Merge remote-tracking branch 'origin/released'

* origin/released:
  libxtrans fontconfig glproto libX11 libxcb xcbproto mesa xserver pixman xkeyboard-config git update 22 Dec 2013

Conflicts:
	mesalib/include/GL/glext.h
	mesalib/src/mesa/drivers/dri/common/dri_util.c
	mesalib/src/mesa/drivers/dri/swrast/swrast.c
	xorg-server/damageext/damageext.c
	xorg-server/dix/dispatch.c
	xorg-server/glx/glxdriswrast.c
	xorg-server/glx/indirect_dispatch.c
	xorg-server/glx/indirect_dispatch_swap.c
	xorg-server/glx/indirect_program.c
	xorg-server/glx/render2.c
	xorg-server/glx/render2swap.c
	xorg-server/hw/xwin/glx/gen_gl_wrappers.py
	xorg-server/hw/xwin/glx/glthunk.c
	xorg-server/hw/xwin/glx/indirect.c
	xorg-server/include/os.h
	xorg-server/present/present_request.c

140 files changed, 4348 insertions, 4275 deletions

diff --git a/mesalib/src/gallium/auxiliary/util/u_blitter.c b/mesalib/src/gallium/auxiliary/util/u_blitter.c
index b95cbab12..9246bd722 100644
--- a/mesalib/src/gallium/auxiliary/util/u_blitter.c
+++ b/mesalib/src/gallium/auxiliary/util/u_blitter.c
@@ -51,6 +51,12 @@
 
 #define INVALID_PTR ((void*)~0)
 
+#define GET_CLEAR_BLEND_STATE_IDX(clear_buffers) \
+   ((clear_buffers) / PIPE_CLEAR_COLOR0)
+
+#define NUM_RESOLVE_FRAG_SHADERS 5 /* MSAA 2x, 4x, 8x, 16x, 32x */
+#define GET_MSAA_RESOLVE_FS_IDX(nr_samples) (util_logbase2(nr_samples)-1)
+
 struct blitter_context_priv
 {
    struct blitter_context base;
@@ -65,6 +71,7 @@ struct blitter_context_priv
    /* Vertex shaders. */
    void *vs; /**< Vertex shader which passes {pos, generic} to the output.*/
    void *vs_pos_only; /**< Vertex shader which passes pos to the output.*/
+   void *vs_layered; /**< Vertex shader which sets LAYER = INSTANCEID. */
 
    /* Fragment shaders. */
    void *fs_empty;
@@ -87,8 +94,14 @@ struct blitter_context_priv
    void *fs_texfetch_depthstencil_msaa[PIPE_MAX_TEXTURE_TYPES];
    void *fs_texfetch_stencil_msaa[PIPE_MAX_TEXTURE_TYPES];
 
+   /* FS which outputs an average of all samples. */
+   void *fs_resolve[PIPE_MAX_TEXTURE_TYPES][NUM_RESOLVE_FRAG_SHADERS][2];
+   void *fs_resolve_sint[PIPE_MAX_TEXTURE_TYPES][NUM_RESOLVE_FRAG_SHADERS][2];
+   void *fs_resolve_uint[PIPE_MAX_TEXTURE_TYPES][NUM_RESOLVE_FRAG_SHADERS][2];
+
    /* Blend state. */
    void *blend[PIPE_MASK_RGBA+1]; /**< blend state with writemask */
+   void *blend_clear[GET_CLEAR_BLEND_STATE_IDX(PIPE_CLEAR_COLOR)+1];
 
    /* Depth stencil alpha state. */
    void *dsa_write_depth_stencil;
@@ -295,6 +308,7 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
          util_make_vertex_passthrough_shader(pipe, 2, semantic_names,
                                              semantic_indices);
    }
+
    if (ctx->has_stream_out) {
       struct pipe_stream_output_info so;
       const uint semantic_names[] = { TGSI_SEMANTIC_POSITION };
@@ -310,6 +324,11 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe)
                                                      semantic_indices, &so);
    }
 
+   if (pipe->screen->get_param(pipe->screen, PIPE_CAP_TGSI_INSTANCEID) &&
+       pipe->screen->get_param(pipe->screen, PIPE_CAP_TGSI_VS_LAYER)) {
+      ctx->vs_layered = util_make_layered_clear_vertex_shader(pipe);
+   }
+
    /* set invariant vertex coordinates */
    for (i = 0; i < 4; i++)
       ctx->vertices[i][0][3] = 1; /*v.w*/
@@ -323,11 +342,15 @@ void util_blitter_destroy(struct blitter_context *blitter)
 {
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_context *pipe = blitter->pipe;
-   int i;
+   int i, j, f;
 
    for (i = 0; i <= PIPE_MASK_RGBA; i++) {
       pipe->delete_blend_state(pipe, ctx->blend[i]);
    }
+   for (i = 0; i < Elements(ctx->blend_clear); i++) {
+      if (ctx->blend_clear[i])
+         pipe->delete_blend_state(pipe, ctx->blend_clear[i]);
+   }
    pipe->delete_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
    pipe->delete_depth_stencil_alpha_state(pipe,
                                           ctx->dsa_write_depth_keep_stencil);
@@ -357,7 +380,23 @@ void util_blitter_destroy(struct blitter_context *blitter)
          ctx->delete_fs_state(pipe, ctx->fs_texfetch_depthstencil[i]);
       if (ctx->fs_texfetch_stencil[i])
          ctx->delete_fs_state(pipe, ctx->fs_texfetch_stencil[i]);
+
+      for (j = 0; j< Elements(ctx->fs_resolve[i]); j++)
+         for (f = 0; f < 2; f++)
+            if (ctx->fs_resolve[i][j][f])
+               ctx->delete_fs_state(pipe, ctx->fs_resolve[i][j][f]);
+
+      for (j = 0; j< Elements(ctx->fs_resolve_sint[i]); j++)
+         for (f = 0; f < 2; f++)
+            if (ctx->fs_resolve_sint[i][j][f])
+               ctx->delete_fs_state(pipe, ctx->fs_resolve_sint[i][j][f]);
+
+      for (j = 0; j< Elements(ctx->fs_resolve_uint[i]); j++)
+         for (f = 0; f < 2; f++)
+            if (ctx->fs_resolve_uint[i][j][f])
+               ctx->delete_fs_state(pipe, ctx->fs_resolve_uint[i][j][f]);
    }
+
    ctx->delete_fs_state(pipe, ctx->fs_empty);
    ctx->delete_fs_state(pipe, ctx->fs_write_one_cbuf);
    ctx->delete_fs_state(pipe, ctx->fs_write_all_cbufs);
@@ -711,22 +750,60 @@ static void blitter_set_dst_dimensions(struct blitter_context_priv *ctx,
 }
 
 static void *blitter_get_fs_texfetch_col(struct blitter_context_priv *ctx,
+                                         enum pipe_format format,
                                          enum pipe_texture_target target,
-                                         unsigned nr_samples)
+                                         unsigned src_nr_samples,
+                                         unsigned dst_nr_samples,
+                                         unsigned filter)
 {
    struct pipe_context *pipe = ctx->base.pipe;
+   unsigned tgsi_tex = util_pipe_tex_to_tgsi_tex(target, src_nr_samples);
 
    assert(target < PIPE_MAX_TEXTURE_TYPES);
 
-   if (nr_samples > 1) {
-      void **shader = &ctx->fs_texfetch_col_msaa[target];
+   if (src_nr_samples > 1) {
+      void **shader;
 
-      /* Create the fragment shader on-demand. */
-      if (!*shader) {
-         unsigned tgsi_tex = util_pipe_tex_to_tgsi_tex(target,
-                                                       nr_samples);
+      if (dst_nr_samples <= 1) {
+         /* The destination has one sample, so we'll do color resolve. */
+         boolean is_uint, is_sint;
+         unsigned index = GET_MSAA_RESOLVE_FS_IDX(src_nr_samples);
 
-         *shader = util_make_fs_blit_msaa_color(pipe, tgsi_tex);
+         is_uint = util_format_is_pure_uint(format);
+         is_sint = util_format_is_pure_sint(format);
+
+         assert(filter < 2);
+
+         if (is_uint)
+            shader = &ctx->fs_resolve_uint[target][index][filter];
+         else if (is_sint)
+            shader = &ctx->fs_resolve_sint[target][index][filter];
+         else
+            shader = &ctx->fs_resolve[target][index][filter];
+
+         if (!*shader) {
+            if (filter == PIPE_TEX_FILTER_LINEAR) {
+               *shader = util_make_fs_msaa_resolve_bilinear(pipe, tgsi_tex,
+                                                   src_nr_samples,
+                                                   is_uint, is_sint);
+            }
+            else {
+               *shader = util_make_fs_msaa_resolve(pipe, tgsi_tex,
+                                                   src_nr_samples,
+                                                   is_uint, is_sint);
+            }
+         }
+      }
+      else {
+         /* The destination has multiple samples, we'll do
+          * an MSAA->MSAA copy.
+          */
+         shader = &ctx->fs_texfetch_col_msaa[target];
+
+         /* Create the fragment shader on-demand. */
+         if (!*shader) {
+            *shader = util_make_fs_blit_msaa_color(pipe, tgsi_tex);
+         }
       }
 
       return *shader;
@@ -735,11 +812,8 @@ static void *blitter_get_fs_texfetch_col(struct blitter_context_priv *ctx,
 
       /* Create the fragment shader on-demand. */
       if (!*shader) {
-         unsigned tgsi_tex = util_pipe_tex_to_tgsi_tex(target, 0);
-
-         *shader =
-            util_make_fragment_tex_shader(pipe, tgsi_tex,
-                                          TGSI_INTERPOLATE_LINEAR);
+         *shader = util_make_fragment_tex_shader(pipe, tgsi_tex,
+                                                 TGSI_INTERPOLATE_LINEAR);
       }
 
       return *shader;
@@ -864,7 +938,7 @@ void util_blitter_cache_all_shaders(struct blitter_context *blitter)
 {
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_screen *screen = blitter->pipe->screen;
-   unsigned i, target, max_samples;
+   unsigned samples, j, f, target, max_samples;
    boolean has_arraytex, has_cubearraytex;
 
    max_samples = ctx->has_texture_multisample ? 2 : 1;
@@ -874,7 +948,7 @@ void util_blitter_cache_all_shaders(struct blitter_context *blitter)
                                     PIPE_CAP_CUBE_MAP_ARRAY) != 0;
 
    /* It only matters if i <= 1 or > 1. */
-   for (i = 1; i <= max_samples; i++) {
+   for (samples = 1; samples <= max_samples; samples++) {
       for (target = PIPE_TEXTURE_1D; target < PIPE_MAX_TEXTURE_TYPES; target++) {
          if (!has_arraytex &&
              (target == PIPE_TEXTURE_1D_ARRAY ||
@@ -885,29 +959,55 @@ void util_blitter_cache_all_shaders(struct blitter_context *blitter)
              (target == PIPE_TEXTURE_CUBE_ARRAY))
             continue;
 
-	 if (i > 1 &&
+	 if (samples > 1 &&
 	     (target != PIPE_TEXTURE_2D &&
 	      target != PIPE_TEXTURE_2D_ARRAY))
 	    continue;
 
-         blitter_get_fs_texfetch_col(ctx, target, i);
-         blitter_get_fs_texfetch_depth(ctx, target, i);
+         /* If samples == 1, the shaders read one texel. If samples >= 1,
+          * they read one sample.
+          */
+         blitter_get_fs_texfetch_col(ctx, PIPE_FORMAT_R32_FLOAT, target,
+                                     samples, samples, 0);
+         blitter_get_fs_texfetch_depth(ctx, target, samples);
          if (ctx->has_stencil_export) {
-            blitter_get_fs_texfetch_depthstencil(ctx, target, i);
-            blitter_get_fs_texfetch_stencil(ctx, target, i);
+            blitter_get_fs_texfetch_depthstencil(ctx, target, samples);
+            blitter_get_fs_texfetch_stencil(ctx, target, samples);
+         }
+
+         if (samples == 1)
+            continue;
+
+         /* MSAA resolve shaders. */
+         for (j = 2; j < 32; j++) {
+            if (!screen->is_format_supported(screen, PIPE_FORMAT_R32_FLOAT,
+                                             target, j,
+                                             PIPE_BIND_SAMPLER_VIEW)) {
+               continue;
+            }
+
+            for (f = 0; f < 2; f++) {
+               blitter_get_fs_texfetch_col(ctx, PIPE_FORMAT_R32_FLOAT, target,
+                                           j, 1, f);
+               blitter_get_fs_texfetch_col(ctx, PIPE_FORMAT_R32_UINT, target,
+                                           j, 1, f);
+               blitter_get_fs_texfetch_col(ctx, PIPE_FORMAT_R32_SINT, target,
+                                           j, 1, f);
+            }
          }
       }
    }
 }
 
 static void blitter_set_common_draw_rect_state(struct blitter_context_priv *ctx,
-                                               boolean scissor)
+                                               boolean scissor,
+                                               boolean vs_layered)
 {
    struct pipe_context *pipe = ctx->base.pipe;
 
    pipe->bind_rasterizer_state(pipe, scissor ? ctx->rs_state_scissor
                                              : ctx->rs_state);
-   pipe->bind_vs_state(pipe, ctx->vs);
+   pipe->bind_vs_state(pipe, vs_layered ? ctx->vs_layered : ctx->vs);
    if (ctx->has_geometry_shader)
       pipe->bind_gs_state(pipe, NULL);
    if (ctx->has_stream_out)
@@ -915,19 +1015,24 @@ static void blitter_set_common_draw_rect_state(struct blitter_context_priv *ctx,
 }
 
 static void blitter_draw(struct blitter_context_priv *ctx,
-                         int x1, int y1, int x2, int y2, float depth)
+                         int x1, int y1, int x2, int y2, float depth,
+                         unsigned num_instances)
 {
-   struct pipe_resource *buf = NULL;
-   unsigned offset = 0;
+   struct pipe_context *pipe = ctx->base.pipe;
+   struct pipe_vertex_buffer vb = {0};
 
    blitter_set_rectangle(ctx, x1, y1, x2, y2, depth);
 
+   vb.stride = 8 * sizeof(float);
+
    u_upload_data(ctx->upload, 0, sizeof(ctx->vertices), ctx->vertices,
-                 &offset, &buf);
+                 &vb.buffer_offset, &vb.buffer);
    u_upload_unmap(ctx->upload);
-   util_draw_vertex_buffer(ctx->base.pipe, NULL, buf, ctx->base.vb_slot,
-                           offset, PIPE_PRIM_TRIANGLE_FAN, 4, 2);
-   pipe_resource_reference(&buf, NULL);
+
+   pipe->set_vertex_buffers(pipe, ctx->base.vb_slot, 1, &vb);
+   util_draw_arrays_instanced(pipe, PIPE_PRIM_TRIANGLE_FAN, 0, 4,
+                              0, num_instances);
+   pipe_resource_reference(&vb.buffer, NULL);
 }
 
 void util_blitter_draw_rectangle(struct blitter_context *blitter,
@@ -949,11 +1054,47 @@ void util_blitter_draw_rectangle(struct blitter_context *blitter,
       default:;
    }
 
-   blitter_draw(ctx, x1, y1, x2, y2, depth);
+   blitter_draw(ctx, x1, y1, x2, y2, depth, 1);
+}
+
+static void *get_clear_blend_state(struct blitter_context_priv *ctx,
+                                   unsigned clear_buffers)
+{
+   struct pipe_context *pipe = ctx->base.pipe;
+   int index;
+
+   clear_buffers &= PIPE_CLEAR_COLOR;
+
+   /* Return an existing blend state. */
+   if (!clear_buffers)
+      return ctx->blend[0];
+
+   index = GET_CLEAR_BLEND_STATE_IDX(clear_buffers);
+
+   if (ctx->blend_clear[index])
+      return ctx->blend_clear[index];
+
+   /* Create a new one. */
+   {
+      struct pipe_blend_state blend = {0};
+      unsigned i;
+
+      blend.independent_blend_enable = 1;
+
+      for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
+         if (clear_buffers & (PIPE_CLEAR_COLOR0 << i)) {
+            blend.rt[i].colormask = PIPE_MASK_RGBA;
+         }
+      }
+
+      ctx->blend_clear[index] = pipe->create_blend_state(pipe, &blend);
+   }
+   return ctx->blend_clear[index];
 }
 
 static void util_blitter_clear_custom(struct blitter_context *blitter,
                                       unsigned width, unsigned height,
+                                      unsigned num_layers,
                                       unsigned clear_buffers,
                                       const union pipe_color_union *color,
                                       double depth, unsigned stencil,
@@ -963,6 +1104,8 @@ static void util_blitter_clear_custom(struct blitter_context *blitter,
    struct pipe_context *pipe = ctx->base.pipe;
    struct pipe_stencil_ref sr = { { 0 } };
 
+   assert(ctx->vs_layered || num_layers <= 1);
+
    blitter_set_running_flag(ctx);
    blitter_check_saved_vertex_states(ctx);
    blitter_check_saved_fragment_states(ctx);
@@ -971,10 +1114,8 @@ static void util_blitter_clear_custom(struct blitter_context *blitter,
    /* bind states */
    if (custom_blend) {
       pipe->bind_blend_state(pipe, custom_blend);
-   } else if (clear_buffers & PIPE_CLEAR_COLOR) {
-      pipe->bind_blend_state(pipe, ctx->blend[PIPE_MASK_RGBA]);
    } else {
-      pipe->bind_blend_state(pipe, ctx->blend[0]);
+      pipe->bind_blend_state(pipe, get_clear_blend_state(ctx, clear_buffers));
    }
 
    if (custom_dsa) {
@@ -996,10 +1137,18 @@ static void util_blitter_clear_custom(struct blitter_context *blitter,
    ctx->bind_fs_state(pipe, ctx->fs_write_all_cbufs);
    pipe->set_sample_mask(pipe, ~0);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
    blitter_set_dst_dimensions(ctx, width, height);
-   blitter->draw_rectangle(blitter, 0, 0, width, height, (float) depth,
-                           UTIL_BLITTER_ATTRIB_COLOR, color);
+
+   if (num_layers > 1 && ctx->vs_layered) {
+      blitter_set_common_draw_rect_state(ctx, FALSE, TRUE);
+      blitter_set_clear_color(ctx, color);
+      blitter_draw(ctx, 0, 0, width, height, depth, num_layers);
+   }
+   else {
+      blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
+      blitter->draw_rectangle(blitter, 0, 0, width, height, (float) depth,
+                              UTIL_BLITTER_ATTRIB_COLOR, color);
+   }
 
    blitter_restore_vertex_states(ctx);
    blitter_restore_fragment_states(ctx);
@@ -1008,12 +1157,12 @@ static void util_blitter_clear_custom(struct blitter_context *blitter,
 }
 
 void util_blitter_clear(struct blitter_context *blitter,
-                        unsigned width, unsigned height,
+                        unsigned width, unsigned height, unsigned num_layers,
                         unsigned clear_buffers,
                         const union pipe_color_union *color,
                         double depth, unsigned stencil)
 {
-   util_blitter_clear_custom(blitter, width, height,
+   util_blitter_clear_custom(blitter, width, height, num_layers,
                              clear_buffers, color, depth, stencil,
                              NULL, NULL);
 }
@@ -1023,7 +1172,7 @@ void util_blitter_custom_clear_depth(struct blitter_context *blitter,
                                      double depth, void *custom_dsa)
 {
     static const union pipe_color_union color;
-    util_blitter_clear_custom(blitter, width, height, 0, &color, depth, 0,
+    util_blitter_clear_custom(blitter, width, height, 0, 0, &color, depth, 0,
                               NULL, custom_dsa);
 }
 
@@ -1137,11 +1286,10 @@ static boolean is_blit_generic_supported(struct blitter_context *blitter,
 
 boolean util_blitter_is_copy_supported(struct blitter_context *blitter,
                                        const struct pipe_resource *dst,
-                                       const struct pipe_resource *src,
-                                       unsigned mask)
+                                       const struct pipe_resource *src)
 {
    return is_blit_generic_supported(blitter, dst, dst->format,
-                                    src, src->format, mask);
+                                    src, src->format, PIPE_MASK_RGBAZS);
 }
 
 boolean util_blitter_is_blit_supported(struct blitter_context *blitter,
@@ -1159,8 +1307,7 @@ void util_blitter_copy_texture(struct blitter_context *blitter,
                                unsigned dstx, unsigned dsty, unsigned dstz,
                                struct pipe_resource *src,
                                unsigned src_level,
-                               const struct pipe_box *srcbox, unsigned mask,
-                               boolean copy_all_samples)
+                               const struct pipe_box *srcbox)
 {
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_context *pipe = ctx->base.pipe;
@@ -1185,8 +1332,7 @@ void util_blitter_copy_texture(struct blitter_context *blitter,
    /* Copy. */
    util_blitter_blit_generic(blitter, dst_view, &dstbox,
                              src_view, srcbox, src->width0, src->height0,
-                             mask, PIPE_TEX_FILTER_NEAREST, NULL,
-                             copy_all_samples);
+                             PIPE_MASK_RGBAZS, PIPE_TEX_FILTER_NEAREST, NULL);
 
    pipe_surface_reference(&dst_view, NULL);
    pipe_sampler_view_reference(&src_view, NULL);
@@ -1199,14 +1345,14 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
                                const struct pipe_box *srcbox,
                                unsigned src_width0, unsigned src_height0,
                                unsigned mask, unsigned filter,
-                               const struct pipe_scissor_state *scissor,
-                               boolean copy_all_samples)
+                               const struct pipe_scissor_state *scissor)
 {
    struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
    struct pipe_context *pipe = ctx->base.pipe;
    struct pipe_framebuffer_state fb_state;
    enum pipe_texture_target src_target = src->texture->target;
    unsigned src_samples = src->texture->nr_samples;
+   unsigned dst_samples = dst->texture->nr_samples;
    boolean has_depth, has_stencil, has_color;
    boolean blit_stencil, blit_depth, blit_color;
    void *sampler_state;
@@ -1231,6 +1377,12 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
       return;
    }
 
+   if (blit_stencil ||
+       (dstbox->width == abs(srcbox->width) &&
+        dstbox->height == abs(srcbox->height))) {
+      filter = PIPE_TEX_FILTER_NEAREST;
+   }
+
    /* Check whether the states are properly saved. */
    blitter_set_running_flag(ctx);
    blitter_check_saved_vertex_states(ctx);
@@ -1273,16 +1425,12 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
       pipe->bind_blend_state(pipe, ctx->blend[mask & PIPE_MASK_RGBA]);
       pipe->bind_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
       ctx->bind_fs_state(pipe,
-            blitter_get_fs_texfetch_col(ctx, src_target,
-                                        src_samples));
+            blitter_get_fs_texfetch_col(ctx, src->format, src_target,
+                                        src_samples, dst_samples, filter));
    }
 
    /* Set the linear filter only for scaled color non-MSAA blits. */
-   if (filter == PIPE_TEX_FILTER_LINEAR &&
-       !blit_depth && !blit_stencil &&
-       src_samples <= 1 &&
-       (dstbox->width != abs(srcbox->width) ||
-        dstbox->height != abs(srcbox->height))) {
+   if (filter == PIPE_TEX_FILTER_LINEAR) {
       if (src_target == PIPE_TEXTURE_RECT) {
          sampler_state = ctx->sampler_state_rect_linear;
       } else {
@@ -1341,7 +1489,7 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
       pipe->set_scissor_states(pipe, 0, 1, scissor);
    }
 
-   blitter_set_common_draw_rect_state(ctx, scissor != NULL);
+   blitter_set_common_draw_rect_state(ctx, scissor != NULL, FALSE);
    blitter_set_dst_dimensions(ctx, dst->width, dst->height);
 
    if ((src_target == PIPE_TEXTURE_1D ||
@@ -1388,10 +1536,9 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
          pipe->set_framebuffer_state(pipe, &fb_state);
 
          /* See if we need to blit a multisample or singlesample buffer. */
-         if (copy_all_samples &&
-             src_samples == dst->texture->nr_samples &&
-             dst->texture->nr_samples > 1) {
-            unsigned i, max_sample = MAX2(dst->texture->nr_samples, 1) - 1;
+         if (src_samples == dst_samples && dst_samples > 1) {
+            /* MSAA copy. */
+            unsigned i, max_sample = dst_samples - 1;
 
             for (i = 0; i <= max_sample; i++) {
                pipe->set_sample_mask(pipe, 1 << i);
@@ -1402,9 +1549,10 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
                                      srcbox->y + srcbox->height);
                blitter_draw(ctx, dstbox->x, dstbox->y,
                             dstbox->x + dstbox->width,
-                            dstbox->y + dstbox->height, 0);
+                            dstbox->y + dstbox->height, 0, 1);
             }
          } else {
+            /* Normal copy, MSAA upsampling, or MSAA resolve. */
             pipe->set_sample_mask(pipe, ~0);
             blitter_set_texcoords(ctx, src, src_width0, src_height0,
                                   srcbox->z + z, 0,
@@ -1413,7 +1561,7 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
                                   srcbox->y + srcbox->height);
             blitter_draw(ctx, dstbox->x, dstbox->y,
                          dstbox->x + dstbox->width,
-                         dstbox->y + dstbox->height, 0);
+                         dstbox->y + dstbox->height, 0, 1);
          }
 
          /* Get the next surface or (if this is the last iteration)
@@ -1465,7 +1613,7 @@ util_blitter_blit(struct blitter_context *blitter,
    util_blitter_blit_generic(blitter, dst_view, &info->dst.box,
                              src_view, &info->src.box, src->width0, src->height0,
                              info->mask, info->filter,
-                             info->scissor_enable ? &info->scissor : NULL, TRUE);
+                             info->scissor_enable ? &info->scissor : NULL);
 
    pipe_surface_reference(&dst_view, NULL);
    pipe_sampler_view_reference(&src_view, NULL);
@@ -1508,7 +1656,7 @@ void util_blitter_clear_render_target(struct blitter_context *blitter,
    pipe->set_framebuffer_state(pipe, &fb_state);
    pipe->set_sample_mask(pipe, ~0);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
+   blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
    blitter_set_dst_dimensions(ctx, dstsurf->width, dstsurf->height);
    blitter->draw_rectangle(blitter, dstx, dsty, dstx+width, dsty+height, 0,
                            UTIL_BLITTER_ATTRIB_COLOR, color);
@@ -1576,7 +1724,7 @@ void util_blitter_clear_depth_stencil(struct blitter_context *blitter,
    pipe->set_framebuffer_state(pipe, &fb_state);
    pipe->set_sample_mask(pipe, ~0);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
+   blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
    blitter_set_dst_dimensions(ctx, dstsurf->width, dstsurf->height);
    blitter->draw_rectangle(blitter, dstx, dsty, dstx+width, dsty+height,
                            (float) depth,
@@ -1633,7 +1781,7 @@ void util_blitter_custom_depth_stencil(struct blitter_context *blitter,
    pipe->set_framebuffer_state(pipe, &fb_state);
    pipe->set_sample_mask(pipe, sample_mask);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
+   blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
    blitter_set_dst_dimensions(ctx, zsurf->width, zsurf->height);
    blitter->draw_rectangle(blitter, 0, 0, zsurf->width, zsurf->height, depth,
                            UTIL_BLITTER_ATTRIB_NONE, NULL);
@@ -1818,7 +1966,7 @@ void util_blitter_custom_resolve_color(struct blitter_context *blitter,
    fb_state.zsbuf = NULL;
    pipe->set_framebuffer_state(pipe, &fb_state);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
+   blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
    blitter_set_dst_dimensions(ctx, src->width0, src->height0);
    blitter->draw_rectangle(blitter, 0, 0, src->width0, src->height0,
                            0, 0, NULL);
@@ -1868,7 +2016,7 @@ void util_blitter_custom_color(struct blitter_context *blitter,
    pipe->set_framebuffer_state(pipe, &fb_state);
    pipe->set_sample_mask(pipe, ~0);
 
-   blitter_set_common_draw_rect_state(ctx, FALSE);
+   blitter_set_common_draw_rect_state(ctx, FALSE, FALSE);
    blitter_set_dst_dimensions(ctx, dstsurf->width, dstsurf->height);
    blitter->draw_rectangle(blitter, 0, 0, dstsurf->width, dstsurf->height,
                            0, 0, NULL);
diff --git a/mesalib/src/gallium/auxiliary/util/u_blitter.h b/mesalib/src/gallium/auxiliary/util/u_blitter.h
index d9cefde50..a30cdc36a 100644
--- a/mesalib/src/gallium/auxiliary/util/u_blitter.h
+++ b/mesalib/src/gallium/auxiliary/util/u_blitter.h
@@ -183,7 +183,7 @@ void util_blitter_draw_rectangle(struct blitter_context *blitter,
  * - blend state
  */
 void util_blitter_clear(struct blitter_context *blitter,
-                        unsigned width, unsigned height,
+                        unsigned width, unsigned height, unsigned num_layers,
                         unsigned clear_buffers,
                         const union pipe_color_union *color,
                         double depth, unsigned stencil);
@@ -191,13 +191,10 @@ void util_blitter_clear(struct blitter_context *blitter,
 /**
  * Check if the blitter (with the help of the driver) can blit between
  * the two resources.
- * The mask is a combination of the PIPE_MASK_* flags.
- * Set to PIPE_MASK_RGBAZS if unsure.
  */
 boolean util_blitter_is_copy_supported(struct blitter_context *blitter,
                                        const struct pipe_resource *dst,
-                                       const struct pipe_resource *src,
-                                       unsigned mask);
+                                       const struct pipe_resource *src);
 
 boolean util_blitter_is_blit_supported(struct blitter_context *blitter,
 				       const struct pipe_blit_info *info);
@@ -221,8 +218,7 @@ void util_blitter_copy_texture(struct blitter_context *blitter,
                                unsigned dstx, unsigned dsty, unsigned dstz,
                                struct pipe_resource *src,
                                unsigned src_level,
-                               const struct pipe_box *srcbox, unsigned mask,
-                               boolean copy_all_samples);
+                               const struct pipe_box *srcbox);
 
 /**
  * This is a generic implementation of pipe->blit, which accepts
@@ -250,8 +246,7 @@ void util_blitter_blit_generic(struct blitter_context *blitter,
                                const struct pipe_box *srcbox,
                                unsigned src_width0, unsigned src_height0,
                                unsigned mask, unsigned filter,
-                               const struct pipe_scissor_state *scissor,
-                               boolean copy_all_samples);
+                               const struct pipe_scissor_state *scissor);
 
 void util_blitter_blit(struct blitter_context *blitter,
 		       const struct pipe_blit_info *info);
diff --git a/mesalib/src/gallium/auxiliary/util/u_clear.h b/mesalib/src/gallium/auxiliary/util/u_clear.h
index e9fd874b1..75047c16b 100644
--- a/mesalib/src/gallium/auxiliary/util/u_clear.h
+++ b/mesalib/src/gallium/auxiliary/util/u_clear.h
@@ -42,9 +42,10 @@ util_clear(struct pipe_context *pipe,
            struct pipe_framebuffer_state *framebuffer, unsigned buffers,
            const union pipe_color_union *color, double depth, unsigned stencil)
 {
-   if (buffers & PIPE_CLEAR_COLOR) {
-      unsigned i;
-      for (i = 0; i < framebuffer->nr_cbufs; i++) {
+   unsigned i;
+
+   for (i = 0; i < framebuffer->nr_cbufs; i++) {
+      if (buffers & (PIPE_CLEAR_COLOR0 << i)) {
          struct pipe_surface *ps = framebuffer->cbufs[i];
          pipe->clear_render_target(pipe, ps, color, 0, 0, ps->width, ps->height);
       }
diff --git a/mesalib/src/gallium/auxiliary/util/u_framebuffer.c b/mesalib/src/gallium/auxiliary/util/u_framebuffer.c
index f84485d1f..683967237 100644
--- a/mesalib/src/gallium/auxiliary/util/u_framebuffer.c
+++ b/mesalib/src/gallium/auxiliary/util/u_framebuffer.c
@@ -147,3 +147,27 @@ util_framebuffer_min_size(const struct pipe_framebuffer_state *fb,
       return TRUE;
    }
 }
+
+
+/**
+ * Return the number of layers set in the framebuffer state.
+ */
+unsigned
+util_framebuffer_get_num_layers(const struct pipe_framebuffer_state *fb)
+{
+	unsigned i, num_layers = 0;
+
+	for (i = 0; i < fb->nr_cbufs; i++) {
+		if (fb->cbufs[i]) {
+			unsigned num = fb->cbufs[i]->u.tex.last_layer -
+				       fb->cbufs[i]->u.tex.first_layer + 1;
+			num_layers = MAX2(num_layers, num);
+		}
+	}
+	if (fb->zsbuf) {
+		unsigned num = fb->zsbuf->u.tex.last_layer -
+			       fb->zsbuf->u.tex.first_layer + 1;
+		num_layers = MAX2(num_layers, num);
+	}
+	return num_layers;
+}
diff --git a/mesalib/src/gallium/auxiliary/util/u_framebuffer.h b/mesalib/src/gallium/auxiliary/util/u_framebuffer.h
index a89066230..0e6c98363 100644
--- a/mesalib/src/gallium/auxiliary/util/u_framebuffer.h
+++ b/mesalib/src/gallium/auxiliary/util/u_framebuffer.h
@@ -55,6 +55,11 @@ util_framebuffer_min_size(const struct pipe_framebuffer_state *fb,
                           unsigned *width,
                           unsigned *height);
 
+
+extern unsigned
+util_framebuffer_get_num_layers(const struct pipe_framebuffer_state *fb);
+
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/mesalib/src/gallium/auxiliary/util/u_simple_shaders.c b/mesalib/src/gallium/auxiliary/util/u_simple_shaders.c
index c93d75469..82f23ebec 100644
--- a/mesalib/src/gallium/auxiliary/util/u_simple_shaders.c
+++ b/mesalib/src/gallium/auxiliary/util/u_simple_shaders.c
@@ -99,6 +99,32 @@ util_make_vertex_passthrough_shader_with_so(struct pipe_context *pipe,
 }
 
 
+void *util_make_layered_clear_vertex_shader(struct pipe_context *pipe)
+{
+   static const char text[] =
+         "VERT\n"
+         "DCL IN[0]\n"
+         "DCL IN[1]\n"
+         "DCL SV[0], INSTANCEID\n"
+         "DCL OUT[0], POSITION\n"
+         "DCL OUT[1], GENERIC[0]\n"
+         "DCL OUT[2], LAYER\n"
+
+         "MOV OUT[0], IN[0]\n"
+         "MOV OUT[1], IN[1]\n"
+         "MOV OUT[2], SV[0]\n"
+         "END\n";
+   struct tgsi_token tokens[1000];
+   struct pipe_shader_state state = {tokens};
+
+   if (!tgsi_text_translate(text, tokens, Elements(tokens))) {
+      assert(0);
+      return NULL;
+   }
+   return pipe->create_vs_state(pipe, &state);
+}
+
+
 /**
  * Make simple fragment texture shader:
  *  IMM {0,0,0,1}                         // (if writemask != 0xf)
@@ -527,3 +553,157 @@ util_make_fs_blit_msaa_depthstencil(struct pipe_context *pipe,
 
    return pipe->create_fs_state(pipe, &state);
 }
+
+
+void *
+util_make_fs_msaa_resolve(struct pipe_context *pipe,
+                          unsigned tgsi_tex, unsigned nr_samples,
+                          boolean is_uint, boolean is_sint)
+{
+   struct ureg_program *ureg;
+   struct ureg_src sampler, coord;
+   struct ureg_dst out, tmp_sum, tmp_coord, tmp;
+   int i;
+
+   ureg = ureg_create(TGSI_PROCESSOR_FRAGMENT);
+   if (!ureg)
+      return NULL;
+
+   /* Declarations. */
+   sampler = ureg_DECL_sampler(ureg, 0);
+   coord = ureg_DECL_fs_input(ureg, TGSI_SEMANTIC_GENERIC, 0,
+                              TGSI_INTERPOLATE_LINEAR);
+   out = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 0);
+   tmp_sum = ureg_DECL_temporary(ureg);
+   tmp_coord = ureg_DECL_temporary(ureg);
+   tmp = ureg_DECL_temporary(ureg);
+
+   /* Instructions. */
+   ureg_MOV(ureg, tmp_sum, ureg_imm1f(ureg, 0));
+   ureg_F2U(ureg, tmp_coord, coord);
+
+   for (i = 0; i < nr_samples; i++) {
+      /* Read one sample. */
+      ureg_MOV(ureg, ureg_writemask(tmp_coord, TGSI_WRITEMASK_W),
+               ureg_imm1u(ureg, i));
+      ureg_TXF(ureg, tmp, tgsi_tex, ureg_src(tmp_coord), sampler);
+
+      if (is_uint)
+         ureg_U2F(ureg, tmp, ureg_src(tmp));
+      else if (is_sint)
+         ureg_I2F(ureg, tmp, ureg_src(tmp));
+
+      /* Add it to the sum.*/
+      ureg_ADD(ureg, tmp_sum, ureg_src(tmp_sum), ureg_src(tmp));
+   }
+
+   /* Calculate the average and return. */
+   ureg_MUL(ureg, tmp_sum, ureg_src(tmp_sum),
+            ureg_imm1f(ureg, 1.0 / nr_samples));
+
+   if (is_uint)
+      ureg_F2U(ureg, out, ureg_src(tmp_sum));
+   else if (is_sint)
+      ureg_F2I(ureg, out, ureg_src(tmp_sum));
+   else
+      ureg_MOV(ureg, out, ureg_src(tmp_sum));
+
+   ureg_END(ureg);
+
+   return ureg_create_shader_and_destroy(ureg, pipe);
+}
+
+
+void *
+util_make_fs_msaa_resolve_bilinear(struct pipe_context *pipe,
+                                   unsigned tgsi_tex, unsigned nr_samples,
+                                   boolean is_uint, boolean is_sint)
+{
+   struct ureg_program *ureg;
+   struct ureg_src sampler, coord;
+   struct ureg_dst out, tmp, top, bottom;
+   struct ureg_dst tmp_coord[4], tmp_sum[4];
+   int i, c;
+
+   ureg = ureg_create(TGSI_PROCESSOR_FRAGMENT);
+   if (!ureg)
+      return NULL;
+
+   /* Declarations. */
+   sampler = ureg_DECL_sampler(ureg, 0);
+   coord = ureg_DECL_fs_input(ureg, TGSI_SEMANTIC_GENERIC, 0,
+                              TGSI_INTERPOLATE_LINEAR);
+   out = ureg_DECL_output(ureg, TGSI_SEMANTIC_COLOR, 0);
+   for (c = 0; c < 4; c++)
+      tmp_sum[c] = ureg_DECL_temporary(ureg);
+   for (c = 0; c < 4; c++)
+      tmp_coord[c] = ureg_DECL_temporary(ureg);
+   tmp = ureg_DECL_temporary(ureg);
+   top = ureg_DECL_temporary(ureg);
+   bottom = ureg_DECL_temporary(ureg);
+
+   /* Instructions. */
+   for (c = 0; c < 4; c++)
+      ureg_MOV(ureg, tmp_sum[c], ureg_imm1f(ureg, 0));
+
+   /* Get 4 texture coordinates for the bilinear filter. */
+   ureg_F2U(ureg, tmp_coord[0], coord); /* top-left */
+   ureg_UADD(ureg, tmp_coord[1], ureg_src(tmp_coord[0]),
+             ureg_imm4u(ureg, 1, 0, 0, 0)); /* top-right */
+   ureg_UADD(ureg, tmp_coord[2], ureg_src(tmp_coord[0]),
+             ureg_imm4u(ureg, 0, 1, 0, 0)); /* bottom-left */
+   ureg_UADD(ureg, tmp_coord[3], ureg_src(tmp_coord[0]),
+             ureg_imm4u(ureg, 1, 1, 0, 0)); /* bottom-right */
+
+   for (i = 0; i < nr_samples; i++) {
+      for (c = 0; c < 4; c++) {
+         /* Read one sample. */
+         ureg_MOV(ureg, ureg_writemask(tmp_coord[c], TGSI_WRITEMASK_W),
+                  ureg_imm1u(ureg, i));
+         ureg_TXF(ureg, tmp, tgsi_tex, ureg_src(tmp_coord[c]), sampler);
+
+         if (is_uint)
+            ureg_U2F(ureg, tmp, ureg_src(tmp));
+         else if (is_sint)
+            ureg_I2F(ureg, tmp, ureg_src(tmp));
+
+         /* Add it to the sum.*/
+         ureg_ADD(ureg, tmp_sum[c], ureg_src(tmp_sum[c]), ureg_src(tmp));
+      }
+   }
+
+   /* Calculate the average. */
+   for (c = 0; c < 4; c++)
+      ureg_MUL(ureg, tmp_sum[c], ureg_src(tmp_sum[c]),
+               ureg_imm1f(ureg, 1.0 / nr_samples));
+
+   /* Take the 4 average values and apply a standard bilinear filter. */
+   ureg_FRC(ureg, tmp, coord);
+
+   ureg_LRP(ureg, top,
+            ureg_scalar(ureg_src(tmp), 0),
+            ureg_src(tmp_sum[1]),
+            ureg_src(tmp_sum[0]));
+
+   ureg_LRP(ureg, bottom,
+            ureg_scalar(ureg_src(tmp), 0),
+            ureg_src(tmp_sum[3]),
+            ureg_src(tmp_sum[2]));
+
+   ureg_LRP(ureg, tmp,
+            ureg_scalar(ureg_src(tmp), 1),
+            ureg_src(bottom),
+            ureg_src(top));
+
+   /* Convert to the texture format and return. */
+   if (is_uint)
+      ureg_F2U(ureg, out, ureg_src(tmp));
+   else if (is_sint)
+      ureg_F2I(ureg, out, ureg_src(tmp));
+   else
+      ureg_MOV(ureg, out, ureg_src(tmp));
+
+   ureg_END(ureg);
+
+   return ureg_create_shader_and_destroy(ureg, pipe);
+}
diff --git a/mesalib/src/gallium/auxiliary/util/u_simple_shaders.h b/mesalib/src/gallium/auxiliary/util/u_simple_shaders.h
index 016664d1b..e81d99414 100644
--- a/mesalib/src/gallium/auxiliary/util/u_simple_shaders.h
+++ b/mesalib/src/gallium/auxiliary/util/u_simple_shaders.h
@@ -56,6 +56,8 @@ util_make_vertex_passthrough_shader_with_so(struct pipe_context *pipe,
                                     const uint *semantic_indexes,
                                     const struct pipe_stream_output_info *so);
 
+extern void *
+util_make_layered_clear_vertex_shader(struct pipe_context *pipe);
 
 extern void *
 util_make_fragment_tex_shader_writemask(struct pipe_context *pipe, 
@@ -122,6 +124,18 @@ void *
 util_make_fs_blit_msaa_stencil(struct pipe_context *pipe,
                                unsigned tgsi_tex);
 
+
+void *
+util_make_fs_msaa_resolve(struct pipe_context *pipe,
+                          unsigned tgsi_tex, unsigned nr_samples,
+                          boolean is_uint, boolean is_sint);
+
+
+void *
+util_make_fs_msaa_resolve_bilinear(struct pipe_context *pipe,
+                                   unsigned tgsi_tex, unsigned nr_samples,
+                                   boolean is_uint, boolean is_sint);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/mesalib/src/glsl/ast.h b/mesalib/src/glsl/ast.h
index 9c43920a4..344277803 100644
--- a/mesalib/src/glsl/ast.h
+++ b/mesalib/src/glsl/ast.h
@@ -361,6 +361,7 @@ struct ast_type_qualifier {
 	 unsigned in:1;
 	 unsigned out:1;
 	 unsigned centroid:1;
+         unsigned sample:1;
 	 unsigned uniform:1;
 	 unsigned smooth:1;
 	 unsigned flat:1;
diff --git a/mesalib/src/glsl/ast_array_index.cpp b/mesalib/src/glsl/ast_array_index.cpp
index f7b5e8350..a5f23206a 100644
--- a/mesalib/src/glsl/ast_array_index.cpp
+++ b/mesalib/src/glsl/ast_array_index.cpp
@@ -41,8 +41,8 @@ update_max_array_access(ir_rvalue *ir, unsigned idx, YYLTYPE *loc,
 {
    if (ir_dereference_variable *deref_var = ir->as_dereference_variable()) {
       ir_variable *var = deref_var->var;
-      if (idx > var->max_array_access) {
-         var->max_array_access = idx;
+      if (idx > var->data.max_array_access) {
+         var->data.max_array_access = idx;
 
          /* Check whether this access will, as a side effect, implicitly cause
           * the size of a built-in array to be too large.
@@ -168,7 +168,7 @@ _mesa_ast_array_index_to_hir(void *mem_ctx,
       if (array->type->is_unsized_array()) {
 	 _mesa_glsl_error(&loc, state, "unsized array index must be constant");
       } else if (array->type->fields.array->is_interface()
-                 && array->variable_referenced()->mode == ir_var_uniform) {
+                 && array->variable_referenced()->data.mode == ir_var_uniform) {
 	 /* Page 46 in section 4.3.7 of the OpenGL ES 3.00 spec says:
 	  *
 	  *     "All indexes used to index a uniform block array must be
@@ -184,7 +184,7 @@ _mesa_ast_array_index_to_hir(void *mem_ctx,
 	  */
 	 ir_variable *v = array->whole_variable_referenced();
 	 if (v != NULL)
-	    v->max_array_access = array->type->array_size() - 1;
+	    v->data.max_array_access = array->type->array_size() - 1;
       }
 
       /* From page 23 (29 of the PDF) of the GLSL 1.30 spec:
diff --git a/mesalib/src/glsl/ast_function.cpp b/mesalib/src/glsl/ast_function.cpp
index 2707522ef..e4c0fd1c4 100644
--- a/mesalib/src/glsl/ast_function.cpp
+++ b/mesalib/src/glsl/ast_function.cpp
@@ -123,7 +123,7 @@ verify_parameter_modes(_mesa_glsl_parse_state *state,
       YYLTYPE loc = actual_ast->get_location();
 
       /* Verify that 'const_in' parameters are ir_constants. */
-      if (formal->mode == ir_var_const_in &&
+      if (formal->data.mode == ir_var_const_in &&
 	  actual->ir_type != ir_type_constant) {
 	 _mesa_glsl_error(&loc, state,
 			  "parameter `in %s' must be a constant expression",
@@ -132,10 +132,10 @@ verify_parameter_modes(_mesa_glsl_parse_state *state,
       }
 
       /* Verify that 'out' and 'inout' actual parameters are lvalues. */
-      if (formal->mode == ir_var_function_out
-          || formal->mode == ir_var_function_inout) {
+      if (formal->data.mode == ir_var_function_out
+          || formal->data.mode == ir_var_function_inout) {
 	 const char *mode = NULL;
-	 switch (formal->mode) {
+	 switch (formal->data.mode) {
 	 case ir_var_function_out:   mode = "out";   break;
 	 case ir_var_function_inout: mode = "inout"; break;
 	 default:                    assert(false);  break;
@@ -155,9 +155,9 @@ verify_parameter_modes(_mesa_glsl_parse_state *state,
 
 	 ir_variable *var = actual->variable_referenced();
 	 if (var)
-	    var->assigned = true;
+	    var->data.assigned = true;
 
-	 if (var && var->read_only) {
+	 if (var && var->data.read_only) {
 	    _mesa_glsl_error(&loc, state,
 			     "function parameter '%s %s' references the "
 			     "read-only variable '%s'",
@@ -274,20 +274,20 @@ fix_parameter(void *mem_ctx, ir_rvalue *actual, const glsl_type *formal_type,
 }
 
 /**
- * If a function call is generated, \c call_ir will point to it on exit.
- * Otherwise \c call_ir will be set to \c NULL.
+ * Generate a function call.
+ *
+ * For non-void functions, this returns a dereference of the temporary variable
+ * which stores the return value for the call.  For void functions, this returns
+ * NULL.
  */
 static ir_rvalue *
 generate_call(exec_list *instructions, ir_function_signature *sig,
 	      exec_list *actual_parameters,
-	      ir_call **call_ir,
 	      struct _mesa_glsl_parse_state *state)
 {
    void *ctx = state;
    exec_list post_call_conversions;
 
-   *call_ir = NULL;
-
    /* Perform implicit conversion of arguments.  For out parameters, we need
     * to place them in a temporary variable and do the conversion after the
     * call takes place.  Since we haven't emitted the call yet, we'll place
@@ -304,7 +304,7 @@ generate_call(exec_list *instructions, ir_function_signature *sig,
       assert(formal != NULL);
 
       if (formal->type->is_numeric() || formal->type->is_boolean()) {
-	 switch (formal->mode) {
+	 switch (formal->data.mode) {
 	 case ir_var_const_in:
 	 case ir_var_function_in: {
 	    ir_rvalue *converted
@@ -316,7 +316,7 @@ generate_call(exec_list *instructions, ir_function_signature *sig,
 	 case ir_var_function_inout:
             fix_parameter(ctx, actual, formal->type,
                           instructions, &post_call_conversions,
-                          formal->mode == ir_var_function_inout);
+                          formal->data.mode == ir_var_function_inout);
 	    break;
 	 default:
 	    assert (!"Illegal formal parameter mode");
@@ -421,6 +421,25 @@ done:
    return sig;
 }
 
+static void
+print_function_prototypes(_mesa_glsl_parse_state *state, YYLTYPE *loc,
+                          ir_function *f)
+{
+   if (f == NULL)
+      return;
+
+   foreach_list (node, &f->signatures) {
+      ir_function_signature *sig = (ir_function_signature *) node;
+
+      if (sig->is_builtin() && !sig->is_builtin_available(state))
+         continue;
+
+      char *str = prototype_string(sig->return_type, f->name, &sig->parameters);
+      _mesa_glsl_error(loc, state, "   %s", str);
+      ralloc_free(str);
+   }
+}
+
 /**
  * Raise a "no matching function" error, listing all possible overloads the
  * compiler considered so developers can figure out what went wrong.
@@ -431,30 +450,23 @@ no_matching_function_error(const char *name,
 			   exec_list *actual_parameters,
 			   _mesa_glsl_parse_state *state)
 {
-   char *str = prototype_string(NULL, name, actual_parameters);
-   _mesa_glsl_error(loc, state, "no matching function for call to `%s'", str);
-   ralloc_free(str);
-
-   const char *prefix = "candidates are: ";
+   gl_shader *sh = _mesa_glsl_get_builtin_function_shader();
 
-   for (int i = -1; i < (int) state->num_builtins_to_link; i++) {
-      glsl_symbol_table *syms = i >= 0 ? state->builtins_to_link[i]->symbols
-				       : state->symbols;
-      ir_function *f = syms->get_function(name);
-      if (f == NULL)
-	 continue;
-
-      foreach_list (node, &f->signatures) {
-	 ir_function_signature *sig = (ir_function_signature *) node;
-
-         if (sig->is_builtin() && !sig->is_builtin_available(state))
-            continue;
+   if (state->symbols->get_function(name) == NULL
+      && (!state->uses_builtin_functions
+          || sh->symbols->get_function(name) == NULL)) {
+      _mesa_glsl_error(loc, state, "no function with name '%s'", name);
+   } else {
+      char *str = prototype_string(NULL, name, actual_parameters);
+      _mesa_glsl_error(loc, state,
+                       "no matching function for call to `%s'; candidates are:",
+                       str);
+      ralloc_free(str);
 
-	 str = prototype_string(sig->return_type, f->name, &sig->parameters);
-	 _mesa_glsl_error(loc, state, "%s%s", prefix, str);
-	 ralloc_free(str);
+      print_function_prototypes(state, loc, state->symbols->get_function(name));
 
-	 prefix = "                ";
+      if (state->uses_builtin_functions) {
+         print_function_prototypes(state, loc, sh->symbols->get_function(name));
       }
    }
 }
@@ -1651,7 +1663,7 @@ ast_function_expression::hir(exec_list *instructions,
    } else {
       const ast_expression *id = subexpressions[0];
       const char *func_name = id->primary_expression.identifier;
-      YYLTYPE loc = id->get_location();
+      YYLTYPE loc = get_location();
       exec_list actual_parameters;
 
       process_parameters(instructions, &actual_parameters, &this->expressions,
@@ -1660,7 +1672,6 @@ ast_function_expression::hir(exec_list *instructions,
       ir_function_signature *sig =
 	 match_function_by_name(func_name, &actual_parameters, state);
 
-      ir_call *call = NULL;
       ir_rvalue *value = NULL;
       if (sig == NULL) {
 	 no_matching_function_error(func_name, &loc, &actual_parameters, state);
@@ -1669,8 +1680,7 @@ ast_function_expression::hir(exec_list *instructions,
 	 /* an error has already been emitted */
 	 value = ir_rvalue::error_value(ctx);
       } else {
-	 value = generate_call(instructions, sig, &actual_parameters,
-			       &call, state);
+	 value = generate_call(instructions, sig, &actual_parameters, state);
       }
 
       return value;
diff --git a/mesalib/src/glsl/ast_to_hir.cpp b/mesalib/src/glsl/ast_to_hir.cpp
index 43cf49745..3bc181e87 100644
--- a/mesalib/src/glsl/ast_to_hir.cpp
+++ b/mesalib/src/glsl/ast_to_hir.cpp
@@ -728,7 +728,7 @@ mark_whole_array_access(ir_rvalue *access)
    ir_dereference_variable *deref = access->as_dereference_variable();
 
    if (deref && deref->var) {
-      deref->var->max_array_access = deref->type->length - 1;
+      deref->var->data.max_array_access = deref->type->length - 1;
    }
 }
 
@@ -767,7 +767,7 @@ do_assignment(exec_list *instructions, struct _mesa_glsl_parse_state *state,
 
    ir_variable *lhs_var = lhs->variable_referenced();
    if (lhs_var)
-      lhs_var->assigned = true;
+      lhs_var->data.assigned = true;
 
    if (!error_emitted) {
       if (non_lvalue_description != NULL) {
@@ -776,7 +776,7 @@ do_assignment(exec_list *instructions, struct _mesa_glsl_parse_state *state,
 			  non_lvalue_description);
 	 error_emitted = true;
       } else if (lhs->variable_referenced() != NULL
-		 && lhs->variable_referenced()->read_only) {
+		 && lhs->variable_referenced()->data.read_only) {
          _mesa_glsl_error(&lhs_loc, state,
                           "assignment to read-only variable '%s'",
                           lhs->variable_referenced()->name);
@@ -819,11 +819,11 @@ do_assignment(exec_list *instructions, struct _mesa_glsl_parse_state *state,
 
 	 assert(var != NULL);
 
-	 if (var->max_array_access >= unsigned(rhs->type->array_size())) {
+	 if (var->data.max_array_access >= unsigned(rhs->type->array_size())) {
 	    /* FINISHME: This should actually log the location of the RHS. */
 	    _mesa_glsl_error(& lhs_loc, state, "array size must be > %u due to "
 			     "previous access",
-			     var->max_array_access);
+			     var->data.max_array_access);
 	 }
 
 	 var->type = glsl_type::get_array_instance(lhs->type->element_type(),
@@ -866,7 +866,7 @@ get_lvalue_copy(exec_list *instructions, ir_rvalue *lvalue)
    var = new(ctx) ir_variable(lvalue->type, "_post_incdec_tmp",
 			      ir_var_temporary);
    instructions->push_tail(var);
-   var->mode = ir_var_auto;
+   var->data.mode = ir_var_auto;
 
    instructions->push_tail(new(ctx) ir_assignment(new(ctx) ir_dereference_variable(var),
 						  lvalue));
@@ -1639,7 +1639,7 @@ ast_expression::hir(exec_list *instructions,
 	 state->symbols->get_variable(this->primary_expression.identifier);
 
       if (var != NULL) {
-	 var->used = true;
+	 var->data.used = true;
 	 result = new(ctx) ir_dereference_variable(var);
       } else {
 	 _mesa_glsl_error(& loc, state, "`%s' undeclared",
@@ -1860,7 +1860,7 @@ ast_fully_specified_type::glsl_type(const char **name,
 
    if (type->base_type == GLSL_TYPE_FLOAT
        && state->es_shader
-       && state->target == fragment_shader
+       && state->target == MESA_SHADER_FRAGMENT
        && this->qualifier.precision == ast_precision_none
        && state->symbols->get_variable("#default precision") == NULL) {
       YYLTYPE loc = this->get_location();
@@ -1882,15 +1882,15 @@ ast_fully_specified_type::glsl_type(const char **name,
  * this function will produce undefined results.
  */
 static bool
-is_varying_var(ir_variable *var, _mesa_glsl_parser_targets target)
+is_varying_var(ir_variable *var, gl_shader_type target)
 {
    switch (target) {
-   case vertex_shader:
-      return var->mode == ir_var_shader_out;
-   case fragment_shader:
-      return var->mode == ir_var_shader_in;
+   case MESA_SHADER_VERTEX:
+      return var->data.mode == ir_var_shader_out;
+   case MESA_SHADER_FRAGMENT:
+      return var->data.mode == ir_var_shader_in;
    default:
-      return var->mode == ir_var_shader_out || var->mode == ir_var_shader_in;
+      return var->data.mode == ir_var_shader_out || var->data.mode == ir_var_shader_in;
    }
 }
 
@@ -1941,7 +1941,7 @@ validate_binding_qualifier(struct _mesa_glsl_parse_state *state,
                            ir_variable *var,
                            const ast_type_qualifier *qual)
 {
-   if (var->mode != ir_var_uniform) {
+   if (var->data.mode != ir_var_uniform) {
       _mesa_glsl_error(loc, state,
                        "the \"binding\" qualifier only applies to uniforms");
       return false;
@@ -1985,13 +1985,13 @@ validate_binding_qualifier(struct _mesa_glsl_parse_state *state,
        */
       unsigned limit = 0;
       switch (state->target) {
-      case vertex_shader:
+      case MESA_SHADER_VERTEX:
          limit = ctx->Const.VertexProgram.MaxTextureImageUnits;
          break;
-      case geometry_shader:
+      case MESA_SHADER_GEOMETRY:
          limit = ctx->Const.GeometryProgram.MaxTextureImageUnits;
          break;
-      case fragment_shader:
+      case MESA_SHADER_FRAGMENT:
          limit = ctx->Const.FragmentProgram.MaxTextureImageUnits;
          break;
       }
@@ -2049,8 +2049,8 @@ interpret_interpolation_qualifier(const struct ast_type_qualifier *qual,
 
       }
 
-      if ((state->target == vertex_shader && mode == ir_var_shader_in) ||
-          (state->target == fragment_shader && mode == ir_var_shader_out)) {
+      if ((state->target == MESA_SHADER_VERTEX && mode == ir_var_shader_in) ||
+          (state->target == MESA_SHADER_FRAGMENT && mode == ir_var_shader_out)) {
          _mesa_glsl_error(loc, state,
                           "interpolation qualifier `%s' cannot be applied to "
                           "vertex shader inputs or fragment shader outputs",
@@ -2077,8 +2077,8 @@ validate_explicit_location(const struct ast_type_qualifier *qual,
     * locations.
     */
    switch (state->target) {
-   case vertex_shader:
-      if (var->mode == ir_var_shader_in) {
+   case MESA_SHADER_VERTEX:
+      if (var->data.mode == ir_var_shader_in) {
          if (!state->check_explicit_attrib_location_allowed(loc, var))
             return;
 
@@ -2088,14 +2088,14 @@ validate_explicit_location(const struct ast_type_qualifier *qual,
       fail = true;
       break;
 
-   case geometry_shader:
+   case MESA_SHADER_GEOMETRY:
       _mesa_glsl_error(loc, state,
                        "geometry shader variables cannot be given "
                        "explicit locations");
       return;
 
-   case fragment_shader:
-      if (var->mode == ir_var_shader_out) {
+   case MESA_SHADER_FRAGMENT:
+      if (var->data.mode == ir_var_shader_out) {
          if (!state->check_explicit_attrib_location_allowed(loc, var))
             return;
 
@@ -2112,7 +2112,7 @@ validate_explicit_location(const struct ast_type_qualifier *qual,
                        mode_string(var),
 		       _mesa_glsl_shader_target_name(state->target));
    } else {
-      var->explicit_location = true;
+      var->data.explicit_location = true;
 
       /* This bit of silliness is needed because invalid explicit locations
        * are supposed to be flagged during linking.  Small negative values
@@ -2122,11 +2122,11 @@ validate_explicit_location(const struct ast_type_qualifier *qual,
        * ensures that negative values stay negative.
        */
       if (qual->location >= 0) {
-         var->location = (state->target == vertex_shader)
+         var->data.location = (state->target == MESA_SHADER_VERTEX)
             ? (qual->location + VERT_ATTRIB_GENERIC0)
             : (qual->location + FRAG_RESULT_DATA0);
       } else {
-         var->location = qual->location;
+         var->data.location = qual->location;
       }
 
       if (qual->flags.q.explicit_index) {
@@ -2143,8 +2143,8 @@ validate_explicit_location(const struct ast_type_qualifier *qual,
             _mesa_glsl_error(loc, state,
                              "explicit index may only be 0 or 1");
          } else {
-            var->explicit_index = true;
-            var->index = qual->index;
+            var->data.explicit_index = true;
+            var->data.index = qual->index;
          }
       }
    }
@@ -2162,25 +2162,28 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
    STATIC_ASSERT(sizeof(qual->flags.q) <= sizeof(qual->flags.i));
 
    if (qual->flags.q.invariant) {
-      if (var->used) {
+      if (var->data.used) {
 	 _mesa_glsl_error(loc, state,
 			  "variable `%s' may not be redeclared "
 			  "`invariant' after being used",
 			  var->name);
       } else {
-	 var->invariant = 1;
+	 var->data.invariant = 1;
       }
    }
 
    if (qual->flags.q.constant || qual->flags.q.attribute
        || qual->flags.q.uniform
-       || (qual->flags.q.varying && (state->target == fragment_shader)))
-      var->read_only = 1;
+       || (qual->flags.q.varying && (state->target == MESA_SHADER_FRAGMENT)))
+      var->data.read_only = 1;
 
    if (qual->flags.q.centroid)
-      var->centroid = 1;
+      var->data.centroid = 1;
 
-   if (qual->flags.q.attribute && state->target != vertex_shader) {
+   if (qual->flags.q.sample)
+      var->data.sample = 1;
+
+   if (qual->flags.q.attribute && state->target != MESA_SHADER_VERTEX) {
       var->type = glsl_type::error_type;
       _mesa_glsl_error(loc, state,
 		       "`attribute' variables may not be declared in the "
@@ -2207,18 +2210,18 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
     * the setting alone.
     */
    if (qual->flags.q.in && qual->flags.q.out)
-      var->mode = ir_var_function_inout;
+      var->data.mode = ir_var_function_inout;
    else if (qual->flags.q.in)
-      var->mode = is_parameter ? ir_var_function_in : ir_var_shader_in;
+      var->data.mode = is_parameter ? ir_var_function_in : ir_var_shader_in;
    else if (qual->flags.q.attribute
-	    || (qual->flags.q.varying && (state->target == fragment_shader)))
-      var->mode = ir_var_shader_in;
+	    || (qual->flags.q.varying && (state->target == MESA_SHADER_FRAGMENT)))
+      var->data.mode = ir_var_shader_in;
    else if (qual->flags.q.out)
-      var->mode = is_parameter ? ir_var_function_out : ir_var_shader_out;
-   else if (qual->flags.q.varying && (state->target == vertex_shader))
-      var->mode = ir_var_shader_out;
+      var->data.mode = is_parameter ? ir_var_function_out : ir_var_shader_out;
+   else if (qual->flags.q.varying && (state->target == MESA_SHADER_VERTEX))
+      var->data.mode = ir_var_shader_out;
    else if (qual->flags.q.uniform)
-      var->mode = ir_var_uniform;
+      var->data.mode = ir_var_uniform;
 
    if (!is_parameter && is_varying_var(var, state->target)) {
       /* This variable is being used to link data between shader stages (in
@@ -2270,28 +2273,28 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
 
    if (state->all_invariant && (state->current_function == NULL)) {
       switch (state->target) {
-      case vertex_shader:
-	 if (var->mode == ir_var_shader_out)
-	    var->invariant = true;
+      case MESA_SHADER_VERTEX:
+	 if (var->data.mode == ir_var_shader_out)
+	    var->data.invariant = true;
 	 break;
-      case geometry_shader:
-	 if ((var->mode == ir_var_shader_in)
-             || (var->mode == ir_var_shader_out))
-	    var->invariant = true;
+      case MESA_SHADER_GEOMETRY:
+	 if ((var->data.mode == ir_var_shader_in)
+             || (var->data.mode == ir_var_shader_out))
+	    var->data.invariant = true;
 	 break;
-      case fragment_shader:
-	 if (var->mode == ir_var_shader_in)
-	    var->invariant = true;
+      case MESA_SHADER_FRAGMENT:
+	 if (var->data.mode == ir_var_shader_in)
+	    var->data.invariant = true;
 	 break;
       }
    }
 
-   var->interpolation =
-      interpret_interpolation_qualifier(qual, (ir_variable_mode) var->mode,
+   var->data.interpolation =
+      interpret_interpolation_qualifier(qual, (ir_variable_mode) var->data.mode,
                                         state, loc);
 
-   var->pixel_center_integer = qual->flags.q.pixel_center_integer;
-   var->origin_upper_left = qual->flags.q.origin_upper_left;
+   var->data.pixel_center_integer = qual->flags.q.pixel_center_integer;
+   var->data.origin_upper_left = qual->flags.q.origin_upper_left;
    if ((qual->flags.q.origin_upper_left || qual->flags.q.pixel_center_integer)
        && (strcmp(var->name, "gl_FragCoord") != 0)) {
       const char *const qual_string = (qual->flags.q.origin_upper_left)
@@ -2312,27 +2315,28 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
 
    if (qual->flags.q.explicit_binding &&
        validate_binding_qualifier(state, loc, var, qual)) {
-      var->explicit_binding = true;
-      var->binding = qual->binding;
+      var->data.explicit_binding = true;
+      var->data.binding = qual->binding;
    }
 
    if (var->type->contains_atomic()) {
-      if (var->mode == ir_var_uniform) {
-         if (var->explicit_binding) {
-            unsigned *offset = &state->atomic_counter_offsets[var->binding];
+      if (var->data.mode == ir_var_uniform) {
+         if (var->data.explicit_binding) {
+            unsigned *offset =
+               &state->atomic_counter_offsets[var->data.binding];
 
             if (*offset % ATOMIC_COUNTER_SIZE)
                _mesa_glsl_error(loc, state,
                                 "misaligned atomic counter offset");
 
-            var->atomic.offset = *offset;
+            var->data.atomic.offset = *offset;
             *offset += var->type->atomic_size();
 
          } else {
             _mesa_glsl_error(loc, state,
                              "atomic counters require explicit binding point");
          }
-      } else if (var->mode != ir_var_function_in) {
+      } else if (var->data.mode != ir_var_function_in) {
          _mesa_glsl_error(loc, state, "atomic counters may only be declared as "
                           "function parameters or uniform-qualified "
                           "global variables");
@@ -2406,15 +2410,15 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
                        "gl_FragDepth");
    }
    if (qual->flags.q.depth_any)
-      var->depth_layout = ir_depth_layout_any;
+      var->data.depth_layout = ir_depth_layout_any;
    else if (qual->flags.q.depth_greater)
-      var->depth_layout = ir_depth_layout_greater;
+      var->data.depth_layout = ir_depth_layout_greater;
    else if (qual->flags.q.depth_less)
-      var->depth_layout = ir_depth_layout_less;
+      var->data.depth_layout = ir_depth_layout_less;
    else if (qual->flags.q.depth_unchanged)
-       var->depth_layout = ir_depth_layout_unchanged;
+       var->data.depth_layout = ir_depth_layout_unchanged;
    else
-       var->depth_layout = ir_depth_layout_none;
+       var->data.depth_layout = ir_depth_layout_none;
 
    if (qual->flags.q.std140 ||
        qual->flags.q.packed ||
@@ -2475,10 +2479,10 @@ get_variable_being_redeclared(ir_variable *var, YYLTYPE loc,
 
       const unsigned size = unsigned(var->type->array_size());
       check_builtin_array_max_size(var->name, size, loc, state);
-      if ((size > 0) && (size <= earlier->max_array_access)) {
+      if ((size > 0) && (size <= earlier->data.max_array_access)) {
 	 _mesa_glsl_error(& loc, state, "array size must be > %u due to "
 			  "previous access",
-			  earlier->max_array_access);
+			  earlier->data.max_array_access);
       }
 
       earlier->type = var->type;
@@ -2488,12 +2492,12 @@ get_variable_being_redeclared(ir_variable *var, YYLTYPE loc,
                state->is_version(150, 0))
 	      && strcmp(var->name, "gl_FragCoord") == 0
 	      && earlier->type == var->type
-	      && earlier->mode == var->mode) {
+	      && earlier->data.mode == var->data.mode) {
       /* Allow redeclaration of gl_FragCoord for ARB_fcc layout
        * qualifiers.
        */
-      earlier->origin_upper_left = var->origin_upper_left;
-      earlier->pixel_center_integer = var->pixel_center_integer;
+      earlier->data.origin_upper_left = var->data.origin_upper_left;
+      earlier->data.pixel_center_integer = var->data.pixel_center_integer;
 
       /* According to section 4.3.7 of the GLSL 1.30 spec,
        * the following built-in varaibles can be redeclared with an
@@ -2513,41 +2517,41 @@ get_variable_being_redeclared(ir_variable *var, YYLTYPE loc,
 		  || strcmp(var->name, "gl_Color") == 0
 		  || strcmp(var->name, "gl_SecondaryColor") == 0)
 	      && earlier->type == var->type
-	      && earlier->mode == var->mode) {
-      earlier->interpolation = var->interpolation;
+	      && earlier->data.mode == var->data.mode) {
+      earlier->data.interpolation = var->data.interpolation;
 
       /* Layout qualifiers for gl_FragDepth. */
    } else if ((state->AMD_conservative_depth_enable ||
                state->ARB_conservative_depth_enable)
 	      && strcmp(var->name, "gl_FragDepth") == 0
 	      && earlier->type == var->type
-	      && earlier->mode == var->mode) {
+	      && earlier->data.mode == var->data.mode) {
 
       /** From the AMD_conservative_depth spec:
        *     Within any shader, the first redeclarations of gl_FragDepth
        *     must appear before any use of gl_FragDepth.
        */
-      if (earlier->used) {
+      if (earlier->data.used) {
 	 _mesa_glsl_error(&loc, state,
 			  "the first redeclaration of gl_FragDepth "
 			  "must appear before any use of gl_FragDepth");
       }
 
       /* Prevent inconsistent redeclaration of depth layout qualifier. */
-      if (earlier->depth_layout != ir_depth_layout_none
-	  && earlier->depth_layout != var->depth_layout) {
+      if (earlier->data.depth_layout != ir_depth_layout_none
+	  && earlier->data.depth_layout != var->data.depth_layout) {
 	 _mesa_glsl_error(&loc, state,
 			  "gl_FragDepth: depth layout is declared here "
 			  "as '%s, but it was previously declared as "
 			  "'%s'",
-			  depth_layout_string(var->depth_layout),
-			  depth_layout_string(earlier->depth_layout));
+			  depth_layout_string(var->data.depth_layout),
+			  depth_layout_string(earlier->data.depth_layout));
       }
 
-      earlier->depth_layout = var->depth_layout;
+      earlier->data.depth_layout = var->data.depth_layout;
 
    } else if (allow_all_redeclarations) {
-      if (earlier->mode != var->mode) {
+      if (earlier->data.mode != var->data.mode) {
          _mesa_glsl_error(&loc, state,
                           "redeclaration of `%s' with incorrect qualifiers",
                           var->name);
@@ -2582,7 +2586,7 @@ process_initializer(ir_variable *var, ast_declaration *decl,
     *    directly by an application via API commands, or indirectly by
     *    OpenGL."
     */
-   if (var->mode == ir_var_uniform) {
+   if (var->data.mode == ir_var_uniform) {
       state->check_version(120, 0, &initializer_loc,
                            "cannot initialize uniforms");
    }
@@ -2592,11 +2596,11 @@ process_initializer(ir_variable *var, ast_declaration *decl,
 		       "cannot initialize samplers");
    }
 
-   if ((var->mode == ir_var_shader_in) && (state->current_function == NULL)) {
+   if ((var->data.mode == ir_var_shader_in) && (state->current_function == NULL)) {
       _mesa_glsl_error(& initializer_loc, state,
 		       "cannot initialize %s shader input / %s",
 		       _mesa_glsl_shader_target_name(state->target),
-		       (state->target == vertex_shader)
+		       (state->target == MESA_SHADER_VERTEX)
 		       ? "attribute" : "varying");
    }
 
@@ -2647,9 +2651,9 @@ process_initializer(ir_variable *var, ast_declaration *decl,
    }
 
    if (rhs && !rhs->type->is_error()) {
-      bool temp = var->read_only;
+      bool temp = var->data.read_only;
       if (type->qualifier.flags.q.constant)
-	 var->read_only = false;
+	 var->data.read_only = false;
 
