch, 16);
+}
+
+static INLINE void
+util_format_dxtn_rgb_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height,
+                                       util_format_dxtn_fetch_t fetch,
+                                       unsigned block_size)
+{
+   unsigned x, y, i, j;
+   for(y = 0; y < height; y += 4) {
+      const uint8_t *src = src_row;
+      for(x = 0; x < width; x += 4) {
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+               uint8_t tmp[4];
+               fetch(0, src, i, j, tmp);
+               dst[0] = ubyte_to_float(tmp[0]);
+               dst[1] = ubyte_to_float(tmp[1]);
+               dst[2] = ubyte_to_float(tmp[2]);
+               dst[3] = ubyte_to_float(tmp[3]);
+            }
+         }
+         src += block_size;
+      }
+      src_row += src_stride;
+   }
+}
+
+void
+util_format_dxt1_rgb_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt1_rgb_fetch, 8);
+}
+
+void
+util_format_dxt1_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt1_rgba_fetch, 8);
+}
+
+void
+util_format_dxt3_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt3_rgba_fetch, 16);
+}
+
+void
+util_format_dxt5_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt5_rgba_fetch, 16);
+}
+
+
+/*
+ * Block compression.
+ */
+
+void
+util_format_dxt1_rgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                      const uint8_t *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   const unsigned bw = 4, bh = 4, bytes_per_block = 8;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][3];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < 3; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*4 + k];
+               }
+            }
+         }
+         util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, 0);
+         dst += bytes_per_block;
+      }
+      dst_row += dst_stride / sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt1_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 8;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, 0);
+         dst += bytes_per_block;
+      }
+      dst_row += dst_stride / sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt3_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 16;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, 0);
+         dst += bytes_per_block;
+      }
+      dst_row += dst_stride / sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt5_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 16;
+   unsigned x, y, i, j, k;
+
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, 0);
+         dst += bytes_per_block;
+      }
+      dst_row += dst_stride / sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt1_rgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                     const float *src, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][3];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 3; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
+               }
+            }
+         }
+         util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, 0);
+         dst += 8;
+      }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt1_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, 0);
+         dst += 8;
+      }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt3_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, 0);
+         dst += 16;
+      }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_dxt5_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
+               }
+            }
+         }
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, 0);
+         dst += 16;
+      }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+/*
+ * SRGB variants.
+ *
+ * FIXME: shunts to RGB for now
+ */
+
+void
+util_format_dxt1_srgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgb_fetch_rgba_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgba_fetch_rgba_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt3_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt3_rgba_fetch_rgba_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt5_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt5_rgba_fetch_rgba_8unorm(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgb_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgb_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgb_fetch_rgba_float(dst, src, i, j);
+}
+
+void
+util_format_dxt1_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt1_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt1_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt1_rgba_fetch_rgba_float(dst, src, i, j);
+}
+
+void
+util_format_dxt3_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt3_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt3_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt3_rgba_fetch_rgba_float(dst, src, i, j);
+}
+
+void
+util_format_dxt5_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+{
+   util_format_dxt5_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
+}
+
+void
+util_format_dxt5_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+{
+   util_format_dxt5_rgba_fetch_rgba_float(dst, src, i, j);
+}
+
diff --git a/src/gallium/auxiliary/util/u_format_s3tc.h b/src/gallium/auxiliary/util/u_format_s3tc.h
new file mode 100644
index 00000000000..97770abd422
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_s3tc.h
@@ -0,0 +1,218 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#ifndef U_FORMAT_S3TC_H_
+#define U_FORMAT_S3TC_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+enum util_format_dxtn {
+  UTIL_FORMAT_DXT1_RGB = 0x83F0,
+  UTIL_FORMAT_DXT1_RGBA = 0x83F1,
+  UTIL_FORMAT_DXT3_RGBA = 0x83F2,
+  UTIL_FORMAT_DXT5_RGBA = 0x83F3
+};
+
+
+typedef void
+(*util_format_dxtn_fetch_t)( int src_stride,
+                             const uint8_t *src,
+                             int col, int row,
+                             uint8_t *dst );
+
+typedef void
+(*util_format_dxtn_pack_t)( int src_comps,
+                            int width, int height,
+                            const uint8_t *src,
+                            enum util_format_dxtn dst_format,
+                            uint8_t *dst,
+                            int dst_stride);
+
+extern boolean util_format_s3tc_enabled;
+
+extern util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch;
+extern util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch;
+
+extern util_format_dxtn_pack_t util_format_dxtn_pack;
+
+
+void
+util_format_s3tc_init(void);
+
+
+void
+util_format_dxt1_rgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j);
+
+
+void
+util_format_dxt1_rgb_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgb_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt1_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt1_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt3_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt3_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+void
+util_format_dxt5_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+void
+util_format_dxt5_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j);
+
+
+#endif /* U_FORMAT_S3TC_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_srgb.h b/src/gallium/auxiliary/util/u_format_srgb.h
new file mode 100644
index 00000000000..43213fbeb3e
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_srgb.h
@@ -0,0 +1,106 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * SRGB translation.
+ *
+ * @author Brian Paul <[email protected]>
+ * @author Michal Krol <[email protected]>
+ * @author Jose Fonseca <[email protected]>
+ */
+
+#ifndef U_FORMAT_SRGB_H_
+#define U_FORMAT_SRGB_H_
+
+
+#include "pipe/p_compiler.h"
+#include "u_math.h"
+
+
+extern const float
+util_format_srgb_8unorm_to_linear_float_table[256];
+
+extern const uint8_t
+util_format_srgb_to_linear_8unorm_table[256];
+
+extern const uint8_t
+util_format_linear_to_srgb_8unorm_table[256];
+
+
+/**
+ * Convert a unclamped linear float to srgb value in the [0,255].
+ * XXX this hasn't been tested (render to srgb surface).
+ * XXX this needs optimization.
+ */
+static INLINE uint8_t
+util_format_linear_float_to_srgb_8unorm(float x)
+{
+   if (x >= 1.0f)
+      return 255;
+   else if (x >= 0.0031308f)
+      return float_to_ubyte(1.055f * powf(x, 0.41666f) - 0.055f);
+   else if (x > 0.0f)
+      return float_to_ubyte(12.92f * x);
+   else
+      return 0;
+}
+
+
+/**
+ * Convert an 8-bit sRGB value from non-linear space to a
+ * linear RGB value in [0, 1].
+ * Implemented with a 256-entry lookup table.
+ */
+static INLINE float
+util_format_srgb_8unorm_to_linear_float(uint8_t x)
+{
+   return util_format_srgb_8unorm_to_linear_float_table[x];
+}
+
+
+/**
+ * Convert a 8bit normalized value from linear to srgb.
+ */
+static INLINE uint8_t
+util_format_linear_to_srgb_8unorm(uint8_t x)
+{
+   return util_format_linear_to_srgb_8unorm_table[x];
+}
+
+
+/**
+ * Convert a 8bit normalized value from srgb to linear.
+ */
+static INLINE uint8_t
+util_format_srgb_to_linear_8unorm(uint8_t x)
+{
+   return util_format_srgb_to_linear_8unorm_table[x];
+}
+
+
+#endif /* U_FORMAT_SRGB_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_srgb.py b/src/gallium/auxiliary/util/u_format_srgb.py
new file mode 100644
index 00000000000..a4c76dc00b3
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_srgb.py
@@ -0,0 +1,101 @@
+#!/usr/bin/env python
+
+'''
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE 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.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * SRGB translation.
+ *
+ * @author Brian Paul <[email protected]>
+ * @author Michal Krol <[email protected]>
+ * @author Jose Fonseca <[email protected]>
+ */
+'''
+
+
+import sys
+import math
+
+
+def srgb_to_linear(x):
+    if x <= 0.04045:
+        return x / 12.92
+    else:
+        return math.pow((x + 0.055) / 1.055, 2.4)
+
+
+def linear_to_srgb(x):
+    if x >= 0.0031308:
+        return 1.055 * math.pow(x, 0.41666) - 0.055
+    else:
+        return 12.92 * x
+
+def generate_srgb_tables():
+    print 'const float'
+    print 'util_format_srgb_8unorm_to_linear_float_table[256] = {'
+    for j in range(0, 256, 4):
+        print '   ',
+        for i in range(j, j + 4):
+            print '%.7e,' % (srgb_to_linear(i / 255.0),),
+        print
+    print '};'
+    print
+    print 'const uint8_t'
+    print 'util_format_srgb_to_linear_8unorm_table[256] = {'
+    for j in range(0, 256, 16):
+        print '   ',
+        for i in range(j, j + 16):
+            print '%3u,' % (int(srgb_to_linear(i / 255.0) * 255.0 + 0.5),),
+        print
+    print '};'
+    print
+    print 'const uint8_t'
+    print 'util_format_linear_to_srgb_8unorm_table[256] = {'
+    for j in range(0, 256, 16):
+        print '   ',
+        for i in range(j, j + 16):
+            print '%3u,' % (int(linear_to_srgb(i / 255.0) * 255.0 + 0.5),),
+        print
+    print '};'
+    print
+
+
+def main():
+    print '/* This file is autogenerated by u_format_srgb.py. Do not edit directly. */'
+    print
+    # This will print the copyright message on the top of this file
+    print __doc__.strip()
+    print
+    print '#include "u_format_srgb.h"'
+    print
+    generate_srgb_tables()    
+
+
+if __name__ == '__main__':
+    main()
diff --git a/src/gallium/auxiliary/util/u_format_table.py b/src/gallium/auxiliary/util/u_format_table.py
index 4e29d15f3bb..ae9a5981973 100755
--- a/src/gallium/auxiliary/util/u_format_table.py
+++ b/src/gallium/auxiliary/util/u_format_table.py
@@ -3,7 +3,7 @@
 '''
 /**************************************************************************
  *
- * Copyright 2009 VMware, Inc.
+ * Copyright 2010 VMware, Inc.
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -33,6 +33,7 @@
 import sys
 
