Merge remote-tracking branch 'origin/released'

Conflicts:
	mesalib/src/mesa/main/imports.h

73 files changed, 5085 insertions, 810 deletions

diff --git a/mesalib/src/gallium/Automake.inc b/mesalib/src/gallium/Automake.inc
index 1f5d532e4..4519c161c 100644
--- a/mesalib/src/gallium/Automake.inc
+++ b/mesalib/src/gallium/Automake.inc
@@ -51,22 +51,20 @@ GALLIUM_VIDEO_CFLAGS = \
 	$(VISIBILITY_CFLAGS)
 
 
-DRI_VERSION_SCRIPT ?= $(top_srcdir)/src/gallium/state_trackers/dri/dri.link
-
 GALLIUM_DRI_LINKER_FLAGS = \
 	-shared \
 	-shrext .so \
 	-module \
 	-avoid-version \
-	$(GC_SECTIONS) \
-	-Wl,--version-script=$(DRI_VERSION_SCRIPT)
+	-Wl,--version-script=$(top_srcdir)/src/gallium/targets/dri.sym \
+	$(GC_SECTIONS)
 
 GALLIUM_VDPAU_LINKER_FLAGS = \
 	-shared \
 	-module \
 	-no-undefined \
 	-version-number $(VDPAU_MAJOR):$(VDPAU_MINOR) \
-	-export-symbols-regex $(VDPAU_EXPORTS) \
+	-Wl,--version-script=$(top_srcdir)/src/gallium/targets/vdpau.sym \
 	$(GC_SECTIONS) \
 	$(LD_NO_UNDEFINED)
 
@@ -75,7 +73,7 @@ GALLIUM_XVMC_LINKER_FLAGS = \
 	-module \
 	-no-undefined \
 	-version-number $(XVMC_MAJOR):$(XVMC_MINOR) \
-	-export-symbols-regex '^XvMC' \
+	-Wl,--version-script=$(top_srcdir)/src/gallium/targets/xvmc.sym \
 	$(GC_SECTIONS) \
 	$(LD_NO_UNDEFINED)
 
@@ -83,7 +81,8 @@ GALLIUM_OMX_LINKER_FLAGS = \
 	-shared \
 	-module \
 	-no-undefined \
-	-export-symbols-regex $(EXPORTS) \
+	-avoid-version \
+	-Wl,--version-script=$(top_srcdir)/src/gallium/targets/omx.sym \
 	$(GC_SECTIONS) \
 	$(LD_NO_UNDEFINED)
 
diff --git a/mesalib/src/gallium/auxiliary/util/u_debug_flush.c b/mesalib/src/gallium/auxiliary/util/u_debug_flush.c
index 9cf70db58..fdb248c23 100644
--- a/mesalib/src/gallium/auxiliary/util/u_debug_flush.c
+++ b/mesalib/src/gallium/auxiliary/util/u_debug_flush.c
@@ -46,6 +46,7 @@
 #include "util/u_hash_table.h"
 #include "util/u_double_list.h"
 #include "util/u_inlines.h"
+#include "util/u_string.h"
 #include "os/os_thread.h"
 #include <stdio.h>
 
@@ -320,8 +321,8 @@ debug_flush_might_flush_cb(void *key, void *value, void *data)
    const char *reason = (const char *) data;
    char message[80];
 
-   snprintf(message, sizeof(message),
-            "%s referenced mapped buffer detected.", reason);
+   util_snprintf(message, sizeof(message),
+                 "%s referenced mapped buffer detected.", reason);
 
    pipe_mutex_lock(fbuf->mutex);
    if (fbuf->mapped_sync) {
diff --git a/mesalib/src/gallium/auxiliary/util/u_math.h b/mesalib/src/gallium/auxiliary/util/u_math.h
index a60e1830a..2ade64af4 100644
--- a/mesalib/src/gallium/auxiliary/util/u_math.h
+++ b/mesalib/src/gallium/auxiliary/util/u_math.h
@@ -138,8 +138,13 @@ roundf(float x)
 }
 #endif
 
+#ifndef INFINITY
 #define INFINITY (DBL_MAX + DBL_MAX)
+#endif
+
+#ifndef NAN
 #define NAN (INFINITY - INFINITY)
+#endif
 
 #endif /* _MSC_VER */
 
diff --git a/mesalib/src/glsl/Makefile.am b/mesalib/src/glsl/Makefile.am
index 534eaa385..fd0e837d1 100644
--- a/mesalib/src/glsl/Makefile.am
+++ b/mesalib/src/glsl/Makefile.am
@@ -61,7 +61,9 @@ tests_general_ir_test_SOURCES =		\
 	$(GLSL_SRCDIR)/standalone_scaffolding.cpp \
 	tests/builtin_variable_test.cpp			\
 	tests/invalidate_locations_test.cpp		\
-	tests/general_ir_test.cpp
+	tests/general_ir_test.cpp			\
+	tests/varyings_test.cpp				\
+	tests/common.c
 tests_general_ir_test_CFLAGS =				\
 	$(PTHREAD_CFLAGS)
 tests_general_ir_test_LDADD =				\
@@ -76,7 +78,8 @@ tests_uniform_initializer_test_SOURCES =		\
 	$(top_srcdir)/src/mesa/program/symbol_table.c	\
 	tests/copy_constant_to_storage_tests.cpp	\
 	tests/set_uniform_initializer_tests.cpp		\
-	tests/uniform_initializer_utils.cpp
+	tests/uniform_initializer_utils.cpp		\
+	tests/common.c
 tests_uniform_initializer_test_CFLAGS =			\
 	$(PTHREAD_CFLAGS)
 tests_uniform_initializer_test_LDADD =			\
@@ -95,7 +98,8 @@ tests_ralloc_test_LDADD =				\
 tests_sampler_types_test_SOURCES =			\
 	$(top_srcdir)/src/mesa/program/prog_hash_table.c\
 	$(top_srcdir)/src/mesa/program/symbol_table.c	\
-	tests/sampler_types_test.cpp
+	tests/sampler_types_test.cpp			\
+	tests/common.c
 tests_sampler_types_test_CFLAGS =			\
 	$(PTHREAD_CFLAGS)
 tests_sampler_types_test_LDADD =			\
@@ -138,6 +142,7 @@ glsl_test_SOURCES = \
 	$(top_srcdir)/src/mesa/program/prog_hash_table.c \
 	$(top_srcdir)/src/mesa/program/symbol_table.c \
 	$(GLSL_SRCDIR)/standalone_scaffolding.cpp \
+	tests/common.c \
 	test.cpp \
 	test_optpass.cpp
 
diff --git a/mesalib/src/glsl/ast_to_hir.cpp b/mesalib/src/glsl/ast_to_hir.cpp
index 0411befa9..7516c33e1 100644
--- a/mesalib/src/glsl/ast_to_hir.cpp
+++ b/mesalib/src/glsl/ast_to_hir.cpp
@@ -123,6 +123,11 @@ _mesa_ast_to_hir(exec_list *instructions, struct _mesa_glsl_parse_state *state)
       instructions->push_head(var);
    }
 
+   /* Figure out if gl_FragCoord is actually used in fragment shader */
+   ir_variable *const var = state->symbols->get_variable("gl_FragCoord");
+   if (var != NULL)
+      state->fs_uses_gl_fragcoord = var->data.used;
+
    /* From section 7.1 (Built-In Language Variables) of the GLSL 4.10 spec:
     *
     *     If multiple shaders using members of a built-in block belonging to
@@ -2341,6 +2346,34 @@ apply_image_qualifier_to_variable(const struct ast_type_qualifier *qual,
    }
 }
 
+static inline const char*
+get_layout_qualifier_string(bool origin_upper_left, bool pixel_center_integer)
+{
+   if (origin_upper_left && pixel_center_integer)
+      return "origin_upper_left, pixel_center_integer";
+   else if (origin_upper_left)
+      return "origin_upper_left";
+   else if (pixel_center_integer)
+      return "pixel_center_integer";
+   else
+      return " ";
+}
+
+static inline bool
+is_conflicting_fragcoord_redeclaration(struct _mesa_glsl_parse_state *state,
+                                       const struct ast_type_qualifier *qual)
+{
+   /* If gl_FragCoord was previously declared, and the qualifiers were
+    * different in any way, return true.
+    */
+   if (state->fs_redeclares_gl_fragcoord) {
+      return (state->fs_pixel_center_integer != qual->flags.q.pixel_center_integer
+         || state->fs_origin_upper_left != qual->flags.q.origin_upper_left);
+   }
+
+   return false;
+}
+
 static void
 apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
                                  ir_variable *var,
@@ -2505,6 +2538,53 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
 		       qual_string);
    }
 
+   if (var->name != NULL && strcmp(var->name, "gl_FragCoord") == 0) {
+
+      /* Section 4.3.8.1, page 39 of GLSL 1.50 spec says:
+       *
+       *    "Within any shader, the first redeclarations of gl_FragCoord
+       *     must appear before any use of gl_FragCoord."
+       *
+       * Generate a compiler error if above condition is not met by the
+       * fragment shader.
+       */
+      ir_variable *earlier = state->symbols->get_variable("gl_FragCoord");
+      if (earlier != NULL &&
+          earlier->data.used &&
+          !state->fs_redeclares_gl_fragcoord) {
+         _mesa_glsl_error(loc, state,
+                          "gl_FragCoord used before its first redeclaration "
+                          "in fragment shader");
+      }
+
+      /* Make sure all gl_FragCoord redeclarations specify the same layout
+       * qualifiers.
+       */
+      if (is_conflicting_fragcoord_redeclaration(state, qual)) {
+         const char *const qual_string =
+            get_layout_qualifier_string(qual->flags.q.origin_upper_left,
+                                        qual->flags.q.pixel_center_integer);
+
+         const char *const state_string =
+            get_layout_qualifier_string(state->fs_origin_upper_left,
+                                        state->fs_pixel_center_integer);
+
+         _mesa_glsl_error(loc, state,
+                          "gl_FragCoord redeclared with different layout "
+                          "qualifiers (%s) and (%s) ",
+                          state_string,
+                          qual_string);
+      }
+      state->fs_origin_upper_left = qual->flags.q.origin_upper_left;
+      state->fs_pixel_center_integer = qual->flags.q.pixel_center_integer;
+      state->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers =
+         !qual->flags.q.origin_upper_left && !qual->flags.q.pixel_center_integer;
+      state->fs_redeclares_gl_fragcoord =
+         state->fs_origin_upper_left ||
+         state->fs_pixel_center_integer ||
+         state->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers;
+   }
+
    if (qual->flags.q.explicit_location) {
       validate_explicit_location(qual, var, state, loc);
    } else if (qual->flags.q.explicit_index) {
diff --git a/mesalib/src/glsl/builtin_types.cpp b/mesalib/src/glsl/builtin_types.cpp
index dd42ecb05..0a0fa8cd3 100644
--- a/mesalib/src/glsl/builtin_types.cpp
+++ b/mesalib/src/glsl/builtin_types.cpp
@@ -241,7 +241,7 @@ const static struct builtin_type_versions {
    T(atomic_uint,                     420, 999)
 };
 
-const glsl_type *const deprecated_types[] = {
+static const glsl_type *const deprecated_types[] = {
    glsl_type::struct_gl_PointParameters_type,
    glsl_type::struct_gl_MaterialParameters_type,
    glsl_type::struct_gl_LightSourceParameters_type,
diff --git a/mesalib/src/glsl/builtin_variables.cpp b/mesalib/src/glsl/builtin_variables.cpp
index 4176ae6e6..9b35850ee 100644
--- a/mesalib/src/glsl/builtin_variables.cpp
+++ b/mesalib/src/glsl/builtin_variables.cpp
@@ -30,21 +30,21 @@
 #include "program/prog_statevars.h"
 #include "program/prog_instruction.h"
 
-static struct gl_builtin_uniform_element gl_NumSamples_elements[] = {
+static const struct gl_builtin_uniform_element gl_NumSamples_elements[] = {
    {NULL, {STATE_NUM_SAMPLES, 0, 0}, SWIZZLE_XXXX}
 };
 
-static struct gl_builtin_uniform_element gl_DepthRange_elements[] = {
+static const struct gl_builtin_uniform_element gl_DepthRange_elements[] = {
    {"near", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_XXXX},
    {"far", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_YYYY},
    {"diff", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_ZZZZ},
 };
 
-static struct gl_builtin_uniform_element gl_ClipPlane_elements[] = {
+static const struct gl_builtin_uniform_element gl_ClipPlane_elements[] = {
    {NULL, {STATE_CLIPPLANE, 0, 0}, SWIZZLE_XYZW}
 };
 
-static struct gl_builtin_uniform_element gl_Point_elements[] = {
+static const struct gl_builtin_uniform_element gl_Point_elements[] = {
    {"size", {STATE_POINT_SIZE}, SWIZZLE_XXXX},
    {"sizeMin", {STATE_POINT_SIZE}, SWIZZLE_YYYY},
    {"sizeMax", {STATE_POINT_SIZE}, SWIZZLE_ZZZZ},
@@ -54,7 +54,7 @@ static struct gl_builtin_uniform_element gl_Point_elements[] = {
    {"distanceQuadraticAttenuation", {STATE_POINT_ATTENUATION}, SWIZZLE_ZZZZ},
 };
 
-static struct gl_builtin_uniform_element gl_FrontMaterial_elements[] = {
+static const struct gl_builtin_uniform_element gl_FrontMaterial_elements[] = {
    {"emission", {STATE_MATERIAL, 0, STATE_EMISSION}, SWIZZLE_XYZW},
    {"ambient", {STATE_MATERIAL, 0, STATE_AMBIENT}, SWIZZLE_XYZW},
    {"diffuse", {STATE_MATERIAL, 0, STATE_DIFFUSE}, SWIZZLE_XYZW},
@@ -62,7 +62,7 @@ static struct gl_builtin_uniform_element gl_FrontMaterial_elements[] = {
    {"shininess", {STATE_MATERIAL, 0, STATE_SHININESS}, SWIZZLE_XXXX},
 };
 
-static struct gl_builtin_uniform_element gl_BackMaterial_elements[] = {
+static const struct gl_builtin_uniform_element gl_BackMaterial_elements[] = {
    {"emission", {STATE_MATERIAL, 1, STATE_EMISSION}, SWIZZLE_XYZW},
    {"ambient", {STATE_MATERIAL, 1, STATE_AMBIENT}, SWIZZLE_XYZW},
    {"diffuse", {STATE_MATERIAL, 1, STATE_DIFFUSE}, SWIZZLE_XYZW},
@@ -70,7 +70,7 @@ static struct gl_builtin_uniform_element gl_BackMaterial_elements[] = {
    {"shininess", {STATE_MATERIAL, 1, STATE_SHININESS}, SWIZZLE_XXXX},
 };
 
-static struct gl_builtin_uniform_element gl_LightSource_elements[] = {
+static const struct gl_builtin_uniform_element gl_LightSource_elements[] = {
    {"ambient", {STATE_LIGHT, 0, STATE_AMBIENT}, SWIZZLE_XYZW},
    {"diffuse", {STATE_LIGHT, 0, STATE_DIFFUSE}, SWIZZLE_XYZW},
    {"specular", {STATE_LIGHT, 0, STATE_SPECULAR}, SWIZZLE_XYZW},
@@ -89,67 +89,67 @@ static struct gl_builtin_uniform_element gl_LightSource_elements[] = {
    {"quadraticAttenuation", {STATE_LIGHT, 0, STATE_ATTENUATION}, SWIZZLE_ZZZZ},
 };
 
-static struct gl_builtin_uniform_element gl_LightModel_elements[] = {
+static const struct gl_builtin_uniform_element gl_LightModel_elements[] = {
    {"ambient", {STATE_LIGHTMODEL_AMBIENT, 0}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_FrontLightModelProduct_elements[] = {
+static const struct gl_builtin_uniform_element gl_FrontLightModelProduct_elements[] = {
    {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 0}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_BackLightModelProduct_elements[] = {
+static const struct gl_builtin_uniform_element gl_BackLightModelProduct_elements[] = {
    {"sceneColor", {STATE_LIGHTMODEL_SCENECOLOR, 1}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_FrontLightProduct_elements[] = {
+static const struct gl_builtin_uniform_element gl_FrontLightProduct_elements[] = {
    {"ambient", {STATE_LIGHTPROD, 0, 0, STATE_AMBIENT}, SWIZZLE_XYZW},
    {"diffuse", {STATE_LIGHTPROD, 0, 0, STATE_DIFFUSE}, SWIZZLE_XYZW},
    {"specular", {STATE_LIGHTPROD, 0, 0, STATE_SPECULAR}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_BackLightProduct_elements[] = {
+static const struct gl_builtin_uniform_element gl_BackLightProduct_elements[] = {
    {"ambient", {STATE_LIGHTPROD, 0, 1, STATE_AMBIENT}, SWIZZLE_XYZW},
    {"diffuse", {STATE_LIGHTPROD, 0, 1, STATE_DIFFUSE}, SWIZZLE_XYZW},
    {"specular", {STATE_LIGHTPROD, 0, 1, STATE_SPECULAR}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_TextureEnvColor_elements[] = {
+static const struct gl_builtin_uniform_element gl_TextureEnvColor_elements[] = {
    {NULL, {STATE_TEXENV_COLOR, 0}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_EyePlaneS_elements[] = {
+static const struct gl_builtin_uniform_element gl_EyePlaneS_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_S}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_EyePlaneT_elements[] = {
+static const struct gl_builtin_uniform_element gl_EyePlaneT_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_T}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_EyePlaneR_elements[] = {
+static const struct gl_builtin_uniform_element gl_EyePlaneR_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_R}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_EyePlaneQ_elements[] = {
+static const struct gl_builtin_uniform_element gl_EyePlaneQ_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_EYE_Q}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_ObjectPlaneS_elements[] = {
+static const struct gl_builtin_uniform_element gl_ObjectPlaneS_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_S}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_ObjectPlaneT_elements[] = {
+static const struct gl_builtin_uniform_element gl_ObjectPlaneT_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_T}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_ObjectPlaneR_elements[] = {
+static const struct gl_builtin_uniform_element gl_ObjectPlaneR_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_R}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_ObjectPlaneQ_elements[] = {
+static const struct gl_builtin_uniform_element gl_ObjectPlaneQ_elements[] = {
    {NULL, {STATE_TEXGEN, 0, STATE_TEXGEN_OBJECT_Q}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_Fog_elements[] = {
+static const struct gl_builtin_uniform_element gl_Fog_elements[] = {
    {"color", {STATE_FOG_COLOR}, SWIZZLE_XYZW},
    {"density", {STATE_FOG_PARAMS}, SWIZZLE_XXXX},
    {"start", {STATE_FOG_PARAMS}, SWIZZLE_YYYY},
@@ -157,32 +157,32 @@ static struct gl_builtin_uniform_element gl_Fog_elements[] = {
    {"scale", {STATE_FOG_PARAMS}, SWIZZLE_WWWW},
 };
 
-static struct gl_builtin_uniform_element gl_NormalScale_elements[] = {
+static const struct gl_builtin_uniform_element gl_NormalScale_elements[] = {
    {NULL, {STATE_NORMAL_SCALE}, SWIZZLE_XXXX},
 };
 
-static struct gl_builtin_uniform_element gl_BumpRotMatrix0MESA_elements[] = {
+static const struct gl_builtin_uniform_element gl_BumpRotMatrix0MESA_elements[] = {
    {NULL, {STATE_INTERNAL, STATE_ROT_MATRIX_0}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_BumpRotMatrix1MESA_elements[] = {
+static const struct gl_builtin_uniform_element gl_BumpRotMatrix1MESA_elements[] = {
    {NULL, {STATE_INTERNAL, STATE_ROT_MATRIX_1}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_FogParamsOptimizedMESA_elements[] = {
+static const struct gl_builtin_uniform_element gl_FogParamsOptimizedMESA_elements[] = {
    {NULL, {STATE_INTERNAL, STATE_FOG_PARAMS_OPTIMIZED}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_CurrentAttribVertMESA_elements[] = {
+static const struct gl_builtin_uniform_element gl_CurrentAttribVertMESA_elements[] = {
    {NULL, {STATE_INTERNAL, STATE_CURRENT_ATTRIB, 0}, SWIZZLE_XYZW},
 };
 
-static struct gl_builtin_uniform_element gl_CurrentAttribFragMESA_elements[] = {
+static const struct gl_builtin_uniform_element gl_CurrentAttribFragMESA_elements[] = {
    {NULL, {STATE_INTERNAL, STATE_CURRENT_ATTRIB_MAYBE_VP_CLAMPED, 0}, SWIZZLE_XYZW},
 };
 
 #define MATRIX(name, statevar, modifier)				\
-   static struct gl_builtin_uniform_element name ## _elements[] = {	\
+   static const struct gl_builtin_uniform_element name ## _elements[] = { \
       { NULL, { statevar, 0, 0, 0, modifier}, SWIZZLE_XYZW },		\
       { NULL, { statevar, 0, 1, 1, modifier}, SWIZZLE_XYZW },		\
       { NULL, { statevar, 0, 2, 2, modifier}, SWIZZLE_XYZW },		\
@@ -225,7 +225,7 @@ MATRIX(gl_TextureMatrixTranspose,
 MATRIX(gl_TextureMatrixInverseTranspose,
        STATE_TEXTURE_MATRIX, STATE_MATRIX_INVERSE);
 
-static struct gl_builtin_uniform_element gl_NormalMatrix_elements[] = {
+static const struct gl_builtin_uniform_element gl_NormalMatrix_elements[] = {
    { NULL, { STATE_MODELVIEW_MATRIX, 0, 0, 0, STATE_MATRIX_INVERSE},
      MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z) },
    { NULL, { STATE_MODELVIEW_MATRIX, 0, 1, 1, STATE_MATRIX_INVERSE},
@@ -498,7 +498,8 @@ builtin_variable_generator::add_uniform(const glsl_type *type,
 
    for (unsigned a = 0; a < array_count; a++) {
       for (unsigned j = 0; j < statevar->num_elements; j++) {
-	 struct gl_builtin_uniform_element *element = &statevar->elements[j];
+	 const struct gl_builtin_uniform_element *element =
+	    &statevar->elements[j];
 
 	 memcpy(slots->tokens, element->tokens, sizeof(element->tokens));
 	 if (type->is_array()) {
diff --git a/mesalib/src/glsl/glcpp/glcpp-parse.y b/mesalib/src/glsl/glcpp/glcpp-parse.y
index f28d8531e..98875837c 100644
--- a/mesalib/src/glsl/glcpp/glcpp-parse.y
+++ b/mesalib/src/glsl/glcpp/glcpp-parse.y
@@ -2062,6 +2062,7 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio
 	/* Add pre-defined macros. */
 	if (parser->is_gles) {
 	   add_builtin_define(parser, "GL_ES", 1);
+           add_builtin_define(parser, "GL_EXT_separate_shader_objects", 1);
 
 	   if (extensions != NULL) {
 	      if (extensions->OES_EGL_image_external)
@@ -2069,6 +2070,7 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio
 	   }
 	} else {
 	   add_builtin_define(parser, "GL_ARB_draw_buffers", 1);
+           add_builtin_define(parser, "GL_ARB_separate_shader_objects", 1);
 	   add_builtin_define(parser, "GL_ARB_texture_rectangle", 1);
            add_builtin_define(parser, "GL_AMD_shader_trinary_minmax", 1);
 
@@ -2134,9 +2136,6 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio
 	      if (extensions->ARB_texture_gather)
 	         add_builtin_define(parser, "GL_ARB_texture_gather", 1);
 
-              if (extensions->ARB_separate_shader_objects)
-                 add_builtin_define(parser, "GL_ARB_separate_shader_objects", 1);
-
 	      if (extensions->ARB_shader_atomic_counters)
 	         add_builtin_define(parser, "GL_ARB_shader_atomic_counters", 1);
 
diff --git a/mesalib/src/glsl/glsl_parser.yy b/mesalib/src/glsl/glsl_parser.yy
index 531ae394b..b71b240de 100644
--- a/mesalib/src/glsl/glsl_parser.yy
+++ b/mesalib/src/glsl/glsl_parser.yy
@@ -1215,7 +1215,7 @@ layout_qualifier_id:
 
       /* Layout qualifiers for GLSL 1.50 geometry shaders. */
       if (!$$.flags.i) {
-         struct {
+         static const struct {
             const char *s;
             GLenum e;
          } map[] = {
@@ -1320,6 +1320,13 @@ layout_qualifier_id:
       if (match_layout_qualifier("location", $1, state) == 0) {
          $$.flags.q.explicit_location = 1;
 
+         if ($$.flags.q.attribute == 1 &&
+             state->ARB_explicit_attrib_location_warn) {
+            _mesa_glsl_warning(& @1, state,
+                               "GL_ARB_explicit_attrib_location layout "
+                               "identifier `%s' used", $1);
+         }
+
          if ($3 >= 0) {
             $$.location = $3;
          } else {
@@ -1369,7 +1376,7 @@ layout_qualifier_id:
          }
       }
 
-      static const char *local_size_qualifiers[3] = {
+      static const char * const local_size_qualifiers[3] = {
          "local_size_x",
          "local_size_y",
          "local_size_z",
@@ -1427,10 +1434,6 @@ layout_qualifier_id:
          _mesa_glsl_error(& @1, state, "unrecognized layout identifier "
                           "`%s'", $1);
          YYERROR;
-      } else if (state->ARB_explicit_attrib_location_warn) {
-         _mesa_glsl_warning(& @1, state,
-                            "GL_ARB_explicit_attrib_location layout "
-                            "identifier `%s' used", $1);
       }
    }
    | interface_block_layout_qualifier
diff --git a/mesalib/src/glsl/glsl_parser_extras.cpp b/mesalib/src/glsl/glsl_parser_extras.cpp
index 03c2a972a..d3339e779 100644
--- a/mesalib/src/glsl/glsl_parser_extras.cpp
+++ b/mesalib/src/glsl/glsl_parser_extras.cpp
@@ -49,7 +49,7 @@ glsl_compute_version_string(void *mem_ctx, bool is_es, unsigned version)
 }
 