       /* Never emit code to initialize a uniform.
        */
@@ -2664,7 +2668,7 @@ process_initializer(ir_variable *var, ast_declaration *decl,
 	 initializer_type = rhs->type;
 
       var->constant_initializer = rhs->constant_expression_value();
-      var->has_initializer = true;
+      var->data.has_initializer = true;
 
       /* If the declared variable is an unsized array, it must inherrit
        * its full type from the initializer.  A declaration such as
@@ -2688,7 +2692,7 @@ process_initializer(ir_variable *var, ast_declaration *decl,
        */
       var->type = initializer_type;
 
-      var->read_only = temp;
+      var->data.read_only = temp;
    }
 
    return result;
@@ -2840,23 +2844,23 @@ ast_declarator_list::hir(exec_list *instructions,
 	    _mesa_glsl_error(& loc, state,
 			     "undeclared variable `%s' cannot be marked "
 			     "invariant", decl->identifier);
-	 } else if ((state->target == vertex_shader)
-	       && (earlier->mode != ir_var_shader_out)) {
+	 } else if ((state->target == MESA_SHADER_VERTEX)
+	       && (earlier->data.mode != ir_var_shader_out)) {
 	    _mesa_glsl_error(& loc, state,
 			     "`%s' cannot be marked invariant, vertex shader "
 			     "outputs only", decl->identifier);
-	 } else if ((state->target == fragment_shader)
-	       && (earlier->mode != ir_var_shader_in)) {
+	 } else if ((state->target == MESA_SHADER_FRAGMENT)
+	       && (earlier->data.mode != ir_var_shader_in)) {
 	    _mesa_glsl_error(& loc, state,
 			     "`%s' cannot be marked invariant, fragment shader "
 			     "inputs only", decl->identifier);
-	 } else if (earlier->used) {
+	 } else if (earlier->data.used) {
 	    _mesa_glsl_error(& loc, state,
 			     "variable `%s' may not be redeclared "
 			     "`invariant' after being used",
 			     earlier->name);
 	 } else {
-	    earlier->invariant = true;
+	    earlier->data.invariant = true;
 	 }
       }
 
@@ -2940,7 +2944,7 @@ ast_declarator_list::hir(exec_list *instructions,
                                precision_names[this->type->qualifier.precision],
                                type_name);
          }
-      } else {
+      } else if (this->type->specifier->structure == NULL) {
          _mesa_glsl_warning(&loc, state, "empty declaration");
       }
    }
@@ -3030,13 +3034,13 @@ ast_declarator_list::hir(exec_list *instructions,
 				       & loc, false);
 
       if (this->type->qualifier.flags.q.invariant) {
-	 if ((state->target == vertex_shader) &&
-             var->mode != ir_var_shader_out) {
+	 if ((state->target == MESA_SHADER_VERTEX) &&
+             var->data.mode != ir_var_shader_out) {
 	    _mesa_glsl_error(& loc, state,
 			     "`%s' cannot be marked invariant, vertex shader "
 			     "outputs only", var->name);
-	 } else if ((state->target == fragment_shader) &&
-		    var->mode != ir_var_shader_in) {
+	 } else if ((state->target == MESA_SHADER_FRAGMENT) &&
+		    var->data.mode != ir_var_shader_in) {
 	    /* FINISHME: Note that this doesn't work for invariant on
 	     * a function signature inval
 	     */
@@ -3073,10 +3077,10 @@ ast_declarator_list::hir(exec_list *instructions,
 			     "global scope%s",
 			     mode, var->name, extra);
 	 }
-      } else if (var->mode == ir_var_shader_in) {
-         var->read_only = true;
+      } else if (var->data.mode == ir_var_shader_in) {
+         var->data.read_only = true;
 
-	 if (state->target == vertex_shader) {
+	 if (state->target == MESA_SHADER_VERTEX) {
 	    bool error_emitted = false;
 
 	    /* From page 31 (page 37 of the PDF) of the GLSL 1.50 spec:
@@ -3131,7 +3135,7 @@ ast_declarator_list::hir(exec_list *instructions,
                                       "cannot have array type")) {
 	       error_emitted = true;
 	    }
-	 } else if (state->target == geometry_shader) {
+	 } else if (state->target == MESA_SHADER_GEOMETRY) {
             /* From section 4.3.4 (Inputs) of the GLSL 1.50 spec:
              *
              *     Geometry shader input variables get the per-vertex values
@@ -3180,11 +3184,11 @@ ast_declarator_list::hir(exec_list *instructions,
        */
       if (state->is_version(130, 300) &&
           var->type->contains_integer() &&
-          var->interpolation != INTERP_QUALIFIER_FLAT &&
-          ((state->target == fragment_shader && var->mode == ir_var_shader_in)
-           || (state->target == vertex_shader && var->mode == ir_var_shader_out
+          var->data.interpolation != INTERP_QUALIFIER_FLAT &&
+          ((state->target == MESA_SHADER_FRAGMENT && var->data.mode == ir_var_shader_in)
+           || (state->target == MESA_SHADER_VERTEX && var->data.mode == ir_var_shader_out
                && state->es_shader))) {
-         const char *var_type = (state->target == vertex_shader) ?
+         const char *var_type = (state->target == MESA_SHADER_VERTEX) ?
             "vertex output" : "fragment input";
          _mesa_glsl_error(&loc, state, "if a %s is (or contains) "
                           "an integer, then it must be qualified with 'flat'",
@@ -3241,14 +3245,14 @@ ast_declarator_list::hir(exec_list *instructions,
          assert(i != NULL);
 
          switch (state->target) {
-         case vertex_shader:
+         case MESA_SHADER_VERTEX:
             if (this->type->qualifier.flags.q.in) {
                _mesa_glsl_error(&loc, state,
                                 "qualifier '%s' cannot be applied to vertex "
                                 "shader inputs", i);
             }
             break;
-         case fragment_shader:
+         case MESA_SHADER_FRAGMENT:
             if (this->type->qualifier.flags.q.out) {
                _mesa_glsl_error(&loc, state,
                                 "qualifier '%s' cannot be applied to fragment "
@@ -3271,12 +3275,20 @@ ast_declarator_list::hir(exec_list *instructions,
       if (state->is_version(130, 300)
           && this->type->qualifier.flags.q.centroid
           && this->type->qualifier.flags.q.in
-          && state->target == vertex_shader) {
+          && state->target == MESA_SHADER_VERTEX) {
 
          _mesa_glsl_error(&loc, state,
                           "'centroid in' cannot be used in a vertex shader");
       }
 
+      if (state->target == MESA_SHADER_VERTEX
+          && this->type->qualifier.flags.q.sample
+          && this->type->qualifier.flags.q.in) {
+
+         _mesa_glsl_error(&loc, state,
+                        "'sample in' cannot be used in a vertex shader");
+      }
+
       /* Section 4.3.6 of the GLSL 1.30 specification states:
        * "It is an error to use centroid out in a fragment shader."
        *
@@ -3284,7 +3296,7 @@ ast_declarator_list::hir(exec_list *instructions,
        * "It is an error to use auxiliary storage qualifiers or interpolation
        *  qualifiers on an output in a fragment shader."
        */
-      if (state->target == fragment_shader &&
+      if (state->target == MESA_SHADER_FRAGMENT &&
           this->type->qualifier.flags.q.out &&
           this->type->qualifier.has_auxiliary_storage()) {
          _mesa_glsl_error(&loc, state,
@@ -3363,12 +3375,12 @@ ast_declarator_list::hir(exec_list *instructions,
                                        false /* allow_all_redeclarations */);
       if (earlier != NULL) {
          if (strncmp(var->name, "gl_", 3) == 0 &&
-             earlier->how_declared == ir_var_declared_in_block) {
+             earlier->data.how_declared == ir_var_declared_in_block) {
             _mesa_glsl_error(&loc, state,
                              "`%s' has already been redeclared using "
                              "gl_PerVertex", var->name);
          }
-         earlier->how_declared = ir_var_declared_normally;
+         earlier->data.how_declared = ir_var_declared_normally;
       }
 
       if (decl->initializer != NULL) {
@@ -3548,7 +3560,7 @@ ast_parameter_declarator::hir(exec_list *instructions,
     *    as out or inout function parameters, nor can they be assigned
     *    into."
     */
-   if ((var->mode == ir_var_function_inout || var->mode == ir_var_function_out)
+   if ((var->data.mode == ir_var_function_inout || var->data.mode == ir_var_function_out)
        && type->contains_sampler()) {
       _mesa_glsl_error(&loc, state, "out and inout parameters cannot contain samplers");
       type = glsl_type::error_type;
@@ -3568,7 +3580,7 @@ ast_parameter_declarator::hir(exec_list *instructions,
     * So for GLSL 1.10, passing an array as an out or inout parameter is not
     * allowed.  This restriction is removed in GLSL 1.20, and in GLSL ES.
     */
-   if ((var->mode == ir_var_function_inout || var->mode == ir_var_function_out)
+   if ((var->data.mode == ir_var_function_inout || var->data.mode == ir_var_function_out)
        && type->is_array()
        && !state->check_version(120, 100, &loc,
                                 "arrays cannot be out or inout parameters")) {
@@ -3942,7 +3954,7 @@ ast_jump_statement::hir(exec_list *instructions,
    }
 
    case ast_discard:
-      if (state->target != fragment_shader) {
+      if (state->target != MESA_SHADER_FRAGMENT) {
 	 YYLTYPE loc = this->get_location();
 
 	 _mesa_glsl_error(& loc, state,
@@ -4480,7 +4492,7 @@ ast_type_specifier::hir(exec_list *instructions,
 
       if (type->base_type == GLSL_TYPE_FLOAT
           && state->es_shader
-          && state->target == fragment_shader) {
+          && state->target == MESA_SHADER_FRAGMENT) {
          /* Section 4.5.3 (Default Precision Qualifiers) of the GLSL ES 1.00
           * spec says:
           *
@@ -4662,6 +4674,7 @@ ast_process_structure_or_interface_block(exec_list *instructions,
          fields[i].interpolation =
             interpret_interpolation_qualifier(qual, var_mode, state, &loc);
          fields[i].centroid = qual->flags.q.centroid ? 1 : 0;
+         fields[i].sample = qual->flags.q.sample ? 1 : 0;
 
          if (qual->flags.q.row_major || qual->flags.q.column_major) {
             if (!qual->flags.q.uniform) {
@@ -4782,7 +4795,7 @@ public:
 
    virtual ir_visitor_status visit(ir_dereference_variable *ir)
    {
-      if (ir->var->mode == mode && ir->var->get_interface_type() == block) {
+      if (ir->var->data.mode == mode && ir->var->get_interface_type() == block) {
          found = true;
          return visit_stop;
       }
@@ -4930,6 +4943,8 @@ ast_interface_block::hir(exec_list *instructions,
                earlier_per_vertex->fields.structure[j].interpolation;
             fields[i].centroid =
                earlier_per_vertex->fields.structure[j].centroid;
+            fields[i].sample =
+               earlier_per_vertex->fields.structure[j].sample;
          }
       }
 
@@ -4980,7 +4995,7 @@ ast_interface_block::hir(exec_list *instructions,
     *     variable (or input block, see interface blocks below) needs to be
     *     declared as an array.
     */
-   if (state->target == geometry_shader && !this->is_array &&
+   if (state->target == MESA_SHADER_GEOMETRY && !this->is_array &&
        var_mode == ir_var_shader_in) {
       _mesa_glsl_error(&loc, state, "geometry shader inputs must be arrays");
    }
@@ -5035,7 +5050,7 @@ ast_interface_block::hir(exec_list *instructions,
           * geometry shader input.
           */
          if (this->array_size == NULL &&
-             (state->target != geometry_shader || !this->layout.flags.q.in)) {
+             (state->target != MESA_SHADER_GEOMETRY || !this->layout.flags.q.in)) {
             _mesa_glsl_error(&loc, state,
                              "only geometry shader inputs may be unsized "
                              "instance block arrays");
@@ -5054,7 +5069,7 @@ ast_interface_block::hir(exec_list *instructions,
                                       var_mode);
       }
 
-      if (state->target == geometry_shader && var_mode == ir_var_shader_in)
+      if (state->target == MESA_SHADER_GEOMETRY && var_mode == ir_var_shader_in)
          handle_geometry_shader_input_decl(state, loc, var);
 
       if (ir_variable *earlier =
@@ -5063,7 +5078,7 @@ ast_interface_block::hir(exec_list *instructions,
             _mesa_glsl_error(&loc, state, "`%s' redeclared",
                              this->instance_name);
          }
-         earlier->how_declared = ir_var_declared_normally;
+         earlier->data.how_declared = ir_var_declared_normally;
          earlier->type = var->type;
          earlier->reinit_interface_type(block_type);
          delete var;
@@ -5082,8 +5097,9 @@ ast_interface_block::hir(exec_list *instructions,
             new(state) ir_variable(fields[i].type,
                                    ralloc_strdup(state, fields[i].name),
                                    var_mode);
-         var->interpolation = fields[i].interpolation;
-         var->centroid = fields[i].centroid;
+         var->data.interpolation = fields[i].interpolation;
+         var->data.centroid = fields[i].centroid;
+         var->data.sample = fields[i].sample;
          var->init_interface_type(block_type);
 
          if (redeclaring_per_vertex) {
@@ -5094,11 +5110,11 @@ ast_interface_block::hir(exec_list *instructions,
                _mesa_glsl_error(&loc, state,
                                 "redeclaration of gl_PerVertex can only "
                                 "include built-in variables");
-            } else if (earlier->how_declared == ir_var_declared_normally) {
+            } else if (earlier->data.how_declared == ir_var_declared_normally) {
                _mesa_glsl_error(&loc, state,
                                 "`%s' has already been redeclared", var->name);
             } else {
-               earlier->how_declared = ir_var_declared_in_block;
+               earlier->data.how_declared = ir_var_declared_in_block;
                earlier->reinit_interface_type(block_type);
             }
             continue;
@@ -5111,8 +5127,8 @@ ast_interface_block::hir(exec_list *instructions,
           * the UBO declaration itself doesn't get an ir_variable unless it
           * has an instance name.  This is ugly.
           */
-         var->explicit_binding = this->layout.flags.q.explicit_binding;
-         var->binding = this->layout.binding;
+         var->data.explicit_binding = this->layout.flags.q.explicit_binding;
+         var->data.binding = this->layout.binding;
 
          state->symbols->add_variable(var);
          instructions->push_tail(var);
@@ -5144,8 +5160,8 @@ ast_interface_block::hir(exec_list *instructions,
             ir_variable *const var = ((ir_instruction *) node)->as_variable();
             if (var != NULL &&
                 var->get_interface_type() == earlier_per_vertex &&
-                var->mode == var_mode) {
-               if (var->how_declared == ir_var_declared_normally) {
+                var->data.mode == var_mode) {
+               if (var->data.how_declared == ir_var_declared_normally) {
                   _mesa_glsl_error(&loc, state,
                                    "redeclaration of gl_PerVertex cannot "
                                    "follow a redeclaration of `%s'",
@@ -5200,7 +5216,7 @@ ast_gs_input_layout::hir(exec_list *instructions,
     */
    foreach_list (node, instructions) {
       ir_variable *var = ((ir_instruction *) node)->as_variable();
-      if (var == NULL || var->mode != ir_var_shader_in)
+      if (var == NULL || var->data.mode != ir_var_shader_in)
          continue;
 
       /* Note: gl_PrimitiveIDIn has mode ir_var_shader_in, but it's not an
@@ -5208,12 +5224,12 @@ ast_gs_input_layout::hir(exec_list *instructions,
        */
 
       if (var->type->is_unsized_array()) {
-         if (var->max_array_access >= num_vertices) {
+         if (var->data.max_array_access >= num_vertices) {
             _mesa_glsl_error(&loc, state,
                              "this geometry shader input layout implies %u"
                              " vertices, but an access to element %u of input"
                              " `%s' already exists", num_vertices,
-                             var->max_array_access, var->name);
+                             var->data.max_array_access, var->name);
          } else {
             var->type = glsl_type::get_array_instance(var->type->fields.array,
                                                       num_vertices);
@@ -5241,7 +5257,7 @@ detect_conflicting_assignments(struct _mesa_glsl_parse_state *state,
    foreach_list(node, instructions) {
       ir_variable *var = ((ir_instruction *)node)->as_variable();
 
-      if (!var || !var->assigned)
+      if (!var || !var->data.assigned)
 	 continue;
 
       if (strcmp(var->name, "gl_FragColor") == 0)
@@ -5249,8 +5265,8 @@ detect_conflicting_assignments(struct _mesa_glsl_parse_state *state,
       else if (strcmp(var->name, "gl_FragData") == 0)
 	 gl_FragData_assigned = true;
       else if (strncmp(var->name, "gl_", 3) != 0) {
-	 if (state->target == fragment_shader &&
-	     var->mode == ir_var_shader_out) {
+	 if (state->target == MESA_SHADER_FRAGMENT &&
+	     var->data.mode == ir_var_shader_out) {
 	    user_defined_fs_output_assigned = true;
 	    user_defined_fs_output = var;
 	 }
@@ -5331,7 +5347,7 @@ remove_per_vertex_blocks(exec_list *instructions,
    foreach_list_safe(node, instructions) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
       if (var != NULL && var->get_interface_type() == per_vertex &&
-          var->mode == mode) {
+          var->data.mode == mode) {
          state->symbols->disable_variable(var->name);
          var->remove();
       }
diff --git a/mesalib/src/glsl/ast_type.cpp b/mesalib/src/glsl/ast_type.cpp
index 2b088bf8b..d758bfa1f 100644
--- a/mesalib/src/glsl/ast_type.cpp
+++ b/mesalib/src/glsl/ast_type.cpp
@@ -90,7 +90,8 @@ ast_type_qualifier::has_storage() const
 bool
 ast_type_qualifier::has_auxiliary_storage() const
 {
-   return this->flags.q.centroid;
+   return this->flags.q.centroid
+          || this->flags.q.sample;
 }
 
 const char*
diff --git a/mesalib/src/glsl/builtin_functions.cpp b/mesalib/src/glsl/builtin_functions.cpp
index a45de091b..f22ace19d 100644
--- a/mesalib/src/glsl/builtin_functions.cpp
+++ b/mesalib/src/glsl/builtin_functions.cpp
@@ -77,7 +77,7 @@ always_available(const _mesa_glsl_parse_state *state)
 static bool
 compatibility_vs_only(const _mesa_glsl_parse_state *state)
 {
-   return state->target == vertex_shader &&
+   return state->target == MESA_SHADER_VERTEX &&
           state->language_version <= 130 &&
           !state->es_shader;
 }
@@ -85,13 +85,13 @@ compatibility_vs_only(const _mesa_glsl_parse_state *state)
 static bool
 fs_only(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader;
+   return state->target == MESA_SHADER_FRAGMENT;
 }
 
 static bool
 gs_only(const _mesa_glsl_parse_state *state)
 {
-   return state->target == geometry_shader;
+   return state->target == MESA_SHADER_GEOMETRY;
 }
 
 static bool
@@ -103,7 +103,7 @@ v110(const _mesa_glsl_parse_state *state)
 static bool
 v110_fs_only(const _mesa_glsl_parse_state *state)
 {
-   return !state->es_shader && state->target == fragment_shader;
+   return !state->es_shader && state->target == MESA_SHADER_FRAGMENT;
 }
 
 static bool
@@ -122,7 +122,7 @@ static bool
 v130_fs_only(const _mesa_glsl_parse_state *state)
 {
    return state->is_version(130, 300) &&
-          state->target == fragment_shader;
+          state->target == MESA_SHADER_FRAGMENT;
 }
 
 static bool
@@ -155,7 +155,7 @@ lod_exists_in_stage(const _mesa_glsl_parse_state *state)
     * Since ARB_shader_texture_lod can only be enabled on desktop GLSL, we
     * don't need to explicitly check state->es_shader.
     */
-   return state->target == vertex_shader ||
+   return state->target == MESA_SHADER_VERTEX ||
           state->is_version(130, 300) ||
           state->ARB_shader_texture_lod_enable;
 }
@@ -223,7 +223,7 @@ texture_array_lod(const _mesa_glsl_parse_state *state)
 static bool
 fs_texture_array(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader &&
+   return state->target == MESA_SHADER_FRAGMENT &&
           state->EXT_texture_array_enable;
 }
 
@@ -243,7 +243,7 @@ texture_multisample(const _mesa_glsl_parse_state *state)
 static bool
 fs_texture_cube_map_array(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader &&
+   return state->target == MESA_SHADER_FRAGMENT &&
           (state->is_version(400, 0) ||
            state->ARB_texture_cube_map_array_enable);
 }
@@ -265,7 +265,7 @@ texture_query_levels(const _mesa_glsl_parse_state *state)
 static bool
 texture_query_lod(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader &&
+   return state->target == MESA_SHADER_FRAGMENT &&
           state->ARB_texture_query_lod_enable;
 }
 
@@ -292,7 +292,7 @@ texture_gather_only(const _mesa_glsl_parse_state *state)
 static bool
 fs_oes_derivatives(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader &&
+   return state->target == MESA_SHADER_FRAGMENT &&
           (state->is_version(110, 300) ||
            state->OES_standard_derivatives_enable);
 }
@@ -318,7 +318,7 @@ tex3d(const _mesa_glsl_parse_state *state)
 static bool
 fs_tex3d(const _mesa_glsl_parse_state *state)
 {
-   return state->target == fragment_shader &&
+   return state->target == MESA_SHADER_FRAGMENT &&
           (!state->es_shader || state->OES_texture_3D_enable);
 }
 
@@ -357,8 +357,6 @@ public:
    ir_function_signature *find(_mesa_glsl_parse_state *state,
                                const char *name, exec_list *actual_parameters);
 
-private:
-   void *mem_ctx;
    /**
     * A shader to hold all the built-in signatures; created by this module.
     *
@@ -368,6 +366,9 @@ private:
     */
    gl_shader *shader;
 
+private:
+   void *mem_ctx;
+
    /** Global variables used by built-in functions. */
    ir_variable *gl_ModelViewProjectionMatrix;
    ir_variable *gl_Vertex;
@@ -608,8 +609,7 @@ builtin_builder::find(_mesa_glsl_parse_state *state,
     * that the "no matching signature" error will list potential candidates
     * from the available built-ins.
     */
-   state->builtins_to_link[0] = shader;
-   state->num_builtins_to_link = 1;
+   state->uses_builtin_functions = true;
 
    ir_function *f = shader->symbols->get_function(name);
    if (f == NULL)
@@ -4020,4 +4020,11 @@ _mesa_glsl_find_builtin_function(_mesa_glsl_parse_state *state,
 {
    return builtins.find(state, name, actual_parameters);
 }
+
+gl_shader *
+_mesa_glsl_get_builtin_function_shader()
+{
+   return builtins.shader;
+}
+
 /** @} */
diff --git a/mesalib/src/glsl/builtin_variables.cpp b/mesalib/src/glsl/builtin_variables.cpp
index d57324c2f..ff9acb8d0 100644
--- a/mesalib/src/glsl/builtin_variables.cpp
+++ b/mesalib/src/glsl/builtin_variables.cpp
@@ -332,6 +332,7 @@ per_vertex_accumulator::add_field(int slot, const glsl_type *type,
    this->fields[this->num_fields].location = slot;
    this->fields[this->num_fields].interpolation = INTERP_QUALIFIER_NONE;
    this->fields[this->num_fields].centroid = 0;
+   this->fields[this->num_fields].sample = 0;
    this->num_fields++;
 }
 
@@ -434,14 +435,14 @@ builtin_variable_generator::add_variable(const char *name,
                                          enum ir_variable_mode mode, int slot)
 {
    ir_variable *var = new(symtab) ir_variable(type, name, mode);
-   var->how_declared = ir_var_declared_implicitly;
+   var->data.how_declared = ir_var_declared_implicitly;
 
-   switch (var->mode) {
+   switch (var->data.mode) {
    case ir_var_auto:
    case ir_var_shader_in:
    case ir_var_uniform:
    case ir_var_system_value:
-      var->read_only = true;
+      var->data.read_only = true;
       break;
    case ir_var_shader_out:
       break;
@@ -454,9 +455,9 @@ builtin_variable_generator::add_variable(const char *name,
       break;
    }
 
-   var->location = slot;
-   var->explicit_location = (slot >= 0);
-   var->explicit_index = 0;
+   var->data.location = slot;
+   var->data.explicit_location = (slot >= 0);
+   var->data.explicit_index = 0;
 
    /* Once the variable is created an initialized, add it to the symbol table
     * and add the declaration to the IR stream.
@@ -523,7 +524,7 @@ builtin_variable_generator::add_const(const char *name, int value)
 					 ir_var_auto, -1);
    var->constant_value = new(var) ir_constant(value);
    var->constant_initializer = new(var) ir_constant(value);
-   var->has_initializer = true;
+   var->data.has_initializer = true;
    return var;
 }
 
@@ -792,9 +793,9 @@ builtin_variable_generator::generate_gs_special_vars()
     */
    ir_variable *var;
    var = add_input(VARYING_SLOT_PRIMITIVE_ID, int_t, "gl_PrimitiveIDIn");
-   var->interpolation = INTERP_QUALIFIER_FLAT;
+   var->data.interpolation = INTERP_QUALIFIER_FLAT;
    var = add_output(VARYING_SLOT_PRIMITIVE_ID, int_t, "gl_PrimitiveID");
-   var->interpolation = INTERP_QUALIFIER_FLAT;
+   var->data.interpolation = INTERP_QUALIFIER_FLAT;
 }
 
 
@@ -812,7 +813,7 @@ builtin_variable_generator::generate_fs_special_vars()
    if (state->is_version(150, 0)) {
       ir_variable *var =
          add_input(VARYING_SLOT_PRIMITIVE_ID, int_t, "gl_PrimitiveID");
-      var->interpolation = INTERP_QUALIFIER_FLAT;
+      var->data.interpolation = INTERP_QUALIFIER_FLAT;
    }
 
    /* gl_FragColor and gl_FragData were deprecated starting in desktop GLSL
@@ -857,6 +858,10 @@ builtin_variable_generator::generate_fs_special_vars()
        */
       add_output(FRAG_RESULT_SAMPLE_MASK, array(int_t, 1), "gl_SampleMask");
    }
+
+   if (state->ARB_gpu_shader5_enable) {
+      add_system_value(SYSTEM_VALUE_SAMPLE_MASK_IN, array(int_t, 1), "gl_SampleMaskIn");
+   }
 }
 
 
@@ -872,13 +877,13 @@ builtin_variable_generator::add_varying(int slot, const glsl_type *type,
                                         const char *name_as_gs_input)
 {
    switch (state->target) {
-   case geometry_shader:
+   case MESA_SHADER_GEOMETRY:
       this->per_vertex_in.add_field(slot, type, name);
       /* FALLTHROUGH */
-   case vertex_shader:
+   case MESA_SHADER_VERTEX:
       this->per_vertex_out.add_field(slot, type, name);
       break;
-   case fragment_shader:
+   case MESA_SHADER_FRAGMENT:
       add_input(slot, type, name);
       break;
    }
@@ -896,7 +901,7 @@ builtin_variable_generator::generate_varyings()
    add_varying(loc, type, name, name "In")
 
    /* gl_Position and gl_PointSize are not visible from fragment shaders. */
-   if (state->target != fragment_shader) {
+   if (state->target != MESA_SHADER_FRAGMENT) {
       ADD_VARYING(VARYING_SLOT_POS, vec4_t, "gl_Position");
       ADD_VARYING(VARYING_SLOT_PSIZ, float_t, "gl_PointSize");
    }
@@ -909,7 +914,7 @@ builtin_variable_generator::generate_varyings()
    if (compatibility) {
       ADD_VARYING(VARYING_SLOT_TEX0, array(vec4_t, 0), "gl_TexCoord");
       ADD_VARYING(VARYING_SLOT_FOGC, float_t, "gl_FogFragCoord");
-      if (state->target == fragment_shader) {
+      if (state->target == MESA_SHADER_FRAGMENT) {
          ADD_VARYING(VARYING_SLOT_COL0, vec4_t, "gl_Color");
          ADD_VARYING(VARYING_SLOT_COL1, vec4_t, "gl_SecondaryColor");
       } else {
@@ -921,13 +926,13 @@ builtin_variable_generator::generate_varyings()
       }
    }
 
-   if (state->target == geometry_shader) {
+   if (state->target == MESA_SHADER_GEOMETRY) {
       const glsl_type *per_vertex_in_type =
          this->per_vertex_in.construct_interface_instance();
       add_variable("gl_in", array(per_vertex_in_type, 0),
                    ir_var_shader_in, -1);
    }
-   if (state->target == vertex_shader || state->target == geometry_shader) {
+   if (state->target == MESA_SHADER_VERTEX || state->target == MESA_SHADER_GEOMETRY) {
       const glsl_type *per_vertex_out_type =
          this->per_vertex_out.construct_interface_instance();
       const glsl_struct_field *fields = per_vertex_out_type->fields.structure;
@@ -935,8 +940,9 @@ builtin_variable_generator::generate_varyings()
          ir_variable *var =
             add_variable(fields[i].name, fields[i].type, ir_var_shader_out,
                          fields[i].location);
-         var->interpolation = fields[i].interpolation;
-         var->centroid = fields[i].centroid;
+         var->data.interpolation = fields[i].interpolation;
+         var->data.centroid = fields[i].centroid;
+         var->data.sample = fields[i].sample;
          var->init_interface_type(per_vertex_out_type);
       }
    }
@@ -958,13 +964,13 @@ _mesa_glsl_initialize_variables(exec_list *instructions,
    gen.generate_varyings();
 
    switch (state->target) {
-   case vertex_shader:
+   case MESA_SHADER_VERTEX:
       gen.generate_vs_special_vars();
       break;
-   case geometry_shader:
+   case MESA_SHADER_GEOMETRY:
       gen.generate_gs_special_vars();
       break;
-   case fragment_shader:
+   case MESA_SHADER_FRAGMENT:
       gen.generate_fs_special_vars();
       break;
    }
diff --git a/mesalib/src/glsl/glsl_lexer.ll b/mesalib/src/glsl/glsl_lexer.ll
index 822d70d6b..50875bf3b 100644
--- a/mesalib/src/glsl/glsl_lexer.ll
+++ b/mesalib/src/glsl/glsl_lexer.ll
@@ -520,7 +520,7 @@ readonly	KEYWORD(0, 300, 0, 0, READONLY);
 writeonly	KEYWORD(0, 300, 0, 0, WRITEONLY);
 resource	KEYWORD(0, 300, 0, 0, RESOURCE);
 patch		KEYWORD(0, 300, 0, 0, PATCH);
-sample		KEYWORD(0, 300, 0, 0, SAMPLE);
+sample		KEYWORD_WITH_ALT(400, 300, 400, 0, yyextra->ARB_gpu_shader5_enable, SAMPLE);
 subroutine	KEYWORD(0, 300, 0, 0, SUBROUTINE);
 
 
diff --git a/mesalib/src/glsl/glsl_parser.yy b/mesalib/src/glsl/glsl_parser.yy
index 8ad7484fb..6cd129cae 100644
--- a/mesalib/src/glsl/glsl_parser.yy
+++ b/mesalib/src/glsl/glsl_parser.yy
@@ -1522,7 +1522,7 @@ type_qualifier:
    {
       if ($2.has_auxiliary_storage()) {
          _mesa_glsl_error(&@1, state,
-                          "duplicate auxiliary storage qualifier (centroid)");
+                          "duplicate auxiliary storage qualifier (centroid or sample)");
       }
 
       if (!state->ARB_shading_language_420pack_enable &&
@@ -1572,7 +1572,12 @@ auxiliary_storage_qualifier:
       memset(& $$, 0, sizeof($$));
       $$.flags.q.centroid = 1;
    }
-   /* TODO: "sample" and "patch" also go here someday. */
+   | SAMPLE
+   {
+      memset(& $$, 0, sizeof($$));
+      $$.flags.q.sample = 1;
+   }
+   /* TODO: "patch" also goes here someday. */
 
 storage_qualifier:
    CONST_TOK
@@ -2215,11 +2220,11 @@ basic_interface_block:
        * "It is illegal to have an input block in a vertex shader
        *  or an output block in a fragment shader"
        */
-      if ((state->target == vertex_shader) && $1.flags.q.in) {
+      if ((state->target == MESA_SHADER_VERTEX) && $1.flags.q.in) {
          _mesa_glsl_error(& @1, state,
                           "`in' interface block is not allowed for "
                           "a vertex shader");
-      } else if ((state->target == fragment_shader) && $1.flags.q.out) {
+      } else if ((state->target == MESA_SHADER_FRAGMENT) && $1.flags.q.out) {
          _mesa_glsl_error(& @1, state,
                           "`out' interface block is not allowed for "
                           "a fragment shader");
@@ -2373,7 +2378,7 @@ layout_defaults:
    {
       void *ctx = state;
       $$ = NULL;
-      if (state->target != geometry_shader) {
+      if (state->target != MESA_SHADER_GEOMETRY) {
          _mesa_glsl_error(& @1, state,
                           "input layout qualifiers only valid in "
                           "geometry shaders");
@@ -2401,7 +2406,7 @@ layout_defaults:
 
    | layout_qualifier OUT_TOK ';'
    {
-      if (state->target != geometry_shader) {
+      if (state->target != MESA_SHADER_GEOMETRY) {
          _mesa_glsl_error(& @1, state,
                           "out layout qualifiers only valid in "
                           "geometry shaders");
diff --git a/mesalib/src/glsl/glsl_parser_extras.cpp b/mesalib/src/glsl/glsl_parser_extras.cpp
index d76d94b7a..5f19368d8 100644
--- a/mesalib/src/glsl/glsl_parser_extras.cpp
+++ b/mesalib/src/glsl/glsl_parser_extras.cpp
@@ -58,9 +58,9 @@ _mesa_glsl_parse_state::_mesa_glsl_parse_state(struct gl_context *_ctx,
    : ctx(_ctx), switch_state()
 {
    switch (target) {
-   case GL_VERTEX_SHADER:   this->target = vertex_shader; break;
-   case GL_FRAGMENT_SHADER: this->target = fragment_shader; break;
-   case GL_GEOMETRY_SHADER: this->target = geometry_shader; break;
+   case GL_VERTEX_SHADER:   this->target = MESA_SHADER_VERTEX; break;
+   case GL_FRAGMENT_SHADER: this->target = MESA_SHADER_FRAGMENT; break;
+   case GL_GEOMETRY_SHADER: this->target = MESA_SHADER_GEOMETRY; break;
    }
 
    this->scanner = NULL;
@@ -76,7 +76,8 @@ _mesa_glsl_parse_state::_mesa_glsl_parse_state(struct gl_context *_ctx,
    this->loop_nesting_ast = NULL;
 
    this->struct_specifier_depth = 0;
-   this->num_builtins_to_link = 0;
+
+   this->uses_builtin_functions = false;
 
    /* Set default language version and extensions */
    this->language_version = ctx->Const.ForceGLSLVersion ?
@@ -366,12 +367,12 @@ _mesa_glsl_shader_target_name(GLenum type)
  * our internal enum into short stage names.
  */
 const char *
-_mesa_glsl_shader_target_name(enum _mesa_glsl_parser_targets target)
+_mesa_glsl_shader_target_name(gl_shader_type target)
 {
    switch (target) {
-   case vertex_shader:   return "vertex";
-   case fragment_shader: return "fragment";
-   case geometry_shader: return "geometry";
+   case MESA_SHADER_VERTEX:   return "vertex";
+   case MESA_SHADER_FRAGMENT: return "fragment";
+   case MESA_SHADER_GEOMETRY: return "geometry";
    }
 
    assert(!"Should not get here.");
@@ -877,6 +878,8 @@ _mesa_ast_type_qualifier_print(const struct ast_type_qualifier *q)
 
    if (q->flags.q.centroid)
       printf("centroid ");
+   if (q->flags.q.sample)
+      printf("sample ");
    if (q->flags.q.uniform)
       printf("uniform ");
    if (q->flags.q.smooth)
@@ -1532,10 +1535,7 @@ _mesa_glsl_compile_shader(struct gl_context *ctx, struct gl_shader *shader,
    shader->InfoLog = state->info_log;
    shader->Version = state->language_version;
    shader->IsES = state->es_shader;
-
-   memcpy(shader->builtins_to_link, state->builtins_to_link,
-          sizeof(shader->builtins_to_link[0]) * state->num_builtins_to_link);
-   shader->num_builtins_to_link = state->num_builtins_to_link;
+   shader->uses_builtin_functions = state->uses_builtin_functions;
 
    if (shader->UniformBlocks)
       ralloc_free(shader->UniformBlocks);
diff --git a/mesalib/src/glsl/glsl_parser_extras.h b/mesalib/src/glsl/glsl_parser_extras.h
index cb998d80e..7aa78d611 100644
--- a/mesalib/src/glsl/glsl_parser_extras.h
+++ b/mesalib/src/glsl/glsl_parser_extras.h
@@ -34,12 +34,6 @@
 #include <stdlib.h>
 #include "glsl_symbol_table.h"
 
-enum _mesa_glsl_parser_targets {
-   vertex_shader,
-   geometry_shader,
-   fragment_shader
-};
-
 struct gl_context;
 
 struct glsl_switch_state {
@@ -175,7 +169,7 @@ struct _mesa_glsl_parse_state {
 
    bool es_shader;
    unsigned language_version;
-   enum _mesa_glsl_parser_targets target;
+   gl_shader_type target;
 
    /**
     * Number of nested struct_specifier levels
@@ -365,9 +359,7 @@ struct _mesa_glsl_parse_state {
    /** Extensions supported by the OpenGL implementation. */
    const struct gl_extensions *extensions;
 
-   /** Shaders containing built-in functions that are used for linking. */
-   struct gl_shader *builtins_to_link[16];
-   unsigned num_builtins_to_link;
+   bool uses_builtin_functions;
 
    /**
     * For geometry shaders, size of the most recently seen input declaration
@@ -437,7 +429,7 @@ extern bool _mesa_glsl_process_extension(const char *name, YYLTYPE *name_locp,
  * Get the textual name of the specified shader target
  */
 extern const char *
-_mesa_glsl_shader_target_name(enum _mesa_glsl_parser_targets target);
+_mesa_glsl_shader_target_name(gl_shader_type target);
 
 
 #endif /* __cplusplus */
diff --git a/mesalib/src/glsl/glsl_types.cpp b/mesalib/src/glsl/glsl_types.cpp
index f74013096..12d4ac0ee 100644
--- a/mesalib/src/glsl/glsl_types.cpp
+++ b/mesalib/src/glsl/glsl_types.cpp
@@ -103,6 +103,7 @@ glsl_type::glsl_type(const glsl_struct_field *fields, unsigned num_fields,
       this->fields.structure[i].location = fields[i].location;
       this->fields.structure[i].interpolation = fields[i].interpolation;
       this->fields.structure[i].centroid = fields[i].centroid;
+      this->fields.structure[i].sample = fields[i].sample;
       this->fields.structure[i].row_major = fields[i].row_major;
    }
 }
@@ -130,6 +131,7 @@ glsl_type::glsl_type(const glsl_struct_field *fields, unsigned num_fields,
       this->fields.structure[i].location = fields[i].location;
       this->fields.structure[i].interpolation = fields[i].interpolation;
       this->fields.structure[i].centroid = fields[i].centroid;
+      this->fields.structure[i].sample = fields[i].sample;
       this->fields.structure[i].row_major = fields[i].row_major;
    }
 }
@@ -483,6 +485,9 @@ glsl_type::record_key_compare(const void *a, const void *b)
       if (key1->fields.structure[i].centroid
           != key2->fields.structure[i].centroid)
          return 1;
+      if (key1->fields.structure[i].sample
+          != key2->fields.structure[i].sample)
+         return 1;
    }
 
    return 0;
diff --git a/mesalib/src/glsl/glsl_types.h b/mesalib/src/glsl/glsl_types.h
index 96eee5e64..fb7c9288d 100644
--- a/mesalib/src/glsl/glsl_types.h
+++ b/mesalib/src/glsl/glsl_types.h
@@ -634,6 +634,12 @@ struct glsl_struct_field {
     * in ir_variable::centroid).  0 otherwise.
     */
    unsigned centroid:1;
+
+   /**
+    * For interface blocks, 1 if this variable uses sample interpolation (as
+    * in ir_variable::sample). 0 otherwise.
+    */
+   unsigned sample:1;
 };
 
 static inline unsigned int
diff --git a/mesalib/src/glsl/ir.cpp b/mesalib/src/glsl/ir.cpp
index 64a2c5af9..8af58b847 100644
--- a/mesalib/src/glsl/ir.cpp
+++ b/mesalib/src/glsl/ir.cpp
@@ -1277,11 +1277,6 @@ ir_constant::is_basis() const
 ir_loop::ir_loop()
 {
    this->ir_type = ir_type_loop;
-   this->cmp = ir_unop_neg;
-   this->from = NULL;
-   this->to = NULL;
-   this->increment = NULL;
-   this->counter = NULL;
 }
 
 
@@ -1364,7 +1359,7 @@ ir_dereference::is_lvalue() const
 
    /* Every l-value derference chain eventually ends in a variable.
     */
-   if ((var == NULL) || var->read_only)
+   if ((var == NULL) || var->data.read_only)
       return false;
 
    /* From page 17 (page 23 of the PDF) of the GLSL 1.20 spec:
@@ -1584,29 +1579,36 @@ ir_swizzle::variable_referenced() const
 
 ir_variable::ir_variable(const struct glsl_type *type, const char *name,
 			 ir_variable_mode mode)
-   : max_array_access(0), max_ifc_array_access(NULL),
-     read_only(false), centroid(false), invariant(false),
-     how_declared(ir_var_declared_normally), mode(mode),
-     interpolation(INTERP_QUALIFIER_NONE), atomic()
+   : max_ifc_array_access(NULL)
 {
    this->ir_type = ir_type_variable;
    this->type = type;
    this->name = ralloc_strdup(this, name);
-   this->explicit_location = false;
-   this->has_initializer = false;
-   this->location = -1;
-   this->location_frac = 0;
+   this->data.explicit_location = false;
+   this->data.has_initializer = false;
+   this->data.location = -1;
+   this->data.location_frac = 0;
    this->warn_extension = NULL;
    this->constant_value = NULL;
    this->constant_initializer = NULL;
-   this->origin_upper_left = false;
-   this->pixel_center_integer = false;
-   this->depth_layout = ir_depth_layout_none;
-   this->used = false;
+   this->data.origin_upper_left = false;
+   this->data.pixel_center_integer = false;
+   this->data.depth_layout = ir_depth_layout_none;
+   this->data.used = false;
+   this->data.read_only = false;
+   this->data.centroid = false;
+   this->data.sample = false;
+   this->data.invariant = false;
+   this->data.how_declared = ir_var_declared_normally;
+   this->data.mode = mode;
+   this->data.interpolation = INTERP_QUALIFIER_NONE;
+   this->data.max_array_access = 0;
+   this->data.atomic.buffer_index = 0;
+   this->data.atomic.offset = 0;
 
    if (type != NULL) {
       if (type->base_type == GLSL_TYPE_SAMPLER)
-         this->read_only = true;
+         this->data.read_only = true;
 
       if (type->is_interface())
          this->init_interface_type(type);
@@ -1634,9 +1636,9 @@ interpolation_string(unsigned interpolation)
 glsl_interp_qualifier
 ir_variable::determine_interpolation_mode(bool flat_shade)
 {
-   if (this->interpolation != INTERP_QUALIFIER_NONE)
-      return (glsl_interp_qualifier) this->interpolation;
-   int location = this->location;
+   if (this->data.interpolation != INTERP_QUALIFIER_NONE)
+      return (glsl_interp_qualifier) this->data.interpolation;
+   int location = this->data.location;
    bool is_gl_Color =
       location == VARYING_SLOT_COL0 || location == VARYING_SLOT_COL1;
    if (flat_shade && is_gl_Color)
@@ -1706,10 +1708,11 @@ ir_function_signature::qualifiers_match(exec_list *params)
       ir_variable *a = (ir_variable *)iter_a.get();
       ir_variable *b = (ir_variable *)iter_b.get();
 
-      if (a->read_only != b->read_only ||
-	  !modes_match(a->mode, b->mode) ||
-	  a->interpolation != b->interpolation ||
-	  a->centroid != b->centroid) {
+      if (a->data.read_only != b->data.read_only ||
+	  !modes_match(a->data.mode, b->data.mode) ||
+	  a->data.interpolation != b->data.interpolation ||
+	  a->data.centroid != b->data.centroid ||
+         a->data.sample != b->data.sample) {
 
 	 /* parameter a's qualifiers don't match */
 	 return a->name;
@@ -1729,12 +1732,6 @@ ir_function_signature::replace_parameters(exec_list *new_params)
     * parameter information comes from the function prototype, it may either
     * specify incorrect parameter names or not have names at all.
     */
-   foreach_iter(exec_list_iterator, iter, parameters) {
-      assert(((ir_instruction *) iter.get())->as_variable() != NULL);
-
-      iter.remove();
-   }
-
    new_params->move_nodes_to(&parameters);
 }
 
@@ -1899,9 +1896,9 @@ vertices_per_prim(GLenum prim)
 const char *
 mode_string(const ir_variable *var)
 {
-   switch (var->mode) {
+   switch (var->data.mode) {
    case ir_var_auto:
-      return (var->read_only) ? "global constant" : "global variable";
+      return (var->data.read_only) ? "global constant" : "global variable";
 
    case ir_var_uniform:
       return "uniform";
diff --git a/mesalib/src/glsl/ir.h b/mesalib/src/glsl/ir.h
index 4f775da4b..780959b73 100644
--- a/mesalib/src/glsl/ir.h
+++ b/mesalib/src/glsl/ir.h
@@ -86,6 +86,7 @@ enum ir_node_type {
    ir_type_max /**< maximum ir_type enum number, for validation */
 };
 
+
 /**
  * Base class of all IR instructions
  */
@@ -402,7 +403,7 @@ public:
     */
    inline bool is_in_uniform_block() const
    {
-      return this->mode == ir_var_uniform && this->interface_type != NULL;
+      return this->data.mode == ir_var_uniform && this->interface_type != NULL;
    }
 
    /**
@@ -502,13 +503,6 @@ public:
    const char *name;
 
    /**
-    * Highest element accessed with a constant expression array index
-    *
-    * Not used for non-array variables.
-    */
-   unsigned max_array_access;
-
-   /**
     * For variables which satisfy the is_interface_instance() predicate, this
     * points to an array of integers such that if the ith member of the
     * interface block is an array, max_ifc_array_access[i] is the maximum
@@ -521,177 +515,189 @@ public:
     */
    unsigned *max_ifc_array_access;
 
-   /**
-    * Is the variable read-only?
-    *
-    * This is set for variables declared as \c const, shader inputs,
-    * and uniforms.
-    */
-   unsigned read_only:1;
-   unsigned centroid:1;
-   unsigned invariant:1;
-
-   /**
-    * Has this variable been used for reading or writing?
-    *
-    * Several GLSL semantic checks require knowledge of whether or not a
-    * variable has been used.  For example, it is an error to redeclare a
-    * variable as invariant after it has been used.
-    *
-    * This is only maintained in the ast_to_hir.cpp path, not in
-    * Mesa's fixed function or ARB program paths.
-    */
-   unsigned used:1;
-
-   /**
-    * Has this variable been statically assigned?
-    *
-    * This answers whether the variable was assigned in any path of
-    * the shader during ast_to_hir.  This doesn't answer whether it is
-    * still written after dead code removal, nor is it maintained in
-    * non-ast_to_hir.cpp (GLSL parsing) paths.
-    */
-   unsigned assigned:1;
-
-   /**
-    * Enum indicating how the variable was declared.  See
-    * ir_var_declaration_type.
-    *
-    * This is used to detect certain kinds of illegal variable redeclarations.
-    */
-   unsigned how_declared:2;
-
-   /**
-    * Storage class of the variable.
-    *
-    * \sa ir_variable_mode
-    */
-   unsigned mode:4;
-
-   /**
-    * Interpolation mode for shader inputs / outputs
-    *
-    * \sa ir_variable_interpolation
-    */
-   unsigned interpolation:2;
-
-   /**
-    * \name ARB_fragment_coord_conventions
-    * @{
-    */
-   unsigned origin_upper_left:1;
-   unsigned pixel_center_integer:1;
-   /*@}*/
+   struct ir_variable_data {
 
-   /**
-    * Was the location explicitly set in the shader?
-    *
-    * If the location is explicitly set in the shader, it \b cannot be changed
-    * by the linker or by the API (e.g., calls to \c glBindAttribLocation have
-    * no effect).
-    */
-   unsigned explicit_location:1;
-   unsigned explicit_index:1;
-
-   /**
-    * Was an initial binding explicitly set in the shader?
-    *
-    * If so, constant_value contains an integer ir_constant representing the
-    * initial binding point.
-    */
-   unsigned explicit_binding:1;
+      /**
+       * Is the variable read-only?
+       *
+       * This is set for variables declared as \c const, shader inputs,
+       * and uniforms.
+       */
+      unsigned read_only:1;
+      unsigned centroid:1;
+      unsigned sample:1;
+      unsigned invariant:1;
+
+      /**
+       * Has this variable been used for reading or writing?
+       *
+       * Several GLSL semantic checks require knowledge of whether or not a
+       * variable has been used.  For example, it is an error to redeclare a
+       * variable as invariant after it has been used.
+       *
+       * This is only maintained in the ast_to_hir.cpp path, not in
+       * Mesa's fixed function or ARB program paths.
+       */
+      unsigned used:1;
+
+      /**
+       * Has this variable been statically assigned?
+       *
+       * This answers whether the variable was assigned in any path of
+       * the shader during ast_to_hir.  This doesn't answer whether it is
+       * still written after dead code removal, nor is it maintained in
+       * non-ast_to_hir.cpp (GLSL parsing) paths.
+       */
+      unsigned assigned:1;
 
-   /**
-    * Does this variable have an initializer?
-    *
-    * This is used by the linker to cross-validiate initializers of global
-    * variables.
-    */
-   unsigned has_initializer:1;
+      /**
+       * Enum indicating how the variable was declared.  See
+       * ir_var_declaration_type.
+       *
+       * This is used to detect certain kinds of illegal variable redeclarations.
+       */
+      unsigned how_declared:2;
 
-   /**
-    * Is this variable a generic output or input that has not yet been matched
-    * up to a variable in another stage of the pipeline?
-    *
-    * This is used by the linker as scratch storage while assigning locations
-    * to generic inputs and outputs.
-    */
-   unsigned is_unmatched_generic_inout:1;
+      /**
+       * Storage class of the variable.
+       *
+       * \sa ir_variable_mode
+       */
+      unsigned mode:4;
 
-   /**
-    * If non-zero, then this variable may be packed along with other variables
-    * into a single varying slot, so this offset should be applied when
-    * accessing components.  For example, an offset of 1 means that the x
-    * component of this variable is actually stored in component y of the
-    * location specified by \c location.
-    */
-   unsigned location_frac:2;
+      /**
+       * Interpolation mode for shader inputs / outputs
+       *
+       * \sa ir_variable_interpolation
+       */
+      unsigned interpolation:2;
 
-   /**
-    * Non-zero if this variable was created by lowering a named interface
-    * block which was not an array.
-    *
-    * Note that this variable and \c from_named_ifc_block_array will never
-    * both be non-zero.
-    */
-   unsigned from_named_ifc_block_nonarray:1;
+      /**
+       * \name ARB_fragment_coord_conventions
+       * @{
+       */
+      unsigned origin_upper_left:1;
+      unsigned pixel_center_integer:1;
+      /*@}*/
+
+      /**
+       * Was the location explicitly set in the shader?
+       *
+       * If the location is explicitly set in the shader, it \b cannot be changed
+       * by the linker or by the API (e.g., calls to \c glBindAttribLocation have
+       * no effect).
+       */
+      unsigned explicit_location:1;
+      unsigned explicit_index:1;
+
+      /**
+       * Was an initial binding explicitly set in the shader?
+       *
+       * If so, constant_value contains an integer ir_constant representing the
+       * initial binding point.
+       */
+      unsigned explicit_binding:1;
 
-   /**
-    * Non-zero if this variable was created by lowering a named interface
-    * block which was an array.
-    *
-    * Note that this variable and \c from_named_ifc_block_nonarray will never
-    * both be non-zero.
-    */
-   unsigned from_named_ifc_block_array:1;
+      /**
+       * Does this variable have an initializer?
+       *
+       * This is used by the linker to cross-validiate initializers of global
+       * variables.
+       */
+      unsigned has_initializer:1;
+
+      /**
+       * Is this variable a generic output or input that has not yet been matched
+       * up to a variable in another stage of the pipeline?
+       *
+       * This is used by the linker as scratch storage while assigning locations
+       * to generic inputs and outputs.
+       */
+      unsigned is_unmatched_generic_inout:1;
+
+      /**
+       * If non-zero, then this variable may be packed along with other variables
+       * into a single varying slot, so this offset should be applied when
+       * accessing components.  For example, an offset of 1 means that the x
+       * component of this variable is actually stored in component y of the
+       * location specified by \c location.
+       */
+      unsigned location_frac:2;
+
+      /**
+       * Non-zero if this variable was created by lowering a named interface
+       * block which was not an array.
+       *
+       * Note that this variable and \c from_named_ifc_block_array will never
+       * both be non-zero.
+       */
+      unsigned from_named_ifc_block_nonarray:1;
+
+      /**
+       * Non-zero if this variable was created by lowering a named interface
+       * block which was an array.
+       *
+       * Note that this variable and \c from_named_ifc_block_nonarray will never
+       * both be non-zero.
+       */
+      unsigned from_named_ifc_block_array:1;
 
-   /**
-    * \brief Layout qualifier for gl_FragDepth.
-    *
-    * This is not equal to \c ir_depth_layout_none if and only if this
-    * variable is \c gl_FragDepth and a layout qualifier is specified.
-    */
-   ir_depth_layout depth_layout;
+      /**
+       * \brief Layout qualifier for gl_FragDepth.
+       *
+       * This is not equal to \c ir_depth_layout_none if and only if this
+       * variable is \c gl_FragDepth and a layout qualifier is specified.
+       */
+      ir_depth_layout depth_layout;
+
+      /**
+       * Storage location of the base of this variable
+       *
+       * The precise meaning of this field depends on the nature of the variable.
+       *
+       *   - Vertex shader input: one of the values from \c gl_vert_attrib.
+       *   - Vertex shader output: one of the values from \c gl_varying_slot.
+       *   - Geometry shader input: one of the values from \c gl_varying_slot.
+       *   - Geometry shader output: one of the values from \c gl_varying_slot.
+       *   - Fragment shader input: one of the values from \c gl_varying_slot.
+       *   - Fragment shader output: one of the values from \c gl_frag_result.
+       *   - Uniforms: Per-stage uniform slot number for default uniform block.
+       *   - Uniforms: Index within the uniform block definition for UBO members.
+       *   - Other: This field is not currently used.
+       *
+       * If the variable is a uniform, shader input, or shader output, and the
+       * slot has not been assigned, the value will be -1.
+       */
+      int location;
 
-   /**
-    * Storage location of the base of this variable
-    *
-    * The precise meaning of this field depends on the nature of the variable.
-    *
-    *   - Vertex shader input: one of the values from \c gl_vert_attrib.
-    *   - Vertex shader output: one of the values from \c gl_varying_slot.
-    *   - Geometry shader input: one of the values from \c gl_varying_slot.
-    *   - Geometry shader output: one of the values from \c gl_varying_slot.
-    *   - Fragment shader input: one of the values from \c gl_varying_slot.
-    *   - Fragment shader output: one of the values from \c gl_frag_result.
-    *   - Uniforms: Per-stage uniform slot number for default uniform block.
-    *   - Uniforms: Index within the uniform block definition for UBO members.
-    *   - Other: This field is not currently used.
-    *
-    * If the variable is a uniform, shader input, or shader output, and the
-    * slot has not been assigned, the value will be -1.
-    */
-   int location;
+      /**
+       * output index for dual source blending.
+       */
+      int index;
 
-   /**
-    * output index for dual source blending.
-    */
-   int index;
+      /**
+       * Initial binding point for a sampler or UBO.
+       *
+       * For array types, this represents the binding point for the first element.
+       */
+      int binding;
 
-   /**
-    * Initial binding point for a sampler or UBO.
-    *
-    * For array types, this represents the binding point for the first element.
-    */
-   int binding;
+      /**
+       * Location an atomic counter is stored at.
+       */
+      struct {
+         unsigned buffer_index;
+         unsigned offset;
+      } atomic;
+
+      /**
+       * Highest element accessed with a constant expression array index
+       *
+       * Not used for non-array variables.
+       */
+      unsigned max_array_access;
 
-   /**
-    * Location an atomic counter is stored at.
-    */
-   struct {
-      unsigned buffer_index;
-      unsigned offset;
-   } atomic;
+   } data;
 
    /**
     * Built-in state that backs this uniform
@@ -1022,34 +1028,6 @@ public:
 
    /** List of ir_instruction that make up the body of the loop. */
    exec_list body_instructions;
-
-   /**
-    * \name Loop counter and controls
-    *
-    * Represents a loop like a FORTRAN \c do-loop.
-    *
-    * \note
-    * If \c from and \c to are the same value, the loop will execute once.
-    */
-   /*@{*/
-   ir_rvalue *from;             /** Value of the loop counter on the first
-				 * iteration of the loop.
-				 */
-   ir_rvalue *to;               /** Value of the loop counter on the last
-				 * iteration of the loop.
-				 */
-   ir_rvalue *increment;
-   ir_variable *counter;
-
-   /**
-    * Comparison operation in the loop terminator.
-    *
-    * If any of the loop control fields are non-\c NULL, this field must be
-    * one of \c ir_binop_less, \c ir_binop_greater, \c ir_binop_lequal,
-    * \c ir_binop_gequal, \c ir_binop_equal, or \c ir_binop_nequal.
-    */
-   int cmp;
-   /*@}*/
 };
 
 
@@ -2344,6 +2322,9 @@ extern ir_function_signature *
 _mesa_glsl_find_builtin_function(_mesa_glsl_parse_state *state,
                                  const char *name, exec_list *actual_parameters);
 
+extern gl_shader *
+_mesa_glsl_get_builtin_function_shader(void);
+
 extern void
 _mesa_glsl_release_functions(void);
 
diff --git a/mesalib/src/glsl/ir_clone.cpp b/mesalib/src/glsl/ir_clone.cpp
index 40ed33afc..4e5cf68ca 100644
--- a/mesalib/src/glsl/ir_clone.cpp
+++ b/mesalib/src/glsl/ir_clone.cpp
@@ -41,35 +41,19 @@ ir_variable *
 ir_variable::clone(void *mem_ctx, struct hash_table *ht) const
 {
    ir_variable *var = new(mem_ctx) ir_variable(this->type, this->name,
-					       (ir_variable_mode) this->mode);
+					       (ir_variable_mode) this->data.mode);
 
-   var->max_array_access = this->max_array_access;
+   var->data.max_array_access = this->data.max_array_access;
    if (this->is_interface_instance()) {
       var->max_ifc_array_access =
          rzalloc_array(var, unsigned, this->interface_type->length);
       memcpy(var->max_ifc_array_access, this->max_ifc_array_access,
              this->interface_type->length * sizeof(unsigned));
    }
-   var->read_only = this->read_only;
-   var->centroid = this->centroid;
-   var->invariant = this->invariant;
-   var->interpolation = this->interpolation;
-   var->location = this->location;
-   var->index = this->index;
-   var->binding = this->binding;
-   var->atomic.buffer_index = this->atomic.buffer_index;
-   var->atomic.offset = this->atomic.offset;
+
+   memcpy(&var->data, &this->data, sizeof(var->data));
+
    var->warn_extension = this->warn_extension;
-   var->origin_upper_left = this->origin_upper_left;
-   var->pixel_center_integer = this->pixel_center_integer;
-   var->explicit_location = this->explicit_location;
-   var->explicit_index = this->explicit_index;
-   var->explicit_binding = this->explicit_binding;
-   var->has_initializer = this->has_initializer;
-   var->depth_layout = this->depth_layout;
-   var->assigned = this->assigned;
-   var->how_declared = this->how_declared;
-   var->used = this->used;
 
    var->num_state_slots = this->num_state_slots;
    if (this->state_slots) {
@@ -157,20 +141,11 @@ ir_loop::clone(void *mem_ctx, struct hash_table *ht) const
 {
    ir_loop *new_loop = new(mem_ctx) ir_loop();
 
-   if (this->from)
-      new_loop->from = this->from->clone(mem_ctx, ht);
-   if (this->to)
-      new_loop->to = this->to->clone(mem_ctx, ht);
-   if (this->increment)
-      new_loop->increment = this->increment->clone(mem_ctx, ht);
-   new_loop->counter = counter;
-
    foreach_iter(exec_list_iterator, iter, this->body_instructions) {
       ir_instruction *ir = (ir_instruction *)iter.get();
       new_loop->body_instructions.push_tail(ir->clone(mem_ctx, ht));
    }
 
-   new_loop->cmp = this->cmp;
    return new_loop;
 }
 
diff --git a/mesalib/src/glsl/ir_constant_expression.cpp b/mesalib/src/glsl/ir_constant_expression.cpp
index 54bfe5832..c24d3c96b 100755
--- a/mesalib/src/glsl/ir_constant_expression.cpp
+++ b/mesalib/src/glsl/ir_constant_expression.cpp
@@ -1576,7 +1576,7 @@ ir_dereference_variable::constant_expression_value(struct hash_table *variable_c
    /* The constant_value of a uniform variable is its initializer,
     * not the lifetime constant value of the uniform.
     */
-   if (var->mode == ir_var_uniform)
+   if (var->data.mode == ir_var_uniform)
       return NULL;
 
    if (!var->constant_value)
diff --git a/mesalib/src/glsl/ir_function.cpp b/mesalib/src/glsl/ir_function.cpp
index 53cf469d9..bd5318d23 100644
--- a/mesalib/src/glsl/ir_function.cpp
+++ b/mesalib/src/glsl/ir_function.cpp
@@ -66,7 +66,7 @@ parameter_lists_match(const exec_list *list_a, const exec_list *list_b)
 
       /* Try to find an implicit conversion from actual to param. */
       inexact_match = true;
-      switch ((enum ir_variable_mode)(param->mode)) {
+      switch ((enum ir_variable_mode)(param->data.mode)) {
       case ir_var_auto:
       case ir_var_uniform:
       case ir_var_temporary:
diff --git a/mesalib/src/glsl/ir_hv_accept.cpp b/mesalib/src/glsl/ir_hv_accept.cpp
index 941b25e97..2a1f70e5b 100644
--- a/mesalib/src/glsl/ir_hv_accept.cpp
+++ b/mesalib/src/glsl/ir_hv_accept.cpp
@@ -91,26 +91,6 @@ ir_loop::accept(ir_hierarchical_visitor *v)
    if (s == visit_stop)
       return s;
 
-   if (s != visit_continue_with_parent) {
-      if (this->from) {
-	 s = this->from->accept(v);
-	 if (s != visit_continue)
-	    return (s == visit_continue_with_parent) ? visit_continue : s;
-      }
-
-      if (this->to) {
-	 s = this->to->accept(v);
-	 if (s != visit_continue)
-	    return (s == visit_continue_with_parent) ? visit_continue : s;
-      }
-
-      if (this->increment) {
-	 s = this->increment->accept(v);
-	 if (s != visit_continue)
-	    return (s == visit_continue_with_parent) ? visit_continue : s;
-      }
-   }
-
    return v->visit_leave(this);
 }
 
diff --git a/mesalib/src/glsl/ir_print_visitor.cpp b/mesalib/src/glsl/ir_print_visitor.cpp
index f85e573c4..01c5f7f1c 100644
--- a/mesalib/src/glsl/ir_print_visitor.cpp
+++ b/mesalib/src/glsl/ir_print_visitor.cpp
@@ -148,8 +148,9 @@ void ir_print_visitor::visit(ir_variable *ir)
 {
    printf("(declare ");
 
-   const char *const cent = (ir->centroid) ? "centroid " : "";
-   const char *const inv = (ir->invariant) ? "invariant " : "";
+   const char *const cent = (ir->data.centroid) ? "centroid " : "";
+   const char *const samp = (ir->data.sample) ? "sample " : "";
+   const char *const inv = (ir->data.invariant) ? "invariant " : "";
    const char *const mode[] = { "", "uniform ", "shader_in ", "shader_out ",
                                 "in ", "out ", "inout ",
 			        "const_in ", "sys ", "temporary " };
@@ -157,8 +158,8 @@ void ir_print_visitor::visit(ir_variable *ir)
    const char *const interp[] = { "", "smooth", "flat", "noperspective" };
    STATIC_ASSERT(ARRAY_SIZE(interp) == INTERP_QUALIFIER_COUNT);
 
-   printf("(%s%s%s%s) ",
-	  cent, inv, mode[ir->mode], interp[ir->interpolation]);
+   printf("(%s%s%s%s%s) ",
+	  cent, samp, inv, mode[ir->data.mode], interp[ir->data.interpolation]);
 
    print_type(ir->type);
    printf(" %s)", unique_name(ir));
@@ -416,9 +417,9 @@ void ir_print_visitor::visit(ir_constant *ir)
             if (ir->value.f[i] == 0.0f)
                /* 0.0 == -0.0, so print with %f to get the proper sign. */
                printf("%.1f", ir->value.f[i]);
-            else if (abs(ir->value.f[i]) < 0.000001f)
+            else if (fabs(ir->value.f[i]) < 0.000001f)
                printf("%a", ir->value.f[i]);
-            else if (abs(ir->value.f[i]) > 1000000.0f)
+            else if (fabs(ir->value.f[i]) > 1000000.0f)
                printf("%e", ir->value.f[i]);
             else
                printf("%f", ir->value.f[i]);
@@ -522,19 +523,7 @@ ir_print_visitor::visit(ir_if *ir)
 void
 ir_print_visitor::visit(ir_loop *ir)
 {
-   printf("(loop (");
-   if (ir->counter != NULL)
-      ir->counter->accept(this);
-   printf(") (");
-   if (ir->from != NULL)
-      ir->from->accept(this);
-   printf(") (");
-   if (ir->to != NULL)
-      ir->to->accept(this);
-   printf(") (");
-   if (ir->increment != NULL)
-      ir->increment->accept(this);
-   printf(") (\n");
+   printf("(loop (\n");
    indentation++;
 
    foreach_iter(exec_list_iterator, iter, ir->body_instructions) {
diff --git a/mesalib/src/glsl/ir_reader.cpp b/mesalib/src/glsl/ir_reader.cpp
index 00e2db9a3..7970112ec 100644
--- a/mesalib/src/glsl/ir_reader.cpp
+++ b/mesalib/src/glsl/ir_reader.cpp
@@ -412,33 +412,35 @@ ir_reader::read_declaration(s_expression *expr)
 
       // FINISHME: Check for duplicate/conflicting qualifiers.
       if (strcmp(qualifier->value(), "centroid") == 0) {
-	 var->centroid = 1;
+	 var->data.centroid = 1;
+      } else if (strcmp(qualifier->value(), "sample") == 0) {
+         var->data.sample = 1;
       } else if (strcmp(qualifier->value(), "invariant") == 0) {
-	 var->invariant = 1;
+	 var->data.invariant = 1;
       } else if (strcmp(qualifier->value(), "uniform") == 0) {
-	 var->mode = ir_var_uniform;
+	 var->data.mode = ir_var_uniform;
       } else if (strcmp(qualifier->value(), "auto") == 0) {
-	 var->mode = ir_var_auto;
+	 var->data.mode = ir_var_auto;
       } else if (strcmp(qualifier->value(), "in") == 0) {
-	 var->mode = ir_var_function_in;
+	 var->data.mode = ir_var_function_in;
       } else if (strcmp(qualifier->value(), "shader_in") == 0) {
-         var->mode = ir_var_shader_in;
+         var->data.mode = ir_var_shader_in;
       } else if (strcmp(qualifier->value(), "const_in") == 0) {
-	 var->mode = ir_var_const_in;
+	 var->data.mode = ir_var_const_in;
       } else if (strcmp(qualifier->value(), "out") == 0) {
-	 var->mode = ir_var_function_out;
+	 var->data.mode = ir_var_function_out;
       } else if (strcmp(qualifier->value(), "shader_out") == 0) {
-	 var->mode = ir_var_shader_out;
+	 var->data.mode = ir_var_shader_out;
       } else if (strcmp(qualifier->value(), "inout") == 0) {
-	 var->mode = ir_var_function_inout;
+	 var->data.mode = ir_var_function_inout;
       } else if (strcmp(qualifier->value(), "temporary") == 0) {
-	 var->mode = ir_var_temporary;
+	 var->data.mode = ir_var_temporary;
       } else if (strcmp(qualifier->value(), "smooth") == 0) {
-	 var->interpolation = INTERP_QUALIFIER_SMOOTH;
+	 var->data.interpolation = INTERP_QUALIFIER_SMOOTH;
       } else if (strcmp(qualifier->value(), "flat") == 0) {
-	 var->interpolation = INTERP_QUALIFIER_FLAT;
+	 var->data.interpolation = INTERP_QUALIFIER_FLAT;
       } else if (strcmp(qualifier->value(), "noperspective") == 0) {
-	 var->interpolation = INTERP_QUALIFIER_NOPERSPECTIVE;
+	 var->data.interpolation = INTERP_QUALIFIER_NOPERSPECTIVE;
       } else {
 	 ir_read_error(expr, "unknown qualifier: %s", qualifier->value());
 	 return NULL;
@@ -486,18 +488,16 @@ ir_reader::read_if(s_expression *expr, ir_loop *loop_ctx)
 ir_loop *
 ir_reader::read_loop(s_expression *expr)
 {
-   s_expression *s_counter, *s_from, *s_to, *s_inc, *s_body;
+   s_expression *s_body;
 
-   s_pattern pat[] = { "loop", s_counter, s_from, s_to, s_inc, s_body };
-   if (!MATCH(expr, pat)) {
-      ir_read_error(expr, "expected (loop <counter> <from> <to> "
-			  "<increment> <body>)");
+   s_pattern loop_pat[] = { "loop", s_body };
+   if (!MATCH(expr, loop_pat)) {
+      ir_read_error(expr, "expected (loop <body>)");
       return NULL;
    }
 
-   // FINISHME: actually read the count/from/to fields.
-
    ir_loop *loop = new(mem_ctx) ir_loop;
+
    read_instructions(&loop->body_instructions, s_body, loop);
    if (state->error) {
       delete loop;
diff --git a/mesalib/src/glsl/ir_set_program_inouts.cpp b/mesalib/src/glsl/ir_set_program_inouts.cpp
index ab23538c3..0b49eb2b6 100644
--- a/mesalib/src/glsl/ir_set_program_inouts.cpp
+++ b/mesalib/src/glsl/ir_set_program_inouts.cpp
@@ -27,7 +27,7 @@
  * Sets the InputsRead and OutputsWritten of Mesa programs.
  *
  * Additionally, for fragment shaders, sets the InterpQualifier array, the
- * IsCentroid bitfield, and the UsesDFdy flag.
+ * IsCentroid and IsSample bitfields, and the UsesDFdy flag.
  *
  * Mesa programs (gl_program, not gl_shader_program) have a set of
  * flags indicating which varyings are read and written.  Computing
@@ -75,9 +75,9 @@ private:
 static inline bool
 is_shader_inout(ir_variable *var)
 {
-   return var->mode == ir_var_shader_in ||
-          var->mode == ir_var_shader_out ||
-          var->mode == ir_var_system_value;
+   return var->data.mode == ir_var_shader_in ||
+          var->data.mode == ir_var_shader_out ||
+          var->data.mode == ir_var_system_value;
 }
 
 static void
@@ -93,20 +93,24 @@ mark(struct gl_program *prog, ir_variable *var, int offset, int len,
     */
 
    for (int i = 0; i < len; i++) {
-      GLbitfield64 bitfield = BITFIELD64_BIT(var->location + var->index + offset + i);
-      if (var->mode == ir_var_shader_in) {
+      GLbitfield64 bitfield =
+         BITFIELD64_BIT(var->data.location + var->data.index + offset + i);
+      if (var->data.mode == ir_var_shader_in) {
 	 prog->InputsRead |= bitfield;
          if (is_fragment_shader) {
             gl_fragment_program *fprog = (gl_fragment_program *) prog;
-            fprog->InterpQualifier[var->location + var->index + offset + i] =
-               (glsl_interp_qualifier) var->interpolation;
-            if (var->centroid)
+            fprog->InterpQualifier[var->data.location +
+                                   var->data.index + offset + i] =
+               (glsl_interp_qualifier) var->data.interpolation;
+            if (var->data.centroid)
                fprog->IsCentroid |= bitfield;
+            if (var->data.sample)
+               fprog->IsSample |= bitfield;
          }
-      } else if (var->mode == ir_var_system_value) {
+      } else if (var->data.mode == ir_var_system_value) {
          prog->SystemValuesRead |= bitfield;
       } else {
-         assert(var->mode == ir_var_shader_out);
+         assert(var->data.mode == ir_var_shader_out);
 	 prog->OutputsWritten |= bitfield;
       }
    }
@@ -121,7 +125,7 @@ ir_set_program_inouts_visitor::mark_whole_variable(ir_variable *var)
 {
    const glsl_type *type = var->type;
    if (this->shader_type == GL_GEOMETRY_SHADER &&
-       var->mode == ir_var_shader_in && type->is_array()) {
+       var->data.mode == ir_var_shader_in && type->is_array()) {
       type = type->fields.array;
    }
 