 from u_format_parse import *
+import u_format_pack
 
 
 def layout_map(layout):
@@ -85,30 +86,22 @@ def write_format_table(formats):
     print __doc__.strip()
     print
     print '#include "u_format.h"'
+    print '#include "u_format_s3tc.h"'
     print
-    print 'const struct util_format_description'
-    print 'util_format_none_description = {'
-    print "   PIPE_FORMAT_NONE,"
-    print "   \"PIPE_FORMAT_NONE\","
-    print "   {0, 0, 0},"
-    print "   0,"
-    print "   0,"
-    print "   0,"
-    print "   0,"
-    print "   {{0, 0, 0}, {0, 0, 0}, {0, 0, 0}, {0, 0, 0}},"
-    print "   {0, 0, 0, 0},"
-    print "   0"
-    print "};"
-    print
+    
+    u_format_pack.generate(formats)
+    
     for format in formats:
         print 'const struct util_format_description'
         print 'util_format_%s_description = {' % (format.short_name(),)
         print "   %s," % (format.name,)
         print "   \"%s\"," % (format.name,)
+        print "   \"%s\"," % (format.short_name(),)
         print "   {%u, %u, %u},\t/* block */" % (format.block_width, format.block_height, format.block_size())
         print "   %s," % (layout_map(format.layout),)
         print "   %u,\t/* nr_channels */" % (format.nr_channels(),)
         print "   %s,\t/* is_array */" % (bool_map(format.is_array()),)
+        print "   %s,\t/* is_bitmask */" % (bool_map(format.is_bitmask()),)
         print "   %s,\t/* is_mixed */" % (bool_map(format.is_mixed()),)
         print "   {"
         for i in range(4):
@@ -136,8 +129,37 @@ def write_format_table(formats):
             print "      %s%s\t/* %s */" % (swizzle_map[swizzle], sep, comment)
         print "   },"
         print "   %s," % (colorspace_map(format.colorspace),)
+        if format.colorspace != ZS:
+            print "   &util_format_%s_unpack_rgba_8unorm," % format.short_name() 
+            print "   &util_format_%s_pack_rgba_8unorm," % format.short_name() 
+            print "   &util_format_%s_unpack_rgba_float," % format.short_name() 
+            print "   &util_format_%s_pack_rgba_float," % format.short_name() 
+            print "   &util_format_%s_fetch_rgba_float," % format.short_name()
+        else:
+            print "   NULL, /* unpack_rgba_8unorm */" 
+            print "   NULL, /* pack_rgba_8unorm */" 
+            print "   NULL, /* unpack_rgba_float */" 
+            print "   NULL, /* pack_rgba_float */" 
+            print "   NULL, /* fetch_rgba_float */" 
+        if format.colorspace == ZS and format.swizzles[0] != SWIZZLE_NONE:
+            print "   &util_format_%s_unpack_z_32unorm," % format.short_name() 
+            print "   &util_format_%s_pack_z_32unorm," % format.short_name() 
+            print "   &util_format_%s_unpack_z_float," % format.short_name() 
+            print "   &util_format_%s_pack_z_float," % format.short_name() 
+        else:
+            print "   NULL, /* unpack_z_32unorm */" 
+            print "   NULL, /* pack_z_32unorm */" 
+            print "   NULL, /* unpack_z_float */" 
+            print "   NULL, /* pack_z_float */" 
+        if format.colorspace == ZS and format.swizzles[1] != SWIZZLE_NONE:
+            print "   &util_format_%s_unpack_s_8uscaled," % format.short_name() 
+            print "   &util_format_%s_pack_s_8uscaled" % format.short_name() 
+        else:
+            print "   NULL, /* unpack_s_8uscaled */" 
+            print "   NULL /* pack_s_8uscaled */" 
         print "};"
         print
+        
     print "const struct util_format_description *"
     print "util_format_description(enum pipe_format format)"
     print "{"
@@ -146,13 +168,10 @@ def write_format_table(formats):
     print "   }"
     print
     print "   switch (format) {"
-    print "   case PIPE_FORMAT_NONE:"
-    print "      return &util_format_none_description;"
     for format in formats:
         print "   case %s:" % format.name
         print "      return &util_format_%s_description;" % (format.short_name(),)
     print "   default:"
-    print "      assert(0);"
     print "      return NULL;"
     print "   }"
     print "}"
diff --git a/src/gallium/auxiliary/util/u_format_tests.c b/src/gallium/auxiliary/util/u_format_tests.c
new file mode 100644
index 00000000000..b1df6c49c42
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_tests.c
@@ -0,0 +1,970 @@
+/**************************************************************************
+ *
+ * Copyright 2009-2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+#include "u_memory.h"
+#include "u_format_tests.h"
+
+
+/*
+ * Helper macros to create the packed bytes for longer words.
+ */
+
+#define PACKED_1x8(x)          {x, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_2x8(x, y)       {x, y, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_3x8(x, y, z)    {x, y, z, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_4x8(x, y, z, w) {x, y, z, w, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_8x8(a, b, c, d, e, f, g, h) {a, b, c, d, e, f, g, h, 0, 0, 0, 0, 0, 0, 0, 0}
+
+#define PACKED_1x16(x)          {(x) & 0xff, (x) >> 8,          0,        0,          0,        0,          0,        0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_2x16(x, y)       {(x) & 0xff, (x) >> 8, (y) & 0xff, (y) >> 8,          0,        0,          0,        0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_3x16(x, y, z)    {(x) & 0xff, (x) >> 8, (y) & 0xff, (y) >> 8, (z) & 0xff, (z) >> 8,          0,        0, 0, 0, 0, 0, 0, 0, 0, 0}
+#define PACKED_4x16(x, y, z, w) {(x) & 0xff, (x) >> 8, (y) & 0xff, (y) >> 8, (z) & 0xff, (z) >> 8, (w) & 0xff, (w) >> 8, 0, 0, 0, 0, 0, 0, 0, 0}
+
+#define PACKED_1x32(x)          {(x) & 0xff, ((x) >> 8) & 0xff, ((x) >> 16) & 0xff, (x) >> 24,          0,                 0,                  0,         0,          0,                 0,                  0,         0,          0,                 0,                  0,         0}
+#define PACKED_2x32(x, y)       {(x) & 0xff, ((x) >> 8) & 0xff, ((x) >> 16) & 0xff, (x) >> 24, (y) & 0xff, ((y) >> 8) & 0xff, ((y) >> 16) & 0xff, (y) >> 24,          0,                 0,                  0,         0,          0,                 0,                  0,         0}
+#define PACKED_3x32(x, y, z)    {(x) & 0xff, ((x) >> 8) & 0xff, ((x) >> 16) & 0xff, (x) >> 24, (y) & 0xff, ((y) >> 8) & 0xff, ((y) >> 16) & 0xff, (y) >> 24, (z) & 0xff, ((z) >> 8) & 0xff, ((z) >> 16) & 0xff, (z) >> 24,          0,                 0,                  0,         0}
+#define PACKED_4x32(x, y, z, w) {(x) & 0xff, ((x) >> 8) & 0xff, ((x) >> 16) & 0xff, (x) >> 24, (y) & 0xff, ((y) >> 8) & 0xff, ((y) >> 16) & 0xff, (y) >> 24, (z) & 0xff, ((z) >> 8) & 0xff, ((z) >> 16) & 0xff, (z) >> 24, (w) & 0xff, ((w) >> 8) & 0xff, ((w) >> 16) & 0xff, (w) >> 24}
+
+#define UNPACKED_1x1(r, g, b, a) \
+      {{{r, g, b, a}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}, {0, 0, 0, 0}}}
+
+#define UNPACKED_2x1(r0, g0, b0, a0, r1, g1, b1, a1) \
+      {{{r0, g0, b0, a0}, {r1, g1, b1, a1}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{ 0,  0,  0,  0}, { 0,  0,  0,  0}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{ 0,  0,  0,  0}, { 0,  0,  0,  0}, {0, 0, 0, 0}, {0, 0, 0, 0}}, \
+       {{ 0,  0,  0,  0}, { 0,  0,  0,  0}, {0, 0, 0, 0}, {0, 0, 0, 0}}}
+
+
+/**
+ * Test cases.
+ *
+ * These were manually entered. We could generate these
+ *
+ * To keep this to a we cover only the corner cases, which should produce
+ * good enough coverage since that pixel format transformations are afine for
+ * non SRGB formats.
+ */
+const struct util_format_test_case
+util_format_test_cases[] =
+{
+
+   /*
+    * 32-bit rendertarget formats
+    */
+
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8A8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000ff00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0xff000000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000003ff), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000ffc00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x3ff00000), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xc0000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000003ff), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000ffc00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x3ff00000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xc0000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B10G10R10A2_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * 16-bit rendertarget formats
+    */
+
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x001f), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x03e0), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7c00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7fff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x001f), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x03e0), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x7c00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x8000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B4G4R4X4_UNORM, PACKED_1x16(0x0fff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B4G4R4X4_UNORM, PACKED_1x16(0x0fff), PACKED_1x16(0x000f), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_B4G4R4X4_UNORM, PACKED_1x16(0x0fff), PACKED_1x16(0x00f0), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B4G4R4X4_UNORM, PACKED_1x16(0x0fff), PACKED_1x16(0x0f00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B4G4R4X4_UNORM, PACKED_1x16(0x0fff), PACKED_1x16(0x0fff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x000f), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x00f0), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0f00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xf000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B4G4R4A4_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B5G6R5_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G6R5_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x001f), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_B5G6R5_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x07e0), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G6R5_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xf800), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B5G6R5_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * Luminance/intensity/alpha formats
+    */
+
+   {PIPE_FORMAT_L8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_A8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_I8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_I8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_L4A4_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_L4A4_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x0f), UNPACKED_1x1(1.0, 1.0, 1.0, 0.0)},
+   {PIPE_FORMAT_L4A4_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xf0), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L4A4_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_L8A8_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_L8A8_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x00ff), UNPACKED_1x1(1.0, 1.0, 1.0, 0.0)},
+   {PIPE_FORMAT_L8A8_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xff00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L8A8_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_L16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * SRGB formats
+    */
+
+   {PIPE_FORMAT_L8_SRGB, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L8_SRGB, PACKED_1x8(0xff), PACKED_1x8(0xbc), UNPACKED_1x1(0.502886458033, 0.502886458033, 0.502886458033, 1.0)},
+   {PIPE_FORMAT_L8_SRGB, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0x00bc), UNPACKED_1x1(0.502886458033, 0.502886458033, 0.502886458033, 0.0)},
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0x00ff), UNPACKED_1x1(1.0, 1.0, 1.0, 0.0)},
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0xcc00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.8)},
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0xff00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_L8A8_SRGB, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xbc, 0x00, 0x00), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0x00, 0x00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0xbc, 0x00), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0xff, 0x00), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0xbc), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0xff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SRGB, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0xff, 0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xbc, 0x00, 0x00, 0x00), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0x00, 0x00, 0x00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xbc, 0x00, 0x00), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xff, 0x00, 0x00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xbc, 0x00), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xcc), UNPACKED_1x1(0.0, 0.0, 0.0, 0.8)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8A8_SRGB, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0xff, 0xff, 0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000bc), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xcc000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.8)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8A8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x000000bc), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_B8G8R8X8_SRGB, PACKED_1x32(0x00ffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000cc), UNPACKED_1x1(0.0, 0.0, 0.0, 0.8)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xbc000000), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_A8R8G8B8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000ff00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xbc000000), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_X8R8G8B8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000cc), UNPACKED_1x1(0.0, 0.0, 0.0, 0.8)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xbc000000), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_A8B8G8R8_SRGB, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000bc00), UNPACKED_1x1(0.0, 0.0, 0.502886458033, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x0000ff00), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00bc0000), UNPACKED_1x1(0.0, 0.502886458033, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0x00ff0000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xbc000000), UNPACKED_1x1(0.502886458033, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xff000000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_X8B8G8R8_SRGB, PACKED_1x32(0xffffff00), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * Mixed-signed formats
+    */
+
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1( 0.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x7f, 0x00, 0x00, 0x00), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x81, 0x00, 0x00, 0x00), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x00, 0x7f, 0x00, 0x00), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x00, 0x81, 0x00, 0x00), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8SG8SB8UX8U_NORM, PACKED_4x8(0xff, 0xff, 0xff, 0x00), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_1x1( 0.0,  0.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000001ff), UNPACKED_1x1( 1.0,  0.0,  0.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000201), UNPACKED_1x1(-1.0,  0.0,  0.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x0007fc00), UNPACKED_1x1( 0.0,  1.0,  0.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00080400), UNPACKED_1x1( 0.0, -1.0,  0.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x1ff00000), UNPACKED_1x1( 0.0,  0.0,  1.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x20100000), UNPACKED_1x1( 0.0,  0.0, -1.0, 0.0)},
+   {PIPE_FORMAT_R10SG10SB10SA2U_NORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xc0000000), UNPACKED_1x1( 0.0,  0.0,  0.0, 1.0)},
+
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1( 0.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0x000f), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0x0011), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0x01e0), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0x0220), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R5SG5SB6U_NORM, PACKED_1x16(0xffff), PACKED_1x16(0xfc00), UNPACKED_1x1( 0.0,  0.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8Bx_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x00), UNPACKED_1x1( 0.0,  0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R8G8Bx_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x7f, 0x00), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8Bx_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x81, 0x00), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8Bx_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x7f), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8Bx_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x81), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+
+   /*
+    * Depth-stencil formats
+    */
+
+   {PIPE_FORMAT_S8_USCALED, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_S8_USCALED, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(0.0, 255.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z32_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z32_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z32_FLOAT, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z32_FLOAT, PACKED_1x32(0xffffffff), PACKED_1x32(0x3f800000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z24_UNORM_S8_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z24_UNORM_S8_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00ffffff), UNPACKED_1x1(1.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z24_UNORM_S8_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(0.0, 255.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z24_UNORM_S8_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 255.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_S8_USCALED_Z24_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_S8_USCALED_Z24_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffff00), UNPACKED_1x1(1.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_S8_USCALED_Z24_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x000000ff), UNPACKED_1x1(0.0, 255.0, 0.0, 0.0)},
+   {PIPE_FORMAT_S8_USCALED_Z24_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 255.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z24X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z24X8_UNORM, PACKED_1x32(0x00ffffff), PACKED_1x32(0x00ffffff), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_X8Z24_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_X8Z24_UNORM, PACKED_1x32(0xffffff00), PACKED_1x32(0xffffff00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+
+   {PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED, PACKED_2x32(0xffffffff, 0x000000ff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1( 0.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED, PACKED_2x32(0xffffffff, 0x000000ff), PACKED_2x32(0x3f800000, 0x00000000), UNPACKED_1x1( 1.0,   0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_Z32_FLOAT_S8X24_USCALED, PACKED_2x32(0xffffffff, 0x000000ff), PACKED_2x32(0x00000000, 0x000000ff), UNPACKED_1x1( 0.0, 255.0, 0.0, 0.0)},
+
+   /*
+    * YUV formats
+    */
+
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0x00, 0x00, 0x00), UNPACKED_2x1(1.0, 0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xff, 0x00, 0x00), UNPACKED_2x1(0.0, 1.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_2x1(0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xff), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_B8G8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0xff, 0xff, 0xff), UNPACKED_2x1(1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0x00, 0x00, 0x00), UNPACKED_2x1(0.0, 1.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xff, 0x00, 0x00), UNPACKED_2x1(1.0, 0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xff), UNPACKED_2x1(0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_G8R8_G8B8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0xff, 0xff, 0xff), UNPACKED_2x1(1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * TODO: Exercise the UV channels as well.
+    */
+   {PIPE_FORMAT_UYVY, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x80, 0x10, 0x80, 0x10), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_UYVY, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x80, 0xeb, 0x80, 0x10), UNPACKED_2x1(1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_UYVY, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x80, 0x10, 0x80, 0xeb), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_YUYV, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x10, 0x80, 0x10, 0x80), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_YUYV, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xeb, 0x80, 0x10, 0x80), UNPACKED_2x1(1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_YUYV, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x10, 0x80, 0xeb, 0x80), UNPACKED_2x1(0.0, 0.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0)},
+
+   /*
+    * Compressed formats
+    */
+
+   {
+      PIPE_FORMAT_DXT1_RGB,
+      PACKED_8x8(0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff),
+      PACKED_8x8(0xf2, 0xd7, 0xb0, 0x20, 0xae, 0x2c, 0x6f, 0x97),
+      {
+         {
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0},
+            {0x5d/255.0, 0x62/255.0, 0x89/255.0, 0xff/255.0},
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0},
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0}
+         },
+         {
+            {0xd6/255.0, 0xff/255.0, 0x94/255.0, 0xff/255.0},
+            {0x5d/255.0, 0x62/255.0, 0x89/255.0, 0xff/255.0},
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0},
+            {0xd6/255.0, 0xff/255.0, 0x94/255.0, 0xff/255.0}
+         },
+         {
+            {0x5d/255.0, 0x62/255.0, 0x89/255.0, 0xff/255.0},
+            {0x5d/255.0, 0x62/255.0, 0x89/255.0, 0xff/255.0},
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0},
+            {0x21/255.0, 0x14/255.0, 0x84/255.0, 0xff/255.0}
+         },
+         {
+            {0x5d/255.0, 0x62/255.0, 0x89/255.0, 0xff/255.0},
+            {0x21/255.0, 0x14/255.0, 0x84/255.0, 0xff/255.0},
+            {0x21/255.0, 0x14/255.0, 0x84/255.0, 0xff/255.0},
+            {0x99/255.0, 0xb0/255.0, 0x8e/255.0, 0xff/255.0}
+         }
+      }
+   },
+   {
+      PIPE_FORMAT_DXT1_RGBA,
+      PACKED_8x8(0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff),
+      PACKED_8x8(0xff, 0x2f, 0xa4, 0x72, 0xeb, 0xb2, 0xbd, 0xbe),
+      {
+         {
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0},
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0}
+         },
+         {
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0},
+            {0x29/255.0, 0xff/255.0, 0xff/255.0, 0xff/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0}
+         },
+         {
+            {0x73/255.0, 0x55/255.0, 0x21/255.0, 0xff/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0}
+         },
+         {
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x00/255.0, 0x00/255.0, 0x00/255.0, 0x00/255.0},
+            {0x4e/255.0, 0xaa/255.0, 0x90/255.0, 0xff/255.0}
+         }
+      }
+   },
+   {
+      PIPE_FORMAT_DXT3_RGBA,
+      {0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff},
+      {0xe7, 0x4a, 0x8f, 0x96, 0x5b, 0xc1, 0x1c, 0x84, 0xf6, 0x8f, 0xab, 0x32, 0x2a, 0x9a, 0x95, 0x5a},
+      {
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x77/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xee/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xaa/255.0},
+            {0x8c/255.0, 0xff/255.0, 0xb5/255.0, 0x44/255.0}
+         },
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xff/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x88/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x66/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x99/255.0}
+         },
+         {
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0xbb/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x55/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x11/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xcc/255.0}
+         },
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xcc/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x11/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x44/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x88/255.0}
+         }
+      }
+   },
+   {
+      PIPE_FORMAT_DXT5_RGBA,
+      {0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff},
+      {0xf8, 0x11, 0xc5, 0x0c, 0x9a, 0x73, 0xb4, 0x9c, 0xf6, 0x8f, 0xab, 0x32, 0x2a, 0x9a, 0x95, 0x5a},
+      {
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x74/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xf8/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xb6/255.0},
+            {0x8c/255.0, 0xff/255.0, 0xb5/255.0, 0x53/255.0}
+         },
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xf8/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x95/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x53/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x95/255.0}
+         },
+         {
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0xb6/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x53/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x11/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xd7/255.0}
+         },
+         {
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0xb6/255.0},
+            {0x6d/255.0, 0xc6/255.0, 0x96/255.0, 0x11/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x32/255.0},
+            {0x31/255.0, 0x55/255.0, 0x5a/255.0, 0x95/255.0}
+         }
+      }
+   },
+
+
+   /*
+    * Standard 8-bit integer formats
+    */
+
+   {PIPE_FORMAT_R8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8_UNORM, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8_UNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_UNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0xff, 0x00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_UNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0xff), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_UNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0xff, 0xff), UNPACKED_1x1(1.0, 1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8_UNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_UNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0x00, 0x00), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_UNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0xff, 0x00), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_UNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0xff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_UNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0xff, 0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0x00, 0x00, 0x00), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xff, 0x00, 0x00), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8A8_UNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0xff, 0xff, 0xff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R8_USCALED, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(  0.0,   0.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8_USCALED, PACKED_1x8(0xff), PACKED_1x8(0xff), UNPACKED_1x1(255.0,   0.0,   0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8_USCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x00), UNPACKED_1x1(  0.0,   0.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_USCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0xff, 0x00), UNPACKED_1x1(255.0,   0.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_USCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0xff), UNPACKED_1x1(  0.0, 255.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_USCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0xff, 0xff), UNPACKED_1x1(255.0, 255.0,   0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8_USCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x00), UNPACKED_1x1(  0.0,   0.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_USCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0x00, 0x00), UNPACKED_1x1(255.0,   0.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_USCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0xff, 0x00), UNPACKED_1x1(  0.0, 255.0,   0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_USCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0xff), UNPACKED_1x1(  0.0,   0.0, 255.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_USCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0xff, 0xff, 0xff), UNPACKED_1x1(255.0, 255.0, 255.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1(  0.0,   0.0,   0.0,   0.0)},
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0x00, 0x00, 0x00), UNPACKED_1x1(255.0,   0.0,   0.0,   0.0)},
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0xff, 0x00, 0x00), UNPACKED_1x1(  0.0, 255.0,   0.0,   0.0)},
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0xff, 0x00), UNPACKED_1x1(  0.0,   0.0, 255.0,   0.0)},
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0xff), UNPACKED_1x1(  0.0,   0.0,   0.0, 255.0)},
+   {PIPE_FORMAT_R8G8B8A8_USCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0xff, 0xff, 0xff, 0xff), UNPACKED_1x1(255.0, 255.0, 255.0, 255.0)},
+
+   {PIPE_FORMAT_R8_SNORM, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8_SNORM, PACKED_1x8(0xff), PACKED_1x8(0x7f), UNPACKED_1x1( 1.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8_SNORM, PACKED_1x8(0xff), PACKED_1x8(0x81), UNPACKED_1x1(-1.0,  0.0,  0.0,  1.0)},
+
+   {PIPE_FORMAT_R8G8_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x00), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x7f, 0x00), UNPACKED_1x1( 1.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x81, 0x00), UNPACKED_1x1(-1.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x7f), UNPACKED_1x1( 0.0,  1.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8_SNORM, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x81), UNPACKED_1x1( 0.0, -1.0,  0.0,  1.0)},
+
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x00), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x7f, 0x00, 0x00), UNPACKED_1x1( 1.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x81, 0x00, 0x00), UNPACKED_1x1(-1.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x7f, 0x00), UNPACKED_1x1( 0.0,  1.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x81, 0x00), UNPACKED_1x1( 0.0, -1.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x7f), UNPACKED_1x1( 0.0,  0.0,  1.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8_SNORM, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x81), UNPACKED_1x1( 0.0,  0.0, -1.0,  1.0)},
+
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1( 0.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x7f, 0x00, 0x00, 0x00), UNPACKED_1x1( 1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x81, 0x00, 0x00, 0x00), UNPACKED_1x1(-1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x7f, 0x00, 0x00), UNPACKED_1x1( 0.0,  1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x81, 0x00, 0x00), UNPACKED_1x1( 0.0, -1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x7f, 0x00), UNPACKED_1x1( 0.0,  0.0,  1.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x81, 0x00), UNPACKED_1x1( 0.0,  0.0, -1.0,  0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x7f), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R8G8B8A8_SNORM, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x81), UNPACKED_1x1( 0.0,  0.0,  0.0, -1.0)},
+
+   {PIPE_FORMAT_R8_SSCALED, PACKED_1x8(0xff), PACKED_1x8(0x00), UNPACKED_1x1(   0.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8_SSCALED, PACKED_1x8(0xff), PACKED_1x8(0x7f), UNPACKED_1x1( 127.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8_SSCALED, PACKED_1x8(0xff), PACKED_1x8(0x80), UNPACKED_1x1(-128.0,    0.0,    0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8_SSCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x00), UNPACKED_1x1(   0.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_SSCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x7f, 0x00), UNPACKED_1x1( 127.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_SSCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x80, 0x00), UNPACKED_1x1(-128.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_SSCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x7f), UNPACKED_1x1(   0.0,  127.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8_SSCALED, PACKED_2x8(0xff, 0xff), PACKED_2x8(0x00, 0x80), UNPACKED_1x1(   0.0, -128.0,    0.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x00), UNPACKED_1x1(   0.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x7f, 0x00, 0x00), UNPACKED_1x1( 127.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x80, 0x00, 0x00), UNPACKED_1x1(-128.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x7f, 0x00), UNPACKED_1x1(   0.0,  127.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x80, 0x00), UNPACKED_1x1(   0.0, -128.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x7f), UNPACKED_1x1(   0.0,    0.0,  127.0, 1.0)},
+   {PIPE_FORMAT_R8G8B8_SSCALED, PACKED_3x8(0xff, 0xff, 0xff), PACKED_3x8(0x00, 0x00, 0x80), UNPACKED_1x1(   0.0,    0.0, -128.0, 1.0)},
+
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x00), UNPACKED_1x1(   0.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x7f, 0x00, 0x00, 0x00), UNPACKED_1x1( 127.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x80, 0x00, 0x00, 0x00), UNPACKED_1x1(-128.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x7f, 0x00, 0x00), UNPACKED_1x1(   0.0,  127.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x80, 0x00, 0x00), UNPACKED_1x1(   0.0, -128.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x7f, 0x00), UNPACKED_1x1(   0.0,    0.0,  127.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x80, 0x00), UNPACKED_1x1(   0.0,    0.0, -128.0,    0.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x7f), UNPACKED_1x1(   0.0,    0.0,    0.0,  127.0)},
+   {PIPE_FORMAT_R8G8B8A8_SSCALED, PACKED_4x8(0xff, 0xff, 0xff, 0xff), PACKED_4x8(0x00, 0x00, 0x00, 0x80), UNPACKED_1x1(   0.0,    0.0,    0.0, -128.0)},
+
+   /*
+    * Standard 16-bit integer formats
+    */
+
+   {PIPE_FORMAT_R16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16_UNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_UNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0xffff, 0x0000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_UNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0xffff), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_UNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0xffff, 0xffff), UNPACKED_1x1(1.0, 1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16B16_UNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_UNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0xffff, 0x0000, 0x0000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_UNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0xffff, 0x0000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_UNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0xffff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_UNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0xffff, 0xffff, 0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0xffff, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0xffff, 0x0000, 0x0000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0xffff, 0x0000), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0xffff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16A16_UNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R16_USCALED, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(    0.0,     0.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16_USCALED, PACKED_1x16(0xffff), PACKED_1x16(0xffff), UNPACKED_1x1(65535.0,     0.0,     0.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16_USCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x0000), UNPACKED_1x1(    0.0,     0.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_USCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0xffff, 0x0000), UNPACKED_1x1(65535.0,     0.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_USCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0xffff), UNPACKED_1x1(    0.0, 65535.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_USCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0xffff, 0xffff), UNPACKED_1x1(65535.0, 65535.0,     0.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16B16_USCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x0000), UNPACKED_1x1(    0.0,     0.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_USCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0xffff, 0x0000, 0x0000), UNPACKED_1x1(65535.0,     0.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_USCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0xffff, 0x0000), UNPACKED_1x1(    0.0, 65535.0,     0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_USCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0xffff), UNPACKED_1x1(    0.0,     0.0, 65535.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_USCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0xffff, 0xffff, 0xffff), UNPACKED_1x1(65535.0, 65535.0, 65535.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(    0.0,     0.0,     0.0,     0.0)},
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0xffff, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(65535.0,     0.0,     0.0,     0.0)},
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0xffff, 0x0000, 0x0000), UNPACKED_1x1(    0.0, 65535.0,     0.0,     0.0)},
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0xffff, 0x0000), UNPACKED_1x1(    0.0,     0.0, 65535.0,     0.0)},
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0xffff), UNPACKED_1x1(    0.0,     0.0,     0.0, 65535.0)},
+   {PIPE_FORMAT_R16G16B16A16_USCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), UNPACKED_1x1(65535.0, 65535.0, 65535.0, 65535.0)},
+
+   {PIPE_FORMAT_R16_SNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16_SNORM, PACKED_1x16(0xffff), PACKED_1x16(0x7fff), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16_SNORM, PACKED_1x16(0xffff), PACKED_1x16(0x8001), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+
+   {PIPE_FORMAT_R16G16_SNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x0000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16_SNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x7fff, 0x0000), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16_SNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x8001, 0x0000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16_SNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x7fff), UNPACKED_1x1(   0.0,    1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16_SNORM, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x8001), UNPACKED_1x1(   0.0,   -1.0,    0.0,    1.0)},
+
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x0000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x7fff, 0x0000, 0x0000), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x8001, 0x0000, 0x0000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x7fff, 0x0000), UNPACKED_1x1(   0.0,    1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x8001, 0x0000), UNPACKED_1x1(   0.0,   -1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x7fff), UNPACKED_1x1(   0.0,    0.0,    1.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16_SNORM, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x8001), UNPACKED_1x1(   0.0,    0.0,   -1.0,    1.0)},
+
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(   0.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x7fff, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(   1.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x8001, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x7fff, 0x0000, 0x0000), UNPACKED_1x1(   0.0,    1.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x8001, 0x0000, 0x0000), UNPACKED_1x1(   0.0,   -1.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x7fff, 0x0000), UNPACKED_1x1(   0.0,    0.0,    1.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x8001, 0x0000), UNPACKED_1x1(   0.0,    0.0,   -1.0,    0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x7fff), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R16G16B16A16_SNORM, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x8001), UNPACKED_1x1(   0.0,    0.0,    0.0,   -1.0)},
+
+   {PIPE_FORMAT_R16_SSCALED, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(     0.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16_SSCALED, PACKED_1x16(0xffff), PACKED_1x16(0x7fff), UNPACKED_1x1( 32767.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16_SSCALED, PACKED_1x16(0xffff), PACKED_1x16(0x8000), UNPACKED_1x1(-32768.0,      0.0,      0.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16_SSCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x0000), UNPACKED_1x1(     0.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_SSCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x7fff, 0x0000), UNPACKED_1x1( 32767.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_SSCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x8000, 0x0000), UNPACKED_1x1(-32768.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_SSCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x7fff), UNPACKED_1x1(     0.0,  32767.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16_SSCALED, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x8000), UNPACKED_1x1(     0.0, -32768.0,      0.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x0000), UNPACKED_1x1(     0.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x7fff, 0x0000, 0x0000), UNPACKED_1x1( 32767.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x8000, 0x0000, 0x0000), UNPACKED_1x1(-32768.0,      0.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x7fff, 0x0000), UNPACKED_1x1(     0.0,  32767.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x8000, 0x0000), UNPACKED_1x1(     0.0, -32768.0,      0.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x7fff), UNPACKED_1x1(     0.0,      0.0,  32767.0,   1.0)},
+   {PIPE_FORMAT_R16G16B16_SSCALED, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x8000), UNPACKED_1x1(     0.0,      0.0, -32768.0,   1.0)},
+
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(     0.0,      0.0,      0.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x7fff, 0x0000, 0x0000, 0x0000), UNPACKED_1x1( 32767.0,      0.0,      0.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x8000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(-32768.0,      0.0,      0.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x7fff, 0x0000, 0x0000), UNPACKED_1x1(     0.0,  32767.0,      0.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x8000, 0x0000, 0x0000), UNPACKED_1x1(     0.0, -32768.0,      0.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x7fff, 0x0000), UNPACKED_1x1(     0.0,      0.0,  32767.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x8000, 0x0000), UNPACKED_1x1(     0.0,      0.0, -32768.0,      0.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x7fff), UNPACKED_1x1(     0.0,      0.0,      0.0,  32767.0)},
+   {PIPE_FORMAT_R16G16B16A16_SSCALED, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x8000), UNPACKED_1x1(     0.0,      0.0,      0.0, -32768.0)},
+
+   /*
+    * Standard 32-bit integer formats
+    *
+    * NOTE: We can't accurately represent integers larger than +/-0x1000000
+    * with single precision floats, so that's as far as we test.
+    */
+
+   {PIPE_FORMAT_R32_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32_UNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0xffffffff), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32_UNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_UNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0xffffffff, 0x00000000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_UNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0xffffffff), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_UNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0xffffffff, 0xffffffff), UNPACKED_1x1(1.0, 1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32_UNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_UNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0xffffffff, 0x00000000, 0x00000000), UNPACKED_1x1(1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_UNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0xffffffff, 0x00000000), UNPACKED_1x1(0.0, 1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_UNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0xffffffff), UNPACKED_1x1(0.0, 0.0, 1.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_UNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(0.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0xffffffff, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(1.0, 0.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0xffffffff, 0x00000000, 0x00000000), UNPACKED_1x1(0.0, 1.0, 0.0, 0.0)},
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0xffffffff, 0x00000000), UNPACKED_1x1(0.0, 0.0, 1.0, 0.0)},
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0xffffffff), UNPACKED_1x1(0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32A32_UNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), UNPACKED_1x1(1.0, 1.0, 1.0, 1.0)},
+
+   {PIPE_FORMAT_R32_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(       0.0,        0.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32_USCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x01000000), UNPACKED_1x1(16777216.0,        0.0,        0.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32_USCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1(       0.0,        0.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_USCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x01000000, 0x00000000), UNPACKED_1x1(16777216.0,        0.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_USCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x01000000), UNPACKED_1x1(       0.0, 16777216.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_USCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x01000000, 0x01000000), UNPACKED_1x1(16777216.0, 16777216.0,        0.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32B32_USCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(       0.0,        0.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_USCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x01000000, 0x00000000, 0x00000000), UNPACKED_1x1(16777216.0,        0.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_USCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x01000000, 0x00000000), UNPACKED_1x1(       0.0, 16777216.0,        0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_USCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x01000000), UNPACKED_1x1(       0.0,        0.0, 16777216.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_USCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x01000000, 0x01000000, 0x01000000), UNPACKED_1x1(16777216.0, 16777216.0, 16777216.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(       0.0,        0.0,        0.0,        0.0)},
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x01000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(16777216.0,        0.0,        0.0,        0.0)},
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x01000000, 0x00000000, 0x00000000), UNPACKED_1x1(       0.0, 16777216.0,        0.0,        0.0)},
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x01000000, 0x00000000), UNPACKED_1x1(       0.0,        0.0, 16777216.0,        0.0)},
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x01000000), UNPACKED_1x1(       0.0,        0.0,        0.0, 16777216.0)},
+   {PIPE_FORMAT_R32G32B32A32_USCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x01000000, 0x01000000, 0x01000000, 0x01000000), UNPACKED_1x1(16777216.0, 16777216.0, 16777216.0, 16777216.0)},
+
+   {PIPE_FORMAT_R32_SNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32_SNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x7fffffff), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32_SNORM, PACKED_1x32(0xffffffff), PACKED_1x32(0x80000001), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+
+   {PIPE_FORMAT_R32G32_SNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32_SNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x7fffffff, 0x00000000), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32_SNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x80000001, 0x00000000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32_SNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x7fffffff), UNPACKED_1x1(   0.0,    1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32_SNORM, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x80000001), UNPACKED_1x1(   0.0,   -1.0,    0.0,    1.0)},
+
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x7fffffff, 0x00000000, 0x00000000), UNPACKED_1x1(   1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x80000001, 0x00000000, 0x00000000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x7fffffff, 0x00000000), UNPACKED_1x1(   0.0,    1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x80000001, 0x00000000), UNPACKED_1x1(   0.0,   -1.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x7fffffff), UNPACKED_1x1(   0.0,    0.0,    1.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32_SNORM, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x80000001), UNPACKED_1x1(   0.0,    0.0,   -1.0,    1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(   0.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x7fffffff, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(   1.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x80000001, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(  -1.0,    0.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x7fffffff, 0x00000000, 0x00000000), UNPACKED_1x1(   0.0,    1.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x80000001, 0x00000000, 0x00000000), UNPACKED_1x1(   0.0,   -1.0,    0.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x7fffffff, 0x00000000), UNPACKED_1x1(   0.0,    0.0,    1.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x80000001, 0x00000000), UNPACKED_1x1(   0.0,    0.0,   -1.0,    0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x7fffffff), UNPACKED_1x1(   0.0,    0.0,    0.0,    1.0)},
+   {PIPE_FORMAT_R32G32B32A32_SNORM, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x80000001), UNPACKED_1x1(   0.0,    0.0,    0.0,   -1.0)},
+
+   {PIPE_FORMAT_R32_SSCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(        0.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32_SSCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0x01000000), UNPACKED_1x1( 16777216.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32_SSCALED, PACKED_1x32(0xffffffff), PACKED_1x32(0xff000000), UNPACKED_1x1(-16777216.0,         0.0,         0.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32_SSCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1(        0.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_SSCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x01000000, 0x00000000), UNPACKED_1x1( 16777216.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_SSCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0xff000000, 0x00000000), UNPACKED_1x1(-16777216.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_SSCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x01000000), UNPACKED_1x1(        0.0,  16777216.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32_SSCALED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0xff000000), UNPACKED_1x1(        0.0, -16777216.0,         0.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(        0.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x01000000, 0x00000000, 0x00000000), UNPACKED_1x1( 16777216.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0xff000000, 0x00000000, 0x00000000), UNPACKED_1x1(-16777216.0,         0.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x01000000, 0x00000000), UNPACKED_1x1(        0.0,  16777216.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0xff000000, 0x00000000), UNPACKED_1x1(        0.0, -16777216.0,         0.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x01000000), UNPACKED_1x1(        0.0,         0.0,  16777216.0,   1.0)},
+   {PIPE_FORMAT_R32G32B32_SSCALED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0xff000000), UNPACKED_1x1(        0.0,         0.0, -16777216.0,   1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(        0.0,         0.0,         0.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x01000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 16777216.0,         0.0,         0.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0xff000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(-16777216.0,         0.0,         0.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x01000000, 0x00000000, 0x00000000), UNPACKED_1x1(        0.0,  16777216.0,         0.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0xff000000, 0x00000000, 0x00000000), UNPACKED_1x1(        0.0, -16777216.0,         0.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x01000000, 0x00000000), UNPACKED_1x1(        0.0,         0.0,  16777216.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0xff000000, 0x00000000), UNPACKED_1x1(        0.0,         0.0, -16777216.0,         0.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x01000000), UNPACKED_1x1(        0.0,         0.0,         0.0,  16777216.0)},
+   {PIPE_FORMAT_R32G32B32A32_SSCALED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0xff000000), UNPACKED_1x1(        0.0,         0.0,         0.0, -16777216.0)},
+
+   /*
+    * Standard 32-bit float formats
+    */
+
+   {PIPE_FORMAT_R32_FLOAT, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(  0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32_FLOAT, PACKED_1x32(0xffffffff), PACKED_1x32(0x3f800000), UNPACKED_1x1(  1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32_FLOAT, PACKED_1x32(0xffffffff), PACKED_1x32(0xbf800000), UNPACKED_1x1( -1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x3f800000, 0x00000000), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0xbf800000, 0x00000000), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x3f800000), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0xbf800000), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FLOAT, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x3f800000, 0x3f800000), UNPACKED_1x1( 1.0,  1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x3f800000, 0x00000000, 0x00000000), UNPACKED_1x1( 1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0xbf800000, 0x00000000, 0x00000000), UNPACKED_1x1(-1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x3f800000, 0x00000000), UNPACKED_1x1( 0.0,  1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0xbf800000, 0x00000000), UNPACKED_1x1( 0.0, -1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x3f800000), UNPACKED_1x1( 0.0,  0.0,  1.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0xbf800000), UNPACKED_1x1( 0.0,  0.0, -1.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FLOAT, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x3f800000, 0x3f800000, 0x3f800000), UNPACKED_1x1( 1.0,  1.0,  1.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x3f800000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0xbf800000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(-1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x3f800000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0xbf800000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0, -1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x3f800000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  1.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0xbf800000, 0x00000000), UNPACKED_1x1( 0.0,  0.0, -1.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x3f800000), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0xbf800000), UNPACKED_1x1( 0.0,  0.0,  0.0, -1.0)},
+   {PIPE_FORMAT_R32G32B32A32_FLOAT, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000), UNPACKED_1x1( 1.0,  1.0,  1.0,  1.0)},
+
+   /*
+    * Half float formats
+    */
+
+   {PIPE_FORMAT_R16_FLOAT, PACKED_1x16(0xffff), PACKED_1x16(0x0000), UNPACKED_1x1(  0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16_FLOAT, PACKED_1x16(0xffff), PACKED_1x16(0x3c00), UNPACKED_1x1(  1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16_FLOAT, PACKED_1x16(0xffff), PACKED_1x16(0xbc00), UNPACKED_1x1( -1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x0000), UNPACKED_1x1( 0.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x3c00, 0x0000), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0xbc00, 0x0000), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0x3c00), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x0000, 0xbc00), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R16G16_FLOAT, PACKED_2x16(0xffff, 0xffff), PACKED_2x16(0x3c00, 0x3c00), UNPACKED_1x1( 1.0,  1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x0000), UNPACKED_1x1( 0.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x3c00, 0x0000, 0x0000), UNPACKED_1x1( 1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0xbc00, 0x0000, 0x0000), UNPACKED_1x1(-1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x3c00, 0x0000), UNPACKED_1x1( 0.0,  1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0xbc00, 0x0000), UNPACKED_1x1( 0.0, -1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0x3c00), UNPACKED_1x1( 0.0,  0.0,  1.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x0000, 0x0000, 0xbc00), UNPACKED_1x1( 0.0,  0.0, -1.0, 1.0)},
+   {PIPE_FORMAT_R16G16B16_FLOAT, PACKED_3x16(0xffff, 0xffff, 0xffff), PACKED_3x16(0x3c00, 0x3c00, 0x3c00), UNPACKED_1x1( 1.0,  1.0,  1.0, 1.0)},
+
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x0000), UNPACKED_1x1( 0.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x3c00, 0x0000, 0x0000, 0x0000), UNPACKED_1x1( 1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0xbc00, 0x0000, 0x0000, 0x0000), UNPACKED_1x1(-1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x3c00, 0x0000, 0x0000), UNPACKED_1x1( 0.0,  1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0xbc00, 0x0000, 0x0000), UNPACKED_1x1( 0.0, -1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x3c00, 0x0000), UNPACKED_1x1( 0.0,  0.0,  1.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0xbc00, 0x0000), UNPACKED_1x1( 0.0,  0.0, -1.0,  0.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0x3c00), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x0000, 0x0000, 0x0000, 0xbc00), UNPACKED_1x1( 0.0,  0.0,  0.0, -1.0)},
+   {PIPE_FORMAT_R16G16B16A16_FLOAT, PACKED_4x16(0xffff, 0xffff, 0xffff, 0xffff), PACKED_4x16(0x3c00, 0x3c00, 0x3c00, 0x3c00), UNPACKED_1x1( 1.0,  1.0,  1.0,  1.0)},
+
+   /*
+    * 32-bit fixed point formats
+    */
+
+   {PIPE_FORMAT_R32_FIXED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(  0.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32_FIXED, PACKED_1x32(0xffffffff), PACKED_1x32(0x00010000), UNPACKED_1x1(  1.0, 0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32_FIXED, PACKED_1x32(0xffffffff), PACKED_1x32(0xffff0000), UNPACKED_1x1( -1.0, 0.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00010000, 0x00000000), UNPACKED_1x1( 1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0xffff0000, 0x00000000), UNPACKED_1x1(-1.0,  0.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0x00010000), UNPACKED_1x1( 0.0,  1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00000000, 0xffff0000), UNPACKED_1x1( 0.0, -1.0, 0.0, 1.0)},
+   {PIPE_FORMAT_R32G32_FIXED, PACKED_2x32(0xffffffff, 0xffffffff), PACKED_2x32(0x00010000, 0x00010000), UNPACKED_1x1( 1.0,  1.0, 0.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00010000, 0x00000000, 0x00000000), UNPACKED_1x1( 1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0xffff0000, 0x00000000, 0x00000000), UNPACKED_1x1(-1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00010000, 0x00000000), UNPACKED_1x1( 0.0,  1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0xffff0000, 0x00000000), UNPACKED_1x1( 0.0, -1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0x00010000), UNPACKED_1x1( 0.0,  0.0,  1.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00000000, 0x00000000, 0xffff0000), UNPACKED_1x1( 0.0,  0.0, -1.0, 1.0)},
+   {PIPE_FORMAT_R32G32B32_FIXED, PACKED_3x32(0xffffffff, 0xffffffff, 0xffffffff), PACKED_3x32(0x00010000, 0x00010000, 0x00010000), UNPACKED_1x1( 1.0,  1.0,  1.0, 1.0)},
+
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00010000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1( 1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0xffff0000, 0x00000000, 0x00000000, 0x00000000), UNPACKED_1x1(-1.0,  0.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00010000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0,  1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0xffff0000, 0x00000000, 0x00000000), UNPACKED_1x1( 0.0, -1.0,  0.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00010000, 0x00000000), UNPACKED_1x1( 0.0,  0.0,  1.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0xffff0000, 0x00000000), UNPACKED_1x1( 0.0,  0.0, -1.0,  0.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0x00010000), UNPACKED_1x1( 0.0,  0.0,  0.0,  1.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00000000, 0x00000000, 0x00000000, 0xffff0000), UNPACKED_1x1( 0.0,  0.0,  0.0, -1.0)},
+   {PIPE_FORMAT_R32G32B32A32_FIXED, PACKED_4x32(0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff), PACKED_4x32(0x00010000, 0x00010000, 0x00010000, 0x00010000), UNPACKED_1x1( 1.0,  1.0,  1.0,  1.0)},
+
+   /*
+    * D3D9 specific vertex formats
+    */
+
+   {PIPE_FORMAT_R10G10B10X2_USCALED, PACKED_1x32(0x3fffffff), PACKED_1x32(0x00000000), UNPACKED_1x1(   0.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_USCALED, PACKED_1x32(0x3fffffff), PACKED_1x32(0x000003ff), UNPACKED_1x1(1023.0,    0.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_USCALED, PACKED_1x32(0x3fffffff), PACKED_1x32(0x000ffc00), UNPACKED_1x1(   0.0, 1023.0,    0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_USCALED, PACKED_1x32(0x3fffffff), PACKED_1x32(0x3ff00000), UNPACKED_1x1(   0.0,    0.0, 1023.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_USCALED, PACKED_1x32(0x3fffffff), PACKED_1x32(0x3fffffff), UNPACKED_1x1(1023.0, 1023.0, 1023.0, 1.0)},
+
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x00000000), UNPACKED_1x1( 0.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x000001ff), UNPACKED_1x1( 1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x00000201), UNPACKED_1x1(-1.0,  0.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x0007fc00), UNPACKED_1x1( 0.0,  1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x00080400), UNPACKED_1x1( 0.0, -1.0,  0.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x1ff00000), UNPACKED_1x1( 0.0,  0.0,  1.0, 1.0)},
+   {PIPE_FORMAT_R10G10B10X2_SNORM, PACKED_1x32(0x3fffffff), PACKED_1x32(0x20100000), UNPACKED_1x1( 0.0,  0.0, -1.0, 1.0)},
+
+   /*
+    * Special formats that not fit anywhere else
+    */
+
+};
+
+
+const unsigned util_format_nr_test_cases = Elements(util_format_test_cases);
diff --git a/src/gallium/auxiliary/util/u_timed_winsys.h b/src/gallium/auxiliary/util/u_format_tests.h
index 542365112d7..f59563f4f4f 100644
--- a/src/gallium/auxiliary/util/u_timed_winsys.h
+++ b/src/gallium/auxiliary/util/u_format_tests.h
@@ -1,8 +1,8 @@
 /**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
+ *
+ * Copyright 2010 VMware, Inc.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,32 +10,62 @@
  * 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 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
  * THE COPYRIGHT HOLDERS, AUTHORS 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 
+ * 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.
  *
  * The above copyright notice and this permission notice (including the
  * next paragraph) shall be included in all copies or substantial portions
  * of the Software.
- * 
- * 
+ *
  **************************************************************************/
-/*
- * Authors: Keith Whitwell <keithw-at-tungstengraphics-dot-com>
+
+
+#ifndef U_FORMAT_TESTS_H_
+#define U_FORMAT_TESTS_H_
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_format.h"
+
+
+#define UTIL_FORMAT_MAX_PACKED_BYTES 16
+#define UTIL_FORMAT_MAX_UNPACKED_WIDTH 4
+#define UTIL_FORMAT_MAX_UNPACKED_HEIGHT 4
+
+
+/**
+ * A (packed, unpacked) color pair.
  */
+struct util_format_test_case
+{
+   enum pipe_format format;
+
+   /**
+    * Mask of the bits that actually meaningful data. Used to mask out the
+    * "X" channels.
+    */
+   uint8_t mask[UTIL_FORMAT_MAX_PACKED_BYTES];
+
+   uint8_t packed[UTIL_FORMAT_MAX_PACKED_BYTES];
+
+   /**
+    * RGBA.
+    */
+   double unpacked[UTIL_FORMAT_MAX_UNPACKED_HEIGHT][UTIL_FORMAT_MAX_UNPACKED_WIDTH][4];
+};
 