 
-static unsigned known_desktop_glsl_versions[] =
+static const unsigned known_desktop_glsl_versions[] =
    { 110, 120, 130, 140, 150, 330, 400, 410, 420, 430, 440 };
 
 
@@ -197,6 +197,12 @@ _mesa_glsl_parse_state::_mesa_glsl_parse_state(struct gl_context *_ctx,
    this->default_uniform_qualifier->flags.q.shared = 1;
    this->default_uniform_qualifier->flags.q.column_major = 1;
 
+   this->fs_uses_gl_fragcoord = false;
+   this->fs_redeclares_gl_fragcoord = false;
+   this->fs_origin_upper_left = false;
+   this->fs_pixel_center_integer = false;
+   this->fs_redeclares_gl_fragcoord_with_no_layout_qualifiers = false;
+
    this->gs_input_prim_type_specified = false;
    this->gs_input_size = 0;
    this->in_qualifier = new(this) ast_type_qualifier();
@@ -500,40 +506,53 @@ struct _mesa_glsl_extension {
 static const _mesa_glsl_extension _mesa_glsl_supported_extensions[] = {
    /*                                  API availability */
    /* name                             GL     ES         supported flag */
+
+   /* ARB extensions go here, sorted alphabetically.
+    */
    EXT(ARB_arrays_of_arrays,           true,  false,     ARB_arrays_of_arrays),
+   EXT(ARB_compute_shader,             true,  false,     ARB_compute_shader),
    EXT(ARB_conservative_depth,         true,  false,     ARB_conservative_depth),
    EXT(ARB_draw_buffers,               true,  false,     dummy_true),
    EXT(ARB_draw_instanced,             true,  false,     ARB_draw_instanced),
    EXT(ARB_explicit_attrib_location,   true,  false,     ARB_explicit_attrib_location),
    EXT(ARB_fragment_coord_conventions, true,  false,     ARB_fragment_coord_conventions),
-   EXT(ARB_texture_rectangle,          true,  false,     dummy_true),
-   EXT(EXT_texture_array,              true,  false,     EXT_texture_array),
-   EXT(ARB_separate_shader_objects,    true,  false,     ARB_separate_shader_objects),
-   EXT(ARB_shader_texture_lod,         true,  false,     ARB_shader_texture_lod),
-   EXT(ARB_shader_stencil_export,      true,  false,     ARB_shader_stencil_export),
-   EXT(AMD_conservative_depth,         true,  false,     ARB_conservative_depth),
-   EXT(AMD_shader_stencil_export,      true,  false,     ARB_shader_stencil_export),
-   EXT(OES_texture_3D,                 false, true,      EXT_texture3D),
-   EXT(OES_EGL_image_external,         false, true,      OES_EGL_image_external),
+   EXT(ARB_gpu_shader5,                true,  false,     ARB_gpu_shader5),
+   EXT(ARB_sample_shading,             true,  false,     ARB_sample_shading),
+   EXT(ARB_separate_shader_objects,    true,  false,     dummy_true),
+   EXT(ARB_shader_atomic_counters,     true,  false,     ARB_shader_atomic_counters),
    EXT(ARB_shader_bit_encoding,        true,  false,     ARB_shader_bit_encoding),
-   EXT(ARB_uniform_buffer_object,      true,  false,     ARB_uniform_buffer_object),
-   EXT(OES_standard_derivatives,       false,  true,     OES_standard_derivatives),
-   EXT(ARB_texture_cube_map_array,     true,  false,     ARB_texture_cube_map_array),
-   EXT(ARB_shading_language_packing,   true,  false,     ARB_shading_language_packing),
+   EXT(ARB_shader_image_load_store,    true,  false,     ARB_shader_image_load_store),
+   EXT(ARB_shader_stencil_export,      true,  false,     ARB_shader_stencil_export),
+   EXT(ARB_shader_texture_lod,         true,  false,     ARB_shader_texture_lod),
    EXT(ARB_shading_language_420pack,   true,  false,     ARB_shading_language_420pack),
+   EXT(ARB_shading_language_packing,   true,  false,     ARB_shading_language_packing),
+   EXT(ARB_texture_cube_map_array,     true,  false,     ARB_texture_cube_map_array),
+   EXT(ARB_texture_gather,             true,  false,     ARB_texture_gather),
    EXT(ARB_texture_multisample,        true,  false,     ARB_texture_multisample),
    EXT(ARB_texture_query_levels,       true,  false,     ARB_texture_query_levels),
    EXT(ARB_texture_query_lod,          true,  false,     ARB_texture_query_lod),
-   EXT(ARB_gpu_shader5,                true,  false,     ARB_gpu_shader5),
+   EXT(ARB_texture_rectangle,          true,  false,     dummy_true),
+   EXT(ARB_uniform_buffer_object,      true,  false,     ARB_uniform_buffer_object),
+   EXT(ARB_viewport_array,             true,  false,     ARB_viewport_array),
+
+   /* KHR extensions go here, sorted alphabetically.
+    */
+
+   /* OES extensions go here, sorted alphabetically.
+    */
+   EXT(OES_EGL_image_external,         false, true,      OES_EGL_image_external),
+   EXT(OES_standard_derivatives,       false, true,      OES_standard_derivatives),
+   EXT(OES_texture_3D,                 false, true,      EXT_texture3D),
+
+   /* All other extensions go here, sorted alphabetically.
+    */
+   EXT(AMD_conservative_depth,         true,  false,     ARB_conservative_depth),
+   EXT(AMD_shader_stencil_export,      true,  false,     ARB_shader_stencil_export),
+   EXT(AMD_shader_trinary_minmax,      true,  false,     dummy_true),
    EXT(AMD_vertex_shader_layer,        true,  false,     AMD_vertex_shader_layer),
+   EXT(EXT_separate_shader_objects,    false, true,      dummy_true),
    EXT(EXT_shader_integer_mix,         true,  true,      EXT_shader_integer_mix),
-   EXT(ARB_texture_gather,             true,  false,     ARB_texture_gather),
-   EXT(ARB_shader_atomic_counters,     true,  false,     ARB_shader_atomic_counters),
-   EXT(ARB_sample_shading,             true,  false,     ARB_sample_shading),
-   EXT(AMD_shader_trinary_minmax,      true,  false,     dummy_true),
-   EXT(ARB_viewport_array,             true,  false,     ARB_viewport_array),
-   EXT(ARB_compute_shader,             true,  false,     ARB_compute_shader),
-   EXT(ARB_shader_image_load_store,    true,  false,     ARB_shader_image_load_store),
+   EXT(EXT_texture_array,              true,  false,     EXT_texture_array),
 };
 
 #undef EXT
@@ -637,7 +656,7 @@ _mesa_glsl_process_extension(const char *name, YYLTYPE *name_locp,
       if (extension && extension->compatible_with_state(state)) {
          extension->set_flags(state, behavior);
       } else {
-         static const char *const fmt = "extension `%s' unsupported in %s shader";
+         static const char fmt[] = "extension `%s' unsupported in %s shader";
 
          if (behavior == extension_require) {
             _mesa_glsl_error(name_locp, state, fmt,
@@ -1356,6 +1375,14 @@ set_shader_inout_layout(struct gl_shader *shader,
       assert(!state->cs_input_local_size_specified);
    }
 
+   if (shader->Stage != MESA_SHADER_FRAGMENT) {
+      /* Should have been prevented by the parser. */
+      assert(!state->fs_uses_gl_fragcoord);
+      assert(!state->fs_redeclares_gl_fragcoord);
+      assert(!state->fs_pixel_center_integer);
+      assert(!state->fs_origin_upper_left);
+   }
+
    switch (shader->Stage) {
    case MESA_SHADER_GEOMETRY:
       shader->Geom.VerticesOut = 0;
@@ -1389,6 +1416,15 @@ set_shader_inout_layout(struct gl_shader *shader,
       }
       break;
 
+   case MESA_SHADER_FRAGMENT:
+      shader->redeclares_gl_fragcoord = state->fs_redeclares_gl_fragcoord;
+      shader->uses_gl_fragcoord = state->fs_uses_gl_fragcoord;
+      shader->pixel_center_integer = state->fs_pixel_center_integer;
+      shader->origin_upper_left = state->fs_origin_upper_left;
+      shader->ARB_fragment_coord_conventions_enable =
+         state->ARB_fragment_coord_conventions_enable;
+      break;
+
    default:
       /* Nothing to do. */
       break;
diff --git a/mesalib/src/glsl/glsl_parser_extras.h b/mesalib/src/glsl/glsl_parser_extras.h
index 0fd778dc3..49402fa21 100644
--- a/mesalib/src/glsl/glsl_parser_extras.h
+++ b/mesalib/src/glsl/glsl_parser_extras.h
@@ -144,8 +144,7 @@ struct _mesa_glsl_parse_state {
    {
       if (!this->has_separate_shader_objects()) {
          const char *const requirement = this->es_shader
-            ? "GL_EXT_separate_shader_objects (not supported by this "
-              "implementation)"
+            ? "GL_EXT_separate_shader_objects extension"
             : "GL_ARB_separate_shader_objects extension or GLSL 420";
 
          _mesa_glsl_error(locp, this, "%s explicit location requires %s",
@@ -168,7 +167,8 @@ struct _mesa_glsl_parse_state {
 
    bool has_separate_shader_objects() const
    {
-      return ARB_separate_shader_objects_enable || is_version(410, 0);
+      return ARB_separate_shader_objects_enable || is_version(410, 0)
+         || EXT_separate_shader_objects_enable;
    }
 
    void process_version_directive(YYLTYPE *locp, int version,
@@ -204,6 +204,18 @@ struct _mesa_glsl_parse_state {
    struct ast_type_qualifier *default_uniform_qualifier;
 
    /**
+    * Variables to track different cases if a fragment shader redeclares
+    * built-in variable gl_FragCoord.
+    *
+    * Note: These values are computed at ast_to_hir time rather than at parse
+    * time.
+    */
+   bool fs_redeclares_gl_fragcoord;
+   bool fs_origin_upper_left;
+   bool fs_pixel_center_integer;
+   bool fs_redeclares_gl_fragcoord_with_no_layout_qualifiers;
+
+   /**
     * True if a geometry shader input primitive type was specified using a
     * layout directive.
     *
@@ -341,8 +353,14 @@ struct _mesa_glsl_parse_state {
     * \name Enable bits for GLSL extensions
     */
    /*@{*/
+   /* ARB extensions go here, sorted alphabetically.
+    */
    bool ARB_arrays_of_arrays_enable;
    bool ARB_arrays_of_arrays_warn;
+   bool ARB_compute_shader_enable;
+   bool ARB_compute_shader_warn;
+   bool ARB_conservative_depth_enable;
+   bool ARB_conservative_depth_warn;
    bool ARB_draw_buffers_enable;
    bool ARB_draw_buffers_warn;
    bool ARB_draw_instanced_enable;
@@ -351,70 +369,78 @@ struct _mesa_glsl_parse_state {
    bool ARB_explicit_attrib_location_warn;
    bool ARB_fragment_coord_conventions_enable;
    bool ARB_fragment_coord_conventions_warn;
-   bool ARB_texture_rectangle_enable;
-   bool ARB_texture_rectangle_warn;
-   bool ARB_texture_gather_enable;
-   bool ARB_texture_gather_warn;
-   bool EXT_texture_array_enable;
-   bool EXT_texture_array_warn;
+   bool ARB_gpu_shader5_enable;
+   bool ARB_gpu_shader5_warn;
+   bool ARB_sample_shading_enable;
+   bool ARB_sample_shading_warn;
    bool ARB_separate_shader_objects_enable;
    bool ARB_separate_shader_objects_warn;
-   bool ARB_shader_texture_lod_enable;
-   bool ARB_shader_texture_lod_warn;
-   bool ARB_shader_stencil_export_enable;
-   bool ARB_shader_stencil_export_warn;
-   bool AMD_conservative_depth_enable;
-   bool AMD_conservative_depth_warn;
-   bool ARB_conservative_depth_enable;
-   bool ARB_conservative_depth_warn;
-   bool AMD_shader_stencil_export_enable;
-   bool AMD_shader_stencil_export_warn;
-   bool OES_texture_3D_enable;
-   bool OES_texture_3D_warn;
-   bool OES_EGL_image_external_enable;
-   bool OES_EGL_image_external_warn;
+   bool ARB_shader_atomic_counters_enable;
+   bool ARB_shader_atomic_counters_warn;
    bool ARB_shader_bit_encoding_enable;
    bool ARB_shader_bit_encoding_warn;
-   bool ARB_uniform_buffer_object_enable;
-   bool ARB_uniform_buffer_object_warn;
-   bool OES_standard_derivatives_enable;
-   bool OES_standard_derivatives_warn;
-   bool ARB_texture_cube_map_array_enable;
-   bool ARB_texture_cube_map_array_warn;
+   bool ARB_shader_image_load_store_enable;
+   bool ARB_shader_image_load_store_warn;
+   bool ARB_shader_stencil_export_enable;
+   bool ARB_shader_stencil_export_warn;
+   bool ARB_shader_texture_lod_enable;
+   bool ARB_shader_texture_lod_warn;
+   bool ARB_shading_language_420pack_enable;
+   bool ARB_shading_language_420pack_warn;
    bool ARB_shading_language_packing_enable;
    bool ARB_shading_language_packing_warn;
+   bool ARB_texture_cube_map_array_enable;
+   bool ARB_texture_cube_map_array_warn;
+   bool ARB_texture_gather_enable;
+   bool ARB_texture_gather_warn;
    bool ARB_texture_multisample_enable;
    bool ARB_texture_multisample_warn;
    bool ARB_texture_query_levels_enable;
    bool ARB_texture_query_levels_warn;
    bool ARB_texture_query_lod_enable;
    bool ARB_texture_query_lod_warn;
-   bool ARB_gpu_shader5_enable;
-   bool ARB_gpu_shader5_warn;
+   bool ARB_texture_rectangle_enable;
+   bool ARB_texture_rectangle_warn;
+   bool ARB_uniform_buffer_object_enable;
+   bool ARB_uniform_buffer_object_warn;
+   bool ARB_viewport_array_enable;
+   bool ARB_viewport_array_warn;
+
+   /* KHR extensions go here, sorted alphabetically.
+    */
+
+   /* OES extensions go here, sorted alphabetically.
+    */
+   bool OES_EGL_image_external_enable;
+   bool OES_EGL_image_external_warn;
+   bool OES_standard_derivatives_enable;
+   bool OES_standard_derivatives_warn;
+   bool OES_texture_3D_enable;
+   bool OES_texture_3D_warn;
+
+   /* All other extensions go here, sorted alphabetically.
+    */
+   bool AMD_conservative_depth_enable;
+   bool AMD_conservative_depth_warn;
+   bool AMD_shader_stencil_export_enable;
+   bool AMD_shader_stencil_export_warn;
+   bool AMD_shader_trinary_minmax_enable;
+   bool AMD_shader_trinary_minmax_warn;
    bool AMD_vertex_shader_layer_enable;
    bool AMD_vertex_shader_layer_warn;
-   bool ARB_shading_language_420pack_enable;
-   bool ARB_shading_language_420pack_warn;
-   bool ARB_sample_shading_enable;
-   bool ARB_sample_shading_warn;
+   bool EXT_separate_shader_objects_enable;
+   bool EXT_separate_shader_objects_warn;
    bool EXT_shader_integer_mix_enable;
    bool EXT_shader_integer_mix_warn;
-   bool ARB_shader_atomic_counters_enable;
-   bool ARB_shader_atomic_counters_warn;
-   bool AMD_shader_trinary_minmax_enable;
-   bool AMD_shader_trinary_minmax_warn;
-   bool ARB_viewport_array_enable;
-   bool ARB_viewport_array_warn;
-   bool ARB_compute_shader_enable;
-   bool ARB_compute_shader_warn;
-   bool ARB_shader_image_load_store_enable;
-   bool ARB_shader_image_load_store_warn;
+   bool EXT_texture_array_enable;
+   bool EXT_texture_array_warn;
    /*@}*/
 
    /** Extensions supported by the OpenGL implementation. */
    const struct gl_extensions *extensions;
 
    bool uses_builtin_functions;
+   bool fs_uses_gl_fragcoord;
 
    /**
     * For geometry shaders, size of the most recently seen input declaration
diff --git a/mesalib/src/glsl/ir.cpp b/mesalib/src/glsl/ir.cpp
index f8f661d5e..dec49bd64 100644
--- a/mesalib/src/glsl/ir.cpp
+++ b/mesalib/src/glsl/ir.cpp
@@ -1333,7 +1333,7 @@ ir_dereference::is_lvalue() const
 }
 
 
-static const char *tex_opcode_strs[] = { "tex", "txb", "txl", "txd", "txf", "txf_ms", "txs", "lod", "tg4", "query_levels" };
+static const char * const tex_opcode_strs[] = { "tex", "txb", "txl", "txd", "txf", "txf_ms", "txs", "lod", "tg4", "query_levels" };
 
 const char *ir_texture::opcode_string()
 {
diff --git a/mesalib/src/glsl/ir_optimization.h b/mesalib/src/glsl/ir_optimization.h
index 40bb61392..c63921c26 100644
--- a/mesalib/src/glsl/ir_optimization.h
+++ b/mesalib/src/glsl/ir_optimization.h
@@ -38,6 +38,8 @@
 #define INT_DIV_TO_MUL_RCP 0x40
 #define BITFIELD_INSERT_TO_BFM_BFI 0x80
 #define LDEXP_TO_ARITH     0x100
+#define CARRY_TO_ARITH     0x200
+#define BORROW_TO_ARITH    0x400
 
 /**
  * \see class lower_packing_builtins_visitor
@@ -112,7 +114,7 @@ bool lower_clip_distance(gl_shader *shader);
 void lower_output_reads(exec_list *instructions);
 bool lower_packing_builtins(exec_list *instructions, int op_mask);
 void lower_ubo_reference(struct gl_shader *shader, exec_list *instructions);
-void lower_packed_varyings(void *mem_ctx, unsigned location_base,
+void lower_packed_varyings(void *mem_ctx,
                            unsigned locations_used, ir_variable_mode mode,
                            unsigned gs_input_vertices, gl_shader *shader);
 bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index);
diff --git a/mesalib/src/glsl/link_varyings.cpp b/mesalib/src/glsl/link_varyings.cpp
index c925c00e3..71998dfa9 100644
--- a/mesalib/src/glsl/link_varyings.cpp
+++ b/mesalib/src/glsl/link_varyings.cpp
@@ -172,6 +172,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
 				 gl_shader *producer, gl_shader *consumer)
 {
    glsl_symbol_table parameters;
+   ir_variable *explicit_locations[MAX_VARYING] = { NULL, };
 
    /* Find all shader outputs in the "producer" stage.
     */
@@ -181,7 +182,26 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
       if ((var == NULL) || (var->data.mode != ir_var_shader_out))
 	 continue;
 
-      parameters.add_variable(var);
+      if (!var->data.explicit_location
+          || var->data.location < VARYING_SLOT_VAR0)
+         parameters.add_variable(var);
+      else {
+         /* User-defined varyings with explicit locations are handled
+          * differently because they do not need to have matching names.
+          */
+         const unsigned idx = var->data.location - VARYING_SLOT_VAR0;
+
+         if (explicit_locations[idx] != NULL) {
+            linker_error(prog,
+                         "%s shader has multiple outputs explicitly "
+                         "assigned to location %d\n",
+                         _mesa_shader_stage_to_string(producer->Stage),
+                         idx);
+            return;
+         }
+
+         explicit_locations[idx] = var;
+      }
    }
 
 
@@ -220,7 +240,27 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
                                              front_color, back_color,
                                              consumer->Stage, producer->Stage);
       } else {
-         ir_variable *const output = parameters.get_variable(input->name);
+         /* The rules for connecting inputs and outputs change in the presence
+          * of explicit locations.  In this case, we no longer care about the
+          * names of the variables.  Instead, we care only about the
+          * explicitly assigned location.
+          */
+         ir_variable *output = NULL;
+         if (input->data.explicit_location
+             && input->data.location >= VARYING_SLOT_VAR0) {
+            output = explicit_locations[input->data.location - VARYING_SLOT_VAR0];
+
+            if (output == NULL) {
+               linker_error(prog,
+                            "%s shader input `%s' with explicit location "
+                            "has no matching output\n",
+                            _mesa_shader_stage_to_string(consumer->Stage),
+                            input->name);
+            }
+         } else {
+            output = parameters.get_variable(input->name);
+         }
+
          if (output != NULL) {
             cross_validate_types_and_qualifiers(prog, input, output,
                                                 consumer->Stage, producer->Stage);
@@ -622,7 +662,7 @@ public:
    ~varying_matches();
    void record(ir_variable *producer_var, ir_variable *consumer_var);
    unsigned assign_locations();
-   void store_locations(unsigned producer_base, unsigned consumer_base) const;
+   void store_locations() const;
 
 private:
    /**
@@ -648,8 +688,8 @@ private:
       PACKING_ORDER_VEC3,
    };
 
-   static unsigned compute_packing_class(ir_variable *var);
-   static packing_order_enum compute_packing_order(ir_variable *var);
+   static unsigned compute_packing_class(const ir_variable *var);
+   static packing_order_enum compute_packing_order(const ir_variable *var);
    static int match_comparator(const void *x_generic, const void *y_generic);
 
    /**
@@ -746,7 +786,10 @@ varying_matches::~varying_matches()
 void
 varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var)
 {
-   if (!producer_var->data.is_unmatched_generic_inout) {
+   assert(producer_var != NULL || consumer_var != NULL);
+
+   if ((producer_var && !producer_var->data.is_unmatched_generic_inout)
+       || (consumer_var && !consumer_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.
@@ -781,24 +824,28 @@ varying_matches::record(ir_variable *producer_var, ir_variable *consumer_var)
          realloc(this->matches,
                  sizeof(*this->matches) * this->matches_capacity);
    }
+
+   const ir_variable *const var = (producer_var != NULL)
+      ? producer_var : consumer_var;
+
    this->matches[this->num_matches].packing_class
-      = this->compute_packing_class(producer_var);
+      = this->compute_packing_class(var);
    this->matches[this->num_matches].packing_order
-      = this->compute_packing_order(producer_var);
+      = this->compute_packing_order(var);
    if (this->disable_varying_packing) {
-      unsigned slots = producer_var->type->is_array()
-         ? (producer_var->type->length
-            * producer_var->type->fields.array->matrix_columns)
-         : producer_var->type->matrix_columns;
+      unsigned slots = var->type->is_array()
+         ? (var->type->length * var->type->fields.array->matrix_columns)
+         : var->type->matrix_columns;
       this->matches[this->num_matches].num_components = 4 * slots;
    } else {
       this->matches[this->num_matches].num_components
-         = producer_var->type->component_slots();
+         = var->type->component_slots();
    }
    this->matches[this->num_matches].producer_var = producer_var;
    this->matches[this->num_matches].consumer_var = consumer_var;
    this->num_matches++;
-   producer_var->data.is_unmatched_generic_inout = 0;
+   if (producer_var)
+      producer_var->data.is_unmatched_generic_inout = 0;
    if (consumer_var)
       consumer_var->data.is_unmatched_generic_inout = 0;
 }
@@ -842,8 +889,7 @@ varying_matches::assign_locations()
  * assignments that were made by varying_matches::assign_locations().
  */
 void
-varying_matches::store_locations(unsigned producer_base,
-                                 unsigned consumer_base) const
+varying_matches::store_locations() const
 {
    for (unsigned i = 0; i < this->num_matches; i++) {
       ir_variable *producer_var = this->matches[i].producer_var;
@@ -852,11 +898,14 @@ varying_matches::store_locations(unsigned producer_base,
       unsigned slot = generic_location / 4;
       unsigned offset = generic_location % 4;
 
-      producer_var->data.location = producer_base + slot;
-      producer_var->data.location_frac = offset;
+      if (producer_var) {
+         producer_var->data.location = VARYING_SLOT_VAR0 + slot;
+         producer_var->data.location_frac = offset;
+      }
+
       if (consumer_var) {
          assert(consumer_var->data.location == -1);
-         consumer_var->data.location = consumer_base + slot;
+         consumer_var->data.location = VARYING_SLOT_VAR0 + slot;
          consumer_var->data.location_frac = offset;
       }
    }
@@ -869,7 +918,7 @@ varying_matches::store_locations(unsigned producer_base,
  * be safely backed into the same vec4.
  */
 unsigned
-varying_matches::compute_packing_class(ir_variable *var)
+varying_matches::compute_packing_class(const ir_variable *var)
 {
    /* Without help from the back-end, there is no way to pack together
     * variables with different interpolation types, because
@@ -900,7 +949,7 @@ varying_matches::compute_packing_class(ir_variable *var)
  * other varyings in the same packing class.
  */
 varying_matches::packing_order_enum
-varying_matches::compute_packing_order(ir_variable *var)
+varying_matches::compute_packing_order(const ir_variable *var)
 {
    const glsl_type *element_type = var->type;
 