@@ -160,7 +164,7 @@ ir_set_program_inouts_visitor::try_mark_partial_variable(ir_variable *var,
    const glsl_type *type = var->type;
 
    if (this->shader_type == GL_GEOMETRY_SHADER &&
-       var->mode == ir_var_shader_in) {
+       var->data.mode == ir_var_shader_in) {
       /* The only geometry shader input that is not an array is
        * gl_PrimitiveIDIn, and in that case, this code will never be reached,
        * because gl_PrimitiveIDIn can't be indexed into in array fashion.
@@ -242,7 +246,7 @@ ir_set_program_inouts_visitor::visit_enter(ir_dereference_array *ir)
       if (ir_dereference_variable * const deref_var =
           inner_array->array->as_dereference_variable()) {
          if (this->shader_type == GL_GEOMETRY_SHADER &&
-             deref_var->var->mode == ir_var_shader_in) {
+             deref_var->var->data.mode == ir_var_shader_in) {
             /* foo is a geometry shader input, so i is the vertex, and j the
              * part of the input we're accessing.
              */
@@ -261,7 +265,7 @@ ir_set_program_inouts_visitor::visit_enter(ir_dereference_array *ir)
               ir->array->as_dereference_variable()) {
       /* ir => foo[i], where foo is a variable. */
       if (this->shader_type == GL_GEOMETRY_SHADER &&
-          deref_var->var->mode == ir_var_shader_in) {
+          deref_var->var->data.mode == ir_var_shader_in) {
          /* foo is a geometry shader input, so i is the vertex, and we're
           * accessing the entire input.
           */
@@ -341,6 +345,7 @@ do_set_program_inouts(exec_list *instructions, struct gl_program *prog,
       gl_fragment_program *fprog = (gl_fragment_program *) prog;
       memset(fprog->InterpQualifier, 0, sizeof(fprog->InterpQualifier));
       fprog->IsCentroid = 0;
+      fprog->IsSample = 0;
       fprog->UsesDFdy = false;
       fprog->UsesKill = false;
    }
diff --git a/mesalib/src/glsl/ir_validate.cpp b/mesalib/src/glsl/ir_validate.cpp
index 13e41a089..4bbb3ce65 100644
--- a/mesalib/src/glsl/ir_validate.cpp
+++ b/mesalib/src/glsl/ir_validate.cpp
@@ -63,7 +63,6 @@ public:
 
    virtual ir_visitor_status visit_enter(ir_if *ir);
 
-   virtual ir_visitor_status visit_leave(ir_loop *ir);
    virtual ir_visitor_status visit_enter(ir_function *ir);
    virtual ir_visitor_status visit_leave(ir_function *ir);
    virtual ir_visitor_status visit_enter(ir_function_signature *ir);
@@ -149,42 +148,6 @@ ir_validate::visit_enter(ir_if *ir)
 
 
 ir_visitor_status
-ir_validate::visit_leave(ir_loop *ir)
-{
-   if (ir->counter != NULL) {
-      if ((ir->from == NULL) || (ir->to == NULL) || (ir->increment == NULL)) {
-	 printf("ir_loop has invalid loop controls:\n"
-		"    counter:   %p\n"
-		"    from:      %p\n"
-		"    to:        %p\n"
-		"    increment: %p\n",
-		(void *) ir->counter, (void *) ir->from, (void *) ir->to,
-                (void *) ir->increment);
-	 abort();
-      }
-
-      if ((ir->cmp < ir_binop_less) || (ir->cmp > ir_binop_nequal)) {
-	 printf("ir_loop has invalid comparitor %d\n", ir->cmp);
-	 abort();
-      }
-   } else {
-      if ((ir->from != NULL) || (ir->to != NULL) || (ir->increment != NULL)) {
-	 printf("ir_loop has invalid loop controls:\n"
-		"    counter:   %p\n"
-		"    from:      %p\n"
-		"    to:        %p\n"
-		"    increment: %p\n",
-		(void *) ir->counter, (void *) ir->from, (void *) ir->to,
-                (void *) ir->increment);
-	 abort();
-      }
-   }
-
-   return visit_continue;
-}
-
-
-ir_visitor_status
 ir_validate::visit_enter(ir_function *ir)
 {
    /* Function definitions cannot be nested.
@@ -679,9 +642,9 @@ ir_validate::visit(ir_variable *ir)
     * to be out of bounds.
     */
    if (ir->type->array_size() > 0) {
-      if (ir->max_array_access >= ir->type->length) {
+      if (ir->data.max_array_access >= ir->type->length) {
 	 printf("ir_variable has maximum access out of bounds (%d vs %d)\n",
-		ir->max_array_access, ir->type->length - 1);
+		ir->data.max_array_access, ir->type->length - 1);
 	 ir->print();
 	 abort();
       }
@@ -707,7 +670,7 @@ ir_validate::visit(ir_variable *ir)
       }
    }
 
-   if (ir->constant_initializer != NULL && !ir->has_initializer) {
+   if (ir->constant_initializer != NULL && !ir->data.has_initializer) {
       printf("ir_variable didn't have an initializer, but has a constant "
 	     "initializer value.\n");
       ir->print();
@@ -789,8 +752,8 @@ ir_validate::visit_enter(ir_call *ir)
          printf("ir_call parameter type mismatch:\n");
          goto dump_ir;
       }
-      if (formal_param->mode == ir_var_function_out
-          || formal_param->mode == ir_var_function_inout) {
+      if (formal_param->data.mode == ir_var_function_out
+          || formal_param->data.mode == ir_var_function_inout) {
          if (!actual_param->is_lvalue()) {
             printf("ir_call out/inout parameters must be lvalues:\n");
             goto dump_ir;
diff --git a/mesalib/src/glsl/link_atomics.cpp b/mesalib/src/glsl/link_atomics.cpp
index 2466bbd79..33903ad54 100644
--- a/mesalib/src/glsl/link_atomics.cpp
+++ b/mesalib/src/glsl/link_atomics.cpp
@@ -73,16 +73,16 @@ namespace {
       const active_atomic_counter *const first = (active_atomic_counter *) a;
       const active_atomic_counter *const second = (active_atomic_counter *) b;
 
-      return int(first->var->atomic.offset) - int(second->var->atomic.offset);
+      return int(first->var->data.atomic.offset) - int(second->var->data.atomic.offset);
    }
 
    bool
    check_atomic_counters_overlap(const ir_variable *x, const ir_variable *y)
    {
-      return ((x->atomic.offset >= y->atomic.offset &&
-               x->atomic.offset < y->atomic.offset + y->type->atomic_size()) ||
-              (y->atomic.offset >= x->atomic.offset &&
-               y->atomic.offset < x->atomic.offset + x->type->atomic_size()));
+      return ((x->data.atomic.offset >= y->data.atomic.offset &&
+               x->data.atomic.offset < y->data.atomic.offset + y->type->atomic_size()) ||
+              (y->data.atomic.offset >= x->data.atomic.offset &&
+               y->data.atomic.offset < x->data.atomic.offset + x->type->atomic_size()));
    }
 
    active_atomic_buffer *
@@ -107,7 +107,7 @@ namespace {
                unsigned id;
                bool found = prog->UniformHash->get(id, var->name);
                assert(found);
-               active_atomic_buffer *buf = &buffers[var->binding];
+               active_atomic_buffer *buf = &buffers[var->data.binding];
 
                /* If this is the first time the buffer is used, increment
                 * the counter of buffers used.
@@ -118,7 +118,7 @@ namespace {
                buf->push_back(id, var);
 
                buf->stage_references[i]++;
-               buf->size = MAX2(buf->size, var->atomic.offset +
+               buf->size = MAX2(buf->size, var->data.atomic.offset +
                                 var->type->atomic_size());
             }
          }
@@ -143,7 +143,7 @@ namespace {
                linker_error(prog, "Atomic counter %s declared at offset %d "
                             "which is already in use.",
                             buffers[i].counters[j].var->name,
-                            buffers[i].counters[j].var->atomic.offset);
+                            buffers[i].counters[j].var->data.atomic.offset);
             }
          }
       }
@@ -190,9 +190,9 @@ link_assign_atomic_counter_resources(struct gl_context *ctx,
          gl_uniform_storage *const storage = &prog->UniformStorage[id];
 
          mab.Uniforms[j] = id;
-         var->atomic.buffer_index = i;
+         var->data.atomic.buffer_index = i;
          storage->atomic_buffer_index = i;
-         storage->offset = var->atomic.offset;
+         storage->offset = var->data.atomic.offset;
          storage->array_stride = (var->type->is_array() ?
                                   var->type->element_type()->atomic_size() : 0);
       }
diff --git a/mesalib/src/glsl/link_functions.cpp b/mesalib/src/glsl/link_functions.cpp
index 68aa62032..56f3f207e 100644
--- a/mesalib/src/glsl/link_functions.cpp
+++ b/mesalib/src/glsl/link_functions.cpp
@@ -201,8 +201,9 @@ public:
             if (formal_param->type->is_array()) {
                ir_dereference_variable *deref = actual_param->as_dereference_variable();
                if (deref && deref->var && deref->var->type->is_array()) {
-                  deref->var->max_array_access =
-                     MAX2(formal_param->max_array_access, deref->var->max_array_access);
+                  deref->var->data.max_array_access =
+                     MAX2(formal_param->data.max_array_access,
+                         deref->var->data.max_array_access);
                }
             }
          }
@@ -234,8 +235,9 @@ public:
                 * we need to track the maximal access to the array as linking
                 * pulls more functions in that access the array.
                 */
-               var->max_array_access =
-                  MAX2(var->max_array_access, ir->var->max_array_access);
+               var->data.max_array_access =
+                  MAX2(var->data.max_array_access,
+                       ir->var->data.max_array_access);
 
                if (var->type->length == 0 && ir->var->type->length != 0)
                   var->type = ir->var->type;
diff --git a/mesalib/src/glsl/link_interface_blocks.cpp b/mesalib/src/glsl/link_interface_blocks.cpp
index 6900fa94e..476963642 100644
--- a/mesalib/src/glsl/link_interface_blocks.cpp
+++ b/mesalib/src/glsl/link_interface_blocks.cpp
@@ -60,7 +60,7 @@ struct interface_block_definition
          if (var->type->is_array())
             array_size = var->type->length;
       }
-      explicitly_declared = (var->how_declared != ir_var_declared_implicitly);
+      explicitly_declared = (var->data.how_declared != ir_var_declared_implicitly);
    }
 
    /**
@@ -270,7 +270,7 @@ validate_intrastage_interface_blocks(struct gl_shader_program *prog,
             continue;
 
          interface_block_definitions *definitions;
-         switch (var->mode) {
+         switch (var->data.mode) {
          case ir_var_shader_in:
             definitions = &in_interfaces;
             break;
@@ -298,7 +298,7 @@ validate_intrastage_interface_blocks(struct gl_shader_program *prog,
              */
             definitions->store(def);
          } else if (!intrastage_match(prev_def, &def,
-                                      (ir_variable_mode) var->mode)) {
+                                      (ir_variable_mode) var->data.mode)) {
             linker_error(prog, "definitions of interface block `%s' do not"
                          " match\n", iface_type->name);
             return;
@@ -318,7 +318,7 @@ validate_interstage_inout_blocks(struct gl_shader_program *prog,
    /* Add input interfaces from the consumer to the symbol table. */
    foreach_list(node, consumer->ir) {
       ir_variable *var = ((ir_instruction *) node)->as_variable();
-      if (!var || !var->get_interface_type() || var->mode != ir_var_shader_in)
+      if (!var || !var->get_interface_type() || var->data.mode != ir_var_shader_in)
          continue;
 
       definitions.store(interface_block_definition(var));
@@ -327,7 +327,7 @@ validate_interstage_inout_blocks(struct gl_shader_program *prog,
    /* Verify that the producer's output interfaces match. */
    foreach_list(node, producer->ir) {
       ir_variable *var = ((ir_instruction *) node)->as_variable();
-      if (!var || !var->get_interface_type() || var->mode != ir_var_shader_out)
+      if (!var || !var->get_interface_type() || var->data.mode != ir_var_shader_out)
          continue;
 
       interface_block_definition *consumer_def =
@@ -361,7 +361,7 @@ validate_interstage_uniform_blocks(struct gl_shader_program *prog,
       const gl_shader *stage = stages[i];
       foreach_list(node, stage->ir) {
          ir_variable *var = ((ir_instruction *) node)->as_variable();
-         if (!var || !var->get_interface_type() || var->mode != ir_var_uniform)
+         if (!var || !var->get_interface_type() || var->data.mode != ir_var_uniform)
             continue;
 
          interface_block_definition *old_def =
diff --git a/mesalib/src/glsl/link_uniform_initializers.cpp b/mesalib/src/glsl/link_uniform_initializers.cpp
index 786aaf0b4..04daa1760 100644
--- a/mesalib/src/glsl/link_uniform_initializers.cpp
+++ b/mesalib/src/glsl/link_uniform_initializers.cpp
@@ -224,15 +224,15 @@ link_set_uniform_initializers(struct gl_shader_program *prog)
       foreach_list(node, shader->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-	 if (!var || var->mode != ir_var_uniform)
+	 if (!var || var->data.mode != ir_var_uniform)
 	    continue;
 
 	 if (!mem_ctx)
 	    mem_ctx = ralloc_context(NULL);
 
-         if (var->explicit_binding) {
+         if (var->data.explicit_binding) {
             linker::set_uniform_binding(mem_ctx, prog, var->name,
-                                        var->type, var->binding);
+                                        var->type, var->data.binding);
          } else if (var->constant_value) {
             linker::set_uniform_initializer(mem_ctx, prog, var->name,
                                             var->type, var->constant_value);
diff --git a/mesalib/src/glsl/link_uniforms.cpp b/mesalib/src/glsl/link_uniforms.cpp
index 95e75d73d..1cbe80b2e 100644
--- a/mesalib/src/glsl/link_uniforms.cpp
+++ b/mesalib/src/glsl/link_uniforms.cpp
@@ -75,7 +75,7 @@ program_resource_visitor::process(ir_variable *var)
     */
 
    /* Only strdup the name if we actually will need to modify it. */
-   if (var->from_named_ifc_block_array) {
+   if (var->data.from_named_ifc_block_array) {
       /* lower_named_interface_blocks created this variable by lowering an
        * interface block array to an array variable.  For example if the
        * original source code was:
@@ -108,7 +108,7 @@ program_resource_visitor::process(ir_variable *var)
          recursion(var->type, &name, new_length, false, NULL);
       }
       ralloc_free(name);
-   } else if (var->from_named_ifc_block_nonarray) {
+   } else if (var->data.from_named_ifc_block_nonarray) {
       /* lower_named_interface_blocks created this variable by lowering a
        * named interface block (non-array) to an ordinary variable.  For
        * example if the original source code was:
@@ -408,10 +408,10 @@ public:
             const struct gl_uniform_block *const block =
                &prog->UniformBlocks[ubo_block_index];
 
-            assert(var->location != -1);
+            assert(var->data.location != -1);
 
             const struct gl_uniform_buffer_variable *const ubo_var =
-               &block->Uniforms[var->location];
+               &block->Uniforms[var->data.location];
 
             ubo_row_major = ubo_var->RowMajor;
             ubo_byte_offset = ubo_var->Offset;
@@ -637,10 +637,10 @@ link_update_uniform_buffer_variables(struct gl_shader *shader)
       if ((var == NULL) || !var->is_in_uniform_block())
 	 continue;
 
-      assert(var->mode == ir_var_uniform);
+      assert(var->data.mode == ir_var_uniform);
 
       if (var->is_interface_instance()) {
-         var->location = 0;
+         var->data.location = 0;
          continue;
       }
 
@@ -669,13 +669,13 @@ link_update_uniform_buffer_variables(struct gl_shader *shader)
 
                if (strncmp(var->name, begin, l) == 0) {
                   found = true;
-                  var->location = j;
+                  var->data.location = j;
                   break;
                }
             } else if (!strcmp(var->name,
                                shader->UniformBlocks[i].Uniforms[j].Name)) {
 	       found = true;
-	       var->location = j;
+	       var->data.location = j;
 	       break;
 	    }
 	 }
@@ -767,7 +767,7 @@ link_assign_uniform_locations(struct gl_shader_program *prog)
       foreach_list(node, sh->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-	 if ((var == NULL) || (var->mode != ir_var_uniform))
+	 if ((var == NULL) || (var->data.mode != ir_var_uniform))
 	    continue;
 
 	 /* FINISHME: Update code to process built-in uniforms!
@@ -818,7 +818,7 @@ link_assign_uniform_locations(struct gl_shader_program *prog)
       foreach_list(node, prog->_LinkedShaders[i]->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-	 if ((var == NULL) || (var->mode != ir_var_uniform))
+	 if ((var == NULL) || (var->data.mode != ir_var_uniform))
 	    continue;
 
 	 /* FINISHME: Update code to process built-in uniforms!
diff --git a/mesalib/src/glsl/link_varyings.cpp b/mesalib/src/glsl/link_varyings.cpp
index be36b5f8f..229a4cb85 100644
--- a/mesalib/src/glsl/link_varyings.cpp
+++ b/mesalib/src/glsl/link_varyings.cpp
@@ -93,31 +93,43 @@ cross_validate_types_and_qualifiers(struct gl_shader_program *prog,
 
    /* Check that all of the qualifiers match between stages.
     */
-   if (input->centroid != output->centroid) {
+   if (input->data.centroid != output->data.centroid) {
       linker_error(prog,
                    "%s shader output `%s' %s centroid qualifier, "
                    "but %s shader input %s centroid qualifier\n",
                    _mesa_glsl_shader_target_name(producer_type),
                    output->name,
-                   (output->centroid) ? "has" : "lacks",
+                   (output->data.centroid) ? "has" : "lacks",
                    _mesa_glsl_shader_target_name(consumer_type),
-                   (input->centroid) ? "has" : "lacks");
+                   (input->data.centroid) ? "has" : "lacks");
       return;
    }
 
-   if (input->invariant != output->invariant) {
+   if (input->data.sample != output->data.sample) {
+      linker_error(prog,
+                   "%s shader output `%s' %s sample qualifier, "
+                   "but %s shader input %s sample qualifier\n",
+                   _mesa_glsl_shader_target_name(producer_type),
+                   output->name,
+                   (output->data.sample) ? "has" : "lacks",
+                   _mesa_glsl_shader_target_name(consumer_type),
+                   (input->data.sample) ? "has" : "lacks");
+      return;
+   }
+
+   if (input->data.invariant != output->data.invariant) {
       linker_error(prog,
                    "%s shader output `%s' %s invariant qualifier, "
                    "but %s shader input %s invariant qualifier\n",
                    _mesa_glsl_shader_target_name(producer_type),
                    output->name,
-                   (output->invariant) ? "has" : "lacks",
+                   (output->data.invariant) ? "has" : "lacks",
                    _mesa_glsl_shader_target_name(consumer_type),
-                   (input->invariant) ? "has" : "lacks");
+                   (input->data.invariant) ? "has" : "lacks");
       return;
    }
 
-   if (input->interpolation != output->interpolation) {
+   if (input->data.interpolation != output->data.interpolation) {
       linker_error(prog,
                    "%s shader output `%s' specifies %s "
                    "interpolation qualifier, "
@@ -125,9 +137,9 @@ cross_validate_types_and_qualifiers(struct gl_shader_program *prog,
                    "interpolation qualifier\n",
                    _mesa_glsl_shader_target_name(producer_type),
                    output->name,
-                   interpolation_string(output->interpolation),
+                   interpolation_string(output->data.interpolation),
                    _mesa_glsl_shader_target_name(consumer_type),
-                   interpolation_string(input->interpolation));
+                   interpolation_string(input->data.interpolation));
       return;
    }
 }
@@ -143,11 +155,11 @@ cross_validate_front_and_back_color(struct gl_shader_program *prog,
                                     GLenum consumer_type,
                                     GLenum producer_type)
 {
-   if (front_color != NULL && front_color->assigned)
+   if (front_color != NULL && front_color->data.assigned)
       cross_validate_types_and_qualifiers(prog, input, front_color,
                                           consumer_type, producer_type);
 
-   if (back_color != NULL && back_color->assigned)
+   if (back_color != NULL && back_color->data.assigned)
       cross_validate_types_and_qualifiers(prog, input, back_color,
                                           consumer_type, producer_type);
 }
@@ -166,7 +178,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
    foreach_list(node, producer->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if ((var == NULL) || (var->mode != ir_var_shader_out))
+      if ((var == NULL) || (var->data.mode != ir_var_shader_out))
 	 continue;
 
       parameters.add_variable(var);
@@ -184,10 +196,10 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
    foreach_list(node, consumer->ir) {
       ir_variable *const input = ((ir_instruction *) node)->as_variable();
 
-      if ((input == NULL) || (input->mode != ir_var_shader_in))
+      if ((input == NULL) || (input->data.mode != ir_var_shader_in))
 	 continue;
 
-      if (strcmp(input->name, "gl_Color") == 0 && input->used) {
+      if (strcmp(input->name, "gl_Color") == 0 && input->data.used) {
          const ir_variable *const front_color =
             parameters.get_variable("gl_FrontColor");
 
@@ -197,7 +209,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
          cross_validate_front_and_back_color(prog, input,
                                              front_color, back_color,
                                              consumer->Type, producer->Type);
-      } else if (strcmp(input->name, "gl_SecondaryColor") == 0 && input->used) {
+      } else if (strcmp(input->name, "gl_SecondaryColor") == 0 && input->data.used) {
          const ir_variable *const front_color =
             parameters.get_variable("gl_FrontSecondaryColor");
 
@@ -317,8 +329,8 @@ tfeedback_decl::assign_location(struct gl_context *ctx,
    assert(this->is_varying());
 
    unsigned fine_location
-      = this->matched_candidate->toplevel_var->location * 4
-      + this->matched_candidate->toplevel_var->location_frac
+      = this->matched_candidate->toplevel_var->data.location * 4
+      + this->matched_candidate->toplevel_var->data.location_frac
       + this->matched_candidate->offset;
 
    if (this->matched_candidate->type->is_array()) {
@@ -734,7 +746,7 @@ varying_matches::~varying_matches()
 void
 varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var)
 {
-   if (!producer_var->is_unmatched_generic_inout) {
+   if (!producer_var->data.is_unmatched_generic_inout) {
       /* Either a location already exists for this variable (since it is part
        * of fixed functionality), or it has already been recorded as part of a
        * previous match.
@@ -752,12 +764,14 @@ varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var)
        * regardless of where they appear.  We can trivially satisfy that
        * requirement by changing the interpolation type to flat here.
        */
-      producer_var->centroid = false;
-      producer_var->interpolation = INTERP_QUALIFIER_FLAT;
+      producer_var->data.centroid = false;
+      producer_var->data.sample = false;
+      producer_var->data.interpolation = INTERP_QUALIFIER_FLAT;
 
       if (consumer_var) {
-         consumer_var->centroid = false;
-         consumer_var->interpolation = INTERP_QUALIFIER_FLAT;
+         consumer_var->data.centroid = false;
+         consumer_var->data.sample = false;
+         consumer_var->data.interpolation = INTERP_QUALIFIER_FLAT;
       }
    }
 
@@ -784,9 +798,9 @@ varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var)
    this->matches[this->num_matches].producer_var = producer_var;
    this->matches[this->num_matches].consumer_var = consumer_var;
    this->num_matches++;
-   producer_var->is_unmatched_generic_inout = 0;
+   producer_var->data.is_unmatched_generic_inout = 0;
    if (consumer_var)
-      consumer_var->is_unmatched_generic_inout = 0;
+      consumer_var->data.is_unmatched_generic_inout = 0;
 }
 
 
@@ -838,12 +852,12 @@ varying_matches::store_locations(unsigned producer_base,
       unsigned slot = generic_location / 4;
       unsigned offset = generic_location % 4;
 
-      producer_var->location = producer_base + slot;
-      producer_var->location_frac = offset;
+      producer_var->data.location = producer_base + slot;
+      producer_var->data.location_frac = offset;
       if (consumer_var) {
-         assert(consumer_var->location == -1);
-         consumer_var->location = consumer_base + slot;
-         consumer_var->location_frac = offset;
+         assert(consumer_var->data.location == -1);
+         consumer_var->data.location = consumer_base + slot;
+         consumer_var->data.location_frac = offset;
       }
    }
 }
@@ -873,9 +887,9 @@ varying_matches::compute_packing_class(ir_variable *var)
     *
     * Therefore, the packing class depends only on the interpolation type.
     */
-   unsigned packing_class = var->centroid ? 1 : 0;
+   unsigned packing_class = var->data.centroid | (var->data.sample << 1);
    packing_class *= 4;
-   packing_class += var->interpolation;
+   packing_class += var->data.interpolation;
    return packing_class;
 }
 
@@ -933,8 +947,8 @@ is_varying_var(GLenum shaderType, const ir_variable *var)
 {
    /* Only fragment shaders will take a varying variable as an input */
    if (shaderType == GL_FRAGMENT_SHADER &&
-       var->mode == ir_var_shader_in) {
-      switch (var->location) {
+       var->data.mode == ir_var_shader_in) {
+      switch (var->data.location) {
       case VARYING_SLOT_POS:
       case VARYING_SLOT_FACE:
       case VARYING_SLOT_PNTC:
@@ -1082,7 +1096,7 @@ assign_varying_locations(struct gl_context *ctx,
          ir_variable *const input_var =
             ((ir_instruction *) node)->as_variable();
 
-         if ((input_var != NULL) && (input_var->mode == ir_var_shader_in)) {
+         if ((input_var != NULL) && (input_var->data.mode == ir_var_shader_in)) {
             if (input_var->get_interface_type() != NULL) {
                char *const iface_field_name =
                   ralloc_asprintf(mem_ctx, "%s.%s",
@@ -1101,7 +1115,7 @@ assign_varying_locations(struct gl_context *ctx,
    foreach_list(node, producer->ir) {
       ir_variable *const output_var = ((ir_instruction *) node)->as_variable();
 
-      if ((output_var == NULL) || (output_var->mode != ir_var_shader_out))
+      if ((output_var == NULL) || (output_var->data.mode != ir_var_shader_out))
 	 continue;
 
       tfeedback_candidate_generator g(mem_ctx, tfeedback_candidates);
@@ -1121,7 +1135,7 @@ assign_varying_locations(struct gl_context *ctx,
             (ir_variable *) hash_table_find(consumer_inputs, output_var->name);
       }
 
-      if (input_var && input_var->mode != ir_var_shader_in)
+      if (input_var && input_var->data.mode != ir_var_shader_in)
          input_var = NULL;
 
       if (input_var) {
@@ -1143,7 +1157,7 @@ assign_varying_locations(struct gl_context *ctx,
          return false;
       }
 
-      if (matched_candidate->toplevel_var->is_unmatched_generic_inout)
+      if (matched_candidate->toplevel_var->data.is_unmatched_generic_inout)
          matches.record(matched_candidate->toplevel_var, NULL);
    }
 
@@ -1185,8 +1199,8 @@ assign_varying_locations(struct gl_context *ctx,
       foreach_list(node, consumer->ir) {
          ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-         if (var && var->mode == ir_var_shader_in &&
-             var->is_unmatched_generic_inout) {
+         if (var && var->data.mode == ir_var_shader_in &&
+             var->data.is_unmatched_generic_inout) {
             if (prog->Version <= 120) {
                /* On page 25 (page 31 of the PDF) of the GLSL 1.20 spec:
                 *
@@ -1211,7 +1225,7 @@ assign_varying_locations(struct gl_context *ctx,
             /* An 'in' variable is only really a shader input if its
              * value is written by the previous stage.
              */
-            var->mode = ir_var_auto;
+            var->data.mode = ir_var_auto;
          }
       }
    }
@@ -1229,7 +1243,7 @@ check_against_output_limit(struct gl_context *ctx,
    foreach_list(node, producer->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if (var && var->mode == ir_var_shader_out &&
+      if (var && var->data.mode == ir_var_shader_out &&
           is_varying_var(producer->Type, var)) {
          output_vectors += var->type->count_attribute_slots();
       }
@@ -1278,7 +1292,7 @@ check_against_input_limit(struct gl_context *ctx,
    foreach_list(node, consumer->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if (var && var->mode == ir_var_shader_in &&
+      if (var && var->data.mode == ir_var_shader_in &&
           is_varying_var(consumer->Type, var)) {
          input_vectors += var->type->count_attribute_slots();
       }
diff --git a/mesalib/src/glsl/linker.cpp b/mesalib/src/glsl/linker.cpp
index 1366077f7..a6133ea9c 100644
--- a/mesalib/src/glsl/linker.cpp
+++ b/mesalib/src/glsl/linker.cpp
@@ -114,8 +114,8 @@ public:
 	 ir_rvalue *param_rval = (ir_rvalue *)iter.get();
 	 ir_variable *sig_param = (ir_variable *)sig_iter.get();
 
-	 if (sig_param->mode == ir_var_function_out ||
-	     sig_param->mode == ir_var_function_inout) {
+	 if (sig_param->data.mode == ir_var_function_out ||
+	     sig_param->data.mode == ir_var_function_inout) {
 	    ir_variable *var = param_rval->variable_referenced();
 	    if (var && strcmp(name, var->name) == 0) {
 	       found = true;
@@ -198,7 +198,7 @@ public:
 
    virtual ir_visitor_status visit(ir_variable *var)
    {
-      if (!var->type->is_array() || var->mode != ir_var_shader_in)
+      if (!var->type->is_array() || var->data.mode != ir_var_shader_in)
          return visit_continue;
 
       unsigned size = var->type->length;
@@ -217,16 +217,16 @@ public:
        * array using an index too large for its actual size assigned at link
        * time.
        */
-      if (var->max_array_access >= this->num_vertices) {
+      if (var->data.max_array_access >= this->num_vertices) {
          linker_error(this->prog, "geometry shader accesses element %i of "
                       "%s, but only %i input vertices\n",
-                      var->max_array_access, var->name, this->num_vertices);
+                      var->data.max_array_access, var->name, this->num_vertices);
          return visit_continue;
       }
 
       var->type = glsl_type::get_array_instance(var->type->element_type(),
                                                 this->num_vertices);
-      var->max_array_access = this->num_vertices - 1;
+      var->data.max_array_access = this->num_vertices - 1;
 
       return visit_continue;
    }
@@ -379,9 +379,9 @@ link_invalidate_variable_locations(exec_list *ir)
        * shader inputs (via layout(location=...)), and generic fragment shader
        * outputs (also via layout(location=...)).
        */
-      if (!var->explicit_location) {
-         var->location = -1;
-         var->location_frac = 0;
+      if (!var->data.explicit_location) {
+         var->data.location = -1;
+         var->data.location_frac = 0;
       }
 
       /* ir_variable::is_unmatched_generic_inout is used by the linker while
@@ -396,10 +396,10 @@ link_invalidate_variable_locations(exec_list *ir)
        * GL_ARB_separate_shader_objects is supported.  When that extension is
        * implemented, this function will need some modifications.
        */
-      if (!var->explicit_location) {
-         var->is_unmatched_generic_inout = 1;
+      if (!var->data.explicit_location) {
+         var->data.is_unmatched_generic_inout = 1;
       } else {
-         var->is_unmatched_generic_inout = 0;
+         var->data.is_unmatched_generic_inout = 0;
       }
    }
 }
@@ -580,13 +580,13 @@ cross_validate_globals(struct gl_shader_program *prog,
 	 if (var == NULL)
 	    continue;
 
-	 if (uniforms_only && (var->mode != ir_var_uniform))
+	 if (uniforms_only && (var->data.mode != ir_var_uniform))
 	    continue;
 
 	 /* Don't cross validate temporaries that are at global scope.  These
 	  * will eventually get pulled into the shaders 'main'.
 	  */
-	 if (var->mode == ir_var_temporary)
+	 if (var->data.mode == ir_var_temporary)
 	    continue;
 
 	 /* If a global with this name has already been seen, verify that the
@@ -619,17 +619,17 @@ cross_validate_globals(struct gl_shader_program *prog,
 	       }
 	    }
 
-	    if (var->explicit_location) {
-	       if (existing->explicit_location
-		   && (var->location != existing->location)) {
+	    if (var->data.explicit_location) {
+	       if (existing->data.explicit_location
+		   && (var->data.location != existing->data.location)) {
 		     linker_error(prog, "explicit locations for %s "
 				  "`%s' have differing values\n",
 				  mode_string(var), var->name);
 		     return;
 	       }
 
-	       existing->location = var->location;
-	       existing->explicit_location = true;
+	       existing->data.location = var->data.location;
+	       existing->data.explicit_location = true;
 	    }
 
             /* From the GLSL 4.20 specification:
@@ -638,21 +638,21 @@ cross_validate_globals(struct gl_shader_program *prog,
              *  opaque-uniform name.  However, it is not an error to specify a
              *  binding on some but not all declarations for the same name"
              */
-            if (var->explicit_binding) {
-               if (existing->explicit_binding &&
-                   var->binding != existing->binding) {
+            if (var->data.explicit_binding) {
+               if (existing->data.explicit_binding &&
+                   var->data.binding != existing->data.binding) {
                   linker_error(prog, "explicit bindings for %s "
                                "`%s' have differing values\n",
                                mode_string(var), var->name);
                   return;
                }
 
-               existing->binding = var->binding;
-               existing->explicit_binding = true;
+               existing->data.binding = var->data.binding;
+               existing->data.explicit_binding = true;
             }
 
             if (var->type->contains_atomic() &&
-                var->atomic.offset != existing->atomic.offset) {
+                var->data.atomic.offset != existing->data.atomic.offset) {
                linker_error(prog, "offset specifications for %s "
                             "`%s' have differing values\n",
                             mode_string(var), var->name);
@@ -671,9 +671,9 @@ cross_validate_globals(struct gl_shader_program *prog,
 	     *    of qualifiers."
 	     */
 	    if (strcmp(var->name, "gl_FragDepth") == 0) {
-	       bool layout_declared = var->depth_layout != ir_depth_layout_none;
+	       bool layout_declared = var->data.depth_layout != ir_depth_layout_none;
 	       bool layout_differs =
-		  var->depth_layout != existing->depth_layout;
+		  var->data.depth_layout != existing->data.depth_layout;
 
 	       if (layout_declared && layout_differs) {
 		  linker_error(prog,
@@ -682,7 +682,7 @@ cross_validate_globals(struct gl_shader_program *prog,
 			       "the same set of qualifiers.");
 	       }
 
-	       if (var->used && layout_differs) {
+	       if (var->data.used && layout_differs) {
 		  linker_error(prog,
 			       "If gl_FragDepth is redeclared with a layout "
 			       "qualifier in any fragment shader, it must be "
@@ -734,8 +734,8 @@ cross_validate_globals(struct gl_shader_program *prog,
 	       }
 	    }
 
-	    if (var->has_initializer) {
-	       if (existing->has_initializer
+	    if (var->data.has_initializer) {
+	       if (existing->data.has_initializer
 		   && (var->constant_initializer == NULL
 		       || existing->constant_initializer == NULL)) {
 		  linker_error(prog,
@@ -750,21 +750,27 @@ cross_validate_globals(struct gl_shader_program *prog,
 		* otherwise) will propagate the existence to the variable
 		* stored in the symbol table.
 		*/
-	       existing->has_initializer = true;
+	       existing->data.has_initializer = true;
 	    }
 
-	    if (existing->invariant != var->invariant) {
+	    if (existing->data.invariant != var->data.invariant) {
 	       linker_error(prog, "declarations for %s `%s' have "
 			    "mismatching invariant qualifiers\n",
 			    mode_string(var), var->name);
 	       return;
 	    }
-            if (existing->centroid != var->centroid) {
+            if (existing->data.centroid != var->data.centroid) {
                linker_error(prog, "declarations for %s `%s' have "
 			    "mismatching centroid qualifiers\n",
 			    mode_string(var), var->name);
                return;
             }
+            if (existing->data.sample != var->data.sample) {
+               linker_error(prog, "declarations for %s `%s` have "
+                            "mismatching sample qualifiers\n",
+                            mode_string(var), var->name);
+               return;
+            }
 	 } else
 	    variables.add_variable(var);
       }
@@ -884,7 +890,7 @@ remap_variables(ir_instruction *inst, struct gl_shader *target,
 
       virtual ir_visitor_status visit(ir_dereference_variable *ir)
       {
-	 if (ir->var->mode == ir_var_temporary) {
+	 if (ir->var->data.mode == ir_var_temporary) {
 	    ir_variable *var = (ir_variable *) hash_table_find(temps, ir->var);
 
 	    assert(var != NULL);
@@ -958,13 +964,13 @@ move_non_declarations(exec_list *instructions, exec_node *last,
 	 continue;
 
       ir_variable *var = inst->as_variable();
-      if ((var != NULL) && (var->mode != ir_var_temporary))
+      if ((var != NULL) && (var->data.mode != ir_var_temporary))
 	 continue;
 
       assert(inst->as_assignment()
              || inst->as_call()
              || inst->as_if() /* for initializers with the ?: operator */
-	     || ((var != NULL) && (var->mode == ir_var_temporary)));
+	     || ((var != NULL) && (var->data.mode == ir_var_temporary)));
 
       if (make_copies) {
 	 inst = inst->clone(target, NULL);
@@ -1036,7 +1042,7 @@ public:
 
    virtual ir_visitor_status visit(ir_variable *var)
    {
-      fixup_type(&var->type, var->max_array_access);
+      fixup_type(&var->type, var->data.max_array_access);
       if (var->type->is_interface()) {
          if (interface_contains_unsized_arrays(var->type)) {
             const glsl_type *new_type =
@@ -1405,36 +1411,38 @@ link_intrastage_shaders(void *mem_ctx,
 					      insertion_point, true, linked);
    }
 
-   /* Resolve initializers for global variables in the linked shader.
-    */
-   unsigned num_linking_shaders = num_shaders;
-   for (unsigned i = 0; i < num_shaders; i++)
-      num_linking_shaders += shader_list[i]->num_builtins_to_link;
+   /* Check if any shader needs built-in functions. */
+   bool need_builtins = false;
+   for (unsigned i = 0; i < num_shaders; i++) {
+      if (shader_list[i]->uses_builtin_functions) {
+         need_builtins = true;
+         break;
+      }
+   }
 
-   gl_shader **linking_shaders =
-      (gl_shader **) calloc(num_linking_shaders, sizeof(gl_shader *));
+   bool ok;
+   if (need_builtins) {
+      /* Make a temporary array one larger than shader_list, which will hold
+       * the built-in function shader as well.
+       */
+      gl_shader **linking_shaders = (gl_shader **)
+         calloc(num_shaders + 1, sizeof(gl_shader *));
+      memcpy(linking_shaders, shader_list, num_shaders * sizeof(gl_shader *));
+      linking_shaders[num_shaders] = _mesa_glsl_get_builtin_function_shader();
 
-   memcpy(linking_shaders, shader_list,
-	  sizeof(linking_shaders[0]) * num_shaders);
+      ok = link_function_calls(prog, linked, linking_shaders, num_shaders + 1);
 
-   unsigned idx = num_shaders;
-   for (unsigned i = 0; i < num_shaders; i++) {
-      memcpy(&linking_shaders[idx], shader_list[i]->builtins_to_link,
-	     sizeof(linking_shaders[0]) * shader_list[i]->num_builtins_to_link);
-      idx += shader_list[i]->num_builtins_to_link;
+      free(linking_shaders);
+   } else {
+      ok = link_function_calls(prog, linked, shader_list, num_shaders);
    }
 
-   assert(idx == num_linking_shaders);
 
-   if (!link_function_calls(prog, linked, linking_shaders,
-			    num_linking_shaders)) {
+   if (!ok) {
       ctx->Driver.DeleteShader(ctx, linked);
-      free(linking_shaders);
       return NULL;
    }
 
-   free(linking_shaders);
-
    /* At this point linked should contain all of the linked IR, so
     * validate it to make sure nothing went wrong.
     */
@@ -1486,7 +1494,7 @@ update_array_sizes(struct gl_shader_program *prog)
       foreach_list(node, prog->_LinkedShaders[i]->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-	 if ((var == NULL) || (var->mode != ir_var_uniform) ||
+	 if ((var == NULL) || (var->data.mode != ir_var_uniform) ||
 	     !var->type->is_array())
 	    continue;
 
@@ -1501,7 +1509,7 @@ update_array_sizes(struct gl_shader_program *prog)
 	 if (var->is_in_uniform_block() || var->type->contains_atomic())
 	    continue;
 
-	 unsigned int size = var->max_array_access;
+	 unsigned int size = var->data.max_array_access;
 	 for (unsigned j = 0; j < MESA_SHADER_TYPES; j++) {
 	       if (prog->_LinkedShaders[j] == NULL)
 		  continue;
@@ -1512,8 +1520,8 @@ update_array_sizes(struct gl_shader_program *prog)
 		  continue;
 
 	       if (strcmp(var->name, other_var->name) == 0 &&
-		   other_var->max_array_access > size) {
-		  size = other_var->max_array_access;
+		   other_var->data.max_array_access > size) {
+		  size = other_var->data.max_array_access;
 	       }
 	    }
 	 }
@@ -1652,16 +1660,17 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
    foreach_list(node, sh->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if ((var == NULL) || (var->mode != (unsigned) direction))
+      if ((var == NULL) || (var->data.mode != (unsigned) direction))
 	 continue;
 
-      if (var->explicit_location) {
-	 if ((var->location >= (int)(max_index + generic_base))
-	     || (var->location < 0)) {
+      if (var->data.explicit_location) {
+	 if ((var->data.location >= (int)(max_index + generic_base))
+	     || (var->data.location < 0)) {
 	    linker_error(prog,
 			 "invalid explicit location %d specified for `%s'\n",
-			 (var->location < 0)
-			 ? var->location : var->location - generic_base,
+			 (var->data.location < 0)
+			 ? var->data.location
+                         : var->data.location - generic_base,
 			 var->name);
 	    return false;
 	 }
@@ -1670,8 +1679,8 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 
 	 if (prog->AttributeBindings->get(binding, var->name)) {
 	    assert(binding >= VERT_ATTRIB_GENERIC0);
-	    var->location = binding;
-            var->is_unmatched_generic_inout = 0;
+	    var->data.location = binding;
+            var->data.is_unmatched_generic_inout = 0;
 	 }
       } else if (target_index == MESA_SHADER_FRAGMENT) {
 	 unsigned binding;
@@ -1679,11 +1688,11 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 
 	 if (prog->FragDataBindings->get(binding, var->name)) {
 	    assert(binding >= FRAG_RESULT_DATA0);
-	    var->location = binding;
-            var->is_unmatched_generic_inout = 0;
+	    var->data.location = binding;
+            var->data.is_unmatched_generic_inout = 0;
 
 	    if (prog->FragDataIndexBindings->get(index, var->name)) {
-	       var->index = index;
+	       var->data.index = index;
 	    }
 	 }
       }
@@ -1694,8 +1703,8 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
        * add it to the list of variables that need linker-assigned locations.
        */
       const unsigned slots = var->type->count_attribute_slots();
-      if (var->location != -1) {
-	 if (var->location >= generic_base && var->index < 1) {
+      if (var->data.location != -1) {
+	 if (var->data.location >= generic_base && var->data.index < 1) {
 	    /* From page 61 of the OpenGL 4.0 spec:
 	     *
 	     *     "LinkProgram will fail if the attribute bindings assigned
@@ -1729,7 +1738,7 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 	    /* Mask representing the contiguous slots that will be used by
 	     * this attribute.
 	     */
-	    const unsigned attr = var->location - generic_base;
+	    const unsigned attr = var->data.location - generic_base;
 	    const unsigned use_mask = (1 << slots) - 1;
 
 	    /* Generate a link error if the set of bits requested for this
@@ -1795,8 +1804,8 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 	 return false;
       }
 
-      to_assign[i].var->location = generic_base + location;
-      to_assign[i].var->is_unmatched_generic_inout = 0;
+      to_assign[i].var->data.location = generic_base + location;
+      to_assign[i].var->data.is_unmatched_generic_inout = 0;
       used_locations |= (use_mask << location);
    }
 
@@ -1813,15 +1822,15 @@ demote_shader_inputs_and_outputs(gl_shader *sh, enum ir_variable_mode mode)
    foreach_list(node, sh->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if ((var == NULL) || (var->mode != int(mode)))
+      if ((var == NULL) || (var->data.mode != int(mode)))
 	 continue;
 
       /* A shader 'in' or 'out' variable is only really an input or output if
        * its value is used by other shader stages.  This will cause the variable
        * to have a location assigned.
        */
-      if (var->is_unmatched_generic_inout) {
-	 var->mode = ir_var_auto;
+      if (var->data.is_unmatched_generic_inout) {
+	 var->data.mode = ir_var_auto;
       }
    }
 }
@@ -1849,12 +1858,12 @@ store_fragdepth_layout(struct gl_shader_program *prog)
    foreach_list(node, ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if (var == NULL || var->mode != ir_var_shader_out) {
+      if (var == NULL || var->data.mode != ir_var_shader_out) {
          continue;
       }
 
       if (strcmp(var->name, "gl_FragDepth") == 0) {
-         switch (var->depth_layout) {
+         switch (var->data.depth_layout) {
          case ir_depth_layout_none:
             prog->FragDepthLayout = FRAG_DEPTH_LAYOUT_NONE;
             return;
diff --git a/mesalib/src/glsl/loop_analysis.cpp b/mesalib/src/glsl/loop_analysis.cpp
index b08241af5..fd2b6c923 100644
--- a/mesalib/src/glsl/loop_analysis.cpp
+++ b/mesalib/src/glsl/loop_analysis.cpp
@@ -33,6 +33,46 @@ static bool all_expression_operands_are_loop_constant(ir_rvalue *,
 static ir_rvalue *get_basic_induction_increment(ir_assignment *, hash_table *);
 
 
+/**
+ * Record the fact that the given loop variable was referenced inside the loop.
+ *
+ * \arg in_assignee is true if the reference was on the LHS of an assignment.
+ *
+ * \arg in_conditional_code_or_nested_loop is true if the reference occurred
+ * inside an if statement or a nested loop.
+ *
+ * \arg current_assignment is the ir_assignment node that the loop variable is
+ * on the LHS of, if any (ignored if \c in_assignee is false).
+ */
+void
+loop_variable::record_reference(bool in_assignee,
+                                bool in_conditional_code_or_nested_loop,
+                                ir_assignment *current_assignment)
+{
+   if (in_assignee) {
+      assert(current_assignment != NULL);
+
+      if (in_conditional_code_or_nested_loop ||
+          current_assignment->condition != NULL) {
+         this->conditional_or_nested_assignment = true;
+      }
+
+      if (this->first_assignment == NULL) {
+         assert(this->num_assignments == 0);
+
+         this->first_assignment = current_assignment;
+      }
+
+      this->num_assignments++;
+   } else if (this->first_assignment == current_assignment) {
+      /* This catches the case where the variable is used in the RHS of an
+       * assignment where it is also in the LHS.
+       */
+      this->read_before_write = true;
+   }
+}
+
+
 loop_state::loop_state()
 {
    this->ht = hash_table_ctor(0, hash_table_pointer_hash,
@@ -94,7 +134,7 @@ loop_terminator *
 loop_variable_state::insert(ir_if *if_stmt)
 {
    void *mem_ctx = ralloc_parent(this);
-   loop_terminator *t = rzalloc(mem_ctx, loop_terminator);
+   loop_terminator *t = new(mem_ctx) loop_terminator();
 
    t->ir = if_stmt;
    this->terminators.push_tail(t);
@@ -102,6 +142,33 @@ loop_variable_state::insert(ir_if *if_stmt)
    return t;
 }
 
+
+/**
+ * If the given variable already is recorded in the state for this loop,
+ * return the corresponding loop_variable object that records information
+ * about it.
+ *
+ * Otherwise, create a new loop_variable object to record information about
+ * the variable, and set its \c read_before_write field appropriately based on
+ * \c in_assignee.
+ *
+ * \arg in_assignee is true if this variable was encountered on the LHS of an
+ * assignment.
+ */
+loop_variable *
+loop_variable_state::get_or_insert(ir_variable *var, bool in_assignee)
+{
+   loop_variable *lv = this->get(var);
+
+   if (lv == NULL) {
+      lv = this->insert(var);
+      lv->read_before_write = !in_assignee;
+   }
+
+   return lv;
+}
+
+
 namespace {
 
 class loop_analysis : public ir_hierarchical_visitor {
@@ -157,14 +224,14 @@ loop_analysis::visit(ir_loop_jump *ir)
 ir_visitor_status
 loop_analysis::visit_enter(ir_call *ir)
 {
-   /* If we're not somewhere inside a loop, there's nothing to do. */
-   if (this->state.is_empty())
-      return visit_continue;
-
-   loop_variable_state *const ls =
-      (loop_variable_state *) this->state.get_head();
+   /* Mark every loop that we're currently analyzing as containing an ir_call
+    * (even those at outer nesting levels).
+    */
+   foreach_list(node, &this->state) {
+      loop_variable_state *const ls = (loop_variable_state *) node;
+      ls->contains_calls = true;
+   }
 
-   ls->contains_calls = true;
    return visit_continue_with_parent;
 }
 
@@ -177,35 +244,18 @@ loop_analysis::visit(ir_dereference_variable *ir)
    if (this->state.is_empty())
       return visit_continue;
 
-   loop_variable_state *const ls =
-      (loop_variable_state *) this->state.get_head();
-
-   ir_variable *var = ir->variable_referenced();
-   loop_variable *lv = ls->get(var);
-
-   if (lv == NULL) {
-      lv = ls->insert(var);
-      lv->read_before_write = !this->in_assignee;
-   }
-
-   if (this->in_assignee) {
-      assert(this->current_assignment != NULL);
+   bool nested = false;
 
-      lv->conditional_assignment = (this->if_statement_depth > 0)
-	 || (this->current_assignment->condition != NULL);
+   foreach_list(node, &this->state) {
+      loop_variable_state *const ls = (loop_variable_state *) node;
 
-      if (lv->first_assignment == NULL) {
-	 assert(lv->num_assignments == 0);
-
-	 lv->first_assignment = this->current_assignment;
-      }
+      ir_variable *var = ir->variable_referenced();
+      loop_variable *lv = ls->get_or_insert(var, this->in_assignee);
 
-      lv->num_assignments++;
-   } else if (lv->first_assignment == this->current_assignment) {
-      /* This catches the case where the variable is used in the RHS of an
-       * assignment where it is also in the LHS.
-       */
-      lv->read_before_write = true;
+      lv->record_reference(this->in_assignee,
+                           nested || this->if_statement_depth > 0,
+                           this->current_assignment);
+      nested = true;
    }
 
    return visit_continue;
@@ -286,7 +336,7 @@ loop_analysis::visit_leave(ir_loop *ir)
       foreach_list_safe(node, &ls->variables) {
 	 loop_variable *lv = (loop_variable *) node;
 
-	 if (lv->conditional_assignment || (lv->num_assignments > 1))
+	 if (lv->conditional_or_nested_assignment || (lv->num_assignments > 1))
 	    continue;
 
 	 /* Process the RHS of the assignment.  If all of the variables
@@ -326,9 +376,10 @@ loop_analysis::visit_leave(ir_loop *ir)
       assert(lv->num_assignments == 1);
       assert(lv->first_assignment != NULL);
 
-      /* The assignmnet to the variable in the loop must be unconditional.
+      /* The assignment to the variable in the loop must be unconditional and
+       * not inside a nested loop.
        */
-      if (lv->conditional_assignment)
+      if (lv->conditional_or_nested_assignment)
 	 continue;
 
       /* Basic loop induction variables have a single assignment in the loop
@@ -338,8 +389,6 @@ loop_analysis::visit_leave(ir_loop *ir)
       ir_rvalue *const inc =
 	 get_basic_induction_increment(lv->first_assignment, ls->var_hash);
       if (inc != NULL) {
-	 lv->iv_scale = NULL;
-	 lv->biv = lv->var;
 	 lv->increment = inc;
 
 	 lv->remove();
@@ -347,6 +396,75 @@ loop_analysis::visit_leave(ir_loop *ir)
       }
    }
 
+   /* Search the loop terminating conditions for those of the form 'i < c'
+    * where i is a loop induction variable, c is a constant, and < is any
+    * relative operator.  From each of these we can infer an iteration count.
+    * Also figure out which terminator (if any) produces the smallest
+    * iteration count--this is the limiting terminator.
+    */
+   foreach_list(node, &ls->terminators) {
+      loop_terminator *t = (loop_terminator *) node;
+      ir_if *if_stmt = t->ir;
+
+      /* If-statements can be either 'if (expr)' or 'if (deref)'.  We only care
+       * about the former here.
+       */
+      ir_expression *cond = if_stmt->condition->as_expression();
+      if (cond == NULL)
+	 continue;
+
+      switch (cond->operation) {
+      case ir_binop_less:
+      case ir_binop_greater:
+      case ir_binop_lequal:
+      case ir_binop_gequal: {
+	 /* The expressions that we care about will either be of the form
+	  * 'counter < limit' or 'limit < counter'.  Figure out which is
+	  * which.
+	  */
+	 ir_rvalue *counter = cond->operands[0]->as_dereference_variable();
+	 ir_constant *limit = cond->operands[1]->as_constant();
+	 enum ir_expression_operation cmp = cond->operation;
+
+	 if (limit == NULL) {
+	    counter = cond->operands[1]->as_dereference_variable();
+	    limit = cond->operands[0]->as_constant();
+
+	    switch (cmp) {
+	    case ir_binop_less:    cmp = ir_binop_greater; break;
+	    case ir_binop_greater: cmp = ir_binop_less;    break;
+	    case ir_binop_lequal:  cmp = ir_binop_gequal;  break;
+	    case ir_binop_gequal:  cmp = ir_binop_lequal;  break;
+	    default: assert(!"Should not get here.");
+	    }
+	 }
+
+	 if ((counter == NULL) || (limit == NULL))
+	    break;
+
+	 ir_variable *var = counter->variable_referenced();
+
+	 ir_rvalue *init = find_initial_value(ir, var);
+
+         loop_variable *lv = ls->get(var);
+         if (lv != NULL && lv->is_induction_var()) {
+            t->iterations = calculate_iterations(init, limit, lv->increment,
+                                                 cmp);
+
+            if (t->iterations >= 0 &&
+                (ls->limiting_terminator == NULL ||
+                 t->iterations < ls->limiting_terminator->iterations)) {
+               ls->limiting_terminator = t;
+            }
+         }
+         break;
+      }
+
+      default:
+         break;
+      }
+   }
+
    return visit_continue;
 }
 
diff --git a/mesalib/src/glsl/loop_analysis.h b/mesalib/src/glsl/loop_analysis.h
index 769d62661..f841042f0 100644
--- a/mesalib/src/glsl/loop_analysis.h
+++ b/mesalib/src/glsl/loop_analysis.h
@@ -39,16 +39,12 @@ analyze_loop_variables(exec_list *instructions);
 /**
  * Fill in loop control fields
  *
- * Based on analysis of loop variables, this function tries to remove sequences
- * in the loop of the form
+ * Based on analysis of loop variables, this function tries to remove
+ * redundant sequences in the loop of the form
  *
  *  (if (expression bool ...) (break))
  *
- * and fill in the \c ir_loop::from, \c ir_loop::to, and \c ir_loop::counter
- * fields of the \c ir_loop.
- *
- * In this process, some conditional break-statements may be eliminated
- * altogether.  For example, if it is provable that one loop exit condition will
+ * For example, if it is provable that one loop exit condition will
  * always be satisfied before another, the unnecessary exit condition will be
  * removed.
  */
@@ -59,6 +55,13 @@ set_loop_controls(exec_list *instructions, loop_state *ls);
 extern bool
 unroll_loops(exec_list *instructions, loop_state *ls, unsigned max_iterations);
 
+ir_rvalue *
+find_initial_value(ir_loop *loop, ir_variable *var);
+
+int
+calculate_iterations(ir_rvalue *from, ir_rvalue *to, ir_rvalue *increment,
+		     enum ir_expression_operation op);
+
 
 /**
  * Tracking for all variables used in a loop
@@ -67,15 +70,11 @@ class loop_variable_state : public exec_node {
 public:
    class loop_variable *get(const ir_variable *);
    class loop_variable *insert(ir_variable *);
+   class loop_variable *get_or_insert(ir_variable *, bool in_assignee);
    class loop_terminator *insert(ir_if *);
 
 
    /**
-    * Loop whose variable state is being tracked by this structure
-    */
-   ir_loop *loop;
-
-   /**
     * Variables that have not yet been classified
     */
    exec_list variables;
@@ -104,18 +103,17 @@ public:
    exec_list terminators;
 
    /**
-    * Hash table containing all variables accessed in this loop
+    * If any of the terminators in \c terminators leads to termination of the
+    * loop after a constant number of iterations, this is the terminator that
+    * leads to termination after the smallest number of iterations.  Otherwise
+    * NULL.
     */
-   hash_table *var_hash;
+   loop_terminator *limiting_terminator;
 
    /**
-    * Maximum number of loop iterations.
-    *
-    * If this value is negative, then the loop may be infinite.  This actually
-    * means that analysis was unable to determine an upper bound on the number
-    * of loop iterations.
+    * Hash table containing all variables accessed in this loop
     */
-   int max_iterations;
+   hash_table *var_hash;
 
    /**
     * Number of ir_loop_jump instructions that operate on this loop
@@ -129,11 +127,11 @@ public:
 
    loop_variable_state()
    {
-      this->max_iterations = -1;
       this->num_loop_jumps = 0;
       this->contains_calls = false;
       this->var_hash = hash_table_ctor(0, hash_table_pointer_hash,
 				       hash_table_pointer_compare);
+      this->limiting_terminator = NULL;
    }
 
    ~loop_variable_state()
@@ -171,8 +169,11 @@ public:
    /** Are all variables in the RHS of the assignment loop constants? */
    bool rhs_clean;
 
-   /** Is there an assignment to the variable that is conditional? */
-   bool conditional_assignment;
+   /**
+    * Is there an assignment to the variable that is conditional, or inside a
+    * nested loop?
+    */
+   bool conditional_or_nested_assignment;
 
    /** Reference to the first assignment to the variable in the loop body. */
    ir_assignment *first_assignment;
@@ -181,27 +182,30 @@ public:
    unsigned num_assignments;
 
    /**
-    * Increment values for loop induction variables
+    * Increment value for a loop induction variable
     *
-    * Loop induction variables have a single increment of the form
-    * \c b * \c biv + \c c, where \c b and \c c are loop constants and \c i
-    * is a basic loop induction variable.
+    * If this is a loop induction variable, the amount by which the variable
+    * is incremented on each iteration through the loop.
     *
-    * If \c iv_scale is \c NULL, 1 is used.  If \c biv is the same as \c var,
-    * then \c var is a basic loop induction variable.
+    * If this is not a loop induction variable, NULL.
     */
-   /*@{*/
-   ir_rvalue *iv_scale;
-   ir_variable *biv;
    ir_rvalue *increment;
-   /*@}*/
+
+
+   inline bool is_induction_var() const
+   {
+      /* Induction variables always have a non-null increment, and vice
+       * versa.
+       */
+      return this->increment != NULL;
+   }
 
 
    inline bool is_loop_constant() const
    {
       const bool is_const = (this->num_assignments == 0)
 	 || ((this->num_assignments == 1)
-	     && !this->conditional_assignment
+	     && !this->conditional_or_nested_assignment
 	     && !this->read_before_write
 	     && this->rhs_clean);
 
@@ -213,16 +217,35 @@ public:
 
       /* Variables that are marked read-only *MUST* be loop constant.
        */
-      assert(!this->var->read_only || (this->var->read_only && is_const));
+      assert(!this->var->data.read_only
+            || (this->var->data.read_only && is_const));
 
       return is_const;
    }
+
+   void record_reference(bool in_assignee,
+                         bool in_conditional_code_or_nested_loop,
+                         ir_assignment *current_assignment);
 };
 
 
 class loop_terminator : public exec_node {
 public:
+   loop_terminator()
+      : ir(NULL), iterations(-1)
+   {
+   }
+
+   /**
+    * Statement which terminates the loop.
+    */
    ir_if *ir;
+
+   /**
+    * The number of iterations after which the terminator is known to
+    * terminate the loop (if that is a fixed value).  Otherwise -1.
+    */
+   int iterations;
 };
 
 
diff --git a/mesalib/src/glsl/loop_controls.cpp b/mesalib/src/glsl/loop_controls.cpp
index 26481930d..3db06ad18 100644
--- a/mesalib/src/glsl/loop_controls.cpp
+++ b/mesalib/src/glsl/loop_controls.cpp
@@ -183,114 +183,41 @@ loop_control_visitor::visit_leave(ir_loop *ir)
       return visit_continue;
    }
 
-   /* Search the loop terminating conditions for one of the form 'i < c' where
-    * i is a loop induction variable, c is a constant, and < is any relative
-    * operator.
-    */
-   int max_iterations = ls->max_iterations;
-
-   if(ir->from && ir->to && ir->increment)
-      max_iterations = calculate_iterations(ir->from, ir->to, ir->increment, (ir_expression_operation)ir->cmp);
-
-   if(max_iterations < 0)
-      max_iterations = INT_MAX;
+   if (ls->limiting_terminator != NULL) {
+      /* If the limiting terminator has an iteration count of zero, then we've
+       * proven that the loop cannot run, so delete it.
+       */
+      int iterations = ls->limiting_terminator->iterations;
+      if (iterations == 0) {
+         ir->remove();
+         this->progress = true;
+         return visit_continue;
+      }
+   }
 
+   /* Remove the conditional break statements associated with all terminators
+    * that are associated with a fixed iteration count, except for the one
+    * associated with the limiting terminator--that one needs to stay, since
+    * it terminates the loop.  Exception: if the loop still has a normative
+    * bound, then that terminates the loop, so we don't even need the limiting
+    * terminator.
+    */
    foreach_list(node, &ls->terminators) {
       loop_terminator *t = (loop_terminator *) node;
-      ir_if *if_stmt = t->ir;
 
-      /* If-statements can be either 'if (expr)' or 'if (deref)'.  We only care
-       * about the former here.
-       */
-      ir_expression *cond = if_stmt->condition->as_expression();
-      if (cond == NULL)
-	 continue;
-
-      switch (cond->operation) {
-      case ir_binop_less:
-      case ir_binop_greater:
-      case ir_binop_lequal:
-      case ir_binop_gequal: {
-	 /* The expressions that we care about will either be of the form
-	  * 'counter < limit' or 'limit < counter'.  Figure out which is
-	  * which.
-	  */
-	 ir_rvalue *counter = cond->operands[0]->as_dereference_variable();
-	 ir_constant *limit = cond->operands[1]->as_constant();
-	 enum ir_expression_operation cmp = cond->operation;
-
-	 if (limit == NULL) {
-	    counter = cond->operands[1]->as_dereference_variable();
-	    limit = cond->operands[0]->as_constant();
-
-	    switch (cmp) {
-	    case ir_binop_less:    cmp = ir_binop_greater; break;
-	    case ir_binop_greater: cmp = ir_binop_less;    break;
-	    case ir_binop_lequal:  cmp = ir_binop_gequal;  break;
-	    case ir_binop_gequal:  cmp = ir_binop_lequal;  break;
-	    default: assert(!"Should not get here.");
-	    }
-	 }
-
-	 if ((counter == NULL) || (limit == NULL))
-	    break;
-
-	 ir_variable *var = counter->variable_referenced();
-
-	 ir_rvalue *init = find_initial_value(ir, var);
-
-	 foreach_list(iv_node, &ls->induction_variables) {
-	    loop_variable *lv = (loop_variable *) iv_node;
-
-	    if (lv->var == var) {
-	       const int iterations = calculate_iterations(init, limit,
-							   lv->increment,
-							   cmp);
-	       if (iterations >= 0) {
-		  /* If the new iteration count is lower than the previously
-		   * believed iteration count, update the loop control values.
-		   */
-		  if (iterations < max_iterations) {
-		     ir->from = init->clone(ir, NULL);
-		     ir->to = limit->clone(ir, NULL);
-		     ir->increment = lv->increment->clone(ir, NULL);
-		     ir->counter = lv->var;
-		     ir->cmp = cmp;
-
-		     max_iterations = iterations;
-		  }
-
-		  /* Remove the conditional break statement.  The loop
-		   * controls are now set such that the exit condition will be
-		   * satisfied.
-		   */
-		  if_stmt->remove();
-
-		  assert(ls->num_loop_jumps > 0);
-		  ls->num_loop_jumps--;
-
-		  this->progress = true;
-	       }
-
-	       break;
-	    }
-	 }
-	 break;
-      }
+      if (t->iterations < 0)
+         continue;
 
-      default:
-	 break;
+      if (t != ls->limiting_terminator) {
+         t->ir->remove();
+
+         assert(ls->num_loop_jumps > 0);
+         ls->num_loop_jumps--;
+
+         this->progress = true;
       }
    }
 
-   /* If we have proven the one of the loop exit conditions is satisifed before
-    * running the loop once, remove the loop.
-    */
-   if (max_iterations == 0)
-      ir->remove();
-   else
-      ls->max_iterations = max_iterations;
-
    return visit_continue;
 }
 
diff --git a/mesalib/src/glsl/loop_unroll.cpp b/mesalib/src/glsl/loop_unroll.cpp
index ff97766f1..6eced1736 100644
--- a/mesalib/src/glsl/loop_unroll.cpp
+++ b/mesalib/src/glsl/loop_unroll.cpp
@@ -37,6 +37,10 @@ public:
    }
 
    virtual ir_visitor_status visit_leave(ir_loop *ir);
+   void simple_unroll(ir_loop *ir, int iterations);
+   void complex_unroll(ir_loop *ir, int iterations,
+                       bool continue_from_then_branch);
+   void splice_post_if_instructions(ir_if *ir_if, exec_list *splice_dest);
 
    loop_state *state;
 
@@ -86,6 +90,138 @@ public:
 };
 
 
+/**
+ * Unroll a loop which does not contain any jumps.  For example, if the input
+ * is:
+ *
+ *     (loop (...) ...instrs...)
+ *
+ * And the iteration count is 3, the output will be:
+ *
+ *     ...instrs... ...instrs... ...instrs...
+ */
+void
+loop_unroll_visitor::simple_unroll(ir_loop *ir, int iterations)
+{
+   void *const mem_ctx = ralloc_parent(ir);
+
+   for (int i = 0; i < iterations; i++) {
+      exec_list copy_list;
+
+      copy_list.make_empty();
+      clone_ir_list(mem_ctx, &copy_list, &ir->body_instructions);
+
+      ir->insert_before(&copy_list);
+   }
+
+   /* The loop has been replaced by the unrolled copies.  Remove the original
+    * loop from the IR sequence.
+    */
+   ir->remove();
+
+   this->progress = true;
+}
+
+
+/**
+ * Unroll a loop whose last statement is an ir_if.  If \c
+ * continue_from_then_branch is true, the loop is repeated only when the
+ * "then" branch of the if is taken; otherwise it is repeated only when the
+ * "else" branch of the if is taken.
+ *
+ * For example, if the input is:
+ *
+ *     (loop (...)
+ *      ...body...
+ *      (if (cond)
+ *          (...then_instrs...)
+ *        (...else_instrs...)))
+ *
+ * And the iteration count is 3, and \c continue_from_then_branch is true,
+ * then the output will be:
+ *
+ *     ...body...
+ *     (if (cond)
+ *         (...then_instrs...
+ *          ...body...
+ *          (if (cond)
+ *              (...then_instrs...
+ *               ...body...
+ *               (if (cond)
+ *                   (...then_instrs...)
+ *                 (...else_instrs...)))
+ *            (...else_instrs...)))
+ *       (...else_instrs))
+ */
+void
+loop_unroll_visitor::complex_unroll(ir_loop *ir, int iterations,
+                                    bool continue_from_then_branch)
+{
+   void *const mem_ctx = ralloc_parent(ir);
+   ir_instruction *ir_to_replace = ir;
+
+   for (int i = 0; i < iterations; i++) {
+      exec_list copy_list;
+
+      copy_list.make_empty();
+      clone_ir_list(mem_ctx, &copy_list, &ir->body_instructions);
+
+      ir_if *ir_if = ((ir_instruction *) copy_list.get_tail())->as_if();
+      assert(ir_if != NULL);
+
+      ir_to_replace->insert_before(&copy_list);
+      ir_to_replace->remove();
+
+      /* placeholder that will be removed in the next iteration */
+      ir_to_replace =
+         new(mem_ctx) ir_loop_jump(ir_loop_jump::jump_continue);
+
+      exec_list *const list = (continue_from_then_branch)
+         ? &ir_if->then_instructions : &ir_if->else_instructions;
+
+      list->push_tail(ir_to_replace);
+   }
+
+   ir_to_replace->remove();
+
+   this->progress = true;
+}
+
+
+/**
+ * Move all of the instructions which follow \c ir_if to the end of
+ * \c splice_dest.
+ *
+ * For example, in the code snippet:
+ *
+ *     (if (cond)
+ *         (...then_instructions...
+ *          break)
+ *       (...else_instructions...))
+ *     ...post_if_instructions...
+ *
+ * If \c ir_if points to the "if" instruction, and \c splice_dest points to
+ * (...else_instructions...), the code snippet is transformed into:
+ *
+ *     (if (cond)
+ *         (...then_instructions...
+ *          break)
+ *       (...else_instructions...
+ *        ...post_if_instructions...))
+ */
+void
+loop_unroll_visitor::splice_post_if_instructions(ir_if *ir_if,
+                                                 exec_list *splice_dest)
+{
+   while (!ir_if->get_next()->is_tail_sentinel()) {
+      ir_instruction *move_ir = (ir_instruction *) ir_if->get_next();
+
+      move_ir->remove();
+      splice_dest->push_tail(move_ir);
+   }
+}
+
+
 ir_visitor_status
 loop_unroll_visitor::visit_leave(ir_loop *ir)
 {
@@ -100,14 +236,14 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
       return visit_continue;
    }
 
-   iterations = ls->max_iterations;
-
    /* Don't try to unroll loops where the number of iterations is not known
     * at compile-time.
     */
-   if (iterations < 0)
+   if (ls->limiting_terminator == NULL)
       return visit_continue;
 
+   iterations = ls->limiting_terminator->iterations;
+
    /* Don't try to unroll loops that have zillions of iterations either.
     */
    if (iterations > (int) max_iterations)
@@ -120,127 +256,83 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
    if (count.fail || count.nodes * iterations > (int)max_iterations * 5)
       return visit_continue;
 
-   if (ls->num_loop_jumps > 1)
-      return visit_continue;
-   else if (ls->num_loop_jumps) {
-      ir_instruction *last_ir = (ir_instruction *) ir->body_instructions.get_tail();
-      assert(last_ir != NULL);
-
-      if (is_break(last_ir)) {
-         /* If the only loop-jump is a break at the end of the loop, the loop
-          * will execute exactly once.  Remove the break, set the iteration
-          * count, and fall through to the normal unroller.
-          */
-         last_ir->remove();
-         iterations = 1;
-
-         this->progress = true;
-      } else {
-         ir_if *ir_if = NULL;
-         ir_instruction *break_ir = NULL;
-         bool continue_from_then_branch = false;
-
-         foreach_list(node, &ir->body_instructions) {
-            /* recognize loops in the form produced by ir_lower_jumps */
-            ir_instruction *cur_ir = (ir_instruction *) node;
-
-            ir_if = cur_ir->as_if();
-            if (ir_if != NULL) {
-	       /* Determine which if-statement branch, if any, ends with a
-		* break.  The branch that did *not* have the break will get a
-		* temporary continue inserted in each iteration of the loop
-		* unroll.
-		*
-		* Note that since ls->num_loop_jumps is <= 1, it is impossible
-		* for both branches to end with a break.
-		*/
-               ir_instruction *ir_if_last =
-                  (ir_instruction *) ir_if->then_instructions.get_tail();
-
-               if (is_break(ir_if_last)) {
-                  continue_from_then_branch = false;
-                  break_ir = ir_if_last;
-                  break;
-               } else {
-                  ir_if_last =
-		     (ir_instruction *) ir_if->else_instructions.get_tail();
-
-                  if (is_break(ir_if_last)) {
-                     break_ir = ir_if_last;
-                     continue_from_then_branch = true;
-                     break;
-                  }
-               }
-            }
-         }
-
-         if (break_ir == NULL)
-            return visit_continue;
-
-         /* move instructions after then if in the continue branch */
-         while (!ir_if->get_next()->is_tail_sentinel()) {
-            ir_instruction *move_ir = (ir_instruction *) ir_if->get_next();
-
-            move_ir->remove();
-            if (continue_from_then_branch)
-               ir_if->then_instructions.push_tail(move_ir);
-            else
-               ir_if->else_instructions.push_tail(move_ir);
-         }
-
-         /* Remove the break from the if-statement.
-          */
-         break_ir->remove();
-
-         void *const mem_ctx = ralloc_parent(ir);
-         ir_instruction *ir_to_replace = ir;
-
-         for (int i = 0; i < iterations; i++) {
-            exec_list copy_list;
+   /* Note: the limiting terminator contributes 1 to ls->num_loop_jumps.
+    * We'll be removing the limiting terminator before we unroll.
+    */
+   assert(ls->num_loop_jumps > 0);
+   unsigned predicted_num_loop_jumps = ls->num_loop_jumps - 1;
 