 
-#ifndef U_TIMED_WINSYS_H
-#define U_TIMED_WINSYS_H
+extern const struct util_format_test_case
+util_format_test_cases[];
 
 
-struct pipe_winsys;
-struct pipe_winsys *u_timed_winsys_create( struct pipe_winsys *backend );
+extern const unsigned util_format_nr_test_cases;
 
 
-#endif
+#endif /* U_FORMAT_TESTS_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_yuv.c b/src/gallium/auxiliary/util/u_format_yuv.c
new file mode 100644
index 00000000000..64ea0b35347
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_yuv.c
@@ -0,0 +1,1182 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+/**
+ * @file
+ * YUV and RGB subsampled formats conversion.
+ *
+ * @author Jose Fonseca <[email protected]>
+ */
+
+
+#include "util/u_format_yuv.h"
+
+
+void
+util_format_r8g8_b8g8_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      float r, g0, g1, b;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r  = ubyte_to_float((value >>  0) & 0xff);
+         g0 = ubyte_to_float((value >>  8) & 0xff);
+         b  = ubyte_to_float((value >> 16) & 0xff);
+         g1 = ubyte_to_float((value >> 24) & 0xff);
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+
+         dst[0] = r;    /* r */
+         dst[1] = g1;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r  = ubyte_to_float((value >>  0) & 0xff);
+         g0 = ubyte_to_float((value >>  8) & 0xff);
+         b  = ubyte_to_float((value >> 16) & 0xff);
+         g1 = ubyte_to_float((value >> 24) & 0xff);
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_r8g8_b8g8_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                          const uint8_t *src_row, unsigned src_stride,
+                                          unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t r, g0, g1, b;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r  = (value >>  0) & 0xff;
+         g0 = (value >>  8) & 0xff;
+         b  = (value >> 16) & 0xff;
+         g1 = (value >> 24) & 0xff;
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+         dst += 4;
+
+         dst[0] = r;    /* r */
+         dst[1] = g1;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         r  = (value >>  0) & 0xff;
+         g0 = (value >>  8) & 0xff;
+         b  = (value >> 16) & 0xff;
+         g1 = (value >> 24) & 0xff;
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_r8g8_b8g8_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                       const float *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      float r, g0, g1, b;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         r  = 0.5f*(src[0] + src[4]);
+         g0 = src[1];
+         g1 = src[5];
+         b  = 0.5f*(src[2] + src[6]);
+
+         value  = float_to_ubyte(r);
+         value |= float_to_ubyte(g0) <<  8;
+         value |= float_to_ubyte(b)  << 16;
+         value |= float_to_ubyte(g1) << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         r  = src[0];
+         g0 = src[1];
+         g1 = 0;
+         b  = src[2];
+
+         value  = float_to_ubyte(r);
+         value |= float_to_ubyte(g0) <<  8;
+         value |= float_to_ubyte(b)  << 16;
+         value |= float_to_ubyte(g1) << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_r8g8_b8g8_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint32_t r, g0, g1, b;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         r  = (src[0] + src[4] + 1) >> 1;
+         g0 = src[1];
+         g1 = src[5];
+         b  = (src[2] + src[6] + 1) >> 1;
+
+         value  = r;
+         value |= g0 <<  8;
+         value |= b  << 16;
+         value |= g1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         r  = src[0];
+         g0 = src[1];
+         g1 = 0;
+         b  = src[2];
+
+         value  = r;
+         value |= g0 <<  8;
+         value |= b  << 16;
+         value |= g1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_r8g8_b8g8_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                        unsigned i, unsigned j)
+{
+   assert(i < 2);
+   assert(j < 1);
+
+   dst[0] = ubyte_to_float(src[0]);             /* r */
+   dst[1] = ubyte_to_float(src[1 + 2*i]);       /* g */
+   dst[2] = ubyte_to_float(src[2]);             /* b */
+   dst[3] = 1.0f;                               /* a */
+}
+
+
+void
+util_format_g8r8_g8b8_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      float r, g0, g1, b;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         g0 = ubyte_to_float((value >>  0) & 0xff);
+         r  = ubyte_to_float((value >>  8) & 0xff);
+         g1 = ubyte_to_float((value >> 16) & 0xff);
+         b  = ubyte_to_float((value >> 24) & 0xff);
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+
+         dst[0] = r;    /* r */
+         dst[1] = g1;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         g0 = ubyte_to_float((value >>  0) & 0xff);
+         r  = ubyte_to_float((value >>  8) & 0xff);
+         g1 = ubyte_to_float((value >> 16) & 0xff);
+         b  = ubyte_to_float((value >> 24) & 0xff);
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 1.0f; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_g8r8_g8b8_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                          const uint8_t *src_row, unsigned src_stride,
+                                          unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t r, g0, g1, b;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         g0 = (value >>  0) & 0xff;
+         r  = (value >>  8) & 0xff;
+         g1 = (value >> 16) & 0xff;
+         b  = (value >> 24) & 0xff;
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+         dst += 4;
+
+         dst[0] = r;    /* r */
+         dst[1] = g1;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         g0 = (value >>  0) & 0xff;
+         r  = (value >>  8) & 0xff;
+         g1 = (value >> 16) & 0xff;
+         b  = (value >> 24) & 0xff;
+
+         dst[0] = r;    /* r */
+         dst[1] = g0;   /* g */
+         dst[2] = b;    /* b */
+         dst[3] = 0xff; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_g8r8_g8b8_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                       const float *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      float r, g0, g1, b;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         r  = 0.5f*(src[0] + src[4]);
+         g0 = src[1];
+         g1 = src[5];
+         b  = 0.5f*(src[2] + src[6]);
+
+         value  = float_to_ubyte(g0);
+         value |= float_to_ubyte(r)  <<  8;
+         value |= float_to_ubyte(g1) << 16;
+         value |= float_to_ubyte(b)  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         r  = src[0];
+         g0 = src[1];
+         g1 = 0;
+         b  = src[2];
+
+         value  = float_to_ubyte(g0);
+         value |= float_to_ubyte(r)  <<  8;
+         value |= float_to_ubyte(g1) << 16;
+         value |= float_to_ubyte(b)  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_g8r8_g8b8_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint32_t r, g0, g1, b;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         r  = (src[0] + src[4] + 1) >> 1;
+         g0 = src[1];
+         g1 = src[5];
+         b  = (src[2] + src[6] + 1) >> 1;
+
+         value  = g0;
+         value |= r  <<  8;
+         value |= g1 << 16;
+         value |= b  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         r  = src[0];
+         g0 = src[1];
+         g1 = 0;
+         b  = src[2];
+
+         value  = g0;
+         value |= r  <<  8;
+         value |= g1 << 16;
+         value |= b  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_g8r8_g8b8_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                        unsigned i, unsigned j)
+{
+   assert(i < 2);
+   assert(j < 1);
+
+   dst[0] = ubyte_to_float(src[1]);             /* r */
+   dst[1] = ubyte_to_float(src[0 + 2*i]);       /* g */
+   dst[2] = ubyte_to_float(src[3]);             /* b */
+   dst[3] = 1.0f;                               /* a */
+}
+
+
+void
+util_format_uyvy_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                              const uint8_t *src_row, unsigned src_stride,
+                              unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t y0, y1, u, v;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         u  = (value >>  0) & 0xff;
+         y0 = (value >>  8) & 0xff;
+         v  = (value >> 16) & 0xff;
+         y1 = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_float(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+
+         util_format_yuv_to_rgb_float(y1, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         u  = (value >>  0) & 0xff;
+         y0 = (value >>  8) & 0xff;
+         v  = (value >> 16) & 0xff;
+         y1 = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_float(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_uyvy_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                               const uint8_t *src_row, unsigned src_stride,
+                               unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t y0, y1, u, v;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         u  = (value >>  0) & 0xff;
+         y0 = (value >>  8) & 0xff;
+         v  = (value >> 16) & 0xff;
+         y1 = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_8unorm(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+         dst += 4;
+
+         util_format_yuv_to_rgb_8unorm(y1, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         u  = (value >>  0) & 0xff;
+         y0 = (value >>  8) & 0xff;
+         v  = (value >> 16) & 0xff;
+         y1 = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_8unorm(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_uyvy_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                            const float *src_row, unsigned src_stride,
+                            unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint8_t y0, y1, u, v;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         uint8_t y0, y1, u0, u1, v0, v1, u, v;
+
+         util_format_rgb_float_to_yuv(src[0], src[1], src[2],
+                                      &y0, &u0, &v0);
+         util_format_rgb_float_to_yuv(src[4], src[5], src[6],
+                                      &y1, &u1, &v1);
+
+         u = (u0 + u1 + 1) >> 1;
+         v = (v0 + v1 + 1) >> 1;
+
+         value  = u;
+         value |= y0 <<  8;
+         value |= v  << 16;
+         value |= y1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         util_format_rgb_float_to_yuv(src[0], src[1], src[2],
+                                      &y0, &u, &v);
+         y1 = 0;
+
+         value  = u;
+         value |= y0 <<  8;
+         value |= v  << 16;
+         value |= y1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_uyvy_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint8_t y0, y1, u, v;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         uint8_t y0, y1, u0, u1, v0, v1, u, v;
+
+         util_format_rgb_8unorm_to_yuv(src[0], src[1], src[2],
+                                       &y0, &u0, &v0);
+         util_format_rgb_8unorm_to_yuv(src[4], src[5], src[6],
+                                       &y1, &u1, &v1);
+
+         u = (u0 + u1 + 1) >> 1;
+         v = (v0 + v1 + 1) >> 1;
+
+         value  = u;
+         value |= y0 <<  8;
+         value |= v  << 16;
+         value |= y1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         util_format_rgb_8unorm_to_yuv(src[0], src[1], src[2],
+                                       &y0, &u, &v);
+         y1 = 0;
+
+         value  = u;
+         value |= y0 <<  8;
+         value |= v  << 16;
+         value |= y1 << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_uyvy_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j)
+{
+   uint8_t y, u, v;
+
+   assert(i < 2);
+   assert(j < 1);
+
+   y = src[1 + i*2];
+   u = src[0];
+   v = src[2];
+
+   util_format_yuv_to_rgb_float(y, u, v, &dst[0], &dst[1], &dst[2]);
+
+   dst[3] = 1.0f;
+}
+
+
+void
+util_format_yuyv_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                              const uint8_t *src_row, unsigned src_stride,
+                              unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t y0, y1, u, v;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         y0 = (value >>  0) & 0xff;
+         u  = (value >>  8) & 0xff;
+         y1 = (value >> 16) & 0xff;
+         v  = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_float(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+
+         util_format_yuv_to_rgb_float(y1, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         y0 = (value >>  0) & 0xff;
+         u  = (value >>  8) & 0xff;
+         y1 = (value >> 16) & 0xff;
+         v  = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_float(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 1.0f; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_yuyv_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                               const uint8_t *src_row, unsigned src_stride,
+                               unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      uint32_t value;
+      uint8_t y0, y1, u, v;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         value = *src++;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         y0 = (value >>  0) & 0xff;
+         u  = (value >>  8) & 0xff;
+         y1 = (value >> 16) & 0xff;
+         v  = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_8unorm(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+         dst += 4;
+
+         util_format_yuv_to_rgb_8unorm(y1, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+         dst += 4;
+      }
+
+      if (x < width) {
+         value = *src;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         y0 = (value >>  0) & 0xff;
+         u  = (value >>  8) & 0xff;
+         y1 = (value >> 16) & 0xff;
+         v  = (value >> 24) & 0xff;
+
+         util_format_yuv_to_rgb_8unorm(y0, u, v, &dst[0], &dst[1], &dst[2]);
+         dst[3] = 0xff; /* a */
+      }
+
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+
+void
+util_format_yuyv_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                            const float *src_row, unsigned src_stride,
+                            unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint8_t y0, y1, u, v;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         uint8_t y0, y1, u0, u1, v0, v1, u, v;
+
+         util_format_rgb_float_to_yuv(src[0], src[1], src[2],
+                                      &y0, &u0, &v0);
+         util_format_rgb_float_to_yuv(src[4], src[5], src[6],
+                                      &y1, &u1, &v1);
+
+         u = (u0 + u1 + 1) >> 1;
+         v = (v0 + v1 + 1) >> 1;
+
+         value  = y0;
+         value |= u  <<  8;
+         value |= y1 << 16;
+         value |= v  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         util_format_rgb_float_to_yuv(src[0], src[1], src[2],
+                                      &y0, &u, &v);
+         y1 = 0;
+
+         value  = y0;
+         value |= u  <<  8;
+         value |= y1 << 16;
+         value |= v  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_yuyv_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height)
+{
+   unsigned x, y;
+
+   for (y = 0; y < height; y += 1) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      uint8_t y0, y1, u, v;
+      uint32_t value;
+
+      for (x = 0; x + 1 < width; x += 2) {
+         uint8_t y0, y1, u0, u1, v0, v1, u, v;
+
+         util_format_rgb_8unorm_to_yuv(src[0], src[1], src[2],
+                                       &y0, &u0, &v0);
+         util_format_rgb_8unorm_to_yuv(src[4], src[5], src[6],
+                                       &y1, &u1, &v1);
+
+         u = (u0 + u1 + 1) >> 1;
+         v = (v0 + v1 + 1) >> 1;
+
+         value  = y0;
+         value |= u  <<  8;
+         value |= y1 << 16;
+         value |= v  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst++ = value;
+
+         src += 8;
+      }
+
+      if (x < width) {
+         util_format_rgb_8unorm_to_yuv(src[0], src[1], src[2],
+                                       &y0, &u, &v);
+         y1 = 0;
+
+         value  = y0;
+         value |= u  <<  8;
+         value |= y1 << 16;
+         value |= v  << 24;
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+
+         *dst = value;
+      }
+
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+
+void
+util_format_yuyv_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j)
+{
+   uint8_t y, u, v;
+
+   assert(i < 2);
+   assert(j < 1);
+
+   y = src[0 + i*2];
+   u = src[1];
+   v = src[3];
+
+   util_format_yuv_to_rgb_float(y, u, v, &dst[0], &dst[1], &dst[2]);
+
+   dst[3] = 1.0f;
+}
+
+/* XXX: Stubbed for now */
+void
+util_format_yv12_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv12_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv12_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv12_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv12_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_yv16_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv16_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv16_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv16_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_yv16_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_iyuv_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_iyuv_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_iyuv_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_iyuv_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_iyuv_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_nv12_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv12_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv12_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv12_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv12_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_nv21_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv21_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv21_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv21_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_nv21_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_ia44_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ia44_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ia44_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ia44_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ia44_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
+void
+util_format_ai44_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ai44_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ai44_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ai44_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height) {}
+void
+util_format_ai44_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j) {}
diff --git a/src/gallium/auxiliary/util/u_format_yuv.h b/src/gallium/auxiliary/util/u_format_yuv.h
new file mode 100644
index 00000000000..9f2365a5266
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_yuv.h
@@ -0,0 +1,358 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+/**
+ * @file
+ * YUV colorspace conversion.
+ *
+ * @author Brian Paul <[email protected]>
+ * @author Michal Krol <[email protected]>
+ * @author Jose Fonseca <[email protected]>
+ *
+ * See also:
+ * - http://www.fourcc.org/fccyvrgb.php
+ * - http://msdn.microsoft.com/en-us/library/ms893078
+ * - http://en.wikipedia.org/wiki/YUV
+ */
+
+
+#ifndef U_FORMAT_YUV_H_
+#define U_FORMAT_YUV_H_
+
+
+#include "pipe/p_compiler.h"
+#include "u_math.h"
+
+
+/*
+ * TODO: Ensure we use consistent and right floating formulas, with enough
+ * precision in the coefficients.
+ */
+
+static INLINE void
+util_format_rgb_float_to_yuv(float r, float g, float b,
+                             uint8_t *y, uint8_t *u, uint8_t *v)
+{
+   const float _r = CLAMP(r, 0.0f, 1.0f);
+   const float _g = CLAMP(g, 0.0f, 1.0f);
+   const float _b = CLAMP(b, 0.0f, 1.0f);
+
+   const float scale = 255.0f;
+
+   const int _y = scale * ( (0.257f * _r) + (0.504f * _g) + (0.098f * _b));
+   const int _u = scale * (-(0.148f * _r) - (0.291f * _g) + (0.439f * _b));
+   const int _v = scale * ( (0.439f * _r) - (0.368f * _g) - (0.071f * _b));
+
+   *y = _y + 16; 
+   *u = _u + 128; 
+   *v = _v + 128; 
+}
+
+
+static INLINE void
+util_format_yuv_to_rgb_float(uint8_t y, uint8_t u, uint8_t v,
+                             float *r, float *g, float *b)
+{
+   const int _y = y - 16;
+   const int _u = u - 128;
+   const int _v = v - 128;
+
+   const float y_factor = 255.0f / 219.0f;
+
+   const float scale = 1.0f / 255.0f;
+
+   *r = scale * (y_factor * _y               + 1.596f * _v);
+   *g = scale * (y_factor * _y - 0.391f * _u - 0.813f * _v);
+   *b = scale * (y_factor * _y + 2.018f * _u              );
+}
+
+
+static INLINE void
+util_format_rgb_8unorm_to_yuv(uint8_t r, uint8_t g, uint8_t b,
+                	      uint8_t *y, uint8_t *u, uint8_t *v)
+{
+   *y = ((  66 * r + 129 * g +  25 * b + 128) >> 8) +  16;
+   *u = (( -38 * r -  74 * g + 112 * b + 128) >> 8) + 128;
+   *v = (( 112 * r -  94 * g -  18 * b + 128) >> 8) + 128;
+}
+
+
+static INLINE void
+util_format_yuv_to_rgb_8unorm(uint8_t y, uint8_t u, uint8_t v,
+                              uint8_t *r, uint8_t *g, uint8_t *b)
+{
+   const int _y = y - 16;
+   const int _u = u - 128;
+   const int _v = v - 128;
+
+   const int _r = (298 * _y            + 409 * _v + 128) >> 8;
+   const int _g = (298 * _y - 100 * _u - 208 * _v + 128) >> 8;
+   const int _b = (298 * _y + 516 * _u            + 128) >> 8;
+
+   *r = CLAMP(_r, 0, 255);
+   *g = CLAMP(_g, 0, 255);
+   *b = CLAMP(_b, 0, 255);
+}
+
+
+
+void
+util_format_uyvy_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                              const uint8_t *src_row, unsigned src_stride,
+                              unsigned width, unsigned height);
+
+void
+util_format_uyvy_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                               const uint8_t *src_row, unsigned src_stride,
+                               unsigned width, unsigned height);
+
+void
+util_format_uyvy_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                            const float *src_row, unsigned src_stride,
+                            unsigned width, unsigned height);
+
+void
+util_format_uyvy_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+
+void
+util_format_uyvy_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+
+void
+util_format_yuyv_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                              const uint8_t *src_row, unsigned src_stride,
+                              unsigned width, unsigned height);
+
+void
+util_format_yuyv_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                               const uint8_t *src_row, unsigned src_stride,
+                               unsigned width, unsigned height);
+
+void
+util_format_yuyv_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                            const float *src_row, unsigned src_stride,
+                            unsigned width, unsigned height);
+
+void
+util_format_yuyv_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+
+void
+util_format_yuyv_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+
+/* XXX: Stubbed for now */
+void
+util_format_yv12_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv12_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv12_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv12_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv12_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_yv16_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv16_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv16_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv16_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_yv16_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_iyuv_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_iyuv_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_iyuv_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_iyuv_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_iyuv_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_nv12_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv12_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv12_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv12_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv12_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_nv21_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv21_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv21_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv21_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_nv21_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_ia44_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ia44_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ia44_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ia44_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ia44_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+void
+util_format_ai44_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ai44_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ai44_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                             const uint8_t *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ai44_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                             const float *src_row, unsigned src_stride,
+                             unsigned width, unsigned height);
+void
+util_format_ai44_fetch_rgba_float(float *dst, const uint8_t *src,
+                             unsigned i, unsigned j);
+
+
+void
+util_format_r8g8_b8g8_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height);
+
+void
+util_format_r8g8_b8g8_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                          const uint8_t *src_row, unsigned src_stride,
+                                          unsigned width, unsigned height);
+
+void
+util_format_r8g8_b8g8_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                       const float *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height);
+
+void
+util_format_r8g8_b8g8_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height);
+
+void
+util_format_r8g8_b8g8_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                        unsigned i, unsigned j);
+
+void
+util_format_g8r8_g8b8_unorm_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height);
+
+void
+util_format_g8r8_g8b8_unorm_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                          const uint8_t *src_row, unsigned src_stride,
+                                          unsigned width, unsigned height);
+
+void
+util_format_g8r8_g8b8_unorm_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                       const float *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height);
+
+void
+util_format_g8r8_g8b8_unorm_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height);
+
+void
+util_format_g8r8_g8b8_unorm_fetch_rgba_float(float *dst, const uint8_t *src,
+                                        unsigned i, unsigned j);
+
+
+
+#endif /* U_FORMAT_YUV_H_ */
diff --git a/src/gallium/auxiliary/util/u_format_zs.c b/src/gallium/auxiliary/util/u_format_zs.c
new file mode 100644
index 00000000000..792d69c214c
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_zs.c
@@ -0,0 +1,920 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#include "u_debug.h"
+#include "u_math.h"
+#include "u_format_zs.h"
+
+
+/*
+ * z32_unorm conversion functions
+ */
+
+static INLINE uint16_t
+z32_unorm_to_z16_unorm(uint32_t z)
+{
+   /* z * 0xffff / 0xffffffff */
+   return z >> 16;
+}
+
+static INLINE uint32_t
+z16_unorm_to_z32_unorm(uint16_t z)
+{
+   /* z * 0xffffffff / 0xffff */
+   return (z << 16) | z;
+}
+
+static INLINE uint32_t
+z32_unorm_to_z24_unorm(uint32_t z)
+{
+   /* z * 0xffffff / 0xffffffff */
+   return z >> 8;
+}
+
+static INLINE uint32_t
+z24_unorm_to_z32_unorm(uint32_t z)
+{
+   /* z * 0xffffffff / 0xffffff */
+   return (z << 8) | (z >> 16);
+}
+
+
+/*
+ * z32_float conversion functions
+ */
+
+static INLINE uint16_t
+z32_float_to_z16_unorm(float z)
+{
+   const float scale = 0xffff;
+   return (uint16_t)(z * scale);
+}
+
+static INLINE float
+z16_unorm_to_z32_float(uint16_t z)
+{
+   const float scale = 1.0 / 0xffff;
+   return (float)(z * scale);
+}
+
+static INLINE uint32_t
+z32_float_to_z24_unorm(float z)
+{
+   const double scale = 0xffffff;
+   return (uint32_t)(z * scale) & 0xffffff;
+}
+
+static INLINE float
+z24_unorm_to_z32_float(uint32_t z)
+{
+   const double scale = 1.0 / 0xffffff;
+   return (float)(z * scale);
+}
+
+static INLINE uint32_t
+z32_float_to_z32_unorm(float z)
+{
+   const double scale = 0xffffffff;
+   return (uint32_t)(z * scale);
+}
+
+static INLINE float
+z32_unorm_to_z32_float(uint32_t z)
+{
+   const double scale = 1.0 / 0xffffffff;
+   return (float)(z * scale);
+}
+
+
+void
+util_format_s8_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_s8_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z16_unorm_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                     const uint8_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint16_t *src = (const uint16_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint16_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap16(value);
+#endif
+         *dst++ = z16_unorm_to_z32_float(value);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z16_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                   const float *src_row, unsigned src_stride,
+                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint16_t *dst = (uint16_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint16_t value;
+         value = z32_float_to_z16_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap16(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z16_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const uint16_t *src = (const uint16_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint16_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap16(value);
+#endif
+         *dst++ = z16_unorm_to_z32_unorm(value);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z16_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                     const uint32_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      uint16_t *dst = (uint16_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint16_t value;
+         value = z32_unorm_to_z16_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap16(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z32_unorm_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                     const uint8_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z32_unorm_to_z32_float(value);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                   const float *src_row, unsigned src_stride,
+                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value;
+         value = z32_float_to_z32_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z32_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width * 4);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                     const uint32_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width * 4);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                     const uint8_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width * 4);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                   const float *src_row, unsigned src_stride,
+                                   unsigned width, unsigned height)
+{
+   unsigned y;
+   for(y = 0; y < height; ++y) {
+      memcpy(dst_row, src_row, width * 4);
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const float *src = (const float *)src_row;
+      for(x = 0; x < width; ++x) {
+         *dst++ = z32_float_to_z32_unorm(*src++);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                     const uint32_t *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      float *dst = (float *)dst_row;
+      for(x = 0; x < width; ++x) {
+         *dst++ = z32_unorm_to_z32_float(*src++);
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                                const uint8_t *src_row, unsigned src_stride,
+                                                unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_float(value & 0xffffff);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                              const float *src_row, unsigned src_stride,
+                                              unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0xff000000;
+         value |= z32_float_to_z24_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                                  const uint8_t *src_row, unsigned src_stride,
+                                                  unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_unorm(value & 0xffffff);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                                const uint32_t *src_row, unsigned src_stride,
+                                                unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value= *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0xff000000;
+         value |= z32_unorm_to_z24_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                   const uint8_t *src_row, unsigned src_stride,
+                                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value >> 24;
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z24_unorm_s8_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                 const uint8_t *src_row, unsigned src_stride,
+                                                 unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0x00ffffff;
+         value |= *src++ << 24;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                                const uint8_t *src_row, unsigned src_stride,
+                                                unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_float(value >> 8);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                              const float *src_row, unsigned src_stride,
+                                              unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0x000000ff;
+         value |= z32_float_to_z24_unorm(*src++) << 8;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                                  const uint8_t *src_row, unsigned src_stride,
+                                                  unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_unorm(value >> 8);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                                const uint32_t *src_row, unsigned src_stride,
+                                                unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0x000000ff;
+         value |= *src++ & 0xffffff00;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                   const uint8_t *src_row, unsigned src_stride,
+                                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint8_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value & 0xff;
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_s8_uscaled_z24_unorm_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                 const uint8_t *src_row, unsigned src_stride,
+                                                 unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint8_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *dst;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         value &= 0xffffff00;
+         value |= *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z24x8_unorm_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_float(value & 0xffffff);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z24x8_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                     const float *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value;
+         value = z32_float_to_z24_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z24x8_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_unorm(value & 0xffffff);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z24x8_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint32_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value;
+         value = z32_unorm_to_z24_unorm(*src++);
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_x8z24_unorm_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const uint32_t *src = (uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_float(value >> 8);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_x8z24_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                     const float *src_row, unsigned src_stride,
+                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value;
+         value = z32_float_to_z24_unorm(*src++) << 8;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_x8z24_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const uint32_t *src = (const uint32_t *)src_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value = *src++;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = z24_unorm_to_z32_unorm(value >> 8);
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_x8z24_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint32_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      uint32_t *dst = (uint32_t *)dst_row;
+      for(x = 0; x < width; ++x) {
+         uint32_t value;
+         value = z32_unorm_to_z24_unorm(*src++) << 8;
+#ifdef PIPE_ARCH_BIG_ENDIAN
+         value = util_bswap32(value);
+#endif
+         *dst++ = value;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_z_float(float *dst_row, unsigned dst_stride,
+                                                   const uint8_t *src_row, unsigned src_stride,
+                                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      float *dst = dst_row;
+      const float *src = (const float *)src_row;
+      for(x = 0; x < width; ++x) {
+         *dst = *src;
+         src += 2;
+         dst += 1;
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_pack_z_float(uint8_t *dst_row, unsigned dst_stride,
+                                                 const float *src_row, unsigned src_stride,
+                                                 unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const float *src = src_row;
+      float *dst = (float *)dst_row;
+      for(x = 0; x < width; ++x) {
+         *dst = *src;
+         src += 1;
+         dst += 2;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride,
+                                                     const uint8_t *src_row, unsigned src_stride,
+                                                     unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint32_t *dst = dst_row;
+      const float *src = (const float *)src_row;
+      for(x = 0; x < width; ++x) {
+         *dst = z32_float_to_z32_unorm(*src);
+         src += 2;
+         dst += 1;
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride,
+                                                   const uint32_t *src_row, unsigned src_stride,
+                                                   unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint32_t *src = src_row;
+      float *dst = (float *)dst_row;
+      for(x = 0; x < width; ++x) {
+         *dst++ = z32_unorm_to_z32_float(*src++);
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                      const uint8_t *src_row, unsigned src_stride,
+                                                      unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      uint8_t *dst = dst_row;
+      const uint8_t *src = src_row + 4;
+      for(x = 0; x < width; ++x) {
+         *dst = *src;
+         src += 8;
+         dst += 1;
+      }
+      src_row += src_stride/sizeof(*src_row);
+      dst_row += dst_stride/sizeof(*dst_row);
+   }
+}
+
+void
+util_format_z32_float_s8x24_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride,
+                                                    const uint8_t *src_row, unsigned src_stride,
+                                                    unsigned width, unsigned height)
+{
+   unsigned x, y;
+   for(y = 0; y < height; ++y) {
+      const uint8_t *src = src_row;
+      uint8_t *dst = dst_row + 4;
+      for(x = 0; x < width; ++x) {
+         *dst = *src;
+         src += 1;
+         dst += 8;
+      }
+      dst_row += dst_stride/sizeof(*dst_row);
+      src_row += src_stride/sizeof(*src_row);
+   }
+}
+
diff --git a/src/gallium/auxiliary/util/u_format_zs.h b/src/gallium/auxiliary/util/u_format_zs.h
new file mode 100644
index 00000000000..650db4b95fd
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_format_zs.h
@@ -0,0 +1,196 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE 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
+ * THE COPYRIGHT HOLDERS, AUTHORS 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.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#ifndef U_FORMAT_ZS_H_
+#define U_FORMAT_ZS_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+void
+util_format_s8_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z16_unorm_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z16_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z16_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z16_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_unorm_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24_unorm_s8_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_s8_uscaled_z24_unorm_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24x8_unorm_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24x8_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24x8_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z24x8_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_x8z24_unorm_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_x8z24_unorm_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_x8z24_unorm_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_x8z24_unorm_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_z_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_pack_z_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_z_32unorm(uint32_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_pack_z_32unorm(uint8_t *dst_row, unsigned dst_stride, const uint32_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_unpack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+void
+util_format_z32_float_s8x24_uscaled_pack_s_8uscaled(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height);
+
+
+#endif /* U_FORMAT_ZS_H_ */
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c
index fc027e48e4e..eee6030ddcc 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.c
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.c
@@ -62,11 +62,12 @@ struct gen_mipmap_state
    struct pipe_rasterizer_state rasterizer;
    struct pipe_sampler_state sampler;
    struct pipe_clip_state clip;
+   struct pipe_vertex_element velem[2];
 