@@ -943,7 +992,7 @@ varying_matches::match_comparator(const void *x_generic, const void *y_generic)
  * varyings, but excludes variables such as gl_FrontFacing and gl_FragCoord.
  */
 static bool
-is_varying_var(gl_shader_stage stage, const ir_variable *var)
+var_counts_against_varying_limit(gl_shader_stage stage, const ir_variable *var)
 {
    /* Only fragment shaders will take a varying variable as an input */
    if (stage == MESA_SHADER_FRAGMENT &&
@@ -1037,6 +1086,157 @@ private:
 };
 
 
+namespace linker {
+
+bool
+populate_consumer_input_sets(void *mem_ctx, exec_list *ir,
+                             hash_table *consumer_inputs,
+                             hash_table *consumer_interface_inputs,
+                             ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX])
+{
+   memset(consumer_inputs_with_locations,
+          0,
+          sizeof(consumer_inputs_with_locations[0]) * VARYING_SLOT_MAX);
+
+   foreach_list(node, ir) {
+      ir_variable *const input_var = ((ir_instruction *) node)->as_variable();
+
+      if ((input_var != NULL) && (input_var->data.mode == ir_var_shader_in)) {
+         if (input_var->type->is_interface())
+            return false;
+
+         if (input_var->data.explicit_location) {
+            /* assign_varying_locations only cares about finding the
+             * ir_variable at the start of a contiguous location block.
+             *
+             *     - For !producer, consumer_inputs_with_locations isn't used.
+             *
+             *     - For !consumer, consumer_inputs_with_locations is empty.
+             *
+             * For consumer && producer, if you were trying to set some
+             * ir_variable to the middle of a location block on the other side
+             * of producer/consumer, cross_validate_outputs_to_inputs() should
+             * be link-erroring due to either type mismatch or location
+             * overlaps.  If the variables do match up, then they've got a
+             * matching data.location and you only looked at
+             * consumer_inputs_with_locations[var->data.location], not any
+             * following entries for the array/structure.
+             */
+            consumer_inputs_with_locations[input_var->data.location] =
+               input_var;
+         } else if (input_var->get_interface_type() != NULL) {
+            char *const iface_field_name =
+               ralloc_asprintf(mem_ctx, "%s.%s",
+                               input_var->get_interface_type()->name,
+                               input_var->name);
+            hash_table_insert(consumer_interface_inputs, input_var,
+                              iface_field_name);
+         } else {
+            hash_table_insert(consumer_inputs, input_var,
+                              ralloc_strdup(mem_ctx, input_var->name));
+         }
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Find a variable from the consumer that "matches" the specified variable
+ *
+ * This function only finds inputs with names that match.  There is no
+ * validation (here) that the types, etc. are compatible.
+ */
+ir_variable *
+get_matching_input(void *mem_ctx,
+                   const ir_variable *output_var,
+                   hash_table *consumer_inputs,
+                   hash_table *consumer_interface_inputs,
+                   ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX])
+{
+   ir_variable *input_var;
+
+   if (output_var->data.explicit_location) {
+      input_var = consumer_inputs_with_locations[output_var->data.location];
+   } else if (output_var->get_interface_type() != NULL) {
+      char *const iface_field_name =
+         ralloc_asprintf(mem_ctx, "%s.%s",
+                         output_var->get_interface_type()->name,
+                         output_var->name);
+      input_var =
+         (ir_variable *) hash_table_find(consumer_interface_inputs,
+                                         iface_field_name);
+   } else {
+      input_var =
+         (ir_variable *) hash_table_find(consumer_inputs, output_var->name);
+   }
+
+   return (input_var == NULL || input_var->data.mode != ir_var_shader_in)
+      ? NULL : input_var;
+}
+
+}
+
+static int
+io_variable_cmp(const void *_a, const void *_b)
+{
+   const ir_variable *const a = *(const ir_variable **) _a;
+   const ir_variable *const b = *(const ir_variable **) _b;
+
+   if (a->data.explicit_location && b->data.explicit_location)
+      return b->data.location - a->data.location;
+
+   if (a->data.explicit_location && !b->data.explicit_location)
+      return 1;
+
+   if (!a->data.explicit_location && b->data.explicit_location)
+      return -1;
+
+   return -strcmp(a->name, b->name);
+}
+
+/**
+ * Sort the shader IO variables into canonical order
+ */
+static void
+canonicalize_shader_io(exec_list *ir, enum ir_variable_mode io_mode)
+{
+   ir_variable *var_table[MAX_PROGRAM_OUTPUTS * 4];
+   unsigned num_variables = 0;
+
+   foreach_list(node, ir) {
+      ir_variable *const var = ((ir_instruction *) node)->as_variable();
+
+      if (var == NULL || var->data.mode != io_mode)
+         continue;
+
+      /* If we have already encountered more I/O variables that could
+       * successfully link, bail.
+       */
+      if (num_variables == ARRAY_SIZE(var_table))
+         return;
+
+      var_table[num_variables++] = var;
+   }
+
+   if (num_variables == 0)
+      return;
+
+   /* Sort the list in reverse order (io_variable_cmp handles this).  Later
+    * we're going to push the variables on to the IR list as a stack, so we
+    * want the last variable (in canonical order) to be first in the list.
+    */
+   qsort(var_table, num_variables, sizeof(var_table[0]), io_variable_cmp);
+
+   /* Remove the variable from it's current location in the IR, and put it at
+    * the front.
+    */
+   for (unsigned i = 0; i < num_variables; i++) {
+      var_table[i]->remove();
+      ir->push_head(var_table[i]);
+   }
+}
+
 /**
  * Assign locations for all variables that are produced in one pipeline stage
  * (the "producer") and consumed in the next stage (the "consumer").
@@ -1069,8 +1269,6 @@ assign_varying_locations(struct gl_context *ctx,
                          tfeedback_decl *tfeedback_decls,
                          unsigned gs_input_vertices)
 {
-   const unsigned producer_base = VARYING_SLOT_VAR0;
-   const unsigned consumer_base = VARYING_SLOT_VAR0;
    varying_matches matches(ctx->Const.DisableVaryingPacking,
                            consumer && consumer->Stage == MESA_SHADER_FRAGMENT);
    hash_table *tfeedback_candidates
@@ -1079,67 +1277,85 @@ assign_varying_locations(struct gl_context *ctx,
       = hash_table_ctor(0, hash_table_string_hash, hash_table_string_compare);
    hash_table *consumer_interface_inputs
       = hash_table_ctor(0, hash_table_string_hash, hash_table_string_compare);
+   ir_variable *consumer_inputs_with_locations[VARYING_SLOT_MAX] = {
+      NULL,
+   };
 
-   /* Operate in a total of three passes.
+   /* Operate in a total of four passes.
+    *
+    * 1. Sort inputs / outputs into a canonical order.  This is necessary so
+    *    that inputs / outputs of separable shaders will be assigned
+    *    predictable locations regardless of the order in which declarations
+    *    appeared in the shader source.
     *
-    * 1. Assign locations for any matching inputs and outputs.
+    * 2. Assign locations for any matching inputs and outputs.
     *
-    * 2. Mark output variables in the producer that do not have locations as
+    * 3. Mark output variables in the producer that do not have locations as
     *    not being outputs.  This lets the optimizer eliminate them.
     *
-    * 3. Mark input variables in the consumer that do not have locations as
+    * 4. Mark input variables in the consumer that do not have locations as
     *    not being inputs.  This lets the optimizer eliminate them.
     */
+   if (consumer)
+      canonicalize_shader_io(consumer->ir, ir_var_shader_in);
+
+   if (producer)
+      canonicalize_shader_io(producer->ir, ir_var_shader_out);
+
+   if (consumer
+       && !linker::populate_consumer_input_sets(mem_ctx,
+                                                consumer->ir,
+                                                consumer_inputs,
+                                                consumer_interface_inputs,
+                                                consumer_inputs_with_locations)) {
+      assert(!"populate_consumer_input_sets failed");
+      hash_table_dtor(tfeedback_candidates);
+      hash_table_dtor(consumer_inputs);
+      hash_table_dtor(consumer_interface_inputs);
+      return false;
+   }
 
-   if (consumer) {
-      foreach_list(node, consumer->ir) {
-         ir_variable *const input_var =
+   if (producer) {
+      foreach_list(node, producer->ir) {
+         ir_variable *const output_var =
             ((ir_instruction *) node)->as_variable();
 
-         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",
-                                  input_var->get_interface_type()->name,
-                                  input_var->name);
-               hash_table_insert(consumer_interface_inputs, input_var,
-                                 iface_field_name);
-            } else {
-               hash_table_insert(consumer_inputs, input_var,
-                                 ralloc_strdup(mem_ctx, input_var->name));
-            }
-         }
-      }
-   }
-
-   foreach_list(node, producer->ir) {
-      ir_variable *const output_var = ((ir_instruction *) node)->as_variable();
+         if ((output_var == NULL) ||
+             (output_var->data.mode != ir_var_shader_out))
+            continue;
 
-      if ((output_var == NULL) || (output_var->data.mode != ir_var_shader_out))
-	 continue;
+         tfeedback_candidate_generator g(mem_ctx, tfeedback_candidates);
+         g.process(output_var);
 
-      tfeedback_candidate_generator g(mem_ctx, tfeedback_candidates);
-      g.process(output_var);
-
-      ir_variable *input_var;
-      if (output_var->get_interface_type() != NULL) {
-         char *const iface_field_name =
-            ralloc_asprintf(mem_ctx, "%s.%s",
-                            output_var->get_interface_type()->name,
-                            output_var->name);
-         input_var =
-            (ir_variable *) hash_table_find(consumer_interface_inputs,
-                                            iface_field_name);
-      } else {
-         input_var =
-            (ir_variable *) hash_table_find(consumer_inputs, output_var->name);
+         ir_variable *const input_var =
+            linker::get_matching_input(mem_ctx, output_var, consumer_inputs,
+                                       consumer_interface_inputs,
+                                       consumer_inputs_with_locations);
+
+         /* If a matching input variable was found, add this ouptut (and the
+          * input) to the set.  If this is a separable program and there is no
+          * consumer stage, add the output.
+          */
+         if (input_var || (prog->SeparateShader && consumer == NULL)) {
+            matches.record(output_var, input_var);
+         }
       }
+   } else {
+      /* If there's no producer stage, then this must be a separable program.
+       * For example, we may have a program that has just a fragment shader.
+       * Later this program will be used with some arbitrary vertex (or
+       * geometry) shader program.  This means that locations must be assigned
+       * for all the inputs.
+       */
+      foreach_list(node, consumer->ir) {
+         ir_variable *const input_var =
+            ((ir_instruction *) node)->as_variable();
 
-      if (input_var && input_var->data.mode != ir_var_shader_in)
-         input_var = NULL;
+         if ((input_var == NULL) ||
+             (input_var->data.mode != ir_var_shader_in))
+            continue;
 
-      if (input_var) {
-         matches.record(output_var, input_var);
+         matches.record(NULL, input_var);
       }
    }
 
@@ -1162,7 +1378,7 @@ assign_varying_locations(struct gl_context *ctx,
    }
 
    const unsigned slots_used = matches.assign_locations();
-   matches.store_locations(producer_base, consumer_base);
+   matches.store_locations();
 
    for (unsigned i = 0; i < num_tfeedback_decls; ++i) {
       if (!tfeedback_decls[i].is_varying())
@@ -1187,15 +1403,17 @@ assign_varying_locations(struct gl_context *ctx,
        */
       assert(!ctx->Extensions.EXT_transform_feedback);
    } else {
-      lower_packed_varyings(mem_ctx, producer_base, slots_used,
-                            ir_var_shader_out, 0, producer);
+      if (producer) {
+         lower_packed_varyings(mem_ctx, slots_used, ir_var_shader_out,
+                               0, producer);
+      }
       if (consumer) {
-         lower_packed_varyings(mem_ctx, consumer_base, slots_used,
-                               ir_var_shader_in, gs_input_vertices, consumer);
+         lower_packed_varyings(mem_ctx, slots_used, ir_var_shader_in,
+                               gs_input_vertices, consumer);
       }
    }
 
-   if (consumer) {
+   if (consumer && producer) {
       foreach_list(node, consumer->ir) {
          ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
@@ -1244,7 +1462,7 @@ check_against_output_limit(struct gl_context *ctx,
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
       if (var && var->data.mode == ir_var_shader_out &&
-          is_varying_var(producer->Stage, var)) {
+          var_counts_against_varying_limit(producer->Stage, var)) {
          output_vectors += var->type->count_attribute_slots();
       }
    }
@@ -1283,7 +1501,7 @@ check_against_input_limit(struct gl_context *ctx,
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
       if (var && var->data.mode == ir_var_shader_in &&
-          is_varying_var(consumer->Stage, var)) {
+          var_counts_against_varying_limit(consumer->Stage, var)) {
          input_vectors += var->type->count_attribute_slots();
       }
    }
diff --git a/mesalib/src/glsl/linker.cpp b/mesalib/src/glsl/linker.cpp
index c2f7f4863..a43d23082 100644
--- a/mesalib/src/glsl/linker.cpp
+++ b/mesalib/src/glsl/linker.cpp
@@ -1195,6 +1195,83 @@ private:
 };
 
 /**
+ * Performs the cross-validation of layout qualifiers specified in
+ * redeclaration of gl_FragCoord for the attached fragment shaders,
+ * and propagates them to the linked FS and linked shader program.
+ */
+static void
+link_fs_input_layout_qualifiers(struct gl_shader_program *prog,
+	                        struct gl_shader *linked_shader,
+	                        struct gl_shader **shader_list,
+	                        unsigned num_shaders)
+{
+   linked_shader->redeclares_gl_fragcoord = false;
+   linked_shader->uses_gl_fragcoord = false;
+   linked_shader->origin_upper_left = false;
+   linked_shader->pixel_center_integer = false;
+
+   if (linked_shader->Stage != MESA_SHADER_FRAGMENT ||
+       (prog->Version < 150 && !prog->ARB_fragment_coord_conventions_enable))
+      return;
+
+   for (unsigned i = 0; i < num_shaders; i++) {
+      struct gl_shader *shader = shader_list[i];
+      /* From the GLSL 1.50 spec, page 39:
+       *
+       *   "If gl_FragCoord is redeclared in any fragment shader in a program,
+       *    it must be redeclared in all the fragment shaders in that program
+       *    that have a static use gl_FragCoord."
+       *
+       * Exclude the case when one of the 'linked_shader' or 'shader' redeclares
+       * gl_FragCoord with no layout qualifiers but the other one doesn't
+       * redeclare it. If we strictly follow GLSL 1.50 spec's language, it
+       * should be a link error. But, generating link error for this case will
+       * be a wrong behaviour which spec didn't intend to do and it could also
+       * break some applications.
+       */
+      if ((linked_shader->redeclares_gl_fragcoord
+           && !shader->redeclares_gl_fragcoord
+           && shader->uses_gl_fragcoord
+           && (linked_shader->origin_upper_left
+               || linked_shader->pixel_center_integer))
+          || (shader->redeclares_gl_fragcoord
+              && !linked_shader->redeclares_gl_fragcoord
+              && linked_shader->uses_gl_fragcoord
+              && (shader->origin_upper_left
+                  || shader->pixel_center_integer))) {
+             linker_error(prog, "fragment shader defined with conflicting "
+                         "layout qualifiers for gl_FragCoord\n");
+      }
+
+      /* From the GLSL 1.50 spec, page 39:
+       *
+       *   "All redeclarations of gl_FragCoord in all fragment shaders in a
+       *    single program must have the same set of qualifiers."
+       */
+      if (linked_shader->redeclares_gl_fragcoord && shader->redeclares_gl_fragcoord
+          && (shader->origin_upper_left != linked_shader->origin_upper_left
+          || shader->pixel_center_integer != linked_shader->pixel_center_integer)) {
+         linker_error(prog, "fragment shader defined with conflicting "
+                      "layout qualifiers for gl_FragCoord\n");
+      }
+
+      /* Update the linked shader state.  Note that uses_gl_fragcoord should
+       * accumulate the results.  The other values should replace.  If there
+       * are multiple redeclarations, all the fields except uses_gl_fragcoord
+       * are already known to be the same.
+       */
+      if (shader->redeclares_gl_fragcoord || shader->uses_gl_fragcoord) {
+         linked_shader->redeclares_gl_fragcoord =
+            shader->redeclares_gl_fragcoord;
+         linked_shader->uses_gl_fragcoord = linked_shader->uses_gl_fragcoord
+            || shader->uses_gl_fragcoord;
+         linked_shader->origin_upper_left = shader->origin_upper_left;
+         linked_shader->pixel_center_integer = shader->pixel_center_integer;
+      }
+   }
+}
+
+/**
  * Performs the cross-validation of geometry shader max_vertices and
  * primitive type layout qualifiers for the attached geometry shaders,
  * and propagates them to the linked GS and linked shader program.
@@ -1471,6 +1548,7 @@ link_intrastage_shaders(void *mem_ctx,
    linked->NumUniformBlocks = num_uniform_blocks;
    ralloc_steal(linked, linked->UniformBlocks);
 
+   link_fs_input_layout_qualifiers(prog, linked, shader_list, num_shaders);
    link_gs_inout_layout_qualifiers(prog, linked, shader_list, num_shaders);
    link_cs_input_layout_qualifiers(prog, linked, shader_list, num_shaders);
 
@@ -1798,10 +1876,12 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 	     *     active attribute array, both of which require multiple
 	     *     contiguous generic attributes."
 	     *
-	     * Previous versions of the spec contain similar language but omit
-	     * the bit about attribute arrays.
+	     * I think above text prohibits the aliasing of explicit and
+	     * automatic assignments. But, aliasing is allowed in manual
+	     * assignments of attribute locations. See below comments for
+	     * the details.
 	     *
-	     * Page 61 of the OpenGL 4.0 spec also says:
+	     * From OpenGL 4.0 spec, page 61:
 	     *
 	     *     "It is possible for an application to bind more than one
 	     *     attribute name to the same location. This is referred to as
@@ -1814,29 +1894,84 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
 	     *     but implementations are not required to generate an error
 	     *     in this case."
 	     *
-	     * These two paragraphs are either somewhat contradictory, or I
-	     * don't fully understand one or both of them.
-	     */
-	    /* FINISHME: The code as currently written does not support
-	     * FINISHME: attribute location aliasing (see comment above).
+	     * From GLSL 4.30 spec, page 54:
+	     *
+	     *    "A program will fail to link if any two non-vertex shader
+	     *     input variables are assigned to the same location. For
+	     *     vertex shaders, multiple input variables may be assigned
+	     *     to the same location using either layout qualifiers or via
+	     *     the OpenGL API. However, such aliasing is intended only to
+	     *     support vertex shaders where each execution path accesses
+	     *     at most one input per each location. Implementations are
+	     *     permitted, but not required, to generate link-time errors
+	     *     if they detect that every path through the vertex shader
+	     *     executable accesses multiple inputs assigned to any single
+	     *     location. For all shader types, a program will fail to link
+	     *     if explicit location assignments leave the linker unable
+	     *     to find space for other variables without explicit
+	     *     assignments."
+	     *
+	     * From OpenGL ES 3.0 spec, page 56:
+	     *
+	     *    "Binding more than one attribute name to the same location
+	     *     is referred to as aliasing, and is not permitted in OpenGL
+	     *     ES Shading Language 3.00 vertex shaders. LinkProgram will
+	     *     fail when this condition exists. However, aliasing is
+	     *     possible in OpenGL ES Shading Language 1.00 vertex shaders.
+	     *     This will only work if only one of the aliased attributes
+	     *     is active in the executable program, or if no path through
+	     *     the shader consumes more than one attribute of a set of
+	     *     attributes aliased to the same location. A link error can
+	     *     occur if the linker determines that every path through the
+	     *     shader consumes multiple aliased attributes, but implemen-
+	     *     tations are not required to generate an error in this case."
+	     *
+	     * After looking at above references from OpenGL, OpenGL ES and
+	     * GLSL specifications, we allow aliasing of vertex input variables
+	     * in: OpenGL 2.0 (and above) and OpenGL ES 2.0.
+	     *
+	     * NOTE: This is not required by the spec but its worth mentioning
+	     * here that we're not doing anything to make sure that no path
+	     * through the vertex shader executable accesses multiple inputs
+	     * assigned to any single location.
 	     */
+
 	    /* Mask representing the contiguous slots that will be used by
 	     * this attribute.
 	     */
 	    const unsigned attr = var->data.location - generic_base;
 	    const unsigned use_mask = (1 << slots) - 1;
+            const char *const string = (target_index == MESA_SHADER_VERTEX)
+               ? "vertex shader input" : "fragment shader output";
+
+            /* Generate a link error if the requested locations for this
+             * attribute exceed the maximum allowed attribute location.
+             */
+            if (attr + slots > max_index) {
+               linker_error(prog,
+                           "insufficient contiguous locations "
+                           "available for %s `%s' %d %d %d", string,
+                           var->name, used_locations, use_mask, attr);
+               return false;
+            }
 
 	    /* Generate a link error if the set of bits requested for this
 	     * attribute overlaps any previously allocated bits.
 	     */
 	    if ((~(use_mask << attr) & used_locations) != used_locations) {
-	       const char *const string = (target_index == MESA_SHADER_VERTEX)
-		  ? "vertex shader input" : "fragment shader output";
-	       linker_error(prog,
-			    "insufficient contiguous locations "
-			    "available for %s `%s' %d %d %d", string,
-			    var->name, used_locations, use_mask, attr);
-	       return false;
+               if (target_index == MESA_SHADER_FRAGMENT ||
+                   (prog->IsES && prog->Version >= 300)) {
+                  linker_error(prog,
+                               "overlapping location is assigned "
+                               "to %s `%s' %d %d %d\n", string,
+                               var->name, used_locations, use_mask, attr);
+                  return false;
+               } else {
+                  linker_warning(prog,
+                                 "overlapping location is assigned "
+                                 "to %s `%s' %d %d %d\n", string,
+                                 var->name, used_locations, use_mask, attr);
+               }
 	    }
 
 	    used_locations |= (use_mask << attr);
@@ -2115,6 +2250,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
    ralloc_free(prog->AtomicBuffers);
    prog->AtomicBuffers = NULL;
    prog->NumAtomicBuffers = 0;
+   prog->ARB_fragment_coord_conventions_enable = false;
 
    /* Separate the shaders into groups based on their type.
     */
@@ -2141,6 +2277,9 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
 	 goto done;
       }
 
+      prog->ARB_fragment_coord_conventions_enable |=
+         prog->Shaders[i]->ARB_fragment_coord_conventions_enable;
+
       gl_shader_stage shader_type = prog->Shaders[i]->Stage;
       shader_list[shader_type][num_shaders[shader_type]] = prog->Shaders[i];
       num_shaders[shader_type]++;
@@ -2161,7 +2300,8 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
    /* Geometry shaders have to be linked with vertex shaders.
     */
    if (num_shaders[MESA_SHADER_GEOMETRY] > 0 &&
-       num_shaders[MESA_SHADER_VERTEX] == 0) {
+       num_shaders[MESA_SHADER_VERTEX] == 0 &&
+       !prog->SeparateShader) {
       linker_error(prog, "Geometry shader must be linked with "
 		   "vertex shader\n");
       goto done;
@@ -2363,7 +2503,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
    if (last >= 0 && last < MESA_SHADER_FRAGMENT) {
       gl_shader *const sh = prog->_LinkedShaders[last];
 
-      if (num_tfeedback_decls != 0) {
+      if (num_tfeedback_decls != 0 || prog->SeparateShader) {
          /* There was no fragment shader, but we still have to assign varying
           * locations for use by transform feedback.
           */
@@ -2377,7 +2517,8 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
       do_dead_builtin_varyings(ctx, sh, NULL,
                                num_tfeedback_decls, tfeedback_decls);
 
-      demote_shader_inputs_and_outputs(sh, ir_var_shader_out);
+      if (!prog->SeparateShader)
+         demote_shader_inputs_and_outputs(sh, ir_var_shader_out);
 
       /* Eliminate code that is now dead due to unused outputs being demoted.
        */
@@ -2392,7 +2533,16 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
       do_dead_builtin_varyings(ctx, NULL, sh,
                                num_tfeedback_decls, tfeedback_decls);
 
-      demote_shader_inputs_and_outputs(sh, ir_var_shader_in);
+      if (prog->SeparateShader) {
+         if (!assign_varying_locations(ctx, mem_ctx, prog,
+                                       NULL /* producer */,
+                                       sh /* consumer */,
+                                       0 /* num_tfeedback_decls */,
+                                       NULL /* tfeedback_decls */,
+                                       0 /* gs_input_vertices */))
+            goto done;
+      } else
+         demote_shader_inputs_and_outputs(sh, ir_var_shader_in);
 
       while (do_dead_code(sh->ir, false))
          ;
@@ -2457,7 +2607,7 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
     * fragment shader) is absent. So, the extension shouldn't change the
     * behavior specified in GLSL specification.
     */
-   if (!prog->InternalSeparateShader && ctx->API == API_OPENGLES2) {
+   if (!prog->SeparateShader && ctx->API == API_OPENGLES2) {
       if (prog->_LinkedShaders[MESA_SHADER_VERTEX] == NULL) {
 	 linker_error(prog, "program lacks a vertex shader\n");
       } else if (prog->_LinkedShaders[MESA_SHADER_FRAGMENT] == NULL) {
diff --git a/mesalib/src/glsl/lower_instructions.cpp b/mesalib/src/glsl/lower_instructions.cpp
index 49316d002..176070c87 100644
--- a/mesalib/src/glsl/lower_instructions.cpp
+++ b/mesalib/src/glsl/lower_instructions.cpp
@@ -39,6 +39,8 @@
  * - MOD_TO_FRACT
  * - LDEXP_TO_ARITH
  * - BITFIELD_INSERT_TO_BFM_BFI
+ * - CARRY_TO_ARITH
+ * - BORROW_TO_ARITH
  *
  * SUB_TO_ADD_NEG:
  * ---------------
@@ -94,6 +96,14 @@
  * Many GPUs implement the bitfieldInsert() built-in from ARB_gpu_shader_5
  * with a pair of instructions.
  *
+ * CARRY_TO_ARITH:
+ * ---------------
+ * Converts ir_carry into (x + y) < x.
+ *
+ * BORROW_TO_ARITH:
+ * ----------------
+ * Converts ir_borrow into (x < y).
+ *
  */
 