-            copy_list.make_empty();
-            clone_ir_list(mem_ctx, &copy_list, &ir->body_instructions);
+   if (predicted_num_loop_jumps > 1)
+      return visit_continue;
 
-            ir_if = ((ir_instruction *) copy_list.get_tail())->as_if();
-            assert(ir_if != NULL);
+   if (predicted_num_loop_jumps == 0) {
+      ls->limiting_terminator->ir->remove();
+      simple_unroll(ir, iterations);
+      return visit_continue;
+   }
 
-            ir_to_replace->insert_before(&copy_list);
-            ir_to_replace->remove();
+   ir_instruction *last_ir = (ir_instruction *) ir->body_instructions.get_tail();
+   assert(last_ir != NULL);
 
-            /* placeholder that will be removed in the next iteration */
-            ir_to_replace =
-	       new(mem_ctx) ir_loop_jump(ir_loop_jump::jump_continue);
+   if (is_break(last_ir)) {
+      /* If the only loop-jump is a break at the end of the loop, the loop
+       * will execute exactly once.  Remove the break and use the simple
+       * unroller with an iteration count of 1.
+       */
+      last_ir->remove();
 
-            exec_list *const list = (continue_from_then_branch)
-               ? &ir_if->then_instructions : &ir_if->else_instructions;
+      ls->limiting_terminator->ir->remove();
+      simple_unroll(ir, 1);
+      return visit_continue;
+   }
 
-            list->push_tail(ir_to_replace);
+   foreach_list(node, &ir->body_instructions) {
+      /* recognize loops in the form produced by ir_lower_jumps */
+      ir_instruction *cur_ir = (ir_instruction *) node;
+
+      /* Skip the limiting terminator, since it will go away when we
+       * unroll.
+       */
+      if (cur_ir == ls->limiting_terminator->ir)
+         continue;
+
+      ir_if *ir_if = cur_ir->as_if();
+      if (ir_if != NULL) {
+         /* Determine which if-statement branch, if any, ends with a
+          * break.  The branch that did *not* have the break will get a
+          * temporary continue inserted in each iteration of the loop
+          * unroll.
+          *
+          * Note that since ls->num_loop_jumps is <= 1, it is impossible
+          * for both branches to end with a break.
+          */
+         ir_instruction *ir_if_last =
+            (ir_instruction *) ir_if->then_instructions.get_tail();
+
+         if (is_break(ir_if_last)) {
+            ls->limiting_terminator->ir->remove();
+            splice_post_if_instructions(ir_if, &ir_if->else_instructions);
+            ir_if_last->remove();
+            complex_unroll(ir, iterations, false);
+            return visit_continue;
+         } else {
+            ir_if_last =
+               (ir_instruction *) ir_if->else_instructions.get_tail();
+
+            if (is_break(ir_if_last)) {
+               ls->limiting_terminator->ir->remove();
+               splice_post_if_instructions(ir_if, &ir_if->then_instructions);
+               ir_if_last->remove();
+               complex_unroll(ir, iterations, true);
+               return visit_continue;
+            }
          }
-
-         ir_to_replace->remove();
-
-         this->progress = true;
-         return visit_continue;
       }
    }
 
-   void *const mem_ctx = ralloc_parent(ir);
-
-   for (int i = 0; i < iterations; i++) {
-      exec_list copy_list;
-
-      copy_list.make_empty();
-      clone_ir_list(mem_ctx, &copy_list, &ir->body_instructions);
-
-      ir->insert_before(&copy_list);
-   }
-
-   /* The loop has been replaced by the unrolled copies.  Remove the original
-    * loop from the IR sequence.
+   /* Did not find the break statement.  It must be in a complex if-nesting,
+    * so don't try to unroll.
     */
-   ir->remove();
-
-   this->progress = true;
    return visit_continue;
 }
 
diff --git a/mesalib/src/glsl/lower_clip_distance.cpp b/mesalib/src/glsl/lower_clip_distance.cpp
index 04fa6d410..bb4f6ab11 100644
--- a/mesalib/src/glsl/lower_clip_distance.cpp
+++ b/mesalib/src/glsl/lower_clip_distance.cpp
@@ -135,13 +135,13 @@ lower_clip_distance_visitor::visit(ir_variable *ir)
                          "gl_ClipDistanceMESA");
       this->new_clip_distance_1d_var->type
          = glsl_type::get_array_instance(glsl_type::vec4_type, new_size);
-      this->new_clip_distance_1d_var->max_array_access
-         = ir->max_array_access / 4;
+      this->new_clip_distance_1d_var->data.max_array_access
+         = ir->data.max_array_access / 4;
 
       ir->replace_with(this->new_clip_distance_1d_var);
    } else {
       /* 2D gl_ClipDistance (used for geometry input). */
-      assert(ir->mode == ir_var_shader_in &&
+      assert(ir->data.mode == ir_var_shader_in &&
              this->shader_type == GL_GEOMETRY_SHADER_ARB);
       if (this->old_clip_distance_2d_var)
          return visit_continue;
@@ -161,8 +161,8 @@ lower_clip_distance_visitor::visit(ir_variable *ir)
          glsl_type::get_array_instance(glsl_type::vec4_type,
             new_size),
          ir->type->array_size());
-      this->new_clip_distance_2d_var->max_array_access
-         = ir->max_array_access / 4;
+      this->new_clip_distance_2d_var->data.max_array_access
+         = ir->data.max_array_access / 4;
 
       ir->replace_with(this->new_clip_distance_2d_var);
    }
@@ -500,8 +500,8 @@ lower_clip_distance_visitor::visit_leave(ir_call *ir)
          this->base_ir->insert_before(temp_clip_distance);
          actual_param->replace_with(
             new(ctx) ir_dereference_variable(temp_clip_distance));
-         if (formal_param->mode == ir_var_function_in
-             || formal_param->mode == ir_var_function_inout) {
+         if (formal_param->data.mode == ir_var_function_in
+             || formal_param->data.mode == ir_var_function_inout) {
             /* Copy from gl_ClipDistance to the temporary before the call.
              * Since we are going to insert this copy before the current
              * instruction, we need to visit it afterwards to make sure it
@@ -513,8 +513,8 @@ lower_clip_distance_visitor::visit_leave(ir_call *ir)
             this->base_ir->insert_before(new_assignment);
             this->visit_new_assignment(new_assignment);
          }
-         if (formal_param->mode == ir_var_function_out
-             || formal_param->mode == ir_var_function_inout) {
+         if (formal_param->data.mode == ir_var_function_out
+             || formal_param->data.mode == ir_var_function_inout) {
             /* Copy from the temporary to gl_ClipDistance after the call.
              * Since visit_list_elements() has already decided which
              * instruction it's going to visit next, we need to visit
diff --git a/mesalib/src/glsl/lower_instructions.cpp b/mesalib/src/glsl/lower_instructions.cpp
index d01879cbd..8f8d448ea 100644
--- a/mesalib/src/glsl/lower_instructions.cpp
+++ b/mesalib/src/glsl/lower_instructions.cpp
@@ -383,7 +383,6 @@ lower_instructions_visitor::ldexp_to_arith(ir_expression *ir)
 
    /* Constants */
    ir_constant *zeroi = ir_constant::zero(ir, ivec);
-   ir_constant *zerof = ir_constant::zero(ir, ir->type);
 
    ir_constant *sign_mantissa_mask = new(ir) ir_constant(0x807fffffu, vec_elem);
    ir_constant *sign_mask = new(ir) ir_constant(0x80000000u, vec_elem);
@@ -429,8 +428,7 @@ lower_instructions_visitor::ldexp_to_arith(ir_expression *ir)
     */
    i.insert_before(zero_sign_x);
    i.insert_before(assign(zero_sign_x,
-                          bitcast_u2f(bit_or(bit_and(bitcast_f2u(x), sign_mask),
-                                             bitcast_f2u(zerof)))));
+                          bitcast_u2f(bit_and(bitcast_f2u(x), sign_mask))));
 
    i.insert_before(is_not_zero_or_underflow);
    i.insert_before(assign(is_not_zero_or_underflow,
diff --git a/mesalib/src/glsl/lower_named_interface_blocks.cpp b/mesalib/src/glsl/lower_named_interface_blocks.cpp
index d59d11150..09d867ea3 100644
--- a/mesalib/src/glsl/lower_named_interface_blocks.cpp
+++ b/mesalib/src/glsl/lower_named_interface_blocks.cpp
@@ -108,7 +108,7 @@ flatten_named_interface_blocks_declarations::run(exec_list *instructions)
        * but, this will require changes to the other uniform block
        * support code.
        */
-      if (var->mode == ir_var_uniform)
+      if (var->data.mode == ir_var_uniform)
          continue;
 
       const glsl_type * iface_t = var->type;
@@ -139,8 +139,8 @@ flatten_named_interface_blocks_declarations::run(exec_list *instructions)
                new_var =
                   new(mem_ctx) ir_variable(iface_t->fields.structure[i].type,
                                            var_name,
-                                           (ir_variable_mode) var->mode);
-               new_var->from_named_ifc_block_nonarray = 1;
+                                           (ir_variable_mode) var->data.mode);
+               new_var->data.from_named_ifc_block_nonarray = 1;
             } else {
                const glsl_type *new_array_type =
                   glsl_type::get_array_instance(
@@ -149,14 +149,15 @@ flatten_named_interface_blocks_declarations::run(exec_list *instructions)
                new_var =
                   new(mem_ctx) ir_variable(new_array_type,
                                            var_name,
-                                           (ir_variable_mode) var->mode);
-               new_var->from_named_ifc_block_array = 1;
+                                           (ir_variable_mode) var->data.mode);
+               new_var->data.from_named_ifc_block_array = 1;
             }
-            new_var->location = iface_t->fields.structure[i].location;
-            new_var->explicit_location = (new_var->location >= 0);
-            new_var->interpolation =
+            new_var->data.location = iface_t->fields.structure[i].location;
+            new_var->data.explicit_location = (new_var->data.location >= 0);
+            new_var->data.interpolation =
                iface_t->fields.structure[i].interpolation;
-            new_var->centroid = iface_t->fields.structure[i].centroid;
+            new_var->data.centroid = iface_t->fields.structure[i].centroid;
+            new_var->data.sample = iface_t->fields.structure[i].sample;
 
             new_var->init_interface_type(iface_t);
             hash_table_insert(interface_namespace, new_var,
@@ -211,7 +212,7 @@ flatten_named_interface_blocks_declarations::handle_rvalue(ir_rvalue **rvalue)
     * but, this will require changes to the other uniform block
     * support code.
     */
-   if (var->mode == ir_var_uniform)
+   if (var->data.mode == ir_var_uniform)
       return;
 
    if (var->get_interface_type() != NULL) {
diff --git a/mesalib/src/glsl/lower_output_reads.cpp b/mesalib/src/glsl/lower_output_reads.cpp
index 128b0b8cd..afe17766b 100644
--- a/mesalib/src/glsl/lower_output_reads.cpp
+++ b/mesalib/src/glsl/lower_output_reads.cpp
@@ -91,7 +91,7 @@ output_read_remover::~output_read_remover()
 ir_visitor_status
 output_read_remover::visit(ir_dereference_variable *ir)
 {
-   if (ir->var->mode != ir_var_shader_out)
+   if (ir->var->data.mode != ir_var_shader_out)
       return visit_continue;
 
    ir_variable *temp = (ir_variable *) hash_table_find(replacements, ir->var);
diff --git a/mesalib/src/glsl/lower_packed_varyings.cpp b/mesalib/src/glsl/lower_packed_varyings.cpp
index 61ee692f6..9edef5d04 100644
--- a/mesalib/src/glsl/lower_packed_varyings.cpp
+++ b/mesalib/src/glsl/lower_packed_varyings.cpp
@@ -258,8 +258,8 @@ lower_packed_varyings_visitor::run(exec_list *instructions)
       if (var == NULL)
          continue;
 
-      if (var->mode != this->mode ||
-          var->location < (int) this->location_base ||
+      if (var->data.mode != this->mode ||
+          var->data.location < (int) this->location_base ||
           !this->needs_lowering(var))
          continue;
 
@@ -268,18 +268,18 @@ lower_packed_varyings_visitor::run(exec_list *instructions)
        * safe, caller should ensure that integral varyings always use flat
        * interpolation, even when this is not required by GLSL.
        */
-      assert(var->interpolation == INTERP_QUALIFIER_FLAT ||
+      assert(var->data.interpolation == INTERP_QUALIFIER_FLAT ||
              !var->type->contains_integer());
 
       /* Change the old varying into an ordinary global. */
-      var->mode = ir_var_auto;
+      var->data.mode = ir_var_auto;
 
       /* Create a reference to the old varying. */
       ir_dereference_variable *deref
          = new(this->mem_ctx) ir_dereference_variable(var);
 
       /* Recursively pack or unpack it. */
-      this->lower_rvalue(deref, var->location * 4 + var->location_frac, var,
+      this->lower_rvalue(deref, var->data.location * 4 + var->data.location_frac, var,
                          var->name, this->gs_input_vertices != 0, 0);
    }
 }
@@ -547,7 +547,7 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
    if (this->packed_varyings[slot] == NULL) {
       char *packed_name = ralloc_asprintf(this->mem_ctx, "packed:%s", name);
       const glsl_type *packed_type;
-      if (unpacked_var->interpolation == INTERP_QUALIFIER_FLAT)
+      if (unpacked_var->data.interpolation == INTERP_QUALIFIER_FLAT)
          packed_type = glsl_type::ivec4_type;
       else
          packed_type = glsl_type::vec4_type;
@@ -562,11 +562,12 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
          /* Prevent update_array_sizes() from messing with the size of the
           * array.
           */
-         packed_var->max_array_access = this->gs_input_vertices - 1;
+         packed_var->data.max_array_access = this->gs_input_vertices - 1;
       }
-      packed_var->centroid = unpacked_var->centroid;
-      packed_var->interpolation = unpacked_var->interpolation;
-      packed_var->location = location;
+      packed_var->data.centroid = unpacked_var->data.centroid;
+      packed_var->data.sample = unpacked_var->data.sample;
+      packed_var->data.interpolation = unpacked_var->data.interpolation;
+      packed_var->data.location = location;
       unpacked_var->insert_before(packed_var);
       this->packed_varyings[slot] = packed_var;
    } else {
diff --git a/mesalib/src/glsl/lower_ubo_reference.cpp b/mesalib/src/glsl/lower_ubo_reference.cpp
index 16b6801b8..c73b148b0 100644
--- a/mesalib/src/glsl/lower_ubo_reference.cpp
+++ b/mesalib/src/glsl/lower_ubo_reference.cpp
@@ -143,7 +143,7 @@ lower_ubo_reference_visitor::handle_rvalue(ir_rvalue **rvalue)
          struct gl_uniform_block *block = &shader->UniformBlocks[i];
 
          this->ubo_var = var->is_interface_instance()
-            ? &block->Uniforms[0] : &block->Uniforms[var->location];
+            ? &block->Uniforms[0] : &block->Uniforms[var->data.location];
 
          break;
       }
diff --git a/mesalib/src/glsl/lower_variable_index_to_cond_assign.cpp b/mesalib/src/glsl/lower_variable_index_to_cond_assign.cpp
index 699fb3903..7c5d80f43 100644
--- a/mesalib/src/glsl/lower_variable_index_to_cond_assign.cpp
+++ b/mesalib/src/glsl/lower_variable_index_to_cond_assign.cpp
@@ -365,7 +365,7 @@ public:
       if (var == NULL)
 	 return this->lower_temps;
 
-      switch (var->mode) {
+      switch (var->data.mode) {
       case ir_var_auto:
       case ir_var_temporary:
 	 return this->lower_temps;
diff --git a/mesalib/src/glsl/opt_array_splitting.cpp b/mesalib/src/glsl/opt_array_splitting.cpp
index c7c5f6712..e946e0ae5 100644
--- a/mesalib/src/glsl/opt_array_splitting.cpp
+++ b/mesalib/src/glsl/opt_array_splitting.cpp
@@ -122,8 +122,8 @@ ir_array_reference_visitor::get_variable_entry(ir_variable *var)
 {
    assert(var);
 
-   if (var->mode != ir_var_auto &&
-       var->mode != ir_var_temporary)
+   if (var->data.mode != ir_var_auto &&
+       var->data.mode != ir_var_temporary)
       return NULL;
 
    if (!(var->type->is_array() || var->type->is_matrix()))
diff --git a/mesalib/src/glsl/opt_constant_folding.cpp b/mesalib/src/glsl/opt_constant_folding.cpp
index 072fefe9a..08a47b96b 100644
--- a/mesalib/src/glsl/opt_constant_folding.cpp
+++ b/mesalib/src/glsl/opt_constant_folding.cpp
@@ -127,8 +127,8 @@ ir_constant_folding_visitor::visit_enter(ir_call *ir)
       ir_rvalue *param_rval = (ir_rvalue *)iter.get();
       ir_variable *sig_param = (ir_variable *)sig_iter.get();
 
-      if (sig_param->mode == ir_var_function_in
-          || sig_param->mode == ir_var_const_in) {
+      if (sig_param->data.mode == ir_var_function_in
+          || sig_param->data.mode == ir_var_const_in) {
 	 ir_rvalue *new_param = param_rval;
 
 	 handle_rvalue(&new_param);
diff --git a/mesalib/src/glsl/opt_constant_propagation.cpp b/mesalib/src/glsl/opt_constant_propagation.cpp
index 2f65937fe..f1a6fbdaa 100644
--- a/mesalib/src/glsl/opt_constant_propagation.cpp
+++ b/mesalib/src/glsl/opt_constant_propagation.cpp
@@ -285,8 +285,8 @@ ir_constant_propagation_visitor::visit_enter(ir_call *ir)
    foreach_iter(exec_list_iterator, iter, ir->actual_parameters) {
       ir_variable *sig_param = (ir_variable *)sig_param_iter.get();
       ir_rvalue *param = (ir_rvalue *)iter.get();
-      if (sig_param->mode != ir_var_function_out
-          && sig_param->mode != ir_var_function_inout) {
+      if (sig_param->data.mode != ir_var_function_out
+          && sig_param->data.mode != ir_var_function_inout) {
 	 ir_rvalue *new_param = param;
 	 handle_rvalue(&new_param);
          if (new_param != param)
diff --git a/mesalib/src/glsl/opt_constant_variable.cpp b/mesalib/src/glsl/opt_constant_variable.cpp
index cbe6450c6..a026c51c3 100644
--- a/mesalib/src/glsl/opt_constant_variable.cpp
+++ b/mesalib/src/glsl/opt_constant_variable.cpp
@@ -137,8 +137,8 @@ ir_constant_variable_visitor::visit_enter(ir_call *ir)
       ir_rvalue *param_rval = (ir_rvalue *)iter.get();
       ir_variable *param = (ir_variable *)sig_iter.get();
 
-      if (param->mode == ir_var_function_out ||
-	  param->mode == ir_var_function_inout) {
+      if (param->data.mode == ir_var_function_out ||
+	  param->data.mode == ir_var_function_inout) {
 	 ir_variable *var = param_rval->variable_referenced();
 	 struct assignment_entry *entry;
 
diff --git a/mesalib/src/glsl/opt_copy_propagation.cpp b/mesalib/src/glsl/opt_copy_propagation.cpp
index 7282b611e..db5dfc153 100644
--- a/mesalib/src/glsl/opt_copy_propagation.cpp
+++ b/mesalib/src/glsl/opt_copy_propagation.cpp
@@ -189,8 +189,8 @@ ir_copy_propagation_visitor::visit_enter(ir_call *ir)
    foreach_iter(exec_list_iterator, iter, ir->actual_parameters) {
       ir_variable *sig_param = (ir_variable *)sig_param_iter.get();
       ir_instruction *ir = (ir_instruction *)iter.get();
-      if (sig_param->mode != ir_var_function_out
-          && sig_param->mode != ir_var_function_inout) {
+      if (sig_param->data.mode != ir_var_function_out
+          && sig_param->data.mode != ir_var_function_inout) {
          ir->accept(this);
       }
       sig_param_iter.next();
diff --git a/mesalib/src/glsl/opt_copy_propagation_elements.cpp b/mesalib/src/glsl/opt_copy_propagation_elements.cpp
index 6a19da40d..ba8a0f532 100644
--- a/mesalib/src/glsl/opt_copy_propagation_elements.cpp
+++ b/mesalib/src/glsl/opt_copy_propagation_elements.cpp
@@ -297,8 +297,8 @@ ir_copy_propagation_elements_visitor::visit_enter(ir_call *ir)
    foreach_iter(exec_list_iterator, iter, ir->actual_parameters) {
       ir_variable *sig_param = (ir_variable *)sig_param_iter.get();
       ir_instruction *ir = (ir_instruction *)iter.get();
-      if (sig_param->mode != ir_var_function_out
-          && sig_param->mode != ir_var_function_inout) {
+      if (sig_param->data.mode != ir_var_function_out
+          && sig_param->data.mode != ir_var_function_inout) {
          ir->accept(this);
       }
       sig_param_iter.next();
diff --git a/mesalib/src/glsl/opt_cse.cpp b/mesalib/src/glsl/opt_cse.cpp
index 8f73940d8..a2b63ee99 100644
--- a/mesalib/src/glsl/opt_cse.cpp
+++ b/mesalib/src/glsl/opt_cse.cpp
@@ -193,7 +193,7 @@ is_cse_candidate_visitor::visit(ir_dereference_variable *ir)
    /* Currently, since we don't handle kills of the ae based on variables
     * getting assigned, we can only handle constant variables.
     */
-   if (ir->var->read_only) {
+   if (ir->var->data.read_only) {
       return visit_continue;
    } else {
       ok = false;
diff --git a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
index b336bc0a8..a939a2b64 100644
--- a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
+++ b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
@@ -85,10 +85,10 @@ public:
    {
       ir_variable *var = ir->variable_referenced();
 
-      if (!var || var->mode != this->mode)
+      if (!var || var->data.mode != this->mode)
          return visit_continue;
 
-      if (this->find_frag_outputs && var->location == FRAG_RESULT_DATA0) {
+      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0) {
          this->fragdata_array = var;
 
          ir_constant *index = ir->array_index->as_constant();
@@ -105,7 +105,7 @@ public:
          return visit_continue_with_parent;
       }
 
-      if (!this->find_frag_outputs && var->location == VARYING_SLOT_TEX0) {
+      if (!this->find_frag_outputs && var->data.location == VARYING_SLOT_TEX0) {
          this->texcoord_array = var;
 
          ir_constant *index = ir->array_index->as_constant();
@@ -130,17 +130,17 @@ public:
    {
       ir_variable *var = ir->variable_referenced();
 
-      if (var->mode != this->mode || !var->type->is_array())
+      if (var->data.mode != this->mode || !var->type->is_array())
          return visit_continue;
 
-      if (this->find_frag_outputs && var->location == FRAG_RESULT_DATA0) {
+      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0) {
          /* This is a whole array dereference. */
          this->fragdata_usage |= (1 << var->type->array_size()) - 1;
          this->lower_fragdata_array = false;
          return visit_continue;
       }
 
-      if (!this->find_frag_outputs && var->location == VARYING_SLOT_TEX0) {
+      if (!this->find_frag_outputs && var->data.location == VARYING_SLOT_TEX0) {
          /* This is a whole array dereference like "gl_TexCoord = x;",
           * there's probably no point in lowering that.
           */
@@ -152,7 +152,7 @@ public:
 
    virtual ir_visitor_status visit(ir_variable *var)
    {
-      if (var->mode != this->mode)
+      if (var->data.mode != this->mode)
          return visit_continue;
 
       /* Nothing to do here for fragment outputs. */
@@ -160,7 +160,7 @@ public:
          return visit_continue;
 
       /* Handle colors and fog. */
-      switch (var->location) {
+      switch (var->data.location) {
       case VARYING_SLOT_COL0:
          this->color[0] = var;
          this->color_usage |= 1;
@@ -358,9 +358,9 @@ public:
                new_var[i] =
                   new(ctx) ir_variable(glsl_type::vec4_type, name,
                                        this->info->mode);
-               new_var[i]->location = start_location + i;
-               new_var[i]->explicit_location = true;
-               new_var[i]->explicit_index = 0;
+               new_var[i]->data.location = start_location + i;
+               new_var[i]->data.explicit_location = true;
+               new_var[i]->data.explicit_index = 0;
             }
 
             ir->head->insert_before(new_var[i]);
diff --git a/mesalib/src/glsl/opt_dead_code.cpp b/mesalib/src/glsl/opt_dead_code.cpp
index b65e5c2ce..a8d8b4a39 100644
--- a/mesalib/src/glsl/opt_dead_code.cpp
+++ b/mesalib/src/glsl/opt_dead_code.cpp
@@ -79,9 +79,9 @@ do_dead_code(exec_list *instructions, bool uniform_locations_assigned)
 	 /* Remove a single dead assignment to the variable we found.
 	  * Don't do so if it's a shader or function output, though.
 	  */
-	 if (entry->var->mode != ir_var_function_out &&
-	     entry->var->mode != ir_var_function_inout &&
-             entry->var->mode != ir_var_shader_out) {
+	 if (entry->var->data.mode != ir_var_function_out &&
+	     entry->var->data.mode != ir_var_function_inout &&
+             entry->var->data.mode != ir_var_shader_out) {
 	    entry->assign->remove();
 	    progress = true;
 
@@ -99,7 +99,7 @@ do_dead_code(exec_list *instructions, bool uniform_locations_assigned)
 	  * stage.  Also, once uniform locations have been assigned, the
 	  * declaration cannot be deleted.
 	  */
-	 if (entry->var->mode == ir_var_uniform &&
+	 if (entry->var->data.mode == ir_var_uniform &&
 	     (uniform_locations_assigned ||
 	      entry->var->constant_value))
 	    continue;
diff --git a/mesalib/src/glsl/opt_dead_code_local.cpp b/mesalib/src/glsl/opt_dead_code_local.cpp
index 42a30b3d8..43a01662a 100644
--- a/mesalib/src/glsl/opt_dead_code_local.cpp
+++ b/mesalib/src/glsl/opt_dead_code_local.cpp
@@ -121,7 +121,7 @@ public:
        */
       foreach_iter(exec_list_iterator, iter, *this->assignments) {
          assignment_entry *entry = (assignment_entry *)iter.get();
-         if (entry->lhs->mode == ir_var_shader_out) {
+         if (entry->lhs->data.mode == ir_var_shader_out) {
             if (debug)
                printf("kill %s\n", entry->lhs->name);
             entry->remove();
diff --git a/mesalib/src/glsl/opt_flip_matrices.cpp b/mesalib/src/glsl/opt_flip_matrices.cpp
index 2107b1d47..9044fd680 100644
--- a/mesalib/src/glsl/opt_flip_matrices.cpp
+++ b/mesalib/src/glsl/opt_flip_matrices.cpp
@@ -104,8 +104,8 @@ matrix_flipper::visit_enter(ir_expression *ir)
 
       var_ref->var = texmat_transpose;
 
-      texmat_transpose->max_array_access =
-         MAX2(texmat_transpose->max_array_access, mat_var->max_array_access);
+      texmat_transpose->data.max_array_access =
+         MAX2(texmat_transpose->data.max_array_access, mat_var->data.max_array_access);
 
       progress = true;
    }
diff --git a/mesalib/src/glsl/opt_function_inlining.cpp b/mesalib/src/glsl/opt_function_inlining.cpp
index f8033a095..c8f42a424 100644
--- a/mesalib/src/glsl/opt_function_inlining.cpp
+++ b/mesalib/src/glsl/opt_function_inlining.cpp
@@ -132,21 +132,21 @@ ir_call::generate_inline(ir_instruction *next_ir)
 	 parameters[i] = NULL;
       } else {
 	 parameters[i] = sig_param->clone(ctx, ht);
-	 parameters[i]->mode = ir_var_auto;
+	 parameters[i]->data.mode = ir_var_auto;
 
 	 /* Remove the read-only decoration becuase we're going to write
 	  * directly to this variable.  If the cloned variable is left
 	  * read-only and the inlined function is inside a loop, the loop
 	  * analysis code will get confused.
 	  */
-	 parameters[i]->read_only = false;
+	 parameters[i]->data.read_only = false;
 	 next_ir->insert_before(parameters[i]);
       }
 
       /* Move the actual param into our param variable if it's an 'in' type. */
-      if (parameters[i] && (sig_param->mode == ir_var_function_in ||
-			    sig_param->mode == ir_var_const_in ||
-			    sig_param->mode == ir_var_function_inout)) {
+      if (parameters[i] && (sig_param->data.mode == ir_var_function_in ||
+			    sig_param->data.mode == ir_var_const_in ||
+			    sig_param->data.mode == ir_var_function_inout)) {
 	 ir_assignment *assign;
 
 	 assign = new(ctx) ir_assignment(new(ctx) ir_dereference_variable(parameters[i]),
@@ -202,8 +202,8 @@ ir_call::generate_inline(ir_instruction *next_ir)
       const ir_variable *const sig_param = (ir_variable *) sig_param_iter.get();
 
       /* Move our param variable into the actual param if it's an 'out' type. */
-      if (parameters[i] && (sig_param->mode == ir_var_function_out ||
-			    sig_param->mode == ir_var_function_inout)) {
+      if (parameters[i] && (sig_param->data.mode == ir_var_function_out ||
+			    sig_param->data.mode == ir_var_function_inout)) {
 	 ir_assignment *assign;
 
 	 assign = new(ctx) ir_assignment(param->clone(ctx, NULL)->as_rvalue(),
diff --git a/mesalib/src/glsl/opt_structure_splitting.cpp b/mesalib/src/glsl/opt_structure_splitting.cpp
index 9f4b3dd8f..414f0000d 100644
--- a/mesalib/src/glsl/opt_structure_splitting.cpp
+++ b/mesalib/src/glsl/opt_structure_splitting.cpp
@@ -103,8 +103,8 @@ ir_structure_reference_visitor::get_variable_entry(ir_variable *var)
 {
    assert(var);
 
-   if (!var->type->is_record() || var->mode == ir_var_uniform
-       || var->mode == ir_var_shader_in || var->mode == ir_var_shader_out)
+   if (!var->type->is_record() || var->data.mode == ir_var_uniform
+       || var->data.mode == ir_var_shader_in || var->data.mode == ir_var_shader_out)
       return NULL;
 
    foreach_iter(exec_list_iterator, iter, this->variable_list) {
diff --git a/mesalib/src/glsl/opt_tree_grafting.cpp b/mesalib/src/glsl/opt_tree_grafting.cpp
index 46c06e6c4..6d75a1573 100644
--- a/mesalib/src/glsl/opt_tree_grafting.cpp
+++ b/mesalib/src/glsl/opt_tree_grafting.cpp
@@ -211,8 +211,8 @@ ir_tree_grafting_visitor::visit_enter(ir_call *ir)
       ir_rvalue *ir = (ir_rvalue *)iter.get();
       ir_rvalue *new_ir = ir;
 
-      if (sig_param->mode != ir_var_function_in
-          && sig_param->mode != ir_var_const_in) {
+      if (sig_param->data.mode != ir_var_function_in
+          && sig_param->data.mode != ir_var_const_in) {
 	 if (check_graft(ir, sig_param) == visit_stop)
 	    return visit_stop;
 	 continue;
@@ -361,9 +361,9 @@ tree_grafting_basic_block(ir_instruction *bb_first,
       if (!lhs_var)
 	 continue;
 
-      if (lhs_var->mode == ir_var_function_out ||
-	  lhs_var->mode == ir_var_function_inout ||
-          lhs_var->mode == ir_var_shader_out)
+      if (lhs_var->data.mode == ir_var_function_out ||
+	  lhs_var->data.mode == ir_var_function_inout ||
+          lhs_var->data.mode == ir_var_shader_out)
 	 continue;
 
       ir_variable_refcount_entry *entry = info->refs->get_variable_entry(lhs_var);
diff --git a/mesalib/src/glsl/standalone_scaffolding.h b/mesalib/src/glsl/standalone_scaffolding.h
index 7afb1c313..9f4818a8a 100644
--- a/mesalib/src/glsl/standalone_scaffolding.h
+++ b/mesalib/src/glsl/standalone_scaffolding.h
@@ -60,7 +60,7 @@ _mesa_shader_type_to_index(GLenum v)
       return MESA_SHADER_GEOMETRY;
    default:
       assert(!"bad value in _mesa_shader_type_to_index()");
-      return MESA_SHADER_TYPES;
+      return MESA_SHADER_VERTEX;
    }
 }
 
diff --git a/mesalib/src/mapi/glapi/gen/ARB_clear_buffer_object.xml b/mesalib/src/mapi/glapi/gen/ARB_clear_buffer_object.xml
new file mode 100644
index 000000000..cb97a0185
--- /dev/null
+++ b/mesalib/src/mapi/glapi/gen/ARB_clear_buffer_object.xml
@@ -0,0 +1,50 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<!-- Note: no GLX protocol info yet. -->
+
+
+<OpenGLAPI>
+
+<category name="GL_ARB_clear_buffer_object" number="121">
+
+    <function name ="ClearBufferData" offset="assign">
+        <param name="target" type="GLenum"/>
+        <param name="internalformat" type="GLenum"/>
+        <param name="format" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="data" type="const GLvoid *"/>
+    </function>
+
+    <function name ="ClearBufferSubData" offset="assign">
+        <param name="target" type="GLenum"/>
+        <param name="internalformat" type="GLenum"/>
+        <param name="offset" type="GLintptr"/>
+        <param name="size" type="GLsizeiptr"/>
+        <param name="format" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="data" type="const GLvoid *"/>
+    </function>
+
+<!--    <function name="ClearNamedBufferDataEXT" offset="assign">
+        <param name="buffer" type="GLuint"/>
+        <param name="internalformat" type="GLenum"/>
+        <param name="format" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="data" type="const GLvoid *"/>
+    </function>
+
+
+    <function name="ClearNamedBufferSubDataEXT" offset="assign">
+        <param name="buffer" type="GLuint"/>
+        <param name="internalformat" type="GLenum"/>
+        <param name="offset" type="GLintptr"/>
+        <param name="size" type="GLsizeiptr"/>
+        <param name="format" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="data" type="const GLvoid *"/>
+    </function> -->
+
+</category>
+
+</OpenGLAPI>
diff --git a/mesalib/src/mapi/glapi/gen/ARB_texture_view.xml b/mesalib/src/mapi/glapi/gen/ARB_texture_view.xml
new file mode 100644
index 000000000..3e6b8c904
--- /dev/null
+++ b/mesalib/src/mapi/glapi/gen/ARB_texture_view.xml
@@ -0,0 +1,23 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<!-- Note: no GLX protocol info yet. -->
+
+<OpenGLAPI>
+
+<category name="GL_ARB_texture_view" number="124">
+
+   <function name="TextureView" offset="assign">
+      <param name="texture" type="GLuint"/>
+      <param name="target" type="GLenum"/>
+      <param name="origtexture" type="GLuint"/>
+      <param name="internalformat" type="GLenum"/>
+      <param name="minlevel" type="GLuint"/>
+      <param name="numlevels" type="GLuint"/>
+      <param name="minlayer" type="GLuint"/>
+      <param name="numlayers" type="GLuint"/>
+   </function>
+
+</category>
+
+</OpenGLAPI>
diff --git a/mesalib/src/mapi/glapi/gen/Makefile.am b/mesalib/src/mapi/glapi/gen/Makefile.am
index 7af769acf..65bd9137c 100644
--- a/mesalib/src/mapi/glapi/gen/Makefile.am
+++ b/mesalib/src/mapi/glapi/gen/Makefile.am
@@ -125,6 +125,7 @@ API_XML = \
 	ARB_texture_rg.xml \
 	ARB_texture_storage_multisample.xml \
 	ARB_texture_storage.xml \
+	ARB_texture_view.xml \
 	ARB_vertex_array_object.xml \
 	ARB_vertex_attrib_binding.xml \
 	AMD_draw_buffers_blend.xml \
diff --git a/mesalib/src/mapi/glapi/gen/gl_API.xml b/mesalib/src/mapi/glapi/gen/gl_API.xml
index 5c204f9a1..c51eb5f4c 100755
--- a/mesalib/src/mapi/glapi/gen/gl_API.xml
+++ b/mesalib/src/mapi/glapi/gen/gl_API.xml
@@ -8460,7 +8460,13 @@
 
 </category>
 
-<!-- ARB extensions #120...#124 -->
+<!-- ARB extension #120 -->
+
+<xi:include href="ARB_clear_buffer_object.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+
+<!-- ARB extensions #122...#123 -->
+
+<xi:include href="ARB_texture_view.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
 <xi:include href="ARB_vertex_attrib_binding.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
 
diff --git a/mesalib/src/mapi/glapi/gen/gl_genexec.py b/mesalib/src/mapi/glapi/gen/gl_genexec.py
index 3ce190fe3..b557b3b99 100644
--- a/mesalib/src/mapi/glapi/gen/gl_genexec.py
+++ b/mesalib/src/mapi/glapi/gen/gl_genexec.py
@@ -102,6 +102,7 @@ header = """/**
 #include "main/texstate.h"
 #include "main/texstorage.h"
 #include "main/texturebarrier.h"
+#include "main/textureview.h"
 #include "main/transformfeedback.h"
 #include "main/mtypes.h"
 #include "main/varray.h"
diff --git a/mesalib/src/mapi/glapi/gen/gl_gentable.py b/mesalib/src/mapi/glapi/gen/gl_gentable.py
index fb0228939..1c704452f 100755
--- a/mesalib/src/mapi/glapi/gen/gl_gentable.py
+++ b/mesalib/src/mapi/glapi/gen/gl_gentable.py
@@ -50,7 +50,9 @@ header = """/* GLXEXT is the define used in the xserver when the GLX extension i
 #include <execinfo.h>
 #endif
 
+#ifndef _WIN32
 #include <dlfcn.h>
+#endif
 #include <stdlib.h>
 #include <stdio.h>
 
diff --git a/mesalib/src/mesa/Makefile.am b/mesalib/src/mesa/Makefile.am
index a60600e03..884383652 100644
--- a/mesalib/src/mesa/Makefile.am
+++ b/mesalib/src/mesa/Makefile.am
@@ -103,7 +103,11 @@ noinst_PROGRAMS = gen_matypes
 gen_matypes_SOURCES = x86/gen_matypes.c
 BUILT_SOURCES += matypes.h
 
-ARCH_LIBS = libmesa_sse41.la
+ARCH_LIBS =
+
+if SSE41_SUPPORTED
+ARCH_LIBS += libmesa_sse41.la
+endif
 
 if HAVE_X86_64_ASM
 MESA_ASM_FILES_FOR_ARCH += $(X86_64_FILES)
diff --git a/mesalib/src/mesa/Makefile.sources b/mesalib/src/mesa/Makefile.sources
index a84f8a788..39525bc5b 100644
--- a/mesalib/src/mesa/Makefile.sources
+++ b/mesalib/src/mesa/Makefile.sources
@@ -103,6 +103,7 @@ MAIN_FILES = \
 	$(SRCDIR)main/texstate.c \
 	$(SRCDIR)main/texstorage.c \
 	$(SRCDIR)main/texstore.c \
+        $(SRCDIR)main/textureview.c \
 	$(SRCDIR)main/texturebarrier.c \
 	$(SRCDIR)main/transformfeedback.c \
 	$(SRCDIR)main/uniforms.c \
diff --git a/mesalib/src/mesa/SConscript b/mesalib/src/mesa/SConscript
index a2bb9f131..bb9b304ef 100644
--- a/mesalib/src/mesa/SConscript
+++ b/mesalib/src/mesa/SConscript
@@ -133,6 +133,7 @@ main_sources = [
     'main/texstorage.c',
     'main/texstore.c',
     'main/texturebarrier.c',
+    'main/textureview.c',
     'main/transformfeedback.c',
     'main/uniform_query.cpp',
     'main/uniforms.c',
diff --git a/mesalib/src/mesa/drivers/common/driverfuncs.c b/mesalib/src/mesa/drivers/common/driverfuncs.c
index 5faa98af1..f18568827 100644
--- a/mesalib/src/mesa/drivers/common/driverfuncs.c
+++ b/mesalib/src/mesa/drivers/common/driverfuncs.c
@@ -211,6 +211,9 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    /* GL_ARB_texture_storage */
    driver->AllocTextureStorage = _mesa_alloc_texture_storage;
 
+   /* GL_ARB_texture_view */
+   driver->TextureView = NULL;
+
    /* GL_ARB_texture_multisample */
    driver->GetSamplePosition = NULL;
 }
diff --git a/mesalib/src/mesa/drivers/dri/common/dri_util.c b/mesalib/src/mesa/drivers/dri/common/dri_util.c
index 379935015..652140f6a 100644
--- a/mesalib/src/mesa/drivers/dri/common/dri_util.c
+++ b/mesalib/src/mesa/drivers/dri/common/dri_util.c
@@ -442,16 +442,19 @@ driCreateContextAttribs(__DRIscreen *screen, int api,
         return NULL;
     }
 
-    ctx = context->driverPrivate;
+    *error = __DRI_CTX_ERROR_SUCCESS;
+    return context;
+}
+
+void
+driContextSetFlags(struct gl_context *ctx, uint32_t flags)
+{
     if ((flags & __DRI_CTX_FLAG_FORWARD_COMPATIBLE) != 0)
         ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT;
     if ((flags & __DRI_CTX_FLAG_DEBUG) != 0) {
         ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_DEBUG_BIT;
         ctx->Debug.DebugOutput = GL_TRUE;
     }
-
-    *error = __DRI_CTX_ERROR_SUCCESS;
-    return context;
 }
 
 static __DRIcontext *
@@ -877,3 +880,18 @@ const __DRIimageDriverExtension driImageDriverExtension = {
     /*.createContextAttribs       =*/ driCreateContextAttribs,
     /*.getAPIMask                 =*/ driGetAPIMask,
 };
+
+/* swrast copy sub buffer entrypoint. */
+static void driCopySubBuffer(__DRIdrawable *pdp, int x, int y,
+                             int w, int h)
+{
+    assert(pdp->driScreenPriv->swrast_loader);
+
+    pdp->driScreenPriv->driver->CopySubBuffer(pdp, x, y, w, h);
+}
+
+/* for swrast only */
+const __DRIcopySubBufferExtension driCopySubBufferExtension = {
+   { __DRI_COPY_SUB_BUFFER, 1 },
+   /*.copySubBuffer =*/ driCopySubBuffer,
+};
diff --git a/mesalib/src/mesa/drivers/dri/common/dri_util.h b/mesalib/src/mesa/drivers/dri/common/dri_util.h
index 79a8564ad..a79a4ed7a 100644
--- a/mesalib/src/mesa/drivers/dri/common/dri_util.h
+++ b/mesalib/src/mesa/drivers/dri/common/dri_util.h
@@ -66,7 +66,7 @@ extern const __DRIcoreExtension driCoreExtension;
 extern const __DRIswrastExtension driSWRastExtension;
 extern const __DRIdri2Extension driDRI2Extension;
 extern const __DRI2configQueryExtension dri2ConfigQueryExtension;
-
+extern const __DRIcopySubBufferExtension driCopySubBufferExtension;
 /**
  * Driver callback functions.
  *
@@ -115,6 +115,9 @@ struct __DriverAPIRec {
                                     int width, int height);
 
     void (*ReleaseBuffer) (__DRIscreen *screenPrivate, __DRIbuffer *buffer);
+
+    void (*CopySubBuffer)(__DRIdrawable *driDrawPriv, int x, int y,
+                          int w, int h);
 };
 
 extern const struct __DriverAPIRec driDriverAPI;
@@ -289,6 +292,9 @@ dri2InvalidateDrawable(__DRIdrawable *drawable);
 extern void
 driUpdateFramebufferSize(struct gl_context *ctx, const __DRIdrawable *dPriv);
 
+extern void
+driContextSetFlags(struct gl_context *ctx, uint32_t flags);
+
 extern const __DRIimageDriverExtension driImageDriverExtension;
 
 #endif /* _DRI_UTIL_H_ */
diff --git a/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c b/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
index 3a33d74ab..5664af4da 100644
--- a/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
+++ b/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
@@ -23,6 +23,132 @@
 
 #include <stdio.h>
 #include "dri_util.h"
+#include <dlfcn.h>
+#include "main/macros.h"
+
+/* We need GNU extensions to dlfcn.h in order to provide backward
+ * compatibility for the older DRI driver loader mechanism. (dladdr,
+ * Dl_info, and RTLD_DEFAULT are only defined when _GNU_SOURCE is
+ * defined.)
+ */
+#ifdef _GNU_SOURCE
+
+#define MEGADRIVER_STUB_MAX_EXTENSIONS 10
+#define LIB_PATH_SUFFIX "_dri.so"
+#define LIB_PATH_SUFFIX_LENGTH (sizeof(LIB_PATH_SUFFIX)-1)
+
+/* This is the table of extensions that the loader will dlsym() for.
+ *
+ * Initially it is empty for the megadriver stub, but the library
+ * constructor may initialize it based on the name of the library that
+ * is being loaded.
+ */
+PUBLIC const __DRIextension *
+__driDriverExtensions[MEGADRIVER_STUB_MAX_EXTENSIONS] = {
+   NULL
+};
+
+/**
+ * This is a constructor function for the megadriver dynamic library.
+ *
+ * When the driver is dlopen'ed, this function will run. It will
+ * search for the name of the foo_dri.so file that was opened using
+ * the dladdr function.
+ *
+ * After finding foo's name, it will call __driDriverGetExtensions_foo
+ * and use the return to update __driDriverExtensions to enable
+ * compatibility with older DRI driver loaders.
+ */
+__attribute__((constructor)) static void
+megadriver_stub_init(void)
+{
+   Dl_info info;
+   char *driver_name;
+   size_t name_len;
+   char *get_extensions_name;
+   const __DRIextension **(*get_extensions)(void);
+   const __DRIextension **extensions;
+   int i;
+
+   /* Call dladdr on __driDriverExtensions. We are really
+    * interested in the returned info.dli_fname so we can
+    * figure out the path name of the library being loaded.
+    */
+   i = dladdr((void*) __driDriverExtensions, &info);
+   if (i == 0)
+      return;
+
+   /* Search for the last '/' character in the path. */
+   driver_name = strrchr(info.dli_fname, '/');
+   if (driver_name != NULL) {
+      /* Skip '/' character */
+      driver_name++;
+   } else {
+      /* Try using the start of the path */
+      driver_name = (char*) info.dli_fname;
+   }
+
+   /* Make sure the path ends with _dri.so */
+   name_len = strlen(driver_name);
+   i = name_len - LIB_PATH_SUFFIX_LENGTH;
+   if (i < 0 || strcmp(driver_name + i, LIB_PATH_SUFFIX) != 0)
+      return;
+
+   /* Duplicate the string so we can modify it.
+    * So far we've been using info.dli_fname.
+    */
+   driver_name = strdup(driver_name);
+   if (!driver_name)
+      return;
+
+   /* The path ends with _dri.so. Chop this part of the
+    * string off. Then we'll have the driver's final name.
+    */
+   driver_name[i] = '\0';
+
+   i = asprintf(&get_extensions_name, "%s_%s",
+                __DRI_DRIVER_GET_EXTENSIONS, driver_name);
+   free(driver_name);
+   if (i == -1)
+      return;
+
+   /* dlsym to get the driver's get extensions function. We
+    * don't have the dlopen handle, so we have to use
+    * RTLD_DEFAULT. It seems unlikely that the symbol will
+    * be found in another library, but this isn't optimal.
+    */
+   get_extensions = dlsym(RTLD_DEFAULT, get_extensions_name);
+   free(get_extensions_name);
+   if (!get_extensions)
+      return;
+
+   /* Use the newer DRI loader entrypoint to find extensions.
+    * We will then expose these extensions via the older
+    * __driDriverExtensions symbol.
+    */
+   extensions = get_extensions();
+
+   /* Copy the extensions into the __driDriverExtensions array
+    * we declared.
+    */
+   for (i = 0; i < ARRAY_SIZE(__driDriverExtensions); i++) {
+      __driDriverExtensions[i] = extensions[i];
+      if (extensions[i] == NULL)
+         break;
+   }
+
+   /* If the driver had more extensions than we reserved, then
+    * bail out.
+    */
+   if (i == ARRAY_SIZE(__driDriverExtensions)) {
+      __driDriverExtensions[0] = NULL;
+      fprintf(stderr, "Megadriver stub did not reserve enough extension "
+              "slots.\n");
+      return;
+   }
+}
+
+#endif /* _GNU_SOURCE */
 
 static const
 __DRIconfig **stub_error_init_screen(__DRIscreen *psp)
diff --git a/mesalib/src/mesa/drivers/dri/swrast/swrast.c b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
index d0ca80fa7..1848da5d9 100755..100644
--- a/mesalib/src/mesa/drivers/dri/swrast/swrast.c
+++ b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
@@ -407,7 +407,7 @@ swrast_map_renderbuffer(struct gl_context *ctx,
       stride = w * cpp;
       xrb->Base.Buffer = (GLubyte*)malloc(h * stride);
 
-      sPriv->swrast_loader->getImage(dPriv, x, y, w, h,
+      sPriv->swrast_loader->getImage(dPriv, x, rb->Height - y - h, w, h,
 				     (char *) xrb->Base.Buffer,
 				     dPriv->loaderPrivate);
 
@@ -710,6 +710,8 @@ dri_create_context(gl_api api,
 	goto context_fail;
     }
 
+    driContextSetFlags(mesaCtx, flags);
+
     /* do bounds checking to prevent segfaults and server crashes! */
     mesaCtx->Const.CheckArrayBounds = GL_TRUE;
 
@@ -825,6 +827,39 @@ dri_unbind_context(__DRIcontext * cPriv)
     return GL_TRUE;
 }
 
+static void
+dri_copy_sub_buffer(__DRIdrawable *dPriv, int x, int y,
+                    int w, int h)
+{
+    __DRIscreen *sPriv = dPriv->driScreenPriv;
+    void *data;
+    int iy;
+    struct dri_drawable *drawable = dri_drawable(dPriv);
+    struct gl_framebuffer *fb;
+    struct dri_swrast_renderbuffer *frontrb, *backrb;
+
+    TRACE;
+
+    fb = &drawable->Base;
+
+    frontrb =
+	dri_swrast_renderbuffer(fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer);
+    backrb =
+	dri_swrast_renderbuffer(fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer);
+
+    /* check for signle-buffered */
+    if (backrb == NULL)
+       return;
+
+    iy = frontrb->Base.Base.Height - y - h;
+    data = (char *)backrb->Base.Buffer + (iy * backrb->pitch) + (x * ((backrb->bpp + 7) / 8));
+    sPriv->swrast_loader->putImage2(dPriv, __DRI_SWRAST_IMAGE_OP_SWAP,
+                                    x, iy, w, h,
+                                    frontrb->pitch,
+                                    data,
+                                    dPriv->loaderPrivate);
+}
+
 
 static const struct __DriverAPIRec swrast_driver_api = {
     /*.InitScreen = */dri_init_screen,
@@ -836,6 +871,7 @@ static const struct __DriverAPIRec swrast_driver_api = {
     /*.SwapBuffers = */dri_swap_buffers,
     /*.MakeCurrent = */dri_make_current,
     /*.UnbindContext = */dri_unbind_context,
+    /*.CopySubBuffer = */dri_copy_sub_buffer,
     /*.AllocateBuffer = */NULL,
     /*.ReleaseBuffer = */NULL
 };
@@ -848,6 +884,7 @@ static const struct __DRIDriverVtableExtensionRec swrast_vtable = {
 static const __DRIextension *swrast_driver_extensions[] = {
     &driCoreExtension.base,
     &driSWRastExtension.base,
+    &driCopySubBufferExtension.base,
     &swrast_vtable.base,
     NULL
 };
diff --git a/mesalib/src/mesa/main/attrib.c b/mesalib/src/mesa/main/attrib.c
index c9332bd52..30c815d67 100644
--- a/mesalib/src/mesa/main/attrib.c
+++ b/mesalib/src/mesa/main/attrib.c
@@ -182,7 +182,7 @@ struct texture_state
  * Allocate new attribute node of given type/kind.  Attach payload data.
  * Insert it into the linked list named by 'head'.
  */
-static void
+static bool
 save_attrib_data(struct gl_attrib_node **head,
                  GLbitfield kind, void *payload)
 {
@@ -196,7 +196,42 @@ save_attrib_data(struct gl_attrib_node **head,
    }
    else {
       /* out of memory! */
+      return false;
+   }
+   return true;
+}
+
+
+/**
+ * Helper function for_mesa_PushAttrib for simple attributes.
+ * Allocates memory for attribute data and copies the given attribute data.
+ * \param head  head of linked list to insert attribute data into
+ * \param attr_bit  one of the GL_<attrib>_BIT flags
+ * \param attr_size  number of bytes to allocate for attribute data
+ * \param attr_data  the attribute data to copy
+ * \return true for success, false for out of memory
+ */
+static bool
+push_attrib(struct gl_context *ctx, struct gl_attrib_node **head,
+            GLbitfield attr_bit, GLuint attr_size, const void *attr_data)
+{
+   void *attribute;
+
+   attribute = MALLOC(attr_size);
+   if (attribute == NULL) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+      return false;
+   }
+
+   if (save_attrib_data(head, attr_bit, attribute)) {
+      memcpy(attribute, attr_data, attr_size);
    }
+   else {
+      FREE(attribute);
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+      return false;
+   }
+   return true;
 }
 
 
@@ -220,42 +255,58 @@ _mesa_PushAttrib(GLbitfield mask)
    head = NULL;
 
    if (mask & GL_ACCUM_BUFFER_BIT) {
-      struct gl_accum_attrib *attr;
-      attr = MALLOC_STRUCT( gl_accum_attrib );
-      memcpy( attr, &ctx->Accum, sizeof(struct gl_accum_attrib) );
-      save_attrib_data(&head, GL_ACCUM_BUFFER_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_ACCUM_BUFFER_BIT,
+                       sizeof(struct gl_accum_attrib),
+                       (void*)&ctx->Accum))
+         goto end;
    }
 
    if (mask & GL_COLOR_BUFFER_BIT) {
       GLuint i;
       struct gl_colorbuffer_attrib *attr;
       attr = MALLOC_STRUCT( gl_colorbuffer_attrib );
-      memcpy( attr, &ctx->Color, sizeof(struct gl_colorbuffer_attrib) );
-      /* push the Draw FBO's DrawBuffer[] state, not ctx->Color.DrawBuffer[] */
-      for (i = 0; i < ctx->Const.MaxDrawBuffers; i ++)
-         attr->DrawBuffer[i] = ctx->DrawBuffer->ColorDrawBuffer[i];
-      save_attrib_data(&head, GL_COLOR_BUFFER_BIT, attr);
+      if (attr == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
+
+    if (save_attrib_data(&head, GL_COLOR_BUFFER_BIT, attr)) {
+         memcpy(attr, &ctx->Color, sizeof(struct gl_colorbuffer_attrib));
+         /* push the Draw FBO's DrawBuffer[] state, not ctx->Color.DrawBuffer[] */
+         for (i = 0; i < ctx->Const.MaxDrawBuffers; i ++)
+            attr->DrawBuffer[i] = ctx->DrawBuffer->ColorDrawBuffer[i];
+      }
+      else {
+         FREE(attr);
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
    }
 
    if (mask & GL_CURRENT_BIT) {
-      struct gl_current_attrib *attr;
-      FLUSH_CURRENT( ctx, 0 );
-      attr = MALLOC_STRUCT( gl_current_attrib );
-      memcpy( attr, &ctx->Current, sizeof(struct gl_current_attrib) );
-      save_attrib_data(&head, GL_CURRENT_BIT, attr);
+      FLUSH_CURRENT(ctx, 0);
+      if (!push_attrib(ctx, &head, GL_CURRENT_BIT,
+                       sizeof(struct gl_current_attrib),
+                       (void*)&ctx->Current))
+         goto end;
    }
 
    if (mask & GL_DEPTH_BUFFER_BIT) {
-      struct gl_depthbuffer_attrib *attr;
-      attr = MALLOC_STRUCT( gl_depthbuffer_attrib );
-      memcpy( attr, &ctx->Depth, sizeof(struct gl_depthbuffer_attrib) );
-      save_attrib_data(&head, GL_DEPTH_BUFFER_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_DEPTH_BUFFER_BIT,
+                       sizeof(struct gl_depthbuffer_attrib),
+                       (void*)&ctx->Depth))
+         goto end;
    }
 
    if (mask & GL_ENABLE_BIT) {
       struct gl_enable_attrib *attr;
       GLuint i;
       attr = MALLOC_STRUCT( gl_enable_attrib );
+      if (attr == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
+
       /* Copy enable flags from all other attributes into the enable struct. */
       attr->AlphaTest = ctx->Color.AlphaEnabled;
       attr->AutoNormal = ctx->Eval.AutoNormal;
@@ -322,97 +373,112 @@ _mesa_PushAttrib(GLbitfield mask)
       /* GL_ARB_fragment_program */
       attr->FragmentProgram = ctx->FragmentProgram.Enabled;
 
-      save_attrib_data(&head, GL_ENABLE_BIT, attr);
+      if (!save_attrib_data(&head, GL_ENABLE_BIT, attr)) {
+         FREE(attr);
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
 
       /* GL_ARB_framebuffer_sRGB / GL_EXT_framebuffer_sRGB */
       attr->sRGBEnabled = ctx->Color.sRGBEnabled;
    }
 
    if (mask & GL_EVAL_BIT) {
-      struct gl_eval_attrib *attr;
-      attr = MALLOC_STRUCT( gl_eval_attrib );
-      memcpy( attr, &ctx->Eval, sizeof(struct gl_eval_attrib) );
-      save_attrib_data(&head, GL_EVAL_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_EVAL_BIT,
+                       sizeof(struct gl_eval_attrib),
+                       (void*)&ctx->Eval))
+         goto end;
    }
 
    if (mask & GL_FOG_BIT) {
-      struct gl_fog_attrib *attr;
-      attr = MALLOC_STRUCT( gl_fog_attrib );
-      memcpy( attr, &ctx->Fog, sizeof(struct gl_fog_attrib) );
-      save_attrib_data(&head, GL_FOG_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_FOG_BIT,
+                       sizeof(struct gl_fog_attrib),
+                       (void*)&ctx->Fog))
+         goto end;
    }
 
    if (mask & GL_HINT_BIT) {
-      struct gl_hint_attrib *attr;
-      attr = MALLOC_STRUCT( gl_hint_attrib );
-      memcpy( attr, &ctx->Hint, sizeof(struct gl_hint_attrib) );
-      save_attrib_data(&head, GL_HINT_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_HINT_BIT,
+                       sizeof(struct gl_hint_attrib),
+                       (void*)&ctx->Hint))
+         goto end;
    }
 
    if (mask & GL_LIGHTING_BIT) {
-      struct gl_light_attrib *attr;
-      FLUSH_CURRENT(ctx, 0);	/* flush material changes */
-      attr = MALLOC_STRUCT( gl_light_attrib );
-      memcpy( attr, &ctx->Light, sizeof(struct gl_light_attrib) );
-      save_attrib_data(&head, GL_LIGHTING_BIT, attr);
+      FLUSH_CURRENT(ctx, 0);   /* flush material changes */
+      if (!push_attrib(ctx, &head, GL_LIGHTING_BIT,
+                       sizeof(struct gl_light_attrib),
+                       (void*)&ctx->Light))
+         goto end;
    }
 
    if (mask & GL_LINE_BIT) {
-      struct gl_line_attrib *attr;
-      attr = MALLOC_STRUCT( gl_line_attrib );
-      memcpy( attr, &ctx->Line, sizeof(struct gl_line_attrib) );
-      save_attrib_data(&head, GL_LINE_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_LINE_BIT,
+                       sizeof(struct gl_line_attrib),
+                       (void*)&ctx->Line))
+         goto end;
    }
 
    if (mask & GL_LIST_BIT) {
-      struct gl_list_attrib *attr;
-      attr = MALLOC_STRUCT( gl_list_attrib );
-      memcpy( attr, &ctx->List, sizeof(struct gl_list_attrib) );
-      save_attrib_data(&head, GL_LIST_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_LIST_BIT,
+                       sizeof(struct gl_list_attrib),
+                       (void*)&ctx->List))
+         goto end;
    }
 
    if (mask & GL_PIXEL_MODE_BIT) {
       struct gl_pixel_attrib *attr;
       attr = MALLOC_STRUCT( gl_pixel_attrib );
-      memcpy( attr, &ctx->Pixel, sizeof(struct gl_pixel_attrib) );
-      /* push the Read FBO's ReadBuffer state, not ctx->Pixel.ReadBuffer */
-      attr->ReadBuffer = ctx->ReadBuffer->ColorReadBuffer;
-      save_attrib_data(&head, GL_PIXEL_MODE_BIT, attr);
+      if (attr == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
+
+      if (save_attrib_data(&head, GL_PIXEL_MODE_BIT, attr)) {
+         memcpy(attr, &ctx->Pixel, sizeof(struct gl_pixel_attrib));
+         /* push the Read FBO's ReadBuffer state, not ctx->Pixel.ReadBuffer */
+         attr->ReadBuffer = ctx->ReadBuffer->ColorReadBuffer;
+      }
+      else {
+         FREE(attr);
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib");
+         goto end;
+      }
    }
 
    if (mask & GL_POINT_BIT) {
-      struct gl_point_attrib *attr;
-      attr = MALLOC_STRUCT( gl_point_attrib );
-      memcpy( attr, &ctx->Point, sizeof(struct gl_point_attrib) );
-      save_attrib_data(&head, GL_POINT_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_POINT_BIT,
+                       sizeof(struct gl_point_attrib),
+                       (void*)&ctx->Point))
+         goto end;
    }
 
    if (mask & GL_POLYGON_BIT) {
-      struct gl_polygon_attrib *attr;
-      attr = MALLOC_STRUCT( gl_polygon_attrib );
-      memcpy( attr, &ctx->Polygon, sizeof(struct gl_polygon_attrib) );
-      save_attrib_data(&head, GL_POLYGON_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_POLYGON_BIT,
+                       sizeof(struct gl_polygon_attrib),
+                       (void*)&ctx->Polygon))
+         goto end;
    }
 
    if (mask & GL_POLYGON_STIPPLE_BIT) {
-      GLuint *stipple;
-      stipple = malloc( 32*sizeof(GLuint) );
-      memcpy( stipple, ctx->PolygonStipple, 32*sizeof(GLuint) );
-      save_attrib_data(&head, GL_POLYGON_STIPPLE_BIT, stipple);
+      if (!push_attrib(ctx, &head, GL_POLYGON_STIPPLE_BIT,
+                       sizeof(ctx->PolygonStipple),
+                       (void*)&ctx->PolygonStipple))
+         goto end;
    }
 
    if (mask & GL_SCISSOR_BIT) {
-      struct gl_scissor_attrib *attr;
-      attr = MALLOC_STRUCT( gl_scissor_attrib );
-      memcpy( attr, &ctx->Scissor, sizeof(struct gl_scissor_attrib) );
-      save_attrib_data(&head, GL_SCISSOR_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_SCISSOR_BIT,
+                       sizeof(struct gl_scissor_attrib),
+                       (void*)&ctx->Scissor))
+         goto end;
    }
 
    if (mask & GL_STENCIL_BUFFER_BIT) {
-      struct gl_stencil_attrib *attr;
-      attr = MALLOC_STRUCT( gl_stencil_attrib );
-      memcpy( attr, &ctx->Stencil, sizeof(struct gl_stencil_attrib) );
-      save_attrib_data(&head, GL_STENCIL_BUFFER_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_STENCIL_BUFFER_BIT,
+                       sizeof(struct gl_stencil_attrib),
+                       (void*)&ctx->Stencil))
+         goto end;
    }
 
    if (mask & GL_TEXTURE_BIT) {
@@ -424,6 +490,12 @@ _mesa_PushAttrib(GLbitfield mask)
          goto end;
       }
 
+      if (!save_attrib_data(&head, GL_TEXTURE_BIT, texstate)) {
+         FREE(texstate);
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib(GL_TEXTURE_BIT)");
+         goto end;
+      }
+
       _mesa_lock_context_textures(ctx);
 
       /* copy/save the bulk of texture state here */
@@ -450,35 +522,35 @@ _mesa_PushAttrib(GLbitfield mask)
       _mesa_reference_shared_state(ctx, &texstate->SharedRef, ctx->Shared);
 
       _mesa_unlock_context_textures(ctx);
-
-      save_attrib_data(&head, GL_TEXTURE_BIT, texstate);
    }
 
    if (mask & GL_TRANSFORM_BIT) {
-      struct gl_transform_attrib *attr;
-      attr = MALLOC_STRUCT( gl_transform_attrib );
-      memcpy( attr, &ctx->Transform, sizeof(struct gl_transform_attrib) );
-      save_attrib_data(&head, GL_TRANSFORM_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_TRANSFORM_BIT,
+                       sizeof(struct gl_transform_attrib),
+                       (void*)&ctx->Transform))
+         goto end;
    }
 
    if (mask & GL_VIEWPORT_BIT) {
-      struct gl_viewport_attrib *attr;
-      attr = MALLOC_STRUCT( gl_viewport_attrib );
-      memcpy( attr, &ctx->Viewport, sizeof(struct gl_viewport_attrib) );
-      save_attrib_data(&head, GL_VIEWPORT_BIT, attr);
+      if (!push_attrib(ctx, &head, GL_VIEWPORT_BIT,
+                       sizeof(struct gl_viewport_attrib),
+                       (void*)&ctx->Viewport))
+         goto end;
    }
 
    /* GL_ARB_multisample */
    if (mask & GL_MULTISAMPLE_BIT_ARB) {
-      struct gl_multisample_attrib *attr;
-      attr = MALLOC_STRUCT( gl_multisample_attrib );
-      memcpy( attr, &ctx->Multisample, sizeof(struct gl_multisample_attrib) );
-      save_attrib_data(&head, GL_MULTISAMPLE_BIT_ARB, attr);
+      if (!push_attrib(ctx, &head, GL_MULTISAMPLE_BIT_ARB,
+                       sizeof(struct gl_multisample_attrib),
+                       (void*)&ctx->Multisample))
+         goto end;
    }
 
 end:
-   ctx->AttribStack[ctx->AttribStackDepth] = head;
-   ctx->AttribStackDepth++;
+   if (head != NULL) {
+       ctx->AttribStack[ctx->AttribStackDepth] = head;
+       ctx->AttribStackDepth++;
+   }
 }
 
 
@@ -641,12 +713,6 @@ pop_enable_group(struct gl_context *ctx, const struct gl_enable_attrib *enable)
             _mesa_set_enable(ctx, GL_TEXTURE_CUBE_MAP,
                              !!(enabled & TEXTURE_CUBE_BIT));
          }
-         if (ctx->Extensions.MESA_texture_array) {
-            _mesa_set_enable(ctx, GL_TEXTURE_1D_ARRAY_EXT,
-                             !!(enabled & TEXTURE_1D_ARRAY_BIT));
-            _mesa_set_enable(ctx, GL_TEXTURE_2D_ARRAY_EXT,
-                             !!(enabled & TEXTURE_2D_ARRAY_BIT));
-         }
       }
 
       if (ctx->Texture.Unit[i].TexGenEnabled != genEnabled) {
@@ -688,12 +754,6 @@ pop_texture_group(struct gl_context *ctx, struct texture_state *texstate)
          _mesa_set_enable(ctx, GL_TEXTURE_RECTANGLE_NV,
                           !!(unit->Enabled & TEXTURE_RECT_BIT));
       }
-      if (ctx->Extensions.MESA_texture_array) {
-         _mesa_set_enable(ctx, GL_TEXTURE_1D_ARRAY_EXT,
-                          !!(unit->Enabled & TEXTURE_1D_ARRAY_BIT));
-         _mesa_set_enable(ctx, GL_TEXTURE_2D_ARRAY_EXT,
-                          !!(unit->Enabled & TEXTURE_2D_ARRAY_BIT));
-      }
       _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, unit->EnvMode);
       _mesa_TexEnvfv(GL_TEXTURE_ENV, GL_TEXTURE_ENV_COLOR, unit->EnvColor);
       _mesa_TexGeni(GL_S, GL_TEXTURE_GEN_MODE, unit->GenS.Mode);
@@ -768,7 +828,7 @@ pop_texture_group(struct gl_context *ctx, struct texture_state *texstate)
          }
          else if ((obj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
                    obj->Target == GL_TEXTURE_2D_ARRAY_EXT) &&
-                  !ctx->Extensions.MESA_texture_array) {
+                  !ctx->Extensions.EXT_texture_array) {
             continue;
          }
          else if (obj->Target == GL_TEXTURE_CUBE_MAP_ARRAY &&
@@ -1482,13 +1542,20 @@ restore_array_attrib(struct gl_context *ctx,
  * init/alloc the fields of 'attrib'.
  * Needs to the init part matching free_array_attrib_data below.
  */
-static void
+static bool
 init_array_attrib_data(struct gl_context *ctx,
                        struct gl_array_attrib *attrib)
 {
    /* Get a non driver gl_array_object. */
    attrib->ArrayObj = CALLOC_STRUCT( gl_array_object );
+
+   if (attrib->ArrayObj == NULL) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib");
+      return false;
+   }
+
    _mesa_initialize_array_object(ctx, attrib->ArrayObj, 0);
+   return true;
 }
 
 /**
@@ -1529,24 +1596,65 @@ _mesa_PushClientAttrib(GLbitfield mask)
       struct gl_pixelstore_attrib *attr;
       /* packing attribs */
       attr = CALLOC_STRUCT( gl_pixelstore_attrib );
-      copy_pixelstore(ctx, attr, &ctx->Pack);
-      save_attrib_data(&head, GL_CLIENT_PACK_BIT, attr);
+      if (attr == NULL) {
+         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib" );
+         goto end;
+      }
+      if (save_attrib_data(&head, GL_CLIENT_PACK_BIT, attr)) {
+         copy_pixelstore(ctx, attr, &ctx->Pack);
+      }
+      else {
+         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib" );
+         FREE(attr);
+         goto end;
+      }
+
       /* unpacking attribs */
       attr = CALLOC_STRUCT( gl_pixelstore_attrib );
-      copy_pixelstore(ctx, attr, &ctx->Unpack);
-      save_attrib_data(&head, GL_CLIENT_UNPACK_BIT, attr);
+      if (attr == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib");
+         goto end;
+      }
+
+      if (save_attrib_data(&head, GL_CLIENT_UNPACK_BIT, attr)) {
+         copy_pixelstore(ctx, attr, &ctx->Unpack);
+      }
+      else {
+         _mesa_error( ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib" );
+         FREE(attr);
+         goto end;
+       }
    }
 
    if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) {
       struct gl_array_attrib *attr;
       attr = CALLOC_STRUCT( gl_array_attrib );
-      init_array_attrib_data(ctx, attr);
-      save_array_attrib(ctx, attr, &ctx->Array);
-      save_attrib_data(&head, GL_CLIENT_VERTEX_ARRAY_BIT, attr);
-   }
+      if (attr == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib");
+         goto end;
+      }
 
-   ctx->ClientAttribStack[ctx->ClientAttribStackDepth] = head;
-   ctx->ClientAttribStackDepth++;
+      if (!init_array_attrib_data(ctx, attr)) {
+         FREE(attr);
+         goto end;
+      }
+
+      if (save_attrib_data(&head, GL_CLIENT_VERTEX_ARRAY_BIT, attr)) {
+         save_array_attrib(ctx, attr, &ctx->Array);
+      }
+      else {
+         free_array_attrib_data(ctx, attr);
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushClientAttrib");
+         FREE(attr);
+         /* goto to keep safe from possible later changes */
+         goto end;
+      }
+   }
+end:
+   if (head != NULL) {
+       ctx->ClientAttribStack[ctx->ClientAttribStackDepth] = head;
+       ctx->ClientAttribStackDepth++;
+   }
 }
 
 
diff --git a/mesalib/src/mesa/main/bufferobj.c b/mesalib/src/mesa/main/bufferobj.c
index 3127a8247..45ddeaba9 100644
--- a/mesalib/src/mesa/main/bufferobj.c
+++ b/mesalib/src/mesa/main/bufferobj.c
@@ -41,6 +41,9 @@
 #include "fbobject.h"
 #include "mtypes.h"
 #include "texobj.h"
+#include "teximage.h"
+#include "glformats.h"
+#include "texstore.h"
 #include "transformfeedback.h"
 #include "dispatch.h"
 
@@ -124,11 +127,13 @@ get_buffer_target(struct gl_context *ctx, GLenum target)
  * Get the buffer object bound to the specified target in a GL context.
  * \param ctx  the GL context
  * \param target  the buffer object target to be retrieved.
+ * \param error  the GL error to record if target is illegal.
  * \return   pointer to the buffer object bound to \c target in the
  *           specified context or \c NULL if \c target is invalid.
  */
 static inline struct gl_buffer_object *
-get_buffer(struct gl_context *ctx, const char *func, GLenum target)
+get_buffer(struct gl_context *ctx, const char *func, GLenum target,
+           GLenum error)
 {
    struct gl_buffer_object **bufObj = get_buffer_target(ctx, target);
 