    void *vs;
    void *fs2d, *fsCube;
 
-   struct pipe_buffer *vbuf;  /**< quad vertices */
+   struct pipe_resource *vbuf;  /**< quad vertices */
    unsigned vbuf_slot;
 
    float vertices[4][2][4];   /**< vertex/texcoords for quad */
@@ -937,6 +938,7 @@ format_to_type_comps(enum pipe_format pformat,
       *datatype = DTYPE_UBYTE;
       *comps = 4;
       return;
+   case PIPE_FORMAT_B5G5R5X1_UNORM:
    case PIPE_FORMAT_B5G5R5A1_UNORM:
       *datatype = DTYPE_USHORT_1_5_5_5_REV;
       *comps = 4;
@@ -1114,11 +1116,10 @@ reduce_3d(enum pipe_format pformat,
 
 static void
 make_1d_mipmap(struct gen_mipmap_state *ctx,
-               struct pipe_texture *pt,
+               struct pipe_resource *pt,
                uint face, uint baseLevel, uint lastLevel)
 {
    struct pipe_context *pipe = ctx->pipe;
-   struct pipe_screen *screen = pipe->screen;
    const uint zslice = 0;
    uint dstLevel;
 
@@ -1127,38 +1128,37 @@ make_1d_mipmap(struct gen_mipmap_state *ctx,
       struct pipe_transfer *srcTrans, *dstTrans;
       void *srcMap, *dstMap;
       
-      srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
+      srcTrans = pipe_get_transfer(pipe, pt, face, srcLevel, zslice,
                                           PIPE_TRANSFER_READ, 0, 0,
                                           u_minify(pt->width0, srcLevel),
                                           u_minify(pt->height0, srcLevel));
-      dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
+      dstTrans = pipe_get_transfer(pipe, pt, face, dstLevel, zslice,
                                           PIPE_TRANSFER_WRITE, 0, 0,
                                           u_minify(pt->width0, dstLevel),
                                           u_minify(pt->height0, dstLevel));
 
-      srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
-      dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
+      srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans);
+      dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans);
 
       reduce_1d(pt->format,
-                srcTrans->width, srcMap,
-                dstTrans->width, dstMap);
+                srcTrans->box.width, srcMap,
+                dstTrans->box.width, dstMap);
 
-      screen->transfer_unmap(screen, srcTrans);
-      screen->transfer_unmap(screen, dstTrans);
+      pipe->transfer_unmap(pipe, srcTrans);
+      pipe->transfer_unmap(pipe, dstTrans);
 
-      screen->tex_transfer_destroy(srcTrans);
-      screen->tex_transfer_destroy(dstTrans);
+      pipe->transfer_destroy(pipe, srcTrans);
+      pipe->transfer_destroy(pipe, dstTrans);
    }
 }
 
 
 static void
 make_2d_mipmap(struct gen_mipmap_state *ctx,
-               struct pipe_texture *pt,
+               struct pipe_resource *pt,
                uint face, uint baseLevel, uint lastLevel)
 {
    struct pipe_context *pipe = ctx->pipe;
-   struct pipe_screen *screen = pipe->screen;
    const uint zslice = 0;
    uint dstLevel;
    
@@ -1170,36 +1170,36 @@ make_2d_mipmap(struct gen_mipmap_state *ctx,
       struct pipe_transfer *srcTrans, *dstTrans;
       ubyte *srcMap, *dstMap;
       
-      srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
-                                          PIPE_TRANSFER_READ, 0, 0,
-                                          u_minify(pt->width0, srcLevel),
-                                          u_minify(pt->height0, srcLevel));
-      dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
-                                          PIPE_TRANSFER_WRITE, 0, 0,
-                                          u_minify(pt->width0, dstLevel),
-                                          u_minify(pt->height0, dstLevel));
-
-      srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
-      dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
+      srcTrans = pipe_get_transfer(pipe, pt, face, srcLevel, zslice,
+				   PIPE_TRANSFER_READ, 0, 0,
+				   u_minify(pt->width0, srcLevel),
+				   u_minify(pt->height0, srcLevel));
+      dstTrans = pipe_get_transfer(pipe, pt, face, dstLevel, zslice,
+				   PIPE_TRANSFER_WRITE, 0, 0,
+				   u_minify(pt->width0, dstLevel),
+				   u_minify(pt->height0, dstLevel));
+
+      srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans);
+      dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans);
 
       reduce_2d(pt->format,
-                srcTrans->width, srcTrans->height,
+                srcTrans->box.width, srcTrans->box.height,
                 srcTrans->stride, srcMap,
-                dstTrans->width, dstTrans->height,
+                dstTrans->box.width, dstTrans->box.height,
                 dstTrans->stride, dstMap);
 
-      screen->transfer_unmap(screen, srcTrans);
-      screen->transfer_unmap(screen, dstTrans);
+      pipe->transfer_unmap(pipe, srcTrans);
+      pipe->transfer_unmap(pipe, dstTrans);
 
-      screen->tex_transfer_destroy(srcTrans);
-      screen->tex_transfer_destroy(dstTrans);
+      pipe->transfer_destroy(pipe, srcTrans);
+      pipe->transfer_destroy(pipe, dstTrans);
    }
 }
 
 
 static void
 make_3d_mipmap(struct gen_mipmap_state *ctx,
-               struct pipe_texture *pt,
+               struct pipe_resource *pt,
                uint face, uint baseLevel, uint lastLevel)
 {
 #if 0
@@ -1215,17 +1215,17 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
       struct pipe_transfer *srcTrans, *dstTrans;
       ubyte *srcMap, *dstMap;
       
-      srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
+      srcTrans = pipe->get_transfer(pipe, pt, face, srcLevel, zslice,
                                           PIPE_TRANSFER_READ, 0, 0,
                                           u_minify(pt->width0, srcLevel),
                                           u_minify(pt->height0, srcLevel));
-      dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
+      dstTrans = pipe->get_transfer(pipe, pt, face, dstLevel, zslice,
                                           PIPE_TRANSFER_WRITE, 0, 0,
                                           u_minify(pt->width0, dstLevel),
                                           u_minify(pt->height0, dstLevel));
 
-      srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
-      dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
+      srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans);
+      dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans);
 
       reduce_3d(pt->format,
                 srcTrans->width, srcTrans->height,
@@ -1233,11 +1233,11 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
                 dstTrans->width, dstTrans->height,
                 dstTrans->stride, dstMap);
 
-      screen->transfer_unmap(screen, srcTrans);
-      screen->transfer_unmap(screen, dstTrans);
+      pipe->transfer_unmap(pipe, srcTrans);
+      pipe->transfer_unmap(pipe, dstTrans);
 
-      screen->tex_transfer_destroy(srcTrans);
-      screen->tex_transfer_destroy(dstTrans);
+      pipe->transfer_destroy(pipe, srcTrans);
+      pipe->transfer_destroy(pipe, dstTrans);
    }
 #else
    (void) reduce_3d;
@@ -1247,7 +1247,7 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
 
 static void
 fallback_gen_mipmap(struct gen_mipmap_state *ctx,
-                    struct pipe_texture *pt,
+                    struct pipe_resource *pt,
                     uint face, uint baseLevel, uint lastLevel)
 {
    switch (pt->target) {
@@ -1307,6 +1307,15 @@ util_create_gen_mipmap(struct pipe_context *pipe,
    ctx->sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NEAREST;
    ctx->sampler.normalized_coords = 1;
 
+   /* vertex elements state */
+   memset(&ctx->velem[0], 0, sizeof(ctx->velem[0]) * 2);
+   for (i = 0; i < 2; i++) {
+      ctx->velem[i].src_offset = i * 4 * sizeof(float);
+      ctx->velem[i].instance_divisor = 0;
+      ctx->velem[i].vertex_buffer_index = 0;
+      ctx->velem[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+   }
+
    /* vertex shader - still needed to specify mapping from fragment
     * shader input semantics to vertex elements 
     */
@@ -1349,8 +1358,7 @@ get_next_slot(struct gen_mipmap_state *ctx)
 
    if (!ctx->vbuf) {
       ctx->vbuf = pipe_buffer_create(ctx->pipe->screen,
-                                     32,
-                                     PIPE_BUFFER_USAGE_VERTEX,
+                                     PIPE_BIND_VERTEX_BUFFER,
                                      max_slots * sizeof ctx->vertices);
    }
    
@@ -1411,7 +1419,7 @@ set_vertex_data(struct gen_mipmap_state *ctx,
 
    offset = get_next_slot( ctx );
 
-   pipe_buffer_write_nooverlap(ctx->pipe->screen, ctx->vbuf,
+   pipe_buffer_write_nooverlap(ctx->pipe, ctx->vbuf,
                                offset, sizeof(ctx->vertices), ctx->vertices);
 
    return offset;
@@ -1431,7 +1439,7 @@ util_destroy_gen_mipmap(struct gen_mipmap_state *ctx)
    pipe->delete_fs_state(pipe, ctx->fs2d);
    pipe->delete_fs_state(pipe, ctx->fsCube);
 
-   pipe_buffer_reference(&ctx->vbuf, NULL);
+   pipe_resource_reference(&ctx->vbuf, NULL);
 
    FREE(ctx);
 }
@@ -1443,7 +1451,7 @@ util_destroy_gen_mipmap(struct gen_mipmap_state *ctx)
  */
 void util_gen_mipmap_flush( struct gen_mipmap_state *ctx )
 {
-   pipe_buffer_reference(&ctx->vbuf, NULL);
+   pipe_resource_reference(&ctx->vbuf, NULL);
    ctx->vbuf_slot = 0;
 } 
 
@@ -1452,7 +1460,7 @@ void util_gen_mipmap_flush( struct gen_mipmap_state *ctx )
  * Generate mipmap images.  It's assumed all needed texture memory is
  * already allocated.
  *
- * \param pt  the texture to generate mipmap levels for
+ * \param psv  the sampler view to the texture to generate mipmap levels for
  * \param face  which cube face to generate mipmaps for (0 for non-cube maps)
  * \param baseLevel  the first mipmap level to use as a src
  * \param lastLevel  the last mipmap level to generate
@@ -1461,12 +1469,13 @@ void util_gen_mipmap_flush( struct gen_mipmap_state *ctx )
  */
 void
 util_gen_mipmap(struct gen_mipmap_state *ctx,
-                struct pipe_texture *pt,
+                struct pipe_sampler_view *psv,
                 uint face, uint baseLevel, uint lastLevel, uint filter)
 {
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
    struct pipe_framebuffer_state fb;
+   struct pipe_resource *pt = psv->texture;
    void *fs = (pt->target == PIPE_TEXTURE_CUBE) ? ctx->fsCube : ctx->fs2d;
    uint dstLevel;
    uint zslice = 0;
@@ -1484,8 +1493,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
           filter == PIPE_TEX_FILTER_NEAREST);
 
    /* check if we can render in the texture's format */
-   if (!screen->is_format_supported(screen, pt->format, PIPE_TEXTURE_2D,
-                                    PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) {
+   if (!screen->is_format_supported(screen, psv->format, PIPE_TEXTURE_2D,
+                                    PIPE_BIND_RENDER_TARGET, 0)) {
       fallback_gen_mipmap(ctx, pt, face, baseLevel, lastLevel);
       return;
    }
@@ -1495,18 +1504,20 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
    cso_save_depth_stencil_alpha(ctx->cso);
    cso_save_rasterizer(ctx->cso);
    cso_save_samplers(ctx->cso);
-   cso_save_sampler_textures(ctx->cso);
+   cso_save_fragment_sampler_views(ctx->cso);
    cso_save_framebuffer(ctx->cso);
    cso_save_fragment_shader(ctx->cso);
    cso_save_vertex_shader(ctx->cso);
    cso_save_viewport(ctx->cso);
    cso_save_clip(ctx->cso);
+   cso_save_vertex_elements(ctx->cso);
 
    /* bind our state */
    cso_set_blend(ctx->cso, &ctx->blend);
    cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
    cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
    cso_set_clip(ctx->cso, &ctx->clip);
+   cso_set_vertex_elements(ctx->cso, 2, ctx->velem);
 
    cso_set_fragment_shader_handle(ctx->cso, fs);
    cso_set_vertex_shader_handle(ctx->cso, ctx->vs);
@@ -1529,7 +1540,7 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
 
       struct pipe_surface *surf = 
          screen->get_tex_surface(screen, pt, face, dstLevel, zslice,
-                                 PIPE_BUFFER_USAGE_GPU_WRITE);
+                                 PIPE_BIND_RENDER_TARGET);
 
       /*
        * Setup framebuffer / dest surface
@@ -1562,7 +1573,7 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
       cso_single_sampler(ctx->cso, 0, &ctx->sampler);
       cso_single_sampler_done(ctx->cso);
 
-      cso_set_sampler_textures(ctx->cso, 1, &pt);
+      cso_set_fragment_sampler_views(ctx->cso, 1, &psv);
 
       /* quad coords in clip coords */
       offset = set_vertex_data(ctx,
@@ -1587,10 +1598,11 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
    cso_restore_depth_stencil_alpha(ctx->cso);
    cso_restore_rasterizer(ctx->cso);
    cso_restore_samplers(ctx->cso);
-   cso_restore_sampler_textures(ctx->cso);
+   cso_restore_fragment_sampler_views(ctx->cso);
    cso_restore_framebuffer(ctx->cso);
    cso_restore_fragment_shader(ctx->cso);
    cso_restore_vertex_shader(ctx->cso);
    cso_restore_viewport(ctx->cso);
    cso_restore_clip(ctx->cso);
+   cso_restore_vertex_elements(ctx->cso);
 }
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.h b/src/gallium/auxiliary/util/u_gen_mipmap.h
index 54608f9466d..a7502b9982b 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.h
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.h
@@ -37,7 +37,7 @@ extern "C" {
 
    
 struct pipe_context;
-struct pipe_texture;
+struct pipe_resource;
 struct cso_context;
 
 struct gen_mipmap_state;
@@ -59,7 +59,7 @@ util_gen_mipmap_flush( struct gen_mipmap_state *ctx );
 
 extern void
 util_gen_mipmap(struct gen_mipmap_state *ctx,
-                struct pipe_texture *pt,
+                struct pipe_sampler_view *psv,
                 uint face, uint baseLevel, uint lastLevel, uint filter);
 
 
diff --git a/src/gallium/auxiliary/util/u_half.h b/src/gallium/auxiliary/util/u_half.h
new file mode 100644
index 00000000000..ad030e90c67
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_half.h
@@ -0,0 +1,90 @@
+/**************************************************************************
+ *
+ * Copyright 2010 Luca Barbieri
+ *
+ * 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 (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.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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 U_HALF_H
+#define U_HALF_H
+
+#include "pipe/p_compiler.h"
+#include "util/u_math.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+extern const uint32_t util_half_to_float_mantissa_table[2048];
+extern const uint32_t util_half_to_float_exponent_table[64];
+extern const uint32_t util_half_to_float_offset_table[64];
+extern const uint16_t util_float_to_half_base_table[512];
+extern const uint8_t util_float_to_half_shift_table[512];
+
+/*
+ * Note that if the half float is a signaling NaN, the x87 FPU will turn
+ * it into a quiet NaN immediately upon loading into a float.
+ *
+ * Additionally, denormals may be flushed to zero.
+ *
+ * To avoid this, use the floatui functions instead of the float ones
+ * when just doing conversion rather than computation on the resulting
+ * floats.
+ */
+
+static INLINE uint32_t
+util_half_to_floatui(uint16_t h)
+{
+   unsigned exp = h >> 10;
+   return util_half_to_float_mantissa_table[util_half_to_float_offset_table[exp] + (h & 0x3ff)] + util_half_to_float_exponent_table[exp];
+}
+
+static INLINE float
+util_half_to_float(uint16_t h)
+{
+   union fi r;
+   r.ui = util_half_to_floatui(h);
+   return r.f;
+}
+
+static INLINE uint16_t
+util_floatui_to_half(uint32_t v)
+{
+   unsigned signexp = v >> 23;
+   return util_float_to_half_base_table[signexp] + ((v & 0x007fffff) >> util_float_to_half_shift_table[signexp]);
+}
+
+static INLINE uint16_t
+util_float_to_half(float f)
+{
+   union fi i;
+   i.f = f;
+   return util_floatui_to_half(i.ui);
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* U_HALF_H */
+
diff --git a/src/gallium/auxiliary/util/u_half.py b/src/gallium/auxiliary/util/u_half.py
new file mode 100644
index 00000000000..8007482e971
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_half.py
@@ -0,0 +1,179 @@
+# Copyright 2010 Luca Barbieri
+#
+# 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 (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.
+# IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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.
+#
+# *************************************************************************
+
+# The code is a reimplementation of the algorithm in
+#  www.fox-toolkit.org/ftp/fasthalffloatconversion.pdf
+# "Fast Half Float Conversions" by Jeroen van der Zijp, Nov 2008
+#
+# The table contents have been slightly changed so that the exponent
+# bias is now in the exponent table instead of the mantissa table (mostly
+# for cosmetic reasons, and because it theoretically allows a variant
+# that flushes denormal to zero but uses a mantissa table with 24-bit
+# entries).
+#
+# The tables are also constructed slightly differently.
+#
+
+# Note that using a 64K * 4 table is a terrible idea since it will not fit
+# in the L1 cache and will massively pollute the L2 cache as well
+#
+# These should instead fit in the L1 cache.
+#
+# TODO: we could use a denormal bias table instead of the mantissa/offset
+# tables: this would reduce the L1 cache usage from 8704 to 2304 bytes
+# but would involve more computation
+#
+# Note however that if denormals are never encountered, the L1 cache usage
+# is only about 4608 bytes anyway.
+
+table_index = None
+table_length = None
+
+def begin(t, n, l):
+	global table_length
+	global table_index
+	table_index = 0
+	table_length = l
+	print
+	print "const " + t + " " + n + "[" + str(l) + "] = {"
+
+def value(v):
+	global table_index
+	table_index += 1
+	print "\t" + hex(v) + ","
+
+def end():
+	global table_length
+	global table_index
+	print "};"
+	assert table_index == table_length
+
+print "/* This file is autogenerated by u_half.py. Do not edit directly. */"
+print "#include \"util/u_half.h\""
+
+begin("uint32_t", "util_half_to_float_mantissa_table", 2048)
+# zero
+value(0)
+
+# denormals
+for i in xrange(1, 1024):
+	m = i << 13
+	e = 0
+
+	# normalize number
+	while (m & 0x00800000) == 0:
+		e -= 0x00800000;
+		m <<= 1;
+
+	m &= ~0x00800000;
+	e += 0x38800000;
+	value(m | e)
+
+# normals
+for i in xrange(1024, 2048):
+	value((i - 1024) << 13)
+end()
+
+begin("uint32_t", "util_half_to_float_exponent_table", 64)
+# positive zero or denormals
+value(0)
+
+# positive numbers
+for i in xrange(1, 31):
+	value(0x38000000 + (i << 23))
+
+# positive infinity/NaN
+value(0x7f800000)
+
+# negative zero or denormals
+value(0x80000000)
+
+# negative numbers
+for i in range(33, 63):
+	value(0xb8000000 + ((i - 32) << 23))
+
+# negative infinity/NaN
+value(0xff800000)
+end()
+
+begin("uint32_t", "util_half_to_float_offset_table", 64)
+# positive zero or denormals
+value(0)
+
+# positive normals
+for i in range(1, 32):
+	value(1024)
+
+# negative zero or denormals
+value(0)
+
+# negative normals
+for i in xrange(33, 64):
+	value(1024)
+end()
+
+begin("uint16_t", "util_float_to_half_base_table", 512)
+for sign in (0, 0x8000):
+	# very small numbers mapping to zero
+	for i in xrange(-127, -24):
+		value(sign | 0)
+
+	# small numbers mapping to denormals
+	for i in xrange(-24, -14):
+		value(sign | (0x400 >> (-14 -i)))
+
+	# normal numbers
+	for i in xrange(-14, 16):
+		value(sign | ((i + 15) << 10))
+
+	# large numbers mapping to infinity
+	for i in xrange(16, 128):
+		value(sign | 0x7c00)
+
+	# infinity and NaNs
+	value(sign | 0x7c00)
+end()
+
+begin("uint8_t", "util_float_to_half_shift_table", 512)
+for sign in (0, 0x8000):
+	# very small numbers mapping to zero
+	for i in xrange(-127, -24):
+		value(24)
+
+	# small numbers mapping to denormals
+	for i in xrange(-24, -14):
+		value(-1 - i)
+
+	# normal numbers
+	for i in xrange(-14, 16):
+		value(13)
+
+	# large numbers mapping to infinity
+	for i in xrange(16, 128):
+		value(24)
+
+	# infinity and NaNs
+	value(13)
+end()
+
diff --git a/src/gallium/auxiliary/util/u_init.h b/src/gallium/auxiliary/util/u_init.h
new file mode 100644
index 00000000000..7bc356a7916
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_init.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright 2010 Luca Barbieri
+ *
+ * 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 (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.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) 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 U_INIT_H
+#define U_INIT_H
+
+/* Use UTIL_INIT(f) to have f called at program initialization.
+   Note that it is only guaranteed to be called if any symbol in the
+   .c file it is in sis referenced by the program.
+
+   UTIL_INIT functions are called in arbitrary order.
+*/
+
+#ifdef __cplusplus
+/* use a C++ global constructor */
+#define UTIL_INIT(f) struct f##__gctor_t {f##__gctor_t() {x();}} f##__gctor;
+#elif defined(_MSC_VER)
+/* add a pointer to the section where MSVC stores global constructor pointers */
+/* see http://blogs.msdn.com/vcblog/archive/2006/10/20/crt-initialization.aspx and
+   http://stackoverflow.com/questions/1113409/attribute-constructor-equivalent-in-vc */
+#pragma section(".CRT$XCU",read)
+#define UTIL_INIT(f) static void __cdecl f##__init(void) {f();}; __declspec(allocate(".CRT$XCU")) void (__cdecl* f##__xcu)(void) = f##__init;
+#elif defined(__GNUC__)
+#define UTIL_INIT(f) static void f##__init(void) __attribute__((constructor)); static void f##__init(void) {f();}
+#else
+#error Unsupported compiler: please find out how to implement global initializers in C on it
+#endif
+
+#endif
+
diff --git a/src/gallium/auxiliary/util/u_inlines.h b/src/gallium/auxiliary/util/u_inlines.h
index 0cb3432c6e4..a48689ee8be 100644
--- a/src/gallium/auxiliary/util/u_inlines.h
+++ b/src/gallium/auxiliary/util/u_inlines.h
@@ -34,6 +34,8 @@
 #include "pipe/p_screen.h"
 #include "util/u_debug.h"
 #include "util/u_atomic.h"
+#include "util/u_box.h"
+#include "util/u_math.h"
 