 #include "main/core.h" /* for M_LOG2E */
@@ -127,6 +137,8 @@ private:
    void log_to_log2(ir_expression *);
    void bitfield_insert_to_bfm_bfi(ir_expression *);
    void ldexp_to_arith(ir_expression *);
+   void carry_to_arith(ir_expression *);
+   void borrow_to_arith(ir_expression *);
 };
 
 } /* anonymous namespace */
@@ -436,6 +448,42 @@ lower_instructions_visitor::ldexp_to_arith(ir_expression *ir)
    this->progress = true;
 }
 
+void
+lower_instructions_visitor::carry_to_arith(ir_expression *ir)
+{
+   /* Translates
+    *   ir_binop_carry x y
+    * into
+    *   sum = ir_binop_add x y
+    *   bcarry = ir_binop_less sum x
+    *   carry = ir_unop_b2i bcarry
+    */
+
+   ir_rvalue *x_clone = ir->operands[0]->clone(ir, NULL);
+   ir->operation = ir_unop_i2u;
+   ir->operands[0] = b2i(less(add(ir->operands[0], ir->operands[1]), x_clone));
+   ir->operands[1] = NULL;
+
+   this->progress = true;
+}
+
+void
+lower_instructions_visitor::borrow_to_arith(ir_expression *ir)
+{
+   /* Translates
+    *   ir_binop_borrow x y
+    * into
+    *   bcarry = ir_binop_less x y
+    *   carry = ir_unop_b2i bcarry
+    */
+
+   ir->operation = ir_unop_i2u;
+   ir->operands[0] = b2i(less(ir->operands[0], ir->operands[1]));
+   ir->operands[1] = NULL;
+
+   this->progress = true;
+}
+
 ir_visitor_status
 lower_instructions_visitor::visit_leave(ir_expression *ir)
 {
@@ -482,6 +530,16 @@ lower_instructions_visitor::visit_leave(ir_expression *ir)
          ldexp_to_arith(ir);
       break;
 
+   case ir_binop_carry:
+      if (lowering(CARRY_TO_ARITH))
+         carry_to_arith(ir);
+      break;
+
+   case ir_binop_borrow:
+      if (lowering(BORROW_TO_ARITH))
+         borrow_to_arith(ir);
+      break;
+
    default:
       return visit_continue;
    }
diff --git a/mesalib/src/glsl/lower_packed_varyings.cpp b/mesalib/src/glsl/lower_packed_varyings.cpp
index 8c1b8850b..e8654748f 100644
--- a/mesalib/src/glsl/lower_packed_varyings.cpp
+++ b/mesalib/src/glsl/lower_packed_varyings.cpp
@@ -160,8 +160,7 @@ namespace {
 class lower_packed_varyings_visitor
 {
 public:
-   lower_packed_varyings_visitor(void *mem_ctx, unsigned location_base,
-                                 unsigned locations_used,
+   lower_packed_varyings_visitor(void *mem_ctx, unsigned locations_used,
                                  ir_variable_mode mode,
                                  unsigned gs_input_vertices,
                                  exec_list *out_instructions);
@@ -190,18 +189,10 @@ private:
    void * const mem_ctx;
 
    /**
-    * Location representing the first generic varying slot for this shader
-    * stage (e.g. VARYING_SLOT_VAR0 if we are packing vertex shader outputs).
-    * Varyings whose location is less than this value are assumed to
-    * correspond to special fixed function hardware, so they are not lowered.
-    */
-   const unsigned location_base;
-
-   /**
     * Number of generic varying slots which are used by this shader.  This is
     * used to allocate temporary intermediate data structures.  If any varying
     * used by this shader has a location greater than or equal to
-    * location_base + locations_used, an assertion will fire.
+    * VARYING_SLOT_VAR0 + locations_used, an assertion will fire.
     */
    const unsigned locations_used;
 
@@ -235,11 +226,9 @@ private:
 } /* anonymous namespace */
 
 lower_packed_varyings_visitor::lower_packed_varyings_visitor(
-      void *mem_ctx, unsigned location_base, unsigned locations_used,
-      ir_variable_mode mode, unsigned gs_input_vertices,
-      exec_list *out_instructions)
+      void *mem_ctx, unsigned locations_used, ir_variable_mode mode,
+      unsigned gs_input_vertices, exec_list *out_instructions)
    : mem_ctx(mem_ctx),
-     location_base(location_base),
      locations_used(locations_used),
      packed_varyings((ir_variable **)
                      rzalloc_array_size(mem_ctx, sizeof(*packed_varyings),
@@ -259,7 +248,7 @@ lower_packed_varyings_visitor::run(exec_list *instructions)
          continue;
 
       if (var->data.mode != this->mode ||
-          var->data.location < (int) this->location_base ||
+          var->data.location < VARYING_SLOT_VAR0 ||
           !this->needs_lowering(var))
          continue;
 
@@ -542,7 +531,7 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
       unsigned location, ir_variable *unpacked_var, const char *name,
       unsigned vertex_index)
 {
-   unsigned slot = location - this->location_base;
+   unsigned slot = location - VARYING_SLOT_VAR0;
    assert(slot < locations_used);
    if (this->packed_varyings[slot] == NULL) {
       char *packed_name = ralloc_asprintf(this->mem_ctx, "packed:%s", name);
@@ -595,7 +584,12 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
 bool
 lower_packed_varyings_visitor::needs_lowering(ir_variable *var)
 {
-   /* Things composed of vec4's don't need lowering.  Everything else does. */
+   /* Things composed of vec4's and varyings with explicitly assigned
+    * locations don't need lowering.  Everything else does.
+    */
+   if (var->data.explicit_location)
+      return false;
+
    const glsl_type *type = var->type;
    if (this->gs_input_vertices != 0) {
       assert(type->is_array());
@@ -654,9 +648,9 @@ lower_packed_varyings_gs_splicer::visit(ir_emit_vertex *ev)
 
 
 void
-lower_packed_varyings(void *mem_ctx, unsigned location_base,
-                      unsigned locations_used, ir_variable_mode mode,
-                      unsigned gs_input_vertices, gl_shader *shader)
+lower_packed_varyings(void *mem_ctx, unsigned locations_used,
+                      ir_variable_mode mode, unsigned gs_input_vertices,
+                      gl_shader *shader)
 {
    exec_list *instructions = shader->ir;
    ir_function *main_func = shader->symbols->get_function("main");
@@ -664,8 +658,7 @@ lower_packed_varyings(void *mem_ctx, unsigned location_base,
    ir_function_signature *main_func_sig
       = main_func->matching_signature(NULL, &void_parameters);
    exec_list new_instructions;
-   lower_packed_varyings_visitor visitor(mem_ctx, location_base,
-                                         locations_used, mode,
+   lower_packed_varyings_visitor visitor(mem_ctx, locations_used, mode,
                                          gs_input_vertices, &new_instructions);
    visitor.run(instructions);
    if (mode == ir_var_shader_out) {
diff --git a/mesalib/src/glsl/main.cpp b/mesalib/src/glsl/main.cpp
index 0d8c01f6a..6c229465c 100644
--- a/mesalib/src/glsl/main.cpp
+++ b/mesalib/src/glsl/main.cpp
@@ -51,6 +51,12 @@
 
 static int glsl_version = 330;
 
+extern "C" void
+_mesa_error_no_memory(const char *caller)
+{
+   fprintf(stderr, "Mesa error: out of memory in %s", caller);
+}
+
 static void
 initialize_context(struct gl_context *ctx, gl_api api)
 {
diff --git a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
index c2a306e7b..6612592aa 100644
--- a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
+++ b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp
@@ -518,14 +518,9 @@ do_dead_builtin_varyings(struct gl_context *ctx,
 
    /* Lowering of built-in varyings has no effect with the core context and
     * GLES2, because they are not available there.
-    *
-    * EXT_separate_shader_objects doesn't allow this optimization,
-    * because a program object can be bound partially (e.g. only one
-    * stage of a program object can be bound).
     */
    if (ctx->API == API_OPENGL_CORE ||
-       ctx->API == API_OPENGLES2 ||
-       ctx->Extensions.EXT_separate_shader_objects) {
+       ctx->API == API_OPENGLES2) {
       return;
    }
 
diff --git a/mesalib/src/mapi/glapi/gen/ARB_multi_bind.xml b/mesalib/src/mapi/glapi/gen/ARB_multi_bind.xml
new file mode 100644
index 000000000..4f2f2a259
--- /dev/null
+++ b/mesalib/src/mapi/glapi/gen/ARB_multi_bind.xml
@@ -0,0 +1,53 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<!-- Note: no GLX protocol info yet. -->
+
+<OpenGLAPI>
+
+<category name="GL_ARB_multi_bind" number="147">
+
+    <function name="BindBuffersBase" offset="assign">
+        <param name="target" type="GLenum"/>
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="buffers" type="const GLuint *"/>
+    </function>
+
+    <function name="BindBuffersRange" offset="assign">
+        <param name="target" type="GLenum"/>
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="buffers" type="const GLuint *"/>
+        <param name="offsets" type="const GLintptr *"/>
+        <param name="sizes" type="const GLsizeiptr *"/>
+    </function>
+
+    <function name="BindTextures" offset="assign">
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="textures" type="const GLuint *"/>
+    </function>
+
+    <function name="BindSamplers" offset="assign">
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="samplers" type="const GLuint *"/>
+    </function>
+
+    <function name="BindImageTextures" offset="assign">
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="textures" type="const GLuint *"/>
+    </function>
+
+    <function name="BindVertexBuffers" offset="assign">
+        <param name="first" type="GLuint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="buffers" type="const GLuint *"/>
+        <param name="offsets" type="const GLintptr *"/>
+        <param name="strides" type="const GLsizei *"/>
+    </function>
+
+</category>
+</OpenGLAPI>
diff --git a/mesalib/src/mapi/glapi/gen/EXT_separate_shader_objects.xml b/mesalib/src/mapi/glapi/gen/EXT_separate_shader_objects.xml
index 03f90a1b1..c6163a193 100644
--- a/mesalib/src/mapi/glapi/gen/EXT_separate_shader_objects.xml
+++ b/mesalib/src/mapi/glapi/gen/EXT_separate_shader_objects.xml
@@ -6,21 +6,292 @@
 <OpenGLAPI>
 <category name="GL_EXT_separate_shader_objects" number="377">
     <!-- Alias of CURRENT_PROGRAM -->
-    <enum name="ACTIVE_PROGRAM_EXT"               value="0x8B8D"/>
+<!--    <enum name="ACTIVE_PROGRAM_EXT"               value="0x8B8D"/> -->
 
-    <function name="UseShaderProgramEXT" deprecated="3.1" offset="assign">
+    <function name="UseShaderProgramEXT" deprecated="3.1" offset="assign" exec="skip">
         <param name="type" type="GLenum"/>
 	<param name="program" type="GLuint"/>
     </function>
 
-    <function name="ActiveProgramEXT" deprecated="3.1" offset="assign">
+    <function name="ActiveProgramEXT" deprecated="3.1" offset="assign" exec="skip">
 	<param name="program" type="GLuint"/>
     </function>
 
-    <function name="CreateShaderProgramEXT" deprecated="3.1" offset="assign">
+    <function name="CreateShaderProgramEXT" deprecated="3.1" offset="assign" exec="skip">
         <param name="type" type="GLenum"/>
         <param name="string" type="const GLchar *"/>
         <return type="GLuint"/>
     </function>
+
+
+    <enum name="ACTIVE_PROGRAM_EXT"                           value="0x8259"/>
+    <enum name="PROGRAM_PIPELINE_BINDING_EXT"                 value="0x825A"/>
+    <enum name="VERTEX_SHADER_BIT_EXT"                        value="0x00000001"/>
+    <enum name="FRAGMENT_SHADER_BIT_EXT"                      value="0x00000002"/>
+    <enum name="ALL_SHADER_BITS_EXT"                          value="0xFFFFFFFF"/>
+    <enum name="PROGRAM_SEPARABLE_EXT"                        value="0x8258"/>
+
+    <function name="UseProgramStagesEXT" alias="UseProgramStages" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+        <param name="stages" type="GLbitfield"/>
+        <param name="program" type="GLuint"/>
+    </function>
+    <function name="ActiveShaderProgramEXT" alias="ActiveShaderProgram" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+        <param name="program" type="GLuint"/>
+    </function>
+    <function name="CreateShaderProgramvEXT" alias="CreateShaderProgramv" static_dispatch="false" es2="2.0">
+        <param name="type" type="GLenum"/>
+        <param name="count" type="GLsizei"/>
+        <param name="strings" type="const GLchar * const *"/>
+        <return type="GLuint"/>
+    </function>
+    <function name="BindProgramPipelineEXT" alias="BindProgramPipeline" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+    </function>
+    <function name="DeleteProgramPipelinesEXT" alias="DeleteProgramPipelines" static_dispatch="false" es2="2.0">
+        <param name="n" type="GLsizei"/>
+        <param name="pipelines" type="const GLuint *"/>
+    </function>
+    <function name="GenProgramPipelinesEXT" alias="GenProgramPipelines" static_dispatch="false" es2="2.0">
+        <param name="n" type="GLsizei"/>
+        <param name="pipelines" type="GLuint *"/>
+    </function>
+    <function name="IsProgramPipelineEXT" alias="IsProgramPipeline" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+        <return type="GLboolean"/>
+    </function>
+    <function name="ProgramParameteriEXT" alias="ProgramParameteri" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="pname" type="GLenum"/>
+        <param name="value" type="GLint"/>
+    </function>
+    <function name="GetProgramPipelineivEXT" alias="GetProgramPipelineiv" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+        <param name="pname" type="GLenum"/>
+        <param name="params" type="GLint *"/>
+    </function>
+    <function name="ProgramUniform1iEXT" alias="ProgramUniform1i" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint"/>
+    </function>
+    <function name="ProgramUniform2iEXT" alias="ProgramUniform2i" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint"/>
+        <param name="y" type="GLint"/>
+    </function>
+    <function name="ProgramUniform3iEXT" alias="ProgramUniform3i" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint"/>
+        <param name="y" type="GLint"/>
+        <param name="z" type="GLint"/>
+    </function>
+    <function name="ProgramUniform4iEXT" alias="ProgramUniform4i" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint"/>
+        <param name="y" type="GLint"/>
+        <param name="z" type="GLint"/>
+        <param name="w" type="GLint"/>
+    </function>
+    <function name="ProgramUniform1uiEXT" alias="ProgramUniform1ui" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint"/>
+    </function>
+    <function name="ProgramUniform2uiEXT" alias="ProgramUniform2ui" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint"/>
+        <param name="y" type="GLuint"/>
+    </function>
+    <function name="ProgramUniform3uiEXT" alias="ProgramUniform3ui" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint"/>
+        <param name="y" type="GLuint"/>
+        <param name="z" type="GLuint"/>
+    </function>
+    <function name="ProgramUniform4uiEXT" alias="ProgramUniform4ui" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint"/>
+        <param name="y" type="GLuint"/>
+        <param name="z" type="GLuint"/>
+        <param name="w" type="GLuint"/>
+    </function>
+    <function name="ProgramUniform1fEXT" alias="ProgramUniform1f" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLfloat"/>
+    </function>
+    <function name="ProgramUniform2fEXT" alias="ProgramUniform2f" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLfloat"/>
+        <param name="y" type="GLfloat"/>
+    </function>
+    <function name="ProgramUniform3fEXT" alias="ProgramUniform3f" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLfloat"/>
+        <param name="y" type="GLfloat"/>
+        <param name="z" type="GLfloat"/>
+    </function>
+    <function name="ProgramUniform4fEXT" alias="ProgramUniform4f" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLfloat"/>
+        <param name="y" type="GLfloat"/>
+        <param name="z" type="GLfloat"/>
+        <param name="w" type="GLfloat"/>
+    </function>
+    <function name="ProgramUniform1ivEXT" alias="ProgramUniform1iv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint *"/>
+    </function>
+    <function name="ProgramUniform2ivEXT" alias="ProgramUniform2iv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint *"/>
+    </function>
+    <function name="ProgramUniform3ivEXT" alias="ProgramUniform3iv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint *"/>
+    </function>
+    <function name="ProgramUniform4ivEXT" alias="ProgramUniform4iv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint *"/>
+    </function>
+    <function name="ProgramUniform1uivEXT" alias="ProgramUniform1uiv" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint *"/>
+    </function>
+    <function name="ProgramUniform2uivEXT" alias="ProgramUniform2uiv" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint *"/>
+    </function>
+    <function name="ProgramUniform3uivEXT" alias="ProgramUniform3uiv" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint *"/>
+    </function>
+    <function name="ProgramUniform4uivEXT" alias="ProgramUniform4uiv" static_dispatch="false" es2="3.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint *"/>
+    </function>
+    <function name="ProgramUniform1fvEXT" alias="ProgramUniform1fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniform2fvEXT" alias="ProgramUniform2fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniform3fvEXT" alias="ProgramUniform3fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniform4fvEXT" alias="ProgramUniform4fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix2fvEXT" alias="ProgramUniformMatrix2fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix3fvEXT" alias="ProgramUniformMatrix3fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix4fvEXT" alias="ProgramUniformMatrix4fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix2x3fvEXT" alias="ProgramUniformMatrix2x3fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix3x2fvEXT" alias="ProgramUniformMatrix3x2fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix2x4fvEXT" alias="ProgramUniformMatrix2x4fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix4x2fvEXT" alias="ProgramUniformMatrix4x2fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix3x4fvEXT" alias="ProgramUniformMatrix3x4fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ProgramUniformMatrix4x3fvEXT" alias="ProgramUniformMatrix4x3fv" static_dispatch="false" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="transpose" type="GLboolean"/>
+        <param name="value" type="const GLfloat *"/>
+    </function>
+    <function name="ValidateProgramPipelineEXT" alias="ValidateProgramPipeline" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+    </function>
+    <function name="GetProgramPipelineInfoLogEXT" alias="GetProgramPipelineInfoLog" static_dispatch="false" es2="2.0">
+        <param name="pipeline" type="GLuint"/>
+        <param name="bufSize" type="GLsizei"/>
+        <param name="length" type="GLsizei *"/>
+        <param name="infoLog" type="GLchar *"/>
+    </function>
 </category>
 </OpenGLAPI>
diff --git a/mesalib/src/mapi/glapi/gen/INTEL_performance_query.xml b/mesalib/src/mapi/glapi/gen/INTEL_performance_query.xml
new file mode 100644
index 000000000..25cd1817f
--- /dev/null
+++ b/mesalib/src/mapi/glapi/gen/INTEL_performance_query.xml
@@ -0,0 +1,93 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_INTEL_performance_query" number="443">
+
+  <function name="GetFirstPerfQueryIdINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryId" type="GLuint *"/>
+  </function>
+
+  <function name="GetNextPerfQueryIdINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryId" type="GLuint"/>
+    <param name="nextQueryId" type="GLuint *"/>
+  </function>
+
+  <function name="GetPerfQueryIdByNameINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryName" type="GLchar *"/>
+    <param name="queryId" type="GLuint *"/>
+  </function>
+
+  <function name="GetPerfQueryInfoINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryId" type="GLuint"/>
+    <param name="queryNameLength" type="GLuint"/>
+    <param name="queryName" type="GLchar *"/>
+    <param name="dataSize" type="GLuint *"/>
+    <param name="noCounters" type="GLuint *"/>
+    <param name="noInstances" type="GLuint *"/>
+    <param name="capsMask" type="GLuint *"/>
+  </function>
+
+  <function name="GetPerfCounterInfoINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryId" type="GLuint"/>
+    <param name="counterId" type="GLuint"/>
+    <param name="counterNameLength" type="GLuint"/>
+    <param name="counterName" type="GLchar *"/>
+    <param name="counterDescLength" type="GLuint"/>
+    <param name="counterDesc" type="GLchar *"/>
+    <param name="counterOffset" type="GLuint *"/>
+    <param name="counterDataSize" type="GLuint *"/>
+    <param name="counterTypeEnum" type="GLuint *"/>
+    <param name="counterDataTypeEnum" type="GLuint *"/>
+    <param name="rawCounterMaxValue" type="GLuint64 *"/>
+  </function>
+
+  <function name="CreatePerfQueryINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryId" type="GLuint"/>
+    <param name="queryHandle" type="GLuint *"/>
+  </function>
+
+  <function name="DeletePerfQueryINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryHandle" type="GLuint"/>
+  </function>
+
+  <function name="BeginPerfQueryINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryHandle" type="GLuint"/>
+  </function>
+
+  <function name="EndPerfQueryINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryHandle" type="GLuint"/>
+  </function>
+
+  <function name="GetPerfQueryDataINTEL" offset="assign" static_dispatch="false" es2="2.0">
+    <param name="queryHandle" type="GLuint"/>
+    <param name="flags" type="GLuint"/>
+    <param name="dataSize" type="GLsizei"/>
+    <param name="data" type="GLvoid *"/>
+    <param name="bytesWritten" type="GLuint *"/>
+  </function>
+
+  <enum name="PERFQUERY_SINGLE_CONTEXT_INTEL" value="0x0000"/>
+  <enum name="PERFQUERY_GLOBAL_CONTEXT_INTEL" value="0x0001"/>
+  <enum name="PERFQUERY_WAIT_INTEL" value="0x83FB"/>
+  <enum name="PERFQUERY_FLUSH_INTEL" value="0x83FA"/>
+  <enum name="PERFQUERY_DONOT_FLUSH_INTEL" value="0x83F9"/>
+  <enum name="PERFQUERY_COUNTER_EVENT_INTEL" value="0x94F0"/>
+  <enum name="PERFQUERY_COUNTER_DURATION_NORM_INTEL" value="0x94F1"/>
+  <enum name="PERFQUERY_COUNTER_DURATION_RAW_INTEL" value="0x94F2"/>
+  <enum name="PERFQUERY_COUNTER_THROUGHPUT_INTEL" value="0x94F3"/>
+  <enum name="PERFQUERY_COUNTER_RAW_INTEL" value="0x94F4"/>
+  <enum name="PERFQUERY_COUNTER_TIMESTAMP_INTEL" value="0x94F5"/>
+  <enum name="PERFQUERY_COUNTER_DATA_UINT32_INTEL" value="0x94F8"/>
+  <enum name="PERFQUERY_COUNTER_DATA_UINT64_INTEL" value="0x94F9"/>
+  <enum name="PERFQUERY_COUNTER_DATA_FLOAT_INTEL" value="0x94FA"/>
+  <enum name="PERFQUERY_COUNTER_DATA_DOUBLE_INTEL" value="0x94FB"/>
+  <enum name="PERFQUERY_COUNTER_DATA_BOOL32_INTEL" value="0x94FC"/>
+  <enum name="PERFQUERY_QUERY_NAME_LENGTH_MAX_INTEL" value="0x94FD"/>
+  <enum name="PERFQUERY_COUNTER_NAME_LENGTH_MAX_INTEL" value="0x94FE"/>
+  <enum name="PERFQUERY_COUNTER_DESC_LENGTH_MAX_INTEL" value="0x94FF"/>
+  <enum name="PERFQUERY_GPA_EXTENDED_COUNTERS_INTEL" value="0x9500"/>
+</category>
+
+</OpenGLAPI>
diff --git a/mesalib/src/mapi/glapi/gen/Makefile.am b/mesalib/src/mapi/glapi/gen/Makefile.am
index 6b932e73f..6f36989bc 100644
--- a/mesalib/src/mapi/glapi/gen/Makefile.am
+++ b/mesalib/src/mapi/glapi/gen/Makefile.am
@@ -133,6 +133,7 @@ API_XML = \
 	ARB_internalformat_query.xml \
 	ARB_invalidate_subdata.xml \
 	ARB_map_buffer_range.xml \
+	ARB_multi_bind.xml \
 	ARB_robustness.xml \
 	ARB_sample_shading.xml \
 	ARB_sampler_objects.xml \
@@ -171,6 +172,7 @@ API_XML = \
 	EXT_texture_array.xml \
 	EXT_texture_integer.xml \
 	EXT_transform_feedback.xml \
+	INTEL_performance_query.xml \
 	KHR_debug.xml \
 	NV_conditional_render.xml \
 	NV_primitive_restart.xml \
diff --git a/mesalib/src/mapi/glapi/gen/gl_API.xml b/mesalib/src/mapi/glapi/gen/gl_API.xml
index 7cb10b3ae..4e430fda5 100755
--- a/mesalib/src/mapi/glapi/gen/gl_API.xml
+++ b/mesalib/src/mapi/glapi/gen/gl_API.xml
@@ -8341,6 +8341,10 @@
     </function>
 </category>
 
+<!-- ARB extensions #145...#146 -->
+
+<xi:include href="ARB_multi_bind.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+
 <!-- Non-ARB extensions sorted by extension number. -->
 