@@ -138,7 +143,7 @@ get_buffer(struct gl_context *ctx, const char *func, GLenum target)
    }
 
    if (!_mesa_is_bufferobj(*bufObj)) {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(buffer 0)", func);
+      _mesa_error(ctx, error, "%s(no buffer bound)", func);
       return NULL;
    }
 
@@ -186,25 +191,58 @@ simplified_access_mode(struct gl_context *ctx, GLbitfield access)
 
 
 /**
+ * Test if the buffer is mapped, and if so, if the mapped range overlaps the
+ * given range.
+ * The regions do not overlap if and only if the end of the given
+ * region is before the mapped region or the start of the given region
+ * is after the mapped region.
+ *
+ * \param obj     Buffer object target on which to operate.
+ * \param offset  Offset of the first byte of the subdata range.
+ * \param size    Size, in bytes, of the subdata range.
+ * \return   true if ranges overlap, false otherwise
+ *
+ */
+static bool
+bufferobj_range_mapped(const struct gl_buffer_object *obj,
+                       GLintptr offset, GLsizeiptr size)
+{
+   if (_mesa_bufferobj_mapped(obj)) {
+      const GLintptr end = offset + size;
+      const GLintptr mapEnd = obj->Offset + obj->Length;
+
+      if (!(end <= obj->Offset || offset >= mapEnd)) {
+         return true;
+      }
+   }
+   return false;
+}
+
+
+/**
  * Tests the subdata range parameters and sets the GL error code for
- * \c glBufferSubDataARB and \c glGetBufferSubDataARB.
+ * \c glBufferSubDataARB, \c glGetBufferSubDataARB and
+ * \c glClearBufferSubData.
  *
  * \param ctx     GL context.
  * \param target  Buffer object target on which to operate.
  * \param offset  Offset of the first byte of the subdata range.
  * \param size    Size, in bytes, of the subdata range.
+ * \param mappedRange  If true, checks if an overlapping range is mapped.
+ *                     If false, checks if buffer is mapped.
+ * \param errorNoBuffer  Error code if no buffer is bound to target.
  * \param caller  Name of calling function for recording errors.
  * \return   A pointer to the buffer object bound to \c target in the
  *           specified context or \c NULL if any of the parameter or state
- *           conditions for \c glBufferSubDataARB or \c glGetBufferSubDataARB
- *           are invalid.
+ *           conditions are invalid.
  *
- * \sa glBufferSubDataARB, glGetBufferSubDataARB
+ * \sa glBufferSubDataARB, glGetBufferSubDataARB, glClearBufferSubData
  */
 static struct gl_buffer_object *
-buffer_object_subdata_range_good( struct gl_context * ctx, GLenum target, 
-                                  GLintptrARB offset, GLsizeiptrARB size,
-                                  const char *caller )
+buffer_object_subdata_range_good(struct gl_context * ctx, GLenum target,
+                                 GLintptrARB offset, GLsizeiptrARB size,
+                                 bool mappedRange, GLenum errorNoBuffer,
+                                 const char *caller)
 {
    struct gl_buffer_object *bufObj;
 
@@ -218,22 +256,30 @@ buffer_object_subdata_range_good( struct gl_context * ctx, GLenum target,
       return NULL;
    }
 
-   bufObj = get_buffer(ctx, caller, target);
+   bufObj = get_buffer(ctx, caller, target, errorNoBuffer);
    if (!bufObj)
       return NULL;
 
    if (offset + size > bufObj->Size) {
       _mesa_error(ctx, GL_INVALID_VALUE,
-		  "%s(offset %lu + size %lu > buffer size %lu)", caller,
+                  "%s(offset %lu + size %lu > buffer size %lu)", caller,
                   (unsigned long) offset,
                   (unsigned long) size,
                   (unsigned long) bufObj->Size);
       return NULL;
    }
-   if (_mesa_bufferobj_mapped(bufObj)) {
-      /* Buffer is currently mapped */
-      _mesa_error(ctx, GL_INVALID_OPERATION, "%s", caller);
-      return NULL;
+
+   if (mappedRange) {
+      if (bufferobj_range_mapped(bufObj, offset, size)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "%s", caller);
+         return NULL;
+      }
+   }
+   else {
+      if (_mesa_bufferobj_mapped(bufObj)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "%s", caller);
+         return NULL;
+      }
    }
 
    return bufObj;
@@ -241,6 +287,98 @@ buffer_object_subdata_range_good( struct gl_context * ctx, GLenum target,
 
 
 /**
+ * Test the format and type parameters and set the GL error code for
+ * \c glClearBufferData and \c glClearBufferSubData.
+ *
+ * \param ctx             GL context.
+ * \param internalformat  Format to which the data is to be converted.
+ * \param format          Format of the supplied data.
+ * \param type            Type of the supplied data.
+ * \param caller          Name of calling function for recording errors.
+ * \return   If internalformat, format and type are legal the gl_format
+ *           corresponding to internalformat, otherwise MESA_FORMAT_NONE.
+ *
+ * \sa glClearBufferData and glClearBufferSubData
+ */
+static gl_format
+validate_clear_buffer_format(struct gl_context *ctx,
+                             GLenum internalformat,
+                             GLenum format, GLenum type,
+                             const char *caller)
+{
+   gl_format mesaFormat;
+   GLenum errorFormatType;
+
+   mesaFormat = _mesa_validate_texbuffer_format(ctx, internalformat);
+   if (mesaFormat == MESA_FORMAT_NONE) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(invalid internalformat)", caller);
+      return MESA_FORMAT_NONE;
+   }
+
+   /* NOTE: not mentioned in ARB_clear_buffer_object but according to
+    * EXT_texture_integer there is no conversion between integer and
+    * non-integer formats
+   */
+   if (_mesa_is_enum_format_signed_int(format) !=
+       _mesa_is_format_integer_color(mesaFormat)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(integer vs non-integer)", caller);
+      return MESA_FORMAT_NONE;
+   }
+
+   if (!_mesa_is_color_format(format)) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(format is not a color format)", caller);
+      return MESA_FORMAT_NONE;
+   }
+
+   errorFormatType = _mesa_error_check_format_and_type(ctx, format, type);
+   if (errorFormatType != GL_NO_ERROR) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(invalid format or type)", caller);
+      return MESA_FORMAT_NONE;
+   }
+
+   return mesaFormat;
+}
+
+
+/**
+ * Convert user-specified clear value to the specified internal format.
+ *
+ * \param ctx             GL context.
+ * \param internalformat  Format to which the data is converted.
+ * \param clearValue      Points to the converted clear value.
+ * \param format          Format of the supplied data.
+ * \param type            Type of the supplied data.
+ * \param data            Data which is to be converted to internalformat.
+ * \param caller          Name of calling function for recording errors.
+ * \return   true if data could be converted, false otherwise.
+ *
+ * \sa glClearBufferData, glClearBufferSubData
+ */
+static bool
+convert_clear_buffer_data(struct gl_context *ctx,
+                          gl_format internalformat,
+                          GLubyte *clearValue, GLenum format, GLenum type,
+                          const GLvoid *data, const char *caller)
+{
+   GLenum internalformatBase = _mesa_get_format_base_format(internalformat);
+
+   if (_mesa_texstore(ctx, 1, internalformatBase, internalformat,
+                      0, &clearValue, 1, 1, 1,
+                      format, type, data, &ctx->Unpack)) {
+      return true;
+   }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s", caller);
+      return false;
+   }
+}
+
+
+/**
  * Allocate and initialize a new buffer object.
  * 
  * Default callback for the \c dd_function_table::NewBufferObject() hook.
@@ -505,6 +643,82 @@ _mesa_buffer_get_subdata( struct gl_context *ctx, GLintptrARB offset,
 
 
 /**
+ * Clear a subrange of the buffer object with copies of the supplied data.
+ * If data is NULL the buffer is filled with zeros.
+ *
+ * This is the default callback for \c dd_function_table::ClearBufferSubData()
+ * Note that all GL error checking will have been done already.
+ *
+ * \param ctx             GL context.
+ * \param offset          Offset of the first byte to be cleared.
+ * \param size            Size, in bytes, of the to be cleared range.
+ * \param clearValue      Source of the data.
+ * \param clearValueSize  Size, in bytes, of the supplied data.
+ * \param bufObj          Object to be cleared.
+ *
+ * \sa glClearBufferSubData, glClearBufferData and
+ * dd_function_table::ClearBufferSubData.
+ */
+static void
+_mesa_buffer_clear_subdata(struct gl_context *ctx,
+                           GLintptr offset, GLsizeiptr size,
+                           const GLvoid *clearValue,
+                           GLsizeiptr clearValueSize,
+                           struct gl_buffer_object *bufObj)
+{
+   GLsizeiptr i;
+   GLubyte *dest;
+
+   if (_mesa_bufferobj_mapped(bufObj)) {
+      GLubyte *data = malloc(size);
+      GLubyte *dataStart = data;
+      if (data == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClearBuffer[Sub]Data");
+         return;
+      }
+
+      if (clearValue == NULL) {
+         /* Clear with zeros, per the spec */
+         memset(data, 0, size);
+      }
+      else {
+         for (i = 0; i < size/clearValueSize; ++i) {
+            memcpy(data, clearValue, clearValueSize);
+            data += clearValueSize;
+         }
+      }
+      ctx->Driver.BufferSubData(ctx, offset, size, dataStart, bufObj);
+      return;
+   }
+
+   ASSERT(ctx->Driver.MapBufferRange);
+   dest = ctx->Driver.MapBufferRange(ctx, offset, size,
+                                     GL_MAP_WRITE_BIT |
+                                     GL_MAP_INVALIDATE_RANGE_BIT,
+                                     bufObj);
+
+   if (!dest) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClearBuffer[Sub]Data");
+      return;
+   }
+
+   if (clearValue == NULL) {
+      /* Clear with zeros, per the spec */
+      memset(dest, 0, size);
+      ctx->Driver.UnmapBuffer(ctx, bufObj);
+      return;
+   }
+
+   for (i = 0; i < size/clearValueSize; ++i) {
+      memcpy(dest, clearValue, clearValueSize);
+      dest += clearValueSize;
+   }
+
+   ctx->Driver.UnmapBuffer(ctx, bufObj);
+}
+
+
+/**
  * Default fallback for \c dd_function_table::MapBufferRange().
  * Called via glMapBufferRange().
  */
@@ -810,6 +1024,9 @@ _mesa_init_buffer_object_functions(struct dd_function_table *driver)
    driver->GetBufferSubData = _mesa_buffer_get_subdata;
    driver->UnmapBuffer = _mesa_buffer_unmap;
 
+   /* GL_ARB_clear_buffer_object */
+   driver->ClearBufferSubData = _mesa_buffer_clear_subdata;
+
    /* GL_ARB_map_buffer_range */
    driver->MapBufferRange = _mesa_buffer_map_range;
    driver->FlushMappedBufferRange = _mesa_buffer_flush_mapped_range;
@@ -1064,7 +1281,7 @@ _mesa_BufferData(GLenum target, GLsizeiptrARB size,
       return;
    }
 
-   bufObj = get_buffer(ctx, "glBufferDataARB", target);
+   bufObj = get_buffer(ctx, "glBufferDataARB", target, GL_INVALID_OPERATION);
    if (!bufObj)
       return;
 
@@ -1103,6 +1320,7 @@ _mesa_BufferSubData(GLenum target, GLintptrARB offset,
    struct gl_buffer_object *bufObj;
 
    bufObj = buffer_object_subdata_range_good( ctx, target, offset, size,
+                                              false, GL_INVALID_OPERATION,
                                               "glBufferSubDataARB" );
    if (!bufObj) {
       /* error already recorded */
@@ -1126,8 +1344,9 @@ _mesa_GetBufferSubData(GLenum target, GLintptrARB offset,
    GET_CURRENT_CONTEXT(ctx);
    struct gl_buffer_object *bufObj;
 
-   bufObj = buffer_object_subdata_range_good( ctx, target, offset, size,
-                                              "glGetBufferSubDataARB" );
+   bufObj = buffer_object_subdata_range_good(ctx, target, offset, size,
+                                             false, GL_INVALID_OPERATION,
+                                             "glGetBufferSubDataARB");
    if (!bufObj) {
       /* error already recorded */
       return;
@@ -1138,6 +1357,111 @@ _mesa_GetBufferSubData(GLenum target, GLintptrARB offset,
 }
 
 
+void GLAPIENTRY
+_mesa_ClearBufferData(GLenum target, GLenum internalformat, GLenum format,
+                      GLenum type, const GLvoid* data)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_buffer_object* bufObj;
+   gl_format mesaFormat;
+   GLubyte clearValue[MAX_PIXEL_BYTES];
+   GLsizeiptr clearValueSize;
+
+   bufObj = get_buffer(ctx, "glClearBufferData", target, GL_INVALID_VALUE);
+   if (!bufObj) {
+      return;
+   }
+
+   if (_mesa_bufferobj_mapped(bufObj)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glClearBufferData(buffer currently mapped)");
+      return;
+   }
+
+   mesaFormat = validate_clear_buffer_format(ctx, internalformat,
+                                             format, type,
+                                             "glClearBufferData");
+   if (mesaFormat == MESA_FORMAT_NONE) {
+      return;
+   }
+
+   clearValueSize = _mesa_get_format_bytes(mesaFormat);
+   if (bufObj->Size % clearValueSize != 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glClearBufferData(size is not a multiple of "
+                  "internalformat size)");
+      return;
+   }
+
+   if (data == NULL) {
+      /* clear to zeros, per the spec */
+      ctx->Driver.ClearBufferSubData(ctx, 0, bufObj->Size,
+                                     NULL, 0, bufObj);
+      return;
+   }
+
+   if (!convert_clear_buffer_data(ctx, mesaFormat, clearValue,
+                                  format, type, data, "glClearBufferData")) {
+      return;
+   }
+
+   ctx->Driver.ClearBufferSubData(ctx, 0, bufObj->Size,
+                                  clearValue, clearValueSize, bufObj);
+}
+
+
+void GLAPIENTRY
+_mesa_ClearBufferSubData(GLenum target, GLenum internalformat,
+                         GLintptr offset, GLsizeiptr size,
+                         GLenum format, GLenum type,
+                         const GLvoid* data)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_buffer_object* bufObj;
+   gl_format mesaFormat;
+   GLubyte clearValue[MAX_PIXEL_BYTES];
+   GLsizeiptr clearValueSize;
+
+   bufObj = buffer_object_subdata_range_good(ctx, target, offset, size,
+                                             true, GL_INVALID_VALUE,
+                                             "glClearBufferSubData");
+   if (!bufObj) {
+      return;
+   }
+
+   mesaFormat = validate_clear_buffer_format(ctx, internalformat,
+                                             format, type,
+                                             "glClearBufferSubData");
+   if (mesaFormat == MESA_FORMAT_NONE) {
+      return;
+   }
+
+   clearValueSize = _mesa_get_format_bytes(mesaFormat);
+   if (offset % clearValueSize != 0 || size % clearValueSize != 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glClearBufferSubData(offset or size is not a multiple of "
+                  "internalformat size)");
+      return;
+   }
+
+   if (data == NULL) {
+      /* clear to zeros, per the spec */
+      ctx->Driver.ClearBufferSubData(ctx, offset, size,
+                                     NULL, 0, bufObj);
+      return;
+   }
+
+   if (!convert_clear_buffer_data(ctx, mesaFormat, clearValue,
+                                  format, type, data,
+                                  "glClearBufferSubData")) {
+      return;
+   }
+
+   ctx->Driver.ClearBufferSubData(ctx, offset, size,
+                                  clearValue, clearValueSize, bufObj);
+}
+
+
 void * GLAPIENTRY
 _mesa_MapBuffer(GLenum target, GLenum access)
 {
@@ -1172,7 +1496,7 @@ _mesa_MapBuffer(GLenum target, GLenum access)
       return NULL;
    }
 
-   bufObj = get_buffer(ctx, "glMapBufferARB", target);
+   bufObj = get_buffer(ctx, "glMapBufferARB", target, GL_INVALID_OPERATION);
    if (!bufObj)
       return NULL;
 
@@ -1241,7 +1565,7 @@ _mesa_UnmapBuffer(GLenum target)
    GLboolean status = GL_TRUE;
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);
 
-   bufObj = get_buffer(ctx, "glUnmapBufferARB", target);
+   bufObj = get_buffer(ctx, "glUnmapBufferARB", target, GL_INVALID_OPERATION);
    if (!bufObj)
       return GL_FALSE;
 
@@ -1302,7 +1626,8 @@ _mesa_GetBufferParameteriv(GLenum target, GLenum pname, GLint *params)
    GET_CURRENT_CONTEXT(ctx);
    struct gl_buffer_object *bufObj;
 
-   bufObj = get_buffer(ctx, "glGetBufferParameterivARB", target);
+   bufObj = get_buffer(ctx, "glGetBufferParameterivARB", target,
+                       GL_INVALID_OPERATION);
    if (!bufObj)
       return;
 
@@ -1355,7 +1680,8 @@ _mesa_GetBufferParameteri64v(GLenum target, GLenum pname, GLint64 *params)
    GET_CURRENT_CONTEXT(ctx);
    struct gl_buffer_object *bufObj;
 
-   bufObj = get_buffer(ctx, "glGetBufferParameteri64v", target);
+   bufObj = get_buffer(ctx, "glGetBufferParameteri64v", target,
+                       GL_INVALID_OPERATION);
    if (!bufObj)
       return;
 
@@ -1408,7 +1734,8 @@ _mesa_GetBufferPointerv(GLenum target, GLenum pname, GLvoid **params)
       return;
    }
 
-   bufObj = get_buffer(ctx, "glGetBufferPointervARB", target);
+   bufObj = get_buffer(ctx, "glGetBufferPointervARB", target,
+                       GL_INVALID_OPERATION);
    if (!bufObj)
       return;
 
@@ -1424,11 +1751,13 @@ _mesa_CopyBufferSubData(GLenum readTarget, GLenum writeTarget,
    GET_CURRENT_CONTEXT(ctx);
    struct gl_buffer_object *src, *dst;
 
-   src = get_buffer(ctx, "glCopyBufferSubData", readTarget);
+   src = get_buffer(ctx, "glCopyBufferSubData", readTarget,
+                    GL_INVALID_OPERATION);
    if (!src)
       return;
 
-   dst = get_buffer(ctx, "glCopyBufferSubData", writeTarget);
+   dst = get_buffer(ctx, "glCopyBufferSubData", writeTarget,
+                    GL_INVALID_OPERATION);
    if (!dst)
       return;
 
@@ -1572,7 +1901,7 @@ _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length,
       return NULL;
    }
 
-   bufObj = get_buffer(ctx, "glMapBufferRange", target);
+   bufObj = get_buffer(ctx, "glMapBufferRange", target, GL_INVALID_OPERATION);
    if (!bufObj)
       return NULL;
 
@@ -1651,7 +1980,8 @@ _mesa_FlushMappedBufferRange(GLenum target, GLintptr offset, GLsizeiptr length)
       return;
    }
 
-   bufObj = get_buffer(ctx, "glFlushMappedBufferRange", target);
+   bufObj = get_buffer(ctx, "glFlushMappedBufferRange", target,
+                       GL_INVALID_OPERATION);
    if (!bufObj)
       return;
 
@@ -2331,23 +2661,11 @@ _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
     *     mapped by MapBuffer, or if the invalidate range intersects the range
     *     currently mapped by MapBufferRange."
     */
-   if (_mesa_bufferobj_mapped(bufObj)) {
-      const GLintptr mapEnd = bufObj->Offset + bufObj->Length;
-
-      /* The regions do not overlap if and only if the end of the discard
-       * region is before the mapped region or the start of the discard region
-       * is after the mapped region.
-       *
-       * Note that 'end' and 'mapEnd' are the first byte *after* the discard
-       * region and the mapped region, repsectively.  It is okay for that byte
-       * to be mapped (for 'end') or discarded (for 'mapEnd').
-       */
-      if (!(end <= bufObj->Offset || offset >= mapEnd)) {
-         _mesa_error(ctx, GL_INVALID_OPERATION,
-                     "glInvalidateBufferSubData(intersection with mapped "
-                     "range)");
-         return;
-      }
+   if (bufferobj_range_mapped(bufObj, offset, length)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glInvalidateBufferSubData(intersection with mapped "
+                  "range)");
+      return;
    }
 
    /* We don't actually do anything for this yet.  Just return after
diff --git a/mesalib/src/mesa/main/bufferobj.h b/mesalib/src/mesa/main/bufferobj.h
index 0b898a21b..71988b0d9 100644
--- a/mesalib/src/mesa/main/bufferobj.h
+++ b/mesalib/src/mesa/main/bufferobj.h
@@ -57,10 +57,10 @@ _mesa_is_bufferobj(const struct gl_buffer_object *obj)
 
 
 extern void
-_mesa_init_buffer_objects( struct gl_context *ctx );
+_mesa_init_buffer_objects(struct gl_context *ctx);
 
 extern void
-_mesa_free_buffer_objects( struct gl_context *ctx );
+_mesa_free_buffer_objects(struct gl_context *ctx);
 
 extern bool
 _mesa_handle_bind_buffer_gen(struct gl_context *ctx,
@@ -77,9 +77,9 @@ extern struct gl_buffer_object *
 _mesa_lookup_bufferobj(struct gl_context *ctx, GLuint buffer);
 
 extern void
-_mesa_initialize_buffer_object( struct gl_context *ctx,
-				struct gl_buffer_object *obj,
-				GLuint name, GLenum target );
+_mesa_initialize_buffer_object(struct gl_context *ctx,
+                               struct gl_buffer_object *obj,
+                               GLuint name, GLenum target);
 
 extern void
 _mesa_reference_buffer_object_(struct gl_context *ctx,
@@ -105,55 +105,90 @@ _mesa_init_buffer_object_functions(struct dd_function_table *driver);
 /*
  * API functions
  */
-
 void GLAPIENTRY
 _mesa_BindBuffer(GLenum target, GLuint buffer);
+
 void GLAPIENTRY
 _mesa_DeleteBuffers(GLsizei n, const GLuint * buffer);
+
 void GLAPIENTRY
 _mesa_GenBuffers(GLsizei n, GLuint * buffer);
+
 GLboolean GLAPIENTRY
 _mesa_IsBuffer(GLuint buffer);
+
 void GLAPIENTRY
-_mesa_BufferData(GLenum target, GLsizeiptrARB size, const GLvoid * data, GLenum usage);
+_mesa_BufferData(GLenum target, GLsizeiptrARB size,
+                 const GLvoid * data, GLenum usage);
+
 void GLAPIENTRY
-_mesa_BufferSubData(GLenum target, GLintptrARB offset, GLsizeiptrARB size, const GLvoid * data);
+_mesa_BufferSubData(GLenum target, GLintptrARB offset,
+                    GLsizeiptrARB size, const GLvoid * data);
+
 void GLAPIENTRY
-_mesa_GetBufferSubData(GLenum target, GLintptrARB offset, GLsizeiptrARB size, void * data);
+_mesa_GetBufferSubData(GLenum target, GLintptrARB offset,
+                       GLsizeiptrARB size, void * data);
+
+void GLAPIENTRY
+_mesa_ClearBufferData(GLenum target, GLenum internalformat,
+                      GLenum format, GLenum type,
+                      const GLvoid * data);
+
+void GLAPIENTRY
+_mesa_ClearBufferSubData(GLenum target, GLenum internalformat,
+                         GLintptr offset, GLsizeiptr size,
+                         GLenum format, GLenum type,
+                         const GLvoid * data);
+
 void * GLAPIENTRY
 _mesa_MapBuffer(GLenum target, GLenum access);
+
 GLboolean GLAPIENTRY
 _mesa_UnmapBuffer(GLenum target);
+
 void GLAPIENTRY
 _mesa_GetBufferParameteriv(GLenum target, GLenum pname, GLint *params);
+
 void GLAPIENTRY
 _mesa_GetBufferParameteri64v(GLenum target, GLenum pname, GLint64 *params);
+
 void GLAPIENTRY
 _mesa_GetBufferPointerv(GLenum target, GLenum pname, GLvoid **params);
+
 void GLAPIENTRY
 _mesa_CopyBufferSubData(GLenum readTarget, GLenum writeTarget,
                         GLintptr readOffset, GLintptr writeOffset,
                         GLsizeiptr size);
+
 void * GLAPIENTRY
 _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length,
                      GLbitfield access);
+
 void GLAPIENTRY
-_mesa_FlushMappedBufferRange(GLenum target, GLintptr offset, GLsizeiptr length);
+_mesa_FlushMappedBufferRange(GLenum target,
+                             GLintptr offset, GLsizeiptr length);
+
 GLenum GLAPIENTRY
 _mesa_ObjectPurgeableAPPLE(GLenum objectType, GLuint name, GLenum option);
+
 GLenum GLAPIENTRY
 _mesa_ObjectUnpurgeableAPPLE(GLenum objectType, GLuint name, GLenum option);
+
 void GLAPIENTRY
-_mesa_GetObjectParameterivAPPLE(GLenum objectType, GLuint name, GLenum pname,
-                                GLint* params);
+_mesa_GetObjectParameterivAPPLE(GLenum objectType, GLuint name,
+                                GLenum pname, GLint* params);
+
 void GLAPIENTRY
 _mesa_BindBufferRange(GLenum target, GLuint index,
                       GLuint buffer, GLintptr offset, GLsizeiptr size);
+
 void GLAPIENTRY
 _mesa_BindBufferBase(GLenum target, GLuint index, GLuint buffer);
+
 void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
                               GLsizeiptr length);
+
 void GLAPIENTRY
 _mesa_InvalidateBufferData(GLuint buffer);
 
diff --git a/mesalib/src/mesa/main/clear.c b/mesalib/src/mesa/main/clear.c
index 304d135d1..f0b525fa0 100644
--- a/mesalib/src/mesa/main/clear.c
+++ b/mesalib/src/mesa/main/clear.c
@@ -219,7 +219,25 @@ make_color_buffer_mask(struct gl_context *ctx, GLint drawbuffer)
    const struct gl_renderbuffer_attachment *att = ctx->DrawBuffer->Attachment;
    GLbitfield mask = 0x0;
 
-   switch (drawbuffer) {
+   /* From the GL 4.0 specification:
+    *	If buffer is COLOR, a particular draw buffer DRAW_BUFFERi is
+    *	specified by passing i as the parameter drawbuffer, and value
+    *	points to a four-element vector specifying the R, G, B, and A
+    *	color to clear that draw buffer to. If the draw buffer is one
+    *	of FRONT, BACK, LEFT, RIGHT, or FRONT_AND_BACK, identifying
+    *	multiple buffers, each selected buffer is cleared to the same
+    *	value.
+    *
+    * Note that "drawbuffer" and "draw buffer" have different meaning.
+    * "drawbuffer" specifies DRAW_BUFFERi, while "draw buffer" is what's
+    * assigned to DRAW_BUFFERi. It could be COLOR_ATTACHMENT0, FRONT, BACK,
+    * etc.
+    */
+   if (drawbuffer < 0 || drawbuffer >= (GLint)ctx->Const.MaxDrawBuffers) {
+      return INVALID_MASK;
+   }
+
+   switch (ctx->DrawBuffer->ColorDrawBuffer[drawbuffer]) {
    case GL_FRONT:
       if (att[BUFFER_FRONT_LEFT].Renderbuffer)
          mask |= BUFFER_BIT_FRONT_LEFT;
@@ -255,11 +273,12 @@ make_color_buffer_mask(struct gl_context *ctx, GLint drawbuffer)
          mask |= BUFFER_BIT_BACK_RIGHT;
       break;
    default:
-      if (drawbuffer < 0 || drawbuffer >= (GLint)ctx->Const.MaxDrawBuffers) {
-         mask = INVALID_MASK;
-      }
-      else if (att[BUFFER_COLOR0 + drawbuffer].Renderbuffer) {
-         mask |= (BUFFER_BIT_COLOR0 << drawbuffer);
+      {
+         GLuint buf = ctx->DrawBuffer->_ColorDrawBufferIndexes[drawbuffer];
+
+         if (buf >= 0 && att[buf].Renderbuffer) {
+            mask |= 1 << buf;
+         }
       }
    }
 
diff --git a/mesalib/src/mesa/main/colortab.c b/mesalib/src/mesa/main/colortab.c
index 81e92d71c..a8edb03dd 100644
--- a/mesalib/src/mesa/main/colortab.c
+++ b/mesalib/src/mesa/main/colortab.c
@@ -44,7 +44,7 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat,
                   const GLvoid *data )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glColorTable(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glColorTable");
 }
 
 
@@ -55,7 +55,7 @@ _mesa_ColorSubTable( GLenum target, GLsizei start,
                      const GLvoid *data )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glColorSubTable(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glColorSubTable");
 }
 
 
@@ -65,7 +65,7 @@ _mesa_CopyColorTable(GLenum target, GLenum internalformat,
                      GLint x, GLint y, GLsizei width)
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glCopyColorTable(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyColorTable");
 }
 
 
@@ -75,7 +75,7 @@ _mesa_CopyColorSubTable(GLenum target, GLsizei start,
                         GLint x, GLint y, GLsizei width)
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glCopyColorSubTable(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyColorSubTable");
 }
 
 
@@ -85,7 +85,7 @@ _mesa_GetnColorTableARB( GLenum target, GLenum format, GLenum type,
                          GLsizei bufSize, GLvoid *data )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetnColorTableARB(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetnColorTableARB");
 }
 
 
@@ -94,7 +94,7 @@ _mesa_GetColorTable( GLenum target, GLenum format,
                      GLenum type, GLvoid *data )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTable(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetColorTable");
 }
 
 
@@ -103,7 +103,7 @@ _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 {
    /* no extensions use this function */
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glColorTableParameterfv");
 }
 
 
@@ -113,7 +113,7 @@ _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params)
 {
    /* no extensions use this function */
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameteriv(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glColorTableParameteriv");
 }
 
 
@@ -122,7 +122,7 @@ void GLAPIENTRY
 _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameterfv(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetColorTableParameterfv");
 }
 
 
@@ -131,5 +131,5 @@ void GLAPIENTRY
 _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params )
 {
    GET_CURRENT_CONTEXT(ctx);
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameteriv(target)");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetColorTableParameteriv");
 }
diff --git a/mesalib/src/mesa/main/condrender.c b/mesalib/src/mesa/main/condrender.c
index 3d9b0eca1..2632f7a1a 100644
--- a/mesalib/src/mesa/main/condrender.c
+++ b/mesalib/src/mesa/main/condrender.c
@@ -72,7 +72,9 @@ _mesa_BeginConditionalRender(GLuint queryId, GLenum mode)
    }
    ASSERT(q->Id == queryId);
 
-   if (q->Target != GL_SAMPLES_PASSED || q->Active) {
+   if ((q->Target != GL_SAMPLES_PASSED &&
+        q->Target != GL_ANY_SAMPLES_PASSED &&
+        q->Target != GL_ANY_SAMPLES_PASSED_CONSERVATIVE) || q->Active) {
       _mesa_error(ctx, GL_INVALID_OPERATION, "glBeginConditionalRender()");
       return;
    }
diff --git a/mesalib/src/mesa/main/context.c b/mesalib/src/mesa/main/context.c
index 50c2f385a..4db27da9a 100644
--- a/mesalib/src/mesa/main/context.c
+++ b/mesalib/src/mesa/main/context.c
@@ -1194,6 +1194,7 @@ _mesa_free_context_data( struct gl_context *ctx )
    _mesa_free_sync_data(ctx);
    _mesa_free_varray_data(ctx);
    _mesa_free_transform_feedback(ctx);
+   _mesa_free_performance_monitors(ctx);
 
    _mesa_reference_buffer_object(ctx, &ctx->Pack.BufferObj, NULL);
    _mesa_reference_buffer_object(ctx, &ctx->Unpack.BufferObj, NULL);
diff --git a/mesalib/src/mesa/main/convolve.c b/mesalib/src/mesa/main/convolve.c
index f44031a9b..b13b89535 100644
--- a/mesalib/src/mesa/main/convolve.c
+++ b/mesalib/src/mesa/main/convolve.c
@@ -45,7 +45,7 @@ _mesa_ConvolutionFilter1D(GLenum target, GLenum internalFormat, GLsizei width, G
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionFilter1D");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionFilter1D");
 }
 
 void GLAPIENTRY
@@ -53,7 +53,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionFilter2D");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionFilter2D");
 }
 
 
@@ -62,7 +62,7 @@ _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionParameterf");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionParameterf");
 }
 
 
@@ -71,7 +71,7 @@ _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionParameterfv");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionParameterfv");
 }
 
 
@@ -80,7 +80,7 @@ _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionParameteri");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionParameteri");
 }
 
 
@@ -89,7 +89,7 @@ _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glConvolutionParameteriv");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glConvolutionParameteriv");
 }
 
 
@@ -98,7 +98,7 @@ _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLi
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glCopyConvolutionFilter1D");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyConvolutionFilter1D");
 }
 
 
@@ -107,7 +107,7 @@ _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLi
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glCopyConvolutionFilter2D");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyConvolutionFilter2D");
 }
 
 
@@ -134,7 +134,7 @@ _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetConvolutionParameterfv");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetConvolutionParameterfv");
 }
 
 
@@ -143,7 +143,7 @@ _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params)
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetConvolutionParameteriv");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetConvolutionParameteriv");
 }
 
 
@@ -155,7 +155,7 @@ _mesa_GetnSeparableFilterARB(GLenum target, GLenum format, GLenum type,
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glGetSeparableFilter");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glGetSeparableFilter");
 }
 
 
@@ -173,5 +173,5 @@ _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLs
 {
    GET_CURRENT_CONTEXT(ctx);
 
-   _mesa_error(ctx, GL_INVALID_ENUM, "glSeparableFilter2D");
+   _mesa_error(ctx, GL_INVALID_OPERATION, "glSeparableFilter2D");
 }
diff --git a/mesalib/src/mesa/main/dd.h b/mesalib/src/mesa/main/dd.h
index b5b874f47..6e73691ea 100644
--- a/mesalib/src/mesa/main/dd.h
+++ b/mesalib/src/mesa/main/dd.h
@@ -375,6 +375,11 @@ struct dd_function_table {
                                     GLsizei levels, GLsizei width,
                                     GLsizei height, GLsizei depth);
 
+   /** Called as part of glTextureView to add views to origTexObj */
+   GLboolean (*TextureView)(struct gl_context *ctx,
+                            struct gl_texture_object *texObj,
+                            struct gl_texture_object *origTexObj);
+
    /**
     * Map a renderbuffer into user space.
     * \param mode  bitmask of GL_MAP_READ_BIT, GL_MAP_WRITE_BIT and
@@ -569,6 +574,12 @@ struct dd_function_table {
 			     GLintptrARB offset, GLsizeiptrARB size,
 			     GLvoid *data, struct gl_buffer_object *obj );
 
+   void (*ClearBufferSubData)( struct gl_context *ctx,
+                               GLintptr offset, GLsizeiptr size,
+                               const GLvoid *clearValue,
+                               GLsizeiptr clearValueSize,
+                               struct gl_buffer_object *obj );
+
    void (*CopyBufferSubData)( struct gl_context *ctx,
                               struct gl_buffer_object *src,
                               struct gl_buffer_object *dst,
diff --git a/mesalib/src/mesa/main/dlist.c b/mesalib/src/mesa/main/dlist.c
index 595641915..cb40ff4db 100644
--- a/mesalib/src/mesa/main/dlist.c
+++ b/mesalib/src/mesa/main/dlist.c
@@ -209,18 +209,6 @@ typedef enum
    OPCODE_COLOR_MASK,
    OPCODE_COLOR_MASK_INDEXED,
    OPCODE_COLOR_MATERIAL,
-   OPCODE_COLOR_TABLE,
-   OPCODE_COLOR_TABLE_PARAMETER_FV,
-   OPCODE_COLOR_TABLE_PARAMETER_IV,
-   OPCODE_COLOR_SUB_TABLE,
-   OPCODE_CONVOLUTION_FILTER_1D,
-   OPCODE_CONVOLUTION_FILTER_2D,
-   OPCODE_CONVOLUTION_PARAMETER_I,
-   OPCODE_CONVOLUTION_PARAMETER_IV,
-   OPCODE_CONVOLUTION_PARAMETER_F,
-   OPCODE_CONVOLUTION_PARAMETER_FV,
-   OPCODE_COPY_COLOR_SUB_TABLE,
-   OPCODE_COPY_COLOR_TABLE,
    OPCODE_COPY_PIXELS,
    OPCODE_COPY_TEX_IMAGE1D,
    OPCODE_COPY_TEX_IMAGE2D,
@@ -243,7 +231,6 @@ typedef enum
    OPCODE_FRONT_FACE,
    OPCODE_FRUSTUM,
    OPCODE_HINT,
-   OPCODE_HISTOGRAM,
    OPCODE_INDEX_MASK,
    OPCODE_INIT_NAMES,
    OPCODE_LIGHT,
@@ -260,7 +247,6 @@ typedef enum
    OPCODE_MAPGRID1,
    OPCODE_MAPGRID2,
    OPCODE_MATRIX_MODE,
-   OPCODE_MIN_MAX,
    OPCODE_MULT_MATRIX,
    OPCODE_ORTHO,
    OPCODE_PASSTHROUGH,
@@ -281,8 +267,6 @@ typedef enum
    OPCODE_PUSH_NAME,
    OPCODE_RASTER_POS,
    OPCODE_READ_BUFFER,
-   OPCODE_RESET_HISTOGRAM,
-   OPCODE_RESET_MIN_MAX,
    OPCODE_ROTATE,
    OPCODE_SCALE,
    OPCODE_SCISSOR,
@@ -486,6 +470,10 @@ typedef enum
  * Each instruction in the display list is stored as a sequence of
  * contiguous nodes in memory.
  * Each node is the union of a variety of data types.
+ *
+ * Note, all of these members should be 4 bytes in size or less for the
+ * sake of compact display lists.  We store 8-byte pointers in a pair of
+ * these nodes using the save/get_pointer() functions below.
  */
 union gl_dlist_node
 {
@@ -500,14 +488,61 @@ union gl_dlist_node
    GLenum e;
    GLfloat f;
    GLsizei si;
-   GLvoid *data;
-   void *next;                  /* If prev node's opcode==OPCODE_CONTINUE */
 };
 
 
 typedef union gl_dlist_node Node;
 
 
+/** How many 4-byte dwords to store a pointer */
+#define POINTER_DWORDS (sizeof(void *) / 4)
+
+/* We want to keep sizeof(union gl_dlist_node) == 4 to minimize
+ * space for display lists.  The following types and functions are
+ * used to help store 4- and 8-byte pointers in 1 or 2 dlist_nodes.
+ */
+union pointer
+{
+   void *ptr;
+   GLuint dwords[POINTER_DWORDS];
+};
+
+
+/**
+ * Save a 4 or 8-byte pointer at dest (and dest+1).
+ */
+static inline void
+save_pointer(union gl_dlist_node *dest, void *src)
+{
+   union pointer p;
+   unsigned i;
+
+   STATIC_ASSERT(POINTER_DWORDS == 1 || POINTER_DWORDS == 2);
+   STATIC_ASSERT(sizeof(union gl_dlist_node) == 4);
+
+   p.ptr = src;
+
+   for (i = 0; i < POINTER_DWORDS; i++)
+      dest[i].ui = p.dwords[i];
+}
+
+
+/**
+ * Retrieve a 4 or 8-byte pointer from node (node+1).
+ */
+static inline void *
+get_pointer(const union gl_dlist_node *node)
+{
+   union pointer p;
+   unsigned i;
+
+   for (i = 0; i < POINTER_DWORDS; i++)
+      p.dwords[i] = node[i].ui;
+
+   return p.ptr;
+}
+
+
 /**
  * Used to store a 64-bit uint in a pair of "Nodes" for the sake of 32-bit
  * environment.  In 64-bit env, sizeof(Node)==8 anyway.
@@ -520,9 +555,9 @@ union uint64_pair
 
 
 /**
- * How many nodes to allocate at a time.
- *
- * \note Reduced now that we hold vertices etc. elsewhere.
+ * How many nodes to allocate at a time.  Note that bulk vertex data
+ * from glBegin/glVertex/glEnd primitives will typically wind up in
+ * a VBO, and not directly in the display list itself.
  */
 #define BLOCK_SIZE 256
 
@@ -538,14 +573,9 @@ static GLuint InstSize[OPCODE_END_OF_LIST + 1];
 void mesa_print_display_list(GLuint list);
 
 
-/**********************************************************************/
-/*****                           Private                          *****/
-/**********************************************************************/
-
-
 /**
- * Make an empty display list.  This is used by glGenLists() to
- * reserve display list IDs.
+ * Allocate a gl_display_list object with an initial block of storage.
+ * \param count  how many display list nodes/tokes to allocate
  */
 static struct gl_display_list *
 make_list(GLuint name, GLuint count)
@@ -637,91 +667,75 @@ _mesa_delete_list(struct gl_context *ctx, struct gl_display_list *dlist)
          switch (opcode) {
             /* for some commands, we need to free malloc'd memory */
          case OPCODE_MAP1:
-            free(n[6].data);
+            free(get_pointer(&n[6]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_MAP2:
-            free(n[10].data);
+            free(get_pointer(&n[10]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_DRAW_PIXELS:
-            free(n[5].data);
+            free(get_pointer(&n[5]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_BITMAP:
-            free(n[7].data);
-            n += InstSize[n[0].opcode];
-            break;
-         case OPCODE_COLOR_TABLE:
-            free(n[6].data);
-            n += InstSize[n[0].opcode];
-            break;
-         case OPCODE_COLOR_SUB_TABLE:
-            free(n[6].data);
-            n += InstSize[n[0].opcode];
-            break;
-         case OPCODE_CONVOLUTION_FILTER_1D:
-            free(n[6].data);
-            n += InstSize[n[0].opcode];
-            break;
-         case OPCODE_CONVOLUTION_FILTER_2D:
-            free(n[7].data);
+            free(get_pointer(&n[7]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_POLYGON_STIPPLE:
-            free(n[1].data);
+            free(get_pointer(&n[1]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_IMAGE1D:
-            free(n[8].data);
+            free(get_pointer(&n[8]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_IMAGE2D:
-            free(n[9].data);
+            free(get_pointer(&n[9]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_IMAGE3D:
-            free(n[10].data);
+            free(get_pointer(&n[10]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_SUB_IMAGE1D:
-            free(n[7].data);
+            free(get_pointer(&n[7]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_SUB_IMAGE2D:
-            free(n[9].data);
+            free(get_pointer(&n[9]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_TEX_SUB_IMAGE3D:
-            free(n[11].data);
+            free(get_pointer(&n[11]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_1D:
-            free(n[7].data);
+            free(get_pointer(&n[7]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_2D:
-            free(n[8].data);
+            free(get_pointer(&n[8]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_3D:
-            free(n[9].data);
+            free(get_pointer(&n[9]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D:
-            free(n[7].data);
+            free(get_pointer(&n[7]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D:
-            free(n[9].data);
+            free(get_pointer(&n[9]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D:
-            free(n[11].data);
+            free(get_pointer(&n[11]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_PROGRAM_STRING_ARB:
-            free(n[4].data);      /* program string */
+            free(get_pointer(&n[4]));      /* program string */
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_UNIFORM_1FV:
@@ -736,7 +750,7 @@ _mesa_delete_list(struct gl_context *ctx, struct gl_display_list *dlist)
          case OPCODE_UNIFORM_2UIV:
          case OPCODE_UNIFORM_3UIV:
          case OPCODE_UNIFORM_4UIV:
-            free(n[3].data);
+            free(get_pointer(&n[3]));
             n += InstSize[n[0].opcode];
             break;
          case OPCODE_UNIFORM_MATRIX22:
@@ -748,12 +762,15 @@ _mesa_delete_list(struct gl_context *ctx, struct gl_display_list *dlist)
          case OPCODE_UNIFORM_MATRIX32:
          case OPCODE_UNIFORM_MATRIX34:
          case OPCODE_UNIFORM_MATRIX43:
-            free(n[4].data);
+            free(get_pointer(&n[4]));
             n += InstSize[n[0].opcode];
             break;
+         case OPCODE_PIXEL_MAP:
+            free(get_pointer(&n[3]));
+            break;
 
          case OPCODE_CONTINUE:
-            n = (Node *) n[1].next;
+            n = (Node *) get_pointer(&n[1]);
             free(block);
             block = n;
             break;
@@ -852,12 +869,6 @@ translate_id(GLsizei n, GLenum type, const GLvoid * list)
 }
 
 
-
-
-/**********************************************************************/
-/*****                        Public                              *****/
-/**********************************************************************/
-
 /**
  * Wrapper for _mesa_unpack_image/bitmap() that handles pixel buffer objects.
  * If width < 0 or height < 0 or format or type are invalid we'll just
@@ -929,6 +940,18 @@ unpack_image(struct gl_context *ctx, GLuint dimensions,
    return NULL;
 }
 
+
+/** Return copy of memory */
+static void *
+memdup(const void *src, GLsizei bytes)
+{
+   void *b = bytes >= 0 ? malloc(bytes) : NULL;
+   if (b)
+      memcpy(b, src, bytes);
+   return b;
+}
+
+
 /**
  * Allocate space for a display list instruction (opcode + payload space).
  * \param opcode  the instruction opcode (OPCODE_* value)
@@ -939,6 +962,7 @@ static Node *
 dlist_alloc(struct gl_context *ctx, OpCode opcode, GLuint bytes)
 {
    const GLuint numNodes = 1 + (bytes + sizeof(Node) - 1) / sizeof(Node);
+   const GLuint contNodes = 1 + POINTER_DWORDS;  /* size of continue info */
    Node *n;
 
    if (opcode < (GLuint) OPCODE_EXT_0) {
@@ -952,7 +976,7 @@ dlist_alloc(struct gl_context *ctx, OpCode opcode, GLuint bytes)
       }
    }
 
-   if (ctx->ListState.CurrentPos + numNodes + 2 > BLOCK_SIZE) {
+   if (ctx->ListState.CurrentPos + numNodes + contNodes > BLOCK_SIZE) {
       /* This block is full.  Allocate a new block and chain to it */
       Node *newblock;
       n = ctx->ListState.CurrentBlock + ctx->ListState.CurrentPos;
@@ -962,7 +986,7 @@ dlist_alloc(struct gl_context *ctx, OpCode opcode, GLuint bytes)
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "Building display list");
          return NULL;
       }
-      n[1].next = (Node *) newblock;
+      save_pointer(&n[1], newblock);
       ctx->ListState.CurrentBlock = newblock;
       ctx->ListState.CurrentPos = 0;
    }
@@ -1043,6 +1067,37 @@ alloc_instruction(struct gl_context *ctx, OpCode opcode, GLuint nparams)
 }
 
 
+/**
+ * Called by EndList to try to reduce memory used for the list.
+ */
+static void
+trim_list(struct gl_context *ctx)
+{
+   /* If the list we're ending only has one allocated block of nodes/tokens
+    * and its size isn't a full block size, realloc the block to use less
+    * memory.  This is important for apps that create many small display
+    * lists and apps that use glXUseXFont (many lists each containing one
+    * glBitmap call).
+    * Note: we currently only trim display lists that allocated one block
+    * of tokens.  That hits the short list case which is what we're mainly
+    * concerned with.  Trimming longer lists would involve traversing the
+    * linked list of blocks.
+    */
+   struct gl_dlist_state *list = &ctx->ListState;
+
+   if ((list->CurrentList->Head == list->CurrentBlock) &&
+       (list->CurrentPos < BLOCK_SIZE)) {
+      /* There's only one block and it's not full, so realloc */
+      GLuint newSize = list->CurrentPos * sizeof(Node);
+      list->CurrentList->Head =
+      list->CurrentBlock = realloc(list->CurrentBlock, newSize);
+      if (!list->CurrentBlock) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glEndList");
+      }
+   }
+}
+
+
 
 /*
  * Display List compilation functions
@@ -1106,7 +1161,7 @@ save_Bitmap(GLsizei width, GLsizei height,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_BITMAP, 7);
+   n = alloc_instruction(ctx, OPCODE_BITMAP, 6 + POINTER_DWORDS);
    if (n) {
       n[1].i = (GLint) width;
       n[2].i = (GLint) height;
@@ -1114,8 +1169,9 @@ save_Bitmap(GLsizei width, GLsizei height,
       n[4].f = yorig;
       n[5].f = xmove;
       n[6].f = ymove;
-      n[7].data = unpack_image(ctx, 2, width, height, 1, GL_COLOR_INDEX,
-                               GL_BITMAP, pixels, &ctx->Unpack);
+      save_pointer(&n[7],
+                   unpack_image(ctx, 2, width, height, 1, GL_COLOR_INDEX,
+                                GL_BITMAP, pixels, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_Bitmap(ctx->Exec, (width, height,
@@ -1735,313 +1791,6 @@ save_ColorMaterial(GLenum face, GLenum mode)
 
 
 static void GLAPIENTRY
-save_ColorTable(GLenum target, GLenum internalFormat,
-                GLsizei width, GLenum format, GLenum type,
-                const GLvoid * table)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   if (_mesa_is_proxy_texture(target)) {
-      /* execute immediately */
-      CALL_ColorTable(ctx->Exec, (target, internalFormat, width,
-                                  format, type, table));
-   }
-   else {
-      Node *n;
-      ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = alloc_instruction(ctx, OPCODE_COLOR_TABLE, 6);
-      if (n) {
-         n[1].e = target;
-         n[2].e = internalFormat;
-         n[3].i = width;
-         n[4].e = format;
-         n[5].e = type;
-         n[6].data = unpack_image(ctx, 1, width, 1, 1, format, type, table,
-                                  &ctx->Unpack);
-      }
-      if (ctx->ExecuteFlag) {
-         CALL_ColorTable(ctx->Exec, (target, internalFormat, width,
-                                     format, type, table));
-      }
-   }
-}
-
-
-
-static void GLAPIENTRY
-save_ColorTableParameterfv(GLenum target, GLenum pname,
-                           const GLfloat *params)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-
-   n = alloc_instruction(ctx, OPCODE_COLOR_TABLE_PARAMETER_FV, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].f = params[0];
-      if (pname == GL_COLOR_TABLE_SGI ||
-          pname == GL_POST_CONVOLUTION_COLOR_TABLE_SGI ||
-          pname == GL_TEXTURE_COLOR_TABLE_SGI) {
-         n[4].f = params[1];
-         n[5].f = params[2];
-         n[6].f = params[3];
-      }
-   }
-
-   if (ctx->ExecuteFlag) {
-      CALL_ColorTableParameterfv(ctx->Exec, (target, pname, params));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-
-   n = alloc_instruction(ctx, OPCODE_COLOR_TABLE_PARAMETER_IV, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].i = params[0];
-      if (pname == GL_COLOR_TABLE_SGI ||
-          pname == GL_POST_CONVOLUTION_COLOR_TABLE_SGI ||
-          pname == GL_TEXTURE_COLOR_TABLE_SGI) {
-         n[4].i = params[1];
-         n[5].i = params[2];
-         n[6].i = params[3];
-      }
-   }
-
-   if (ctx->ExecuteFlag) {
-      CALL_ColorTableParameteriv(ctx->Exec, (target, pname, params));
-   }
-}
-
-
-
-static void GLAPIENTRY
-save_ColorSubTable(GLenum target, GLsizei start, GLsizei count,
-                   GLenum format, GLenum type, const GLvoid * table)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_COLOR_SUB_TABLE, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].i = start;
-      n[3].i = count;
-      n[4].e = format;
-      n[5].e = type;
-      n[6].data = unpack_image(ctx, 1, count, 1, 1, format, type, table,
-                               &ctx->Unpack);
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ColorSubTable(ctx->Exec,
-                         (target, start, count, format, type, table));
-   }
-}
-
-
-static void GLAPIENTRY
-save_CopyColorSubTable(GLenum target, GLsizei start,
-                       GLint x, GLint y, GLsizei width)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_COPY_COLOR_SUB_TABLE, 5);
-   if (n) {
-      n[1].e = target;
-      n[2].i = start;
-      n[3].i = x;
-      n[4].i = y;
-      n[5].i = width;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_CopyColorSubTable(ctx->Exec, (target, start, x, y, width));
-   }
-}
-
-
-static void GLAPIENTRY
-save_CopyColorTable(GLenum target, GLenum internalformat,
-                    GLint x, GLint y, GLsizei width)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_COPY_COLOR_TABLE, 5);
-   if (n) {
-      n[1].e = target;
-      n[2].e = internalformat;
-      n[3].i = x;
-      n[4].i = y;
-      n[5].i = width;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_CopyColorTable(ctx->Exec, (target, internalformat, x, y, width));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionFilter1D(GLenum target, GLenum internalFormat, GLsizei width,
-                         GLenum format, GLenum type, const GLvoid * filter)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_FILTER_1D, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].e = internalFormat;
-      n[3].i = width;
-      n[4].e = format;
-      n[5].e = type;
-      n[6].data = unpack_image(ctx, 1, width, 1, 1, format, type, filter,
-                               &ctx->Unpack);
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionFilter1D(ctx->Exec, (target, internalFormat, width,
-                                           format, type, filter));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionFilter2D(GLenum target, GLenum internalFormat,
-                         GLsizei width, GLsizei height, GLenum format,
-                         GLenum type, const GLvoid * filter)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_FILTER_2D, 7);
-   if (n) {
-      n[1].e = target;
-      n[2].e = internalFormat;
-      n[3].i = width;
-      n[4].i = height;
-      n[5].e = format;
-      n[6].e = type;
-      n[7].data = unpack_image(ctx, 2, width, height, 1, format, type, filter,
-                               &ctx->Unpack);
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionFilter2D(ctx->Exec,
-                               (target, internalFormat, width, height, format,
-                                type, filter));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionParameteri(GLenum target, GLenum pname, GLint param)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_I, 3);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].i = param;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionParameteri(ctx->Exec, (target, pname, param));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_IV, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].i = params[0];
-      if (pname == GL_CONVOLUTION_BORDER_COLOR ||
-          pname == GL_CONVOLUTION_FILTER_SCALE ||
-          pname == GL_CONVOLUTION_FILTER_BIAS) {
-         n[4].i = params[1];
-         n[5].i = params[2];
-         n[6].i = params[3];
-      }
-      else {
-         n[4].i = n[5].i = n[6].i = 0;
-      }
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionParameteriv(ctx->Exec, (target, pname, params));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_F, 3);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].f = param;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionParameterf(ctx->Exec, (target, pname, param));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ConvolutionParameterfv(GLenum target, GLenum pname,
-                            const GLfloat *params)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_CONVOLUTION_PARAMETER_FV, 6);
-   if (n) {
-      n[1].e = target;
-      n[2].e = pname;
-      n[3].f = params[0];
-      if (pname == GL_CONVOLUTION_BORDER_COLOR ||
-          pname == GL_CONVOLUTION_FILTER_SCALE ||
-          pname == GL_CONVOLUTION_FILTER_BIAS) {
-         n[4].f = params[1];
-         n[5].f = params[2];
-         n[6].f = params[3];
-      }
-      else {
-         n[4].f = n[5].f = n[6].f = 0.0F;
-      }
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ConvolutionParameterfv(ctx->Exec, (target, pname, params));
-   }
-}
-
-
-static void GLAPIENTRY
 save_CopyPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum type)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -2314,14 +2063,15 @@ save_DrawPixels(GLsizei width, GLsizei height,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_DRAW_PIXELS, 5);
+   n = alloc_instruction(ctx, OPCODE_DRAW_PIXELS, 4 + POINTER_DWORDS);
    if (n) {
       n[1].i = width;
       n[2].i = height;
       n[3].e = format;
       n[4].e = type;
-      n[5].data = unpack_image(ctx, 2, width, height, 1, format, type,
-                               pixels, &ctx->Unpack);
+      save_pointer(&n[5],
+                   unpack_image(ctx, 2, width, height, 1, format, type,
+                                pixels, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_DrawPixels(ctx->Exec, (width, height, format, type, pixels));
@@ -2530,27 +2280,6 @@ save_Hint(GLenum target, GLenum mode)
 
 
 static void GLAPIENTRY
-save_Histogram(GLenum target, GLsizei width, GLenum internalFormat,
-               GLboolean sink)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_HISTOGRAM, 4);
-   if (n) {
-      n[1].e = target;
-      n[2].i = width;
-      n[3].e = internalFormat;
-      n[4].b = sink;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_Histogram(ctx->Exec, (target, width, internalFormat, sink));
-   }
-}
-
-
-static void GLAPIENTRY
 save_IndexMask(GLuint mask)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -2890,7 +2619,7 @@ save_Map1d(GLenum target, GLdouble u1, GLdouble u2, GLint stride,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_MAP1, 6);
+   n = alloc_instruction(ctx, OPCODE_MAP1, 5 + POINTER_DWORDS);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points1d(target, stride, order, points);
       n[1].e = target;
@@ -2898,7 +2627,7 @@ save_Map1d(GLenum target, GLdouble u1, GLdouble u2, GLint stride,
       n[3].f = (GLfloat) u2;
       n[4].i = _mesa_evaluator_components(target);      /* stride */
       n[5].i = order;
-      n[6].data = (void *) pnts;
+      save_pointer(&n[6], pnts);
    }
    if (ctx->ExecuteFlag) {
       CALL_Map1d(ctx->Exec, (target, u1, u2, stride, order, points));
@@ -2912,7 +2641,7 @@ save_Map1f(GLenum target, GLfloat u1, GLfloat u2, GLint stride,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_MAP1, 6);
+   n = alloc_instruction(ctx, OPCODE_MAP1, 5 + POINTER_DWORDS);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points1f(target, stride, order, points);
       n[1].e = target;
@@ -2920,7 +2649,7 @@ save_Map1f(GLenum target, GLfloat u1, GLfloat u2, GLint stride,
       n[3].f = u2;
       n[4].i = _mesa_evaluator_components(target);      /* stride */
       n[5].i = order;
-      n[6].data = (void *) pnts;
+      save_pointer(&n[6], pnts);
    }
    if (ctx->ExecuteFlag) {
       CALL_Map1f(ctx->Exec, (target, u1, u2, stride, order, points));
@@ -2937,7 +2666,7 @@ save_Map2d(GLenum target,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_MAP2, 10);
+   n = alloc_instruction(ctx, OPCODE_MAP2, 9 + POINTER_DWORDS);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points2d(target, ustride, uorder,
                                               vstride, vorder, points);
@@ -2951,7 +2680,7 @@ save_Map2d(GLenum target,
       n[7].i = _mesa_evaluator_components(target);      /*vstride */
       n[8].i = uorder;
       n[9].i = vorder;
-      n[10].data = (void *) pnts;
+      save_pointer(&n[10], pnts);
    }
    if (ctx->ExecuteFlag) {
       CALL_Map2d(ctx->Exec, (target,
@@ -2970,7 +2699,7 @@ save_Map2f(GLenum target,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_MAP2, 10);
+   n = alloc_instruction(ctx, OPCODE_MAP2, 9 + POINTER_DWORDS);
    if (n) {
       GLfloat *pnts = _mesa_copy_map_points2f(target, ustride, uorder,
                                               vstride, vorder, points);
@@ -2984,7 +2713,7 @@ save_Map2f(GLenum target,
       n[7].i = _mesa_evaluator_components(target);      /*vstride */
       n[8].i = uorder;
       n[9].i = vorder;
-      n[10].data = (void *) pnts;
+      save_pointer(&n[10], pnts);
    }
    if (ctx->ExecuteFlag) {
       CALL_Map2f(ctx->Exec, (target, u1, u2, ustride, uorder,
@@ -3067,25 +2796,6 @@ save_MatrixMode(GLenum mode)
 
 
 static void GLAPIENTRY
-save_Minmax(GLenum target, GLenum internalFormat, GLboolean sink)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_MIN_MAX, 3);
-   if (n) {
-      n[1].e = target;
-      n[2].e = internalFormat;
-      n[3].b = sink;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_Minmax(ctx->Exec, (target, internalFormat, sink));
-   }
-}
-
-
-static void GLAPIENTRY
 save_MultMatrixf(const GLfloat * m)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -3156,12 +2866,11 @@ save_PixelMapfv(GLenum map, GLint mapsize, const GLfloat *values)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_PIXEL_MAP, 3);
+   n = alloc_instruction(ctx, OPCODE_PIXEL_MAP, 2 + POINTER_DWORDS);
    if (n) {
       n[1].e = map;
       n[2].i = mapsize;
-      n[3].data = malloc(mapsize * sizeof(GLfloat));
-      memcpy(n[3].data, (void *) values, mapsize * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(values, mapsize * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_PixelMapfv(ctx->Exec, (map, mapsize, values));
@@ -3336,10 +3045,11 @@ save_PolygonStipple(const GLubyte * pattern)
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_POLYGON_STIPPLE, 1);
+   n = alloc_instruction(ctx, OPCODE_POLYGON_STIPPLE, POINTER_DWORDS);
    if (n) {
-      n[1].data = unpack_image(ctx, 2, 32, 32, 1, GL_COLOR_INDEX, GL_BITMAP,
-                               pattern, &ctx->Unpack);
+      save_pointer(&n[1],
+                   unpack_image(ctx, 2, 32, 32, 1, GL_COLOR_INDEX, GL_BITMAP,
+                                pattern, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_PolygonStipple(ctx->Exec, ((GLubyte *) pattern));
@@ -3667,38 +3377,6 @@ save_ReadBuffer(GLenum mode)
 
 
 static void GLAPIENTRY
-save_ResetHistogram(GLenum target)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_RESET_HISTOGRAM, 1);
-   if (n) {
-      n[1].e = target;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ResetHistogram(ctx->Exec, (target));
-   }
-}
-
-
-static void GLAPIENTRY
-save_ResetMinmax(GLenum target)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   Node *n;
-   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_RESET_MIN_MAX, 1);
-   if (n) {
-      n[1].e = target;
-   }
-   if (ctx->ExecuteFlag) {
-      CALL_ResetMinmax(ctx->Exec, (target));
-   }
-}
-
-
-static void GLAPIENTRY
 save_Rotatef(GLfloat angle, GLfloat x, GLfloat y, GLfloat z)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -4140,7 +3818,7 @@ save_TexImage1D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE1D, 8);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE1D, 7 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4149,8 +3827,9 @@ save_TexImage1D(GLenum target,
          n[5].i = border;
          n[6].e = format;
          n[7].e = type;
-         n[8].data = unpack_image(ctx, 1, width, 1, 1, format, type,
-                                  pixels, &ctx->Unpack);
+         save_pointer(&n[8],
+                      unpack_image(ctx, 1, width, 1, 1, format, type,
+                                   pixels, &ctx->Unpack));
       }
       if (ctx->ExecuteFlag) {
          CALL_TexImage1D(ctx->Exec, (target, level, components, width,
@@ -4175,7 +3854,7 @@ save_TexImage2D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE2D, 9);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE2D, 8 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4185,8 +3864,9 @@ save_TexImage2D(GLenum target,
          n[6].i = border;
          n[7].e = format;
          n[8].e = type;
-         n[9].data = unpack_image(ctx, 2, width, height, 1, format, type,
-                                  pixels, &ctx->Unpack);
+         save_pointer(&n[9],
+                      unpack_image(ctx, 2, width, height, 1, format, type,
+                                   pixels, &ctx->Unpack));
       }
       if (ctx->ExecuteFlag) {
          CALL_TexImage2D(ctx->Exec, (target, level, components, width,
@@ -4213,7 +3893,7 @@ save_TexImage3D(GLenum target,
    else {
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE3D, 10);
+      n = alloc_instruction(ctx, OPCODE_TEX_IMAGE3D, 9 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4224,8 +3904,9 @@ save_TexImage3D(GLenum target,
          n[7].i = border;
          n[8].e = format;
          n[9].e = type;
-         n[10].data = unpack_image(ctx, 3, width, height, depth, format, type,
-                                   pixels, &ctx->Unpack);
+         save_pointer(&n[10],
+                      unpack_image(ctx, 3, width, height, depth, format, type,
+                                   pixels, &ctx->Unpack));
       }
       if (ctx->ExecuteFlag) {
          CALL_TexImage3D(ctx->Exec, (target, level, internalFormat, width,
@@ -4246,7 +3927,7 @@ save_TexSubImage1D(GLenum target, GLint level, GLint xoffset,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE1D, 7);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE1D, 6 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4254,8 +3935,9 @@ save_TexSubImage1D(GLenum target, GLint level, GLint xoffset,
       n[4].i = (GLint) width;
       n[5].e = format;
       n[6].e = type;
-      n[7].data = unpack_image(ctx, 1, width, 1, 1, format, type,
-                               pixels, &ctx->Unpack);
+      save_pointer(&n[7],
+                   unpack_image(ctx, 1, width, 1, 1, format, type,
+                                pixels, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_TexSubImage1D(ctx->Exec, (target, level, xoffset, width,
@@ -4275,7 +3957,7 @@ save_TexSubImage2D(GLenum target, GLint level,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE2D, 9);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE2D, 8 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4285,8 +3967,9 @@ save_TexSubImage2D(GLenum target, GLint level,
       n[6].i = (GLint) height;
       n[7].e = format;
       n[8].e = type;
-      n[9].data = unpack_image(ctx, 2, width, height, 1, format, type,
-                               pixels, &ctx->Unpack);
+      save_pointer(&n[9],
+                   unpack_image(ctx, 2, width, height, 1, format, type,
+                                pixels, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_TexSubImage2D(ctx->Exec, (target, level, xoffset, yoffset,
@@ -4306,7 +3989,7 @@ save_TexSubImage3D(GLenum target, GLint level,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE3D, 11);
+   n = alloc_instruction(ctx, OPCODE_TEX_SUB_IMAGE3D, 10 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4318,8 +4001,9 @@ save_TexSubImage3D(GLenum target, GLint level,
       n[8].i = (GLint) depth;
       n[9].e = format;
       n[10].e = type;
-      n[11].data = unpack_image(ctx, 3, width, height, depth, format, type,
-                                pixels, &ctx->Unpack);
+      save_pointer(&n[11],
+                   unpack_image(ctx, 3, width, height, depth, format, type,
+                                pixels, &ctx->Unpack));
    }
    if (ctx->ExecuteFlag) {
       CALL_TexSubImage3D(ctx->Exec, (target, level,
@@ -4626,7 +4310,8 @@ save_CompressedTexImage1DARB(GLenum target, GLint level,
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_1D, 7);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_1D,
+                            6 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4634,7 +4319,8 @@ save_CompressedTexImage1DARB(GLenum target, GLint level,
          n[4].i = (GLint) width;
          n[5].i = border;
          n[6].i = imageSize;
-         n[7].data = copy_data(data, imageSize, "glCompressedTexImage1DARB");
+         save_pointer(&n[7],
+                      copy_data(data, imageSize, "glCompressedTexImage1DARB"));
       }
       if (ctx->ExecuteFlag) {
          CALL_CompressedTexImage1D(ctx->Exec,
@@ -4662,7 +4348,8 @@ save_CompressedTexImage2DARB(GLenum target, GLint level,
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_2D, 8);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_2D,
+                            7 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4671,7 +4358,8 @@ save_CompressedTexImage2DARB(GLenum target, GLint level,
          n[5].i = (GLint) height;
          n[6].i = border;
          n[7].i = imageSize;
-         n[8].data = copy_data(data, imageSize, "glCompressedTexImage2DARB");
+         save_pointer(&n[8],
+                      copy_data(data, imageSize, "glCompressedTexImage2DARB"));
       }
       if (ctx->ExecuteFlag) {
          CALL_CompressedTexImage2D(ctx->Exec,
@@ -4699,7 +4387,8 @@ save_CompressedTexImage3DARB(GLenum target, GLint level,
       Node *n;
       ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_3D, 9);
+      n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_IMAGE_3D,
+                            8 + POINTER_DWORDS);
       if (n) {
          n[1].e = target;
          n[2].i = level;
@@ -4709,7 +4398,8 @@ save_CompressedTexImage3DARB(GLenum target, GLint level,
          n[6].i = (GLint) depth;
          n[7].i = border;
          n[8].i = imageSize;
-         n[9].data = copy_data(data, imageSize, "glCompressedTexImage3DARB");
+         save_pointer(&n[9],
+                      copy_data(data, imageSize, "glCompressedTexImage3DARB"));
       }
       if (ctx->ExecuteFlag) {
          CALL_CompressedTexImage3D(ctx->Exec,
@@ -4730,7 +4420,8 @@ save_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset,
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D, 7);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D,
+                         6 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4738,7 +4429,8 @@ save_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset,
       n[4].i = (GLint) width;
       n[5].e = format;
       n[6].i = imageSize;
-      n[7].data = copy_data(data, imageSize, "glCompressedTexSubImage1DARB");
+      save_pointer(&n[7],
+                   copy_data(data, imageSize, "glCompressedTexSubImage1DARB"));
    }
    if (ctx->ExecuteFlag) {
       CALL_CompressedTexSubImage1D(ctx->Exec, (target, level, xoffset,
@@ -4758,7 +4450,8 @@ save_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset,
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D, 9);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D,
+                         8 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4768,7 +4461,8 @@ save_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset,
       n[6].i = (GLint) height;
       n[7].e = format;
       n[8].i = imageSize;
-      n[9].data = copy_data(data, imageSize, "glCompressedTexSubImage2DARB");
+      save_pointer(&n[9],
+                   copy_data(data, imageSize, "glCompressedTexSubImage2DARB"));
    }
    if (ctx->ExecuteFlag) {
       CALL_CompressedTexSubImage2D(ctx->Exec,
@@ -4788,7 +4482,8 @@ save_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset,
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D, 11);
+   n = alloc_instruction(ctx, OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D,
+                         10 + POINTER_DWORDS);
    if (n) {
       n[1].e = target;
       n[2].i = level;
@@ -4800,7 +4495,8 @@ save_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset,
       n[8].i = (GLint) depth;
       n[9].e = format;
       n[10].i = imageSize;
-      n[11].data = copy_data(data, imageSize, "glCompressedTexSubImage3DARB");
+      save_pointer(&n[11],
+                   copy_data(data, imageSize, "glCompressedTexSubImage3DARB"));
    }
    if (ctx->ExecuteFlag) {
       CALL_CompressedTexSubImage3D(ctx->Exec,
@@ -5098,7 +4794,7 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len,
 
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
 
-   n = alloc_instruction(ctx, OPCODE_PROGRAM_STRING_ARB, 4);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_STRING_ARB, 3 + POINTER_DWORDS);
    if (n) {
       GLubyte *programCopy = malloc(len);
       if (!programCopy) {
@@ -5109,7 +4805,7 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len,
       n[1].e = target;
       n[2].e = format;
       n[3].i = len;
-      n[4].data = programCopy;
+      save_pointer(&n[4], programCopy);
    }
    if (ctx->ExecuteFlag) {
       CALL_ProgramStringARB(ctx->Exec, (target, format, len, string));
@@ -6284,28 +5980,17 @@ save_Uniform4fARB(GLint location, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 }
 