 
 #ifdef __cplusplus
@@ -87,18 +89,6 @@ pipe_reference(struct pipe_reference *ptr, struct pipe_reference *reference)
    return destroy;
 }
 
-static INLINE void
-pipe_buffer_reference(struct pipe_buffer **ptr, struct pipe_buffer *buf)
-{
-   struct pipe_buffer *old_buf;
-
-   assert(ptr);
-   old_buf = *ptr;
-
-   if (pipe_reference(&(*ptr)->reference, &buf->reference))
-      old_buf->screen->buffer_destroy(old_buf);
-   *ptr = buf;
-}
 
 static INLINE void
 pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf)
@@ -110,106 +100,184 @@ pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf)
    *ptr = surf;
 }
 
+
 static INLINE void
-pipe_texture_reference(struct pipe_texture **ptr, struct pipe_texture *tex)
+pipe_resource_reference(struct pipe_resource **ptr, struct pipe_resource *tex)
 {
-   struct pipe_texture *old_tex = *ptr;
+   struct pipe_resource *old_tex = *ptr;
 
    if (pipe_reference(&(*ptr)->reference, &tex->reference))
-      old_tex->screen->texture_destroy(old_tex);
+      old_tex->screen->resource_destroy(old_tex->screen, old_tex);
    *ptr = tex;
 }
 
 
+static INLINE void
+pipe_sampler_view_reference(struct pipe_sampler_view **ptr, struct pipe_sampler_view *view)
+{
+   struct pipe_sampler_view *old_view = *ptr;
+
+   if (pipe_reference(&(*ptr)->reference, &view->reference))
+      old_view->context->sampler_view_destroy(old_view->context, old_view);
+   *ptr = view;
+}
+
+static INLINE void
+pipe_surface_reset(struct pipe_surface* ps, struct pipe_resource *pt,
+		unsigned face, unsigned level, unsigned zslice, unsigned flags)
+{
+   pipe_resource_reference(&ps->texture, pt);
+   ps->format = pt->format;
+   ps->width = u_minify(pt->width0, level);
+   ps->height = u_minify(pt->height0, level);
+   ps->usage = flags;
+   ps->face = face;
+   ps->level = level;
+   ps->zslice = zslice;
+}
+
+static INLINE void
+pipe_surface_init(struct pipe_surface* ps, struct pipe_resource *pt,
+                unsigned face, unsigned level, unsigned zslice, unsigned flags)
+{
+   ps->texture = 0;
+   pipe_reference_init(&ps->reference, 1);
+   pipe_surface_reset(ps, pt, face, level, zslice, flags);
+}
+
 /*
  * Convenience wrappers for screen buffer functions.
  */
 
-static INLINE struct pipe_buffer *
+static INLINE struct pipe_resource *
 pipe_buffer_create( struct pipe_screen *screen,
-                    unsigned alignment, unsigned usage, unsigned size )
+		    unsigned bind,
+		    unsigned size )
 {
-   return screen->buffer_create(screen, alignment, usage, size);
+   struct pipe_resource buffer;
+   memset(&buffer, 0, sizeof buffer);
+   buffer.target = PIPE_BUFFER;
+   buffer.format = PIPE_FORMAT_R8_UNORM; /* want TYPELESS or similar */
+   buffer.bind = bind;
+   buffer.usage = PIPE_USAGE_DEFAULT;
+   buffer.flags = 0;
+   buffer.width0 = size;
+   buffer.height0 = 1;
+   buffer.depth0 = 1;
+   return screen->resource_create(screen, &buffer);
 }
 
-static INLINE struct pipe_buffer *
-pipe_user_buffer_create( struct pipe_screen *screen, void *ptr, unsigned size )
+
+static INLINE struct pipe_resource *
+pipe_user_buffer_create( struct pipe_screen *screen, void *ptr, unsigned size,
+			 unsigned usage )
 {
-   return screen->user_buffer_create(screen, ptr, size);
+   return screen->user_buffer_create(screen, ptr, size, usage);
 }
 
 static INLINE void *
-pipe_buffer_map(struct pipe_screen *screen,
-                struct pipe_buffer *buf,
-                unsigned usage)
+pipe_buffer_map_range(struct pipe_context *pipe,
+		      struct pipe_resource *buffer,
+		      unsigned offset,
+		      unsigned length,
+		      unsigned usage,
+		      struct pipe_transfer **transfer)
 {
-   if(screen->buffer_map_range) {
-      unsigned offset = 0;
-      unsigned length = buf->size;
-      return screen->buffer_map_range(screen, buf, offset, length, usage);
+   struct pipe_box box;
+   void *map;
+
+   assert(offset < buffer->width0);
+   assert(offset + length <= buffer->width0);
+   assert(length);
+   
+   u_box_1d(offset, length, &box);
+
+   *transfer = pipe->get_transfer( pipe,
+				   buffer,
+				   u_subresource(0, 0),
+				   usage,
+				   &box);
+   
+   if (*transfer == NULL)
+      return NULL;
+
+   map = pipe->transfer_map( pipe, *transfer );
+   if (map == NULL) {
+      pipe->transfer_destroy( pipe, *transfer );
+      return NULL;
    }
-   else
-      return screen->buffer_map(screen, buf, usage);
+
+   /* Match old screen->buffer_map_range() behaviour, return pointer
+    * to where the beginning of the buffer would be:
+    */
+   return (void *)((char *)map - offset);
 }
 
-static INLINE void
-pipe_buffer_unmap(struct pipe_screen *screen,
-                  struct pipe_buffer *buf)
+
+static INLINE void *
+pipe_buffer_map(struct pipe_context *pipe,
+                struct pipe_resource *buffer,
+                unsigned usage,
+		struct pipe_transfer **transfer)
 {
-   screen->buffer_unmap(screen, buf);
+   return pipe_buffer_map_range(pipe, buffer, 0, buffer->width0, usage, transfer);
 }
 
-static INLINE void *
-pipe_buffer_map_range(struct pipe_screen *screen,
-                struct pipe_buffer *buf,
-                unsigned offset,
-                unsigned length,
-                unsigned usage)
+
+static INLINE void
+pipe_buffer_unmap(struct pipe_context *pipe,
+                  struct pipe_resource *buf,
+		  struct pipe_transfer *transfer)
 {
-   assert(offset < buf->size);
-   assert(offset + length <= buf->size);
-   assert(length);
-   if(screen->buffer_map_range)
-      return screen->buffer_map_range(screen, buf, offset, length, usage);
-   else
-      return screen->buffer_map(screen, buf, usage);
+   if (transfer) {
+      pipe->transfer_unmap(pipe, transfer);
+      pipe->transfer_destroy(pipe, transfer);
+   }
 }
 
 static INLINE void
-pipe_buffer_flush_mapped_range(struct pipe_screen *screen,
-                               struct pipe_buffer *buf,
+pipe_buffer_flush_mapped_range(struct pipe_context *pipe,
+			       struct pipe_transfer *transfer,
                                unsigned offset,
                                unsigned length)
 {
-   assert(offset < buf->size);
-   assert(offset + length <= buf->size);
+   struct pipe_box box;
+   int transfer_offset;
+
    assert(length);
-   if(screen->buffer_flush_mapped_range)
-      screen->buffer_flush_mapped_range(screen, buf, offset, length);
+   assert(transfer->box.x <= offset);
+   assert(offset + length <= transfer->box.x + transfer->box.width);
+
+   /* Match old screen->buffer_flush_mapped_range() behaviour, where
+    * offset parameter is relative to the start of the buffer, not the
+    * mapped range.
+    */
+   transfer_offset = offset - transfer->box.x;
+   
+   u_box_1d(transfer_offset, length, &box);
+
+   pipe->transfer_flush_region(pipe, transfer, &box);
 }
 
 static INLINE void
-pipe_buffer_write(struct pipe_screen *screen,
-                  struct pipe_buffer *buf,
-                  unsigned offset, unsigned size,
+pipe_buffer_write(struct pipe_context *pipe,
+                  struct pipe_resource *buf,
+                  unsigned offset,
+		  unsigned size,
                   const void *data)
 {
-   void *map;
-   
-   assert(offset < buf->size);
-   assert(offset + size <= buf->size);
-   assert(size);
-
-   map = pipe_buffer_map_range(screen, buf, offset, size, 
-                               PIPE_BUFFER_USAGE_CPU_WRITE | 
-                               PIPE_BUFFER_USAGE_FLUSH_EXPLICIT |
-                               PIPE_BUFFER_USAGE_DISCARD);
-   assert(map);
-   if(map) {
-      memcpy((uint8_t *)map + offset, data, size);
-      pipe_buffer_flush_mapped_range(screen, buf, offset, size);
-      pipe_buffer_unmap(screen, buf);
-   }
+   struct pipe_box box;
+
+   u_box_1d(offset, size, &box);
+
+   pipe->transfer_inline_write( pipe,
+				buf,
+				u_subresource(0,0),
+				PIPE_TRANSFER_WRITE,
+				&box,
+				data,
+				size,
+				0);
 }
 
 /**
@@ -219,87 +287,88 @@ pipe_buffer_write(struct pipe_screen *screen,
  * been written before.
  */
 static INLINE void
-pipe_buffer_write_nooverlap(struct pipe_screen *screen,
-                            struct pipe_buffer *buf,
+pipe_buffer_write_nooverlap(struct pipe_context *pipe,
+                            struct pipe_resource *buf,
                             unsigned offset, unsigned size,
                             const void *data)
 {
-   void *map;
-
-   assert(offset < buf->size);
-   assert(offset + size <= buf->size);
-   assert(size);
-
-   map = pipe_buffer_map_range(screen, buf, offset, size,
-                               PIPE_BUFFER_USAGE_CPU_WRITE |
-                               PIPE_BUFFER_USAGE_FLUSH_EXPLICIT |
-                               PIPE_BUFFER_USAGE_DISCARD |
-                               PIPE_BUFFER_USAGE_UNSYNCHRONIZED);
-   assert(map);
-   if(map) {
-      memcpy((uint8_t *)map + offset, data, size);
-      pipe_buffer_flush_mapped_range(screen, buf, offset, size);
-      pipe_buffer_unmap(screen, buf);
-   }
+   struct pipe_box box;
+
+   u_box_1d(offset, size, &box);
+
+   pipe->transfer_inline_write(pipe, 
+			       buf,
+			       u_subresource(0,0),
+			       (PIPE_TRANSFER_WRITE |
+				PIPE_TRANSFER_NOOVERWRITE),
+			       &box,
+			       data,
+			       0, 0);
 }
 
 static INLINE void
-pipe_buffer_read(struct pipe_screen *screen,
-                 struct pipe_buffer *buf,
-                 unsigned offset, unsigned size,
+pipe_buffer_read(struct pipe_context *pipe,
+                 struct pipe_resource *buf,
+                 unsigned offset,
+		 unsigned size,
                  void *data)
 {
-   void *map;
-   
-   assert(offset < buf->size);
-   assert(offset + size <= buf->size);
-   assert(size);
-
-   map = pipe_buffer_map_range(screen, buf, offset, size, PIPE_BUFFER_USAGE_CPU_READ);
-   assert(map);
-   if(map) {
-      memcpy(data, (const uint8_t *)map + offset, size);
-      pipe_buffer_unmap(screen, buf);
-   }
+   struct pipe_transfer *src_transfer;
+   ubyte *map;
+
+   map = (ubyte *) pipe_buffer_map_range(pipe,
+					 buf,
+					 offset, size,
+					 PIPE_TRANSFER_READ,
+					 &src_transfer);
+
+   if (map)
+      memcpy(data, map + offset, size);
+
+   pipe_buffer_unmap(pipe, buf, src_transfer);
 }
 
-static INLINE void *
-pipe_transfer_map( struct pipe_transfer *transf )
+static INLINE struct pipe_transfer *
+pipe_get_transfer( struct pipe_context *context,
+		       struct pipe_resource *resource,
+		       unsigned face, unsigned level,
+		       unsigned zslice,
+		       enum pipe_transfer_usage usage,
+		       unsigned x, unsigned y,
+		       unsigned w, unsigned h)
 {
-   struct pipe_screen *screen = transf->texture->screen;
-   return screen->transfer_map(screen, transf);
+   struct pipe_box box;
+   u_box_2d_zslice( x, y, zslice, w, h, &box );
+   return context->get_transfer( context,
+				 resource,
+				 u_subresource(face, level),
+				 usage,
+				 &box );
 }
 
-static INLINE void
-pipe_transfer_unmap( struct pipe_transfer *transf )
+static INLINE void *
+pipe_transfer_map( struct pipe_context *context,
+                   struct pipe_transfer *transfer )
 {
-   struct pipe_screen *screen = transf->texture->screen;
-   screen->transfer_unmap(screen, transf);
+   return context->transfer_map( context, transfer );
 }
 
 static INLINE void
-pipe_transfer_destroy( struct pipe_transfer *transf )
+pipe_transfer_unmap( struct pipe_context *context,
+                     struct pipe_transfer *transfer )
 {
-   struct pipe_screen *screen = transf->texture->screen;
-   screen->tex_transfer_destroy(transf);
+   context->transfer_unmap( context, transfer );
 }
 
-static INLINE unsigned
-pipe_transfer_buffer_flags( struct pipe_transfer *transf )
+
+static INLINE void
+pipe_transfer_destroy( struct pipe_context *context, 
+		       struct pipe_transfer *transfer )
 {
-   switch (transf->usage & PIPE_TRANSFER_READ_WRITE) {
-   case PIPE_TRANSFER_READ_WRITE:
-      return PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE;
-   case PIPE_TRANSFER_READ:
-      return PIPE_BUFFER_USAGE_CPU_READ;
-   case PIPE_TRANSFER_WRITE:
-      return PIPE_BUFFER_USAGE_CPU_WRITE;
-   default:
-      debug_assert(0);
-      return 0;
-   }
+   context->transfer_destroy(context, transfer);
 }
 
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/gallium/auxiliary/util/u_memory.h b/src/gallium/auxiliary/util/u_memory.h
index a2fc5973565..53d56599fe4 100644
--- a/src/gallium/auxiliary/util/u_memory.h
+++ b/src/gallium/auxiliary/util/u_memory.h
@@ -88,7 +88,7 @@ mem_dup(const void *src, uint size)
 /**
  * Offset of a field in a struct, in bytes.
  */
-#define Offset(TYPE, MEMBER) ((unsigned)&(((TYPE *)NULL)->MEMBER))
+#define Offset(TYPE, MEMBER) ((uintptr_t)&(((TYPE *)NULL)->MEMBER))
 
 
 
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index 50f1b1670b6..3ebef9fb749 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -37,6 +37,7 @@
 
 #include "pipe/p_compiler.h"
 #include "pipe/p_format.h"
+#include "util/u_debug.h"
 #include "util/u_format.h"
 #include "util/u_math.h"
 
@@ -92,6 +93,11 @@ util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a,
          uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
       }
       return;
+   case PIPE_FORMAT_B5G5R5X1_UNORM:
+      {
+         uc->us = ((0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
+      }
+      return;
    case PIPE_FORMAT_B5G5R5A1_UNORM:
       {
          uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
@@ -216,6 +222,15 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) 0xff;
       }
       return;
+   case PIPE_FORMAT_B5G5R5X1_UNORM:
+      {
+         ushort p = uc->us;
+         *r = (ubyte) (((p >>  7) & 0xf8) | ((p >> 12) & 0x7));
+         *g = (ubyte) (((p >>  2) & 0xf8) | ((p >>  7) & 0x7));
+         *b = (ubyte) (((p <<  3) & 0xf8) | ((p >>  2) & 0x7));
+         *a = (ubyte) 0xff;
+      }
+      return;
    case PIPE_FORMAT_B5G5R5A1_UNORM:
       {
          ushort p = uc->us;
@@ -361,6 +376,11 @@ util_pack_color(const float rgba[4], enum pipe_format format, union util_color *
          uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
       }
       return;
+   case PIPE_FORMAT_B5G5R5X1_UNORM:
+      {
+         uc->us = ((0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
+      }
+      return;
    case PIPE_FORMAT_B5G5R5A1_UNORM:
       {
          uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
@@ -427,17 +447,17 @@ util_pack_z(enum pipe_format format, double z)
       return (uint) (z * 0xffffffff);
    case PIPE_FORMAT_Z32_FLOAT:
       return (uint)z;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       if (z == 1.0)
          return 0xffffff;
       return (uint) (z * 0xffffff);
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       if (z == 1.0)
          return 0xffffff00;
       return ((uint) (z * 0xffffff)) << 8;
-   case PIPE_FORMAT_S8_UNORM:
+   case PIPE_FORMAT_S8_USCALED:
       /* this case can get it via util_pack_z_stencil() */
       return 0;
    default:
@@ -458,13 +478,13 @@ util_pack_z_stencil(enum pipe_format format, double z, uint s)
    unsigned packed = util_pack_z(format, z);
 
    switch (format) {
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       packed |= s << 24;
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       packed |= s;
       break;
-   case PIPE_FORMAT_S8_UNORM:
+   case PIPE_FORMAT_S8_USCALED:
       packed |= s;
       break;
    default:
diff --git a/src/gallium/auxiliary/util/u_rect.c b/src/gallium/auxiliary/util/u_rect.c
index 8479161c744..098cdfd58b1 100644
--- a/src/gallium/auxiliary/util/u_rect.c
+++ b/src/gallium/auxiliary/util/u_rect.c
@@ -35,6 +35,7 @@
 #include "pipe/p_context.h"
 #include "pipe/p_screen.h"
 #include "util/u_format.h"
+#include "util/u_inlines.h"
 #include "util/u_rect.h"
 
 
@@ -169,7 +170,6 @@ util_surface_copy(struct pipe_context *pipe,
                   unsigned src_x, unsigned src_y, 
                   unsigned w, unsigned h)
 {
-   struct pipe_screen *screen = pipe->screen;
    struct pipe_transfer *src_trans, *dst_trans;
    void *dst_map;
    const void *src_map;
@@ -182,28 +182,28 @@ util_surface_copy(struct pipe_context *pipe,
    src_format = src->texture->format;
    dst_format = dst->texture->format;
 
-   src_trans = screen->get_tex_transfer(screen,
-                                        src->texture,
-                                        src->face,
-                                        src->level,
-                                        src->zslice,
-                                        PIPE_TRANSFER_READ,
-                                        src_x, src_y, w, h);
-
-   dst_trans = screen->get_tex_transfer(screen,
-                                        dst->texture,
-                                        dst->face,
-                                        dst->level,
-                                        dst->zslice,
-                                        PIPE_TRANSFER_WRITE,
-                                        dst_x, dst_y, w, h);
+   src_trans = pipe_get_transfer(pipe,
+				 src->texture,
+				 src->face,
+				 src->level,
+				 src->zslice,
+				 PIPE_TRANSFER_READ,
+				 src_x, src_y, w, h);
+
+   dst_trans = pipe_get_transfer(pipe,
+				 dst->texture,
+				 dst->face,
+				 dst->level,
+				 dst->zslice,
+				 PIPE_TRANSFER_WRITE,
+				 dst_x, dst_y, w, h);
 
    assert(util_format_get_blocksize(dst_format) == util_format_get_blocksize(src_format));
    assert(util_format_get_blockwidth(dst_format) == util_format_get_blockwidth(src_format));
    assert(util_format_get_blockheight(dst_format) == util_format_get_blockheight(src_format));
 
-   src_map = pipe->screen->transfer_map(screen, src_trans);
-   dst_map = pipe->screen->transfer_map(screen, dst_trans);
+   src_map = pipe->transfer_map(pipe, src_trans);
+   dst_map = pipe->transfer_map(pipe, dst_trans);
 
    assert(src_map);
    assert(dst_map);
@@ -221,11 +221,11 @@ util_surface_copy(struct pipe_context *pipe,
                      do_flip ? h - 1 : 0);
    }
 
-   pipe->screen->transfer_unmap(pipe->screen, src_trans);
-   pipe->screen->transfer_unmap(pipe->screen, dst_trans);
+   pipe->transfer_unmap(pipe, src_trans);
+   pipe->transfer_unmap(pipe, dst_trans);
 
-   screen->tex_transfer_destroy(src_trans);
-   screen->tex_transfer_destroy(dst_trans);
+   pipe->transfer_destroy(pipe, src_trans);
+   pipe->transfer_destroy(pipe, dst_trans);
 }
 
 
@@ -243,65 +243,65 @@ util_surface_fill(struct pipe_context *pipe,
                   unsigned dstx, unsigned dsty,
                   unsigned width, unsigned height, unsigned value)
 {
-   struct pipe_screen *screen = pipe->screen;
    struct pipe_transfer *dst_trans;
    void *dst_map;
 
    assert(dst->texture);
    if (!dst->texture)
       return;
-   dst_trans = screen->get_tex_transfer(screen,
-                                        dst->texture,
-                                        dst->face,
-                                        dst->level,
-                                        dst->zslice,
-                                        PIPE_TRANSFER_WRITE,
-                                        dstx, dsty, width, height);
+   dst_trans = pipe_get_transfer(pipe,
+				 dst->texture,
+				 dst->face,
+				 dst->level,
+				 dst->zslice,
+				 PIPE_TRANSFER_WRITE,
+				 dstx, dsty, width, height);
 
-   dst_map = pipe->screen->transfer_map(screen, dst_trans);
+   dst_map = pipe->transfer_map(pipe, dst_trans);
 
    assert(dst_map);
 
    if (dst_map) {
       assert(dst_trans->stride > 0);
 
-      switch (util_format_get_blocksize(dst_trans->texture->format)) {
+      switch (util_format_get_blocksize(dst->texture->format)) {
       case 1:
       case 2:
       case 4:
-         util_fill_rect(dst_map, dst_trans->texture->format, dst_trans->stride,
+         util_fill_rect(dst_map, dst->texture->format,
+			dst_trans->stride,
                         0, 0, width, height, value);
          break;
       case 8:
-         {
-            /* expand the 4-byte clear value to an 8-byte value */
-            ushort *row = (ushort *) dst_map;
-            ushort val0 = UBYTE_TO_USHORT((value >>  0) & 0xff);
-            ushort val1 = UBYTE_TO_USHORT((value >>  8) & 0xff);
-            ushort val2 = UBYTE_TO_USHORT((value >> 16) & 0xff);
-            ushort val3 = UBYTE_TO_USHORT((value >> 24) & 0xff);
-            unsigned i, j;
-            val0 = (val0 << 8) | val0;
-            val1 = (val1 << 8) | val1;
-            val2 = (val2 << 8) | val2;
-            val3 = (val3 << 8) | val3;
-            for (i = 0; i < height; i++) {
-               for (j = 0; j < width; j++) {
-                  row[j*4+0] = val0;
-                  row[j*4+1] = val1;
-                  row[j*4+2] = val2;
-                  row[j*4+3] = val3;
-               }
-               row += dst_trans->stride/2;
-            }
-         }
-         break;
+      {
+	 /* expand the 4-byte clear value to an 8-byte value */
+	 ushort *row = (ushort *) dst_map;
+	 ushort val0 = UBYTE_TO_USHORT((value >>  0) & 0xff);
+	 ushort val1 = UBYTE_TO_USHORT((value >>  8) & 0xff);
+	 ushort val2 = UBYTE_TO_USHORT((value >> 16) & 0xff);
+	 ushort val3 = UBYTE_TO_USHORT((value >> 24) & 0xff);
+	 unsigned i, j;
+	 val0 = (val0 << 8) | val0;
+	 val1 = (val1 << 8) | val1;
+	 val2 = (val2 << 8) | val2;
+	 val3 = (val3 << 8) | val3;
+	 for (i = 0; i < height; i++) {
+	    for (j = 0; j < width; j++) {
+	       row[j*4+0] = val0;
+	       row[j*4+1] = val1;
+	       row[j*4+2] = val2;
+	       row[j*4+3] = val3;
+	    }
+	    row += dst_trans->stride/2;
+	 }
+      }
+      break;
       default:
          assert(0);
          break;
       }
    }
 