 <category name="GL_EXT_blend_color" number="2">
@@ -12808,6 +12812,8 @@
     <enum name="SKIP_DECODE_EXT"                      value="0x8A4A"/>
 </category>
 
+<xi:include href="INTEL_performance_query.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/>
+
 <!-- Unnumbered extensions sorted by name. -->
 
 <category name="GL_ATI_blend_equation_separate">
diff --git a/mesalib/src/mesa/drivers/common/meta.c b/mesalib/src/mesa/drivers/common/meta.c
index ab86f9c96..b4c30564f 100644
--- a/mesalib/src/mesa/drivers/common/meta.c
+++ b/mesalib/src/mesa/drivers/common/meta.c
@@ -86,6 +86,9 @@
 /** Return offset in bytes of the field within a vertex struct */
 #define OFFSET(FIELD) ((void *) offsetof(struct vertex, FIELD))
 
+static void
+meta_clear(struct gl_context *ctx, GLbitfield buffers, bool glsl);
+
 static struct blit_shader *
 choose_blit_shader(GLenum target, struct blit_shader_table *table);
 
@@ -389,6 +392,24 @@ _mesa_meta_init(struct gl_context *ctx)
    ctx->Meta = CALLOC_STRUCT(gl_meta_state);
 }
 
+static GLenum
+gl_buffer_index_to_drawbuffers_enum(gl_buffer_index bufindex)
+{
+   assert(bufindex < BUFFER_COUNT);
+
+   if (bufindex >= BUFFER_COLOR0)
+      return GL_COLOR_ATTACHMENT0 + bufindex - BUFFER_COLOR0;
+   else if (bufindex == BUFFER_FRONT_LEFT)
+      return GL_FRONT_LEFT;
+   else if (bufindex == BUFFER_FRONT_RIGHT)
+      return GL_FRONT_RIGHT;
+   else if (bufindex == BUFFER_BACK_LEFT)
+      return GL_BACK_LEFT;
+   else if (bufindex == BUFFER_BACK_RIGHT)
+      return GL_BACK_RIGHT;
+
+   return GL_NONE;
+}
 
 /**
  * Free context meta-op state.
@@ -576,20 +597,21 @@ _mesa_meta_begin(struct gl_context *ctx, GLbitfield state)
          _mesa_set_enable(ctx, GL_FRAGMENT_SHADER_ATI, GL_FALSE);
       }
 
-      if (ctx->Extensions.ARB_separate_shader_objects) {
-         /* Warning it must be done before _mesa_UseProgram call */
-         _mesa_reference_pipeline_object(ctx, &save->_Shader, ctx->_Shader);
+      if (ctx->Pipeline.Current) {
          _mesa_reference_pipeline_object(ctx, &save->Pipeline,
                                          ctx->Pipeline.Current);
          _mesa_BindProgramPipeline(0);
       }
 
-      for (i = 0; i < MESA_SHADER_STAGES; i++) {
+      /* Save the shader state from ctx->Shader (instead of ctx->_Shader) so
+       * that we don't have to worry about the current pipeline state.
+       */
+      for (i = 0; i <= MESA_SHADER_FRAGMENT; i++) {
          _mesa_reference_shader_program(ctx, &save->Shader[i],
-                                     ctx->_Shader->CurrentProgram[i]);
+                                        ctx->Shader.CurrentProgram[i]);
       }
       _mesa_reference_shader_program(ctx, &save->ActiveShader,
-                                     ctx->_Shader->ActiveProgram);
+                                     ctx->Shader.ActiveProgram);
 
       _mesa_UseProgram(0);
    }
@@ -774,6 +796,23 @@ _mesa_meta_begin(struct gl_context *ctx, GLbitfield state)
          _mesa_set_framebuffer_srgb(ctx, GL_FALSE);
    }
 
+   if (state & MESA_META_DRAW_BUFFERS) {
+      int buf, real_color_buffers = 0;
+      memset(save->ColorDrawBuffers, 0, sizeof(save->ColorDrawBuffers));
+
+      for (buf = 0; buf < MAX_DRAW_BUFFERS; buf++) {
+         int buf_index = ctx->DrawBuffer->_ColorDrawBufferIndexes[buf];
+         if (buf_index == -1)
+            continue;
+
+         save->ColorDrawBuffers[buf] =
+            gl_buffer_index_to_drawbuffers_enum(buf_index);
+
+         if (++real_color_buffers >= ctx->DrawBuffer->_NumColorDrawBuffers)
+            break;
+      }
+   }
+
    /* misc */
    {
       save->Lighting = ctx->Light.Enabled;
@@ -908,6 +947,14 @@ _mesa_meta_end(struct gl_context *ctx)
    }
 
    if (state & MESA_META_SHADER) {
+      static const GLenum targets[] = {
+         GL_VERTEX_SHADER,
+         GL_GEOMETRY_SHADER,
+         GL_FRAGMENT_SHADER,
+      };
+
+      bool any_shader;
+
       if (ctx->Extensions.ARB_vertex_program) {
          _mesa_set_enable(ctx, GL_VERTEX_PROGRAM_ARB,
                           save->VertexProgramEnabled);
@@ -929,37 +976,46 @@ _mesa_meta_end(struct gl_context *ctx)
                           save->ATIFragmentShaderEnabled);
       }
 
-      /* Warning it must be done before _mesa_use_shader_program call */
-      if (ctx->Extensions.ARB_separate_shader_objects) {
-         _mesa_reference_pipeline_object(ctx, &ctx->_Shader, save->_Shader);
-         _mesa_reference_pipeline_object(ctx, &ctx->Pipeline.Current,
-                                         save->Pipeline);
-         _mesa_reference_pipeline_object(ctx, &save->Pipeline, NULL);
-      }
+      any_shader = false;
+      for (i = 0; i <= MESA_SHADER_FRAGMENT; i++) {
+         /* It is safe to call _mesa_use_shader_program even if the extension
+          * necessary for that program state is not supported.  In that case,
+          * the saved program object must be NULL and the currently bound
+          * program object must be NULL.  _mesa_use_shader_program is a no-op
+          * in that case.
+          */
+         _mesa_use_shader_program(ctx, targets[i],
+                                  save->Shader[i],
+                                  &ctx->Shader);
 
-      if (ctx->Extensions.ARB_vertex_shader) {
-	 _mesa_use_shader_program(ctx, GL_VERTEX_SHADER,
-                                  save->Shader[MESA_SHADER_VERTEX],
-                                  ctx->_Shader);
+         /* Do this *before* killing the reference. :)
+          */
+         if (save->Shader[i] != NULL)
+            any_shader = true;
+
+         _mesa_reference_shader_program(ctx, &save->Shader[i], NULL);
       }
 
-      if (_mesa_has_geometry_shaders(ctx))
-	 _mesa_use_shader_program(ctx, GL_GEOMETRY_SHADER_ARB,
-                                  save->Shader[MESA_SHADER_GEOMETRY],
-                                  ctx->_Shader);
+      _mesa_reference_shader_program(ctx, &ctx->Shader.ActiveProgram,
+                                     save->ActiveShader);
+      _mesa_reference_shader_program(ctx, &save->ActiveShader, NULL);
 
-      if (ctx->Extensions.ARB_fragment_shader)
-	 _mesa_use_shader_program(ctx, GL_FRAGMENT_SHADER,
-                                  save->Shader[MESA_SHADER_FRAGMENT],
-                                  ctx->_Shader);
+      /* If there were any stages set with programs, use ctx->Shader as the
+       * current shader state.  Otherwise, use Pipeline.Default.  The pipeline
+       * hasn't been restored yet, and that may modify ctx->_Shader further.
+       */
+      if (any_shader)
+         _mesa_reference_pipeline_object(ctx, &ctx->_Shader,
+                                         &ctx->Shader);
+      else
+         _mesa_reference_pipeline_object(ctx, &ctx->_Shader,
+                                         ctx->Pipeline.Default);
 
-      _mesa_reference_shader_program(ctx, &ctx->_Shader->ActiveProgram,
-				     save->ActiveShader);
+      if (save->Pipeline) {
+         _mesa_bind_pipeline(ctx, save->Pipeline);
 
-      for (i = 0; i < MESA_SHADER_STAGES; i++)
-         _mesa_reference_shader_program(ctx, &save->Shader[i], NULL);
-      _mesa_reference_shader_program(ctx, &save->ActiveShader, NULL);
-      _mesa_reference_pipeline_object(ctx, &save->_Shader, NULL);
+         _mesa_reference_pipeline_object(ctx, &save->Pipeline, NULL);
+      }
    }
 
    if (state & MESA_META_STENCIL_TEST) {
@@ -1155,6 +1211,10 @@ _mesa_meta_end(struct gl_context *ctx)
        ctx->CurrentRenderbuffer->Name != save->RenderbufferName)
       _mesa_BindRenderbuffer(GL_RENDERBUFFER, save->RenderbufferName);
 
+   if (state & MESA_META_DRAW_BUFFERS) {
+      _mesa_DrawBuffers(MAX_DRAW_BUFFERS, save->ColorDrawBuffers);
+   }
+
    ctx->Meta->SaveStackDepth--;
 
    ctx->API = save->API;
@@ -1441,100 +1501,13 @@ _mesa_meta_setup_ff_tnl_for_blit(GLuint *VAO, GLuint *VBO,
 void
 _mesa_meta_Clear(struct gl_context *ctx, GLbitfield buffers)
 {
-   struct clear_state *clear = &ctx->Meta->Clear;
-   struct vertex verts[4];
-   /* save all state but scissor, pixel pack/unpack */
-   GLbitfield metaSave = (MESA_META_ALL -
-			  MESA_META_SCISSOR -
-			  MESA_META_PIXEL_STORE -
-			  MESA_META_CONDITIONAL_RENDER -
-                          MESA_META_FRAMEBUFFER_SRGB);
-   const GLuint stencilMax = (1 << ctx->DrawBuffer->Visual.stencilBits) - 1;
-
-   if (buffers & BUFFER_BITS_COLOR) {
-      /* if clearing color buffers, don't save/restore colormask */
-      metaSave -= MESA_META_COLOR_MASK;
-   }
-
-   _mesa_meta_begin(ctx, metaSave);
-
-   _mesa_meta_setup_vertex_objects(&clear->VAO, &clear->VBO, false, 3, 0, 4);
-
-   /* GL_COLOR_BUFFER_BIT */
-   if (buffers & BUFFER_BITS_COLOR) {
-      /* leave colormask, glDrawBuffer state as-is */
-
-      /* Clears never have the color clamped. */
-      if (ctx->Extensions.ARB_color_buffer_float)
-         _mesa_ClampColor(GL_CLAMP_FRAGMENT_COLOR, GL_FALSE);
-   }
-   else {
-      ASSERT(metaSave & MESA_META_COLOR_MASK);
-      _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE);
-   }
-
-   /* GL_DEPTH_BUFFER_BIT */
-   if (buffers & BUFFER_BIT_DEPTH) {
-      _mesa_set_enable(ctx, GL_DEPTH_TEST, GL_TRUE);
-      _mesa_DepthFunc(GL_ALWAYS);
-      _mesa_DepthMask(GL_TRUE);
-   }
-   else {
-      assert(!ctx->Depth.Test);
-   }
-
-   /* GL_STENCIL_BUFFER_BIT */
-   if (buffers & BUFFER_BIT_STENCIL) {
-      _mesa_set_enable(ctx, GL_STENCIL_TEST, GL_TRUE);
-      _mesa_StencilOpSeparate(GL_FRONT_AND_BACK,
-                              GL_REPLACE, GL_REPLACE, GL_REPLACE);
-      _mesa_StencilFuncSeparate(GL_FRONT_AND_BACK, GL_ALWAYS,
-                                ctx->Stencil.Clear & stencilMax,
-                                ctx->Stencil.WriteMask[0]);
-   }
-   else {
-      assert(!ctx->Stencil.Enabled);
-   }
-
-   /* vertex positions/colors */
-   {
-      const GLfloat x0 = (GLfloat) ctx->DrawBuffer->_Xmin;
-      const GLfloat y0 = (GLfloat) ctx->DrawBuffer->_Ymin;
-      const GLfloat x1 = (GLfloat) ctx->DrawBuffer->_Xmax;
-      const GLfloat y1 = (GLfloat) ctx->DrawBuffer->_Ymax;
-      const GLfloat z = invert_z(ctx->Depth.Clear);
-      GLuint i;
-
-      verts[0].x = x0;
-      verts[0].y = y0;
-      verts[0].z = z;
-      verts[1].x = x1;
-      verts[1].y = y0;
-      verts[1].z = z;
-      verts[2].x = x1;
-      verts[2].y = y1;
-      verts[2].z = z;
-      verts[3].x = x0;
-      verts[3].y = y1;
-      verts[3].z = z;
-
-      /* vertex colors */
-      for (i = 0; i < 4; i++) {
-         verts[i].r = ctx->Color.ClearColor.f[0];
-         verts[i].g = ctx->Color.ClearColor.f[1];
-         verts[i].b = ctx->Color.ClearColor.f[2];
-         verts[i].a = ctx->Color.ClearColor.f[3];
-      }
-
-      /* upload new vertex data */
-      _mesa_BufferData(GL_ARRAY_BUFFER_ARB, sizeof(verts), verts,
-			  GL_DYNAMIC_DRAW_ARB);
-   }
-
-   /* draw quad */
-   _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4);
+   meta_clear(ctx, buffers, false);
+}
 
-   _mesa_meta_end(ctx);
+void
+_mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
+{
+   meta_clear(ctx, buffers, true);
 }
 
 static void
@@ -1682,21 +1655,60 @@ meta_glsl_clear_cleanup(struct clear_state *clear)
 }
 
 /**
+ * Given a bitfield of BUFFER_BIT_x draw buffers, call glDrawBuffers to
+ * set GL to only draw to those buffers.
+ *
+ * Since the bitfield has no associated order, the assignment of draw buffer
+ * indices to color attachment indices is rather arbitrary.
+ */
+static void
+drawbuffers_from_bitfield(GLbitfield bits)
+{
+   GLenum enums[MAX_DRAW_BUFFERS];
+   int i = 0;
+   int n;
+
+   /* This function is only legal for color buffer bitfields. */
+   assert((bits & ~BUFFER_BITS_COLOR) == 0);
+
+   /* Make sure we don't overflow any arrays. */
+   assert(_mesa_bitcount(bits) <= MAX_DRAW_BUFFERS);
+
+   enums[0] = GL_NONE;
+
+   if (bits & BUFFER_BIT_FRONT_LEFT)
+      enums[i++] = GL_FRONT_LEFT;
+
+   if (bits & BUFFER_BIT_FRONT_RIGHT)
+      enums[i++] = GL_FRONT_RIGHT;
+
+   if (bits & BUFFER_BIT_BACK_LEFT)
+      enums[i++] = GL_BACK_LEFT;
+
+   if (bits & BUFFER_BIT_BACK_RIGHT)
+      enums[i++] = GL_BACK_RIGHT;
+
+   for (n = 0; n < MAX_COLOR_ATTACHMENTS; n++) {
+      if (bits & (1 << (BUFFER_COLOR0 + n)))
+         enums[i++] = GL_COLOR_ATTACHMENT0 + n;
+   }
+
+   _mesa_DrawBuffers(i, enums);
+}
+
+/**
  * Meta implementation of ctx->Driver.Clear() in terms of polygon rendering.
  */
-void
-_mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
+static void
+meta_clear(struct gl_context *ctx, GLbitfield buffers, bool glsl)
 {
    struct clear_state *clear = &ctx->Meta->Clear;
    GLbitfield metaSave;
    const GLuint stencilMax = (1 << ctx->DrawBuffer->Visual.stencilBits) - 1;
    struct gl_framebuffer *fb = ctx->DrawBuffer;
-   const float x0 = ((float)fb->_Xmin / fb->Width)  * 2.0f - 1.0f;
-   const float y0 = ((float)fb->_Ymin / fb->Height) * 2.0f - 1.0f;
-   const float x1 = ((float)fb->_Xmax / fb->Width)  * 2.0f - 1.0f;
-   const float y1 = ((float)fb->_Ymax / fb->Height) * 2.0f - 1.0f;
-   const float z = -invert_z(ctx->Depth.Clear);
+   float x0, y0, x1, y1, z;
    struct vertex verts[4];
+   int i;
 
    metaSave = (MESA_META_ALPHA_TEST |
 	       MESA_META_BLEND |
@@ -1711,7 +1723,18 @@ _mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
                MESA_META_MULTISAMPLE |
                MESA_META_OCCLUSION_QUERY);
 
-   if (!(buffers & BUFFER_BITS_COLOR)) {
+   if (!glsl) {
+      metaSave |= MESA_META_FOG |
+                  MESA_META_PIXEL_TRANSFER |
+                  MESA_META_TRANSFORM |
+                  MESA_META_TEXTURE |
+                  MESA_META_CLAMP_VERTEX_COLOR |
+                  MESA_META_SELECT_FEEDBACK;
+   }
+
+   if (buffers & BUFFER_BITS_COLOR) {
+      metaSave |= MESA_META_DRAW_BUFFERS;
+   } else {
       /* We'll use colormask to disable color writes.  Otherwise,
        * respect color mask
        */
@@ -1720,13 +1743,30 @@ _mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
 
    _mesa_meta_begin(ctx, metaSave);
 
-   meta_glsl_clear_init(ctx, clear);
+   if (glsl) {
+      meta_glsl_clear_init(ctx, clear);
+
+      x0 = ((float) fb->_Xmin / fb->Width)  * 2.0f - 1.0f;
+      y0 = ((float) fb->_Ymin / fb->Height) * 2.0f - 1.0f;
+      x1 = ((float) fb->_Xmax / fb->Width)  * 2.0f - 1.0f;
+      y1 = ((float) fb->_Ymax / fb->Height) * 2.0f - 1.0f;
+      z = -invert_z(ctx->Depth.Clear);
+   } else {
+      _mesa_meta_setup_vertex_objects(&clear->VAO, &clear->VBO, false, 3, 0, 4);
+
+      x0 = (float) fb->_Xmin;
+      y0 = (float) fb->_Ymin;
+      x1 = (float) fb->_Xmax;
+      y1 = (float) fb->_Ymax;
+      z = invert_z(ctx->Depth.Clear);
+   }
 
    if (fb->_IntegerColor) {
+      assert(glsl);
       _mesa_UseProgram(clear->IntegerShaderProg);
       _mesa_Uniform4iv(clear->IntegerColorLocation, 1,
 			  ctx->Color.ClearColor.i);
-   } else {
+   } else if (glsl) {
       _mesa_UseProgram(clear->ShaderProg);
       _mesa_Uniform4fv(clear->ColorLocation, 1,
 			  ctx->Color.ClearColor.f);
@@ -1734,7 +1774,10 @@ _mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
 
    /* GL_COLOR_BUFFER_BIT */
    if (buffers & BUFFER_BITS_COLOR) {
-      /* leave colormask, glDrawBuffer state as-is */
+      /* Only draw to the buffers we were asked to clear. */
+      drawbuffers_from_bitfield(buffers & BUFFER_BITS_COLOR);
+
+      /* leave colormask state as-is */
 
       /* Clears never have the color clamped. */
       if (ctx->Extensions.ARB_color_buffer_float)
@@ -1782,6 +1825,15 @@ _mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
    verts[3].y = y1;
    verts[3].z = z;
 
+   if (!glsl) {
+      for (i = 0; i < 4; i++) {
+         verts[i].r = ctx->Color.ClearColor.f[0];
+         verts[i].g = ctx->Color.ClearColor.f[1];
+         verts[i].b = ctx->Color.ClearColor.f[2];
+         verts[i].a = ctx->Color.ClearColor.f[3];
+      }
+   }
+
    /* upload new vertex data */
    _mesa_BufferData(GL_ARRAY_BUFFER_ARB, sizeof(verts), verts,
 		       GL_DYNAMIC_DRAW_ARB);
@@ -1789,6 +1841,7 @@ _mesa_meta_glsl_Clear(struct gl_context *ctx, GLbitfield buffers)
    /* draw quad(s) */
    if (fb->MaxNumLayers > 0) {
       unsigned layer;
+      assert(glsl);
       for (layer = 0; layer < fb->MaxNumLayers; layer++) {
          if (fb->_IntegerColor)
             _mesa_Uniform1i(clear->IntegerLayerLocation, layer);
@@ -2756,7 +2809,7 @@ copytexsubimage_using_blit_framebuffer(struct gl_context *ctx, GLuint dims,
 
    _mesa_unlock_texture(ctx, texObj);
 
-   _mesa_meta_begin(ctx, MESA_META_ALL);
+   _mesa_meta_begin(ctx, MESA_META_ALL & ~MESA_META_DRAW_BUFFERS);
 
    _mesa_GenFramebuffers(1, &fbo);
    _mesa_BindFramebuffer(GL_DRAW_FRAMEBUFFER, fbo);
@@ -2978,7 +3031,8 @@ decompress_texture_image(struct gl_context *ctx,
       break;
    }
 
-   _mesa_meta_begin(ctx, MESA_META_ALL & ~MESA_META_PIXEL_STORE);
+   _mesa_meta_begin(ctx, MESA_META_ALL & ~(MESA_META_PIXEL_STORE |
+                                           MESA_META_DRAW_BUFFERS));
 
    samplerSave = ctx->Texture.Unit[ctx->Texture.CurrentUnit].Sampler ?
          ctx->Texture.Unit[ctx->Texture.CurrentUnit].Sampler->Name : 0;
diff --git a/mesalib/src/mesa/drivers/common/meta.h b/mesalib/src/mesa/drivers/common/meta.h
index fde4f9a7a..32b71fae7 100644
--- a/mesalib/src/mesa/drivers/common/meta.h
+++ b/mesalib/src/mesa/drivers/common/meta.h
@@ -58,6 +58,7 @@
 #define MESA_META_MULTISAMPLE          0x100000
 #define MESA_META_FRAMEBUFFER_SRGB     0x200000
 #define MESA_META_OCCLUSION_QUERY      0x400000
+#define MESA_META_DRAW_BUFFERS         0x800000
 /**\}*/
 
 /**
@@ -121,7 +122,6 @@ struct save_state
    GLboolean ATIFragmentShaderEnabled;
    struct gl_shader_program *Shader[MESA_SHADER_STAGES];
    struct gl_shader_program *ActiveShader;
-   struct gl_pipeline_object   *_Shader;
    struct gl_pipeline_object   *Pipeline;
 
    /** MESA_META_STENCIL_TEST */
@@ -181,6 +181,9 @@ struct save_state
    GLboolean TransformFeedbackNeedsResume;
 
    GLuint DrawBufferName, ReadBufferName, RenderbufferName;
+
+   /** MESA_META_DRAW_BUFFERS */
+   GLenum ColorDrawBuffers[MAX_DRAW_BUFFERS];
 };
 
 /**
diff --git a/mesalib/src/mesa/drivers/common/meta_blit.c b/mesalib/src/mesa/drivers/common/meta_blit.c
index 5d72dd2ec..c3dc14614 100644
--- a/mesalib/src/mesa/drivers/common/meta_blit.c
+++ b/mesalib/src/mesa/drivers/common/meta_blit.c
@@ -659,7 +659,7 @@ _mesa_meta_BlitFramebuffer(struct gl_context *ctx,
    /* Only scissor affects blit, but we're doing to set a custom scissor if
     * necessary anyway, so save/clear state.
     */
-   _mesa_meta_begin(ctx, MESA_META_ALL);
+   _mesa_meta_begin(ctx, MESA_META_ALL & ~MESA_META_DRAW_BUFFERS);
 
    /* If the clipping earlier changed the destination rect at all, then
     * enable the scissor to clip to it.
diff --git a/mesalib/src/mesa/drivers/common/meta_generate_mipmap.c b/mesalib/src/mesa/drivers/common/meta_generate_mipmap.c
index 3c9ac89af..d12806c3d 100644
--- a/mesalib/src/mesa/drivers/common/meta_generate_mipmap.c
+++ b/mesalib/src/mesa/drivers/common/meta_generate_mipmap.c
@@ -182,7 +182,7 @@ _mesa_meta_GenerateMipmap(struct gl_context *ctx, GLenum target,
       faceTarget = target;
    }
 
-   _mesa_meta_begin(ctx, MESA_META_ALL);
+   _mesa_meta_begin(ctx, MESA_META_ALL & ~MESA_META_DRAW_BUFFERS);
 
    /* Choose between glsl version and fixed function version of
     * GenerateMipmap function.
diff --git a/mesalib/src/mesa/main/bufferobj.c b/mesalib/src/mesa/main/bufferobj.c
index aed6d4366..2a5ffc3e6 100644
--- a/mesalib/src/mesa/main/bufferobj.c
+++ b/mesalib/src/mesa/main/bufferobj.c
@@ -975,6 +975,78 @@ _mesa_lookup_bufferobj(struct gl_context *ctx, GLuint buffer)
 }
 