 
-/** Return copy of memory */
-static void *
-memdup(const void *src, GLsizei bytes)
-{
-   void *b = bytes >= 0 ? malloc(bytes) : NULL;
-   if (b)
-      memcpy(b, src, bytes);
-   return b;
-}
-
-
 static void GLAPIENTRY
 save_Uniform1fvARB(GLint location, GLsizei count, const GLfloat *v)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_1FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1FV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 1 * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(v, count * 1 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform1fv(ctx->Exec, (location, count, v));
@@ -6318,11 +6003,11 @@ save_Uniform2fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_2FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2FV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 2 * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(v, count * 2 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform2fv(ctx->Exec, (location, count, v));
@@ -6335,11 +6020,11 @@ save_Uniform3fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_3FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3FV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 3 * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(v, count * 3 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform3fv(ctx->Exec, (location, count, v));
@@ -6352,11 +6037,11 @@ save_Uniform4fvARB(GLint location, GLsizei count, const GLfloat *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_4FV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4FV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 4 * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(v, count * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform4fv(ctx->Exec, (location, count, v));
@@ -6442,11 +6127,11 @@ save_Uniform1ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_1IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1IV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 1 * sizeof(GLint));
+      save_pointer(&n[3], memdup(v, count * 1 * sizeof(GLint)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform1iv(ctx->Exec, (location, count, v));
@@ -6459,11 +6144,11 @@ save_Uniform2ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_2IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2IV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 2 * sizeof(GLint));
+      save_pointer(&n[3], memdup(v, count * 2 * sizeof(GLint)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform2iv(ctx->Exec, (location, count, v));
@@ -6476,11 +6161,11 @@ save_Uniform3ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_3IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3IV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 3 * sizeof(GLint));
+      save_pointer(&n[3], memdup(v, count * 3 * sizeof(GLint)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform3iv(ctx->Exec, (location, count, v));
@@ -6493,11 +6178,11 @@ save_Uniform4ivARB(GLint location, GLsizei count, const GLint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_4IV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4IV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 4 * sizeof(GLfloat));
+      save_pointer(&n[3], memdup(v, count * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_Uniform4iv(ctx->Exec, (location, count, v));
@@ -6584,11 +6269,11 @@ save_Uniform1uiv(GLint location, GLsizei count, const GLuint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_1UIV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_1UIV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 1 * sizeof(*v));
+      save_pointer(&n[3], memdup(v, count * 1 * sizeof(*v)));
    }
    if (ctx->ExecuteFlag) {
       /*CALL_Uniform1uiv(ctx->Exec, (location, count, v));*/
@@ -6601,11 +6286,11 @@ save_Uniform2uiv(GLint location, GLsizei count, const GLuint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_2UIV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_2UIV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 2 * sizeof(*v));
+      save_pointer(&n[3], memdup(v, count * 2 * sizeof(*v)));
    }
    if (ctx->ExecuteFlag) {
       /*CALL_Uniform2uiv(ctx->Exec, (location, count, v));*/
@@ -6618,11 +6303,11 @@ save_Uniform3uiv(GLint location, GLsizei count, const GLuint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_3UIV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_3UIV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 3 * sizeof(*v));
+      save_pointer(&n[3], memdup(v, count * 3 * sizeof(*v)));
    }
    if (ctx->ExecuteFlag) {
       /*CALL_Uniform3uiv(ctx->Exec, (location, count, v));*/
@@ -6635,11 +6320,11 @@ save_Uniform4uiv(GLint location, GLsizei count, const GLuint *v)
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_4UIV, 3);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_4UIV, 2 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
-      n[3].data = memdup(v, count * 4 * sizeof(*v));
+      save_pointer(&n[3], memdup(v, count * 4 * sizeof(*v)));
    }
    if (ctx->ExecuteFlag) {
       /*CALL_Uniform4uiv(ctx->Exec, (location, count, v));*/
@@ -6655,12 +6340,12 @@ save_UniformMatrix2fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX22, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX22, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 2 * 2 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 2 * 2 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix2fv(ctx->Exec, (location, count, transpose, m));
@@ -6674,12 +6359,12 @@ save_UniformMatrix3fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX33, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX33, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 3 * 3 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 3 * 3 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix3fv(ctx->Exec, (location, count, transpose, m));
@@ -6693,12 +6378,12 @@ save_UniformMatrix4fvARB(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX44, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX44, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 4 * 4 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 4 * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix4fv(ctx->Exec, (location, count, transpose, m));
@@ -6713,12 +6398,12 @@ save_UniformMatrix2x3fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX23, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX23, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 2 * 3 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 2 * 3 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix2x3fv(ctx->Exec, (location, count, transpose, m));
@@ -6732,12 +6417,12 @@ save_UniformMatrix3x2fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX32, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX32, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 3 * 2 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 3 * 2 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix3x2fv(ctx->Exec, (location, count, transpose, m));
@@ -6752,12 +6437,12 @@ save_UniformMatrix2x4fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX24, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX24, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 2 * 4 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 2 * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix2x4fv(ctx->Exec, (location, count, transpose, m));
@@ -6771,12 +6456,12 @@ save_UniformMatrix4x2fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX42, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX42, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 4 * 2 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 4 * 2 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix4x2fv(ctx->Exec, (location, count, transpose, m));
@@ -6791,12 +6476,12 @@ save_UniformMatrix3x4fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX34, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX34, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 3 * 4 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 3 * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix3x4fv(ctx->Exec, (location, count, transpose, m));
@@ -6810,12 +6495,12 @@ save_UniformMatrix4x3fv(GLint location, GLsizei count, GLboolean transpose,
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX43, 4);
+   n = alloc_instruction(ctx, OPCODE_UNIFORM_MATRIX43, 3 + POINTER_DWORDS);
    if (n) {
       n[1].i = location;
       n[2].i = count;
       n[3].b = transpose;
-      n[4].data = memdup(m, count * 4 * 3 * sizeof(GLfloat));
+      save_pointer(&n[4], memdup(m, count * 4 * 3 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
       CALL_UniformMatrix4x3fv(ctx->Exec, (location, count, transpose, m));
@@ -7185,10 +6870,10 @@ save_WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout)
    if (n) {
       union uint64_pair p;
       p.uint64 = timeout;
-      n[1].data = sync;
-      n[2].e = flags;
-      n[3].ui = p.uint32[0];
-      n[4].ui = p.uint32[1];
+      n[1].bf = flags;
+      n[2].ui = p.uint32[0];
+      n[3].ui = p.uint32[1];
+      save_pointer(&n[4], sync);
    }
    if (ctx->ExecuteFlag) {
       CALL_WaitSync(ctx->Exec, (sync, flags, timeout));
@@ -7252,10 +6937,14 @@ static void
 save_error(struct gl_context *ctx, GLenum error, const char *s)
 {
    Node *n;
-   n = alloc_instruction(ctx, OPCODE_ERROR, 2);
+   n = alloc_instruction(ctx, OPCODE_ERROR, 1 + POINTER_DWORDS);
    if (n) {
       n[1].e = error;
-      n[2].data = (void *) s;
+      save_pointer(&n[2], (void *) s);
+      /* note: the data/string here doesn't have to be freed in
+       * _mesa_delete_list() since the string is never dynamically
+       * allocated.
+       */
    }
 }
 
@@ -7336,7 +7025,7 @@ execute_list(struct gl_context *ctx, GLuint list)
       else {
          switch (opcode) {
          case OPCODE_ERROR:
-            _mesa_error(ctx, n[1].e, "%s", (const char *) n[2].data);
+            _mesa_error(ctx, n[1].e, "%s", (const char *) get_pointer(&n[2]));
             break;
          case OPCODE_ACCUM:
             CALL_Accum(ctx->Exec, (n[1].e, n[2].f));
@@ -7353,7 +7042,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                ctx->Unpack = ctx->DefaultPacking;
                CALL_Bitmap(ctx->Exec, ((GLsizei) n[1].i, (GLsizei) n[2].i,
                                        n[3].f, n[4].f, n[5].f, n[6].f,
-                                       (const GLubyte *) n[7].data));
+                                       get_pointer(&n[7])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7478,102 +7167,6 @@ execute_list(struct gl_context *ctx, GLuint list)
          case OPCODE_COLOR_MATERIAL:
             CALL_ColorMaterial(ctx->Exec, (n[1].e, n[2].e));
             break;
-         case OPCODE_COLOR_TABLE:
-            {
-               const struct gl_pixelstore_attrib save = ctx->Unpack;
-               ctx->Unpack = ctx->DefaultPacking;
-               CALL_ColorTable(ctx->Exec, (n[1].e, n[2].e, n[3].i, n[4].e,
-                                           n[5].e, n[6].data));
-               ctx->Unpack = save;      /* restore */
-            }
-            break;
-         case OPCODE_COLOR_TABLE_PARAMETER_FV:
-            {
-               GLfloat params[4];
-               params[0] = n[3].f;
-               params[1] = n[4].f;
-               params[2] = n[5].f;
-               params[3] = n[6].f;
-               CALL_ColorTableParameterfv(ctx->Exec,
-                                          (n[1].e, n[2].e, params));
-            }
-            break;
-         case OPCODE_COLOR_TABLE_PARAMETER_IV:
-            {
-               GLint params[4];
-               params[0] = n[3].i;
-               params[1] = n[4].i;
-               params[2] = n[5].i;
-               params[3] = n[6].i;
-               CALL_ColorTableParameteriv(ctx->Exec,
-                                          (n[1].e, n[2].e, params));
-            }
-            break;
-         case OPCODE_COLOR_SUB_TABLE:
-            {
-               const struct gl_pixelstore_attrib save = ctx->Unpack;
-               ctx->Unpack = ctx->DefaultPacking;
-               CALL_ColorSubTable(ctx->Exec, (n[1].e, n[2].i, n[3].i,
-                                              n[4].e, n[5].e, n[6].data));
-               ctx->Unpack = save;      /* restore */
-            }
-            break;
-         case OPCODE_CONVOLUTION_FILTER_1D:
-            {
-               const struct gl_pixelstore_attrib save = ctx->Unpack;
-               ctx->Unpack = ctx->DefaultPacking;
-               CALL_ConvolutionFilter1D(ctx->Exec, (n[1].e, n[2].i, n[3].i,
-                                                    n[4].e, n[5].e,
-                                                    n[6].data));
-               ctx->Unpack = save;      /* restore */
-            }
-            break;
-         case OPCODE_CONVOLUTION_FILTER_2D:
-            {
-               const struct gl_pixelstore_attrib save = ctx->Unpack;
-               ctx->Unpack = ctx->DefaultPacking;
-               CALL_ConvolutionFilter2D(ctx->Exec, (n[1].e, n[2].i, n[3].i,
-                                                    n[4].i, n[5].e, n[6].e,
-                                                    n[7].data));
-               ctx->Unpack = save;      /* restore */
-            }
-            break;
-         case OPCODE_CONVOLUTION_PARAMETER_I:
-            CALL_ConvolutionParameteri(ctx->Exec, (n[1].e, n[2].e, n[3].i));
-            break;
-         case OPCODE_CONVOLUTION_PARAMETER_IV:
-            {
-               GLint params[4];
-               params[0] = n[3].i;
-               params[1] = n[4].i;
-               params[2] = n[5].i;
-               params[3] = n[6].i;
-               CALL_ConvolutionParameteriv(ctx->Exec,
-                                           (n[1].e, n[2].e, params));
-            }
-            break;
-         case OPCODE_CONVOLUTION_PARAMETER_F:
-            CALL_ConvolutionParameterf(ctx->Exec, (n[1].e, n[2].e, n[3].f));
-            break;
-         case OPCODE_CONVOLUTION_PARAMETER_FV:
-            {
-               GLfloat params[4];
-               params[0] = n[3].f;
-               params[1] = n[4].f;
-               params[2] = n[5].f;
-               params[3] = n[6].f;
-               CALL_ConvolutionParameterfv(ctx->Exec,
-                                           (n[1].e, n[2].e, params));
-            }
-            break;
-         case OPCODE_COPY_COLOR_SUB_TABLE:
-            CALL_CopyColorSubTable(ctx->Exec, (n[1].e, n[2].i,
-                                               n[3].i, n[4].i, n[5].i));
-            break;
-         case OPCODE_COPY_COLOR_TABLE:
-            CALL_CopyColorSubTable(ctx->Exec, (n[1].e, n[2].i,
-                                               n[3].i, n[4].i, n[5].i));
-            break;
          case OPCODE_COPY_PIXELS:
             CALL_CopyPixels(ctx->Exec, (n[1].i, n[2].i,
                                         (GLsizei) n[3].i, (GLsizei) n[4].i,
@@ -7628,7 +7221,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                const struct gl_pixelstore_attrib save = ctx->Unpack;
                ctx->Unpack = ctx->DefaultPacking;
                CALL_DrawPixels(ctx->Exec, (n[1].i, n[2].i, n[3].e, n[4].e,
-                                           n[5].data));
+                                           get_pointer(&n[5])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7665,9 +7258,6 @@ execute_list(struct gl_context *ctx, GLuint list)
          case OPCODE_HINT:
             CALL_Hint(ctx->Exec, (n[1].e, n[2].e));
             break;
-         case OPCODE_HISTOGRAM:
-            CALL_Histogram(ctx->Exec, (n[1].e, n[2].i, n[3].e, n[4].b));
-            break;
          case OPCODE_INDEX_MASK:
             CALL_IndexMask(ctx->Exec, (n[1].ui));
             break;
@@ -7733,7 +7323,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                GLfloat u1 = n[2].f;
                GLfloat u2 = n[3].f;
                CALL_Map1f(ctx->Exec, (target, u1, u2, ustride, uorder,
-                                      (GLfloat *) n[6].data));
+                                      (GLfloat *) get_pointer(&n[6])));
             }
             break;
          case OPCODE_MAP2:
@@ -7749,7 +7339,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                GLint vorder = n[9].i;
                CALL_Map2f(ctx->Exec, (target, u1, u2, ustride, uorder,
                                       v1, v2, vstride, vorder,
-                                      (GLfloat *) n[10].data));
+                                      (GLfloat *) get_pointer(&n[10])));
             }
             break;
          case OPCODE_MAPGRID1:
@@ -7762,9 +7352,6 @@ execute_list(struct gl_context *ctx, GLuint list)
          case OPCODE_MATRIX_MODE:
             CALL_MatrixMode(ctx->Exec, (n[1].e));
             break;
-         case OPCODE_MIN_MAX:
-            CALL_Minmax(ctx->Exec, (n[1].e, n[2].e, n[3].b));
-            break;
          case OPCODE_MULT_MATRIX:
             if (sizeof(Node) == sizeof(GLfloat)) {
                CALL_MultMatrixf(ctx->Exec, (&n[1].f));
@@ -7787,7 +7374,7 @@ execute_list(struct gl_context *ctx, GLuint list)
             break;
          case OPCODE_PIXEL_MAP:
             CALL_PixelMapfv(ctx->Exec,
-                            (n[1].e, n[2].i, (GLfloat *) n[3].data));
+                            (n[1].e, n[2].i, get_pointer(&n[3])));
             break;
          case OPCODE_PIXEL_TRANSFER:
             CALL_PixelTransferf(ctx->Exec, (n[1].e, n[2].f));
@@ -7814,7 +7401,7 @@ execute_list(struct gl_context *ctx, GLuint list)
             {
                const struct gl_pixelstore_attrib save = ctx->Unpack;
                ctx->Unpack = ctx->DefaultPacking;
-               CALL_PolygonStipple(ctx->Exec, ((GLubyte *) n[1].data));
+               CALL_PolygonStipple(ctx->Exec, (get_pointer(&n[1])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7848,12 +7435,6 @@ execute_list(struct gl_context *ctx, GLuint list)
          case OPCODE_READ_BUFFER:
             CALL_ReadBuffer(ctx->Exec, (n[1].e));
             break;
-         case OPCODE_RESET_HISTOGRAM:
-            CALL_ResetHistogram(ctx->Exec, (n[1].e));
-            break;
-         case OPCODE_RESET_MIN_MAX:
-            CALL_ResetMinmax(ctx->Exec, (n[1].e));
-            break;
          case OPCODE_ROTATE:
             CALL_Rotatef(ctx->Exec, (n[1].f, n[2].f, n[3].f, n[4].f));
             break;
@@ -7930,7 +7511,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                                            n[5].e,      /* border */
                                            n[6].e,      /* format */
                                            n[7].e,      /* type */
-                                           n[8].data));
+                                           get_pointer(&n[8])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7946,7 +7527,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                                            n[6].e,      /* border */
                                            n[7].e,      /* format */
                                            n[8].e,      /* type */
-                                           n[9].data));
+                                           get_pointer(&n[9])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7963,7 +7544,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                                            n[7].e,      /* border */
                                            n[8].e,      /* format */
                                            n[9].e,      /* type */
-                                           n[10].data));
+                                           get_pointer(&n[10])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7973,7 +7554,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                ctx->Unpack = ctx->DefaultPacking;
                CALL_TexSubImage1D(ctx->Exec, (n[1].e, n[2].i, n[3].i,
                                               n[4].i, n[5].e,
-                                              n[6].e, n[7].data));
+                                              n[6].e, get_pointer(&n[7])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7984,7 +7565,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                CALL_TexSubImage2D(ctx->Exec, (n[1].e, n[2].i, n[3].i,
                                               n[4].i, n[5].e,
                                               n[6].i, n[7].e, n[8].e,
-                                              n[9].data));
+                                              get_pointer(&n[9])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -7995,7 +7576,7 @@ execute_list(struct gl_context *ctx, GLuint list)
                CALL_TexSubImage3D(ctx->Exec, (n[1].e, n[2].i, n[3].i,
                                               n[4].i, n[5].i, n[6].i, n[7].i,
                                               n[8].i, n[9].e, n[10].e,
-                                              n[11].data));
+                                              get_pointer(&n[11])));
                ctx->Unpack = save;      /* restore */
             }
             break;
@@ -8014,36 +7595,38 @@ execute_list(struct gl_context *ctx, GLuint list)
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_1D:  /* GL_ARB_texture_compression */
             CALL_CompressedTexImage1D(ctx->Exec, (n[1].e, n[2].i, n[3].e,
-                                                     n[4].i, n[5].i, n[6].i,
-                                                     n[7].data));
+                                                  n[4].i, n[5].i, n[6].i,
+                                                  get_pointer(&n[7])));
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_2D:  /* GL_ARB_texture_compression */
             CALL_CompressedTexImage2D(ctx->Exec, (n[1].e, n[2].i, n[3].e,
-                                                     n[4].i, n[5].i, n[6].i,
-                                                     n[7].i, n[8].data));
+                                                  n[4].i, n[5].i, n[6].i,
+                                                  n[7].i, get_pointer(&n[8])));
             break;
          case OPCODE_COMPRESSED_TEX_IMAGE_3D:  /* GL_ARB_texture_compression */
             CALL_CompressedTexImage3D(ctx->Exec, (n[1].e, n[2].i, n[3].e,
-                                                     n[4].i, n[5].i, n[6].i,
-                                                     n[7].i, n[8].i,
-                                                     n[9].data));
+                                                  n[4].i, n[5].i, n[6].i,
+                                                  n[7].i, n[8].i,
+                                                  get_pointer(&n[9])));
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_1D:      /* GL_ARB_texture_compress */
             CALL_CompressedTexSubImage1D(ctx->Exec,
                                             (n[1].e, n[2].i, n[3].i, n[4].i,
-                                             n[5].e, n[6].i, n[7].data));
+                                             n[5].e, n[6].i,
+                                             get_pointer(&n[7])));
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_2D:      /* GL_ARB_texture_compress */
             CALL_CompressedTexSubImage2D(ctx->Exec,
                                             (n[1].e, n[2].i, n[3].i, n[4].i,
                                              n[5].i, n[6].i, n[7].e, n[8].i,
-                                             n[9].data));
+                                             get_pointer(&n[9])));
             break;
          case OPCODE_COMPRESSED_TEX_SUB_IMAGE_3D:      /* GL_ARB_texture_compress */
             CALL_CompressedTexSubImage3D(ctx->Exec,
                                             (n[1].e, n[2].i, n[3].i, n[4].i,
                                              n[5].i, n[6].i, n[7].i, n[8].i,
-                                             n[9].e, n[10].i, n[11].data));
+                                             n[9].e, n[10].i,
+                                             get_pointer(&n[11])));
             break;
          case OPCODE_SAMPLE_COVERAGE:  /* GL_ARB_multisample */
             CALL_SampleCoverage(ctx->Exec, (n[1].f, n[2].b));
@@ -8067,7 +7650,8 @@ execute_list(struct gl_context *ctx, GLuint list)
             break;
          case OPCODE_PROGRAM_STRING_ARB:
             CALL_ProgramStringARB(ctx->Exec,
-                                  (n[1].e, n[2].e, n[3].i, n[4].data));
+                                  (n[1].e, n[2].e, n[3].i,
+                                   get_pointer(&n[4])));
             break;
          case OPCODE_PROGRAM_ENV_PARAMETER_ARB:
             CALL_ProgramEnvParameter4fARB(ctx->Exec, (n[1].e, n[2].ui, n[3].f,
@@ -8126,16 +7710,16 @@ execute_list(struct gl_context *ctx, GLuint list)
                               (n[1].i, n[2].f, n[3].f, n[4].f, n[5].f));
 	    break;
 	 case OPCODE_UNIFORM_1FV:
-	    CALL_Uniform1fv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform1fv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_2FV:
-	    CALL_Uniform2fv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform2fv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_3FV:
-	    CALL_Uniform3fv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform3fv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_4FV:
-	    CALL_Uniform4fv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform4fv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_1I:
 	    CALL_Uniform1i(ctx->Exec, (n[1].i, n[2].i));
@@ -8151,16 +7735,16 @@ execute_list(struct gl_context *ctx, GLuint list)
                               (n[1].i, n[2].i, n[3].i, n[4].i, n[5].i));
 	    break;
 	 case OPCODE_UNIFORM_1IV:
-	    CALL_Uniform1iv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform1iv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_2IV:
-	    CALL_Uniform2iv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform2iv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_3IV:
-	    CALL_Uniform3iv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform3iv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_4IV:
-	    CALL_Uniform4iv(ctx->Exec, (n[1].i, n[2].i, n[3].data));
+	    CALL_Uniform4iv(ctx->Exec, (n[1].i, n[2].i, get_pointer(&n[3])));
 	    break;
 	 case OPCODE_UNIFORM_1UI:
 	    /*CALL_Uniform1uiARB(ctx->Exec, (n[1].i, n[2].i));*/
@@ -8177,52 +7761,56 @@ execute_list(struct gl_context *ctx, GLuint list)
             */
 	    break;
 	 case OPCODE_UNIFORM_1UIV:
-	    /*CALL_Uniform1uivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data));*/
+	    /*CALL_Uniform1uivARB(ctx->Exec, (n[1].i, n[2].i,
+                                              get_pointer(&n[3])));*/
 	    break;
 	 case OPCODE_UNIFORM_2UIV:
-	    /*CALL_Uniform2uivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data));*/
+	    /*CALL_Uniform2uivARB(ctx->Exec, (n[1].i, n[2].i,
+                                              get_pointer(&n[3])));*/
 	    break;
 	 case OPCODE_UNIFORM_3UIV:
-	    /*CALL_Uniform3uivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data));*/
+	    /*CALL_Uniform3uivARB(ctx->Exec, (n[1].i, n[2].i,
+                                              get_pointer(&n[3])));*/
 	    break;
 	 case OPCODE_UNIFORM_4UIV:
-	    /*CALL_Uniform4uivARB(ctx->Exec, (n[1].i, n[2].i, n[3].data));*/
+	    /*CALL_Uniform4uivARB(ctx->Exec, (n[1].i, n[2].i,
+                                              get_pointer(&n[3])));*/
 	    break;
 	 case OPCODE_UNIFORM_MATRIX22:
 	    CALL_UniformMatrix2fv(ctx->Exec,
-                                     (n[1].i, n[2].i, n[3].b, n[4].data));
+                                  (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX33:
 	    CALL_UniformMatrix3fv(ctx->Exec,
-                                     (n[1].i, n[2].i, n[3].b, n[4].data));
+                                  (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX44:
 	    CALL_UniformMatrix4fv(ctx->Exec,
-                                     (n[1].i, n[2].i, n[3].b, n[4].data));
+                                  (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX23:
 	    CALL_UniformMatrix2x3fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX32:
 	    CALL_UniformMatrix3x2fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX24:
 	    CALL_UniformMatrix2x4fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX42:
 	    CALL_UniformMatrix4x2fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX34:
 	    CALL_UniformMatrix3x4fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 	 case OPCODE_UNIFORM_MATRIX43:
 	    CALL_UniformMatrix4x3fv(ctx->Exec,
-                                    (n[1].i, n[2].i, n[3].b, n[4].data));
+                                    (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 
          case OPCODE_CLAMP_COLOR:
@@ -8476,9 +8064,10 @@ execute_list(struct gl_context *ctx, GLuint list)
          case OPCODE_WAIT_SYNC:
             {
                union uint64_pair p;
-               p.uint32[0] = n[3].ui;
-               p.uint32[1] = n[4].ui;
-               CALL_WaitSync(ctx->Exec, (n[1].data, n[2].bf, p.uint64));
+               p.uint32[0] = n[2].ui;
+               p.uint32[1] = n[3].ui;
+               CALL_WaitSync(ctx->Exec,
+                             (get_pointer(&n[4]), n[1].bf, p.uint64));
             }
             break;
 
@@ -8495,7 +8084,7 @@ execute_list(struct gl_context *ctx, GLuint list)
             break;
 
          case OPCODE_CONTINUE:
-            n = (Node *) n[1].next;
+            n = (Node *) get_pointer(&n[1]);
             break;
          case OPCODE_END_OF_LIST:
             done = GL_TRUE;
@@ -8684,6 +8273,8 @@ _mesa_EndList(void)
 
    (void) alloc_instruction(ctx, OPCODE_END_OF_LIST, 0);
 
+   trim_list(ctx);
+
    /* Destroy old list, if any */
    destroy_list(ctx, ctx->ListState.CurrentList->Name);
 
@@ -8697,6 +8288,8 @@ _mesa_EndList(void)
       mesa_print_display_list(ctx->ListState.CurrentList->Name);
 
    ctx->ListState.CurrentList = NULL;
+   ctx->ListState.CurrentBlock = NULL;
+   ctx->ListState.CurrentPos = 0;
    ctx->ExecuteFlag = GL_TRUE;
    ctx->CompileFlag = GL_FALSE;
 
@@ -8998,22 +8591,6 @@ _mesa_initialize_save_table(const struct gl_context *ctx)
    /* Not all are supported */
    SET_BlendColor(table, save_BlendColor);
    SET_BlendEquation(table, save_BlendEquation);
-   SET_ColorSubTable(table, save_ColorSubTable);
-   SET_ColorTable(table, save_ColorTable);
-   SET_ColorTableParameterfv(table, save_ColorTableParameterfv);
-   SET_ColorTableParameteriv(table, save_ColorTableParameteriv);
-   SET_ConvolutionFilter1D(table, save_ConvolutionFilter1D);
-   SET_ConvolutionFilter2D(table, save_ConvolutionFilter2D);
-   SET_ConvolutionParameterf(table, save_ConvolutionParameterf);
-   SET_ConvolutionParameterfv(table, save_ConvolutionParameterfv);
-   SET_ConvolutionParameteri(table, save_ConvolutionParameteri);
-   SET_ConvolutionParameteriv(table, save_ConvolutionParameteriv);
-   SET_CopyColorSubTable(table, save_CopyColorSubTable);
-   SET_CopyColorTable(table, save_CopyColorTable);
-   SET_Histogram(table, save_Histogram);
-   SET_Minmax(table, save_Minmax);
-   SET_ResetHistogram(table, save_ResetHistogram);
-   SET_ResetMinmax(table, save_ResetMinmax);
 
    /* 2. GL_EXT_blend_color */
 #if 0
@@ -9030,12 +8607,6 @@ _mesa_initialize_save_table(const struct gl_context *ctx)
    SET_TexSubImage3DEXT(table, save_TexSubImage3D);
 #endif
 
-   /* 14. GL_SGI_color_table */
-#if 0
-   SET_ColorTableSGI(table, save_ColorTable);
-   SET_ColorSubTableSGI(table, save_ColorSubTable);
-#endif
-
    /* 37. GL_EXT_blend_minmax */
 #if 0
    SET_BlendEquationEXT(table, save_BlendEquationEXT);
@@ -9332,7 +8903,8 @@ print_list(struct gl_context *ctx, GLuint list)
             break;
          case OPCODE_BITMAP:
             printf("Bitmap %d %d %g %g %g %g %p\n", n[1].i, n[2].i,
-                         n[3].f, n[4].f, n[5].f, n[6].f, (void *) n[7].data);
+                   n[3].f, n[4].f, n[5].f, n[6].f,
+                   get_pointer(&n[7]));
             break;
          case OPCODE_CALL_LIST:
             printf("CallList %d\n", (int) n[1].ui);
@@ -9341,16 +8913,6 @@ print_list(struct gl_context *ctx, GLuint list)
             printf("CallList %d + offset %u = %u\n", (int) n[1].ui,
                          ctx->List.ListBase, ctx->List.ListBase + n[1].ui);
             break;
-         case OPCODE_COLOR_TABLE_PARAMETER_FV:
-            printf("ColorTableParameterfv %s %s %f %f %f %f\n",
-                         enum_string(n[1].e), enum_string(n[2].e),
-                         n[3].f, n[4].f, n[5].f, n[6].f);
-            break;
-         case OPCODE_COLOR_TABLE_PARAMETER_IV:
-            printf("ColorTableParameteriv %s %s %d %d %d %d\n",
-                         enum_string(n[1].e), enum_string(n[2].e),
-                         n[3].i, n[4].i, n[5].i, n[6].i);
-            break;
          case OPCODE_DISABLE:
             printf("Disable %s\n", enum_string(n[1].e));
             break;
@@ -9525,12 +9087,12 @@ print_list(struct gl_context *ctx, GLuint list)
              * meta opcodes/commands
              */
          case OPCODE_ERROR:
-            printf("Error: %s %s\n",
-                         enum_string(n[1].e), (const char *) n[2].data);
+            printf("Error: %s %s\n", enum_string(n[1].e),
+                   (const char *) get_pointer(&n[2]));
             break;
          case OPCODE_CONTINUE:
             printf("DISPLAY-LIST-CONTINUE\n");
-            n = (Node *) n[1].next;
+            n = (Node *) get_pointer(&n[1]);
             break;
          case OPCODE_END_OF_LIST:
             printf("END-LIST %u\n", list);
diff --git a/mesalib/src/mesa/main/enable.c b/mesalib/src/mesa/main/enable.c
index c047f5df2..bb4a23c91 100644
--- a/mesalib/src/mesa/main/enable.c
+++ b/mesalib/src/mesa/main/enable.c
@@ -934,25 +934,6 @@ _mesa_set_enable(struct gl_context *ctx, GLenum cap, GLboolean state)
 	ctx->ATIFragmentShader.Enabled = state;
         break;
 
-      /* GL_MESA_texture_array */
-      case GL_TEXTURE_1D_ARRAY_EXT:
-         if (ctx->API != API_OPENGL_COMPAT)
-            goto invalid_enum_error;
-         CHECK_EXTENSION(MESA_texture_array, cap);
-         if (!enable_texture(ctx, state, TEXTURE_1D_ARRAY_BIT)) {
-            return;
-         }
-         break;
-
-      case GL_TEXTURE_2D_ARRAY_EXT:
-         if (ctx->API != API_OPENGL_COMPAT)
-            goto invalid_enum_error;
-         CHECK_EXTENSION(MESA_texture_array, cap);
-         if (!enable_texture(ctx, state, TEXTURE_2D_ARRAY_BIT)) {
-            return;
-         }
-         break;
-
       case GL_TEXTURE_CUBE_MAP_SEAMLESS:
          if (!_mesa_is_desktop_gl(ctx))
             goto invalid_enum_error;
diff --git a/mesalib/src/mesa/main/extensions.c b/mesalib/src/mesa/main/extensions.c
index 42da9057c..f0e1858e4 100644
--- a/mesalib/src/mesa/main/extensions.c
+++ b/mesalib/src/mesa/main/extensions.c
@@ -82,6 +82,7 @@ static const struct extension extension_table[] = {
    { "GL_ARB_ES3_compatibility",                   o(ARB_ES3_compatibility),                   GL,             2012 },
    { "GL_ARB_base_instance",                       o(ARB_base_instance),                       GL,             2011 },
    { "GL_ARB_blend_func_extended",                 o(ARB_blend_func_extended),                 GL,             2009 },
+   { "GL_ARB_clear_buffer_object",                 o(dummy_true),                              GL,             2012 },
    { "GL_ARB_color_buffer_float",                  o(ARB_color_buffer_float),                  GL,             2004 },
    { "GL_ARB_copy_buffer",                         o(dummy_true),                              GL,             2008 },
    { "GL_ARB_conservative_depth",                  o(ARB_conservative_depth),                  GL,             2011 },
@@ -158,6 +159,7 @@ static const struct extension extension_table[] = {
    { "GL_ARB_texture_rg",                          o(ARB_texture_rg),                          GL,             2008 },
    { "GL_ARB_texture_storage",                     o(dummy_true),                              GL,             2011 },
    { "GL_ARB_texture_storage_multisample",         o(ARB_texture_multisample),                 GL,             2012 },
+   { "GL_ARB_texture_view",                        o(ARB_texture_view),                        GL,             2012 },
    { "GL_ARB_texture_swizzle",                     o(EXT_texture_swizzle),                     GL,             2008 },
    { "GL_ARB_timer_query",                         o(ARB_timer_query),                         GL,             2010 },
    { "GL_ARB_transform_feedback2",                 o(ARB_transform_feedback2),                 GL,             2010 },
@@ -199,7 +201,7 @@ static const struct extension extension_table[] = {
    { "GL_EXT_gpu_shader4",                         o(EXT_gpu_shader4),                         GL,             2006 },
    { "GL_EXT_map_buffer_range",                    o(ARB_map_buffer_range),                          ES1 | ES2, 2012 },
    { "GL_EXT_multi_draw_arrays",                   o(dummy_true),                              GLL | ES1 | ES2, 1999 },
-   { "GL_EXT_packed_depth_stencil",                o(EXT_packed_depth_stencil),                GL,             2005 },
+   { "GL_EXT_packed_depth_stencil",                o(dummy_true),                              GL,             2005 },
    { "GL_EXT_packed_float",                        o(EXT_packed_float),                        GL,             2004 },
    { "GL_EXT_packed_pixels",                       o(dummy_true),                              GLL,            1997 },
    { "GL_EXT_pixel_buffer_object",                 o(EXT_pixel_buffer_object),                 GL,             2004 },
@@ -272,7 +274,7 @@ static const struct extension extension_table[] = {
    { "GL_OES_framebuffer_object",                  o(dummy_true),                                   ES1,       2005 },
    { "GL_OES_get_program_binary",                  o(dummy_true),                                         ES2, 2008 },
    { "GL_OES_mapbuffer",                           o(dummy_true),                                   ES1 | ES2, 2005 },
-   { "GL_OES_packed_depth_stencil",                o(EXT_packed_depth_stencil),                     ES1 | ES2, 2007 },
+   { "GL_OES_packed_depth_stencil",                o(dummy_true),                                   ES1 | ES2, 2007 },
    { "GL_OES_point_size_array",                    o(dummy_true),                                   ES1,       2004 },
    { "GL_OES_point_sprite",                        o(ARB_point_sprite),                             ES1,       2004 },
    { "GL_OES_query_matrix",                        o(dummy_true),                                   ES1,       2003 },
@@ -302,7 +304,7 @@ static const struct extension extension_table[] = {
    { "GL_AMD_performance_monitor",                 o(AMD_performance_monitor),                 GL,             2007 },
    { "GL_AMD_seamless_cubemap_per_texture",        o(AMD_seamless_cubemap_per_texture),        GL,             2009 },
    { "GL_AMD_shader_stencil_export",               o(ARB_shader_stencil_export),               GL,             2009 },
-   { "GL_AMD_vertex_shader_layer",                 o(AMD_vertex_shader_layer),                 GL,             2012 },
+   { "GL_AMD_vertex_shader_layer",                 o(AMD_vertex_shader_layer),                 GLC,            2012 },
    { "GL_APPLE_object_purgeable",                  o(APPLE_object_purgeable),                  GL,             2006 },
    { "GL_APPLE_packed_pixels",                     o(dummy_true),                              GLL,            2002 },
    { "GL_APPLE_texture_max_level",                 o(dummy_true),                                   ES1 | ES2, 2009 },
@@ -321,7 +323,6 @@ static const struct extension extension_table[] = {
    { "GL_IBM_texture_mirrored_repeat",             o(dummy_true),                              GLL,            1998 },
    { "GL_INGR_blend_func_separate",                o(EXT_blend_func_separate),                 GLL,            1999 },
    { "GL_MESA_pack_invert",                        o(MESA_pack_invert),                        GL,             2002 },
-   { "GL_MESA_texture_array",                      o(MESA_texture_array),                      GLL,            2007 },
    { "GL_MESA_texture_signed_rgba",                o(EXT_texture_snorm),                       GL,             2009 },
    { "GL_MESA_window_pos",                         o(dummy_true),                              GLL,            2000 },
    { "GL_MESA_ycbcr_texture",                      o(MESA_ycbcr_texture),                      GL,             2002 },
@@ -333,7 +334,7 @@ static const struct extension extension_table[] = {
    { "GL_NV_fog_distance",                         o(NV_fog_distance),                         GLL,            2001 },
    { "GL_NV_fragment_program_option",              o(NV_fragment_program_option),              GLL,            2005 },
    { "GL_NV_light_max_exponent",                   o(dummy_true),                              GLL,            1999 },
-   { "GL_NV_packed_depth_stencil",                 o(EXT_packed_depth_stencil),                GL,             2000 },
+   { "GL_NV_packed_depth_stencil",                 o(dummy_true),                              GL,             2000 },
    { "GL_NV_point_sprite",                         o(NV_point_sprite),                         GL,             2001 },
    { "GL_NV_primitive_restart",                    o(NV_primitive_restart),                    GLL,            2002 },
    { "GL_NV_read_buffer",                          o(dummy_true),                              ES2,            2011 },
@@ -436,7 +437,6 @@ _mesa_enable_sw_extensions(struct gl_context *ctx)
    ctx->Extensions.EXT_depth_bounds_test = GL_TRUE;
    ctx->Extensions.EXT_draw_buffers2 = GL_TRUE;
    ctx->Extensions.EXT_framebuffer_blit = GL_TRUE;
-   ctx->Extensions.EXT_packed_depth_stencil = GL_TRUE;
    ctx->Extensions.EXT_pixel_buffer_object = GL_TRUE;
    ctx->Extensions.EXT_point_parameters = GL_TRUE;
    ctx->Extensions.EXT_provoking_vertex = GL_TRUE;
@@ -453,7 +453,6 @@ _mesa_enable_sw_extensions(struct gl_context *ctx)
    /*ctx->Extensions.EXT_transform_feedback = GL_TRUE;*/
    ctx->Extensions.EXT_vertex_array_bgra = GL_TRUE;
    ctx->Extensions.MESA_pack_invert = GL_TRUE;
-   ctx->Extensions.MESA_texture_array = GL_TRUE;
    ctx->Extensions.MESA_ycbcr_texture = GL_TRUE;
    ctx->Extensions.NV_conditional_render = GL_TRUE;
    ctx->Extensions.NV_point_sprite = GL_TRUE;
@@ -480,7 +479,8 @@ set_extension( struct gl_context *ctx, const char *name, GLboolean state )
 
    if (ctx->Extensions.String) {
       /* The string was already queried - can't change it now! */
-      _mesa_problem(ctx, "Trying to enable/disable extension after glGetString(GL_EXTENSIONS): %s", name);
+      _mesa_problem(ctx, "Trying to enable/disable extension after "
+                    "glGetString(GL_EXTENSIONS): %s", name);
       return GL_FALSE;
    }
 
@@ -679,10 +679,10 @@ _mesa_make_extension_string(struct gl_context *ctx)
       return NULL;
    }
 
-   /* Sort extensions in chronological order because certain old applications (e.g.,
-    * Quake3 demo) store the extension list in a static size buffer so chronologically
-    * order ensure that the extensions that such applications expect will fit into
-    * that buffer.
+   /* Sort extensions in chronological order because certain old applications
+    * (e.g., Quake3 demo) store the extension list in a static size buffer so
+    * chronologically order ensure that the extensions that such applications
+    * expect will fit into that buffer.
     */
    j = 0;
    for (i = extension_table; i->name != 0; ++i) {
@@ -693,7 +693,8 @@ _mesa_make_extension_string(struct gl_context *ctx)
       }
    }
    assert(j == count);
-   qsort(extension_indices, count, sizeof *extension_indices, extension_compare);
+   qsort(extension_indices, count,
+         sizeof *extension_indices, extension_compare);
 
    /* Build the extension string.*/
    for (j = 0; j < count; ++j) {
diff --git a/mesalib/src/mesa/main/fbobject.c b/mesalib/src/mesa/main/fbobject.c
index 861885dd3..2892784f8 100644
--- a/mesalib/src/mesa/main/fbobject.c
+++ b/mesalib/src/mesa/main/fbobject.c
@@ -781,9 +781,8 @@ test_attachment_completeness(const struct gl_context *ctx, GLenum format,
          if (baseFormat == GL_DEPTH_COMPONENT) {
             /* OK */
          }
-         else if (ctx->Extensions.EXT_packed_depth_stencil &&
-                  ctx->Extensions.ARB_depth_texture &&
-                  baseFormat == GL_DEPTH_STENCIL_EXT) {
+         else if (ctx->Extensions.ARB_depth_texture &&
+                  baseFormat == GL_DEPTH_STENCIL) {
             /* OK */
          }
          else {
@@ -794,9 +793,8 @@ test_attachment_completeness(const struct gl_context *ctx, GLenum format,
       }
       else {
          ASSERT(format == GL_STENCIL);
-         if (ctx->Extensions.EXT_packed_depth_stencil &&
-             ctx->Extensions.ARB_depth_texture &&
-             baseFormat == GL_DEPTH_STENCIL_EXT) {
+         if (ctx->Extensions.ARB_depth_texture &&
+             baseFormat == GL_DEPTH_STENCIL) {
             /* OK */
          }
          else {
@@ -830,8 +828,7 @@ test_attachment_completeness(const struct gl_context *ctx, GLenum format,
          if (baseFormat == GL_DEPTH_COMPONENT) {
             /* OK */
          }
-         else if (ctx->Extensions.EXT_packed_depth_stencil &&
-                  baseFormat == GL_DEPTH_STENCIL_EXT) {
+         else if (baseFormat == GL_DEPTH_STENCIL) {
             /* OK */
          }
          else {
@@ -842,11 +839,8 @@ test_attachment_completeness(const struct gl_context *ctx, GLenum format,
       }
       else {
          assert(format == GL_STENCIL);
-         if (baseFormat == GL_STENCIL_INDEX) {
-            /* OK */
-         }
-         else if (ctx->Extensions.EXT_packed_depth_stencil &&
-                  baseFormat == GL_DEPTH_STENCIL_EXT) {
+         if (baseFormat == GL_STENCIL_INDEX ||
+             baseFormat == GL_DEPTH_STENCIL) {
             /* OK */
          }
          else {
@@ -1457,13 +1451,10 @@ _mesa_base_fbo_format(struct gl_context *ctx, GLenum internalFormat)
    case GL_DEPTH_COMPONENT16:
    case GL_DEPTH_COMPONENT24:
       return GL_DEPTH_COMPONENT;
-   case GL_DEPTH_STENCIL_EXT:
-      return _mesa_is_desktop_gl(ctx)
-         && ctx->Extensions.EXT_packed_depth_stencil
-         ? GL_DEPTH_STENCIL_EXT : 0;
-   case GL_DEPTH24_STENCIL8_EXT:
-      return ctx->Extensions.EXT_packed_depth_stencil
-         ? GL_DEPTH_STENCIL_EXT : 0;
+   case GL_DEPTH_STENCIL:
+      return _mesa_is_desktop_gl(ctx) ? GL_DEPTH_STENCIL : 0;
+   case GL_DEPTH24_STENCIL8:
+      return GL_DEPTH_STENCIL;
    case GL_DEPTH_COMPONENT32F:
       return ctx->Version >= 30
          || (ctx->API == API_OPENGL_COMPAT && ctx->Extensions.ARB_depth_buffer_float)
diff --git a/mesalib/src/mesa/main/ff_fragment_shader.cpp b/mesalib/src/mesa/main/ff_fragment_shader.cpp
index 01edd3ff8..ba6258d89 100644
--- a/mesalib/src/mesa/main/ff_fragment_shader.cpp
+++ b/mesalib/src/mesa/main/ff_fragment_shader.cpp
@@ -543,7 +543,8 @@ get_current_attrib(texenv_fragment_program *p, GLuint attrib)
    ir_rvalue *val;
 
    current = p->shader->symbols->get_variable("gl_CurrentAttribFragMESA");
-   current->max_array_access = MAX2(current->max_array_access, attrib);
+   assert(current);
+   current->data.max_array_access = MAX2(current->data.max_array_access, attrib);
    val = new(p->mem_ctx) ir_dereference_variable(current);
    ir_rvalue *index = new(p->mem_ctx) ir_constant(attrib);
    return new(p->mem_ctx) ir_dereference_array(val, index);
@@ -587,7 +588,7 @@ get_source(texenv_fragment_program *p,
       var = p->shader->symbols->get_variable("gl_TextureEnvColor");
       assert(var);
       deref = new(p->mem_ctx) ir_dereference_variable(var);
-      var->max_array_access = MAX2(var->max_array_access, unit);
+      var->data.max_array_access = MAX2(var->data.max_array_access, unit);
       return new(p->mem_ctx) ir_dereference_array(deref,
 						  new(p->mem_ctx) ir_constant(unit));
 
@@ -927,7 +928,7 @@ static void load_texture( texenv_fragment_program *p, GLuint unit )
       texcoord = new(p->mem_ctx) ir_dereference_variable(tc_array);
       ir_rvalue *index = new(p->mem_ctx) ir_constant(unit);
       texcoord = new(p->mem_ctx) ir_dereference_array(texcoord, index);
-      tc_array->max_array_access = MAX2(tc_array->max_array_access, unit);
+      tc_array->data.max_array_access = MAX2(tc_array->data.max_array_access, unit);
    }
 
    if (!p->state->unit[unit].enabled) {
@@ -1096,14 +1097,16 @@ load_texunit_bumpmap( texenv_fragment_program *p, GLuint unit )
    ir_variable *rot_mat_0, *rot_mat_1;
 
    rot_mat_0 = p->shader->symbols->get_variable("gl_BumpRotMatrix0MESA");
+   assert(rot_mat_0);
    rot_mat_1 = p->shader->symbols->get_variable("gl_BumpRotMatrix1MESA");
+   assert(rot_mat_1);
 
    ir_variable *tc_array = p->shader->symbols->get_variable("gl_TexCoord");
    assert(tc_array);
    texcoord = new(p->mem_ctx) ir_dereference_variable(tc_array);
    ir_rvalue *index = new(p->mem_ctx) ir_constant(bumpedUnitNr);
    texcoord = new(p->mem_ctx) ir_dereference_array(texcoord, index);
-   tc_array->max_array_access = MAX2(tc_array->max_array_access, unit);
+   tc_array->data.max_array_access = MAX2(tc_array->data.max_array_access, unit);
 
    load_texenv_source( p, unit + SRC_TEXTURE0, unit );
 
@@ -1157,8 +1160,11 @@ emit_fog_instructions(texenv_fragment_program *p,
    fragcolor = swizzle_xyz(fog_result);
 
    oparams = p->shader->symbols->get_variable("gl_FogParamsOptimizedMESA");
+   assert(oparams);
    fogcoord = p->shader->symbols->get_variable("gl_FogFragCoord");
+   assert(fogcoord);
    params = p->shader->symbols->get_variable("gl_Fog");
+   assert(params);
    f = new(p->mem_ctx) ir_dereference_variable(fogcoord);
 
    ir_variable *f_var = p->make_temp(glsl_type::float_type, "fog_factor");
@@ -1344,7 +1350,7 @@ create_new_program(struct gl_context *ctx, struct state_key *key)
 
    p.shader->CompileStatus = true;
    p.shader->Version = state->language_version;
-   p.shader->num_builtins_to_link = state->num_builtins_to_link;
+   p.shader->uses_builtin_functions = state->uses_builtin_functions;
    p.shader_program->Shaders =
       (gl_shader **)malloc(sizeof(*p.shader_program->Shaders));
    p.shader_program->Shaders[0] = p.shader;
diff --git a/mesalib/src/mesa/main/get.c b/mesalib/src/mesa/main/get.c
index 7f233409e..691380898 100644
--- a/mesalib/src/mesa/main/get.c
+++ b/mesalib/src/mesa/main/get.c
@@ -327,8 +327,8 @@ static const int extra_EXT_framebuffer_sRGB_and_new_buffers[] = {
    EXTRA_END
 };
 
-static const int extra_MESA_texture_array_es3[] = {
-   EXT(MESA_texture_array),
+static const int extra_EXT_texture_array_es3[] = {
+   EXT(EXT_texture_array),
    EXTRA_API_ES3,
    EXTRA_END
 };
@@ -339,7 +339,7 @@ static const int extra_ARB_shader_atomic_counters_and_geometry_shader[] = {
 };
 
 EXTRA_EXT(ARB_texture_cube_map);
-EXTRA_EXT(MESA_texture_array);
+EXTRA_EXT(EXT_texture_array);
 EXTRA_EXT(NV_fog_distance);
 EXTRA_EXT(EXT_texture_filter_anisotropic);
 EXTRA_EXT(NV_point_sprite);
@@ -565,8 +565,6 @@ find_custom_value(struct gl_context *ctx, const struct value_desc *d, union valu
    case GL_TEXTURE_1D:
    case GL_TEXTURE_2D:
    case GL_TEXTURE_3D:
-   case GL_TEXTURE_1D_ARRAY_EXT:
-   case GL_TEXTURE_2D_ARRAY_EXT:
    case GL_TEXTURE_CUBE_MAP_ARB:
    case GL_TEXTURE_RECTANGLE_NV:
    case GL_TEXTURE_EXTERNAL_OES:
diff --git a/mesalib/src/mesa/main/get_hash_params.py b/mesalib/src/mesa/main/get_hash_params.py
index 781b796ee..653bf6256 100644
--- a/mesalib/src/mesa/main/get_hash_params.py
+++ b/mesalib/src/mesa/main/get_hash_params.py
@@ -376,8 +376,8 @@ descriptor=[
   [ "PIXEL_UNPACK_BUFFER_BINDING_EXT", "LOC_CUSTOM, TYPE_INT, 0, extra_EXT_pixel_buffer_object" ],
 
   # GL_EXT_texture_array
-  [ "TEXTURE_BINDING_2D_ARRAY", "LOC_CUSTOM, TYPE_INT, TEXTURE_2D_ARRAY_INDEX, extra_MESA_texture_array_es3" ],
-  [ "MAX_ARRAY_TEXTURE_LAYERS_EXT", "CONTEXT_INT(Const.MaxArrayTextureLayers), extra_MESA_texture_array_es3" ],
+  [ "TEXTURE_BINDING_2D_ARRAY", "LOC_CUSTOM, TYPE_INT, TEXTURE_2D_ARRAY_INDEX, extra_EXT_texture_array_es3" ],
+  [ "MAX_ARRAY_TEXTURE_LAYERS_EXT", "CONTEXT_INT(Const.MaxArrayTextureLayers), extra_EXT_texture_array_es3" ],
 
 # GL_EXT_transform_feedback
   [ "TRANSFORM_FEEDBACK_BUFFER_BINDING", "LOC_CUSTOM, TYPE_INT, 0, extra_EXT_transform_feedback" ],
@@ -504,10 +504,8 @@ descriptor=[
   [ "STEREO", "BUFFER_INT(Visual.stereoMode), NO_EXTRA" ],
   [ "TEXTURE_1D", "LOC_CUSTOM, TYPE_BOOLEAN, NO_OFFSET, NO_EXTRA" ],
   [ "TEXTURE_3D", "LOC_CUSTOM, TYPE_BOOLEAN, NO_OFFSET, NO_EXTRA" ],
-  [ "TEXTURE_1D_ARRAY_EXT", "LOC_CUSTOM, TYPE_BOOLEAN, NO_OFFSET, NO_EXTRA" ],
-  [ "TEXTURE_2D_ARRAY_EXT", "LOC_CUSTOM, TYPE_BOOLEAN, NO_OFFSET, NO_EXTRA" ],
   [ "TEXTURE_BINDING_1D", "LOC_CUSTOM, TYPE_INT, TEXTURE_1D_INDEX, NO_EXTRA" ],
-  [ "TEXTURE_BINDING_1D_ARRAY", "LOC_CUSTOM, TYPE_INT, TEXTURE_1D_ARRAY_INDEX, extra_MESA_texture_array" ],
+  [ "TEXTURE_BINDING_1D_ARRAY", "LOC_CUSTOM, TYPE_INT, TEXTURE_1D_ARRAY_INDEX, extra_EXT_texture_array" ],
   [ "TEXTURE_GEN_S", "LOC_TEXUNIT, TYPE_BIT_0, offsetof(struct gl_texture_unit, TexGenEnabled), NO_EXTRA" ],
   [ "TEXTURE_GEN_T", "LOC_TEXUNIT, TYPE_BIT_1, offsetof(struct gl_texture_unit, TexGenEnabled), NO_EXTRA" ],
   [ "TEXTURE_GEN_R", "LOC_TEXUNIT, TYPE_BIT_2, offsetof(struct gl_texture_unit, TexGenEnabled), NO_EXTRA" ],
diff --git a/mesalib/src/mesa/main/getstring.c b/mesalib/src/mesa/main/getstring.c
index d8189115a..b66e24788 100644
--- a/mesalib/src/mesa/main/getstring.c
+++ b/mesalib/src/mesa/main/getstring.c
@@ -42,8 +42,6 @@ shading_language_version(struct gl_context *ctx)
    case API_OPENGL_COMPAT:
    case API_OPENGL_CORE:
       switch (ctx->Const.GLSLVersion) {
-      case 110:
-         return (const GLubyte *) "1.10";
       case 120:
          return (const GLubyte *) "1.20";
       case 130:
diff --git a/mesalib/src/mesa/main/glformats.c b/mesalib/src/mesa/main/glformats.c
index 740faa890..bec7a9bbb 100644
--- a/mesalib/src/mesa/main/glformats.c
+++ b/mesalib/src/mesa/main/glformats.c
@@ -1301,9 +1301,6 @@ _mesa_error_check_format_and_type(const struct gl_context *ctx,
       return GL_INVALID_OPERATION;
 
    case GL_UNSIGNED_INT_24_8:
-      if (!ctx->Extensions.EXT_packed_depth_stencil) {
-         return GL_INVALID_ENUM;
-      }
       if (format != GL_DEPTH_STENCIL) {
          return GL_INVALID_OPERATION;
       }
@@ -1484,9 +1481,8 @@ _mesa_error_check_format_and_type(const struct gl_context *ctx,
          else
             return GL_INVALID_OPERATION;
 
-      case GL_DEPTH_STENCIL_EXT:
-         if (ctx->Extensions.EXT_packed_depth_stencil &&
-             type == GL_UNSIGNED_INT_24_8)
+      case GL_DEPTH_STENCIL:
+         if (type == GL_UNSIGNED_INT_24_8)
             return GL_NO_ERROR;
          else if (ctx->Extensions.ARB_depth_buffer_float &&
              type == GL_FLOAT_32_UNSIGNED_INT_24_8_REV)
@@ -1694,8 +1690,6 @@ GLenum
 _mesa_es3_error_check_format_and_type(GLenum format, GLenum type,
                                       GLenum internalFormat)
 {
-   GLboolean type_valid = GL_TRUE;
-
    switch (format) {
    case GL_RGBA:
       switch (type) {
@@ -2116,5 +2110,5 @@ _mesa_es3_error_check_format_and_type(GLenum format, GLenum type,
       break;
    }
 
-   return type_valid ? GL_NO_ERROR : GL_INVALID_OPERATION;
+   return GL_NO_ERROR;
 }
diff --git a/mesalib/src/mesa/main/imports.c b/mesalib/src/mesa/main/imports.c
index 277e9476a..4afe156b0 100644
--- a/mesalib/src/mesa/main/imports.c
+++ b/mesalib/src/mesa/main/imports.c
@@ -168,6 +168,8 @@ _mesa_align_calloc(size_t bytes, unsigned long alignment)
  * \param ptr pointer to the memory to be freed.
  * The actual address to free is stored in the word immediately before the
  * address the client sees.
+ * Note that it is legal to pass NULL pointer to this function and will be
+ * handled accordingly.
  */
 void
 _mesa_align_free(void *ptr)
@@ -177,9 +179,11 @@ _mesa_align_free(void *ptr)
 #elif defined(_WIN32) && defined(_MSC_VER)
    _aligned_free(ptr);
 #else
-   void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
-   void *realAddr = *cubbyHole;
-   free(realAddr);
+   if (ptr) {
+      void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
+      void *realAddr = *cubbyHole;
+      free(realAddr);
+   }
 #endif /* defined(HAVE_POSIX_MEMALIGN) */
 }
 
@@ -199,8 +203,8 @@ _mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
    if (newBuf && oldBuffer && copySize > 0) {
       memcpy(newBuf, oldBuffer, copySize);
    }
-   if (oldBuffer)
-      _mesa_align_free(oldBuffer);
+
+   _mesa_align_free(oldBuffer);
    return newBuf;
 #endif
 }
diff --git a/mesalib/src/mesa/main/mipmap.c b/mesalib/src/mesa/main/mipmap.c
index 180f89116..033015780 100644
--- a/mesalib/src/mesa/main/mipmap.c
+++ b/mesalib/src/mesa/main/mipmap.c
@@ -1767,8 +1767,8 @@ _mesa_generate_mipmap_level(GLenum target,
  * compute next (level+1) image size
  * \return GL_FALSE if no smaller size can be generated (eg. src is 1x1x1 size)
  */
-static GLboolean
-next_mipmap_level_size(GLenum target, GLint border,
+GLboolean
+_mesa_next_mipmap_level_size(GLenum target, GLint border,
                        GLint srcWidth, GLint srcHeight, GLint srcDepth,
                        GLint *dstWidth, GLint *dstHeight, GLint *dstDepth)
 {
@@ -1911,7 +1911,7 @@ generate_mipmap_uncompressed(struct gl_context *ctx, GLenum target,
       srcDepth = srcImage->Depth;
       border = srcImage->Border;
 
-      nextLevel = next_mipmap_level_size(target, border,
+      nextLevel = _mesa_next_mipmap_level_size(target, border,
                                          srcWidth, srcHeight, srcDepth,
                                          &dstWidth, &dstHeight, &dstDepth);
       if (!nextLevel)
@@ -2102,7 +2102,7 @@ generate_mipmap_compressed(struct gl_context *ctx, GLenum target,
       srcDepth = srcImage->Depth;
       border = srcImage->Border;
 
-      nextLevel = next_mipmap_level_size(target, border,
+      nextLevel = _mesa_next_mipmap_level_size(target, border,
                                          srcWidth, srcHeight, srcDepth,
                                          &dstWidth, &dstHeight, &dstDepth);
       if (!nextLevel)
diff --git a/mesalib/src/mesa/main/mipmap.h b/mesalib/src/mesa/main/mipmap.h
index d5bd1d83d..ee91df006 100644
--- a/mesalib/src/mesa/main/mipmap.h
+++ b/mesalib/src/mesa/main/mipmap.h
@@ -51,5 +51,9 @@ extern void
 _mesa_generate_mipmap(struct gl_context *ctx, GLenum target,
                       struct gl_texture_object *texObj);
 
+extern GLboolean
+_mesa_next_mipmap_level_size(GLenum target, GLint border,
+                       GLint srcWidth, GLint srcHeight, GLint srcDepth,
+                       GLint *dstWidth, GLint *dstHeight, GLint *dstDepth);
 
 #endif /* MIPMAP_H */
diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h
index b4b432f40..f93bb5641 100644
--- a/mesalib/src/mesa/main/mtypes.h
+++ b/mesalib/src/mesa/main/mtypes.h
@@ -280,6 +280,7 @@ typedef enum
  */
 #define SYSTEM_BIT_SAMPLE_ID BITFIELD64_BIT(SYSTEM_VALUE_SAMPLE_ID)
 #define SYSTEM_BIT_SAMPLE_POS BITFIELD64_BIT(SYSTEM_VALUE_SAMPLE_POS)
+#define SYSTEM_BIT_SAMPLE_MASK_IN BITFIELD64_BIT(SYSTEM_VALUE_SAMPLE_MASK_IN)
 
 /**
  * Determine if the given gl_varying_slot appears in the fragment shader.
@@ -1193,6 +1194,11 @@ struct gl_texture_object
                                     pressure? */
    GLboolean Immutable;        /**< GL_ARB_texture_storage */
 
+   GLuint MinLevel;            /**< GL_ARB_texture_view */
+   GLuint MinLayer;            /**< GL_ARB_texture_view */
+   GLuint NumLevels;           /**< GL_ARB_texture_view */
+   GLuint NumLayers;           /**< GL_ARB_texture_view */
+
    /** Actual texture images, indexed by [cube face] and [mipmap level] */
    struct gl_texture_image *Image[MAX_FACES][MAX_TEXTURE_LEVELS];
 
@@ -1968,12 +1974,13 @@ typedef enum
  */
 typedef enum
 {
-   SYSTEM_VALUE_FRONT_FACE,  /**< Fragment shader only (not done yet) */
-   SYSTEM_VALUE_VERTEX_ID,   /**< Vertex shader only */
-   SYSTEM_VALUE_INSTANCE_ID, /**< Vertex shader only */
-   SYSTEM_VALUE_SAMPLE_ID,   /**< Fragment shader only */
-   SYSTEM_VALUE_SAMPLE_POS,  /**< Fragment shader only */
-   SYSTEM_VALUE_MAX          /**< Number of values */
+   SYSTEM_VALUE_FRONT_FACE,     /**< Fragment shader only (not done yet) */
+   SYSTEM_VALUE_VERTEX_ID,      /**< Vertex shader only */
+   SYSTEM_VALUE_INSTANCE_ID,    /**< Vertex shader only */
+   SYSTEM_VALUE_SAMPLE_ID,      /**< Fragment shader only */
+   SYSTEM_VALUE_SAMPLE_POS,     /**< Fragment shader only */
+   SYSTEM_VALUE_SAMPLE_MASK_IN, /**< Fragment shader only */
+   SYSTEM_VALUE_MAX             /**< Number of values */
 } gl_system_value;
 
 
@@ -2132,6 +2139,12 @@ struct gl_fragment_program
     * uses centroid interpolation, 0 otherwise.  Unused inputs are 0.
     */
    GLbitfield64 IsCentroid;
+
+   /**
+    * Bitfield indicating, for each fragment shader input, 1 if that input
+    * uses sample interpolation, 0 otherwise.  Unused inputs are 0.
+    */
+   GLbitfield64 IsSample;
 };
 
 
@@ -2352,9 +2365,7 @@ struct gl_shader
    struct exec_list *ir;
    struct glsl_symbol_table *symbols;
 
-   /** Shaders containing built-in functions that are used for linking. */
-   struct gl_shader *builtins_to_link[16];
-   unsigned num_builtins_to_link;
+   bool uses_builtin_functions;
 
    /**
     * Geometry shader state from GLSL 1.50 layout qualifiers.
@@ -2388,9 +2399,10 @@ typedef enum
    MESA_SHADER_VERTEX = 0,
    MESA_SHADER_GEOMETRY = 1,
    MESA_SHADER_FRAGMENT = 2,
-   MESA_SHADER_TYPES = 3
 } gl_shader_type;
 
+#define MESA_SHADER_TYPES (MESA_SHADER_FRAGMENT + 1)
+
 
 struct gl_uniform_buffer_variable
 {
@@ -3374,6 +3386,7 @@ struct gl_extensions
    GLboolean ARB_texture_query_lod;
    GLboolean ARB_texture_rg;
    GLboolean ARB_texture_rgb10_a2ui;
+   GLboolean ARB_texture_view;
    GLboolean ARB_timer_query;
    GLboolean ARB_transform_feedback2;
    GLboolean ARB_transform_feedback3;
@@ -3395,7 +3408,6 @@ struct gl_extensions
    GLboolean EXT_framebuffer_sRGB;
    GLboolean EXT_gpu_program_parameters;
    GLboolean EXT_gpu_shader4;
-   GLboolean EXT_packed_depth_stencil;
    GLboolean EXT_packed_float;
    GLboolean EXT_pixel_buffer_object;
    GLboolean EXT_point_parameters;
@@ -3432,7 +3444,6 @@ struct gl_extensions
    GLboolean ATI_fragment_shader;
    GLboolean ATI_separate_stencil;
    GLboolean MESA_pack_invert;
-   GLboolean MESA_texture_array;
    GLboolean MESA_ycbcr_texture;
    GLboolean NV_conditional_render;
    GLboolean NV_fog_distance;
diff --git a/mesalib/src/mesa/main/performance_monitor.c b/mesalib/src/mesa/main/performance_monitor.c
index 4981e6fb2..e62f77012 100644
--- a/mesalib/src/mesa/main/performance_monitor.c
+++ b/mesalib/src/mesa/main/performance_monitor.c
@@ -93,6 +93,25 @@ fail:
    return NULL;
 }
 
+static void
+free_performance_monitor(GLuint key, void *data, void *user)
+{
+   struct gl_perf_monitor_object *m = data;
+   struct gl_context *ctx = user;
+
+   ralloc_free(m->ActiveGroups);
+   ralloc_free(m->ActiveCounters);
+   ctx->Driver.DeletePerfMonitor(ctx, m);
+}
+
+void
+_mesa_free_performance_monitors(struct gl_context *ctx)
+{
+   _mesa_HashDeleteAll(ctx->PerfMonitor.Monitors,
+                       free_performance_monitor, ctx);
+   _mesa_DeleteHashTable(ctx->PerfMonitor.Monitors);
+}
+
 static inline struct gl_perf_monitor_object *
 lookup_monitor(struct gl_context *ctx, GLuint id)
 {
diff --git a/mesalib/src/mesa/main/performance_monitor.h b/mesalib/src/mesa/main/performance_monitor.h
index a852a4184..76234e5c1 100644
--- a/mesalib/src/mesa/main/performance_monitor.h
+++ b/mesalib/src/mesa/main/performance_monitor.h
@@ -35,6 +35,9 @@
 extern void
 _mesa_init_performance_monitors(struct gl_context *ctx);
 
+extern void
+_mesa_free_performance_monitors(struct gl_context *ctx);
+
 extern void GLAPIENTRY
 _mesa_GetPerfMonitorGroupsAMD(GLint *numGroups, GLsizei groupsSize,
                               GLuint *groups);
diff --git a/mesalib/src/mesa/main/shader_query.cpp b/mesalib/src/mesa/main/shader_query.cpp
index 3014a9778..f14e1a562 100644
--- a/mesalib/src/mesa/main/shader_query.cpp
+++ b/mesalib/src/mesa/main/shader_query.cpp
@@ -106,8 +106,8 @@ _mesa_GetActiveAttrib(GLhandleARB program, GLuint desired_index,
       const ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
       if (var == NULL
-	  || var->mode != ir_var_shader_in
-	  || var->location == -1)
+	  || var->data.mode != ir_var_shader_in
+	  || var->data.location == -1)
 	 continue;
 
       if (current_index == desired_index) {
@@ -169,13 +169,13 @@ _mesa_GetAttribLocation(GLhandleARB program, const GLcharARB * name)
        *     attribute, or if an error occurs, -1 will be returned."
        */
       if (var == NULL
-	  || var->mode != ir_var_shader_in
-	  || var->location == -1
-	  || var->location < VERT_ATTRIB_GENERIC0)
+	  || var->data.mode != ir_var_shader_in
+	  || var->data.location == -1
+	  || var->data.location < VERT_ATTRIB_GENERIC0)
 	 continue;
 
       if (strcmp(var->name, name) == 0)
-	 return var->location - VERT_ATTRIB_GENERIC0;
+	 return var->data.location - VERT_ATTRIB_GENERIC0;
    }
 
    return -1;
@@ -197,8 +197,8 @@ _mesa_count_active_attribs(struct gl_shader_program *shProg)
       const ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
       if (var == NULL
-	  || var->mode != ir_var_shader_in
-	  || var->location == -1)
+	  || var->data.mode != ir_var_shader_in
+	  || var->data.location == -1)
 	 continue;
 
       i++;
@@ -223,8 +223,8 @@ _mesa_longest_attribute_name_length(struct gl_shader_program *shProg)
       const ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
       if (var == NULL
-	  || var->mode != ir_var_shader_in
-	  || var->location == -1)
+	  || var->data.mode != ir_var_shader_in
+	  || var->data.location == -1)
 	 continue;
 
       const size_t len = strlen(var->name);
@@ -333,13 +333,13 @@ _mesa_GetFragDataIndex(GLuint program, const GLchar *name)
        *     attribute, or if an error occurs, -1 will be returned."
        */
       if (var == NULL
-          || var->mode != ir_var_shader_out
-          || var->location == -1
-          || var->location < FRAG_RESULT_DATA0)
+          || var->data.mode != ir_var_shader_out
+          || var->data.location == -1
+          || var->data.location < FRAG_RESULT_DATA0)
          continue;
 
       if (strcmp(var->name, name) == 0)
-         return var->index;
+         return var->data.index;
    }
 
    return -1;
@@ -389,13 +389,13 @@ _mesa_GetFragDataLocation(GLuint program, const GLchar *name)
        *     attribute, or if an error occurs, -1 will be returned."
        */
       if (var == NULL
-	  || var->mode != ir_var_shader_out
-	  || var->location == -1
-	  || var->location < FRAG_RESULT_DATA0)
+	  || var->data.mode != ir_var_shader_out
+	  || var->data.location == -1
+	  || var->data.location < FRAG_RESULT_DATA0)
 	 continue;
 
       if (strcmp(var->name, name) == 0)
-	 return var->location - FRAG_RESULT_DATA0;
+	 return var->data.location - FRAG_RESULT_DATA0;
    }
 
    return -1;
diff --git a/mesalib/src/mesa/main/shaderapi.c b/mesalib/src/mesa/main/shaderapi.c
index 1d9aac39d..4f3be68a4 100644
--- a/mesalib/src/mesa/main/shaderapi.c
+++ b/mesalib/src/mesa/main/shaderapi.c
@@ -414,7 +414,7 @@ detach_shader(struct gl_context *ctx, GLuint program, GLuint shader)
          err = GL_INVALID_OPERATION;
       else
          err = GL_INVALID_VALUE;
-      _mesa_error(ctx, err, "glDetachProgram(shader)");
+      _mesa_error(ctx, err, "glDetachShader(shader)");
       return;
    }
 }
diff --git a/mesalib/src/mesa/main/shaderobj.h b/mesalib/src/mesa/main/shaderobj.h
index de1c9fcaf..aff178f26 100644
--- a/mesalib/src/mesa/main/shaderobj.h
+++ b/mesalib/src/mesa/main/shaderobj.h
@@ -113,7 +113,7 @@ _mesa_shader_type_to_index(GLenum v)
       return MESA_SHADER_GEOMETRY;
    default:
       ASSERT(0 && "bad value in _mesa_shader_type_to_index()");
-      return MESA_SHADER_TYPES;
+      return MESA_SHADER_VERTEX;
    }
 }
 
diff --git a/mesalib/src/mesa/main/texgetimage.c b/mesalib/src/mesa/main/texgetimage.c
index d66ca1abc..200d29c2f 100644
--- a/mesalib/src/mesa/main/texgetimage.c
+++ b/mesalib/src/mesa/main/texgetimage.c
@@ -759,8 +759,7 @@ legal_getteximage_target(struct gl_context *ctx, GLenum target)
       return ctx->Extensions.NV_texture_rectangle;
    case GL_TEXTURE_1D_ARRAY_EXT:
    case GL_TEXTURE_2D_ARRAY_EXT:
-      return (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array);
+      return ctx->Extensions.EXT_texture_array;
    case GL_TEXTURE_CUBE_MAP_ARRAY:
       return ctx->Extensions.ARB_texture_cube_map_array;
    default:
diff --git a/mesalib/src/mesa/main/teximage.c b/mesalib/src/mesa/main/teximage.c
index 793c5d382..9c3f1e86e 100644
--- a/mesalib/src/mesa/main/teximage.c
+++ b/mesalib/src/mesa/main/teximage.c
@@ -48,6 +48,7 @@
 #include "texobj.h"
 #include "texstate.h"
 #include "texstorage.h"
+#include "textureview.h"
 #include "mtypes.h"
 #include "glformats.h"
 
@@ -300,14 +301,12 @@ _mesa_base_tex_format( struct gl_context *ctx, GLint internalFormat )
       }
    }
 