-   pipe->screen->transfer_unmap(pipe->screen, dst_trans);
-   screen->tex_transfer_destroy(dst_trans);
+   pipe->transfer_unmap(pipe, dst_trans);
+   pipe->transfer_destroy(pipe, dst_trans);
 }
diff --git a/src/gallium/auxiliary/util/u_resource.c b/src/gallium/auxiliary/util/u_resource.c
new file mode 100644
index 00000000000..9e6474b83de
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_resource.c
@@ -0,0 +1,96 @@
+
+
+#include "util/u_inlines.h"
+#include "util/u_transfer.h"
+
+static INLINE struct u_resource *
+u_resource( struct pipe_resource *res )
+{
+   return (struct u_resource *)res;
+}
+
+boolean u_resource_get_handle_vtbl(struct pipe_screen *screen,
+			  struct pipe_resource *resource,
+			  struct winsys_handle *handle)
+{
+   struct u_resource *ur = u_resource(resource);
+   return ur->vtbl->resource_get_handle(screen, resource, handle);
+}
+
+void u_resource_destroy_vtbl(struct pipe_screen *screen,
+		    struct pipe_resource *resource)
+{
+   struct u_resource *ur = u_resource(resource);
+   ur->vtbl->resource_destroy(screen, resource);
+}
+
+unsigned u_is_resource_referenced_vtbl( struct pipe_context *pipe,
+					struct pipe_resource *resource,
+					unsigned face, unsigned level)
+{
+   struct u_resource *ur = u_resource(resource);
+   return ur->vtbl->is_resource_referenced(pipe, resource, face, level);
+}
+
+struct pipe_transfer *u_get_transfer_vtbl(struct pipe_context *context,
+					  struct pipe_resource *resource,
+					  struct pipe_subresource sr,
+					  enum pipe_transfer_usage usage,
+					  const struct pipe_box *box)
+{
+   struct u_resource *ur = u_resource(resource);
+   return ur->vtbl->get_transfer(context, resource, sr, usage, box);
+}
+
+void u_transfer_destroy_vtbl(struct pipe_context *pipe,
+			     struct pipe_transfer *transfer)
+{
+   struct u_resource *ur = u_resource(transfer->resource);
+   ur->vtbl->transfer_destroy(pipe, transfer);
+}
+
+void *u_transfer_map_vtbl( struct pipe_context *pipe,
+			   struct pipe_transfer *transfer )
+{
+   struct u_resource *ur = u_resource(transfer->resource);
+   return ur->vtbl->transfer_map(pipe, transfer);
+}
+
+void u_transfer_flush_region_vtbl( struct pipe_context *pipe,
+				   struct pipe_transfer *transfer,
+				   const struct pipe_box *box)
+{
+   struct u_resource *ur = u_resource(transfer->resource);
+   ur->vtbl->transfer_flush_region(pipe, transfer, box);
+}
+
+void u_transfer_unmap_vtbl( struct pipe_context *pipe,
+			    struct pipe_transfer *transfer )
+{
+   struct u_resource *ur = u_resource(transfer->resource);
+   ur->vtbl->transfer_unmap(pipe, transfer);
+}
+
+void u_transfer_inline_write_vtbl( struct pipe_context *pipe,
+				   struct pipe_resource *resource,
+				   struct pipe_subresource sr,
+				   unsigned usage,
+				   const struct pipe_box *box,
+				   const void *data,
+				   unsigned stride,
+				   unsigned slice_stride)
+{
+   struct u_resource *ur = u_resource(resource);
+   ur->vtbl->transfer_inline_write(pipe, 
+				   resource,
+				   sr,
+				   usage,
+				   box,
+				   data,
+				   stride,
+				   slice_stride);
+}
+
+
+
+
diff --git a/src/gallium/auxiliary/util/u_sampler.c b/src/gallium/auxiliary/util/u_sampler.c
new file mode 100644
index 00000000000..e77f562ea22
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_sampler.c
@@ -0,0 +1,100 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+#include "u_format.h"
+#include "u_sampler.h"
+
+
+static void
+default_template(struct pipe_sampler_view *view,
+                 const struct pipe_resource *texture,
+                 enum pipe_format format,
+                 unsigned expand_green_blue)
+{
+   /* XXX: Check if format is compatible with texture->format.
+    */
+
+   view->format = format;
+   view->first_level = 0;
+   view->last_level = texture->last_level;
+   view->swizzle_r = PIPE_SWIZZLE_RED;
+   view->swizzle_g = PIPE_SWIZZLE_GREEN;
+   view->swizzle_b = PIPE_SWIZZLE_BLUE;
+   view->swizzle_a = PIPE_SWIZZLE_ALPHA;
+
+   /* Override default green and blue component expansion to the requested
+    * one.
+    *
+    * Gallium expands nonexistent components to (0,0,0,1), DX9 expands
+    * to (1,1,1,1).  Since alpha is always expanded to 1, and red is
+    * always present, we only really care about green and blue
+    * components.
+    *
+    * To make it look less hackish, one would have to add
+    * UTIL_FORMAT_SWIZZLE_EXPAND to indicate components for expansion
+    * and then override without exceptions or favoring one component
+    * over another.
+    */
+   if (format != PIPE_FORMAT_A8_UNORM) {
+      const struct util_format_description *desc = util_format_description(format);
+
+      assert(desc);
+      if (desc) {
+         if (desc->swizzle[1] == UTIL_FORMAT_SWIZZLE_0) {
+            view->swizzle_g = expand_green_blue;
+         }
+         if (desc->swizzle[2] == UTIL_FORMAT_SWIZZLE_0) {
+            view->swizzle_b = expand_green_blue;
+         }
+      }
+   }
+}
+
+void
+u_sampler_view_default_template(struct pipe_sampler_view *view,
+                                const struct pipe_resource *texture,
+                                enum pipe_format format)
+{
+   /* Expand to (0, 0, 0, 1) */
+   default_template(view,
+                    texture,
+                    format,
+                    PIPE_SWIZZLE_ZERO);
+}
+
+void
+u_sampler_view_default_dx9_template(struct pipe_sampler_view *view,
+                                    const struct pipe_resource *texture,
+                                    enum pipe_format format)
+{
+   /* Expand to (1, 1, 1, 1) */
+   default_template(view,
+                    texture,
+                    format,
+                    PIPE_SWIZZLE_ONE);
+}
diff --git a/src/gallium/auxiliary/util/u_sampler.h b/src/gallium/auxiliary/util/u_sampler.h
new file mode 100644
index 00000000000..f3dad7417e0
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_sampler.h
@@ -0,0 +1,57 @@
+/**************************************************************************
+ *
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE 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 U_SAMPLER_H
+#define U_SAMPLER_H
+
+
+#include "pipe/p_defines.h"
+#include "pipe/p_format.h"
+#include "pipe/p_state.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+
+void
+u_sampler_view_default_template(struct pipe_sampler_view *view,
+                                const struct pipe_resource *texture,
+                                enum pipe_format format);
+
+void
+u_sampler_view_default_dx9_template(struct pipe_sampler_view *view,
+                                    const struct pipe_resource *texture,
+                                    enum pipe_format format);
+
+
+#ifdef __cplusplus
+} /* extern "C" { */
+#endif
+
+#endif /* U_SAMPLER_H */
diff --git a/src/gallium/auxiliary/util/u_simple_screen.c b/src/gallium/auxiliary/util/u_simple_screen.c
deleted file mode 100644
index 53f3c16dbcc..00000000000
--- a/src/gallium/auxiliary/util/u_simple_screen.c
+++ /dev/null
@@ -1,153 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2009 VMware, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-#include "u_simple_screen.h"
-
-#include "pipe/p_screen.h"
-#include "pipe/p_state.h"
-#include "util/u_simple_screen.h"
-
-
-static struct pipe_buffer *
-pass_buffer_create(struct pipe_screen *screen,
-                   unsigned alignment,
-                   unsigned usage,
-                   unsigned size)
-{
-   struct pipe_buffer *buffer =
-      screen->winsys->buffer_create(screen->winsys, alignment, usage, size);
-
-   buffer->screen = screen;
-
-   return buffer;
-}
-
-static struct pipe_buffer *
-pass_user_buffer_create(struct pipe_screen *screen,
-                        void *ptr,
-                        unsigned bytes)
-{
-   struct pipe_buffer *buffer =
-      screen->winsys->user_buffer_create(screen->winsys, ptr, bytes);
-
-   buffer->screen = screen;
-
-   return buffer;
-}
-
-static struct pipe_buffer *
-pass_surface_buffer_create(struct pipe_screen *screen,
-                           unsigned width, unsigned height,
-                           enum pipe_format format,
-                           unsigned usage,
-                           unsigned tex_usage,
-                           unsigned *stride)
-{
-   struct pipe_buffer *buffer =
-      screen->winsys->surface_buffer_create(screen->winsys, width, height,
-                                            format, usage, tex_usage, stride);
-
-   buffer->screen = screen;
-
-   return buffer;
-}
-
-static void *
-pass_buffer_map(struct pipe_screen *screen,
-                struct pipe_buffer *buf,
-                unsigned usage)
-{
-   return screen->winsys->buffer_map(screen->winsys, buf, usage);
-}
-
-static void
-pass_buffer_unmap(struct pipe_screen *screen,
-                  struct pipe_buffer *buf)
-{
-   screen->winsys->buffer_unmap(screen->winsys, buf);
-}
-
-static void
-pass_buffer_destroy(struct pipe_buffer *buf)
-{
-   buf->screen->winsys->buffer_destroy(buf);
-}
-
-
-static void
-pass_flush_frontbuffer(struct pipe_screen *screen,
-                       struct pipe_surface *surf,
-                       void *context_private)
-{
-   screen->winsys->flush_frontbuffer(screen->winsys, surf, context_private);
-}
-
-static void
-pass_fence_reference(struct pipe_screen *screen,
-                     struct pipe_fence_handle **ptr,
-                     struct pipe_fence_handle *fence)
-{
-   screen->winsys->fence_reference(screen->winsys, ptr, fence);
-}
-
-static int
-pass_fence_signalled(struct pipe_screen *screen,
-                     struct pipe_fence_handle *fence,
-                     unsigned flag)
-{
-   return screen->winsys->fence_signalled(screen->winsys, fence, flag);
-}
-
-static int
-pass_fence_finish(struct pipe_screen *screen,
-                  struct pipe_fence_handle *fence,
-                  unsigned flag)
-{
-   return screen->winsys->fence_finish(screen->winsys, fence, flag);
-}
-
-void
-u_simple_screen_init(struct pipe_screen *screen)
-{
-   screen->buffer_create = pass_buffer_create;
-   screen->user_buffer_create = pass_user_buffer_create;
-   screen->surface_buffer_create = pass_surface_buffer_create;
-
-   screen->buffer_map = pass_buffer_map;
-   screen->buffer_unmap = pass_buffer_unmap;
-   screen->buffer_destroy = pass_buffer_destroy;
-   screen->flush_frontbuffer = pass_flush_frontbuffer;
-   screen->fence_reference = pass_fence_reference;
-   screen->fence_signalled = pass_fence_signalled;
-   screen->fence_finish = pass_fence_finish;
-}
-
-const char *
-u_simple_screen_winsys_name(struct pipe_screen *screen)
-{
-   return screen->winsys->get_name(screen->winsys);
-}
diff --git a/src/gallium/auxiliary/util/u_simple_screen.h b/src/gallium/auxiliary/util/u_simple_screen.h
index bb3f5ba102f..b52232f025c 100644
--- a/src/gallium/auxiliary/util/u_simple_screen.h
+++ b/src/gallium/auxiliary/util/u_simple_screen.h
@@ -33,7 +33,7 @@
 struct pipe_screen;
 struct pipe_fence_handle;
 struct pipe_surface;
-struct pipe_buffer;
+struct pipe_resource;
 
 /**
  * Gallium3D drivers are (meant to be!) independent of both GL and the
@@ -53,11 +53,6 @@ struct pipe_winsys
    const char *(*get_name)( struct pipe_winsys *ws );
 
    /**
-    * Do any special operations to ensure buffer size is correct
-    */
-   void (*update_buffer)( struct pipe_winsys *ws,
-                          void *context_private );
-   /**
     * Do any special operations to ensure frontbuffer contents are
     * displayed, eg copy fake frontbuffer.
     */
@@ -73,14 +68,13 @@ struct pipe_winsys
     * window systems must then implement that interface (rather than the
     * other way around...).
     *
-    * usage is a bitmask of PIPE_BUFFER_USAGE_PIXEL/VERTEX/INDEX/CONSTANT. This
-    * usage argument is only an optimization hint, not a guarantee, therefore
-    * proper behavior must be observed in all circumstances.
+    * usage is a bitmask of PIPE_BIND_*.
+    * All possible usages must be present.
     *
     * alignment indicates the client's alignment requirements, eg for
     * SSE instructions.
     */
-   struct pipe_buffer *(*buffer_create)( struct pipe_winsys *ws,
+   struct pipe_resource *(*buffer_create)( struct pipe_winsys *ws,
                                          unsigned alignment,
                                          unsigned usage,
                                          unsigned size );
@@ -106,7 +100,7 @@ struct pipe_winsys
     * Note that ptr may be accessed at any time upto the time when the
     * buffer is destroyed, so the data must not be freed before then.
     */
-   struct pipe_buffer *(*user_buffer_create)(struct pipe_winsys *ws,
+   struct pipe_resource *(*user_buffer_create)(struct pipe_winsys *ws,
                                                     void *ptr,
                                                     unsigned bytes);
 
@@ -117,11 +111,11 @@ struct pipe_winsys
     * display targets) must be allocated with special characteristics, memory
     * pools, or obtained directly from the windowing system.
     *
-    * This callback is invoked by the pipe_screenwhen creating a texture marked
-    * with the PIPE_TEXTURE_USAGE_DISPLAY_TARGET flag  to get the underlying
+    * This callback is invoked by the pipe_screen when creating a texture marked
+    * with the PIPE_BIND_DISPLAY_TARGET flag  to get the underlying
     * buffer storage.
     */
-   struct pipe_buffer *(*surface_buffer_create)(struct pipe_winsys *ws,
+   struct pipe_resource *(*surface_buffer_create)(struct pipe_winsys *ws,
                                                 unsigned width, unsigned height,
                                                 enum pipe_format format,
                                                 unsigned usage,
@@ -134,13 +128,13 @@ struct pipe_winsys
     * flags is bitmask of PIPE_BUFFER_USAGE_CPU_READ/WRITE flags.
     */
    void *(*buffer_map)( struct pipe_winsys *ws,
-                        struct pipe_buffer *buf,
+                        struct pipe_resource *buf,
                         unsigned usage );
 
    void (*buffer_unmap)( struct pipe_winsys *ws,
-                         struct pipe_buffer *buf );
+                         struct pipe_resource *buf );
 
-   void (*buffer_destroy)( struct pipe_buffer *buf );
+   void (*buffer_destroy)( struct pipe_resource *buf );
 
 
    /** Set ptr = fence, with reference counting */
diff --git a/src/gallium/auxiliary/util/u_surface.c b/src/gallium/auxiliary/util/u_surface.c
index 33306bbc2a6..42440d0d673 100644
--- a/src/gallium/auxiliary/util/u_surface.c
+++ b/src/gallium/auxiliary/util/u_surface.c
@@ -50,7 +50,8 @@
 boolean
 util_create_rgba_surface(struct pipe_screen *screen,
                          uint width, uint height,
-                         struct pipe_texture **textureOut,
+			 uint bind,
+                         struct pipe_resource **textureOut,
                          struct pipe_surface **surfaceOut)
 {
    static const enum pipe_format rgbaFormats[] = {
@@ -60,15 +61,14 @@ util_create_rgba_surface(struct pipe_screen *screen,
       PIPE_FORMAT_NONE
    };
    const uint target = PIPE_TEXTURE_2D;
-   const uint usage = PIPE_TEXTURE_USAGE_RENDER_TARGET;
    enum pipe_format format = PIPE_FORMAT_NONE;
-   struct pipe_texture templ;
+   struct pipe_resource templ;
    uint i;
 
    /* Choose surface format */
    for (i = 0; rgbaFormats[i]; i++) {
       if (screen->is_format_supported(screen, rgbaFormats[i],
-                                      target, usage, 0)) {
+                                      target, bind, 0)) {
          format = rgbaFormats[i];
          break;
       }
@@ -84,16 +84,19 @@ util_create_rgba_surface(struct pipe_screen *screen,
    templ.width0 = width;
    templ.height0 = height;
    templ.depth0 = 1;
-   templ.tex_usage = usage;
+   templ.bind = bind;
 
-   *textureOut = screen->texture_create(screen, &templ);
+   *textureOut = screen->resource_create(screen, &templ);
    if (!*textureOut)
       return FALSE;
 
    /* create surface / view into texture */
-   *surfaceOut = screen->get_tex_surface(screen, *textureOut, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE);
+   *surfaceOut = screen->get_tex_surface(screen, 
+					 *textureOut,
+					 0, 0, 0,
+					 bind);
    if (!*surfaceOut) {
-      pipe_texture_reference(textureOut, NULL);
+      pipe_resource_reference(textureOut, NULL);
       return FALSE;
    }
 
@@ -105,11 +108,11 @@ util_create_rgba_surface(struct pipe_screen *screen,
  * Release the surface and texture from util_create_rgba_surface().
  */
 void
-util_destroy_rgba_surface(struct pipe_texture *texture,
+util_destroy_rgba_surface(struct pipe_resource *texture,
                           struct pipe_surface *surface)
 {
    pipe_surface_reference(&surface, NULL);
-   pipe_texture_reference(&texture, NULL);
+   pipe_resource_reference(&texture, NULL);
 }
 
 
diff --git a/src/gallium/auxiliary/util/u_surface.h b/src/gallium/auxiliary/util/u_surface.h
index 3c60df2c3e5..119fcd4ce8e 100644
--- a/src/gallium/auxiliary/util/u_surface.h
+++ b/src/gallium/auxiliary/util/u_surface.h
@@ -52,13 +52,13 @@ util_same_surface(const struct pipe_surface *s1, const struct pipe_surface *s2)
 
 extern boolean
 util_create_rgba_surface(struct pipe_screen *screen,
-                         uint width, uint height,
-                         struct pipe_texture **textureOut,
+                         uint width, uint height, uint bind,
+                         struct pipe_resource **textureOut,
                          struct pipe_surface **surfaceOut);
 
 
 extern void
-util_destroy_rgba_surface(struct pipe_texture *texture,
+util_destroy_rgba_surface(struct pipe_resource *texture,
                           struct pipe_surface *surface);
 
 
diff --git a/src/gallium/auxiliary/util/u_surfaces.c b/src/gallium/auxiliary/util/u_surfaces.c
new file mode 100644
index 00000000000..668da8c5c27
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_surfaces.c
@@ -0,0 +1,112 @@
+#include "u_surfaces.h"
+#include "util/u_hash_table.h"
+#include "util/u_inlines.h"
+#include "util/u_memory.h"
+
+/* TODO: ouch, util_hash_table should do these by default when passed a null function pointer
+ * this indirect function call is quite bad
+ */
+static unsigned
+hash(void *key)
+{
+   return (unsigned)(uintptr_t)key;
+}
+
+static int
+compare(void *key1, void *key2)
+{
+   return (unsigned)(uintptr_t)key1 - (unsigned)(uintptr_t)key2;
+}
+
+struct pipe_surface *
+util_surfaces_do_get(struct util_surfaces *us, unsigned surface_struct_size, struct pipe_screen *pscreen, struct pipe_resource *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags)
+{
+   struct pipe_surface *ps;
+   void *key = NULL;
+
+   if(pt->target == PIPE_TEXTURE_3D || pt->target == PIPE_TEXTURE_CUBE)
+   {	/* or 2D array */
+      if(!us->u.table)
+	 us->u.table = util_hash_table_create(hash, compare);
+      key = (void *)(((zslice + face) << 8) | level);
+      /* TODO: ouch, should have a get-reference function...
+       * also, shouldn't allocate a two-pointer structure for each item... */
+      ps = util_hash_table_get(us->u.table, key);
+   }
+   else
+   {
+      if(!us->u.array)
+	 us->u.array = CALLOC(pt->last_level + 1, sizeof(struct pipe_surface *));
+      ps = us->u.array[level];
+   }
+
+   if(ps)
+   {
+      p_atomic_inc(&ps->reference.count);
+      return ps;
+   }
+
+   ps = (struct pipe_surface *)CALLOC(1, surface_struct_size);
+   if(!ps)
+      return NULL;
+
+   pipe_surface_init(ps, pt, face, level, zslice, flags);
+   ps->offset = ~0;
+
+   if(pt->target == PIPE_TEXTURE_3D || pt->target == PIPE_TEXTURE_CUBE)
+      util_hash_table_set(us->u.table, key, ps);
+   else
+      us->u.array[level] = ps;
+
+   return ps;
+}
+
+void
+util_surfaces_do_detach(struct util_surfaces *us, struct pipe_surface *ps)
+{
+   struct pipe_resource *pt = ps->texture;
+   if(pt->target == PIPE_TEXTURE_3D || pt->target == PIPE_TEXTURE_CUBE)
+   {	/* or 2D array */
+      void* key = (void*)(uintptr_t)(((ps->zslice + ps->face) << 8) | ps->level);
+      util_hash_table_remove(us->u.table, key);
+   }
+   else
+      us->u.array[ps->level] = 0;
+}
+
+static enum pipe_error
+util_surfaces_destroy_callback(void *key, void *value, void *data)
+{
+   void (*destroy_surface) (struct pipe_surface * ps) = data;
+   destroy_surface((struct pipe_surface *)value);
+   return PIPE_OK;
+}
+
+void
+util_surfaces_destroy(struct util_surfaces *us, struct pipe_resource *pt, void (*destroy_surface) (struct pipe_surface *))
+{
+   if(pt->target == PIPE_TEXTURE_3D || pt->target == PIPE_TEXTURE_CUBE)
+   {	/* or 2D array */
+      if(us->u.table)
+      {
+	 util_hash_table_foreach(us->u.table, util_surfaces_destroy_callback, destroy_surface);
+	 util_hash_table_destroy(us->u.table);
+	 us->u.table = NULL;
+      }
+   }
+   else
+   {
+      if(us->u.array)
+      {
+	 unsigned i;
+	 for(i = 0; i < pt->last_level; ++i)
+	 {
+	    struct pipe_surface *ps = us->u.array[i];
+	    if(ps)
+	       destroy_surface(ps);
+	 }
+	 FREE(us->u.array);
+	 us->u.array = NULL;
+      }
+   }
+}
diff --git a/src/gallium/auxiliary/util/u_surfaces.h b/src/gallium/auxiliary/util/u_surfaces.h
new file mode 100644
index 00000000000..0195bf5afba
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_surfaces.h
@@ -0,0 +1,54 @@
+#ifndef U_SURFACES_H_
+#define U_SURFACES_H_
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_state.h"
+#include "util/u_atomic.h"
+
+struct util_hash_table;
+
+struct util_surfaces
+{
+   union
+   {
+      struct util_hash_table *table;
+      struct pipe_surface **array;
+   } u;
+};
+
+struct pipe_surface *util_surfaces_do_get(struct util_surfaces *us, unsigned surface_struct_size, struct pipe_screen *pscreen, struct pipe_resource *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags);
+
+/* fast inline path for the very common case */
+static INLINE struct pipe_surface *
+util_surfaces_get(struct util_surfaces *us, unsigned surface_struct_size, struct pipe_screen *pscreen, struct pipe_resource *pt, unsigned face, unsigned level, unsigned zslice, unsigned flags)
+{
+   if(likely(pt->target == PIPE_TEXTURE_2D && us->u.array))
+   {
+      struct pipe_surface *ps = us->u.array[level];
+      if(ps)
+      {
+	 p_atomic_inc(&ps->reference.count);
+	 return ps;
+      }
+   }
+
+   return util_surfaces_do_get(us, surface_struct_size, pscreen, pt, face, level, zslice, flags);
+}
+
+void util_surfaces_do_detach(struct util_surfaces *us, struct pipe_surface *ps);
+
+static INLINE void
+util_surfaces_detach(struct util_surfaces *us, struct pipe_surface *ps)
+{
+   if(likely(ps->texture->target == PIPE_TEXTURE_2D))
+   {
+      us->u.array[ps->level] = 0;
+      return;
+   }
+
+   util_surfaces_do_detach(us, ps);
+}
+
+void util_surfaces_destroy(struct util_surfaces *us, struct pipe_resource *pt, void (*destroy_surface) (struct pipe_surface *));
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 79481b710bf..f7aa1403d08 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -45,27 +45,27 @@
  * Move raw block of pixels from transfer object to user memory.
  */
 void
-pipe_get_tile_raw(struct pipe_transfer *pt,
+pipe_get_tile_raw(struct pipe_context *pipe,
+                  struct pipe_transfer *pt,
                   uint x, uint y, uint w, uint h,
                   void *dst, int dst_stride)
 {
-   struct pipe_screen *screen = pt->texture->screen;
    const void *src;
 
    if (dst_stride == 0)
-      dst_stride = util_format_get_stride(pt->texture->format, w);
+      dst_stride = util_format_get_stride(pt->resource->format, w);
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
-   src = screen->transfer_map(screen, pt);
+   src = pipe->transfer_map(pipe, pt);
    assert(src);
    if(!src)
       return;
 
-   util_copy_rect(dst, pt->texture->format, dst_stride, 0, 0, w, h, src, pt->stride, x, y);
+   util_copy_rect(dst, pt->resource->format, dst_stride, 0, 0, w, h, src, pt->stride, x, y);
 
-   screen->transfer_unmap(screen, pt);
+   pipe->transfer_unmap(pipe, pt);
 }
 
 
@@ -73,28 +73,28 @@ pipe_get_tile_raw(struct pipe_transfer *pt,
  * Move raw block of pixels from user memory to transfer object.
  */
 void
-pipe_put_tile_raw(struct pipe_transfer *pt,
+pipe_put_tile_raw(struct pipe_context *pipe,
+                  struct pipe_transfer *pt,
                   uint x, uint y, uint w, uint h,
                   const void *src, int src_stride)
 {
-   struct pipe_screen *screen = pt->texture->screen;
    void *dst;
-   enum pipe_format format = pt->texture->format;
+   enum pipe_format format = pt->resource->format;
 
    if (src_stride == 0)
       src_stride = util_format_get_stride(format, w);
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
-   dst = screen->transfer_map(screen, pt);
+   dst = pipe->transfer_map(pipe, pt);
    assert(dst);
    if(!dst)
       return;
 
    util_copy_rect(dst, format, pt->stride, x, y, w, h, src, src_stride, 0, 0);
 