 
+struct gl_buffer_object *
+_mesa_lookup_bufferobj_locked(struct gl_context *ctx, GLuint buffer)
+{
+   return (struct gl_buffer_object *)
+      _mesa_HashLookupLocked(ctx->Shared->BufferObjects, buffer);
+}
+
+
+void
+_mesa_begin_bufferobj_lookups(struct gl_context *ctx)
+{
+   _mesa_HashLockMutex(ctx->Shared->BufferObjects);
+}
+
+
+void
+_mesa_end_bufferobj_lookups(struct gl_context *ctx)
+{
+   _mesa_HashUnlockMutex(ctx->Shared->BufferObjects);
+}
+
+
+/**
+ * Look up a buffer object for a multi-bind function.
+ *
+ * Unlike _mesa_lookup_bufferobj(), this function also takes care
+ * of generating an error if the buffer ID is not zero or the name
+ * of an existing buffer object.
+ *
+ * If the buffer ID refers to an existing buffer object, a pointer
+ * to the buffer object is returned.  If the ID is zero, a pointer
+ * to the shared NullBufferObj is returned.  If the ID is not zero
+ * and does not refer to a valid buffer object, this function
+ * returns NULL.
+ *
+ * This function assumes that the caller has already locked the
+ * hash table mutex by calling _mesa_begin_bufferobj_lookups().
+ */
+struct gl_buffer_object *
+_mesa_multi_bind_lookup_bufferobj(struct gl_context *ctx,
+                                  const GLuint *buffers,
+                                  GLuint index, const char *caller)
+{
+   struct gl_buffer_object *bufObj;
+
+   if (buffers[index] != 0) {
+      bufObj = _mesa_lookup_bufferobj_locked(ctx, buffers[index]);
+
+      /* The multi-bind functions don't create the buffer objects
+         when they don't exist. */
+      if (bufObj == &DummyBufferObject)
+         bufObj = NULL;
+   } else
+      bufObj = ctx->Shared->NullBufferObj;
+
+   if (!bufObj) {
+      /* The ARB_multi_bind spec says:
+       *
+       *    "An INVALID_OPERATION error is generated if any value
+       *     in <buffers> is not zero or the name of an existing
+       *     buffer object (per binding)."
+       */
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(buffers[%u]=%u is not zero or the name "
+                  "of an existing buffer object)",
+                  caller, index, buffers[index]);
+   }
+
+   return bufObj;
+}
+
+
 /**
  * If *ptr points to obj, set ptr = the Null/default buffer object.
  * This is a helper for buffer object deletion.
@@ -2529,17 +2601,45 @@ _mesa_GetObjectParameterivAPPLE(GLenum objectType, GLuint name, GLenum pname,
    }
 }
 
+/**
+ * Binds a buffer object to a uniform buffer binding point.
+ *
+ * The caller is responsible for flushing vertices and updating
+ * NewDriverState.
+ */
 static void
 set_ubo_binding(struct gl_context *ctx,
-		int index,
-		struct gl_buffer_object *bufObj,
-		GLintptr offset,
-		GLsizeiptr size,
-		GLboolean autoSize)
+                struct gl_uniform_buffer_binding *binding,
+                struct gl_buffer_object *bufObj,
+                GLintptr offset,
+                GLsizeiptr size,
+                GLboolean autoSize)
+{
+   _mesa_reference_buffer_object(ctx, &binding->BufferObject, bufObj);
+
+   binding->Offset = offset;
+   binding->Size = size;
+   binding->AutomaticSize = autoSize;
+}
+
+/**
+ * Binds a buffer object to a uniform buffer binding point.
+ *
+ * Unlike set_ubo_binding(), this function also flushes vertices
+ * and updates NewDriverState.  It also checks if the binding
+ * has actually changed before updating it.
+ */
+static void
+bind_uniform_buffer(struct gl_context *ctx,
+                    GLuint index,
+                    struct gl_buffer_object *bufObj,
+                    GLintptr offset,
+                    GLsizeiptr size,
+                    GLboolean autoSize)
 {
-   struct gl_uniform_buffer_binding *binding;
+   struct gl_uniform_buffer_binding *binding =
+      &ctx->UniformBufferBindings[index];
 
-   binding = &ctx->UniformBufferBindings[index];
    if (binding->BufferObject == bufObj &&
        binding->Offset == offset &&
        binding->Size == size &&
@@ -2550,10 +2650,7 @@ set_ubo_binding(struct gl_context *ctx,
    FLUSH_VERTICES(ctx, 0);
    ctx->NewDriverState |= ctx->DriverFlags.NewUniformBuffer;
 
-   _mesa_reference_buffer_object(ctx, &binding->BufferObject, bufObj);
-   binding->Offset = offset;
-   binding->Size = size;
-   binding->AutomaticSize = autoSize;
+   set_ubo_binding(ctx, binding, bufObj, offset, size, autoSize);
 }
 
 /**
@@ -2588,7 +2685,7 @@ bind_buffer_range_uniform_buffer(struct gl_context *ctx,
    }
 
    _mesa_reference_buffer_object(ctx, &ctx->UniformBuffer, bufObj);
-   set_ubo_binding(ctx, index, bufObj, offset, size, GL_FALSE);
+   bind_uniform_buffer(ctx, index, bufObj, offset, size, GL_FALSE);
 }
 
 
@@ -2607,19 +2704,52 @@ bind_buffer_base_uniform_buffer(struct gl_context *ctx,
    }
 
    _mesa_reference_buffer_object(ctx, &ctx->UniformBuffer, bufObj);
+
    if (bufObj == ctx->Shared->NullBufferObj)
-      set_ubo_binding(ctx, index, bufObj, -1, -1, GL_TRUE);
+      bind_uniform_buffer(ctx, index, bufObj, -1, -1, GL_TRUE);
    else
-      set_ubo_binding(ctx, index, bufObj, 0, 0, GL_TRUE);
+      bind_uniform_buffer(ctx, index, bufObj, 0, 0, GL_TRUE);
 }
 
+/**
+ * Binds a buffer object to an atomic buffer binding point.
+ *
+ * The caller is responsible for validating the offset,
+ * flushing the vertices and updating NewDriverState.
+ */
 static void
 set_atomic_buffer_binding(struct gl_context *ctx,
-                          unsigned index,
+                          struct gl_atomic_buffer_binding *binding,
                           struct gl_buffer_object *bufObj,
                           GLintptr offset,
-                          GLsizeiptr size,
-                          const char *name)
+                          GLsizeiptr size)
+{
+   _mesa_reference_buffer_object(ctx, &binding->BufferObject, bufObj);
+
+   if (bufObj == ctx->Shared->NullBufferObj) {
+      binding->Offset = -1;
+      binding->Size = -1;
+   } else {
+      binding->Offset = offset;
+      binding->Size = size;
+   }
+}
+
+/**
+ * Binds a buffer object to an atomic buffer binding point.
+ *
+ * Unlike set_atomic_buffer_binding(), this function also validates the
+ * index and offset, flushes vertices, and updates NewDriverState.
+ * It also checks if the binding has actually changing before
+ * updating it.
+ */
+static void
+bind_atomic_buffer(struct gl_context *ctx,
+                   unsigned index,
+                   struct gl_buffer_object *bufObj,
+                   GLintptr offset,
+                   GLsizeiptr size,
+                   const char *name)
 {
    struct gl_atomic_buffer_binding *binding;
 
@@ -2647,15 +2777,704 @@ set_atomic_buffer_binding(struct gl_context *ctx,
    FLUSH_VERTICES(ctx, 0);
    ctx->NewDriverState |= ctx->DriverFlags.NewAtomicBuffer;
 
-   _mesa_reference_buffer_object(ctx, &binding->BufferObject, bufObj);
+   set_atomic_buffer_binding(ctx, binding, bufObj, offset, size);
+}
 
-   if (bufObj == ctx->Shared->NullBufferObj) {
-      binding->Offset = -1;
-      binding->Size = -1;
-   } else {
-      binding->Offset = offset;
-      binding->Size = size;
+static inline bool
+bind_buffers_check_offset_and_size(struct gl_context *ctx,
+                                   GLuint index,
+                                   const GLintptr *offsets,
+                                   const GLsizeiptr *sizes)
+{
+   if (offsets[index] < 0) {
+     /* The ARB_multi_bind spec says:
+      *
+      *    "An INVALID_VALUE error is generated by BindBuffersRange if any
+      *     value in <offsets> is less than zero (per binding)."
+      */
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glBindBuffersRange(offsets[%u]=%lld < 0)",
+                  index, (long long int) offsets[index]);
+      return false;
+   }
+
+   if (sizes[index] <= 0) {
+     /* The ARB_multi_bind spec says:
+      *
+      *     "An INVALID_VALUE error is generated by BindBuffersRange if any
+      *      value in <sizes> is less than or equal to zero (per binding)."
+      */
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glBindBuffersRange(sizes[%u]=%lld <= 0)",
+                  index, (long long int) sizes[index]);
+      return false;
+   }
+
+   return true;
+}
+
+static bool
+error_check_bind_uniform_buffers(struct gl_context *ctx,
+                                 GLuint first, GLsizei count,
+                                 const char *caller)
+{
+   if (!ctx->Extensions.ARB_uniform_buffer_object) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(target=GL_UNIFORM_BUFFER)", caller);
+      return false;
+   }
+
+   /* The ARB_multi_bind_spec says:
+    *
+    *     "An INVALID_OPERATION error is generated if <first> + <count> is
+    *      greater than the number of target-specific indexed binding points,
+    *      as described in section 6.7.1."
+    */
+   if (first + count > ctx->Const.MaxUniformBufferBindings) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(first=%u + count=%d > the value of "
+                  "GL_MAX_UNIFORM_BUFFER_BINDINGS=%u)",
+                  caller, first, count,
+                  ctx->Const.MaxUniformBufferBindings);
+      return false;
+   }
+
+   return true;
+}
+
+/**
+ * Unbind all uniform buffers in the range
+ * <first> through <first>+<count>-1
+ */
+static void
+unbind_uniform_buffers(struct gl_context *ctx, GLuint first, GLsizei count)
+{
+   struct gl_buffer_object *bufObj = ctx->Shared->NullBufferObj;
+   GLuint i;
+
+   for (i = 0; i < count; i++)
+      set_ubo_binding(ctx, &ctx->UniformBufferBindings[first + i],
+                      bufObj, -1, -1, GL_TRUE);
+}
+
+static void
+bind_uniform_buffers_base(struct gl_context *ctx, GLuint first, GLsizei count,
+                          const GLuint *buffers)
+{
+   GLuint i;
+
+   if (!error_check_bind_uniform_buffers(ctx, first, count, "glBindBuffersBase"))
+      return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewUniformBuffer;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *   "If <buffers> is NULL, all bindings from <first> through
+       *    <first>+<count>-1 are reset to their unbound (zero) state."
+       */
+      unbind_uniform_buffers(ctx, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_uniform_buffer_binding *binding =
+          &ctx->UniformBufferBindings[first + i];
+      struct gl_buffer_object *bufObj;
+
+      if (binding->BufferObject && binding->BufferObject->Name == buffers[i])
+         bufObj = binding->BufferObject;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersBase");
+
+      if (bufObj) {
+         if (bufObj == ctx->Shared->NullBufferObj)
+            set_ubo_binding(ctx, binding, bufObj, -1, -1, GL_TRUE);
+         else
+            set_ubo_binding(ctx, binding, bufObj, 0, 0, GL_TRUE);
+      }
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+static void
+bind_uniform_buffers_range(struct gl_context *ctx, GLuint first, GLsizei count,
+                           const GLuint *buffers,
+                           const GLintptr *offsets, const GLsizeiptr *sizes)
+{
+   GLuint i;
+
+   if (!error_check_bind_uniform_buffers(ctx, first, count,
+                                         "glBindBuffersRange"))
+      return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewUniformBuffer;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *    "If <buffers> is NULL, all bindings from <first> through
+       *     <first>+<count>-1 are reset to their unbound (zero) state.
+       *     In this case, the offsets and sizes associated with the
+       *     binding points are set to default values, ignoring
+       *     <offsets> and <sizes>."
+       */
+      unbind_uniform_buffers(ctx, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_uniform_buffer_binding *binding =
+         &ctx->UniformBufferBindings[first + i];
+      struct gl_buffer_object *bufObj;
+
+      if (!bind_buffers_check_offset_and_size(ctx, i, offsets, sizes))
+         continue;
+
+      /* The ARB_multi_bind spec says:
+       *
+       *     "An INVALID_VALUE error is generated by BindBuffersRange if any
+       *      pair of values in <offsets> and <sizes> does not respectively
+       *      satisfy the constraints described for those parameters for the
+       *      specified target, as described in section 6.7.1 (per binding)."
+       *
+       * Section 6.7.1 refers to table 6.5, which says:
+       *
+       *     "┌───────────────────────────────────────────────────────────────┐
+       *      │ Uniform buffer array bindings (see sec. 7.6)                  │
+       *      ├─────────────────────┬─────────────────────────────────────────┤
+       *      │  ...                │  ...                                    │
+       *      │  offset restriction │  multiple of value of UNIFORM_BUFFER_-  │
+       *      │                     │  OFFSET_ALIGNMENT                       │
+       *      │  ...                │  ...                                    │
+       *      │  size restriction   │  none                                   │
+       *      └─────────────────────┴─────────────────────────────────────────┘"
+       */
+      if (offsets[i] & (ctx->Const.UniformBufferOffsetAlignment - 1)) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindBuffersRange(offsets[%u]=%lld is misaligned; "
+                     "it must be a multiple of the value of "
+                     "GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT=%u when "
+                     "target=GL_UNIFORM_BUFFER)",
+                     i, (long long int) offsets[i],
+                     ctx->Const.UniformBufferOffsetAlignment);
+         continue;
+      }
+
+      if (binding->BufferObject && binding->BufferObject->Name == buffers[i])
+         bufObj = binding->BufferObject;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersRange");
+
+      if (bufObj) {
+         if (bufObj == ctx->Shared->NullBufferObj)
+            set_ubo_binding(ctx, binding, bufObj, -1, -1, GL_FALSE);
+         else
+            set_ubo_binding(ctx, binding, bufObj,
+                            offsets[i], sizes[i], GL_FALSE);
+      }
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+static bool
+error_check_bind_xfb_buffers(struct gl_context *ctx,
+                             struct gl_transform_feedback_object *tfObj,
+                             GLuint first, GLsizei count, const char *caller)
+{
+   if (!ctx->Extensions.EXT_transform_feedback) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(target=GL_TRANSFORM_FEEDBACK_BUFFER)", caller);
+      return false;
    }
+
+   /* Page 398 of the PDF of the OpenGL 4.4 (Core Profile) spec says:
+    *
+    *     "An INVALID_OPERATION error is generated :
+    *
+    *     ...
+    *     • by BindBufferRange or BindBufferBase if target is TRANSFORM_-
+    *       FEEDBACK_BUFFER and transform feedback is currently active."
+    *
+    * We assume that this is also meant to apply to BindBuffersRange
+    * and BindBuffersBase.
+    */
+   if (tfObj->Active) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(Changing transform feedback buffers while "
+                  "transform feedback is active)", caller);
+      return false;
+   }
+
+   /* The ARB_multi_bind_spec says:
+    *
+    *     "An INVALID_OPERATION error is generated if <first> + <count> is
+    *      greater than the number of target-specific indexed binding points,
+    *      as described in section 6.7.1."
+    */
+   if (first + count > ctx->Const.MaxTransformFeedbackBuffers) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(first=%u + count=%d > the value of "
+                  "GL_MAX_TRANSFORM_FEEDBACK_BUFFERS=%u)",
+                  caller, first, count,
+                  ctx->Const.MaxTransformFeedbackBuffers);
+      return false;
+   }
+
+   return true;
+}
+
+/**
+ * Unbind all transform feedback buffers in the range
+ * <first> through <first>+<count>-1
+ */
+static void
+unbind_xfb_buffers(struct gl_context *ctx,
+                   struct gl_transform_feedback_object *tfObj,
+                   GLuint first, GLsizei count)
+{
+   struct gl_buffer_object * const bufObj = ctx->Shared->NullBufferObj;
+   GLuint i;
+
+   for (i = 0; i < count; i++)
+      _mesa_set_transform_feedback_binding(ctx, tfObj, first + i,
+                                           bufObj, 0, 0);
+}
+
+static void
+bind_xfb_buffers_base(struct gl_context *ctx,
+                      GLuint first, GLsizei count,
+                      const GLuint *buffers)
+{
+   struct gl_transform_feedback_object *tfObj =
+      ctx->TransformFeedback.CurrentObject;
+   GLuint i;
+
+   if (!error_check_bind_xfb_buffers(ctx, tfObj, first, count,
+                                     "glBindBuffersBase"))
+      return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewTransformFeedback;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *   "If <buffers> is NULL, all bindings from <first> through
+       *    <first>+<count>-1 are reset to their unbound (zero) state."
+       */
+      unbind_xfb_buffers(ctx, tfObj, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_buffer_object * const boundBufObj = tfObj->Buffers[first + i];
+      struct gl_buffer_object *bufObj;
+
+      if (boundBufObj && boundBufObj->Name == buffers[i])
+         bufObj = boundBufObj;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersBase");
+
+      if (bufObj)
+         _mesa_set_transform_feedback_binding(ctx, tfObj, first + i,
+                                              bufObj, 0, 0);
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+static void
+bind_xfb_buffers_range(struct gl_context *ctx,
+                       GLuint first, GLsizei count,
+                       const GLuint *buffers,
+                       const GLintptr *offsets,
+                       const GLsizeiptr *sizes)
+{
+   struct gl_transform_feedback_object *tfObj =
+       ctx->TransformFeedback.CurrentObject;
+   GLuint i;
+
+   if (!error_check_bind_xfb_buffers(ctx, tfObj, first, count,
+                                     "glBindBuffersRange"))
+      return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewTransformFeedback;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *    "If <buffers> is NULL, all bindings from <first> through
+       *     <first>+<count>-1 are reset to their unbound (zero) state.
+       *     In this case, the offsets and sizes associated with the
+       *     binding points are set to default values, ignoring
+       *     <offsets> and <sizes>."
+       */
+      unbind_xfb_buffers(ctx, tfObj, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      const GLuint index = first + i;
+      struct gl_buffer_object * const boundBufObj = tfObj->Buffers[index];
+      struct gl_buffer_object *bufObj;
+
+      if (!bind_buffers_check_offset_and_size(ctx, i, offsets, sizes))
+         continue;
+
+      /* The ARB_multi_bind spec says:
+       *
+       *     "An INVALID_VALUE error is generated by BindBuffersRange if any
+       *      pair of values in <offsets> and <sizes> does not respectively
+       *      satisfy the constraints described for those parameters for the
+       *      specified target, as described in section 6.7.1 (per binding)."
+       *
+       * Section 6.7.1 refers to table 6.5, which says:
+       *
+       *     "┌───────────────────────────────────────────────────────────────┐
+       *      │ Transform feedback array bindings (see sec. 13.2.2)           │
+       *      ├───────────────────────┬───────────────────────────────────────┤
+       *      │    ...                │    ...                                │
+       *      │    offset restriction │    multiple of 4                      │
+       *      │    ...                │    ...                                │
+       *      │    size restriction   │    multiple of 4                      │
+       *      └───────────────────────┴───────────────────────────────────────┘"
+       */
+      if (offsets[i] & 0x3) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindBuffersRange(offsets[%u]=%lld is misaligned; "
+                     "it must be a multiple of 4 when "
+                     "target=GL_TRANSFORM_FEEDBACK_BUFFER)",
+                     i, (long long int) offsets[i]);
+         continue;
+      }
+
+      if (sizes[i] & 0x3) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindBuffersRange(sizes[%u]=%lld is misaligned; "
+                     "it must be a multiple of 4 when "
+                     "target=GL_TRANSFORM_FEEDBACK_BUFFER)",
+                     i, (long long int) sizes[i]);
+         continue;
+      }
+
+      if (boundBufObj && boundBufObj->Name == buffers[i])
+         bufObj = boundBufObj;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersRange");
+
+      if (bufObj)
+         _mesa_set_transform_feedback_binding(ctx, tfObj, index, bufObj,
+                                              offsets[i], sizes[i]);
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+static bool
+error_check_bind_atomic_buffers(struct gl_context *ctx,
+                                GLuint first, GLsizei count,
+                                const char *caller)
+{
+   if (!ctx->Extensions.ARB_shader_atomic_counters) {
+      _mesa_error(ctx, GL_INVALID_ENUM,
+                  "%s(target=GL_ATOMIC_COUNTER_BUFFER)", caller);
+      return false;
+   }
+
+   /* The ARB_multi_bind_spec says:
+    *
+    *     "An INVALID_OPERATION error is generated if <first> + <count> is
+    *      greater than the number of target-specific indexed binding points,
+    *      as described in section 6.7.1."
+    */
+   if (first + count > ctx->Const.MaxAtomicBufferBindings) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(first=%u + count=%d > the value of "
+                  "GL_MAX_ATOMIC_BUFFER_BINDINGS=%u)",
+                  caller, first, count, ctx->Const.MaxAtomicBufferBindings);
+      return false;
+   }
+
+   return true;
+}
+
+/**
+ * Unbind all atomic counter buffers in the range
+ * <first> through <first>+<count>-1
+ */
+static void
+unbind_atomic_buffers(struct gl_context *ctx, GLuint first, GLsizei count)
+{
+   struct gl_buffer_object * const bufObj = ctx->Shared->NullBufferObj;
+   GLuint i;
+
+   for (i = 0; i < count; i++)
+      set_atomic_buffer_binding(ctx, &ctx->AtomicBufferBindings[first + i],
+                                bufObj, -1, -1);
+}
+
+static void
+bind_atomic_buffers_base(struct gl_context *ctx,
+                         GLuint first,
+                         GLsizei count,
+                         const GLuint *buffers)
+{
+   GLuint i;
+
+   if (!error_check_bind_atomic_buffers(ctx, first, count,
+                                        "glBindBuffersBase"))
+     return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewAtomicBuffer;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *   "If <buffers> is NULL, all bindings from <first> through
+       *    <first>+<count>-1 are reset to their unbound (zero) state."
+       */
+      unbind_atomic_buffers(ctx, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_atomic_buffer_binding *binding =
+         &ctx->AtomicBufferBindings[first + i];
+      struct gl_buffer_object *bufObj;
+
+      if (binding->BufferObject && binding->BufferObject->Name == buffers[i])
+         bufObj = binding->BufferObject;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersBase");
+
+      if (bufObj)
+         set_atomic_buffer_binding(ctx, binding, bufObj, 0, 0);
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+static void
+bind_atomic_buffers_range(struct gl_context *ctx,
+                          GLuint first,
+                          GLsizei count,
+                          const GLuint *buffers,
+                          const GLintptr *offsets,
+                          const GLsizeiptr *sizes)
+{
+   GLuint i;
+
+   if (!error_check_bind_atomic_buffers(ctx, first, count,
+                                        "glBindBuffersRange"))
+     return;
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewAtomicBuffer;
+
+   if (!buffers) {
+      /* The ARB_multi_bind spec says:
+       *
+       *    "If <buffers> is NULL, all bindings from <first> through
+       *     <first>+<count>-1 are reset to their unbound (zero) state.
+       *     In this case, the offsets and sizes associated with the
+       *     binding points are set to default values, ignoring
+       *     <offsets> and <sizes>."
+       */
+      unbind_atomic_buffers(ctx, first, count);
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by a
+    *          command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require a
+    *          first pass to scan the entire list of bound objects for errors
+    *          and then a second pass to actually perform the bindings.
+    *          Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding point
+    *       is not updated and an error will be generated.  However, other
+    *       binding points in the same command will be updated if their
+    *       parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_atomic_buffer_binding *binding =
+         &ctx->AtomicBufferBindings[first + i];
+      struct gl_buffer_object *bufObj;
+
+      if (!bind_buffers_check_offset_and_size(ctx, i, offsets, sizes))
+         continue;
+
+      /* The ARB_multi_bind spec says:
+       *
+       *     "An INVALID_VALUE error is generated by BindBuffersRange if any
+       *      pair of values in <offsets> and <sizes> does not respectively
+       *      satisfy the constraints described for those parameters for the
+       *      specified target, as described in section 6.7.1 (per binding)."
+       *
+       * Section 6.7.1 refers to table 6.5, which says:
+       *
+       *     "┌───────────────────────────────────────────────────────────────┐
+       *      │ Atomic counter array bindings (see sec. 7.7.2)                │
+       *      ├───────────────────────┬───────────────────────────────────────┤
+       *      │    ...                │    ...                                │
+       *      │    offset restriction │    multiple of 4                      │
+       *      │    ...                │    ...                                │
+       *      │    size restriction   │    none                               │
+       *      └───────────────────────┴───────────────────────────────────────┘"
+       */
+      if (offsets[i] & (ATOMIC_COUNTER_SIZE - 1)) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindBuffersRange(offsets[%u]=%lld is misaligned; "
+                     "it must be a multiple of %d when "
+                     "target=GL_ATOMIC_COUNTER_BUFFER)",
+                     i, (long long int) offsets[i], ATOMIC_COUNTER_SIZE);
+         continue;
+      }
+
+      if (binding->BufferObject && binding->BufferObject->Name == buffers[i])
+         bufObj = binding->BufferObject;
+      else
+         bufObj = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindBuffersRange");
+
+      if (bufObj)
+         set_atomic_buffer_binding(ctx, binding, bufObj, offsets[i], sizes[i]);
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
 }
 
 void GLAPIENTRY
@@ -2697,8 +3516,8 @@ _mesa_BindBufferRange(GLenum target, GLuint index,
       bind_buffer_range_uniform_buffer(ctx, index, bufObj, offset, size);
       return;
    case GL_ATOMIC_COUNTER_BUFFER:
-      set_atomic_buffer_binding(ctx, index, bufObj, offset, size,
-                                "glBindBufferRange");
+      bind_atomic_buffer(ctx, index, bufObj, offset, size,
+                         "glBindBufferRange");
       return;
    default:
       _mesa_error(ctx, GL_INVALID_ENUM, "glBindBufferRange(target)");
@@ -2761,8 +3580,8 @@ _mesa_BindBufferBase(GLenum target, GLuint index, GLuint buffer)
       bind_buffer_base_uniform_buffer(ctx, index, bufObj);
       return;
    case GL_ATOMIC_COUNTER_BUFFER:
-      set_atomic_buffer_binding(ctx, index, bufObj, 0, 0,
-                                "glBindBufferBase");
+      bind_atomic_buffer(ctx, index, bufObj, 0, 0,
+                         "glBindBufferBase");
       return;
    default:
       _mesa_error(ctx, GL_INVALID_ENUM, "glBindBufferBase(target)");
@@ -2771,6 +3590,54 @@ _mesa_BindBufferBase(GLenum target, GLuint index, GLuint buffer)
 }
 
 void GLAPIENTRY
+_mesa_BindBuffersRange(GLenum target, GLuint first, GLsizei count,
+                       const GLuint *buffers,
+                       const GLintptr *offsets, const GLsizeiptr *sizes)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   switch (target) {
+   case GL_TRANSFORM_FEEDBACK_BUFFER:
+      bind_xfb_buffers_range(ctx, first, count, buffers, offsets, sizes);
+      return;
+   case GL_UNIFORM_BUFFER:
+      bind_uniform_buffers_range(ctx, first, count, buffers, offsets, sizes);
+      return;
+   case GL_ATOMIC_COUNTER_BUFFER:
+      bind_atomic_buffers_range(ctx, first, count, buffers,
+                                offsets, sizes);
+      return;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glBindBuffersRange(target=%s)",
+                  _mesa_lookup_enum_by_nr(target));
+      break;
+   }
+}
+
+void GLAPIENTRY
+_mesa_BindBuffersBase(GLenum target, GLuint first, GLsizei count,
+                      const GLuint *buffers)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   switch (target) {
+   case GL_TRANSFORM_FEEDBACK_BUFFER:
+      bind_xfb_buffers_base(ctx, first, count, buffers);
+      return;
+   case GL_UNIFORM_BUFFER:
+      bind_uniform_buffers_base(ctx, first, count, buffers);
+      return;
+   case GL_ATOMIC_COUNTER_BUFFER:
+      bind_atomic_buffers_base(ctx, first, count, buffers);
+      return;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glBindBuffersBase(target=%s)",
+                  _mesa_lookup_enum_by_nr(target));
+      break;
+   }
+}
+
+void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
                               GLsizeiptr length)
 {
diff --git a/mesalib/src/mesa/main/bufferobj.h b/mesalib/src/mesa/main/bufferobj.h
index c08c4fdf2..0779605c0 100644
--- a/mesalib/src/mesa/main/bufferobj.h
+++ b/mesalib/src/mesa/main/bufferobj.h
@@ -86,6 +86,20 @@ _mesa_update_default_objects_buffer_objects(struct gl_context *ctx);
 extern struct gl_buffer_object *
 _mesa_lookup_bufferobj(struct gl_context *ctx, GLuint buffer);
 