-   if (ctx->Extensions.EXT_packed_depth_stencil) {
-      switch (internalFormat) {
-         case GL_DEPTH_STENCIL_EXT:
-         case GL_DEPTH24_STENCIL8_EXT:
-            return GL_DEPTH_STENCIL_EXT;
-         default:
-            ; /* fallthrough */
-      }
+   switch (internalFormat) {
+   case GL_DEPTH_STENCIL:
+   case GL_DEPTH24_STENCIL8:
+      return GL_DEPTH_STENCIL;
+   default:
+      ; /* fallthrough */
    }
 
    if (ctx->Extensions.EXT_texture_sRGB) {
@@ -752,8 +751,7 @@ _mesa_select_tex_object(struct gl_context *ctx,
                         const struct gl_texture_unit *texUnit,
                         GLenum target)
 {
-   const GLboolean arrayTex = (ctx->Extensions.MESA_texture_array ||
-                               ctx->Extensions.EXT_texture_array);
+   const GLboolean arrayTex = ctx->Extensions.EXT_texture_array;
 
    switch (target) {
       case GL_TEXTURE_1D:
@@ -1018,8 +1016,7 @@ _mesa_max_texture_levels(struct gl_context *ctx, GLenum target)
    case GL_PROXY_TEXTURE_1D_ARRAY_EXT:
    case GL_TEXTURE_2D_ARRAY_EXT:
    case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
-      return (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array)
+      return ctx->Extensions.EXT_texture_array
          ? ctx->Const.MaxTextureLevels : 0;
    case GL_TEXTURE_CUBE_MAP_ARRAY:
    case GL_PROXY_TEXTURE_CUBE_MAP_ARRAY:
@@ -1726,8 +1723,7 @@ target_can_be_compressed(const struct gl_context *ctx, GLenum target,
       return ctx->Extensions.ARB_texture_cube_map;
    case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
    case GL_TEXTURE_2D_ARRAY_EXT:
-      return (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array);
+      return ctx->Extensions.EXT_texture_array;
    case GL_PROXY_TEXTURE_CUBE_MAP_ARRAY:
    case GL_TEXTURE_CUBE_MAP_ARRAY:
       return ctx->Extensions.ARB_texture_cube_map_array;
@@ -1775,9 +1771,7 @@ legal_teximage_target(struct gl_context *ctx, GLuint dims, GLenum target)
             && ctx->Extensions.NV_texture_rectangle;
       case GL_TEXTURE_1D_ARRAY_EXT:
       case GL_PROXY_TEXTURE_1D_ARRAY_EXT:
-         return _mesa_is_desktop_gl(ctx)
-            && (ctx->Extensions.MESA_texture_array ||
-                ctx->Extensions.EXT_texture_array);
+         return _mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array;
       default:
          return GL_FALSE;
       }
@@ -1788,14 +1782,10 @@ legal_teximage_target(struct gl_context *ctx, GLuint dims, GLenum target)
       case GL_PROXY_TEXTURE_3D:
          return _mesa_is_desktop_gl(ctx);
       case GL_TEXTURE_2D_ARRAY_EXT:
-         return (_mesa_is_desktop_gl(ctx)
-                 && (ctx->Extensions.MESA_texture_array ||
-                     ctx->Extensions.EXT_texture_array))
+         return (_mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array)
             || _mesa_is_gles3(ctx);
       case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
-         return _mesa_is_desktop_gl(ctx)
-            && (ctx->Extensions.MESA_texture_array ||
-                ctx->Extensions.EXT_texture_array);
+         return _mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array;
       case GL_TEXTURE_CUBE_MAP_ARRAY:
       case GL_PROXY_TEXTURE_CUBE_MAP_ARRAY:
          return ctx->Extensions.ARB_texture_cube_map_array;
@@ -1836,9 +1826,7 @@ legal_texsubimage_target(struct gl_context *ctx, GLuint dims, GLenum target)
          return _mesa_is_desktop_gl(ctx)
             && ctx->Extensions.NV_texture_rectangle;
       case GL_TEXTURE_1D_ARRAY_EXT:
-         return _mesa_is_desktop_gl(ctx)
-            && (ctx->Extensions.MESA_texture_array ||
-                ctx->Extensions.EXT_texture_array);
+         return _mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array;
       default:
          return GL_FALSE;
       }
@@ -1847,9 +1835,7 @@ legal_texsubimage_target(struct gl_context *ctx, GLuint dims, GLenum target)
       case GL_TEXTURE_3D:
          return GL_TRUE;
       case GL_TEXTURE_2D_ARRAY_EXT:
-         return (_mesa_is_desktop_gl(ctx)
-                 && (ctx->Extensions.MESA_texture_array ||
-                     ctx->Extensions.EXT_texture_array))
+         return (_mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array)
             || _mesa_is_gles3(ctx);
       case GL_TEXTURE_CUBE_MAP_ARRAY:
       case GL_PROXY_TEXTURE_CUBE_MAP_ARRAY:
@@ -2870,7 +2856,7 @@ _mesa_choose_texture_format(struct gl_context *ctx,
    }
 
    /* If the application requested compression to an S3TC format but we don't
-    * have the DTXn library, force a generic compressed format instead.
+    * have the DXTn library, force a generic compressed format instead.
     */
    if (internalFormat != format && format != GL_NONE) {
       const GLenum before = internalFormat;
@@ -3892,87 +3878,108 @@ _mesa_CompressedTexSubImage3D(GLenum target, GLint level, GLint xoffset,
 static gl_format
 get_texbuffer_format(const struct gl_context *ctx, GLenum internalFormat)
 {
+   if (ctx->API != API_OPENGL_CORE) {
+      switch (internalFormat) {
+      case GL_ALPHA8:
+         return MESA_FORMAT_A8;
+      case GL_ALPHA16:
+         return MESA_FORMAT_A16;
+      case GL_ALPHA16F_ARB:
+         return MESA_FORMAT_ALPHA_FLOAT16;
+      case GL_ALPHA32F_ARB:
+         return MESA_FORMAT_ALPHA_FLOAT32;
+      case GL_ALPHA8I_EXT:
+         return MESA_FORMAT_ALPHA_INT8;
+      case GL_ALPHA16I_EXT:
+         return MESA_FORMAT_ALPHA_INT16;
+      case GL_ALPHA32I_EXT:
+         return MESA_FORMAT_ALPHA_INT32;
+      case GL_ALPHA8UI_EXT:
+         return MESA_FORMAT_ALPHA_UINT8;
+      case GL_ALPHA16UI_EXT:
+         return MESA_FORMAT_ALPHA_UINT16;
+      case GL_ALPHA32UI_EXT:
+         return MESA_FORMAT_ALPHA_UINT32;
+      case GL_LUMINANCE8:
+         return MESA_FORMAT_L8;
+      case GL_LUMINANCE16:
+         return MESA_FORMAT_L16;
+      case GL_LUMINANCE16F_ARB:
+         return MESA_FORMAT_LUMINANCE_FLOAT16;
+      case GL_LUMINANCE32F_ARB:
+         return MESA_FORMAT_LUMINANCE_FLOAT32;
+      case GL_LUMINANCE8I_EXT:
+         return MESA_FORMAT_LUMINANCE_INT8;
+      case GL_LUMINANCE16I_EXT:
+         return MESA_FORMAT_LUMINANCE_INT16;
+      case GL_LUMINANCE32I_EXT:
+         return MESA_FORMAT_LUMINANCE_INT32;
+      case GL_LUMINANCE8UI_EXT:
+         return MESA_FORMAT_LUMINANCE_UINT8;
+      case GL_LUMINANCE16UI_EXT:
+         return MESA_FORMAT_LUMINANCE_UINT16;
+      case GL_LUMINANCE32UI_EXT:
+         return MESA_FORMAT_LUMINANCE_UINT32;
+      case GL_LUMINANCE8_ALPHA8:
+         return MESA_FORMAT_AL88;
+      case GL_LUMINANCE16_ALPHA16:
+         return MESA_FORMAT_AL1616;
+      case GL_LUMINANCE_ALPHA16F_ARB:
+         return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16;
+      case GL_LUMINANCE_ALPHA32F_ARB:
+         return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32;
+      case GL_LUMINANCE_ALPHA8I_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_INT8;
+      case GL_LUMINANCE_ALPHA16I_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_INT8;
+      case GL_LUMINANCE_ALPHA32I_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_INT16;
+      case GL_LUMINANCE_ALPHA8UI_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_UINT8;
+      case GL_LUMINANCE_ALPHA16UI_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_UINT16;
+      case GL_LUMINANCE_ALPHA32UI_EXT:
+         return MESA_FORMAT_LUMINANCE_ALPHA_UINT32;
+      case GL_INTENSITY8:
+         return MESA_FORMAT_I8;
+      case GL_INTENSITY16:
+         return MESA_FORMAT_I16;
+      case GL_INTENSITY16F_ARB:
+         return MESA_FORMAT_INTENSITY_FLOAT16;
+      case GL_INTENSITY32F_ARB:
+         return MESA_FORMAT_INTENSITY_FLOAT32;
+      case GL_INTENSITY8I_EXT:
+         return MESA_FORMAT_INTENSITY_INT8;
+      case GL_INTENSITY16I_EXT:
+         return MESA_FORMAT_INTENSITY_INT16;
+      case GL_INTENSITY32I_EXT:
+         return MESA_FORMAT_INTENSITY_INT32;
+      case GL_INTENSITY8UI_EXT:
+         return MESA_FORMAT_INTENSITY_UINT8;
+      case GL_INTENSITY16UI_EXT:
+         return MESA_FORMAT_INTENSITY_UINT16;
+      case GL_INTENSITY32UI_EXT:
+         return MESA_FORMAT_INTENSITY_UINT32;
+      default:
+         break;
+      }
+   }
+
+   if (ctx->API == API_OPENGL_CORE &&
+       ctx->Extensions.ARB_texture_buffer_object_rgb32) {
+      switch (internalFormat) {
+      case GL_RGB32F:
+         return MESA_FORMAT_RGB_FLOAT32;
+      case GL_RGB32UI:
+         return MESA_FORMAT_RGB_UINT32;
+      case GL_RGB32I:
+         return MESA_FORMAT_RGB_INT32;
+      default:
+         break;
+      }
+   }
+
    switch (internalFormat) {
-   case GL_ALPHA8:
-      return MESA_FORMAT_A8;
-   case GL_ALPHA16:
-      return MESA_FORMAT_A16;
-   case GL_ALPHA16F_ARB:
-      return MESA_FORMAT_ALPHA_FLOAT16;
-   case GL_ALPHA32F_ARB:
-      return MESA_FORMAT_ALPHA_FLOAT32;
-   case GL_ALPHA8I_EXT:
-      return MESA_FORMAT_ALPHA_INT8;
-   case GL_ALPHA16I_EXT:
-      return MESA_FORMAT_ALPHA_INT16;
-   case GL_ALPHA32I_EXT:
-      return MESA_FORMAT_ALPHA_INT32;
-   case GL_ALPHA8UI_EXT:
-      return MESA_FORMAT_ALPHA_UINT8;
-   case GL_ALPHA16UI_EXT:
-      return MESA_FORMAT_ALPHA_UINT16;
-   case GL_ALPHA32UI_EXT:
-      return MESA_FORMAT_ALPHA_UINT32;
-   case GL_LUMINANCE8:
-      return MESA_FORMAT_L8;
-   case GL_LUMINANCE16:
-      return MESA_FORMAT_L16;
-   case GL_LUMINANCE16F_ARB:
-      return MESA_FORMAT_LUMINANCE_FLOAT16;
-   case GL_LUMINANCE32F_ARB:
-      return MESA_FORMAT_LUMINANCE_FLOAT32;
-   case GL_LUMINANCE8I_EXT:
-      return MESA_FORMAT_LUMINANCE_INT8;
-   case GL_LUMINANCE16I_EXT:
-      return MESA_FORMAT_LUMINANCE_INT16;
-   case GL_LUMINANCE32I_EXT:
-      return MESA_FORMAT_LUMINANCE_INT32;
-   case GL_LUMINANCE8UI_EXT:
-      return MESA_FORMAT_LUMINANCE_UINT8;
-   case GL_LUMINANCE16UI_EXT:
-      return MESA_FORMAT_LUMINANCE_UINT16;
-   case GL_LUMINANCE32UI_EXT:
-      return MESA_FORMAT_LUMINANCE_UINT32;
-   case GL_LUMINANCE8_ALPHA8:
-      return MESA_FORMAT_AL88;
-   case GL_LUMINANCE16_ALPHA16:
-      return MESA_FORMAT_AL1616;
-   case GL_LUMINANCE_ALPHA16F_ARB:
-      return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16;
-   case GL_LUMINANCE_ALPHA32F_ARB:
-      return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32;
-   case GL_LUMINANCE_ALPHA8I_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_INT8;
-   case GL_LUMINANCE_ALPHA16I_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_INT8;
-   case GL_LUMINANCE_ALPHA32I_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_INT16;
-   case GL_LUMINANCE_ALPHA8UI_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_UINT8;
-   case GL_LUMINANCE_ALPHA16UI_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_UINT16;
-   case GL_LUMINANCE_ALPHA32UI_EXT:
-      return MESA_FORMAT_LUMINANCE_ALPHA_UINT32;
-   case GL_INTENSITY8:
-      return MESA_FORMAT_I8;
-   case GL_INTENSITY16:
-      return MESA_FORMAT_I16;
-   case GL_INTENSITY16F_ARB:
-      return MESA_FORMAT_INTENSITY_FLOAT16;
-   case GL_INTENSITY32F_ARB:
-      return MESA_FORMAT_INTENSITY_FLOAT32;
-   case GL_INTENSITY8I_EXT:
-      return MESA_FORMAT_INTENSITY_INT8;
-   case GL_INTENSITY16I_EXT:
-      return MESA_FORMAT_INTENSITY_INT16;
-   case GL_INTENSITY32I_EXT:
-      return MESA_FORMAT_INTENSITY_INT32;
-   case GL_INTENSITY8UI_EXT:
-      return MESA_FORMAT_INTENSITY_UINT8;
-   case GL_INTENSITY16UI_EXT:
-      return MESA_FORMAT_INTENSITY_UINT16;
-   case GL_INTENSITY32UI_EXT:
-      return MESA_FORMAT_INTENSITY_UINT32;
    case GL_RGBA8:
       return MESA_FORMAT_RGBA8888_REV;
    case GL_RGBA16:
@@ -4036,21 +4043,15 @@ get_texbuffer_format(const struct gl_context *ctx, GLenum internalFormat)
    case GL_R32UI:
       return MESA_FORMAT_R_UINT32;
 
-   case GL_RGB32F:
-      return MESA_FORMAT_RGB_FLOAT32;
-   case GL_RGB32UI:
-      return MESA_FORMAT_RGB_UINT32;
-   case GL_RGB32I:
-      return MESA_FORMAT_RGB_INT32;
-
    default:
       return MESA_FORMAT_NONE;
    }
 }
 
 
-static gl_format
-validate_texbuffer_format(const struct gl_context *ctx, GLenum internalFormat)
+gl_format
+_mesa_validate_texbuffer_format(const struct gl_context *ctx,
+                                GLenum internalFormat)
 {
    gl_format format = get_texbuffer_format(ctx, internalFormat);
    GLenum datatype;
@@ -4065,15 +4066,10 @@ validate_texbuffer_format(const struct gl_context *ctx, GLenum internalFormat)
    if (datatype == GL_HALF_FLOAT && !ctx->Extensions.ARB_half_float_pixel)
       return MESA_FORMAT_NONE;
 
-   /* The GL_ARB_texture_rg and GL_ARB_texture_buffer_object specs don't make
-    * any mention of R/RG formats, but they appear in the GL 3.1 core
-    * specification.
-    */
-   if (ctx->Version <= 30) {
+   if (!ctx->Extensions.ARB_texture_rg) {
       GLenum base_format = _mesa_get_format_base_format(format);
-
       if (base_format == GL_R || base_format == GL_RG)
-	 return MESA_FORMAT_NONE;
+         return MESA_FORMAT_NONE;
    }
 
    if (!ctx->Extensions.ARB_texture_buffer_object_rgb32) {
@@ -4100,7 +4096,7 @@ texbufferrange(struct gl_context *ctx, GLenum target, GLenum internalFormat,
       return;
    }
 
-   format = validate_texbuffer_format(ctx, internalFormat);
+   format = _mesa_validate_texbuffer_format(ctx, internalFormat);
    if (format == MESA_FORMAT_NONE) {
       _mesa_error(ctx, GL_INVALID_ENUM, "glTexBuffer(internalFormat 0x%x)",
                   internalFormat);
@@ -4348,6 +4344,11 @@ teximagemultisample(GLuint dims, GLenum target, GLsizei samples,
       }
 
       texObj->Immutable = immutable;
+
+      if (immutable) {
+         _mesa_set_texture_view_state(ctx, texObj, target, 1);
+      }
+
       _mesa_update_fbo_texture(ctx, texObj, 0, 0);
    }
 }
diff --git a/mesalib/src/mesa/main/teximage.h b/mesalib/src/mesa/main/teximage.h
index 792383d2f..0b5786340 100644
--- a/mesalib/src/mesa/main/teximage.h
+++ b/mesalib/src/mesa/main/teximage.h
@@ -147,6 +147,10 @@ _mesa_legal_texture_dimensions(struct gl_context *ctx, GLenum target,
                                GLint level, GLint width, GLint height,
                                GLint depth, GLint border);
 
+extern gl_format
+_mesa_validate_texbuffer_format(const struct gl_context *ctx,
+                                GLenum internalFormat);
+
 /**
  * Lock a texture for updating.  See also _mesa_lock_context_textures().
  */
diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c
index abd30c563..d6510fefd 100644
--- a/mesalib/src/mesa/main/texobj.c
+++ b/mesalib/src/mesa/main/texobj.c
@@ -552,7 +552,7 @@ _mesa_test_texobj_completeness( const struct gl_context *ctx,
 
    t->_MaxLevel = MIN3(t->MaxLevel,
                        /* 'p' in the GL spec */
-                       baseLevel + baseImage->MaxNumLevels - 1,
+                       (int) (baseLevel + baseImage->MaxNumLevels - 1),
                        /* 'q' in the GL spec */
                        maxLevels - 1);
 
@@ -889,6 +889,8 @@ count_tex_size(GLuint key, void *data, void *userData)
       (const struct gl_texture_object *) data;
    GLuint *total = (GLuint *) userData;
 
+   (void) key;
+
    *total = *total + texture_size(texObj);
 }
 
@@ -1171,7 +1173,7 @@ _mesa_DeleteTextures( GLsizei n, const GLuint *textures)
  * \return TEXTURE_x_INDEX or -1 if target is invalid
  */
 static GLint
-target_enum_to_index(struct gl_context *ctx, GLenum target)
+target_enum_to_index(const struct gl_context *ctx, GLenum target)
 {
    switch (target) {
    case GL_TEXTURE_1D:
@@ -1179,25 +1181,21 @@ target_enum_to_index(struct gl_context *ctx, GLenum target)
    case GL_TEXTURE_2D:
       return TEXTURE_2D_INDEX;
    case GL_TEXTURE_3D:
-      return TEXTURE_3D_INDEX;
-   case GL_TEXTURE_CUBE_MAP_ARB:
+      return ctx->API != API_OPENGLES ? TEXTURE_3D_INDEX : -1;
+   case GL_TEXTURE_CUBE_MAP:
       return ctx->Extensions.ARB_texture_cube_map
          ? TEXTURE_CUBE_INDEX : -1;
-   case GL_TEXTURE_RECTANGLE_NV:
+   case GL_TEXTURE_RECTANGLE:
       return _mesa_is_desktop_gl(ctx) && ctx->Extensions.NV_texture_rectangle
          ? TEXTURE_RECT_INDEX : -1;
-   case GL_TEXTURE_1D_ARRAY_EXT:
-      return _mesa_is_desktop_gl(ctx)
-         && (ctx->Extensions.EXT_texture_array
-             || ctx->Extensions.MESA_texture_array)
+   case GL_TEXTURE_1D_ARRAY:
+      return _mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array
          ? TEXTURE_1D_ARRAY_INDEX : -1;
-   case GL_TEXTURE_2D_ARRAY_EXT:
-      return (_mesa_is_desktop_gl(ctx)
-              && (ctx->Extensions.EXT_texture_array
-                  || ctx->Extensions.MESA_texture_array))
+   case GL_TEXTURE_2D_ARRAY:
+      return (_mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array)
          || _mesa_is_gles3(ctx)
          ? TEXTURE_2D_ARRAY_INDEX : -1;
-   case GL_TEXTURE_BUFFER_ARB:
+   case GL_TEXTURE_BUFFER:
       return ctx->API == API_OPENGL_CORE &&
              ctx->Extensions.ARB_texture_buffer_object ?
              TEXTURE_BUFFER_INDEX : -1;
@@ -1205,7 +1203,8 @@ target_enum_to_index(struct gl_context *ctx, GLenum target)
       return _mesa_is_gles(ctx) && ctx->Extensions.OES_EGL_image_external
          ? TEXTURE_EXTERNAL_INDEX : -1;
    case GL_TEXTURE_CUBE_MAP_ARRAY:
-      return TEXTURE_CUBE_ARRAY_INDEX;
+      return _mesa_is_desktop_gl(ctx) && ctx->Extensions.ARB_texture_cube_map_array
+         ? TEXTURE_CUBE_ARRAY_INDEX : -1;
    case GL_TEXTURE_2D_MULTISAMPLE:
       return _mesa_is_desktop_gl(ctx) && ctx->Extensions.ARB_texture_multisample
          ? TEXTURE_2D_MULTISAMPLE_INDEX: -1;
diff --git a/mesalib/src/mesa/main/texparam.c b/mesalib/src/mesa/main/texparam.c
index 7092c630b..94e498d20 100644
--- a/mesalib/src/mesa/main/texparam.c
+++ b/mesalib/src/mesa/main/texparam.c
@@ -158,16 +158,13 @@ get_texobj(struct gl_context *ctx, GLenum target, GLboolean get)
       }
       break;
    case GL_TEXTURE_1D_ARRAY_EXT:
-      if (_mesa_is_desktop_gl(ctx)
-          && (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array)) {
+      if (_mesa_is_desktop_gl(ctx) && ctx->Extensions.EXT_texture_array) {
          return texUnit->CurrentTex[TEXTURE_1D_ARRAY_INDEX];
       }
       break;
    case GL_TEXTURE_2D_ARRAY_EXT:
       if ((_mesa_is_desktop_gl(ctx) || _mesa_is_gles3(ctx))
-          && (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array)) {
+          && ctx->Extensions.EXT_texture_array) {
          return texUnit->CurrentTex[TEXTURE_2D_ARRAY_INDEX];
       }
       break;
@@ -1046,8 +1043,7 @@ legal_get_tex_level_parameter_target(struct gl_context *ctx, GLenum target)
    case GL_PROXY_TEXTURE_1D_ARRAY_EXT:
    case GL_TEXTURE_2D_ARRAY_EXT:
    case GL_PROXY_TEXTURE_2D_ARRAY_EXT:
-      return (ctx->Extensions.MESA_texture_array ||
-              ctx->Extensions.EXT_texture_array);
+      return ctx->Extensions.EXT_texture_array;
    case GL_TEXTURE_BUFFER:
       /* GetTexLevelParameter accepts GL_TEXTURE_BUFFER in GL 3.1+ contexts,
        * but not in earlier versions that expose ARB_texture_buffer_object.
@@ -1166,10 +1162,7 @@ get_tex_level_parameter_image(struct gl_context *ctx,
             goto invalid_pname;
          *params = _mesa_get_format_bits(texFormat, pname);
          break;
-      case GL_TEXTURE_STENCIL_SIZE_EXT:
-         if (!ctx->Extensions.EXT_packed_depth_stencil &&
-             !ctx->Extensions.ARB_framebuffer_object)
-            goto invalid_pname;
+      case GL_TEXTURE_STENCIL_SIZE:
          *params = _mesa_get_format_bits(texFormat, pname);
          break;
       case GL_TEXTURE_SHARED_SIZE:
@@ -1562,9 +1555,35 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
          break;
 
       case GL_TEXTURE_IMMUTABLE_LEVELS:
-         if (!_mesa_is_gles3(ctx))
+         if (_mesa_is_gles3(ctx) ||
+             (_mesa_is_desktop_gl(ctx) && ctx->Extensions.ARB_texture_view))
+            *params = (GLfloat) obj->ImmutableLevels;
+         else
             goto invalid_pname;
-         *params = (GLfloat) obj->ImmutableLevels;
+         break;
+
+      case GL_TEXTURE_VIEW_MIN_LEVEL:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLfloat) obj->MinLevel;
+         break;
+
+      case GL_TEXTURE_VIEW_NUM_LEVELS:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLfloat) obj->NumLevels;
+         break;
+
+      case GL_TEXTURE_VIEW_MIN_LAYER:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLfloat) obj->MinLayer;
+         break;
+
+      case GL_TEXTURE_VIEW_NUM_LAYERS:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLfloat) obj->NumLayers;
          break;
 
       case GL_REQUIRED_TEXTURE_IMAGE_UNITS_OES:
@@ -1746,9 +1765,35 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
          break;
 
       case GL_TEXTURE_IMMUTABLE_LEVELS:
-         if (!_mesa_is_gles3(ctx))
+         if (_mesa_is_gles3(ctx) ||
+             (_mesa_is_desktop_gl(ctx) && ctx->Extensions.ARB_texture_view))
+            *params = obj->ImmutableLevels;
+         else
+            goto invalid_pname;
+         break;
+
+      case GL_TEXTURE_VIEW_MIN_LEVEL:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLint) obj->MinLevel;
+         break;
+
+      case GL_TEXTURE_VIEW_NUM_LEVELS:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLint) obj->NumLevels;
+         break;
+
+      case GL_TEXTURE_VIEW_MIN_LAYER:
+         if (!ctx->Extensions.ARB_texture_view)
+            goto invalid_pname;
+         *params = (GLint) obj->MinLayer;
+         break;
+
+      case GL_TEXTURE_VIEW_NUM_LAYERS:
+         if (!ctx->Extensions.ARB_texture_view)
             goto invalid_pname;
-         *params = obj->ImmutableLevels;
+         *params = (GLint) obj->NumLayers;
          break;
 
       case GL_REQUIRED_TEXTURE_IMAGE_UNITS_OES:
diff --git a/mesalib/src/mesa/main/texstorage.c b/mesalib/src/mesa/main/texstorage.c
index 84b8f8224..5062fdb4f 100644
--- a/mesalib/src/mesa/main/texstorage.c
+++ b/mesalib/src/mesa/main/texstorage.c
@@ -37,7 +37,9 @@
 #include "macros.h"
 #include "teximage.h"
 #include "texobj.h"
+#include "mipmap.h"
 #include "texstorage.h"
+#include "textureview.h"
 #include "mtypes.h"
 
 
@@ -73,8 +75,7 @@ legal_texobj_target(struct gl_context *ctx, GLuint dims, GLenum target)
          return ctx->Extensions.NV_texture_rectangle;
       case GL_TEXTURE_1D_ARRAY:
       case GL_PROXY_TEXTURE_1D_ARRAY:
-         return (ctx->Extensions.MESA_texture_array ||
-                 ctx->Extensions.EXT_texture_array);
+         return ctx->Extensions.EXT_texture_array;
       default:
          return GL_FALSE;
       }
@@ -85,8 +86,7 @@ legal_texobj_target(struct gl_context *ctx, GLuint dims, GLenum target)
          return GL_TRUE;
       case GL_TEXTURE_2D_ARRAY:
       case GL_PROXY_TEXTURE_2D_ARRAY:
-         return (ctx->Extensions.MESA_texture_array ||
-                 ctx->Extensions.EXT_texture_array);
+         return ctx->Extensions.EXT_texture_array;
       case GL_TEXTURE_CUBE_MAP_ARRAY:
       case GL_PROXY_TEXTURE_CUBE_MAP_ARRAY:
          return ctx->Extensions.ARB_texture_cube_map_array;
@@ -100,27 +100,6 @@ legal_texobj_target(struct gl_context *ctx, GLuint dims, GLenum target)
 }
 
 
-/**
- * Compute the size of the next mipmap level.
- */
-static void
-next_mipmap_level_size(GLenum target,
-                       GLint *width, GLint *height, GLint *depth)
-{
-   if (*width > 1) {
-      *width /= 2;
-   }
-
-   if ((*height > 1) && (target != GL_TEXTURE_1D_ARRAY)) {
-      *height /= 2;
-   }
-
-   if ((*depth > 1) && (target != GL_TEXTURE_2D_ARRAY)) {
-      *depth /= 2;
-   }
-}
-
-
 /** Helper to get a particular texture image in a texture object */
 static struct gl_texture_image *
 get_tex_image(struct gl_context *ctx, 
@@ -164,7 +143,8 @@ initialize_texture_fields(struct gl_context *ctx,
                                     0, internalFormat, texFormat);
       }
 
-      next_mipmap_level_size(target, &levelWidth, &levelHeight, &levelDepth);
+      _mesa_next_mipmap_level_size(target, 0, levelWidth, levelHeight, levelDepth,
+                                   &levelWidth, &levelHeight, &levelDepth);
    }
    return GL_TRUE;
 }
@@ -436,8 +416,8 @@ texstorage(GLuint dims, GLenum target, GLsizei levels, GLenum internalformat,
          return;
       }
 
-      texObj->Immutable = GL_TRUE;
-      texObj->ImmutableLevels = levels;
+      _mesa_set_texture_view_state(ctx, texObj, target, levels);
+
    }
 }
 
diff --git a/mesalib/src/mesa/main/texstore.c b/mesalib/src/mesa/main/texstore.c
index 76d8d9ba3..5adbd5dc9 100644
--- a/mesalib/src/mesa/main/texstore.c
+++ b/mesalib/src/mesa/main/texstore.c
@@ -3864,6 +3864,7 @@ _mesa_texstore_memcpy(TEXSTORE_PARAMS)
 /**
  * Store user data into texture memory.
  * Called via glTex[Sub]Image1/2/3D()
+ * \return GL_TRUE for success, GL_FALSE for failure (out of memory).
  */
 GLboolean
 _mesa_texstore(TEXSTORE_PARAMS)
diff --git a/mesalib/src/mesa/main/textureview.c b/mesalib/src/mesa/main/textureview.c
new file mode 100644
index 000000000..5f88a4171
--- /dev/null
+++ b/mesalib/src/mesa/main/textureview.c
@@ -0,0 +1,668 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 2013 LunarG, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Courtney Goeltzenleuchter <courtney@lunarg.com>
+ */
+
+
+/**
+ * \file textureview.c
+ * GL_ARB_texture_view functions
+ */
+
+#include "glheader.h"
+#include "context.h"
+#include "enums.h"
+#include "imports.h"
+#include "macros.h"
+#include "teximage.h"
+#include "texobj.h"
+#include "mipmap.h"
+#include "texstorage.h"
+#include "textureview.h"
+#include "stdbool.h"
+#include "mtypes.h"
+
+/* Table 3.X.2 (Compatible internal formats for TextureView)
+    ---------------------------------------------------------------------------
+    | Class                 | Internal formats                                |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_128_BITS   | RGBA32F, RGBA32UI, RGBA32I                      |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_96_BITS    | RGB32F, RGB32UI, RGB32I                         |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_64_BITS    | RGBA16F, RG32F, RGBA16UI, RG32UI, RGBA16I,      |
+    |                       | RG32I, RGBA16, RGBA16_SNORM                     |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_48_BITS    | RGB16, RGB16_SNORM, RGB16F, RGB16UI, RGB16I     |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_32_BITS    | RG16F, R11F_G11F_B10F, R32F,                    |
+    |                       | RGB10_A2UI, RGBA8UI, RG16UI, R32UI,             |
+    |                       | RGBA8I, RG16I, R32I, RGB10_A2, RGBA8, RG16,     |
+    |                       | RGBA8_SNORM, RG16_SNORM, SRGB8_ALPHA8, RGB9_E5  |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_24_BITS    | RGB8, RGB8_SNORM, SRGB8, RGB8UI, RGB8I          |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_16_BITS    | R16F, RG8UI, R16UI, RG8I, R16I, RG8, R16,       |
+    |                       | RG8_SNORM, R16_SNORM                            |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_8_BITS     | R8UI, R8I, R8, R8_SNORM                         |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_RGTC1_RED  | COMPRESSED_RED_RGTC1,                           |
+    |                       | COMPRESSED_SIGNED_RED_RGTC1                     |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_RGTC2_RG   | COMPRESSED_RG_RGTC2,                            |
+    |                       | COMPRESSED_SIGNED_RG_RGTC2                      |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_BPTC_UNORM | COMPRESSED_RGBA_BPTC_UNORM,                     |
+    |                       | COMPRESSED_SRGB_ALPHA_BPTC_UNORM                |
+    ---------------------------------------------------------------------------
+    | VIEW_CLASS_BPTC_FLOAT | COMPRESSED_RGB_BPTC_SIGNED_FLOAT,               |
+    |                       | COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT              |
+    ---------------------------------------------------------------------------
+ */
+struct internal_format_class_info {
+   GLenum view_class;
+   GLenum internal_format;
+};
+static const struct internal_format_class_info compatible_internal_formats[] = {
+   {GL_VIEW_CLASS_128_BITS, GL_RGBA32F},
+   {GL_VIEW_CLASS_128_BITS, GL_RGBA32UI},
+   {GL_VIEW_CLASS_128_BITS, GL_RGBA32I},
+   {GL_VIEW_CLASS_96_BITS, GL_RGB32F},
+   {GL_VIEW_CLASS_96_BITS, GL_RGB32UI},
+   {GL_VIEW_CLASS_96_BITS, GL_RGB32I},
+   {GL_VIEW_CLASS_64_BITS, GL_RGBA16F},
+   {GL_VIEW_CLASS_64_BITS, GL_RG32F},
+   {GL_VIEW_CLASS_64_BITS, GL_RGBA16UI},
+   {GL_VIEW_CLASS_64_BITS, GL_RG32UI},
+   {GL_VIEW_CLASS_64_BITS, GL_RGBA16I},
+   {GL_VIEW_CLASS_64_BITS, GL_RG32I},
+   {GL_VIEW_CLASS_64_BITS, GL_RGBA16},
+   {GL_VIEW_CLASS_64_BITS, GL_RGBA16_SNORM},
+   {GL_VIEW_CLASS_48_BITS, GL_RGB16},
+   {GL_VIEW_CLASS_48_BITS, GL_RGB16_SNORM},
+   {GL_VIEW_CLASS_48_BITS, GL_RGB16F},
+   {GL_VIEW_CLASS_48_BITS, GL_RGB16UI},
+   {GL_VIEW_CLASS_48_BITS, GL_RGB16I},
+   {GL_VIEW_CLASS_32_BITS, GL_RG16F},
+   {GL_VIEW_CLASS_32_BITS, GL_R11F_G11F_B10F},
+   {GL_VIEW_CLASS_32_BITS, GL_R32F},
+   {GL_VIEW_CLASS_32_BITS, GL_RGB10_A2UI},
+   {GL_VIEW_CLASS_32_BITS, GL_RGBA8UI},
+   {GL_VIEW_CLASS_32_BITS, GL_RG16UI},
+   {GL_VIEW_CLASS_32_BITS, GL_R32UI},
+   {GL_VIEW_CLASS_32_BITS, GL_RGBA8I},
+   {GL_VIEW_CLASS_32_BITS, GL_RG16I},
+   {GL_VIEW_CLASS_32_BITS, GL_R32I},
+   {GL_VIEW_CLASS_32_BITS, GL_RGB10_A2},
+   {GL_VIEW_CLASS_32_BITS, GL_RGBA8},
+   {GL_VIEW_CLASS_32_BITS, GL_RG16},
+   {GL_VIEW_CLASS_32_BITS, GL_RGBA8_SNORM},
+   {GL_VIEW_CLASS_32_BITS, GL_RG16_SNORM},
+   {GL_VIEW_CLASS_32_BITS, GL_SRGB8_ALPHA8},
+   {GL_VIEW_CLASS_32_BITS, GL_RGB9_E5},
+   {GL_VIEW_CLASS_24_BITS, GL_RGB8},
+   {GL_VIEW_CLASS_24_BITS, GL_RGB8_SNORM},
+   {GL_VIEW_CLASS_24_BITS, GL_SRGB8},
+   {GL_VIEW_CLASS_24_BITS, GL_RGB8UI},
+   {GL_VIEW_CLASS_24_BITS, GL_RGB8I},
+   {GL_VIEW_CLASS_16_BITS, GL_R16F},
+   {GL_VIEW_CLASS_16_BITS, GL_RG8UI},
+   {GL_VIEW_CLASS_16_BITS, GL_R16UI},
+   {GL_VIEW_CLASS_16_BITS, GL_RG8I},
+   {GL_VIEW_CLASS_16_BITS, GL_R16I},
+   {GL_VIEW_CLASS_16_BITS, GL_RG8},
+   {GL_VIEW_CLASS_16_BITS, GL_R16},
+   {GL_VIEW_CLASS_16_BITS, GL_RG8_SNORM},
+   {GL_VIEW_CLASS_16_BITS, GL_R16_SNORM},
+   {GL_VIEW_CLASS_8_BITS, GL_R8UI},
+   {GL_VIEW_CLASS_8_BITS, GL_R8I},
+   {GL_VIEW_CLASS_8_BITS, GL_R8},
+   {GL_VIEW_CLASS_8_BITS, GL_R8_SNORM},
+   {GL_VIEW_CLASS_RGTC1_RED, GL_COMPRESSED_RED_RGTC1},
+   {GL_VIEW_CLASS_RGTC1_RED, GL_COMPRESSED_SIGNED_RED_RGTC1},
+   {GL_VIEW_CLASS_RGTC2_RG, GL_COMPRESSED_RG_RGTC2},
+   {GL_VIEW_CLASS_RGTC2_RG, GL_COMPRESSED_SIGNED_RG_RGTC2},
+   {GL_VIEW_CLASS_BPTC_UNORM, GL_COMPRESSED_RGBA_BPTC_UNORM_ARB},
+   {GL_VIEW_CLASS_BPTC_UNORM, GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB},
+   {GL_VIEW_CLASS_BPTC_FLOAT, GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB},
+   {GL_VIEW_CLASS_BPTC_FLOAT, GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB},
+};
+
+static const struct internal_format_class_info s3tc_compatible_internal_formats[] = {
+   {GL_VIEW_CLASS_S3TC_DXT1_RGB, GL_COMPRESSED_RGB_S3TC_DXT1_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT1_RGB, GL_COMPRESSED_SRGB_S3TC_DXT1_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT1_RGBA, GL_COMPRESSED_RGBA_S3TC_DXT1_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT1_RGBA, GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT3_RGBA, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT3_RGBA, GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT5_RGBA, GL_COMPRESSED_RGBA_S3TC_DXT5_EXT},
+   {GL_VIEW_CLASS_S3TC_DXT5_RGBA, GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT},
+};
+
+/**
+ * Lookup format view class based on internalformat
+ * \return VIEW_CLASS if internalformat found in table, false otherwise.
+ */
+static GLenum
+lookup_view_class(struct gl_context *ctx, GLenum internalformat)
+{
+   GLuint i;
+
+   for (i = 0; i < ARRAY_SIZE(compatible_internal_formats); i++) {
+      if (compatible_internal_formats[i].internal_format == internalformat)
+         return compatible_internal_formats[i].view_class;
+   }
+
+   if (ctx->Extensions.EXT_texture_compression_s3tc && ctx->Extensions.EXT_texture_sRGB) {
+      for (i = 0; i < ARRAY_SIZE(s3tc_compatible_internal_formats); i++) {
+         if (s3tc_compatible_internal_formats[i].internal_format == internalformat)
+            return s3tc_compatible_internal_formats[i].view_class;
+      }
+   }
+   return GL_FALSE;
+}
+
+/**
+ * Initialize new texture's gl_texture_image structures. Will not call driver
+ * to allocate new space, simply record relevant layer, face, format, etc.
+ * \return GL_FALSE if any error, GL_TRUE otherwise.
+ */
+static GLboolean
+initialize_texture_fields(struct gl_context *ctx,
+                          GLenum target,
+                          struct gl_texture_object *texObj,
+                          GLint levels,
+                          GLsizei width, GLsizei height, GLsizei depth,
+                          GLenum internalFormat, gl_format texFormat)
+{
+   const GLuint numFaces = _mesa_num_tex_faces(target);
+   GLint level, levelWidth = width, levelHeight = height, levelDepth = depth;
+   GLuint face;
+
+   /* Pretend we are bound to initialize the gl_texture_image structs */
+   texObj->Target = target;
+
+   /* Set up all the texture object's gl_texture_images */
+   for (level = 0; level < levels; level++) {
+      for (face = 0; face < numFaces; face++) {
+         struct gl_texture_image *texImage;
+         GLenum faceTarget = target;
+
+         if (target == GL_TEXTURE_CUBE_MAP)
+            faceTarget = GL_TEXTURE_CUBE_MAP_POSITIVE_X + face;
+
+         texImage = _mesa_get_tex_image(ctx, texObj, faceTarget, level);
+
+         if (!texImage) {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexStorage");
+            return GL_FALSE;
+         }
+
+         _mesa_init_teximage_fields(ctx, texImage,
+                                    levelWidth, levelHeight, levelDepth,
+                                    0, internalFormat, texFormat);
+      }
+
+      _mesa_next_mipmap_level_size(target, 0, levelWidth, levelHeight, levelDepth,
+                                   &levelWidth, &levelHeight, &levelDepth);
+   }
+
+   /* "unbind" */
+   texObj->Target = 0;
+
+   return GL_TRUE;
+}
+
+#define RETURN_IF_SUPPORTED(t) do {		\
+   if (newTarget == GL_ ## t)                   \
+      return true;				\
+} while (0)
+
+/**
+ * Check for compatible target
+ * If an error is found, record it with _mesa_error()
+ * \return false if any error, true otherwise.
+ */
+static bool
+target_valid(struct gl_context *ctx, GLenum origTarget, GLenum newTarget)
+{
+   /*
+    * From ARB_texture_view spec:
+   ---------------------------------------------------------------------------------------------------------
+   | Original target              | Valid new targets |
+   ---------------------------------------------------------------------------------------------------------
+   | TEXTURE_1D                   | TEXTURE_1D, TEXTURE_1D_ARRAY |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_2D                   | TEXTURE_2D, TEXTURE_2D_ARRAY |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_3D                   | TEXTURE_3D |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_CUBE_MAP             | TEXTURE_CUBE_MAP, TEXTURE_2D, TEXTURE_2D_ARRAY, TEXTURE_CUBE_MAP_ARRAY |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_RECTANGLE            | TEXTURE_RECTANGLE |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_BUFFER               | <none> |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_1D_ARRAY             | TEXTURE_1D_ARRAY, TEXTURE_1D |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_2D_ARRAY             | TEXTURE_2D_ARRAY, TEXTURE_2D, TEXTURE_CUBE_MAP, TEXTURE_CUBE_MAP_ARRAY |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_CUBE_MAP_ARRAY       | TEXTURE_CUBE_MAP_ARRAY, TEXTURE_2D_ARRAY, TEXTURE_2D, TEXTURE_CUBE_MAP |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_2D_MULTISAMPLE       | TEXTURE_2D_MULTISAMPLE, TEXTURE_2D_MULTISAMPLE_ARRAY |
+   | ------------------------------------------------------------------------------------------------------- |
+   | TEXTURE_2D_MULTISAMPLE_ARRAY | TEXTURE_2D_MULTISAMPLE, TEXTURE_2D_MULTISAMPLE_ARRAY |
+   ---------------------------------------------------------------------------------------------------------
+    */
+
+   switch (origTarget) {
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_1D_ARRAY:
+      RETURN_IF_SUPPORTED(TEXTURE_1D);
+      RETURN_IF_SUPPORTED(TEXTURE_1D_ARRAY);
+      break;
+   case GL_TEXTURE_2D:
+      RETURN_IF_SUPPORTED(TEXTURE_2D);
+      RETURN_IF_SUPPORTED(TEXTURE_2D_ARRAY);
+      break;
+   case GL_TEXTURE_3D:
+      RETURN_IF_SUPPORTED(TEXTURE_3D);
+      break;
+   case GL_TEXTURE_RECTANGLE:
+      RETURN_IF_SUPPORTED(TEXTURE_RECTANGLE);
+      break;
+   case GL_TEXTURE_CUBE_MAP:
+   case GL_TEXTURE_2D_ARRAY:
+   case GL_TEXTURE_CUBE_MAP_ARRAY:
+      RETURN_IF_SUPPORTED(TEXTURE_2D);
+      RETURN_IF_SUPPORTED(TEXTURE_2D_ARRAY);
+      RETURN_IF_SUPPORTED(TEXTURE_CUBE_MAP);
+      RETURN_IF_SUPPORTED(TEXTURE_CUBE_MAP_ARRAY);
+      break;
+   case GL_TEXTURE_2D_MULTISAMPLE:
+   case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+      RETURN_IF_SUPPORTED(TEXTURE_2D_MULTISAMPLE);
+      RETURN_IF_SUPPORTED(TEXTURE_2D_MULTISAMPLE_ARRAY);
+      break;
+   }
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+               "glTextureView(illegal target=%s)",
+               _mesa_lookup_enum_by_nr(newTarget));
+   return false;
+}
+#undef RETURN_IF_SUPPORTED
+
+/**
+ * Check for compatible format
+ * If an error is found, record it with _mesa_error()
+ * \return false if any error, true otherwise.
+ */
+static bool
+compatible_format(struct gl_context *ctx, const struct gl_texture_object *origTexObj,
+                  GLenum internalformat)
+{
+   /* Level 0 of a texture created by glTextureStorage or glTextureView
+    * is always defined.
+    */
+   struct gl_texture_image *texImage = origTexObj->Image[0][0];
+   GLint origInternalFormat = texImage->InternalFormat;
+   unsigned int origViewClass, newViewClass;
+
+   /* The two textures' internal formats must be compatible according to
+    * Table 3.X.2 (Compatible internal formats for TextureView)
+    * if the internal format exists in that table the view class must match.
+    * The internal formats must be identical if not in that table,
+    * or an INVALID_OPERATION error is generated.
+    */
+   if (origInternalFormat == internalformat)
+      return true;
+
+   origViewClass = lookup_view_class(ctx, origInternalFormat);
+   newViewClass = lookup_view_class(ctx, internalformat);
+   if ((origViewClass == newViewClass) && origViewClass != false)
+      return true;
+
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+               "glTextureView(internalformat %s not compatible with origtexture %s)",
+               _mesa_lookup_enum_by_nr(internalformat),
+               _mesa_lookup_enum_by_nr(origInternalFormat));
+   return false;
+}
+/**
+ * Helper function for TexStorage and teximagemultisample to set immutable
+ * texture state needed by ARB_texture_view.
+ */
+void
+_mesa_set_texture_view_state(struct gl_context *ctx,
+                             struct gl_texture_object *texObj,
+                             GLenum target, GLuint levels)
+{
+   struct gl_texture_image *texImage;
+
+   /* Get a reference to what will become this View's base level */
+   texImage = _mesa_select_tex_image(ctx, texObj, target, 0);
+
+   /* When an immutable texture is created via glTexStorage or glTexImageMultisample,
+    * TEXTURE_IMMUTABLE_FORMAT becomes TRUE.
+    * TEXTURE_IMMUTABLE_LEVELS and TEXTURE_VIEW_NUM_LEVELS become levels.
+    * If the texture target is TEXTURE_1D_ARRAY then
+    * TEXTURE_VIEW_NUM_LAYERS becomes height.
+    * If the texture target is TEXTURE_2D_ARRAY, TEXTURE_CUBE_MAP_ARRAY,
+    * or TEXTURE_2D_MULTISAMPLE_ARRAY then TEXTURE_VIEW_NUM_LAYERS becomes depth.
+    * If the texture target is TEXTURE_CUBE_MAP, then
+    * TEXTURE_VIEW_NUM_LAYERS becomes 6.
+    * For any other texture target, TEXTURE_VIEW_NUM_LAYERS becomes 1.
+    * 
+    * ARB_texture_multisample: Multisample textures do
+    * not have multiple image levels.
+    */
+
+   texObj->Immutable = GL_TRUE;
+   texObj->ImmutableLevels = levels;
+   texObj->MinLevel = 0;
+   texObj->NumLevels = levels;
+   texObj->MinLayer = 0;
+   texObj->NumLayers = 1;
+   switch (target) {
+   case GL_TEXTURE_1D_ARRAY:
+      texObj->NumLayers = texImage->Height;
+      break;
+
+   case GL_TEXTURE_2D_MULTISAMPLE:
+      texObj->NumLevels = 1;
+      texObj->ImmutableLevels = 1;
+      break;
+
+   case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+      texObj->NumLevels = 1;
+      texObj->ImmutableLevels = 1;
+      /* fall through to set NumLayers */
+
+   case GL_TEXTURE_2D_ARRAY:
+   case GL_TEXTURE_CUBE_MAP_ARRAY:
+      texObj->NumLayers = texImage->Depth;
+      break;
+
+   case GL_TEXTURE_CUBE_MAP:
+      texObj->NumLayers = 6;
+      break;
+
+   }
+}
+
+/**
+ * glTextureView (ARB_texture_view)
+ * If an error is found, record it with _mesa_error()
+ * \return none.
+ */
+void GLAPIENTRY
+_mesa_TextureView(GLuint texture, GLenum target, GLuint origtexture,
+                  GLenum internalformat,
+                  GLuint minlevel, GLuint numlevels,
+                  GLuint minlayer, GLuint numlayers)
+{
+   struct gl_texture_object *texObj;
+   struct gl_texture_object *origTexObj;
+   struct gl_texture_image *origTexImage;
+   GLuint newViewMinLevel, newViewMinLayer;
+   GLuint newViewNumLevels, newViewNumLayers;
+   GLsizei width, height, depth;
+   gl_format texFormat;
+   GLboolean sizeOK, dimensionsOK;
+   GLenum faceTarget;
+
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & (VERBOSE_API | VERBOSE_TEXTURE))
+      _mesa_debug(ctx, "glTextureView %d %s %d %s %d %d %d %d\n",
+                  texture, _mesa_lookup_enum_by_nr(target), origtexture,
+                  _mesa_lookup_enum_by_nr(internalformat),
+                  minlevel, numlevels, minlayer, numlayers);
+
+   if (origtexture == 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glTextureView(origtexture = %u)", origtexture);
+      return;
+   }
+
+   /* Need original texture information to validate arguments */
+   origTexObj = _mesa_lookup_texture(ctx, origtexture);
+
+   /* If <origtexture> is not the name of a texture, INVALID_VALUE is generated. */
+   if (!origTexObj) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glTextureView(origtexture = %u)", origtexture);
+      return;
+   }
+
+   /* If <origtexture>'s TEXTURE_IMMUTABLE_FORMAT value is not TRUE,
+    * INVALID_OPERATION is generated.
+    */
+   if (!origTexObj->Immutable) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(origtexture not immutable)");
+      return;
+   }
+
+   /* If <texture> is 0, INVALID_VALUE is generated. */
+   if (texture == 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glTextureView(texture = 0)");
+      return;
+   }
+
+   /* If <texture> is not a valid name returned by GenTextures,
+    * the error INVALID_OPERATION is generated.
+    */
+   texObj = _mesa_lookup_texture(ctx, texture);
+   if (texObj == NULL) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(texture = %u non-gen name)", texture);
+      return;
+   }
+
+   /* If <texture> has already been bound and given a target, then
+    * the error INVALID_OPERATION is generated.
+    */
+   if (texObj->Target) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(texture = %u already bound)", texture);
+      return;
+   }
+
+   /* Check for compatible target */
+   if (!target_valid(ctx, origTexObj->Target, target)) {
+      return; /* error was recorded */
+   }
+
+   /* minlevel and minlayer are relative to the view of origtexture
+    * If minlevel or minlayer is greater than level or layer, respectively,
+    * of origtexture return INVALID_VALUE.
+    */
+   newViewMinLevel = origTexObj->MinLevel + minlevel;
+   newViewMinLayer = origTexObj->MinLayer + minlayer;
+   if (newViewMinLevel >= (origTexObj->MinLevel + origTexObj->NumLevels)) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glTextureView(new minlevel (%d) > orig minlevel (%d) + orig numlevels (%d))",
+                  newViewMinLevel, origTexObj->MinLevel, origTexObj->NumLevels);
+      return;
+   }
+
+   if (newViewMinLayer >= (origTexObj->MinLayer + origTexObj->NumLayers)) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glTextureView(new minlayer (%d) > orig minlayer (%d) + orig numlayers (%d))",
+                  newViewMinLayer, origTexObj->MinLayer, origTexObj->NumLayers);
+      return;
+   }
+
+   if (!compatible_format(ctx, origTexObj, internalformat)) {
+      return; /* Error logged */
+   }
+
+   texFormat = _mesa_choose_texture_format(ctx, texObj, target, 0,
+                                           internalformat, GL_NONE, GL_NONE);
+   assert(texFormat != MESA_FORMAT_NONE);
+   if (texFormat == MESA_FORMAT_NONE) return;
+
+   newViewNumLevels = MIN2(numlevels, origTexObj->NumLevels - minlevel);
+   newViewNumLayers = MIN2(numlayers, origTexObj->NumLayers - minlayer);
+
+   faceTarget = origTexObj->Target;
+   if (faceTarget == GL_TEXTURE_CUBE_MAP)
+      faceTarget = GL_TEXTURE_CUBE_MAP_POSITIVE_X + minlayer;
+
+   /* Get a reference to what will become this View's base level */
+   origTexImage = _mesa_select_tex_image(ctx, origTexObj,
+                                         faceTarget, minlevel);
+   width = origTexImage->Width;
+   height = origTexImage->Height;
+   depth = origTexImage->Depth;
+
+   /* Adjust width, height, depth to be appropriate for new target */
+   switch (target) {
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_3D:
+      break;
+
+   case GL_TEXTURE_1D_ARRAY:
+      height = (GLsizei) newViewNumLayers;
+      break;
+
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_2D_MULTISAMPLE:
+   case GL_TEXTURE_RECTANGLE:
+   case GL_TEXTURE_CUBE_MAP:
+      depth = 1;
+      break;
+
+   case GL_TEXTURE_2D_ARRAY:
+   case GL_TEXTURE_CUBE_MAP_ARRAY:
+   case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+      depth = newViewNumLayers;
+      break;
+   }
+
+   /* If the dimensions of the original texture are larger than the maximum
+    * supported dimensions of the new target, the error INVALID_OPERATION is
+    * generated. For example, if the original texture has a TEXTURE_2D_ARRAY
+    * target and its width is greater than MAX_CUBE_MAP_TEXTURE_SIZE, an error
+    * will be generated if TextureView is called to create a TEXTURE_CUBE_MAP
+    * view.
+    */
+   dimensionsOK = _mesa_legal_texture_dimensions(ctx, target, 0,
+                                                 width, height, depth, 0);
+   if (!dimensionsOK) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(invalid width or height or depth)");
+      return;
+   }
+
+   sizeOK = ctx->Driver.TestProxyTexImage(ctx, target, 0, texFormat,
+                                          width, height, depth, 0);
+   if (!sizeOK) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(invalid texture size)");
+      return;
+   }
+
+   /* If <target> is TEXTURE_1D, TEXTURE_2D, TEXTURE_3D, TEXTURE_RECTANGLE,
+    * or TEXTURE_2D_MULTISAMPLE and <numlayers> does not equal 1, the error
+    * INVALID_VALUE is generated.
+    */
+   switch (target) {
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_3D:
+   case GL_TEXTURE_RECTANGLE:
+   case GL_TEXTURE_2D_MULTISAMPLE:
+      if (numlayers != 1) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glTextureView(numlayers %d != 1)", numlayers);
+         return;
+      }
+      break;
+
+   case GL_TEXTURE_CUBE_MAP:
+      /* If the new texture's target is TEXTURE_CUBE_MAP, the clamped <numlayers>
+       * must be equal to 6.
+       */
+      if (newViewNumLayers != 6) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glTextureView(clamped numlayers %d != 6)",
+                     newViewNumLayers);
+         return;
+      }
+      break;
+
+   case GL_TEXTURE_CUBE_MAP_ARRAY:
+      /* If the new texture's target is TEXTURE_CUBE_MAP_ARRAY,
+       * then <numlayers> counts layer-faces rather than layers,
+       * and the clamped <numlayers> must be a multiple of 6.
+       * Otherwise, the error INVALID_VALUE is generated.
+       */
+      if ((newViewNumLayers % 6) != 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glTextureView(clamped numlayers %d is not a multiple of 6)",
+                     newViewNumLayers);
+         return;
+      }
+      break;
+   }
+
+   /* If the new texture's target is TEXTURE_CUBE_MAP or
+    * TEXTURE_CUBE_MAP_ARRAY, the width and height of the original texture's
+    * levels must be equal otherwise the error INVALID_OPERATION is generated.
+    */
+   if ((target == GL_TEXTURE_CUBE_MAP || target == GL_TEXTURE_CUBE_MAP_ARRAY) &&
+       (origTexImage->Width != origTexImage->Height)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glTextureView(origtexture width (%d) != height (%d))",
+                  origTexImage->Width, origTexImage->Height);
+      return;
+   }
+
+   /* When the original texture's target is TEXTURE_CUBE_MAP, the layer
+    * parameters are interpreted in the same order as if it were a
+    * TEXTURE_CUBE_MAP_ARRAY with 6 layer-faces.
+    */
+
+   /* If the internal format does not exactly match the internal format of the
+    * original texture, the contents of the memory are reinterpreted in the
+    * same manner as for image bindings described in
+    * section 3.9.20 (Texture Image Loads and Stores).
+    */
+
+   /* TEXTURE_BASE_LEVEL and TEXTURE_MAX_LEVEL are interpreted
+    * relative to the view and not relative to the original data store.
+    */
+
+   if (!initialize_texture_fields(ctx, target, texObj, newViewNumLevels,
+                                  width, height, depth,
+                                  internalformat, texFormat)) {
+      return; /* Already recorded error */
+   }
+
+   texObj->MinLevel = newViewMinLevel;
+   texObj->MinLayer = newViewMinLayer;
+   texObj->NumLevels = newViewNumLevels;
+   texObj->NumLayers = newViewNumLayers;
+   texObj->Immutable = GL_TRUE;
+   texObj->ImmutableLevels = origTexObj->ImmutableLevels;
+   texObj->Target = target;
+
+   if (ctx->Driver.TextureView != NULL && !ctx->Driver.TextureView(ctx, texObj, origTexObj)) {
+      return; /* driver recorded error */
+   }
+}
diff --git a/mesalib/src/mesa/main/textureview.h b/mesalib/src/mesa/main/textureview.h
new file mode 100644
index 000000000..3088ac193
--- /dev/null
+++ b/mesalib/src/mesa/main/textureview.h
@@ -0,0 +1,43 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 2012-2013 LunarG, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Courtney Goeltzenleuchter <courtney@lunarg.com>
+ */
+
+
+#ifndef TEXTUREVIEW_H
+#define TEXTUREVIEW_H
+
+
+extern void GLAPIENTRY
+_mesa_TextureView(GLuint texture, GLenum target, GLuint origtexture,
+                  GLenum internalformat,
+                  GLuint minlevel, GLuint numlevels,
+                  GLuint minlayer, GLuint numlayers);
+
+extern void
+_mesa_set_texture_view_state(struct gl_context *ctx, struct gl_texture_object *texObj,
+                       GLenum target, GLuint levels);
+
+#endif /* TEXTUREVIEW_H */
diff --git a/mesalib/src/mesa/main/version.c b/mesalib/src/mesa/main/version.c
index ee77947b4..119c7dae3 100644
--- a/mesalib/src/mesa/main/version.c
+++ b/mesalib/src/mesa/main/version.c
@@ -224,7 +224,6 @@ compute_version(struct gl_context *ctx)
 			      (ctx->Extensions.EXT_stencil_two_side
 			       || ctx->Extensions.ATI_separate_stencil));
    const GLboolean ver_2_1 = (ver_2_0 &&
-                              ctx->Const.GLSLVersion >= 120 &&
                               ctx->Extensions.EXT_pixel_buffer_object &&
                               ctx->Extensions.EXT_texture_sRGB);
    const GLboolean ver_3_0 = (ver_2_1 &&
diff --git a/mesalib/src/mesa/math/m_matrix.c b/mesalib/src/mesa/math/m_matrix.c
index 290231527..274f969d2 100644
--- a/mesalib/src/mesa/math/m_matrix.c
+++ b/mesalib/src/mesa/math/m_matrix.c
@@ -1488,14 +1488,11 @@ _math_matrix_ctr( GLmatrix *m )
 void
 _math_matrix_dtr( GLmatrix *m )
 {
-   if (m->m) {
-      _mesa_align_free( m->m );
-      m->m = NULL;
-   }
-   if (m->inv) {
-      _mesa_align_free( m->inv );
-      m->inv = NULL;
-   }
+   _mesa_align_free( m->m );
+   m->m = NULL;
+
+   _mesa_align_free( m->inv );
+   m->inv = NULL;
 }
 
 /*@}*/
diff --git a/mesalib/src/mesa/program/ir_to_mesa.cpp b/mesalib/src/mesa/program/ir_to_mesa.cpp
index c833a12f2..23d479c32 100644
--- a/mesalib/src/mesa/program/ir_to_mesa.cpp
+++ b/mesalib/src/mesa/program/ir_to_mesa.cpp
@@ -681,11 +681,11 @@ ir_to_mesa_visitor::visit(ir_variable *ir)
    if (strcmp(ir->name, "gl_FragCoord") == 0) {
       struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog;
 
-      fp->OriginUpperLeft = ir->origin_upper_left;
-      fp->PixelCenterInteger = ir->pixel_center_integer;
+      fp->OriginUpperLeft = ir->data.origin_upper_left;
+      fp->PixelCenterInteger = ir->data.pixel_center_integer;
    }
 
-   if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
+   if (ir->data.mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
       unsigned int i;
       const ir_state_slot *const slots = ir->state_slots;
       assert(ir->state_slots != NULL);
@@ -759,49 +759,10 @@ ir_to_mesa_visitor::visit(ir_variable *ir)
 void
 ir_to_mesa_visitor::visit(ir_loop *ir)
 {
-   ir_dereference_variable *counter = NULL;
-
-   if (ir->counter != NULL)
-      counter = new(mem_ctx) ir_dereference_variable(ir->counter);
-
-   if (ir->from != NULL) {
-      assert(ir->counter != NULL);
-
-      ir_assignment *a =
-	new(mem_ctx) ir_assignment(counter, ir->from, NULL);
-
-      a->accept(this);
-   }
-
    emit(NULL, OPCODE_BGNLOOP);
 
-   if (ir->to) {
-      ir_expression *e =
-	 new(mem_ctx) ir_expression(ir->cmp, glsl_type::bool_type,
-					  counter, ir->to);
-      ir_if *if_stmt =  new(mem_ctx) ir_if(e);
-
-      ir_loop_jump *brk =
-	new(mem_ctx) ir_loop_jump(ir_loop_jump::jump_break);
-
-      if_stmt->then_instructions.push_tail(brk);
-
-      if_stmt->accept(this);
-   }
-
    visit_exec_list(&ir->body_instructions, this);
 
-   if (ir->increment) {
-      ir_expression *e =
-	 new(mem_ctx) ir_expression(ir_binop_add, counter->type,
-					  counter, ir->increment);
-
-      ir_assignment *a =
-	new(mem_ctx) ir_assignment(counter, e, NULL);
-
-      a->accept(this);
-   }
-
    emit(NULL, OPCODE_ENDLOOP);
 }
 
@@ -1567,10 +1528,10 @@ ir_to_mesa_visitor::visit(ir_dereference_variable *ir)
    ir_variable *var = ir->var;
 
    if (!entry) {
-      switch (var->mode) {
+      switch (var->data.mode) {
       case ir_var_uniform:
 	 entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM,
-					       var->location);
+					       var->data.location);
 	 this->variables.push_tail(entry);
 	 break;
       case ir_var_shader_in:
@@ -1579,21 +1540,21 @@ ir_to_mesa_visitor::visit(ir_dereference_variable *ir)
 	  * user-assigned generic attributes (glBindVertexLocation),
 	  * and user-defined varyings.
 	  */
-	 assert(var->location != -1);
+	 assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_INPUT,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_shader_out:
-	 assert(var->location != -1);
+	 assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_OUTPUT,
-                                               var->location);
+                                               var->data.location);
 	 break;
       case ir_var_system_value:
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_SYSTEM_VALUE,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_auto:
       case ir_var_temporary:
@@ -2443,7 +2404,7 @@ public:
       this->idx = -1;
       this->program_resource_visitor::process(var);
 
-      var->location = this->idx;
+      var->data.location = this->idx;
    }
 
 private:
@@ -2538,7 +2499,7 @@ _mesa_generate_parameters_list_for_uniforms(struct gl_shader_program
    foreach_list(node, sh->ir) {
       ir_variable *var = ((ir_instruction *) node)->as_variable();
 
-      if ((var == NULL) || (var->mode != ir_var_uniform)
+      if ((var == NULL) || (var->data.mode != ir_var_uniform)
 	  || var->is_in_uniform_block() || (strncmp(var->name, "gl_", 3) == 0))
 	 continue;
 
diff --git a/mesalib/src/mesa/program/prog_parameter.c b/mesalib/src/mesa/program/prog_parameter.c
index 4d9cf08d2..54531d255 100644
--- a/mesalib/src/mesa/program/prog_parameter.c
+++ b/mesalib/src/mesa/program/prog_parameter.c
@@ -83,8 +83,7 @@ _mesa_free_parameter_list(struct gl_program_parameter_list *paramList)
       free((void *)paramList->Parameters[i].Name);
    }
    free(paramList->Parameters);
-   if (paramList->ParameterValues)
-      _mesa_align_free(paramList->ParameterValues);
+   _mesa_align_free(paramList->ParameterValues);
    free(paramList);
 }
 
diff --git a/mesalib/src/mesa/program/program.c b/mesalib/src/mesa/program/program.c
index 01f8c6f11..cdf1c03fa 100644
--- a/mesalib/src/mesa/program/program.c
+++ b/mesalib/src/mesa/program/program.c
@@ -1049,6 +1049,14 @@ _mesa_get_min_invocations_per_fragment(struct gl_context *ctx,
     *  has no effect."
     */
    if (ctx->Multisample.Enabled) {
+      /* The ARB_gpu_shader5 specification says:
+       *
+       * "Use of the "sample" qualifier on a fragment shader input
+       *  forces per-sample shading"
+       */
+      if (prog->IsSample)
+         return MAX2(ctx->DrawBuffer->Visual.samples, 1);
+
       if (prog->Base.SystemValuesRead & (SYSTEM_BIT_SAMPLE_ID |
                                          SYSTEM_BIT_SAMPLE_POS))
          return MAX2(ctx->DrawBuffer->Visual.samples, 1);
diff --git a/mesalib/src/mesa/program/program_parse_extra.c b/mesalib/src/mesa/program/program_parse_extra.c
index e8e1912eb..a9e364045 100644
--- a/mesalib/src/mesa/program/program_parse_extra.c
+++ b/mesalib/src/mesa/program/program_parse_extra.c
@@ -256,15 +256,6 @@ _mesa_ARBfp_parse_option(struct asm_parser_state *state, const char *option)
 	    return 1;
 	 }
       }
-   } else if (strncmp(option, "MESA_", 5) == 0) {
-      option += 5;
-
-      if (strcmp(option, "texture_array") == 0) {
-	 if (state->ctx->Extensions.MESA_texture_array) {
-	    state->option.TexArray = 1;
-	    return 1;
-	 }
-      }
    }
 
    return 0;
diff --git a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
index c752640f4..51f079cda 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -44,56 +44,6 @@
 
 
 /**
- * When doing GL render to texture, we have to be sure that finalize_texture()
- * didn't yank out the pipe_resource that we earlier created a surface for.
- * Check for that here and create a new surface if needed.
- */
-static void
-update_renderbuffer_surface(struct st_context *st,
-                            struct st_renderbuffer *strb)
-{
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_resource *resource = strb->rtt ? strb->rtt->pt : strb->texture;
-   int rtt_width = strb->Base.Width;
-   int rtt_height = strb->Base.Height;
-   enum pipe_format format = st->ctx->Color.sRGBEnabled ? resource->format : util_format_linear(resource->format);
-
-   if (!strb->surface ||
-       strb->surface->texture->nr_samples != strb->Base.NumSamples ||
-       strb->surface->format != format ||
-       strb->surface->texture != resource ||
-       strb->surface->width != rtt_width ||
-       strb->surface->height != rtt_height) {
-      GLuint level;
-      /* find matching mipmap level size */
-      for (level = 0; level <= resource->last_level; level++) {
-         if (u_minify(resource->width0, level) == rtt_width &&
-             u_minify(resource->height0, level) == rtt_height) {
-            struct pipe_surface surf_tmpl;
-            memset(&surf_tmpl, 0, sizeof(surf_tmpl));
-            surf_tmpl.format = format;
-            surf_tmpl.u.tex.level = level;
-            surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice;
-            surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice;
-
-            pipe_surface_reference(&strb->surface, NULL);
-
-            strb->surface = pipe->create_surface(pipe,
-                                                 resource,
-                                                 &surf_tmpl);
-#if 0
-            printf("-- alloc new surface %d x %d into tex %p\n",
-                   strb->surface->width, strb->surface->height,
-                   texture);
-#endif
-            break;
-         }
-      }
-   }
-}
-
-
-/**
  * Update framebuffer state (color, depth, stencil, etc. buffers)
  */
 static void
@@ -121,10 +71,10 @@ update_framebuffer_state( struct st_context *st )
 
       if (strb) {
          /*printf("--------- framebuffer surface rtt %p\n", strb->rtt);*/
-         if (strb->rtt ||
+         if (strb->is_rtt ||
              (strb->texture && util_format_is_srgb(strb->texture->format))) {
             /* rendering to a GL texture, may have to update surface */
-            update_renderbuffer_surface(st, strb);
+            st_update_renderbuffer_surface(st, strb);
          }
 
          if (strb->surface) {
@@ -144,9 +94,9 @@ update_framebuffer_state( struct st_context *st )
     */
    strb = st_renderbuffer(fb->Attachment[BUFFER_DEPTH].Renderbuffer);
    if (strb) {
-      if (strb->rtt) {
+      if (strb->is_rtt) {
          /* rendering to a GL texture, may have to update surface */
-         update_renderbuffer_surface(st, strb);
+         st_update_renderbuffer_surface(st, strb);
       }
       pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
    }
diff --git a/mesalib/src/mesa/state_tracker/st_atom_sampler.c b/mesalib/src/mesa/state_tracker/st_atom_sampler.c
index 302e12981..57670ce25 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_sampler.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_sampler.c
@@ -130,15 +130,20 @@ convert_sampler(struct st_context *st,
                 struct pipe_sampler_state *sampler,
                 GLuint texUnit)
 {
-   struct gl_texture_object *texobj;
+   const struct gl_texture_object *texobj;
    struct gl_context *ctx = st->ctx;
    struct gl_sampler_object *msamp;
+   const struct gl_texture_image *teximg;
+   GLenum texBaseFormat;
 
    texobj = ctx->Texture.Unit[texUnit]._Current;
    if (!texobj) {
       texobj = _mesa_get_fallback_texture(ctx, TEXTURE_2D_INDEX);
    }
 
+   teximg = texobj->Image[0][texobj->BaseLevel];
+   texBaseFormat = teximg ? teximg->_BaseFormat : GL_RGBA;
+
    msamp = _mesa_get_samplerobj(ctx, texUnit);
 
    memset(sampler, 0, sizeof(*sampler));
@@ -170,21 +175,15 @@ convert_sampler(struct st_context *st,
       assert(sampler->min_lod <= sampler->max_lod);
    }
 
+   /* For non-black borders... */
    if (msamp->BorderColor.ui[0] ||
        msamp->BorderColor.ui[1] ||
        msamp->BorderColor.ui[2] ||
        msamp->BorderColor.ui[3]) {
-      struct st_texture_object *stobj = st_texture_object(texobj);
-      struct gl_texture_image *teximg;
-      GLboolean is_integer = GL_FALSE;
+      const struct st_texture_object *stobj = st_texture_object_const(texobj);
+      const GLboolean is_integer = texobj->_IsIntegerFormat;
       union pipe_color_union border_color;
 
-      teximg = texobj->Image[0][texobj->BaseLevel];
-
-      if (teximg) {
-         is_integer = _mesa_is_enum_format_integer(teximg->InternalFormat);
-      }
-
       if (st->apply_texture_swizzle_to_border_color && stobj->sampler_view) {
          const unsigned char swz[4] =
          {
@@ -196,25 +195,26 @@ convert_sampler(struct st_context *st,
 
          st_translate_color(&msamp->BorderColor,
                             &border_color,
-                            teximg ? teximg->_BaseFormat : GL_RGBA, is_integer);
+                            texBaseFormat, is_integer);
 
          util_format_apply_color_swizzle(&sampler->border_color,
                                          &border_color, swz, is_integer);
       } else {
          st_translate_color(&msamp->BorderColor,
                             &sampler->border_color,
-                            teximg ? teximg->_BaseFormat : GL_RGBA, is_integer);
+                            texBaseFormat, is_integer);
       }
    }
 
    sampler->max_anisotropy = (msamp->MaxAnisotropy == 1.0 ?
                               0 : (GLuint) msamp->MaxAnisotropy);
 
-   /* only care about ARB_shadow, not SGI shadow */
-   if (msamp->CompareMode == GL_COMPARE_R_TO_TEXTURE) {
+   /* If sampling a depth texture and using shadow comparison */
+   if ((texBaseFormat == GL_DEPTH_COMPONENT ||
+        texBaseFormat == GL_DEPTH_STENCIL) &&
+       msamp->CompareMode == GL_COMPARE_R_TO_TEXTURE) {
       sampler->compare_mode = PIPE_TEX_COMPARE_R_TO_TEXTURE;
-      sampler->compare_func
-         = st_compare_func_to_pipe(msamp->CompareFunc);
+      sampler->compare_func = st_compare_func_to_pipe(msamp->CompareFunc);
    }
 
    sampler->seamless_cube_map =
diff --git a/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c b/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
index 7fa4cbdc4..230ab5449 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_bufferobjects.c
@@ -422,6 +422,9 @@ st_bufferobj_validate_usage(struct st_context *st,
 void
 st_init_bufferobject_functions(struct dd_function_table *functions)
 {
+   /* plug in default driver fallbacks (such as for ClearBufferSubData) */
+   _mesa_init_buffer_object_functions(functions);
+
    functions->NewBufferObject = st_bufferobj_alloc;
    functions->DeleteBuffer = st_bufferobj_free;
    functions->BufferData = st_bufferobj_data;
diff --git a/mesalib/src/mesa/state_tracker/st_cb_clear.c b/mesalib/src/mesa/state_tracker/st_cb_clear.c
index 8da664afb..887e58bd9 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_clear.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_clear.c
@@ -51,6 +51,7 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "util/u_format.h"
+#include "util/u_framebuffer.h"
 #include "util/u_inlines.h"
 #include "util/u_simple_shaders.h"
 #include "util/u_draw_quad.h"
@@ -129,6 +130,26 @@ set_vertex_shader(struct st_context *st)
 }
 