-   screen->transfer_unmap(screen, pt);
+   pipe->transfer_unmap(pipe, pt);
 }
 
 
@@ -108,387 +108,6 @@ pipe_put_tile_raw(struct pipe_transfer *pt,
 
 
 
-/*** PIPE_FORMAT_B8G8R8A8_UNORM ***/
-
-static void
-a8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  0) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >> 24) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B8G8R8X8_UNORM ***/
-
-static void
-x8r8g8b8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  0) & 0xff);
-         pRow[3] = 1.0F;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-x8r8g8b8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         *dst++ = (0xff << 24) | (r << 16) | (g << 8) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
-
-static void
-b8g8r8a8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >> 24) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >>  0) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-b8g8r8a8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (b << 24) | (g << 16) | (r << 8) | a;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_A8B8G8R8_UNORM ***/
-
-static void
-r8g8b8a8_get_tile_rgba(const unsigned *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = ubyte_to_float((pixel >> 24) & 0xff);
-         pRow[1] = ubyte_to_float((pixel >> 16) & 0xff);
-         pRow[2] = ubyte_to_float((pixel >>  8) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >>  0) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r8g8b8a8_put_tile_rgba(unsigned *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (r << 24) | (g << 16) | (b << 8) | a;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B5G5R5A1_UNORM ***/
-
-static void
-a1r5g5b5_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >> 10) & 0x1f) * (1.0f / 31.0f);
-         pRow[1] = ((pixel >>  5) & 0x1f) * (1.0f / 31.0f);
-         pRow[2] = ((pixel      ) & 0x1f) * (1.0f / 31.0f);
-         pRow[3] = ((pixel >> 15)       ) * 1.0f;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a1r5g5b5_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         r = r >> 3;  /* 5 bits */
-         g = g >> 3;  /* 5 bits */
-         b = b >> 3;  /* 5 bits */
-         a = a >> 7;  /* 1 bit */
-         *dst++ = (a << 15) | (r << 10) | (g << 5) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B4G4R4A4_UNORM ***/
-
-static void
-a4r4g4b4_get_tile_rgba(const ushort *src,
-                       unsigned w, unsigned h,
-                       float *p,
-                       unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >>  8) & 0xf) * (1.0f / 15.0f);
-         pRow[1] = ((pixel >>  4) & 0xf) * (1.0f / 15.0f);
-         pRow[2] = ((pixel      ) & 0xf) * (1.0f / 15.0f);
-         pRow[3] = ((pixel >> 12)      ) * (1.0f / 15.0f);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a4r4g4b4_put_tile_rgba(ushort *dst,
-                       unsigned w, unsigned h,
-                       const float *p,
-                       unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = float_to_ubyte(pRow[0]);
-         g = float_to_ubyte(pRow[1]);
-         b = float_to_ubyte(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         r >>= 4;
-         g >>= 4;
-         b >>= 4;
-         a >>= 4;
-         *dst++ = (a << 12) | (r << 8) | (g << 4) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_B5G6R5_UNORM ***/
-
-static void
-r5g6b5_get_tile_rgba(const ushort *src,
-                     unsigned w, unsigned h,
-                     float *p,
-                     unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const ushort pixel = *src++;
-         pRow[0] = ((pixel >> 11) & 0x1f) * (1.0f / 31.0f);
-         pRow[1] = ((pixel >>  5) & 0x3f) * (1.0f / 63.0f);
-         pRow[2] = ((pixel      ) & 0x1f) * (1.0f / 31.0f);
-         pRow[3] = 1.0f;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r5g6b5_put_tile_rgba(ushort *dst,
-                     unsigned w, unsigned h,
-                     const float *p,
-                     unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         uint r = (uint) (CLAMP(pRow[0], 0.0, 1.0) * 31.0);
-         uint g = (uint) (CLAMP(pRow[1], 0.0, 1.0) * 63.0);
-         uint b = (uint) (CLAMP(pRow[2], 0.0, 1.0) * 31.0);
-         *dst++ = (r << 11) | (g << 5) | (b);
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_R8G8B8_UNORM ***/
-
-static void
-r8g8b8_get_tile_rgba(const ubyte *src,
-                     unsigned w, unsigned h,
-                     float *p,
-                     unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         pRow[0] = ubyte_to_float(src[0]);
-         pRow[1] = ubyte_to_float(src[1]);
-         pRow[2] = ubyte_to_float(src[2]);
-         pRow[3] = 1.0f;
-         src += 3;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r8g8b8_put_tile_rgba(ubyte *dst,
-                     unsigned w, unsigned h,
-                     const float *p,
-                     unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         dst[0] = float_to_ubyte(pRow[0]);
-         dst[1] = float_to_ubyte(pRow[1]);
-         dst[2] = float_to_ubyte(pRow[2]);
-         dst += 3;
-      }
-      p += src_stride;
-   }
-}
-
-
-
 /*** PIPE_FORMAT_Z16_UNORM ***/
 
 /**
@@ -518,448 +137,6 @@ z16_get_tile_rgba(const ushort *src,
 
 
 
-/*** PIPE_FORMAT_L8_UNORM ***/
-
-static void
-l8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = ubyte_to_float(*src);
-         pRow[3] = 1.0;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-l8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r;
-         r = float_to_ubyte(pRow[0]);
-         *dst++ = (ubyte) r;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_A8_UNORM ***/
-
-static void
-a8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = 0.0;
-         pRow[3] = ubyte_to_float(*src);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned a;
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (ubyte) a;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-/*** PIPE_FORMAT_R16_SNORM ***/
-
-static void
-r16_get_tile_rgba(const short *src,
-                  unsigned w, unsigned h,
-                  float *p,
-                  unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] = SHORT_TO_FLOAT(src[0]);
-         pRow[1] =
-         pRow[2] = 0.0;
-         pRow[3] = 1.0;
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r16_put_tile_rgba(short *dst,
-                  unsigned w, unsigned h,
-                  const float *p,
-                  unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, dst++, pRow += 4) {
-         UNCLAMPED_FLOAT_TO_SHORT(dst[0], pRow[0]);
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_R16G16B16A16_SNORM ***/
-
-static void
-r16g16b16a16_get_tile_rgba(const short *src,
-                           unsigned w, unsigned h,
-                           float *p,
-                           unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src += 4, pRow += 4) {
-         pRow[0] = SHORT_TO_FLOAT(src[0]);
-         pRow[1] = SHORT_TO_FLOAT(src[1]);
-         pRow[2] = SHORT_TO_FLOAT(src[2]);
-         pRow[3] = SHORT_TO_FLOAT(src[3]);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-r16g16b16a16_put_tile_rgba(short *dst,
-                           unsigned w, unsigned h,
-                           const float *p,
-                           unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, dst += 4, pRow += 4) {
-         UNCLAMPED_FLOAT_TO_SHORT(dst[0], pRow[0]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[1], pRow[1]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[2], pRow[2]);
-         UNCLAMPED_FLOAT_TO_SHORT(dst[3], pRow[3]);
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_A8B8G8R8_SRGB ***/
-
-/**
- * Convert an 8-bit sRGB value from non-linear space to a
- * linear RGB value in [0, 1].
- * Implemented with a 256-entry lookup table.
- */
-static INLINE float
-srgb_to_linear(ubyte cs8)
-{
-   static float table[256];
-   static boolean tableReady = FALSE;
-   if (!tableReady) {
-      /* compute lookup table now */
-      uint i;
-      for (i = 0; i < 256; i++) {
-         const float cs = ubyte_to_float(i);
-         if (cs <= 0.04045) {
-            table[i] = cs / 12.92f;
-         }
-         else {
-            table[i] = (float) powf((cs + 0.055) / 1.055, 2.4);
-         }
-      }
-      tableReady = TRUE;
-   }
-   return table[cs8];
-}
-
-
-/**
- * Convert linear float in [0,1] to an srgb ubyte value in [0,255].
- * XXX this hasn't been tested (render to srgb surface).
- * XXX this needs optimization.
- */
-static INLINE ubyte
-linear_to_srgb(float cl)
-{
-   if (cl >= 1.0F)
-      return 255;
-   else if (cl >= 0.0031308F)
-      return float_to_ubyte(1.055F * powf(cl, 0.41666F) - 0.055F);
-   else if (cl > 0.0F)
-      return float_to_ubyte(12.92F * cl);
-   else
-      return 0.0;
-}
-
-
-static void
-a8r8g8b8_srgb_get_tile_rgba(const unsigned *src,
-                            unsigned w, unsigned h,
-                            float *p,
-                            unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         const unsigned pixel = *src++;
-         pRow[0] = srgb_to_linear((pixel >> 16) & 0xff);
-         pRow[1] = srgb_to_linear((pixel >>  8) & 0xff);
-         pRow[2] = srgb_to_linear((pixel >>  0) & 0xff);
-         pRow[3] = ubyte_to_float((pixel >> 24) & 0xff);
-      }
-      p += dst_stride;
-   }
-}
-
-static void
-a8r8g8b8_srgb_put_tile_rgba(unsigned *dst,
-                            unsigned w, unsigned h,
-                            const float *p,
-                            unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, g, b, a;
-         r = linear_to_srgb(pRow[0]);
-         g = linear_to_srgb(pRow[1]);
-         b = linear_to_srgb(pRow[2]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_L8A8_SRGB ***/
-
-static void
-a8l8_srgb_get_tile_rgba(const ushort *src,
-                        unsigned w, unsigned h,
-                        float *p,
-                        unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         ushort p = *src++;
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = srgb_to_linear(p & 0xff);
-         pRow[3] = ubyte_to_float(p >> 8);
-      }
-      p += dst_stride;
-   }
-}
-
-static void
-a8l8_srgb_put_tile_rgba(ushort *dst,
-                        unsigned w, unsigned h,
-                        const float *p,
-                        unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, a;
-         r = linear_to_srgb(pRow[0]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 8) | r;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_L8_SRGB ***/
-
-static void
-l8_srgb_get_tile_rgba(const ubyte *src,
-                      unsigned w, unsigned h,
-                      float *p,
-                      unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = srgb_to_linear(*src);
-         pRow[3] = 1.0;
-      }
-      p += dst_stride;
-   }
-}
-
-static void
-l8_srgb_put_tile_rgba(ubyte *dst,
-                      unsigned w, unsigned h,
-                      const float *p,
-                      unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r;
-         r = linear_to_srgb(pRow[0]);
-         *dst++ = (ubyte) r;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_I8_UNORM ***/
-
-static void
-i8_get_tile_rgba(const ubyte *src,
-                 unsigned w, unsigned h,
-                 float *p,
-                 unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, src++, pRow += 4) {
-         pRow[0] =
-         pRow[1] =
-         pRow[2] =
-         pRow[3] = ubyte_to_float(*src);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-i8_put_tile_rgba(ubyte *dst,
-                 unsigned w, unsigned h,
-                 const float *p,
-                 unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r;
-         r = float_to_ubyte(pRow[0]);
-         *dst++ = (ubyte) r;
-      }
-      p += src_stride;
-   }
-}
-
-
-/*** PIPE_FORMAT_L8A8_UNORM ***/
-
-static void
-a8l8_get_tile_rgba(const ushort *src,
-                   unsigned w, unsigned h,
-                   float *p,
-                   unsigned dst_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         ushort p = *src++;
-         pRow[0] =
-         pRow[1] =
-         pRow[2] = ubyte_to_float(p & 0xff);
-         pRow[3] = ubyte_to_float(p >> 8);
-      }
-      p += dst_stride;
-   }
-}
-
-
-static void
-a8l8_put_tile_rgba(ushort *dst,
-                   unsigned w, unsigned h,
-                   const float *p,
-                   unsigned src_stride)
-{
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      const float *pRow = p;
-      for (j = 0; j < w; j++, pRow += 4) {
-         unsigned r, a;
-         r = float_to_ubyte(pRow[0]);
-         a = float_to_ubyte(pRow[3]);
-         *dst++ = (a << 8) | r;
-      }
-      p += src_stride;
-   }
-}
-
-
-
-
 /*** PIPE_FORMAT_Z32_UNORM ***/
 
 /**
@@ -987,7 +164,7 @@ z32_get_tile_rgba(const unsigned *src,
 }
 
 
-/*** PIPE_FORMAT_Z24S8_UNORM ***/
+/*** PIPE_FORMAT_Z24_UNORM_S8_USCALED ***/
 
 /**
  * Return Z component as four float in [0,1].  Stencil part ignored.
@@ -1014,7 +191,7 @@ s8z24_get_tile_rgba(const unsigned *src,
 }
 
 
-/*** PIPE_FORMAT_S8Z24_UNORM ***/
+/*** PIPE_FORMAT_S8_USCALED_Z24_UNORM ***/
 
 /**
  * Return Z component as four float in [0,1].  Stencil part ignored.
@@ -1067,94 +244,6 @@ z32f_get_tile_rgba(const float *src,
 }
 
 
-/*** PIPE_FORMAT_UYVY / PIPE_FORMAT_YUYV ***/
-
-/**
- * Convert YCbCr (or YCrCb) to RGBA.
- */
-static void
-ycbcr_get_tile_rgba(const ushort *src,
-                    unsigned w, unsigned h,
-                    float *p,
-                    unsigned dst_stride,
-                    boolean rev)
-{
-   const float scale = 1.0f / 255.0f;
-   unsigned i, j;
-
-   for (i = 0; i < h; i++) {
-      float *pRow = p;
-      /* do two texels at a time */
-      for (j = 0; j < (w & ~1); j += 2, src += 2) {
-         const ushort t0 = src[0];
-         const ushort t1 = src[1];
-         const ubyte y0 = (t0 >> 8) & 0xff;  /* luminance */
-         const ubyte y1 = (t1 >> 8) & 0xff;  /* luminance */
-         ubyte cb, cr;
-         float r, g, b;
-
-         if (rev) {
-            cb = t1 & 0xff;         /* chroma U */
-            cr = t0 & 0xff;         /* chroma V */
-         }
-         else {
-            cb = t0 & 0xff;         /* chroma U */
-            cr = t1 & 0xff;         /* chroma V */
-         }
-
-         /* even pixel: y0,cr,cb */
-         r = 1.164f * (y0-16) + 1.596f * (cr-128);
-         g = 1.164f * (y0-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y0-16) + 2.018f * (cb-128);
-         pRow[0] = r * scale;
-         pRow[1] = g * scale;
-         pRow[2] = b * scale;
-         pRow[3] = 1.0f;
-         pRow += 4;
-
-         /* odd pixel: use y1,cr,cb */
-         r = 1.164f * (y1-16) + 1.596f * (cr-128);
-         g = 1.164f * (y1-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y1-16) + 2.018f * (cb-128);
-         pRow[0] = r * scale;
-         pRow[1] = g * scale;
-         pRow[2] = b * scale;
-         pRow[3] = 1.0f;
-         pRow += 4;
-
-      }
-      /* do the last texel */
-      if (w & 1) {
-         const ushort t0 = src[0];
-         const ushort t1 = src[1];
-         const ubyte y0 = (t0 >> 8) & 0xff;  /* luminance */
-         ubyte cb, cr;
-         float r, g, b;
-
-         if (rev) {
-            cb = t1 & 0xff;         /* chroma U */
-            cr = t0 & 0xff;         /* chroma V */
-         }
-         else {
-            cb = t0 & 0xff;         /* chroma U */
-            cr = t1 & 0xff;         /* chroma V */
-         }
-
-         /* even pixel: y0,cr,cb */
-         r = 1.164f * (y0-16) + 1.596f * (cr-128);
-         g = 1.164f * (y0-16) - 0.813f * (cr-128) - 0.391f * (cb-128);
-         b = 1.164f * (y0-16) + 2.018f * (cb-128);
-         pRow[0] = r * scale;
-         pRow[1] = g * scale;
-         pRow[2] = b * scale;
-         pRow[3] = 1.0f;
-         pRow += 4;
-      }
-      p += dst_stride;
-   }
-}
-
-
 void
 pipe_tile_raw_to_rgba(enum pipe_format format,
                       void *src,
@@ -1162,80 +251,23 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
                       float *dst, unsigned dst_stride)
 {
    switch (format) {
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      a8r8g8b8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
-      x8r8g8b8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      b8g8r8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
-      r8g8b8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5A1_UNORM:
-      a1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B4G4R4A4_UNORM:
-      a4r4g4b4_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B5G6R5_UNORM:
-      r5g6b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      r8g8b8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8A8_UNORM:
-      a8l8_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8A8_SRGB:
-      a8r8g8b8_srgb_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8A8_SRGB:
-      a8l8_srgb_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
-      break;
-   case PIPE_FORMAT_L8_SRGB:
-      l8_srgb_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
-      break;
    case PIPE_FORMAT_Z16_UNORM:
       z16_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
       break;
    case PIPE_FORMAT_Z32_UNORM:
       z32_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       s8z24_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       z24s8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
    case PIPE_FORMAT_Z32_FLOAT:
       z32f_get_tile_rgba((float *) src, w, h, dst, dst_stride);
       break;
-   case PIPE_FORMAT_UYVY:
-      ycbcr_get_tile_rgba((ushort *) src, w, h, dst, dst_stride, FALSE);
-      break;
-   case PIPE_FORMAT_YUYV:
-      ycbcr_get_tile_rgba((ushort *) src, w, h, dst, dst_stride, TRUE);
-      break;
    default:
       util_format_read_4f(format,
                           dst, dst_stride * sizeof(float),
@@ -1246,15 +278,16 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
 
 
 void
-pipe_get_tile_rgba(struct pipe_transfer *pt,
+pipe_get_tile_rgba(struct pipe_context *pipe,
+                   struct pipe_transfer *pt,
                    uint x, uint y, uint w, uint h,
                    float *p)
 {
    unsigned dst_stride = w * 4;
    void *packed;
-   enum pipe_format format = pt->texture->format;
+   enum pipe_format format = pt->resource->format;
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
    packed = MALLOC(util_format_get_nblocks(format, w, h) * util_format_get_blocksize(format));
@@ -1265,24 +298,94 @@ pipe_get_tile_rgba(struct pipe_transfer *pt,
    if(format == PIPE_FORMAT_UYVY || format == PIPE_FORMAT_YUYV)
       assert((x & 1) == 0);
 
-   pipe_get_tile_raw(pt, x, y, w, h, packed, 0);
+   pipe_get_tile_raw(pipe, pt, x, y, w, h, packed, 0);
+
+   pipe_tile_raw_to_rgba(format, packed, w, h, p, dst_stride);
+
+   FREE(packed);
+}
+
+
+void
+pipe_get_tile_swizzle(struct pipe_context *pipe,
+		      struct pipe_transfer *pt,
+                      uint x,
+                      uint y,
+                      uint w,
+                      uint h,
+                      uint swizzle_r,
+                      uint swizzle_g,
+                      uint swizzle_b,
+                      uint swizzle_a,
+                      enum pipe_format format,
+                      float *p)
+{
+   unsigned dst_stride = w * 4;
+   void *packed;
+   uint iy;
+   float rgba01[6];
+
+   if (u_clip_tile(x, y, &w, &h, &pt->box)) {
+      return;
+   }
+
+   packed = MALLOC(util_format_get_nblocks(format, w, h) * util_format_get_blocksize(format));
+   if (!packed) {
+      return;
+   }
+
+   if (format == PIPE_FORMAT_UYVY || format == PIPE_FORMAT_YUYV) {
+      assert((x & 1) == 0);
+   }
+
+   pipe_get_tile_raw(pipe, pt, x, y, w, h, packed, 0);
 
    pipe_tile_raw_to_rgba(format, packed, w, h, p, dst_stride);
 
    FREE(packed);
+
+   if (swizzle_r == PIPE_SWIZZLE_RED &&
+       swizzle_g == PIPE_SWIZZLE_GREEN &&
+       swizzle_b == PIPE_SWIZZLE_BLUE &&
+       swizzle_a == PIPE_SWIZZLE_ALPHA) {
+      /* no-op, skip */
+      return;
+   }
+
+   rgba01[PIPE_SWIZZLE_ZERO] = 0.0f;
+   rgba01[PIPE_SWIZZLE_ONE] = 1.0f;
+
+   for (iy = 0; iy < h; iy++) {
+      float *row = p;
+      uint ix;
+
+      for (ix = 0; ix < w; ix++) {
+         rgba01[PIPE_SWIZZLE_RED] = row[0];
+         rgba01[PIPE_SWIZZLE_GREEN] = row[1];
+         rgba01[PIPE_SWIZZLE_BLUE] = row[2];
+         rgba01[PIPE_SWIZZLE_ALPHA] = row[3];
+
+         *row++ = rgba01[swizzle_r];
+         *row++ = rgba01[swizzle_g];
+         *row++ = rgba01[swizzle_b];
+         *row++ = rgba01[swizzle_a];
+      }
+      p += dst_stride;
+   }
 }
 
 
 void
-pipe_put_tile_rgba(struct pipe_transfer *pt,
+pipe_put_tile_rgba(struct pipe_context *pipe,
+                   struct pipe_transfer *pt,
                    uint x, uint y, uint w, uint h,
                    const float *p)
 {
    unsigned src_stride = w * 4;
    void *packed;
-   enum pipe_format format = pt->texture->format;
+   enum pipe_format format = pt->resource->format;
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
    packed = MALLOC(util_format_get_nblocks(format, w, h) * util_format_get_blocksize(format));
@@ -1291,68 +394,17 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
       return;
 
    switch (format) {
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-      a8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
-      x8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-      b8g8r8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
-      r8g8b8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B5G5R5A1_UNORM:
-      a1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B5G6R5_UNORM:
-      r5g6b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R8G8B8_UNORM:
-      r8g8b8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B4G4R4A4_UNORM:
-      a4r4g4b4_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8_UNORM:
-      l8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_A8_UNORM:
-      a8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_I8_UNORM:
-      i8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8A8_UNORM:
-      a8l8_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R16_SNORM:
-      r16_put_tile_rgba((short *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      r16g16b16a16_put_tile_rgba((short *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_B8G8R8A8_SRGB:
-      a8r8g8b8_srgb_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8A8_SRGB:
-      a8l8_srgb_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
-      break;
-   case PIPE_FORMAT_L8_SRGB:
-      l8_srgb_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
-      break;
    case PIPE_FORMAT_Z16_UNORM:
       /*z16_put_tile_rgba((ushort *) packed, w, h, p, src_stride);*/
       break;
    case PIPE_FORMAT_Z32_UNORM:
       /*z32_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       /*s8z24_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       /*z24s8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
       break;
@@ -1363,7 +415,7 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
                            0, 0, w, h);
    }
 
-   pipe_put_tile_raw(pt, x, y, w, h, packed, 0);
+   pipe_put_tile_raw(pipe, pt, x, y, w, h, packed, 0);
 
    FREE(packed);
 }
@@ -1373,21 +425,21 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
  * Get a block of Z values, converted to 32-bit range.
  */
 void
-pipe_get_tile_z(struct pipe_transfer *pt,
+pipe_get_tile_z(struct pipe_context *pipe,
+                struct pipe_transfer *pt,
                 uint x, uint y, uint w, uint h,
                 uint *z)
 {
-   struct pipe_screen *screen = pt->texture->screen;
    const uint dstStride = w;
    ubyte *map;
    uint *pDest = z;
    uint i, j;
-   enum pipe_format format = pt->texture->format;
+   enum pipe_format format = pt->resource->format;
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
-   map = (ubyte *)screen->transfer_map(screen, pt);
+   map = (ubyte *)pipe->transfer_map(pipe, pt);
    if (!map) {
       assert(0);
       return;
@@ -1405,7 +457,7 @@ pipe_get_tile_z(struct pipe_transfer *pt,
          }
       }
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
    case PIPE_FORMAT_Z24X8_UNORM:
       {
          const uint *ptrc
@@ -1420,7 +472,7 @@ pipe_get_tile_z(struct pipe_transfer *pt,
          }
       }
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
    case PIPE_FORMAT_X8Z24_UNORM:
       {
          const uint *ptrc
@@ -1453,26 +505,26 @@ pipe_get_tile_z(struct pipe_transfer *pt,
       assert(0);
    }
 
-   screen->transfer_unmap(screen, pt);
+   pipe->transfer_unmap(pipe, pt);
 }
 
 
 void
-pipe_put_tile_z(struct pipe_transfer *pt,
+pipe_put_tile_z(struct pipe_context *pipe,
+                struct pipe_transfer *pt,
                 uint x, uint y, uint w, uint h,
                 const uint *zSrc)
 {
-   struct pipe_screen *screen = pt->texture->screen;
    const uint srcStride = w;
    const uint *ptrc = zSrc;
    ubyte *map;
    uint i, j;
-   enum pipe_format format = pt->texture->format;
+   enum pipe_format format = pt->resource->format;
 
-   if (pipe_clip_tile(x, y, &w, &h, pt))
+   if (u_clip_tile(x, y, &w, &h, &pt->box))
       return;
 
-   map = (ubyte *)screen->transfer_map(screen, pt);
+   map = (ubyte *)pipe->transfer_map(pipe, pt);
    if (!map) {
       assert(0);
       return;
@@ -1489,10 +541,10 @@ pipe_put_tile_z(struct pipe_transfer *pt,
          }
       }
       break;
-   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_Z24_UNORM_S8_USCALED:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
-         assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
+         /*assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);*/
          for (i = 0; i < h; i++) {
             for (j = 0; j < w; j++) {
                /* convert 32-bit Z to 24-bit Z, preserve stencil */
@@ -1516,10 +568,10 @@ pipe_put_tile_z(struct pipe_transfer *pt,
          }
       }
       break;
-   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
       {
          uint *pDest = (uint *) (map + y * pt->stride + x*4);
-         assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);
+         /*assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE);*/
          for (i = 0; i < h; i++) {
             for (j = 0; j < w; j++) {
                /* convert 32-bit Z to 24-bit Z, preserve stencil */
@@ -1560,7 +612,7 @@ pipe_put_tile_z(struct pipe_transfer *pt,
       assert(0);
    }
 
-   screen->transfer_unmap(screen, pt);
+   pipe->transfer_unmap(pipe, pt);
 }
 
 
diff --git a/src/gallium/auxiliary/util/u_tile.h b/src/gallium/auxiliary/util/u_tile.h
index 1453af38b8a..986eee07435 100644
--- a/src/gallium/auxiliary/util/u_tile.h
+++ b/src/gallium/auxiliary/util/u_tile.h
@@ -32,22 +32,24 @@
 
 struct pipe_transfer;
 
-
 /**
  * Clip tile against transfer dims.
+ *
+ * XXX: this only clips width and height!
+ *
  * \return TRUE if tile is totally clipped, FALSE otherwise
  */
 static INLINE boolean
-pipe_clip_tile(uint x, uint y, uint *w, uint *h, const struct pipe_transfer *pt)
+u_clip_tile(uint x, uint y, uint *w, uint *h, const struct pipe_box *box)
 {
-   if (x >= pt->width)
+   if (x >= box->width)
       return TRUE;
-   if (y >= pt->height)
+   if (y >= box->height)
       return TRUE;
-   if (x + *w > pt->width)
-      *w = pt->width - x;
-   if (y + *h > pt->height)
-      *h = pt->height - y;
+   if (x + *w > box->width)
+      *w = box->width - x;
+   if (y + *h > box->height)
+      *h = box->height - y;
    return FALSE;
 }
 