+extern struct gl_buffer_object *
+_mesa_lookup_bufferobj_locked(struct gl_context *ctx, GLuint buffer);
+
+extern void
+_mesa_begin_bufferobj_lookups(struct gl_context *ctx);
+
+extern void
+_mesa_end_bufferobj_lookups(struct gl_context *ctx);
+
+extern struct gl_buffer_object *
+_mesa_multi_bind_lookup_bufferobj(struct gl_context *ctx,
+                                  const GLuint *buffers,
+                                  GLuint index, const char *caller);
+
 extern void
 _mesa_initialize_buffer_object(struct gl_context *ctx,
                                struct gl_buffer_object *obj,
@@ -210,6 +224,13 @@ void GLAPIENTRY
 _mesa_BindBufferBase(GLenum target, GLuint index, GLuint buffer);
 
 void GLAPIENTRY
+_mesa_BindBuffersRange(GLenum target, GLuint first, GLsizei count,
+                       const GLuint *buffers,
+                       const GLintptr *offsets, const GLsizeiptr *sizes);
+void GLAPIENTRY
+_mesa_BindBuffersBase(GLenum target, GLuint first, GLsizei count,
+                      const GLuint *buffers);
+void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
                               GLsizeiptr length);
 
diff --git a/mesalib/src/mesa/main/config.h b/mesalib/src/mesa/main/config.h
index b2583e6d5..04175991e 100644
--- a/mesalib/src/mesa/main/config.h
+++ b/mesalib/src/mesa/main/config.h
@@ -285,6 +285,14 @@
 #define MAX_VERTEX_STREAMS                  4
 /*@}*/
 
+/** For GL_INTEL_performance_query */
+/*@{*/
+#define MAX_PERFQUERY_QUERY_NAME_LENGTH     256
+#define MAX_PERFQUERY_COUNTER_NAME_LENGTH   256
+#define MAX_PERFQUERY_COUNTER_DESC_LENGTH   1024
+#define PERFQUERY_HAVE_GPA_EXTENDED_COUNTERS 0
+/*@}*/
+
 /*
  * Color channel component order
  * 
diff --git a/mesalib/src/mesa/main/dd.h b/mesalib/src/mesa/main/dd.h
index 971524135..633ea2c3a 100644
--- a/mesalib/src/mesa/main/dd.h
+++ b/mesalib/src/mesa/main/dd.h
@@ -314,10 +314,10 @@ struct dd_function_table {
    /*@{*/
 
    /**
-    * Called by glBindTexture().
+    * Called by glBindTexture() and glBindTextures().
     */
-   void (*BindTexture)( struct gl_context *ctx, GLenum target,
-                        struct gl_texture_object *tObj );
+   void (*BindTexture)( struct gl_context *ctx, GLuint texUnit,
+                        GLenum target, struct gl_texture_object *tObj );
 
    /**
     * Called to allocate a new texture object.  Drivers will usually
diff --git a/mesalib/src/mesa/main/dlist.c b/mesalib/src/mesa/main/dlist.c
index d431fd221..5874b99f0 100644
--- a/mesalib/src/mesa/main/dlist.c
+++ b/mesalib/src/mesa/main/dlist.c
@@ -364,6 +364,42 @@ typedef enum
    OPCODE_UNIFORM_3UIV,
    OPCODE_UNIFORM_4UIV,
 
+   /* OpenGL 4.2 / GL_ARB_separate_shader_objects */
+   OPCODE_USE_PROGRAM_STAGES,
+   OPCODE_PROGRAM_UNIFORM_1F,
+   OPCODE_PROGRAM_UNIFORM_2F,
+   OPCODE_PROGRAM_UNIFORM_3F,
+   OPCODE_PROGRAM_UNIFORM_4F,
+   OPCODE_PROGRAM_UNIFORM_1FV,
+   OPCODE_PROGRAM_UNIFORM_2FV,
+   OPCODE_PROGRAM_UNIFORM_3FV,
+   OPCODE_PROGRAM_UNIFORM_4FV,
+   OPCODE_PROGRAM_UNIFORM_1I,
+   OPCODE_PROGRAM_UNIFORM_2I,
+   OPCODE_PROGRAM_UNIFORM_3I,
+   OPCODE_PROGRAM_UNIFORM_4I,
+   OPCODE_PROGRAM_UNIFORM_1IV,
+   OPCODE_PROGRAM_UNIFORM_2IV,
+   OPCODE_PROGRAM_UNIFORM_3IV,
+   OPCODE_PROGRAM_UNIFORM_4IV,
+   OPCODE_PROGRAM_UNIFORM_1UI,
+   OPCODE_PROGRAM_UNIFORM_2UI,
+   OPCODE_PROGRAM_UNIFORM_3UI,
+   OPCODE_PROGRAM_UNIFORM_4UI,
+   OPCODE_PROGRAM_UNIFORM_1UIV,
+   OPCODE_PROGRAM_UNIFORM_2UIV,
+   OPCODE_PROGRAM_UNIFORM_3UIV,
+   OPCODE_PROGRAM_UNIFORM_4UIV,
+   OPCODE_PROGRAM_UNIFORM_MATRIX22F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX33F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX44F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX23F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX32F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX24F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX42F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX34F,
+   OPCODE_PROGRAM_UNIFORM_MATRIX43F,
+
    /* GL_ARB_color_buffer_float */
    OPCODE_CLAMP_COLOR,
 
@@ -407,10 +443,6 @@ typedef enum
    OPCODE_TEXPARAMETER_I,
    OPCODE_TEXPARAMETER_UI,
 
-   /* GL_EXT_separate_shader_objects */
-   OPCODE_ACTIVE_PROGRAM_EXT,
-   OPCODE_USE_SHADER_PROGRAM_EXT,
-
    /* GL_ARB_instanced_arrays */
    OPCODE_VERTEX_ATTRIB_DIVISOR,
 
@@ -765,6 +797,33 @@ _mesa_delete_list(struct gl_context *ctx, struct gl_display_list *dlist)
             free(get_pointer(&n[4]));
             n += InstSize[n[0].opcode];
             break;
+         case OPCODE_PROGRAM_UNIFORM_1FV:
+         case OPCODE_PROGRAM_UNIFORM_2FV:
+         case OPCODE_PROGRAM_UNIFORM_3FV:
+         case OPCODE_PROGRAM_UNIFORM_4FV:
+         case OPCODE_PROGRAM_UNIFORM_1IV:
+         case OPCODE_PROGRAM_UNIFORM_2IV:
+         case OPCODE_PROGRAM_UNIFORM_3IV:
+         case OPCODE_PROGRAM_UNIFORM_4IV:
+         case OPCODE_PROGRAM_UNIFORM_1UIV:
+         case OPCODE_PROGRAM_UNIFORM_2UIV:
+         case OPCODE_PROGRAM_UNIFORM_3UIV:
+         case OPCODE_PROGRAM_UNIFORM_4UIV:
+            free(get_pointer(&n[4]));
+            n += InstSize[n[0].opcode];
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX22F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX33F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX44F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX24F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX42F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX23F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX32F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX34F:
+         case OPCODE_PROGRAM_UNIFORM_MATRIX43F:
+            free(get_pointer(&n[5]));
+            n += InstSize[n[0].opcode];
+            break;
          case OPCODE_PIXEL_MAP:
             free(get_pointer(&n[3]));
             n += InstSize[n[0].opcode];
@@ -6510,49 +6569,689 @@ save_UniformMatrix4x3fv(GLint location, GLsizei count, GLboolean transpose,
 }
 
 static void GLAPIENTRY
-save_ClampColorARB(GLenum target, GLenum clamp)
+save_UseProgramStages(GLuint pipeline, GLbitfield stages, GLuint program)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_CLAMP_COLOR, 2);
+   n = alloc_instruction(ctx, OPCODE_USE_PROGRAM_STAGES, 3);
    if (n) {
-      n[1].e = target;
-      n[2].e = clamp;
+      n[1].ui = pipeline;
+      n[2].ui = stages;
+      n[3].ui = program;
    }
    if (ctx->ExecuteFlag) {
-      CALL_ClampColor(ctx->Exec, (target, clamp));
+      CALL_UseProgramStages(ctx->Exec, (pipeline, stages, program));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform1f(GLuint program, GLint location, GLfloat x)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_1F, 3);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].f = x;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform1f(ctx->Exec, (program, location, x));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2f(GLuint program, GLint location, GLfloat x, GLfloat y)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_2F, 4);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].f = x;
+      n[4].f = y;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2f(ctx->Exec, (program, location, x, y));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3f(GLuint program, GLint location,
+                      GLfloat x, GLfloat y, GLfloat z)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_3F, 5);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].f = x;
+      n[4].f = y;
+      n[5].f = z;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3f(ctx->Exec, (program, location, x, y, z));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform4f(GLuint program, GLint location,
+                      GLfloat x, GLfloat y, GLfloat z, GLfloat w)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_4F, 6);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].f = x;
+      n[4].f = y;
+      n[5].f = z;
+      n[6].f = w;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform4f(ctx->Exec, (program, location, x, y, z, w));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform1fv(GLuint program, 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_PROGRAM_UNIFORM_1FV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 1 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform1fv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2fv(GLuint program, 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_PROGRAM_UNIFORM_2FV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 2 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2fv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3fv(GLuint program, 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_PROGRAM_UNIFORM_3FV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 3 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3fv(ctx->Exec, (program, location, count, v));
    }
 }
 
 static void GLAPIENTRY
-save_UseShaderProgramEXT(GLenum type, GLuint program)
+save_ProgramUniform4fv(GLuint program, 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_USE_SHADER_PROGRAM_EXT, 2);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_4FV, 3 + POINTER_DWORDS);
    if (n) {
-      n[1].ui = type;
-      n[2].ui = program;
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 4 * sizeof(GLfloat)));
    }
    if (ctx->ExecuteFlag) {
-      CALL_UseShaderProgramEXT(ctx->Exec, (type, program));
+      CALL_ProgramUniform4fv(ctx->Exec, (program, location, count, v));
    }
 }
 
 static void GLAPIENTRY
-save_ActiveProgramEXT(GLuint program)
+save_ProgramUniform1i(GLuint program, GLint location, GLint x)
 {
    GET_CURRENT_CONTEXT(ctx);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
-   n = alloc_instruction(ctx, OPCODE_ACTIVE_PROGRAM_EXT, 1);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_1I, 3);
    if (n) {
       n[1].ui = program;
+      n[2].i = location;
+      n[3].i = x;
    }
    if (ctx->ExecuteFlag) {
-      CALL_ActiveProgramEXT(ctx->Exec, (program));
+      CALL_ProgramUniform1i(ctx->Exec, (program, location, x));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2i(GLuint program, GLint location, GLint x, GLint y)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_2I, 4);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = x;
+      n[4].i = y;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2i(ctx->Exec, (program, location, x, y));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3i(GLuint program, GLint location,
+                      GLint x, GLint y, GLint z)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_3I, 5);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = x;
+      n[4].i = y;
+      n[5].i = z;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3i(ctx->Exec, (program, location, x, y, z));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform4i(GLuint program, GLint location,
+                      GLint x, GLint y, GLint z, GLint w)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_4I, 6);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = x;
+      n[4].i = y;
+      n[5].i = z;
+      n[6].i = w;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform4i(ctx->Exec, (program, location, x, y, z, w));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform1iv(GLuint program, 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_PROGRAM_UNIFORM_1IV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 1 * sizeof(GLint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform1iv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2iv(GLuint program, 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_PROGRAM_UNIFORM_2IV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 2 * sizeof(GLint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2iv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3iv(GLuint program, 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_PROGRAM_UNIFORM_3IV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 3 * sizeof(GLint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3iv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform4iv(GLuint program, 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_PROGRAM_UNIFORM_4IV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 4 * sizeof(GLint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform4iv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform1ui(GLuint program, GLint location, GLuint x)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_1UI, 3);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].ui = x;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform1ui(ctx->Exec, (program, location, x));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2ui(GLuint program, GLint location, GLuint x, GLuint y)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_2UI, 4);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].ui = x;
+      n[4].ui = y;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2ui(ctx->Exec, (program, location, x, y));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3ui(GLuint program, GLint location,
+                       GLuint x, GLuint y, GLuint z)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_3UI, 5);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].ui = x;
+      n[4].ui = y;
+      n[5].ui = z;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3ui(ctx->Exec, (program, location, x, y, z));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform4ui(GLuint program, GLint location,
+                       GLuint x, GLuint y, GLuint z, GLuint w)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_4UI, 6);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].ui = x;
+      n[4].ui = y;
+      n[5].ui = z;
+      n[6].ui = w;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform4ui(ctx->Exec, (program, location, x, y, z, w));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform1uiv(GLuint program, 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_PROGRAM_UNIFORM_1UIV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 1 * sizeof(GLuint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform1uiv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform2uiv(GLuint program, 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_PROGRAM_UNIFORM_2UIV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 2 * sizeof(GLuint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform2uiv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform3uiv(GLuint program, 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_PROGRAM_UNIFORM_3UIV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 3 * sizeof(GLuint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform3uiv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniform4uiv(GLuint program, 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_PROGRAM_UNIFORM_4UIV, 3 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      save_pointer(&n[4], memdup(v, count * 4 * sizeof(GLuint)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniform4uiv(ctx->Exec, (program, location, count, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix2fv(GLuint program, GLint location, GLsizei count,
+                             GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX22F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 2 * 2 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix2fv(ctx->Exec,
+                                   (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix2x3fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX23F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 2 * 3 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix2x3fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix2x4fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX24F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 2 * 4 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix2x4fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix3x2fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX32F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 3 * 2 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix3x2fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix3fv(GLuint program, GLint location, GLsizei count,
+                             GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX33F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 3 * 3 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix3fv(ctx->Exec,
+                                   (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix3x4fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX34F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 3 * 4 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix3x4fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix4x2fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX42F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 4 * 2 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix4x2fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix4x3fv(GLuint program, GLint location, GLsizei count,
+                               GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX43F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 4 * 3 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix4x3fv(ctx->Exec,
+                                     (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ProgramUniformMatrix4fv(GLuint program, GLint location, GLsizei count,
+                             GLboolean transpose, const GLfloat *v)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_PROGRAM_UNIFORM_MATRIX44F,
+                         4 + POINTER_DWORDS);
+   if (n) {
+      n[1].ui = program;
+      n[2].i = location;
+      n[3].i = count;
+      n[4].b = transpose;
+      save_pointer(&n[5], memdup(v, count * 4 * 4 * sizeof(GLfloat)));
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ProgramUniformMatrix4fv(ctx->Exec,
+                                   (program, location, count, transpose, v));
+   }
+}
+
+static void GLAPIENTRY
+save_ClampColorARB(GLenum target, GLenum clamp)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   n = alloc_instruction(ctx, OPCODE_CLAMP_COLOR, 2);
+   if (n) {
+      n[1].e = target;
+      n[2].e = clamp;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_ClampColor(ctx->Exec, (target, clamp));
    }
 }
 
@@ -7692,12 +8391,6 @@ execute_list(struct gl_context *ctx, GLuint list)
 	 case OPCODE_USE_PROGRAM:
 	    CALL_UseProgram(ctx->Exec, (n[1].ui));
 	    break;
-	 case OPCODE_USE_SHADER_PROGRAM_EXT:
-	    CALL_UseShaderProgramEXT(ctx->Exec, (n[1].ui, n[2].ui));
-	    break;
-	 case OPCODE_ACTIVE_PROGRAM_EXT:
-	    CALL_ActiveProgramEXT(ctx->Exec, (n[1].ui));
-	    break;
 	 case OPCODE_UNIFORM_1F:
 	    CALL_Uniform1f(ctx->Exec, (n[1].i, n[2].f));
 	    break;
@@ -7815,6 +8508,147 @@ execute_list(struct gl_context *ctx, GLuint list)
                                     (n[1].i, n[2].i, n[3].b, get_pointer(&n[4])));
 	    break;
 
+	 case OPCODE_USE_PROGRAM_STAGES:
+	    CALL_UseProgramStages(ctx->Exec, (n[1].ui, n[2].ui, n[3].ui));
+	    break;
+         case OPCODE_PROGRAM_UNIFORM_1F:
+            CALL_ProgramUniform1f(ctx->Exec, (n[1].ui, n[2].i, n[3].f));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2F:
+            CALL_ProgramUniform2f(ctx->Exec, (n[1].ui, n[2].i, n[3].f, n[4].f));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3F:
+            CALL_ProgramUniform3f(ctx->Exec, (n[1].ui, n[2].i,
+                                              n[3].f, n[4].f, n[5].f));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4F:
+            CALL_ProgramUniform4f(ctx->Exec, (n[1].ui, n[2].i,
+                                              n[3].f, n[4].f, n[5].f, n[6].f));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_1FV:
+            CALL_ProgramUniform1fv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2FV:
+            CALL_ProgramUniform2fv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3FV:
+            CALL_ProgramUniform3fv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4FV:
+            CALL_ProgramUniform4fv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_1I:
+            CALL_ProgramUniform1i(ctx->Exec, (n[1].ui, n[2].i, n[3].i));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2I:
+            CALL_ProgramUniform2i(ctx->Exec, (n[1].ui, n[2].i, n[3].i, n[4].i));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3I:
+            CALL_ProgramUniform3i(ctx->Exec, (n[1].ui, n[2].i,
+                                              n[3].i, n[4].i, n[5].i));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4I:
+            CALL_ProgramUniform4i(ctx->Exec, (n[1].ui, n[2].i,
+                                              n[3].i, n[4].i, n[5].i, n[6].i));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_1IV:
+            CALL_ProgramUniform1iv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2IV:
+            CALL_ProgramUniform2iv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3IV:
+            CALL_ProgramUniform3iv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4IV:
+            CALL_ProgramUniform4iv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                               get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_1UI:
+            CALL_ProgramUniform1ui(ctx->Exec, (n[1].ui, n[2].i, n[3].ui));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2UI:
+            CALL_ProgramUniform2ui(ctx->Exec, (n[1].ui, n[2].i,
+                                               n[3].ui, n[4].ui));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3UI:
+            CALL_ProgramUniform3ui(ctx->Exec, (n[1].ui, n[2].i,
+                                               n[3].ui, n[4].ui, n[5].ui));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4UI:
+            CALL_ProgramUniform4ui(ctx->Exec, (n[1].ui, n[2].i,
+                                               n[3].ui,
+                                               n[4].ui, n[5].ui, n[6].ui));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_1UIV:
+            CALL_ProgramUniform1uiv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                                get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_2UIV:
+            CALL_ProgramUniform2uiv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                                get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_3UIV:
+            CALL_ProgramUniform3uiv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                                get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_4UIV:
+            CALL_ProgramUniform4uiv(ctx->Exec, (n[1].ui, n[2].i, n[3].i,
+                                                get_pointer(&n[4])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX22F:
+            CALL_ProgramUniformMatrix2fv(ctx->Exec,
+                                         (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                          get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX23F:
+            CALL_ProgramUniformMatrix2x3fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX24F:
+            CALL_ProgramUniformMatrix2x4fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX32F:
+            CALL_ProgramUniformMatrix3x2fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX33F:
+            CALL_ProgramUniformMatrix3fv(ctx->Exec,
+                                         (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                          get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX34F:
+            CALL_ProgramUniformMatrix3x4fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX42F:
+            CALL_ProgramUniformMatrix4x2fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX43F:
+            CALL_ProgramUniformMatrix4x3fv(ctx->Exec,
+                                           (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                            get_pointer(&n[5])));
+            break;
+         case OPCODE_PROGRAM_UNIFORM_MATRIX44F:
+            CALL_ProgramUniformMatrix4fv(ctx->Exec,
+                                         (n[1].ui, n[2].i, n[3].i, n[4].b,
+                                          get_pointer(&n[5])));
+            break;
+
          case OPCODE_CLAMP_COLOR:
             CALL_ClampColor(ctx->Exec, (n[1].e, n[2].e));
             break;
@@ -8763,10 +9597,6 @@ _mesa_initialize_save_table(const struct gl_context *ctx)
    SET_TexParameterIiv(table, save_TexParameterIiv);
    SET_TexParameterIuiv(table, save_TexParameterIuiv);
 
-   /* 377. GL_EXT_separate_shader_objects */
-   SET_UseShaderProgramEXT(table, save_UseShaderProgramEXT);
-   SET_ActiveProgramEXT(table, save_ActiveProgramEXT);
-
    /* GL_ARB_color_buffer_float */
    SET_ClampColor(table, save_ClampColorARB);
 
@@ -8856,6 +9686,42 @@ _mesa_initialize_save_table(const struct gl_context *ctx)
    SET_DrawArraysInstancedBaseInstance(table, save_DrawArraysInstancedBaseInstance);
    SET_DrawElementsInstancedBaseInstance(table, save_DrawElementsInstancedBaseInstance);
    SET_DrawElementsInstancedBaseVertexBaseInstance(table, save_DrawElementsInstancedBaseVertexBaseInstance);
+
+   /* OpenGL 4.2 / GL_ARB_separate_shader_objects */
+   SET_UseProgramStages(table, save_UseProgramStages);
+   SET_ProgramUniform1f(table, save_ProgramUniform1f);
+   SET_ProgramUniform2f(table, save_ProgramUniform2f);
+   SET_ProgramUniform3f(table, save_ProgramUniform3f);
+   SET_ProgramUniform4f(table, save_ProgramUniform4f);
+   SET_ProgramUniform1fv(table, save_ProgramUniform1fv);
+   SET_ProgramUniform2fv(table, save_ProgramUniform2fv);
+   SET_ProgramUniform3fv(table, save_ProgramUniform3fv);
+   SET_ProgramUniform4fv(table, save_ProgramUniform4fv);
+   SET_ProgramUniform1i(table, save_ProgramUniform1i);
+   SET_ProgramUniform2i(table, save_ProgramUniform2i);
+   SET_ProgramUniform3i(table, save_ProgramUniform3i);
+   SET_ProgramUniform4i(table, save_ProgramUniform4i);
+   SET_ProgramUniform1iv(table, save_ProgramUniform1iv);
+   SET_ProgramUniform2iv(table, save_ProgramUniform2iv);
+   SET_ProgramUniform3iv(table, save_ProgramUniform3iv);
+   SET_ProgramUniform4iv(table, save_ProgramUniform4iv);
+   SET_ProgramUniform1ui(table, save_ProgramUniform1ui);
+   SET_ProgramUniform2ui(table, save_ProgramUniform2ui);
+   SET_ProgramUniform3ui(table, save_ProgramUniform3ui);
+   SET_ProgramUniform4ui(table, save_ProgramUniform4ui);
+   SET_ProgramUniform1uiv(table, save_ProgramUniform1uiv);
+   SET_ProgramUniform2uiv(table, save_ProgramUniform2uiv);
+   SET_ProgramUniform3uiv(table, save_ProgramUniform3uiv);
+   SET_ProgramUniform4uiv(table, save_ProgramUniform4uiv);
+   SET_ProgramUniformMatrix2fv(table, save_ProgramUniformMatrix2fv);
+   SET_ProgramUniformMatrix3fv(table, save_ProgramUniformMatrix3fv);
+   SET_ProgramUniformMatrix4fv(table, save_ProgramUniformMatrix4fv);
+   SET_ProgramUniformMatrix2x3fv(table, save_ProgramUniformMatrix2x3fv);
+   SET_ProgramUniformMatrix3x2fv(table, save_ProgramUniformMatrix3x2fv);
+   SET_ProgramUniformMatrix2x4fv(table, save_ProgramUniformMatrix2x4fv);
+   SET_ProgramUniformMatrix4x2fv(table, save_ProgramUniformMatrix4x2fv);
+   SET_ProgramUniformMatrix3x4fv(table, save_ProgramUniformMatrix3x4fv);
+   SET_ProgramUniformMatrix4x3fv(table, save_ProgramUniformMatrix4x3fv);
 }
 
 
diff --git a/mesalib/src/mesa/main/errors.c b/mesalib/src/mesa/main/errors.c
index 30a867298..aa0ff50ac 100644
--- a/mesalib/src/mesa/main/errors.c
+++ b/mesalib/src/mesa/main/errors.c
@@ -1414,6 +1414,12 @@ _mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... )
    _mesa_record_error(ctx, error);
 }
 