 
+static void
+set_vertex_shader_layered(struct st_context *st)
+{
+   struct pipe_context *pipe = st->pipe;
+
+   if (!pipe->screen->get_param(pipe->screen, PIPE_CAP_TGSI_INSTANCEID) ||
+       !pipe->screen->get_param(pipe->screen, PIPE_CAP_TGSI_VS_LAYER)) {
+      assert(!"Got layered clear, but the VS layer output is unsupported");
+      set_vertex_shader(st);
+      return;
+   }
+
+   if (!st->clear.vs_layered) {
+      st->clear.vs_layered = util_make_layered_clear_vertex_shader(pipe);
+   }
+
+   cso_set_vertex_shader_handle(st->cso_context, st->clear.vs_layered);
+}
+
+
 /**
  * Draw a screen-aligned quadrilateral.
  * Coords are clip coords with y=0=bottom.
@@ -136,15 +157,19 @@ set_vertex_shader(struct st_context *st)
 static void
 draw_quad(struct st_context *st,
           float x0, float y0, float x1, float y1, GLfloat z,
+          unsigned num_instances,
           const union pipe_color_union *color)
 {
-   struct pipe_context *pipe = st->pipe;
-   struct pipe_resource *vbuf = NULL;
-   GLuint i, offset;
+   struct cso_context *cso = st->cso_context;
+   struct pipe_vertex_buffer vb = {0};
+   GLuint i;
    float (*vertices)[2][4];  /**< vertex pos + color */
 
+   vb.stride = 8 * sizeof(float);
+
    if (u_upload_alloc(st->uploader, 0, 4 * sizeof(vertices[0]),
-                      &offset, &vbuf, (void **) &vertices) != PIPE_OK) {
+                      &vb.buffer_offset, &vb.buffer,
+                      (void **) &vertices) != PIPE_OK) {
       return;
    }
 
@@ -174,16 +199,10 @@ draw_quad(struct st_context *st,
    u_upload_unmap(st->uploader);
 
    /* draw */
-   util_draw_vertex_buffer(pipe,
-                           st->cso_context,
-                           vbuf,
-                           cso_get_aux_vertex_buffer_slot(st->cso_context),
-                           offset,
-                           PIPE_PRIM_TRIANGLE_FAN,
-                           4,  /* verts */
-                           2); /* attribs/vert */
-
-   pipe_resource_reference(&vbuf, NULL);
+   cso_set_vertex_buffers(cso, cso_get_aux_vertex_buffer_slot(cso), 1, &vb);
+   cso_draw_arrays_instanced(cso, PIPE_PRIM_TRIANGLE_FAN, 0, 4,
+                             0, num_instances);
+   pipe_resource_reference(&vb.buffer, NULL);
 }
 
 
@@ -194,8 +213,7 @@ draw_quad(struct st_context *st,
  * ctx->DrawBuffer->_X/Ymin/max fields.
  */
 static void
-clear_with_quad(struct gl_context *ctx,
-                GLboolean color, GLboolean depth, GLboolean stencil)
+clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
 {
    struct st_context *st = st_context(ctx);
    const struct gl_framebuffer *fb = ctx->DrawBuffer;
@@ -205,7 +223,8 @@ clear_with_quad(struct gl_context *ctx,
    const GLfloat x1 = (GLfloat) ctx->DrawBuffer->_Xmax / fb_width * 2.0f - 1.0f;
    const GLfloat y0 = (GLfloat) ctx->DrawBuffer->_Ymin / fb_height * 2.0f - 1.0f;
    const GLfloat y1 = (GLfloat) ctx->DrawBuffer->_Ymax / fb_height * 2.0f - 1.0f;
-   union pipe_color_union clearColor;
+   unsigned num_layers =
+      util_framebuffer_get_num_layers(&st->state.framebuffer);
 
    /*
    printf("%s %s%s%s %f,%f %f,%f\n", __FUNCTION__, 
@@ -233,7 +252,7 @@ clear_with_quad(struct gl_context *ctx,
    {
       struct pipe_blend_state blend;
       memset(&blend, 0, sizeof(blend));
-      if (color) {
+      if (clear_buffers & PIPE_CLEAR_COLOR) {
          int num_buffers = ctx->Extensions.EXT_draw_buffers2 ?
                            ctx->DrawBuffer->_NumColorDrawBuffers : 1;
          int i;
@@ -241,6 +260,9 @@ clear_with_quad(struct gl_context *ctx,
          blend.independent_blend_enable = num_buffers > 1;
 
          for (i = 0; i < num_buffers; i++) {
+            if (!(clear_buffers & (PIPE_CLEAR_COLOR0 << i)))
+               continue;
+
             if (ctx->Color.ColorMask[i][0])
                blend.rt[i].colormask |= PIPE_MASK_R;
             if (ctx->Color.ColorMask[i][1])
@@ -261,13 +283,13 @@ clear_with_quad(struct gl_context *ctx,
    {
       struct pipe_depth_stencil_alpha_state depth_stencil;
       memset(&depth_stencil, 0, sizeof(depth_stencil));
-      if (depth) {
+      if (clear_buffers & PIPE_CLEAR_DEPTH) {
          depth_stencil.depth.enabled = 1;
          depth_stencil.depth.writemask = 1;
          depth_stencil.depth.func = PIPE_FUNC_ALWAYS;
       }
 
-      if (stencil) {
+      if (clear_buffers & PIPE_CLEAR_STENCIL) {
          struct pipe_stencil_ref stencil_ref;
          memset(&stencil_ref, 0, sizeof(stencil_ref));
          depth_stencil.stencil[0].enabled = 1;
@@ -305,21 +327,20 @@ clear_with_quad(struct gl_context *ctx,
    }
 
    set_fragment_shader(st);
-   set_vertex_shader(st);
    cso_set_geometry_shader_handle(st->cso_context, NULL);
 
-   if (ctx->DrawBuffer->_ColorDrawBuffers[0]) {
-      struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
-      GLboolean is_integer = _mesa_is_enum_format_integer(rb->InternalFormat);
+   if (num_layers > 1)
+      set_vertex_shader_layered(st);
+   else
+      set_vertex_shader(st);
 
-      st_translate_color(&ctx->Color.ClearColor,
-                         &clearColor,
-                         ctx->DrawBuffer->_ColorDrawBuffers[0]->_BaseFormat,
-                         is_integer);
-   }
+   /* We can't translate the clear color to the colorbuffer format,
+    * because different colorbuffers may have different formats.
+    */
 
    /* draw quad matching scissor rect */
-   draw_quad(st, x0, y0, x1, y1, (GLfloat) ctx->Depth.Clear, &clearColor);
+   draw_quad(st, x0, y0, x1, y1, (GLfloat) ctx->Depth.Clear, num_layers,
+             (union pipe_color_union*)&ctx->Color.ClearColor);
 
    /* Restore pipe state */
    cso_restore_blend(st->cso_context);
@@ -352,6 +373,19 @@ is_scissor_enabled(struct gl_context *ctx, struct gl_renderbuffer *rb)
 
 
 /**
+ * Return if all of the color channels are masked.
+ */
+static INLINE GLboolean
+is_color_disabled(struct gl_context *ctx, int i)
+{
+   return !ctx->Color.ColorMask[i][0] &&
+          !ctx->Color.ColorMask[i][1] &&
+          !ctx->Color.ColorMask[i][2] &&
+          !ctx->Color.ColorMask[i][3];
+}
+
+
+/**
  * Return if any of the color channels are masked.
  */
 static INLINE GLboolean
@@ -408,11 +442,14 @@ st_Clear(struct gl_context *ctx, GLbitfield mask)
             if (!strb || !strb->surface)
                continue;
 
+            if (is_color_disabled(ctx, colormask_index))
+               continue;
+
             if (is_scissor_enabled(ctx, rb) ||
                 is_color_masked(ctx, colormask_index))
-               quad_buffers |= PIPE_CLEAR_COLOR;
+               quad_buffers |= PIPE_CLEAR_COLOR0 << i;
             else
-               clear_buffers |= PIPE_CLEAR_COLOR;
+               clear_buffers |= PIPE_CLEAR_COLOR0 << i;
          }
       }
    }
@@ -445,24 +482,13 @@ st_Clear(struct gl_context *ctx, GLbitfield mask)
     */
    if (quad_buffers) {
       quad_buffers |= clear_buffers;
-      clear_with_quad(ctx,
-                      quad_buffers & PIPE_CLEAR_COLOR,
-                      quad_buffers & PIPE_CLEAR_DEPTH,
-                      quad_buffers & PIPE_CLEAR_STENCIL);
+      clear_with_quad(ctx, quad_buffers);
    } else if (clear_buffers) {
-      union pipe_color_union clearColor;
-
-      if (ctx->DrawBuffer->_ColorDrawBuffers[0]) {
-         struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
-         GLboolean is_integer = _mesa_is_enum_format_integer(rb->InternalFormat);
-
-         st_translate_color(&ctx->Color.ClearColor,
-                            &clearColor,
-			    ctx->DrawBuffer->_ColorDrawBuffers[0]->_BaseFormat,
-			    is_integer);
-      }
-
-      st->pipe->clear(st->pipe, clear_buffers, &clearColor,
+      /* We can't translate the clear color to the colorbuffer format,
+       * because different colorbuffers may have different formats.
+       */
+      st->pipe->clear(st->pipe, clear_buffers,
+                      (union pipe_color_union*)&ctx->Color.ClearColor,
                       ctx->Depth.Clear, ctx->Stencil.Clear);
    }
    if (mask & BUFFER_BIT_ACCUM)
diff --git a/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c b/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
index 2ce4728ad..3058dfb5b 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -878,7 +878,8 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
    }
 
    stmap = pipe_transfer_map(pipe, strb->texture,
-                             strb->rtt_level, strb->rtt_face + strb->rtt_slice,
+                             strb->surface->u.tex.level,
+                             strb->surface->u.tex.first_layer,
                              usage, x, y,
                              width, height, &pt);
 
@@ -1263,8 +1264,8 @@ copy_stencil_pixels(struct gl_context *ctx, GLint srcx, GLint srcy,
    /* map the stencil buffer */
    drawMap = pipe_transfer_map(pipe,
                                rbDraw->texture,
-                               rbDraw->rtt_level,
-                               rbDraw->rtt_face + rbDraw->rtt_slice,
+                               rbDraw->surface->u.tex.level,
+                               rbDraw->surface->u.tex.first_layer,
                                usage, dstx, dsty,
                                width, height, &ptDraw);
 
@@ -1422,20 +1423,20 @@ blit_copy_pixels(struct gl_context *ctx, GLint srcx, GLint srcy,
 
          memset(&blit, 0, sizeof(blit));
          blit.src.resource = rbRead->texture;
-         blit.src.level = rbRead->rtt_level;
+         blit.src.level = rbRead->surface->u.tex.level;
          blit.src.format = rbRead->texture->format;
          blit.src.box.x = readX;
          blit.src.box.y = readY;
-         blit.src.box.z = rbRead->rtt_face + rbRead->rtt_slice;
+         blit.src.box.z = rbRead->surface->u.tex.first_layer;
          blit.src.box.width = readW;
          blit.src.box.height = readH;
          blit.src.box.depth = 1;
          blit.dst.resource = rbDraw->texture;
-         blit.dst.level = rbDraw->rtt_level;
+         blit.dst.level = rbDraw->surface->u.tex.level;
          blit.dst.format = rbDraw->texture->format;
          blit.dst.box.x = drawX;
          blit.dst.box.y = drawY;
-         blit.dst.box.z = rbDraw->rtt_face + rbDraw->rtt_slice;
+         blit.dst.box.z = rbDraw->surface->u.tex.first_layer;
          blit.dst.box.width = drawW;
          blit.dst.box.height = drawH;
          blit.dst.box.depth = 1;
@@ -1633,11 +1634,11 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy,
 
       memset(&blit, 0, sizeof(blit));
       blit.src.resource = rbRead->texture;
-      blit.src.level = rbRead->rtt_level;
+      blit.src.level = rbRead->surface->u.tex.level;
       blit.src.format = rbRead->texture->format;
       blit.src.box.x = readX;
       blit.src.box.y = readY;
-      blit.src.box.z = rbRead->rtt_face + rbRead->rtt_slice;
+      blit.src.box.z = rbRead->surface->u.tex.first_layer;
       blit.src.box.width = readW;
       blit.src.box.height = readH;
       blit.src.box.depth = 1;
diff --git a/mesalib/src/mesa/state_tracker/st_cb_fbo.c b/mesalib/src/mesa/state_tracker/st_cb_fbo.c
index 20894825f..70baa9965 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_fbo.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_fbo.c
@@ -389,6 +389,72 @@ st_bind_framebuffer(struct gl_context *ctx, GLenum target,
 
 
 /**
+ * Create or update the pipe_surface of a FBO renderbuffer.
+ * This is usually called after st_finalize_texture.
+ */
+void
+st_update_renderbuffer_surface(struct st_context *st,
+                               struct st_renderbuffer *strb)
+{
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_resource *resource = strb->texture;
+   int rtt_width = strb->Base.Width;
+   int rtt_height = strb->Base.Height;
+   int rtt_depth = strb->Base.Depth;
+   enum pipe_format format = st->ctx->Color.sRGBEnabled ? resource->format :
+         util_format_linear(resource->format);
+   unsigned first_layer, last_layer, level;
+
+   if (resource->target == PIPE_TEXTURE_1D_ARRAY) {
+      rtt_depth = rtt_height;
+      rtt_height = 1;
+   }
+
+   /* find matching mipmap level size */
+   for (level = 0; level <= resource->last_level; level++) {
+      if (u_minify(resource->width0, level) == rtt_width &&
+          u_minify(resource->height0, level) == rtt_height &&
+          (resource->target != PIPE_TEXTURE_3D ||
+           u_minify(resource->depth0, level) == rtt_depth)) {
+         break;
+      }
+   }
+   assert(level <= resource->last_level);
+
+   /* determine the layer bounds */
+   if (strb->rtt_layered) {
+      first_layer = 0;
+      last_layer = util_max_layer(strb->texture, level);
+   }
+   else {
+      first_layer =
+      last_layer = strb->rtt_face + strb->rtt_slice;
+   }
+
+   if (!strb->surface ||
+       strb->surface->texture->nr_samples != strb->Base.NumSamples ||
+       strb->surface->format != format ||
+       strb->surface->texture != resource ||
+       strb->surface->width != rtt_width ||
+       strb->surface->height != rtt_height ||
+       strb->surface->u.tex.level != level ||
+       strb->surface->u.tex.first_layer != first_layer ||
+       strb->surface->u.tex.last_layer != last_layer) {
+      /* create a new pipe_surface */
+      struct pipe_surface surf_tmpl;
+      memset(&surf_tmpl, 0, sizeof(surf_tmpl));
+      surf_tmpl.format = format;
+      surf_tmpl.u.tex.level = level;
+      surf_tmpl.u.tex.first_layer = first_layer;
+      surf_tmpl.u.tex.last_layer = last_layer;
+
+      pipe_surface_reference(&strb->surface, NULL);
+
+      strb->surface = pipe->create_surface(pipe, resource, &surf_tmpl);
+   }
+}
+
+/**
  * Called by ctx->Driver.RenderTexture
  */
 static void
@@ -401,8 +467,6 @@ st_render_texture(struct gl_context *ctx,
    struct gl_renderbuffer *rb = att->Renderbuffer;
    struct st_renderbuffer *strb = st_renderbuffer(rb);
    struct pipe_resource *pt;
-   struct st_texture_object *stObj;
-   struct pipe_surface surf_tmpl;
 
    if (!st_finalize_texture(ctx, pipe, att->Texture))
       return;
@@ -410,31 +474,16 @@ st_render_texture(struct gl_context *ctx,
    pt = st_get_texobj_resource(att->Texture);
    assert(pt);
 
-   /* get the texture for the texture object */
-   stObj = st_texture_object(att->Texture);
-
    /* point renderbuffer at texobject */
-   strb->rtt = stObj;
-   strb->rtt_level = att->TextureLevel;
+   strb->is_rtt = TRUE;
    strb->rtt_face = att->CubeMapFace;
    strb->rtt_slice = att->Zoffset;
-
-   pipe_resource_reference( &strb->texture, pt );
+   strb->rtt_layered = att->Layered;
+   pipe_resource_reference(&strb->texture, pt);
 
    pipe_surface_release(pipe, &strb->surface);
 
-   assert(strb->rtt_level <= strb->texture->last_level);
-
-   /* new surface for rendering into the texture */
-   memset(&surf_tmpl, 0, sizeof(surf_tmpl));
-   surf_tmpl.format = ctx->Color.sRGBEnabled
-      ? strb->texture->format : util_format_linear(strb->texture->format);
-   surf_tmpl.u.tex.level = strb->rtt_level;
-   surf_tmpl.u.tex.first_layer = strb->rtt_face + strb->rtt_slice;
-   surf_tmpl.u.tex.last_layer = strb->rtt_face + strb->rtt_slice;
-   strb->surface = pipe->create_surface(pipe,
-                                        strb->texture,
-                                        &surf_tmpl);
+   st_update_renderbuffer_surface(st, strb);
 
    strb->Base.Format = st_pipe_format_to_mesa_format(pt->format);
 
@@ -464,7 +513,7 @@ st_finish_render_texture(struct gl_context *ctx, struct gl_renderbuffer *rb)
    if (!strb)
       return;
 
-   strb->rtt = NULL;
+   strb->is_rtt = FALSE;
 
    /* restore previous framebuffer state */
    st_invalidate_state(ctx, _NEW_BUFFERS);
@@ -706,8 +755,8 @@ st_MapRenderbuffer(struct gl_context *ctx,
 
     map = pipe_transfer_map(pipe,
                             strb->texture,
-                            strb->rtt_level,
-                            strb->rtt_face + strb->rtt_slice,
+                            strb->surface->u.tex.level,
+                            strb->surface->u.tex.first_layer,
                             usage, x, y2, w, h, &strb->transfer);
    if (map) {
       if (invert) {
diff --git a/mesalib/src/mesa/state_tracker/st_cb_fbo.h b/mesalib/src/mesa/state_tracker/st_cb_fbo.h
index f335c371b..88fccc298 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_fbo.h
+++ b/mesalib/src/mesa/state_tracker/st_cb_fbo.h
@@ -58,8 +58,10 @@ struct st_renderbuffer
    boolean software;
    void *data;
 
-   struct st_texture_object *rtt;  /**< GL render to texture's texture */
-   unsigned rtt_level, rtt_face, rtt_slice;
+   /* Inputs from Driver.RenderTexture, don't use directly. */
+   boolean is_rtt; /**< whether Driver.RenderTexture was called */
+   unsigned rtt_face, rtt_slice;
+   boolean rtt_layered; /**< whether glFramebufferTexture was called */
 };
 
 
@@ -74,6 +76,10 @@ extern struct gl_renderbuffer *
 st_new_renderbuffer_fb(enum pipe_format format, int samples, boolean sw);
 
 extern void
+st_update_renderbuffer_surface(struct st_context *st,
+                               struct st_renderbuffer *strb);
+
+extern void
 st_init_fbo_functions(struct dd_function_table *functions);
 
 #endif /* ST_CB_FBO_H */
diff --git a/mesalib/src/mesa/state_tracker/st_cb_readpixels.c b/mesalib/src/mesa/state_tracker/st_cb_readpixels.c
index b5df58c03..7547dfd8b 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_readpixels.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_readpixels.c
@@ -166,7 +166,7 @@ st_readpixels(struct gl_context *ctx, GLint x, GLint y,
    }
 
    blit.src.resource = src;
-   blit.src.level = strb->rtt_level;
+   blit.src.level = strb->surface->u.tex.level;
    blit.src.format = src_format;
    blit.dst.resource = dst;
    blit.dst.level = 0;
@@ -175,7 +175,7 @@ st_readpixels(struct gl_context *ctx, GLint x, GLint y,
    blit.dst.box.x = 0;
    blit.src.box.y = y;
    blit.dst.box.y = 0;
-   blit.src.box.z = strb->rtt_face + strb->rtt_slice;
+   blit.src.box.z = strb->surface->u.tex.first_layer;
    blit.dst.box.z = 0;
    blit.src.box.width = blit.dst.box.width = width;
    blit.src.box.height = blit.dst.box.height = height;
diff --git a/mesalib/src/mesa/state_tracker/st_cb_texture.c b/mesalib/src/mesa/state_tracker/st_cb_texture.c
index faa9ee3f6..c09f34e37 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_texture.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_texture.c
@@ -175,10 +175,8 @@ st_FreeTextureImageBuffer(struct gl_context *ctx,
       pipe_resource_reference(&stImage->pt, NULL);
    }
 
-   if (stImage->TexData) {
-      _mesa_align_free(stImage->TexData);
-      stImage->TexData = NULL;
-   }
+   _mesa_align_free(stImage->TexData);
+   stImage->TexData = NULL;
 }
 
 
@@ -1156,8 +1154,8 @@ fallback_copy_texsubimage(struct gl_context *ctx,
 
    map = pipe_transfer_map(pipe,
                            strb->texture,
-                           strb->rtt_level,
-                           strb->rtt_face + strb->rtt_slice,
+                           strb->surface->u.tex.level,
+                           strb->surface->u.tex.first_layer,
                            PIPE_TRANSFER_READ,
                            srcX, srcY,
                            width, height, &src_trans);
diff --git a/mesalib/src/mesa/state_tracker/st_context.h b/mesalib/src/mesa/state_tracker/st_context.h
index ab89b4947..cd0a5ae98 100644
--- a/mesalib/src/mesa/state_tracker/st_context.h
+++ b/mesalib/src/mesa/state_tracker/st_context.h
@@ -178,6 +178,7 @@ struct st_context
       struct pipe_viewport_state viewport;
       void *vs;
       void *fs;
+      void *vs_layered;
    } clear;
 
    /** used for anything using util_draw_vertex_buffer */
diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c
index cd10a0c9d..5e4a3b398 100644
--- a/mesalib/src/mesa/state_tracker/st_extensions.c
+++ b/mesalib/src/mesa/state_tracker/st_extensions.c
@@ -406,7 +406,6 @@ void st_init_extensions(struct st_context *st)
       { o(NV_texture_barrier),               PIPE_CAP_TEXTURE_BARRIER                  },
       /* GL_NV_point_sprite is not supported by gallium because we don't
        * support the GL_POINT_SPRITE_R_MODE_NV option. */
-      { o(MESA_texture_array),               PIPE_CAP_MAX_TEXTURE_ARRAY_LAYERS         },
 
       { o(OES_standard_derivatives),         PIPE_CAP_SM3                              },
       { o(ARB_texture_cube_map_array),       PIPE_CAP_CUBE_MAP_ARRAY                   },
@@ -444,11 +443,6 @@ void st_init_extensions(struct st_context *st)
       { { o(ARB_depth_buffer_float) },
         { PIPE_FORMAT_Z32_FLOAT,
           PIPE_FORMAT_Z32_FLOAT_S8X24_UINT } },
-
-      { { o(EXT_packed_depth_stencil) },
-        { PIPE_FORMAT_S8_UINT_Z24_UNORM,
-          PIPE_FORMAT_Z24_UNORM_S8_UINT },
-        GL_TRUE }, /* at least one format must be supported */
    };
 
    /* Required: sampler support */
@@ -605,6 +599,13 @@ void st_init_extensions(struct st_context *st)
       ctx->Const.ForceGLSLVersion = st->options.force_glsl_version;
    }
 
+   /* This extension needs full OpenGL 3.2, but we don't know if that's
+    * supported at this point. Only check the GLSL version. */
+   if (ctx->Const.GLSLVersion >= 150 &&
+       screen->get_param(screen, PIPE_CAP_TGSI_VS_LAYER)) {
+      ctx->Extensions.AMD_vertex_shader_layer = GL_TRUE;
+   }
+
    if (ctx->Const.GLSLVersion >= 130) {
       ctx->Const.NativeIntegers = GL_TRUE;
       ctx->Const.MaxClipPlanes = 8;
@@ -758,8 +759,7 @@ void st_init_extensions(struct st_context *st)
                              PIPE_BUFFER, PIPE_BIND_SAMPLER_VIEW);
    }
 
-   if (screen->get_param(screen, PIPE_CAP_MIXED_FRAMEBUFFER_SIZES) &&
-       ctx->Extensions.EXT_packed_depth_stencil) {
+   if (screen->get_param(screen, PIPE_CAP_MIXED_FRAMEBUFFER_SIZES)) {
       ctx->Extensions.ARB_framebuffer_object = GL_TRUE;
    }
 
diff --git a/mesalib/src/mesa/state_tracker/st_format.c b/mesalib/src/mesa/state_tracker/st_format.c
index ec2552384..6acf98390 100644
--- a/mesalib/src/mesa/state_tracker/st_format.c
+++ b/mesalib/src/mesa/state_tracker/st_format.c
@@ -1854,12 +1854,12 @@ st_QuerySamplesForFormat(struct gl_context *ctx, GLenum target,
  * Similarly for texture border colors.
  */
 void
-st_translate_color(union gl_color_union *colorIn,
+st_translate_color(const union gl_color_union *colorIn,
                    union pipe_color_union *colorOut,
                    GLenum baseFormat, GLboolean is_integer)
 {
    if (is_integer) {
-      int *in = colorIn->i;
+      const int *in = colorIn->i;
       int *out = colorOut->i;
 
       switch (baseFormat) {
@@ -1901,7 +1901,7 @@ st_translate_color(union gl_color_union *colorIn,
       }
    }
    else {
-      float *in = colorIn->f;
+      const float *in = colorIn->f;
       float *out = colorOut->f;
 
       switch (baseFormat) {
diff --git a/mesalib/src/mesa/state_tracker/st_format.h b/mesalib/src/mesa/state_tracker/st_format.h
index 6e97dcb96..3278748d0 100644
--- a/mesalib/src/mesa/state_tracker/st_format.h
+++ b/mesalib/src/mesa/state_tracker/st_format.h
@@ -73,7 +73,7 @@ st_QuerySamplesForFormat(struct gl_context *ctx, GLenum target,
 
 
 extern void
-st_translate_color(union gl_color_union *colorIn,
+st_translate_color(const union gl_color_union *colorIn,
                    union pipe_color_union *colorOut,
                    GLenum baseFormat, GLboolean is_integer);
 
diff --git a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index ac95968d6..1331c73dc 100644
--- a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -1056,11 +1056,11 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
    if (strcmp(ir->name, "gl_FragCoord") == 0) {
       struct gl_fragment_program *fp = (struct gl_fragment_program *)this->prog;
 
-      fp->OriginUpperLeft = ir->origin_upper_left;
-      fp->PixelCenterInteger = ir->pixel_center_integer;
+      fp->OriginUpperLeft = ir->data.origin_upper_left;
+      fp->PixelCenterInteger = ir->data.pixel_center_integer;
    }
 
-   if (ir->mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
+   if (ir->data.mode == ir_var_uniform && strncmp(ir->name, "gl_", 3) == 0) {
       unsigned int i;
       const ir_state_slot *const slots = ir->state_slots;
       assert(ir->state_slots != NULL);
@@ -1137,53 +1137,10 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
 void
 glsl_to_tgsi_visitor::visit(ir_loop *ir)
 {
-   ir_dereference_variable *counter = NULL;
-
-   if (ir->counter != NULL)
-      counter = new(ir) ir_dereference_variable(ir->counter);
-
-   if (ir->from != NULL) {
-      assert(ir->counter != NULL);
-
-      ir_assignment *a = new(ir) ir_assignment(counter, ir->from, NULL);
-
-      a->accept(this);
-      delete a;
-   }
-
    emit(NULL, TGSI_OPCODE_BGNLOOP);
 
-   if (ir->to) {
-      ir_expression *e =
-         new(ir) ir_expression(ir->cmp, glsl_type::bool_type,
-        		       counter, ir->to);
-      ir_if *if_stmt =  new(ir) ir_if(e);
-
-      ir_loop_jump *brk = new(ir) ir_loop_jump(ir_loop_jump::jump_break);
-
-      if_stmt->then_instructions.push_tail(brk);
-
-      if_stmt->accept(this);
-
-      delete if_stmt;
-      delete e;
-      delete brk;
-   }
-
    visit_exec_list(&ir->body_instructions, this);
 
-   if (ir->increment) {
-      ir_expression *e =
-         new(ir) ir_expression(ir_binop_add, counter->type,
-        		       counter, ir->increment);
-
-      ir_assignment *a = new(ir) ir_assignment(counter, e, NULL);
-
-      a->accept(this);
-      delete a;
-      delete e;
-   }
-
    emit(NULL, TGSI_OPCODE_ENDLOOP);
 }
 
@@ -2063,10 +2020,10 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
    ir_variable *var = ir->var;
 
    if (!entry) {
-      switch (var->mode) {
+      switch (var->data.mode) {
       case ir_var_uniform:
          entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM,
-        				       var->location);
+        				       var->data.location);
          this->variables.push_tail(entry);
          break;
       case ir_var_shader_in:
@@ -2075,21 +2032,22 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
           * generic attributes (glBindVertexLocation), and
           * user-defined varyings.
           */
-         assert(var->location != -1);
+         assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_INPUT,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_shader_out:
-         assert(var->location != -1);
+         assert(var->data.location != -1);
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_OUTPUT,
-                                               var->location + var->index);
+                                               var->data.location
+                                               + var->data.index);
          break;
       case ir_var_system_value:
          entry = new(mem_ctx) variable_storage(var,
                                                PROGRAM_SYSTEM_VALUE,
-                                               var->location);
+                                               var->data.location);
          break;
       case ir_var_auto:
       case ir_var_temporary:
@@ -2388,7 +2346,7 @@ glsl_to_tgsi_visitor::visit(ir_assignment *ir)
       assert(!ir->lhs->type->is_scalar() && !ir->lhs->type->is_vector());
       l.writemask = WRITEMASK_XYZW;
    } else if (ir->lhs->type->is_scalar() &&
-              ir->lhs->variable_referenced()->mode == ir_var_shader_out) {
+              ir->lhs->variable_referenced()->data.mode == ir_var_shader_out) {
       /* FINISHME: This hack makes writing to gl_FragDepth, which lives in the
        * FINISHME: W component of fragment shader output zero, work correctly.
        */
@@ -2664,8 +2622,8 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
       ir_rvalue *param_rval = (ir_rvalue *)iter.get();
       ir_variable *param = (ir_variable *)sig_iter.get();
 
-      if (param->mode == ir_var_function_in ||
-          param->mode == ir_var_function_inout) {
+      if (param->data.mode == ir_var_function_in ||
+          param->data.mode == ir_var_function_inout) {
          variable_storage *storage = find_variable_storage(param);
          assert(storage);
 
@@ -2700,8 +2658,8 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
       ir_rvalue *param_rval = (ir_rvalue *)iter.get();
       ir_variable *param = (ir_variable *)sig_iter.get();
 
-      if (param->mode == ir_var_function_out ||
-          param->mode == ir_var_function_inout) {
+      if (param->data.mode == ir_var_function_out ||
+          param->data.mode == ir_var_function_inout) {
          variable_storage *storage = find_variable_storage(param);
          assert(storage);
 
diff --git a/mesalib/src/mesa/state_tracker/st_program.c b/mesalib/src/mesa/state_tracker/st_program.c
index f010e1889..f72122b4f 100644
--- a/mesalib/src/mesa/state_tracker/st_program.c
+++ b/mesalib/src/mesa/state_tracker/st_program.c
@@ -258,6 +258,10 @@ st_prepare_vertex_program(struct gl_context *ctx,
             stvp->output_semantic_name[slot] = TGSI_SEMANTIC_CLIPVERTEX;
             stvp->output_semantic_index[slot] = 0;
             break;
+         case VARYING_SLOT_LAYER:
+            stvp->output_semantic_name[slot] = TGSI_SEMANTIC_LAYER;
+            stvp->output_semantic_index[slot] = 0;
+            break;
 
          case VARYING_SLOT_TEX0:
          case VARYING_SLOT_TEX1:
@@ -563,6 +567,11 @@ st_translate_fragment_program(struct st_context *st,
             input_semantic_index[slot] = 0;
             interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
             break;
+         case VARYING_SLOT_PRIMITIVE_ID:
+            input_semantic_name[slot] = TGSI_SEMANTIC_PRIMID;
+            input_semantic_index[slot] = 0;
+            interpMode[slot] = TGSI_INTERPOLATE_CONSTANT;
+            break;
          case VARYING_SLOT_CLIP_DIST0:
             input_semantic_name[slot] = TGSI_SEMANTIC_CLIPDIST;
             input_semantic_index[slot] = 0;
diff --git a/mesalib/src/mesa/state_tracker/st_texture.h b/mesalib/src/mesa/state_tracker/st_texture.h
index c15aeaea6..ac93d9655 100644
--- a/mesalib/src/mesa/state_tracker/st_texture.h
+++ b/mesalib/src/mesa/state_tracker/st_texture.h
@@ -112,6 +112,12 @@ st_texture_object(struct gl_texture_object *obj)
    return (struct st_texture_object *) obj;
 }
 
+static INLINE const struct st_texture_object *
+st_texture_object_const(const struct gl_texture_object *obj)
+{
+   return (const struct st_texture_object *) obj;
+}
+
 
 static INLINE struct pipe_resource *
 st_get_texobj_resource(struct gl_texture_object *texObj)
diff --git a/mesalib/src/mesa/swrast/s_texture.c b/mesalib/src/mesa/swrast/s_texture.c
index 27803c553..c08a4e9d1 100644
--- a/mesalib/src/mesa/swrast/s_texture.c
+++ b/mesalib/src/mesa/swrast/s_texture.c
@@ -164,10 +164,9 @@ _swrast_free_texture_image_buffer(struct gl_context *ctx,
                                   struct gl_texture_image *texImage)
 {
    struct swrast_texture_image *swImage = swrast_texture_image(texImage);
-   if (swImage->Buffer) {
-      _mesa_align_free(swImage->Buffer);
-      swImage->Buffer = NULL;
-   }
+
+   _mesa_align_free(swImage->Buffer);
+   swImage->Buffer = NULL;
 
    free(swImage->ImageSlices);
    swImage->ImageSlices = NULL;
diff --git a/mesalib/src/mesa/tnl/t_vertex.c b/mesalib/src/mesa/tnl/t_vertex.c
index c7a745ed7..8c4195eed 100644
--- a/mesalib/src/mesa/tnl/t_vertex.c
+++ b/mesalib/src/mesa/tnl/t_vertex.c
@@ -546,10 +546,8 @@ void _tnl_free_vertices( struct gl_context *ctx )
       struct tnl_clipspace *vtx = GET_VERTEX_STATE(ctx);
       struct tnl_clipspace_fastpath *fp, *tmp;
 
-      if (vtx->vertex_buf) {
-         _mesa_align_free(vtx->vertex_buf);
-         vtx->vertex_buf = NULL;
-      }
+      _mesa_align_free(vtx->vertex_buf);
+      vtx->vertex_buf = NULL;
 
       for (fp = vtx->fastpath ; fp ; fp = tmp) {
          tmp = fp->next;
diff --git a/mesalib/src/mesa/vbo/vbo_exec_api.c b/mesalib/src/mesa/vbo/vbo_exec_api.c
index c84d97f56..6feda1657 100644
--- a/mesalib/src/mesa/vbo/vbo_exec_api.c
+++ b/mesalib/src/mesa/vbo/vbo_exec_api.c
@@ -990,11 +990,10 @@ void vbo_use_buffer_objects(struct gl_context *ctx)
 
    /* Make sure this func is only used once */
    assert(exec->vtx.bufferobj == ctx->Shared->NullBufferObj);
-   if (exec->vtx.buffer_map) {
-      _mesa_align_free(exec->vtx.buffer_map);
-      exec->vtx.buffer_map = NULL;
-      exec->vtx.buffer_ptr = NULL;
-   }
+
+   _mesa_align_free(exec->vtx.buffer_map);
+   exec->vtx.buffer_map = NULL;
+   exec->vtx.buffer_ptr = NULL;
 
    /* Allocate a real buffer object now */
    _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL);
diff --git a/mesalib/src/mesa/x86/rtasm/x86sse.c b/mesalib/src/mesa/x86/rtasm/x86sse.c
deleted file mode 100644
index c93faba79..000000000
--- a/mesalib/src/mesa/x86/rtasm/x86sse.c
+++ /dev/null
@@ -1,1203 +0,0 @@
-#ifdef USE_X86_ASM
-#if defined(__i386__) || defined(__386__)
-
-#include "main/imports.h"
-#include "x86sse.h"
-
-#define DISASSEM 0
-#define X86_TWOB 0x0f
-
-#if 0
-static unsigned char *cptr( void (*label)() )
-{
-   return (unsigned char *)(unsigned long)label;
-}
-#endif
-
-
-static void do_realloc( struct x86_function *p )
-{
-   if (p->size == 0) {
-      p->size = 1024;
-      p->store = _mesa_exec_malloc(p->size);
-      p->csr = p->store;
-   }
-   else {
-      unsigned used = p->csr - p->store;
-      unsigned char *tmp = p->store;
-      p->size *= 2;
-      p->store = _mesa_exec_malloc(p->size);
-      memcpy(p->store, tmp, used);
-      p->csr = p->store + used;
-      _mesa_exec_free(tmp);
-   }
-}
-
-/* Emit bytes to the instruction stream:
- */
-static unsigned char *reserve( struct x86_function *p, int bytes )
-{
-   if (p->csr + bytes - p->store > p->size)
-      do_realloc(p);
-
-   {
-      unsigned char *csr = p->csr;
-      p->csr += bytes;
-      return csr;
-   }
-}
-
-
-
-static void emit_1b( struct x86_function *p, char b0 )
-{
-   char *csr = (char *)reserve(p, 1);
-   *csr = b0;
-}
-
-static void emit_1i( struct x86_function *p, int i0 )
-{
-   int *icsr = (int *)reserve(p, sizeof(i0));
-   *icsr = i0;
-}
-
-static void emit_1ub( struct x86_function *p, unsigned char b0 )
-{
-   unsigned char *csr = reserve(p, 1);
-   *csr++ = b0;
-}
-
-static void emit_2ub( struct x86_function *p, unsigned char b0, unsigned char b1 )
-{
-   unsigned char *csr = reserve(p, 2);
-   *csr++ = b0;
-   *csr++ = b1;
-}
-
-static void emit_3ub( struct x86_function *p, unsigned char b0, unsigned char b1, unsigned char b2 )
-{
-   unsigned char *csr = reserve(p, 3);
-   *csr++ = b0;
-   *csr++ = b1;
-   *csr++ = b2;
-}
-
-
-/* Build a modRM byte + possible displacement.  No treatment of SIB
- * indexing.  BZZT - no way to encode an absolute address.
- */
-static void emit_modrm( struct x86_function *p, 
-			struct x86_reg reg, 
-			struct x86_reg regmem )
-{
-   unsigned char val = 0;
-   
-   assert(reg.mod == mod_REG);
-   
-   val |= regmem.mod << 6;     	/* mod field */
-   val |= reg.idx << 3;		/* reg field */
-   val |= regmem.idx;		/* r/m field */
-   
-   emit_1ub(p, val);
-
-   /* Oh-oh we've stumbled into the SIB thing.
-    */
-   if (regmem.file == file_REG32 &&
-       regmem.idx == reg_SP) {
-      emit_1ub(p, 0x24);		/* simplistic! */
-   }
-
-   switch (regmem.mod) {
-   case mod_REG:
-   case mod_INDIRECT:
-      break;
-   case mod_DISP8:
-      emit_1b(p, regmem.disp);
-      break;
-   case mod_DISP32:
-      emit_1i(p, regmem.disp);
-      break;
-   default:
-      assert(0);
-      break;
-   }
-}
-
-
-static void emit_modrm_noreg( struct x86_function *p,
-			      unsigned op,
-			      struct x86_reg regmem )
-{
-   struct x86_reg dummy = x86_make_reg(file_REG32, op);
-   emit_modrm(p, dummy, regmem);
-}
-
-/* Many x86 instructions have two opcodes to cope with the situations
- * where the destination is a register or memory reference
- * respectively.  This function selects the correct opcode based on
- * the arguments presented.
- */
-static void emit_op_modrm( struct x86_function *p,
-			   unsigned char op_dst_is_reg, 
-			   unsigned char op_dst_is_mem,
-			   struct x86_reg dst,
-			   struct x86_reg src )
-{  
-   switch (dst.mod) {
-   case mod_REG:
-      emit_1ub(p, op_dst_is_reg);
-      emit_modrm(p, dst, src);
-      break;
-   case mod_INDIRECT:
-   case mod_DISP32:
-   case mod_DISP8:
-      assert(src.mod == mod_REG);
-      emit_1ub(p, op_dst_is_mem);
-      emit_modrm(p, src, dst);
-      break;
-   default:
-      assert(0);
-      break;
-   }
-}
-
-
-
-
-
-
-
-/* Create and manipulate registers and regmem values:
- */
-struct x86_reg x86_make_reg( enum x86_reg_file file,
-			     enum x86_reg_name idx )
-{
-   struct x86_reg reg;
-
-   reg.file = file;
-   reg.idx = idx;
-   reg.mod = mod_REG;
-   reg.disp = 0;
-
-   return reg;
-}
-
-struct x86_reg x86_make_disp( struct x86_reg reg,
-			      int disp )
-{
-   assert(reg.file == file_REG32);
-
-   if (reg.mod == mod_REG)
-      reg.disp = disp;
-   else
-      reg.disp += disp;
-
-   if (reg.disp == 0)
-      reg.mod = mod_INDIRECT;
-   else if (reg.disp <= 127 && reg.disp >= -128)
-      reg.mod = mod_DISP8;
-   else
-      reg.mod = mod_DISP32;
-
-   return reg;
-}
-
-struct x86_reg x86_deref( struct x86_reg reg )
-{
-   return x86_make_disp(reg, 0);
-}
-
-struct x86_reg x86_get_base_reg( struct x86_reg reg )
-{
-   return x86_make_reg( reg.file, reg.idx );
-}
-
-unsigned char *x86_get_label( struct x86_function *p )
-{
-   return p->csr;
-}
-
-
-
-/***********************************************************************
- * x86 instructions
- */
-
-
-void x86_jcc( struct x86_function *p,
-	      enum x86_cc cc,
-	      unsigned char *label )
-{
-   int offset = label - (x86_get_label(p) + 2);
-   
-   if (offset <= 127 && offset >= -128) {
-      emit_1ub(p, 0x70 + cc);
-      emit_1b(p, (char) offset);
-   }
-   else {
-      offset = label - (x86_get_label(p) + 6);
-      emit_2ub(p, 0x0f, 0x80 + cc);
-      emit_1i(p, offset);
-   }
-}
-
-/* Always use a 32bit offset for forward jumps:
- */
-unsigned char *x86_jcc_forward( struct x86_function *p,
-			  enum x86_cc cc )
-{
-   emit_2ub(p, 0x0f, 0x80 + cc);
-   emit_1i(p, 0);
-   return x86_get_label(p);
-}
-
-unsigned char *x86_jmp_forward( struct x86_function *p)
-{
-   emit_1ub(p, 0xe9);
-   emit_1i(p, 0);
-   return x86_get_label(p);
-}
-
-unsigned char *x86_call_forward( struct x86_function *p)
-{
-   emit_1ub(p, 0xe8);
-   emit_1i(p, 0);
-   return x86_get_label(p);
-}
-
-/* Fixup offset from forward jump:
- */
-void x86_fixup_fwd_jump( struct x86_function *p,
-			 unsigned char *fixup )
-{
-   *(int *)(fixup - 4) = x86_get_label(p) - fixup;
-}
-
-void x86_jmp( struct x86_function *p, unsigned char *label)
-{
-   emit_1ub(p, 0xe9);
-   emit_1i(p, label - x86_get_label(p) - 4);
-}
-
-#if 0
-/* This doesn't work once we start reallocating & copying the
- * generated code on buffer fills, because the call is relative to the
- * current pc.
- */
-void x86_call( struct x86_function *p, void (*label)())
-{
-   emit_1ub(p, 0xe8);
-   emit_1i(p, cptr(label) - x86_get_label(p) - 4);
-}
-#else
-void x86_call( struct x86_function *p, struct x86_reg reg)
-{
-   emit_1ub(p, 0xff);
-   emit_modrm_noreg(p, 2, reg);
-}
-#endif
-
-
-/* michal:
- * Temporary. As I need immediate operands, and dont want to mess with the codegen,
- * I load the immediate into general purpose register and use it.
- */
-void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm )
-{
-   assert(dst.mod == mod_REG);
-   emit_1ub(p, 0xb8 + dst.idx);
-   emit_1i(p, imm);
-}
-
-void x86_push( struct x86_function *p,
-	       struct x86_reg reg )
-{
-   assert(reg.mod == mod_REG);
-   emit_1ub(p, 0x50 + reg.idx);
-   p->stack_offset += 4;
-}
-
-void x86_pop( struct x86_function *p,
-	      struct x86_reg reg )
-{
-   assert(reg.mod == mod_REG);
-   emit_1ub(p, 0x58 + reg.idx);
-   p->stack_offset -= 4;
-}
-
-void x86_inc( struct x86_function *p,
-	      struct x86_reg reg )
-{
-   assert(reg.mod == mod_REG);
-   emit_1ub(p, 0x40 + reg.idx);
-}
-
-void x86_dec( struct x86_function *p,
-	      struct x86_reg reg )
-{
-   assert(reg.mod == mod_REG);
-   emit_1ub(p, 0x48 + reg.idx);
-}
-
-void x86_ret( struct x86_function *p )
-{
-   emit_1ub(p, 0xc3);
-}
-
-void x86_sahf( struct x86_function *p )
-{
-   emit_1ub(p, 0x9e);
-}
-
-void x86_mov( struct x86_function *p,
-	      struct x86_reg dst,
-	      struct x86_reg src )
-{
-   emit_op_modrm( p, 0x8b, 0x89, dst, src );
-}
-
-void x86_xor( struct x86_function *p,
-	      struct x86_reg dst,
-	      struct x86_reg src )
-{
-   emit_op_modrm( p, 0x33, 0x31, dst, src );
-}
-
-void x86_cmp( struct x86_function *p,
-	      struct x86_reg dst,
-	      struct x86_reg src )
-{
-   emit_op_modrm( p, 0x3b, 0x39, dst, src );
-}
-
-void x86_lea( struct x86_function *p,
-	      struct x86_reg dst,
-	      struct x86_reg src )
-{
-   emit_1ub(p, 0x8d);
-   emit_modrm( p, dst, src );
-}
-
-void x86_test( struct x86_function *p,
-	       struct x86_reg dst,
-	       struct x86_reg src )
-{
-   emit_1ub(p, 0x85);
-   emit_modrm( p, dst, src );
-}
-
-void x86_add( struct x86_function *p,
-	       struct x86_reg dst,
-	       struct x86_reg src )
-{
-   emit_op_modrm(p, 0x03, 0x01, dst, src );
-}
-
-void x86_mul( struct x86_function *p,
-	       struct x86_reg src )
-{
-   assert (src.file == file_REG32 && src.mod == mod_REG);
-   emit_op_modrm(p, 0xf7, 0, x86_make_reg (file_REG32, reg_SP), src );
-}
-
-void x86_sub( struct x86_function *p,
-	       struct x86_reg dst,
-	       struct x86_reg src )
-{
-   emit_op_modrm(p, 0x2b, 0x29, dst, src );
-}
-
-void x86_or( struct x86_function *p,
-             struct x86_reg dst,
-             struct x86_reg src )
-{
-   emit_op_modrm( p, 0x0b, 0x09, dst, src );
-}
-
-void x86_and( struct x86_function *p,
-              struct x86_reg dst,
-              struct x86_reg src )
-{
-   emit_op_modrm( p, 0x23, 0x21, dst, src );
-}
-
-
-
-/***********************************************************************
- * SSE instructions
- */
-
-
-void sse_movss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, 0xF3, X86_TWOB);
-   emit_op_modrm( p, 0x10, 0x11, dst, src );
-}
-
-void sse_movaps( struct x86_function *p,
-		 struct x86_reg dst,
-		 struct x86_reg src )
-{
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x28, 0x29, dst, src );
-}
-
-void sse_movups( struct x86_function *p,
-		 struct x86_reg dst,
-		 struct x86_reg src )
-{
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x10, 0x11, dst, src );
-}
-
-void sse_movhps( struct x86_function *p,
-		 struct x86_reg dst,
-		 struct x86_reg src )
-{
-   assert(dst.mod != mod_REG || src.mod != mod_REG);
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x16, 0x17, dst, src ); /* cf movlhps */
-}
-
-void sse_movlps( struct x86_function *p,
-		 struct x86_reg dst,
-		 struct x86_reg src )
-{
-   assert(dst.mod != mod_REG || src.mod != mod_REG);
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x12, 0x13, dst, src ); /* cf movhlps */
-}
-
-void sse_maxps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x5F);
-   emit_modrm( p, dst, src );
-}
-
-void sse_maxss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x5F);
-   emit_modrm( p, dst, src );
-}
-
-void sse_divss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x5E);
-   emit_modrm( p, dst, src );
-}
-
-void sse_minps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x5D);
-   emit_modrm( p, dst, src );
-}
-
-void sse_subps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x5C);
-   emit_modrm( p, dst, src );
-}
-
-void sse_mulps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x59);
-   emit_modrm( p, dst, src );
-}
-
-void sse_mulss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x59);
-   emit_modrm( p, dst, src );
-}
-
-void sse_addps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x58);
-   emit_modrm( p, dst, src );
-}
-
-void sse_addss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x58);
-   emit_modrm( p, dst, src );
-}
-
-void sse_andnps( struct x86_function *p,
-                 struct x86_reg dst,
-                 struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x55);
-   emit_modrm( p, dst, src );
-}
-
-void sse_andps( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x54);
-   emit_modrm( p, dst, src );
-}
-
-void sse_rsqrtps( struct x86_function *p,
-                  struct x86_reg dst,
-                  struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x52);
-   emit_modrm( p, dst, src );
-}
-
-void sse_rsqrtss( struct x86_function *p,
-		  struct x86_reg dst,
-		  struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x52);
-   emit_modrm( p, dst, src );
-
-}
-
-void sse_movhlps( struct x86_function *p,
-		  struct x86_reg dst,
-		  struct x86_reg src )
-{
-   assert(dst.mod == mod_REG && src.mod == mod_REG);
-   emit_2ub(p, X86_TWOB, 0x12);
-   emit_modrm( p, dst, src );
-}
-
-void sse_movlhps( struct x86_function *p,
-		  struct x86_reg dst,
-		  struct x86_reg src )
-{
-   assert(dst.mod == mod_REG && src.mod == mod_REG);
-   emit_2ub(p, X86_TWOB, 0x16);
-   emit_modrm( p, dst, src );
-}
-
-void sse_orps( struct x86_function *p,
-               struct x86_reg dst,
-               struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x56);
-   emit_modrm( p, dst, src );
-}
-
-void sse_xorps( struct x86_function *p,
-                struct x86_reg dst,
-                struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x57);
-   emit_modrm( p, dst, src );
-}
-
-void sse_cvtps2pi( struct x86_function *p,
-		   struct x86_reg dst,
-		   struct x86_reg src )
-{
-   assert(dst.file == file_MMX && 
-	  (src.file == file_XMM || src.mod != mod_REG));
-
-   p->need_emms = 1;
-
-   emit_2ub(p, X86_TWOB, 0x2d);
-   emit_modrm( p, dst, src );
-}
-
-
-/* Shufps can also be used to implement a reduced swizzle when dest ==
- * arg0.
- */
-void sse_shufps( struct x86_function *p,
-		 struct x86_reg dest,
-		 struct x86_reg arg0,
-		 unsigned char shuf) 
-{
-   emit_2ub(p, X86_TWOB, 0xC6);
-   emit_modrm(p, dest, arg0);
-   emit_1ub(p, shuf); 
-}
-
-void sse_cmpps( struct x86_function *p,
-		struct x86_reg dest,
-		struct x86_reg arg0,
-		unsigned char cc) 
-{
-   emit_2ub(p, X86_TWOB, 0xC2);
-   emit_modrm(p, dest, arg0);
-   emit_1ub(p, cc); 
-}
-
-void sse_pmovmskb( struct x86_function *p,
-                   struct x86_reg dest,
-                   struct x86_reg src)
-{
-    emit_3ub(p, 0x66, X86_TWOB, 0xD7);
-    emit_modrm(p, dest, src);
-}
-
-/***********************************************************************
- * SSE2 instructions
- */
-
-/**
- * Perform a reduced swizzle:
- */
-void sse2_pshufd( struct x86_function *p,
-		  struct x86_reg dest,
-		  struct x86_reg arg0,
-		  unsigned char shuf) 
-{
-   emit_3ub(p, 0x66, X86_TWOB, 0x70);
-   emit_modrm(p, dest, arg0);
-   emit_1ub(p, shuf); 
-}
-
-void sse2_cvttps2dq( struct x86_function *p,
-                     struct x86_reg dst,
-                     struct x86_reg src )
-{
-   emit_3ub( p, 0xF3, X86_TWOB, 0x5B );
-   emit_modrm( p, dst, src );
-}
-
-void sse2_cvtps2dq( struct x86_function *p,
-		    struct x86_reg dst,
-		    struct x86_reg src )
-{
-   emit_3ub(p, 0x66, X86_TWOB, 0x5B);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_packssdw( struct x86_function *p,
-		    struct x86_reg dst,
-		    struct x86_reg src )
-{
-   emit_3ub(p, 0x66, X86_TWOB, 0x6B);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_packsswb( struct x86_function *p,
-		    struct x86_reg dst,
-		    struct x86_reg src )
-{
-   emit_3ub(p, 0x66, X86_TWOB, 0x63);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_packuswb( struct x86_function *p,
-		    struct x86_reg dst,
-		    struct x86_reg src )
-{
-   emit_3ub(p, 0x66, X86_TWOB, 0x67);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_rcpps( struct x86_function *p,
-                 struct x86_reg dst,
-                 struct x86_reg src )
-{
-   emit_2ub(p, X86_TWOB, 0x53);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_rcpss( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_3ub(p, 0xF3, X86_TWOB, 0x53);
-   emit_modrm( p, dst, src );
-}
-
-void sse2_movd( struct x86_function *p,
-		struct x86_reg dst,
-		struct x86_reg src )
-{
-   emit_2ub(p, 0x66, X86_TWOB);
-   emit_op_modrm( p, 0x6e, 0x7e, dst, src );
-}
-
-
-
-
-/***********************************************************************
- * x87 instructions
- */
-void x87_fist( struct x86_function *p, struct x86_reg dst )
-{
-   emit_1ub(p, 0xdb);
-   emit_modrm_noreg(p, 2, dst);
-}
-
-void x87_fistp( struct x86_function *p, struct x86_reg dst )
-{
-   emit_1ub(p, 0xdb);
-   emit_modrm_noreg(p, 3, dst);
-}
-
-void x87_fild( struct x86_function *p, struct x86_reg arg )
-{
-   emit_1ub(p, 0xdf);
-   emit_modrm_noreg(p, 0, arg);
-}
-
-void x87_fldz( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xee);
-}
-
-
-void x87_fldcw( struct x86_function *p, struct x86_reg arg )
-{
-   assert(arg.file == file_REG32);
-   assert(arg.mod != mod_REG);
-   emit_1ub(p, 0xd9);
-   emit_modrm_noreg(p, 5, arg);
-}
-
-void x87_fld1( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xe8);
-}
-
-void x87_fldl2e( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xea);
-}
-
-void x87_fldln2( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xed);
-}
-
-void x87_fwait( struct x86_function *p )
-{
-   emit_1ub(p, 0x9b);
-}
-
-void x87_fnclex( struct x86_function *p )
-{
-   emit_2ub(p, 0xdb, 0xe2);
-}
-
-void x87_fclex( struct x86_function *p )
-{
-   x87_fwait(p);
-   x87_fnclex(p);
-}
-
-
-static void x87_arith_op( struct x86_function *p, struct x86_reg dst, struct x86_reg arg,
-			  unsigned char dst0ub0,
-			  unsigned char dst0ub1,
-			  unsigned char arg0ub0,
-			  unsigned char arg0ub1,
-			  unsigned char argmem_noreg)
-{
-   assert(dst.file == file_x87);
-
-   if (arg.file == file_x87) {
-      if (dst.idx == 0) 
-	 emit_2ub(p, dst0ub0, dst0ub1+arg.idx);
-      else if (arg.idx == 0) 
-	 emit_2ub(p, arg0ub0, arg0ub1+arg.idx);
-      else
-	 assert(0);
-   }
-   else if (dst.idx == 0) {
-      assert(arg.file == file_REG32);
-      emit_1ub(p, 0xd8);
-      emit_modrm_noreg(p, argmem_noreg, arg);
-   }
-   else
-      assert(0);
-}
-
-void x87_fmul( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xc8,
-		0xdc, 0xc8,
-		4);
-}
-
-void x87_fsub( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xe0,
-		0xdc, 0xe8,
-		4);
-}
-
-void x87_fsubr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xe8,
-		0xdc, 0xe0,
-		5);
-}
-
-void x87_fadd( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xc0,
-		0xdc, 0xc0,
-		0);
-}
-
-void x87_fdiv( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xf0,
-		0xdc, 0xf8,
-		6);
-}
-
-void x87_fdivr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg )
-{
-   x87_arith_op(p, dst, arg, 
-		0xd8, 0xf8,
-		0xdc, 0xf0,
-		7);
-}
-
-void x87_fmulp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xc8+dst.idx);
-}
-
-void x87_fsubp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xe8+dst.idx);
-}
-
-void x87_fsubrp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xe0+dst.idx);
-}
-
-void x87_faddp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xc0+dst.idx);
-}
-
-void x87_fdivp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xf8+dst.idx);
-}
-
-void x87_fdivrp( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_x87);
-   assert(dst.idx >= 1);
-   emit_2ub(p, 0xde, 0xf0+dst.idx);
-}
-
-void x87_fucom( struct x86_function *p, struct x86_reg arg )
-{
-   assert(arg.file == file_x87);
-   emit_2ub(p, 0xdd, 0xe0+arg.idx);
-}
-
-void x87_fucomp( struct x86_function *p, struct x86_reg arg )
-{
-   assert(arg.file == file_x87);
-   emit_2ub(p, 0xdd, 0xe8+arg.idx);
-}
-
-void x87_fucompp( struct x86_function *p )
-{
-   emit_2ub(p, 0xda, 0xe9);
-}
-
-void x87_fxch( struct x86_function *p, struct x86_reg arg )
-{
-   assert(arg.file == file_x87);
-   emit_2ub(p, 0xd9, 0xc8+arg.idx);
-}
-
-void x87_fabs( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xe1);
-}
-
-void x87_fchs( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xe0);
-}
-
-void x87_fcos( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xff);
-}
-
-
-void x87_fprndint( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xfc);
-}
-
-void x87_fscale( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xfd);
-}
-
-void x87_fsin( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xfe);
-}
-
-void x87_fsincos( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xfb);
-}
-
-void x87_fsqrt( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xfa);
-}
-
-void x87_fxtract( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xf4);
-}
-
-/* st0 = (2^st0)-1
- *
- * Restrictions: -1.0 <= st0 <= 1.0
- */
-void x87_f2xm1( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xf0);
-}
-
-/* st1 = st1 * log2(st0);
- * pop_stack;
- */
-void x87_fyl2x( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xf1);
-}
-
-/* st1 = st1 * log2(st0 + 1.0);
- * pop_stack;
- *
- * A fast operation, with restrictions: -.29 < st0 < .29 
- */
-void x87_fyl2xp1( struct x86_function *p )
-{
-   emit_2ub(p, 0xd9, 0xf9);
-}
-
-
-void x87_fld( struct x86_function *p, struct x86_reg arg )
-{
-   if (arg.file == file_x87) 
-      emit_2ub(p, 0xd9, 0xc0 + arg.idx);
-   else {
-      emit_1ub(p, 0xd9);
-      emit_modrm_noreg(p, 0, arg);
-   }
-}
-
-void x87_fst( struct x86_function *p, struct x86_reg dst )
-{
-   if (dst.file == file_x87) 
-      emit_2ub(p, 0xdd, 0xd0 + dst.idx);
-   else {
-      emit_1ub(p, 0xd9);
-      emit_modrm_noreg(p, 2, dst);
-   }
-}
-
-void x87_fstp( struct x86_function *p, struct x86_reg dst )
-{
-   if (dst.file == file_x87) 
-      emit_2ub(p, 0xdd, 0xd8 + dst.idx);
-   else {
-      emit_1ub(p, 0xd9);
-      emit_modrm_noreg(p, 3, dst);
-   }
-}
-
-void x87_fcom( struct x86_function *p, struct x86_reg dst )
-{
-   if (dst.file == file_x87) 
-      emit_2ub(p, 0xd8, 0xd0 + dst.idx);
-   else {
-      emit_1ub(p, 0xd8);
-      emit_modrm_noreg(p, 2, dst);
-   }
-}
-
-void x87_fcomp( struct x86_function *p, struct x86_reg dst )
-{
-   if (dst.file == file_x87) 
-      emit_2ub(p, 0xd8, 0xd8 + dst.idx);
-   else {
-      emit_1ub(p, 0xd8);
-      emit_modrm_noreg(p, 3, dst);
-   }
-}
-
-
-void x87_fnstsw( struct x86_function *p, struct x86_reg dst )
-{
-   assert(dst.file == file_REG32);
-
-   if (dst.idx == reg_AX &&
-       dst.mod == mod_REG) 
-      emit_2ub(p, 0xdf, 0xe0);
-   else {
-      emit_1ub(p, 0xdd);
-      emit_modrm_noreg(p, 7, dst);
-   }
-}
-
-
-
-
-/***********************************************************************
- * MMX instructions
- */
-
-void mmx_emms( struct x86_function *p )
-{
-   assert(p->need_emms);
-   emit_2ub(p, 0x0f, 0x77);
-   p->need_emms = 0;
-}
-
-void mmx_packssdw( struct x86_function *p,
-		   struct x86_reg dst,
-		   struct x86_reg src )
-{
-   assert(dst.file == file_MMX && 
-	  (src.file == file_MMX || src.mod != mod_REG));
-
-   p->need_emms = 1;
-
-   emit_2ub(p, X86_TWOB, 0x6b);
-   emit_modrm( p, dst, src );
-}
-
-void mmx_packuswb( struct x86_function *p,
-		   struct x86_reg dst,
-		   struct x86_reg src )
-{
-   assert(dst.file == file_MMX && 
-	  (src.file == file_MMX || src.mod != mod_REG));
-
-   p->need_emms = 1;
-
-   emit_2ub(p, X86_TWOB, 0x67);
-   emit_modrm( p, dst, src );
-}
-
-void mmx_movd( struct x86_function *p,
-	       struct x86_reg dst,
-	       struct x86_reg src )
-{
-   p->need_emms = 1;
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x6e, 0x7e, dst, src );
-}
-
-void mmx_movq( struct x86_function *p,
-	       struct x86_reg dst,
-	       struct x86_reg src )
-{
-   p->need_emms = 1;
-   emit_1ub(p, X86_TWOB);
-   emit_op_modrm( p, 0x6f, 0x7f, dst, src );
-}
-
-
-/***********************************************************************
- * Helper functions
- */
-
-
-/* Retreive a reference to one of the function arguments, taking into
- * account any push/pop activity:
- */
-struct x86_reg x86_fn_arg( struct x86_function *p,
-			   unsigned arg )
-{
-   return x86_make_disp(x86_make_reg(file_REG32, reg_SP), 
-			p->stack_offset + arg * 4);	/* ??? */
-}
-
-
-void x86_init_func( struct x86_function *p )
-{
-   p->size = 0;
-   p->store = NULL;
-   p->csr = p->store;
-}
-
-int x86_init_func_size( struct x86_function *p, unsigned code_size )
-{
-   p->size = code_size;
-   p->store = _mesa_exec_malloc(code_size);
-   p->csr = p->store;
-   return p->store != NULL;
-}
-
-void x86_release_func( struct x86_function *p )
-{
-   _mesa_exec_free(p->store);
-   p->store = NULL;
-   p->csr = NULL;
-   p->size = 0;
-}
-
-
-void (*x86_get_func( struct x86_function *p ))(void)
-{
-   if (DISASSEM && p->store)
-      printf("disassemble %p %p\n", p->store, p->csr);
-   return (void (*)(void)) (unsigned long) p->store;
-}
-
-#else
-
-void x86sse_dummy( void )
-{
-}
-
-#endif
-
-#else  /* USE_X86_ASM */
-
-int x86sse_c_dummy_var; /* silence warning */
-
-#endif /* USE_X86_ASM */
diff --git a/mesalib/src/mesa/x86/rtasm/x86sse.h b/mesalib/src/mesa/x86/rtasm/x86sse.h
deleted file mode 100644
index f6282f5bd..000000000
--- a/mesalib/src/mesa/x86/rtasm/x86sse.h
+++ /dev/null
@@ -1,256 +0,0 @@
-
-#ifndef _X86SSE_H_
-#define _X86SSE_H_
-
-#if defined(__i386__) || defined(__386__)
-
-/* It is up to the caller to ensure that instructions issued are
- * suitable for the host cpu.  There are no checks made in this module
- * for mmx/sse/sse2 support on the cpu.
- */
-struct x86_reg {
-   unsigned file:3;
-   unsigned idx:3;
-   unsigned mod:2;		/* mod_REG if this is just a register */
-   int      disp:24;		/* only +/- 23bits of offset - should be enough... */
-};
-
-struct x86_function {
-   unsigned size;
-   unsigned char *store;
-   unsigned char *csr;
-   unsigned stack_offset;
-   int need_emms;
-   const char *fn;
-};
-
-enum x86_reg_file {
-   file_REG32,
-   file_MMX,
-   file_XMM,
-   file_x87
-};
-
-/* Values for mod field of modr/m byte
- */
-enum x86_reg_mod {
-   mod_INDIRECT,
-   mod_DISP8,
-   mod_DISP32,
-   mod_REG
-};
-
-enum x86_reg_name {
-   reg_AX,
-   reg_CX,
-   reg_DX,
-   reg_BX,
-   reg_SP,
-   reg_BP,
-   reg_SI,
-   reg_DI
-};
-
-
-enum x86_cc {
-   cc_O,			/* overflow */
-   cc_NO,			/* not overflow */
-   cc_NAE,			/* not above or equal / carry */
-   cc_AE,			/* above or equal / not carry */
-   cc_E,			/* equal / zero */
-   cc_NE			/* not equal / not zero */
-};
-
-enum sse_cc {
-   cc_Equal,
-   cc_LessThan,
-   cc_LessThanEqual,
-   cc_Unordered,
-   cc_NotEqual,
-   cc_NotLessThan,
-   cc_NotLessThanEqual,
-   cc_Ordered
-};
-
-#define cc_Z  cc_E
-#define cc_NZ cc_NE
-
-/* Begin/end/retreive function creation:
- */
-
-
-void x86_init_func( struct x86_function *p );
-int x86_init_func_size( struct x86_function *p, unsigned code_size );
-void x86_release_func( struct x86_function *p );
-void (*x86_get_func( struct x86_function *p ))( void );
-
-
-
-/* Create and manipulate registers and regmem values:
- */
-struct x86_reg x86_make_reg( enum x86_reg_file file,
-			     enum x86_reg_name idx );
-
-struct x86_reg x86_make_disp( struct x86_reg reg,
-			      int disp );
-
-struct x86_reg x86_deref( struct x86_reg reg );
-
-struct x86_reg x86_get_base_reg( struct x86_reg reg );
-
-
-/* Labels, jumps and fixup:
- */
-unsigned char *x86_get_label( struct x86_function *p );
-
-void x86_jcc( struct x86_function *p,
-	      enum x86_cc cc,
-	      unsigned char *label );
-
-unsigned char *x86_jcc_forward( struct x86_function *p,
-			  enum x86_cc cc );
-
-unsigned char *x86_jmp_forward( struct x86_function *p);
-
-unsigned char *x86_call_forward( struct x86_function *p);
-
-void x86_fixup_fwd_jump( struct x86_function *p,
-			 unsigned char *fixup );
-
-void x86_jmp( struct x86_function *p, unsigned char *label );
-
-/* void x86_call( struct x86_function *p, void (*label)() ); */
-void x86_call( struct x86_function *p, struct x86_reg reg);
-
-/* michal:
- * Temporary. As I need immediate operands, and dont want to mess with the codegen,
- * I load the immediate into general purpose register and use it.
- */
-void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm );
-
-
-/* Macro for sse_shufps() and sse2_pshufd():
- */
-#define SHUF(_x,_y,_z,_w)       (((_x)<<0) | ((_y)<<2) | ((_z)<<4) | ((_w)<<6))
-#define SHUF_NOOP               RSW(0,1,2,3)
-#define GET_SHUF(swz, idx)      (((swz) >> ((idx)*2)) & 0x3)
-
-void mmx_emms( struct x86_function *p );
-void mmx_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void mmx_movq( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void mmx_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void mmx_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-
-void sse2_cvtps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_cvttps2dq( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_movd( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_packssdw( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_packsswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_packuswb( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_pshufd( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0,
-                  unsigned char shuf );
-void sse2_rcpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse2_rcpss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-
-void sse_addps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_addss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_cvtps2pi( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_divss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_andnps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_andps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_cmpps( struct x86_function *p, struct x86_reg dst, struct x86_reg src,
-                unsigned char cc );
-void sse_maxps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_maxss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_minps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movaps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movhlps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movlhps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movlps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_movups( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_mulps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_mulss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_orps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_xorps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_subps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_rsqrtps( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_rsqrtss( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void sse_shufps( struct x86_function *p, struct x86_reg dest, struct x86_reg arg0,
-                 unsigned char shuf );
-void sse_pmovmskb( struct x86_function *p, struct x86_reg dest, struct x86_reg src );
-
-void x86_add( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_and( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_cmp( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_dec( struct x86_function *p, struct x86_reg reg );
-void x86_inc( struct x86_function *p, struct x86_reg reg );
-void x86_lea( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_mov( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_mul( struct x86_function *p, struct x86_reg src );
-void x86_or( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_pop( struct x86_function *p, struct x86_reg reg );
-void x86_push( struct x86_function *p, struct x86_reg reg );
-void x86_ret( struct x86_function *p );
-void x86_sub( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_test( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_xor( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
-void x86_sahf( struct x86_function *p );
-
-void x87_f2xm1( struct x86_function *p );
-void x87_fabs( struct x86_function *p );
-void x87_fadd( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_faddp( struct x86_function *p, struct x86_reg dst );
-void x87_fchs( struct x86_function *p );
-void x87_fclex( struct x86_function *p );
-void x87_fcom( struct x86_function *p, struct x86_reg dst );
-void x87_fcomp( struct x86_function *p, struct x86_reg dst );
-void x87_fcos( struct x86_function *p );
-void x87_fdiv( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_fdivp( struct x86_function *p, struct x86_reg dst );
-void x87_fdivr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_fdivrp( struct x86_function *p, struct x86_reg dst );
-void x87_fild( struct x86_function *p, struct x86_reg arg );
-void x87_fist( struct x86_function *p, struct x86_reg dst );
-void x87_fistp( struct x86_function *p, struct x86_reg dst );
-void x87_fld( struct x86_function *p, struct x86_reg arg );
-void x87_fld1( struct x86_function *p );
-void x87_fldcw( struct x86_function *p, struct x86_reg arg );
-void x87_fldl2e( struct x86_function *p );
-void x87_fldln2( struct x86_function *p );
-void x87_fldz( struct x86_function *p );
-void x87_fmul( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_fmulp( struct x86_function *p, struct x86_reg dst );
-void x87_fnclex( struct x86_function *p );
-void x87_fprndint( struct x86_function *p );
-void x87_fscale( struct x86_function *p );
-void x87_fsin( struct x86_function *p );
-void x87_fsincos( struct x86_function *p );
-void x87_fsqrt( struct x86_function *p );
-void x87_fst( struct x86_function *p, struct x86_reg dst );
-void x87_fstp( struct x86_function *p, struct x86_reg dst );
-void x87_fsub( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_fsubp( struct x86_function *p, struct x86_reg dst );
-void x87_fsubr( struct x86_function *p, struct x86_reg dst, struct x86_reg arg );
-void x87_fsubrp( struct x86_function *p, struct x86_reg dst );
-void x87_fxch( struct x86_function *p, struct x86_reg dst );
-void x87_fxtract( struct x86_function *p );
-void x87_fyl2x( struct x86_function *p );
-void x87_fyl2xp1( struct x86_function *p );
-void x87_fwait( struct x86_function *p );
-void x87_fnstsw( struct x86_function *p, struct x86_reg dst );
-void x87_fucompp( struct x86_function *p );
-void x87_fucomp( struct x86_function *p, struct x86_reg arg );
-void x87_fucom( struct x86_function *p, struct x86_reg arg );
-
-
-
-/* Retreive a reference to one of the function arguments, taking into
- * account any push/pop activity.  Note - doesn't track explict
- * manipulation of ESP by other instructions.
- */
-struct x86_reg x86_fn_arg( struct x86_function *p, unsigned arg );
-
-#endif
-#endif