@@ -56,34 +58,54 @@ extern "C" {
 #endif
 
 void
-pipe_get_tile_raw(struct pipe_transfer *pt,
+pipe_get_tile_raw(struct pipe_context *pipe,
+                  struct pipe_transfer *pt,
                   uint x, uint y, uint w, uint h,
                   void *p, int dst_stride);
 
 void
-pipe_put_tile_raw(struct pipe_transfer *pt,
+pipe_put_tile_raw(struct pipe_context *pipe,
+                  struct pipe_transfer *pt,
                   uint x, uint y, uint w, uint h,
                   const void *p, int src_stride);
 
 
 void
-pipe_get_tile_rgba(struct pipe_transfer *pt,
+pipe_get_tile_rgba(struct pipe_context *pipe,
+                   struct pipe_transfer *pt,
                    uint x, uint y, uint w, uint h,
                    float *p);
 
 void
-pipe_put_tile_rgba(struct pipe_transfer *pt,
+pipe_get_tile_swizzle(struct pipe_context *pipe,
+		      struct pipe_transfer *pt,
+                      uint x,
+                      uint y,
+                      uint w,
+                      uint h,
+                      uint swizzle_r,
+                      uint swizzle_g,
+                      uint swizzle_b,
+                      uint swizzle_a,
+                      enum pipe_format format,
+                      float *p);
+
+void
+pipe_put_tile_rgba(struct pipe_context *pipe,
+                   struct pipe_transfer *pt,
                    uint x, uint y, uint w, uint h,
                    const float *p);
 
 
 void
-pipe_get_tile_z(struct pipe_transfer *pt,
+pipe_get_tile_z(struct pipe_context *pipe,
+                struct pipe_transfer *pt,
                 uint x, uint y, uint w, uint h,
                 uint *z);
 
 void
-pipe_put_tile_z(struct pipe_transfer *pt,
+pipe_put_tile_z(struct pipe_context *pipe,
+                struct pipe_transfer *pt,
                 uint x, uint y, uint w, uint h,
                 const uint *z);
 
diff --git a/src/gallium/auxiliary/util/u_timed_winsys.c b/src/gallium/auxiliary/util/u_timed_winsys.c
deleted file mode 100644
index d88298bc14c..00000000000
--- a/src/gallium/auxiliary/util/u_timed_winsys.c
+++ /dev/null
@@ -1,312 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * 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 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
- * THE COPYRIGHT HOLDERS, AUTHORS 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.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Keith Whitwell <keithw-at-tungstengraphics-dot-com>
- */
-
-#include "pipe/p_state.h"
-#include "util/u_simple_screen.h"
-#include "u_timed_winsys.h"
-#include "util/u_memory.h"
-#include "os/os_time.h"
-
-
-struct timed_winsys {
-   struct pipe_winsys base;
-   struct pipe_winsys *backend;
-   uint64_t last_dump;
-   struct {
-      const char *name_key;
-      double total;
-      unsigned calls;
-   } funcs[13];
-};
-
-
-static struct timed_winsys *timed_winsys( struct pipe_winsys *winsys )
-{
-   return (struct timed_winsys *)winsys;
-}
-
-
-static void time_display( struct pipe_winsys *winsys )
-{
-   struct timed_winsys *tws = timed_winsys(winsys);
-   unsigned i;
-   double overall = 0;
-
-   for (i = 0; i < Elements(tws->funcs); i++) {
-      if (tws->funcs[i].name_key) {
-         debug_printf("*** %-25s %5.3fms (%d calls, avg %.3fms)\n", 
-                      tws->funcs[i].name_key,
-                      tws->funcs[i].total,
-                      tws->funcs[i].calls,
-                      tws->funcs[i].total / tws->funcs[i].calls);
-         overall += tws->funcs[i].total;
-         tws->funcs[i].calls = 0;
-         tws->funcs[i].total = 0;
-      }
-   }
-
-   debug_printf("*** %-25s %5.3fms\n", 
-                "OVERALL WINSYS",
-                overall);
-}
-
-static void time_finish( struct pipe_winsys *winsys,
-                         long long startval, 
-                         unsigned idx,
-                         const char *name ) 
-{
-   struct timed_winsys *tws = timed_winsys(winsys);
-   int64_t endval = os_time_get();
-   double elapsed = (endval - startval)/1000.0;
-
-   if (endval - startval > 1000LL) 
-      debug_printf("*** %s %.3f\n", name, elapsed );
-
-   assert( tws->funcs[idx].name_key == name ||
-           tws->funcs[idx].name_key == NULL);
-
-   tws->funcs[idx].name_key = name;
-   tws->funcs[idx].total += elapsed;
-   tws->funcs[idx].calls++;
-
-   if (endval - tws->last_dump > 10LL * 1000LL * 1000LL) {
-      time_display( winsys );
-      tws->last_dump = endval;
-   }
-}
-
-
-/* Pipe has no concept of pools, but the psb driver passes a flag that
- * can be mapped onto pools in the backend.
- */
-static struct pipe_buffer *
-timed_buffer_create(struct pipe_winsys *winsys, 
-                    unsigned alignment, 
-                    unsigned usage, 
-                    unsigned size )
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   struct pipe_buffer *buf =
-      backend->buffer_create( backend, alignment, usage, size );
-
-   time_finish(winsys, start, 0, __FUNCTION__);
-   
-   return buf;
-}
-
-
-
-
-static struct pipe_buffer *
-timed_user_buffer_create(struct pipe_winsys *winsys,
-                             void *data, 
-                             unsigned bytes) 
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   struct pipe_buffer *buf = backend->user_buffer_create( backend, data, bytes );
-
-   time_finish(winsys, start, 1, __FUNCTION__);
-   
-   return buf;
-}
-
-
-static void *
-timed_buffer_map(struct pipe_winsys *winsys,
-                     struct pipe_buffer *buf,
-                     unsigned flags)
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   void *map = backend->buffer_map( backend, buf, flags );
-
-   time_finish(winsys, start, 2, __FUNCTION__);
-   
-   return map;
-}
-
-
-static void
-timed_buffer_unmap(struct pipe_winsys *winsys,
-                       struct pipe_buffer *buf)
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   backend->buffer_unmap( backend, buf );
-
-   time_finish(winsys, start, 3, __FUNCTION__);
-}
-
-
-static void
-timed_buffer_destroy(struct pipe_buffer *buf)
-{
-   struct pipe_winsys *winsys = buf->screen->winsys;
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   backend->buffer_destroy( buf );
-
-   time_finish(winsys, start, 4, __FUNCTION__);
-}
-
-
-static void
-timed_flush_frontbuffer( struct pipe_winsys *winsys,
-                         struct pipe_surface *surf,
-                         void *context_private)
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   backend->flush_frontbuffer( backend, surf, context_private );
-
-   time_finish(winsys, start, 5, __FUNCTION__);
-}
-
-
-
-
-static struct pipe_buffer *
-timed_surface_buffer_create(struct pipe_winsys *winsys,
-                              unsigned width, unsigned height,
-                              enum pipe_format format, 
-                              unsigned usage,
-                              unsigned tex_usage,
-                              unsigned *stride)
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   struct pipe_buffer *ret = backend->surface_buffer_create( backend, width, height, 
-                                                             format, usage, tex_usage, stride );
-
-   time_finish(winsys, start, 7, __FUNCTION__);
-   
-   return ret;
-}
-
-
-static const char *
-timed_get_name( struct pipe_winsys *winsys )
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   const char *ret = backend->get_name( backend );
-
-   time_finish(winsys, start, 9, __FUNCTION__);
-   
-   return ret;
-}
-
-static void
-timed_fence_reference(struct pipe_winsys *winsys,
-                    struct pipe_fence_handle **ptr,
-                    struct pipe_fence_handle *fence)
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   backend->fence_reference( backend, ptr, fence );
-
-   time_finish(winsys, start, 10, __FUNCTION__);
-}
-
-
-static int
-timed_fence_signalled( struct pipe_winsys *winsys,
-                       struct pipe_fence_handle *fence,
-                       unsigned flag )
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   int ret = backend->fence_signalled( backend, fence, flag );
-
-   time_finish(winsys, start, 11, __FUNCTION__);
-   
-   return ret;
-}
-
-static int
-timed_fence_finish( struct pipe_winsys *winsys,
-                     struct pipe_fence_handle *fence,
-                     unsigned flag )
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   int64_t start = os_time_get();
-
-   int ret = backend->fence_finish( backend, fence, flag );
-
-   time_finish(winsys, start, 12, __FUNCTION__);
-   
-   return ret;
-}
-
-static void
-timed_winsys_destroy( struct pipe_winsys *winsys )
-{
-   struct pipe_winsys *backend = timed_winsys(winsys)->backend;
-   backend->destroy( backend );
-   FREE(winsys);
-}
-
-
-
-struct pipe_winsys *u_timed_winsys_create( struct pipe_winsys *backend )
-{
-   struct timed_winsys *ws = CALLOC_STRUCT(timed_winsys);
-   
-   ws->base.user_buffer_create = timed_user_buffer_create;
-   ws->base.buffer_map = timed_buffer_map;
-   ws->base.buffer_unmap = timed_buffer_unmap;
-   ws->base.buffer_destroy = timed_buffer_destroy;
-   ws->base.buffer_create = timed_buffer_create;
-   ws->base.surface_buffer_create = timed_surface_buffer_create;
-   ws->base.flush_frontbuffer = timed_flush_frontbuffer;
-   ws->base.get_name = timed_get_name;
-   ws->base.fence_reference = timed_fence_reference;
-   ws->base.fence_signalled = timed_fence_signalled;
-   ws->base.fence_finish = timed_fence_finish;
-   ws->base.destroy = timed_winsys_destroy;
-   
-   ws->backend = backend;
-
-   return &ws->base;
-}
-
diff --git a/src/gallium/auxiliary/util/u_transfer.c b/src/gallium/auxiliary/util/u_transfer.c
new file mode 100644
index 00000000000..bedace3b1dc
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_transfer.c
@@ -0,0 +1,110 @@
+#include "pipe/p_context.h"
+#include "util/u_rect.h"
+#include "util/u_inlines.h"
+#include "util/u_transfer.h"
+#include "util/u_memory.h"
+
+/* One-shot transfer operation with data supplied in a user
+ * pointer.  XXX: strides??
+ */
+void u_default_transfer_inline_write( struct pipe_context *pipe,
+				      struct pipe_resource *resource,
+				      struct pipe_subresource sr,
+				      unsigned usage,
+				      const struct pipe_box *box,
+				      const void *data,
+				      unsigned stride,
+				      unsigned slice_stride)
+{
+   struct pipe_transfer *transfer = NULL;
+   uint8_t *map = NULL;
+
+   transfer = pipe->get_transfer(pipe, 
+				 resource,
+				 sr,
+				 usage,
+				 box );
+   if (transfer == NULL)
+      goto out;
+
+   map = pipe_transfer_map(pipe, transfer);
+   if (map == NULL)
+      goto out;
+
+   assert(box->depth == 1);	/* XXX: fix me */
+   
+   util_copy_rect(map,
+		  resource->format,
+		  transfer->stride, /* bytes? */
+		  0, 0,
+		  box->width,
+		  box->height,
+		  data,
+		  box->width,	/* bytes? texels? */
+		  0, 0);
+
+out:
+   if (map)
+      pipe_transfer_unmap(pipe, transfer);
+
+   if (transfer)
+      pipe_transfer_destroy(pipe, transfer);
+}
+
+
+boolean u_default_resource_get_handle(struct pipe_screen *screen,
+				      struct pipe_resource *resource,
+				      struct winsys_handle *handle)
+{
+   return FALSE;
+}
+
+
+
+void u_default_transfer_flush_region( struct pipe_context *pipe,
+				      struct pipe_transfer *transfer,
+				      const struct pipe_box *box)
+{
+   /* This is a no-op implementation, nothing to do.
+    */
+}
+
+unsigned u_default_is_resource_referenced( struct pipe_context *pipe,
+					   struct pipe_resource *resource,
+					unsigned face, unsigned level)
+{
+   return 0;
+}
+
+struct pipe_transfer * u_default_get_transfer(struct pipe_context *context,
+					      struct pipe_resource *resource,
+					      struct pipe_subresource sr,
+					      unsigned usage,
+					      const struct pipe_box *box)
+{
+   struct pipe_transfer *transfer = CALLOC_STRUCT(pipe_transfer);
+   if (transfer == NULL)
+      return NULL;
+
+   transfer->resource = resource;
+   transfer->sr = sr;
+   transfer->usage = usage;
+   transfer->box = *box;
+
+   /* Note strides are zero, this is ok for buffers, but not for
+    * textures 2d & higher at least. 
+    */
+   return transfer;
+}
+
+void u_default_transfer_unmap( struct pipe_context *pipe,
+			      struct pipe_transfer *transfer )
+{
+}
+
+void u_default_transfer_destroy(struct pipe_context *pipe,
+				struct pipe_transfer *transfer)
+{
+   FREE(transfer);
+}
+
diff --git a/src/gallium/auxiliary/util/u_transfer.h b/src/gallium/auxiliary/util/u_transfer.h
new file mode 100644
index 00000000000..eb07945d15f
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_transfer.h
@@ -0,0 +1,145 @@
+
+#ifndef U_TRANSFER_H
+#define U_TRANSFER_H
+
+/* Fallback implementations for inline read/writes which just go back
+ * to the regular transfer behaviour.
+ */
+#include "pipe/p_state.h"
+
+struct pipe_context;
+
+boolean u_default_resource_get_handle(struct pipe_screen *screen,
+				      struct pipe_resource *resource,
+				      struct winsys_handle *handle);
+
+void u_default_transfer_inline_write( struct pipe_context *pipe,
+			      struct pipe_resource *resource,
+			      struct pipe_subresource sr,
+			      unsigned usage,
+			      const struct pipe_box *box,
+			      const void *data,
+			      unsigned stride,
+			      unsigned slice_stride);
+
+void u_default_transfer_flush_region( struct pipe_context *pipe,
+				      struct pipe_transfer *transfer,
+				      const struct pipe_box *box);
+
+unsigned u_default_is_resource_referenced( struct pipe_context *pipe,
+					   struct pipe_resource *resource,
+					   unsigned face, unsigned level);
+
+struct pipe_transfer * u_default_get_transfer(struct pipe_context *context,
+					      struct pipe_resource *resource,
+					      struct pipe_subresource sr,
+					      unsigned usage,
+					      const struct pipe_box *box);
+
+void u_default_transfer_unmap( struct pipe_context *pipe,
+			       struct pipe_transfer *transfer );
+
+void u_default_transfer_destroy(struct pipe_context *pipe,
+				struct pipe_transfer *transfer);
+
+
+
+/* Useful helper to allow >1 implementation of resource functionality
+ * to exist in a single driver.  This is intended to be transitionary!
+ */
+struct u_resource_vtbl {
+
+   boolean (*resource_get_handle)(struct pipe_screen *,
+				  struct pipe_resource *tex,
+				  struct winsys_handle *handle);
+
+   void (*resource_destroy)(struct pipe_screen *,
+			    struct pipe_resource *pt);
+
+   unsigned (*is_resource_referenced)(struct pipe_context *pipe,
+				      struct pipe_resource *texture,
+				      unsigned face, unsigned level);
+
+   struct pipe_transfer *(*get_transfer)(struct pipe_context *,
+					 struct pipe_resource *resource,
+					 struct pipe_subresource,
+					 unsigned usage,
+					 const struct pipe_box *);
+
+   void (*transfer_destroy)(struct pipe_context *,
+			    struct pipe_transfer *);
+
+   void *(*transfer_map)( struct pipe_context *,
+			  struct pipe_transfer *transfer );
+
+   void (*transfer_flush_region)( struct pipe_context *,
+				  struct pipe_transfer *transfer,
+				  const struct pipe_box *);
+
+   void (*transfer_unmap)( struct pipe_context *,
+			   struct pipe_transfer *transfer );
+
+   void (*transfer_inline_write)( struct pipe_context *pipe,
+				  struct pipe_resource *resource,
+				  struct pipe_subresource sr,
+				  unsigned usage,
+				  const struct pipe_box *box,
+				  const void *data,
+				  unsigned stride,
+				  unsigned slice_stride);
+};
+
+
+struct u_resource {
+   struct pipe_resource b;
+   struct u_resource_vtbl *vtbl;
+};
+
+
+boolean u_resource_get_handle_vtbl(struct pipe_screen *screen,
+			      struct pipe_resource *resource,
+			      struct winsys_handle *handle);
+
+void u_resource_destroy_vtbl(struct pipe_screen *screen,
+			struct pipe_resource *resource);
+
+unsigned u_is_resource_referenced_vtbl( struct pipe_context *pipe,
+					struct pipe_resource *resource,
+					unsigned face, unsigned level);
+
+struct pipe_transfer *u_get_transfer_vtbl(struct pipe_context *context,
+				     struct pipe_resource *resource,
+				     struct pipe_subresource sr,
+				     unsigned usage,
+				     const struct pipe_box *box);
+
+void u_transfer_destroy_vtbl(struct pipe_context *pipe,
+			struct pipe_transfer *transfer);
+
+void *u_transfer_map_vtbl( struct pipe_context *pipe,
+		      struct pipe_transfer *transfer );
+
+void u_transfer_flush_region_vtbl( struct pipe_context *pipe,
+			      struct pipe_transfer *transfer,
+			      const struct pipe_box *box);
+
+void u_transfer_unmap_vtbl( struct pipe_context *rm_ctx,
+		       struct pipe_transfer *transfer );
+
+void u_transfer_inline_write_vtbl( struct pipe_context *rm_ctx,
+				   struct pipe_resource *resource,
+				   struct pipe_subresource sr,
+				   unsigned usage,
+				   const struct pipe_box *box,
+				   const void *data,
+				   unsigned stride,
+				   unsigned slice_stride);
+
+
+
+
+
+
+
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_upload_mgr.c b/src/gallium/auxiliary/util/u_upload_mgr.c
index 012b2ae2336..75d44432d9e 100644
--- a/src/gallium/auxiliary/util/u_upload_mgr.c
+++ b/src/gallium/auxiliary/util/u_upload_mgr.c
@@ -31,7 +31,7 @@
 
 #include "pipe/p_defines.h"
 #include "util/u_inlines.h"
-#include "pipe/p_screen.h"
+#include "pipe/p_context.h"
 #include "util/u_memory.h"
 #include "util/u_math.h"
 
@@ -39,7 +39,7 @@
 
 
 struct u_upload_mgr {
-   struct pipe_screen *screen;
+   struct pipe_context *pipe;
 
    unsigned default_size;
    unsigned alignment;
@@ -47,21 +47,21 @@ struct u_upload_mgr {
 
    /* The active buffer:
     */
-   struct pipe_buffer *buffer;
+   struct pipe_resource *buffer;
    unsigned size;
    unsigned offset;
 };
 
 
-struct u_upload_mgr *u_upload_create( struct pipe_screen *screen,
+struct u_upload_mgr *u_upload_create( struct pipe_context *pipe,
                                       unsigned default_size,
                                       unsigned alignment,
                                       unsigned usage )
 {
    struct u_upload_mgr *upload = CALLOC_STRUCT( u_upload_mgr );
 
+   upload->pipe = pipe;
    upload->default_size = default_size;
-   upload->screen = screen;
    upload->alignment = alignment;
    upload->usage = usage;
    upload->buffer = NULL;
@@ -69,31 +69,44 @@ struct u_upload_mgr *u_upload_create( struct pipe_screen *screen,
    return upload;
 }
 
-
+/* Slightly specialized version of buffer_write designed to maximize
+ * chances of the driver consolidating successive writes into a single
+ * upload.
+ *
+ * dirty_size may be slightly greater than size to cope with
+ * alignment.  We don't want to leave holes between succesively mapped
+ * regions as that may prevent the driver from consolidating uploads.
+ * 
+ * Note that the 'data' pointer has probably come from the application
+ * and we cannot read even a byte past its end without risking
+ * segfaults, or at least complaints from valgrind..
+ */
 static INLINE enum pipe_error
-my_buffer_write(struct pipe_screen *screen,
-                struct pipe_buffer *buf,
+my_buffer_write(struct pipe_context *pipe,
+                struct pipe_resource *buf,
                 unsigned offset, unsigned size, unsigned dirty_size,
                 const void *data)
 {
+   struct pipe_transfer *transfer = NULL;
    uint8_t *map;
    
-   assert(offset < buf->size);
-   assert(offset + size <= buf->size);
+   assert(offset < buf->width0);
+   assert(offset + size <= buf->width0);
    assert(dirty_size >= size);
    assert(size);
 
-   map = pipe_buffer_map_range(screen, buf, offset, size, 
-                               PIPE_BUFFER_USAGE_CPU_WRITE |
-                               PIPE_BUFFER_USAGE_FLUSH_EXPLICIT |
-                               PIPE_BUFFER_USAGE_DISCARD |
-                               PIPE_BUFFER_USAGE_UNSYNCHRONIZED);
+   map = pipe_buffer_map_range(pipe, buf, offset, dirty_size,
+                               PIPE_TRANSFER_WRITE |
+                               PIPE_TRANSFER_FLUSH_EXPLICIT |
+                               PIPE_TRANSFER_DISCARD |
+                               PIPE_TRANSFER_UNSYNCHRONIZED,
+			       &transfer);
    if (map == NULL) 
       return PIPE_ERROR_OUT_OF_MEMORY;
 
    memcpy(map + offset, data, size);
-   pipe_buffer_flush_mapped_range(screen, buf, offset, dirty_size);
-   pipe_buffer_unmap(screen, buf);
+   pipe_buffer_flush_mapped_range(pipe, transfer, offset, dirty_size);
+   pipe_buffer_unmap(pipe, buf, transfer);
 
    return PIPE_OK;
 }
@@ -109,7 +122,7 @@ my_buffer_write(struct pipe_screen *screen,
  */
 void u_upload_flush( struct u_upload_mgr *upload )
 {
-   pipe_buffer_reference( &upload->buffer, NULL );
+   pipe_resource_reference( &upload->buffer, NULL );
    upload->size = 0;
 }
 
@@ -135,9 +148,8 @@ u_upload_alloc_buffer( struct u_upload_mgr *upload,
     */
    size = align(MAX2(upload->default_size, min_size), 4096);
 
-   upload->buffer = pipe_buffer_create( upload->screen,
-                                        upload->alignment,
-                                        upload->usage | PIPE_BUFFER_USAGE_CPU_WRITE,
+   upload->buffer = pipe_buffer_create( upload->pipe->screen,
+                                        upload->usage,
                                         size );
    if (upload->buffer == NULL) 
       goto fail;
@@ -149,7 +161,7 @@ u_upload_alloc_buffer( struct u_upload_mgr *upload,
 
 fail:
    if (upload->buffer)
-      pipe_buffer_reference( &upload->buffer, NULL );
+      pipe_resource_reference( &upload->buffer, NULL );
 
    return PIPE_ERROR_OUT_OF_MEMORY;
 }
@@ -159,7 +171,7 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
                                unsigned size,
                                const void *data,
                                unsigned *out_offset,
-                               struct pipe_buffer **outbuf )
+                               struct pipe_resource **outbuf )
 {
    unsigned alloc_size = align( size, upload->alignment );
    enum pipe_error ret = PIPE_OK;
@@ -172,7 +184,7 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
 
    /* Copy the data, using map_range if available:
     */
-   ret = my_buffer_write( upload->screen, 
+   ret = my_buffer_write( upload->pipe, 
                           upload->buffer,
                           upload->offset,
                           size, 
@@ -183,7 +195,7 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
 
    /* Emit the return values:
     */
-   pipe_buffer_reference( outbuf, upload->buffer );
+   pipe_resource_reference( outbuf, upload->buffer );
    *out_offset = upload->offset;
    upload->offset += alloc_size;
    return PIPE_OK;
@@ -198,15 +210,18 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
 enum pipe_error u_upload_buffer( struct u_upload_mgr *upload,
                                  unsigned offset,
                                  unsigned size,
-                                 struct pipe_buffer *inbuf,
+                                 struct pipe_resource *inbuf,
                                  unsigned *out_offset,
-                                 struct pipe_buffer **outbuf )
+                                 struct pipe_resource **outbuf )
 {
    enum pipe_error ret = PIPE_OK;
+   struct pipe_transfer *transfer = NULL;
    const char *map = NULL;
 
-   map = (const char *)pipe_buffer_map( 
-      upload->screen, inbuf, PIPE_BUFFER_USAGE_CPU_READ );
+   map = (const char *)pipe_buffer_map(upload->pipe,
+				       inbuf,
+				       PIPE_TRANSFER_READ,
+				       &transfer);
 
    if (map == NULL) {
       ret = PIPE_ERROR_OUT_OF_MEMORY;
@@ -226,7 +241,7 @@ enum pipe_error u_upload_buffer( struct u_upload_mgr *upload,
 
 done:
    if (map)
-      pipe_buffer_unmap( upload->screen, inbuf );
+      pipe_buffer_unmap( upload->pipe, inbuf, transfer );
 
    return ret;
 }
diff --git a/src/gallium/auxiliary/util/u_upload_mgr.h b/src/gallium/auxiliary/util/u_upload_mgr.h
index e158bed9d04..a124924fc80 100644
--- a/src/gallium/auxiliary/util/u_upload_mgr.h
+++ b/src/gallium/auxiliary/util/u_upload_mgr.h
@@ -35,11 +35,11 @@
 #include "pipe/p_defines.h"
 
 struct pipe_screen;
-struct pipe_buffer;
+struct pipe_resource;
 struct u_upload_mgr;
 
 
-struct u_upload_mgr *u_upload_create( struct pipe_screen *screen,
+struct u_upload_mgr *u_upload_create( struct pipe_context *pipe,
                                       unsigned default_size,
                                       unsigned alignment,
                                       unsigned usage );
@@ -61,15 +61,15 @@ enum pipe_error u_upload_data( struct u_upload_mgr *upload,
                                unsigned size,
                                const void *data,
                                unsigned *out_offset,
-                               struct pipe_buffer **outbuf );
+                               struct pipe_resource **outbuf );
 
 
 enum pipe_error u_upload_buffer( struct u_upload_mgr *upload,
                                  unsigned offset,
                                  unsigned size,
-                                 struct pipe_buffer *inbuf,
+                                 struct pipe_resource *inbuf,
                                  unsigned *out_offset,
-                                 struct pipe_buffer **outbuf );
+                                 struct pipe_resource **outbuf );
author	Younes Manton <[email protected]>	2010-04-30 20:42:30 -0400
committer	Younes Manton <[email protected]>	2010-04-30 20:42:30 -0400
commit	a8ea1dacc63ac567498049e5756c247b9fec6cd9 (patch)
tree	4031e2e2b6166bd926b43fa4bbb3aab773a30ee5 /src/gallium/auxiliary/util
parent	404fb63b4649f58fce443615e49337d42b8ddece (diff)
parent	35d960cc744c374ccaad48c3d80559b59c74e28a (diff)