+void
+_mesa_error_no_memory(const char *caller)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   _mesa_error(ctx, GL_OUT_OF_MEMORY, "out of memory in %s", caller);
+}
 
 /**
  * Report debug information.  Print error message to stderr via fprintf().
diff --git a/mesalib/src/mesa/main/errors.h b/mesalib/src/mesa/main/errors.h
index 06d0b21fc..b388138e8 100644
--- a/mesalib/src/mesa/main/errors.h
+++ b/mesalib/src/mesa/main/errors.h
@@ -64,6 +64,9 @@ extern void
 _mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... ) PRINTFLIKE(3, 4);
 
 extern void
+_mesa_error_no_memory(const char *caller);
+
+extern void
 _mesa_debug( const struct gl_context *ctx, const char *fmtString, ... ) PRINTFLIKE(2, 3);
 
 extern void
diff --git a/mesalib/src/mesa/main/extensions.c b/mesalib/src/mesa/main/extensions.c
index a72284c9a..c2ff7e3b7 100644
--- a/mesalib/src/mesa/main/extensions.c
+++ b/mesalib/src/mesa/main/extensions.c
@@ -114,6 +114,7 @@ static const struct extension extension_table[] = {
    { "GL_ARB_invalidate_subdata",                  o(dummy_true),                              GL,             2012 },
    { "GL_ARB_map_buffer_alignment",                o(dummy_true),                              GL,             2011 },
    { "GL_ARB_map_buffer_range",                    o(ARB_map_buffer_range),                    GL,             2008 },
+   { "GL_ARB_multi_bind",                          o(dummy_true),                              GL,             2013 },
    { "GL_ARB_multi_draw_indirect",                 o(ARB_draw_indirect),                       GLC,            2012 },
    { "GL_ARB_multisample",                         o(dummy_true),                              GLL,            1994 },
    { "GL_ARB_multitexture",                        o(dummy_true),                              GLL,            1998 },
@@ -127,7 +128,7 @@ static const struct extension extension_table[] = {
    { "GL_ARB_sample_shading",                      o(ARB_sample_shading),                      GL,             2009 },
    { "GL_ARB_sampler_objects",                     o(dummy_true),                              GL,             2009 },
    { "GL_ARB_seamless_cube_map",                   o(ARB_seamless_cube_map),                   GL,             2009 },
-   { "GL_ARB_separate_shader_objects",             o(ARB_separate_shader_objects),             GL,             2010 },
+   { "GL_ARB_separate_shader_objects",             o(dummy_true),                              GL,             2010 },
    { "GL_ARB_shader_atomic_counters",              o(ARB_shader_atomic_counters),              GL,             2011 },
    { "GL_ARB_shader_bit_encoding",                 o(ARB_shader_bit_encoding),                 GL,             2010 },
    { "GL_ARB_shader_image_load_store",             o(ARB_shader_image_load_store),             GL,             2011 },
@@ -217,7 +218,7 @@ static const struct extension extension_table[] = {
    { "GL_EXT_provoking_vertex",                    o(EXT_provoking_vertex),                    GL,             2009 },
    { "GL_EXT_rescale_normal",                      o(dummy_true),                              GLL,            1997 },
    { "GL_EXT_secondary_color",                     o(dummy_true),                              GLL,            1999 },
-   { "GL_EXT_separate_shader_objects",             o(EXT_separate_shader_objects),             GLL,            2008 },
+   { "GL_EXT_separate_shader_objects",             o(dummy_true),                                         ES2, 2013 },
    { "GL_EXT_separate_specular_color",             o(dummy_true),                              GLL,            1997 },
    { "GL_EXT_shader_integer_mix",                  o(EXT_shader_integer_mix),                  GL       | ES3, 2013 },
    { "GL_EXT_shadow_funcs",                        o(ARB_shadow),                              GLL,            2002 },
@@ -330,6 +331,7 @@ static const struct extension extension_table[] = {
    { "GL_IBM_rasterpos_clip",                      o(dummy_true),                              GLL,            1996 },
    { "GL_IBM_texture_mirrored_repeat",             o(dummy_true),                              GLL,            1998 },
    { "GL_INGR_blend_func_separate",                o(EXT_blend_func_separate),                 GLL,            1999 },
+   { "GL_INTEL_performance_query",                 o(INTEL_performance_query),                       GL | ES2, 2013 },
    { "GL_MESA_pack_invert",                        o(MESA_pack_invert),                        GL,             2002 },
    { "GL_MESA_texture_signed_rgba",                o(EXT_texture_snorm),                       GL,             2009 },
    { "GL_MESA_window_pos",                         o(dummy_true),                              GLL,            2000 },
@@ -409,7 +411,6 @@ _mesa_enable_sw_extensions(struct gl_context *ctx)
    ctx->Extensions.ARB_occlusion_query = GL_TRUE;
    ctx->Extensions.ARB_occlusion_query2 = GL_TRUE;
    ctx->Extensions.ARB_point_sprite = GL_TRUE;
-   ctx->Extensions.EXT_separate_shader_objects = GL_TRUE;
    ctx->Extensions.ARB_shadow = GL_TRUE;
    ctx->Extensions.ARB_texture_border_clamp = GL_TRUE;
    ctx->Extensions.ARB_texture_cube_map = GL_TRUE;
diff --git a/mesalib/src/mesa/main/fbobject.c b/mesalib/src/mesa/main/fbobject.c
index ca16ae1ec..97538bc7b 100644
--- a/mesalib/src/mesa/main/fbobject.c
+++ b/mesalib/src/mesa/main/fbobject.c
@@ -1058,6 +1058,8 @@ _mesa_test_framebuffer_completeness(struct gl_context *ctx,
       if (att->Layered) {
          if (att_tex_target == GL_TEXTURE_CUBE_MAP)
             att_layer_count = 6;
+         else if (att_tex_target == GL_TEXTURE_1D_ARRAY)
+            att_layer_count = att->Renderbuffer->Height;
          else
             att_layer_count = att->Renderbuffer->Depth;
       } else {
diff --git a/mesalib/src/mesa/main/ff_fragment_shader.cpp b/mesalib/src/mesa/main/ff_fragment_shader.cpp
index 605f3713e..8c360970f 100644
--- a/mesalib/src/mesa/main/ff_fragment_shader.cpp
+++ b/mesalib/src/mesa/main/ff_fragment_shader.cpp
@@ -1299,7 +1299,7 @@ create_new_program(struct gl_context *ctx, struct state_key *key)
     * fixed function program in a GLES2 context at all, but that's a
     * big mess to clean up.
     */
-   p.shader_program->InternalSeparateShader = GL_TRUE;
+   p.shader_program->SeparateShader = GL_TRUE;
 
    state->language_version = 130;
    state->es_shader = false;
diff --git a/mesalib/src/mesa/main/format_unpack.c b/mesalib/src/mesa/main/format_unpack.c
index cf96609ea..9cc8f4dba 100644
--- a/mesalib/src/mesa/main/format_unpack.c
+++ b/mesalib/src/mesa/main/format_unpack.c
@@ -4263,3 +4263,108 @@ _mesa_unpack_uint_24_8_depth_stencil_row(mesa_format format, GLuint n,
       return;
    }
 }
+
+static void
+unpack_float_32_uint_24_8_Z24_UNORM_S8_UINT(const GLuint *src,
+                                            GLuint *dst, GLuint n)
+{
+   GLuint i;
+   struct z32f_x24s8 *d = (struct z32f_x24s8 *) dst;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
+
+   for (i = 0; i < n; i++) {
+      const GLuint z24 = src[i] & 0xffffff;
+      d[i].z = z24 * scale;
+      d[i].x24s8 = src[i] >> 24;
+      assert(d[i].z >= 0.0f);
+      assert(d[i].z <= 1.0f);
+   }
+}
+
+static void
+unpack_float_32_uint_24_8_Z32_FLOAT_S8X24_UINT(const GLuint *src,
+                                               GLuint *dst, GLuint n)
+{
+   memcpy(dst, src, n * sizeof(struct z32f_x24s8));
+}
+
+static void
+unpack_float_32_uint_24_8_S8_UINT_Z24_UNORM(const GLuint *src,
+                                            GLuint *dst, GLuint n)
+{
+   GLuint i;
+   struct z32f_x24s8 *d = (struct z32f_x24s8 *) dst;
+   const GLdouble scale = 1.0 / (GLdouble) 0xffffff;
+
+   for (i = 0; i < n; i++) {
+      const GLuint z24 = src[i] >> 8;
+      d[i].z = z24 * scale;
+      d[i].x24s8 = src[i] & 0xff;
+      assert(d[i].z >= 0.0f);
+      assert(d[i].z <= 1.0f);
+   }
+}
+
+/**
+ * Unpack depth/stencil returning as GL_FLOAT_32_UNSIGNED_INT_24_8_REV.
+ * \param format  the source data format
+ *
+ * In GL_FLOAT_32_UNSIGNED_INT_24_8_REV lower 4 bytes contain float
+ * component and higher 4 bytes contain packed 24-bit and 8-bit
+ * components.
+ *
+ *    31 30 29 28 ... 4 3 2 1 0    31 30 29 ... 9 8 7 6 5 ... 2 1 0
+ *    +-------------------------+  +--------------------------------+
+ *    |    Float Component      |  | Unused         | 8 bit stencil |
+ *    +-------------------------+  +--------------------------------+
+ *          lower 4 bytes                  higher 4 bytes
+ */
+void
+_mesa_unpack_float_32_uint_24_8_depth_stencil_row(mesa_format format, GLuint n,
+			                          const void *src, GLuint *dst)
+{
+   switch (format) {
+   case MESA_FORMAT_S8_UINT_Z24_UNORM:
+      unpack_float_32_uint_24_8_S8_UINT_Z24_UNORM(src, dst, n);
+      break;
+   case MESA_FORMAT_Z24_UNORM_S8_UINT:
+      unpack_float_32_uint_24_8_Z24_UNORM_S8_UINT(src, dst, n);
+      break;
+   case MESA_FORMAT_Z32_FLOAT_S8X24_UINT:
+      unpack_float_32_uint_24_8_Z32_FLOAT_S8X24_UINT(src, dst, n);
+      break;
+   default:
+      _mesa_problem(NULL,
+                    "bad format %s in _mesa_unpack_uint_24_8_depth_stencil_row",
+                    _mesa_get_format_name(format));
+      return;
+   }
+}
+
+/**
+ * Unpack depth/stencil
+ * \param format  the source data format
+ * \param type the destination data type
+ */
+void
+_mesa_unpack_depth_stencil_row(mesa_format format, GLuint n,
+	                       const void *src, GLenum type,
+                               GLuint *dst)
+{
+   assert(type == GL_UNSIGNED_INT_24_8 ||
+          type == GL_FLOAT_32_UNSIGNED_INT_24_8_REV);
+
+   switch (type) {
+   case GL_UNSIGNED_INT_24_8:
+      _mesa_unpack_uint_24_8_depth_stencil_row(format, n, src, dst);
+      break;
+   case GL_FLOAT_32_UNSIGNED_INT_24_8_REV:
+      _mesa_unpack_float_32_uint_24_8_depth_stencil_row(format, n, src, dst);
+      break;
+   default:
+      _mesa_problem(NULL,
+                    "bad type 0x%x in _mesa_unpack_depth_stencil_row",
+                    type);
+      return;
+   }
+}
diff --git a/mesalib/src/mesa/main/format_unpack.h b/mesalib/src/mesa/main/format_unpack.h
index 1fcfc04b3..51f97df4d 100644
--- a/mesalib/src/mesa/main/format_unpack.h
+++ b/mesalib/src/mesa/main/format_unpack.h
@@ -63,5 +63,13 @@ void
 _mesa_unpack_uint_24_8_depth_stencil_row(mesa_format format, GLuint n,
 					 const void *src, GLuint *dst);
 
-
+void
+_mesa_unpack_float_32_uint_24_8_depth_stencil_row(mesa_format format,
+                                                  GLuint n,
+                                                  const void *src,
+                                                  GLuint *dst);
+void
+_mesa_unpack_depth_stencil_row(mesa_format format, GLuint n,
+                              const void *src, GLenum type,
+                              GLuint *dst);
 #endif /* FORMAT_UNPACK_H */
diff --git a/mesalib/src/mesa/main/get.c b/mesalib/src/mesa/main/get.c
index fe35ff3ef..80a5839b5 100644
--- a/mesalib/src/mesa/main/get.c
+++ b/mesalib/src/mesa/main/get.c
@@ -387,7 +387,6 @@ EXTRA_EXT(ARB_texture_cube_map_array);
 EXTRA_EXT(ARB_texture_buffer_range);
 EXTRA_EXT(ARB_texture_multisample);
 EXTRA_EXT(ARB_texture_gather);
-EXTRA_EXT(ARB_separate_shader_objects);
 EXTRA_EXT(ARB_shader_atomic_counters);
 EXTRA_EXT(ARB_draw_indirect);
 EXTRA_EXT(ARB_shader_image_load_store);
@@ -395,6 +394,7 @@ EXTRA_EXT(ARB_viewport_array);
 EXTRA_EXT(ARB_compute_shader);
 EXTRA_EXT(ARB_gpu_shader5);
 EXTRA_EXT2(ARB_transform_feedback3, ARB_gpu_shader5);
+EXTRA_EXT(INTEL_performance_query);
 
 static const int
 extra_ARB_color_buffer_float_or_glcore[] = {
diff --git a/mesalib/src/mesa/main/get_hash_params.py b/mesalib/src/mesa/main/get_hash_params.py
index 06d0bbacc..d40fa0778 100644
--- a/mesalib/src/mesa/main/get_hash_params.py
+++ b/mesalib/src/mesa/main/get_hash_params.py
@@ -311,6 +311,12 @@ descriptor=[
 # GL_ARB_get_program_binary / GL_OES_get_program_binary
   [ "NUM_PROGRAM_BINARY_FORMATS", "CONST(0), NO_EXTRA" ],
   [ "PROGRAM_BINARY_FORMATS", "LOC_CUSTOM, TYPE_INVALID, 0, NO_EXTRA" ],
+
+# GL_INTEL_performance_query
+  [ "PERFQUERY_QUERY_NAME_LENGTH_MAX_INTEL", "CONST(MAX_PERFQUERY_QUERY_NAME_LENGTH), extra_INTEL_performance_query" ],
+  [ "PERFQUERY_COUNTER_NAME_LENGTH_MAX_INTEL", "CONST(MAX_PERFQUERY_COUNTER_NAME_LENGTH), extra_INTEL_performance_query" ],
+  [ "PERFQUERY_COUNTER_DESC_LENGTH_MAX_INTEL", "CONST(MAX_PERFQUERY_COUNTER_DESC_LENGTH), extra_INTEL_performance_query" ],
+  [ "PERFQUERY_GPA_EXTENDED_COUNTERS_INTEL", "CONST(PERFQUERY_HAVE_GPA_EXTENDED_COUNTERS), extra_INTEL_performance_query" ],
 ]},
 
 # GLES3 is not a typo.
@@ -727,7 +733,7 @@ descriptor=[
   [ "MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS_ARB", "CONTEXT_INT(Const.MaxProgramTextureGatherComponents), extra_ARB_texture_gather"],
 
 # GL_ARB_separate_shader_objects
-  [ "PROGRAM_PIPELINE_BINDING", "LOC_CUSTOM, TYPE_INT, GL_PROGRAM_PIPELINE_BINDING, extra_ARB_separate_shader_objects" ],
+  [ "PROGRAM_PIPELINE_BINDING", "LOC_CUSTOM, TYPE_INT, GL_PROGRAM_PIPELINE_BINDING, NO_EXTRA" ],
 
 # GL_ARB_shader_atomic_counters
   [ "ATOMIC_COUNTER_BUFFER_BINDING", "LOC_CUSTOM, TYPE_INT, 0, extra_ARB_shader_atomic_counters" ],
diff --git a/mesalib/src/mesa/main/hash.c b/mesalib/src/mesa/main/hash.c
index 4c92005e0..23018e9da 100644
--- a/mesalib/src/mesa/main/hash.c
+++ b/mesalib/src/mesa/main/hash.c
@@ -194,15 +194,55 @@ _mesa_HashLookup(struct _mesa_HashTable *table, GLuint key)
 
 
 /**
- * Insert a key/pointer pair into the hash table.  
- * If an entry with this key already exists we'll replace the existing entry.
- * 
+ * Lookup an entry in the hash table without locking the mutex.
+ *
+ * The hash table mutex must be locked manually by calling
+ * _mesa_HashLockMutex() before calling this function.
+ *
+ * \param table the hash table.
+ * \param key the key.
+ *
+ * \return pointer to user's data or NULL if key not in table
+ */
+void *
+_mesa_HashLookupLocked(struct _mesa_HashTable *table, GLuint key)
+{
+   return _mesa_HashLookup_unlocked(table, key);
+}
+
+
+/**
+ * Lock the hash table mutex.
+ *
+ * This function should be used when multiple objects need
+ * to be looked up in the hash table, to avoid having to lock
+ * and unlock the mutex each time.
+ *
  * \param table the hash table.
- * \param key the key (not zero).
- * \param data pointer to user data.
  */
 void
-_mesa_HashInsert(struct _mesa_HashTable *table, GLuint key, void *data)
+_mesa_HashLockMutex(struct _mesa_HashTable *table)
+{
+   assert(table);
+   mtx_lock(&table->Mutex);
+}
+
+
+/**
+ * Unlock the hash table mutex.
+ *
+ * \param table the hash table.
+ */
+void
+_mesa_HashUnlockMutex(struct _mesa_HashTable *table)
+{
+   assert(table);
+   mtx_unlock(&table->Mutex);
+}
+
+
+static inline void
+_mesa_HashInsert_unlocked(struct _mesa_HashTable *table, GLuint key, void *data)
 {
    uint32_t hash = uint_hash(key);
    struct hash_entry *entry;
@@ -210,8 +250,6 @@ _mesa_HashInsert(struct _mesa_HashTable *table, GLuint key, void *data)
    assert(table);
    assert(key);
 
-   mtx_lock(&table->Mutex);
-
    if (key > table->MaxKey)
       table->MaxKey = key;
 
@@ -225,11 +263,44 @@ _mesa_HashInsert(struct _mesa_HashTable *table, GLuint key, void *data)
          _mesa_hash_table_insert(table->ht, hash, uint_key(key), data);
       }
    }
+}
 
-   mtx_unlock(&table->Mutex);
+
+/**
+ * Insert a key/pointer pair into the hash table without locking the mutex.
+ * If an entry with this key already exists we'll replace the existing entry.
+ *
+ * The hash table mutex must be locked manually by calling
+ * _mesa_HashLockMutex() before calling this function.
+ *
+ * \param table the hash table.
+ * \param key the key (not zero).
+ * \param data pointer to user data.
+ */
+void
+_mesa_HashInsertLocked(struct _mesa_HashTable *table, GLuint key, void *data)
+{
+   _mesa_HashInsert_unlocked(table, key, data);
 }
 
 
+/**
+ * Insert a key/pointer pair into the hash table.
+ * If an entry with this key already exists we'll replace the existing entry.
+ *
+ * \param table the hash table.
+ * \param key the key (not zero).
+ * \param data pointer to user data.
+ */
+void
+_mesa_HashInsert(struct _mesa_HashTable *table, GLuint key, void *data)
+{
+   assert(table);
+   mtx_lock(&table->Mutex);
+   _mesa_HashInsert_unlocked(table, key, data);
+   mtx_unlock(&table->Mutex);
+}
+
 
 /**
  * Remove an entry from the hash table.
diff --git a/mesalib/src/mesa/main/hash.h b/mesalib/src/mesa/main/hash.h
index b34f32848..e3e8f492e 100644
--- a/mesalib/src/mesa/main/hash.h
+++ b/mesalib/src/mesa/main/hash.h
@@ -45,6 +45,15 @@ extern void _mesa_HashInsert(struct _mesa_HashTable *table, GLuint key, void *da
 
 extern void _mesa_HashRemove(struct _mesa_HashTable *table, GLuint key);
 
+extern void _mesa_HashLockMutex(struct _mesa_HashTable *table);
+
+extern void _mesa_HashUnlockMutex(struct _mesa_HashTable *table);
+
+extern void *_mesa_HashLookupLocked(struct _mesa_HashTable *table, GLuint key);
+
+extern void _mesa_HashInsertLocked(struct _mesa_HashTable *table,
+                                   GLuint key, void *data);
+
 extern void
 _mesa_HashDeleteAll(struct _mesa_HashTable *table,
                     void (*callback)(GLuint key, void *data, void *userData),
diff --git a/mesalib/src/mesa/main/imports.h b/mesalib/src/mesa/main/imports.h
index db19eda3d..ba869286f 100644..100755
--- a/mesalib/src/mesa/main/imports.h
+++ b/mesalib/src/mesa/main/imports.h
@@ -126,7 +126,8 @@ typedef union { GLfloat f; GLint i; GLuint u; } fi_type;
 #define atanhf(f) ((float) atanh(f))
 #endif
 
-#if defined(_MSC_VER) && (_MSC_VER < 1800)  /* Not req'd on VS2013 and above */
+#if defined(_MSC_VER)
+#if _MSC_VER < 1800  /* Not req'd on VS2013 and above */
 static inline float truncf(float x) { return x < 0.0f ? ceilf(x) : floorf(x); }
 static inline float exp2f(float x) { return powf(2.0f, x); }
 static inline float log2f(float x) { return logf(x) * 1.442695041f; }
@@ -135,6 +136,7 @@ static inline float acoshf(float x) { return logf(x + sqrtf(x * x - 1.0f)); }
 static inline float atanhf(float x) { return (logf(1.0f + x) - logf(1.0f - x)) / 2.0f; }
 static inline int isblank(int ch) { return ch == ' ' || ch == '\t'; }
 #define strtoll(p, e, b) _strtoi64(p, e, b)
+#endif /* _MSC_VER < 1800 */
 #endif
 /*@}*/
 
diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h
index 5fbfffe98..917d071a2 100644
--- a/mesalib/src/mesa/main/mtypes.h
+++ b/mesalib/src/mesa/main/mtypes.h
@@ -1195,6 +1195,8 @@ struct gl_texture_object
    GLuint Name;                /**< the user-visible texture object ID */
    GLchar *Label;               /**< GL_KHR_debug */
    GLenum Target;              /**< GL_TEXTURE_1D, GL_TEXTURE_2D, etc. */
+   gl_texture_index TargetIndex; /**< The gl_texture_unit::CurrentTex index.
+                                      Only valid when Target is valid. */
 
    struct gl_sampler_object Sampler;
 
@@ -1368,6 +1370,9 @@ struct gl_texture_unit
 
    /** Points to highest priority, complete and enabled texture object */
    struct gl_texture_object *_Current;
+
+   /** Texture targets that have a non-default texture bound */
+   GLbitfield _BoundTextures;
 };
 
 
@@ -2432,6 +2437,15 @@ struct gl_shader
    struct glsl_symbol_table *symbols;
 
    bool uses_builtin_functions;
+   bool uses_gl_fragcoord;
+   bool redeclares_gl_fragcoord;
+   bool ARB_fragment_coord_conventions_enable;
+
+   /**
+    * Fragment shader state from GLSL 1.50 layout qualifiers.
+    */
+   bool origin_upper_left;
+   bool pixel_center_integer;
 
    /**
     * Geometry shader state from GLSL 1.50 layout qualifiers.
@@ -2594,17 +2608,6 @@ struct gl_shader_program
    GLboolean BinaryRetreivableHint;
 
    /**
-    * Flags that the linker should not reject the program if it lacks
-    * a vertex or fragment shader.  GLES2 doesn't allow separate
-    * shader objects, and would reject them.  However, we internally
-    * build separate shader objects for fixed function programs, which
-    * we use for drivers/common/meta.c and for handling
-    * _mesa_update_state with no program bound (for example in
-    * glClear()).
-    */
-   GLboolean InternalSeparateShader;
-
-   /**
     * Indicates whether program can be bound for individual pipeline stages
     * using UseProgramStages after it is next linked.
     */
@@ -2757,6 +2760,11 @@ struct gl_shader_program
     * \c NULL.
     */
    struct gl_shader *_LinkedShaders[MESA_SHADER_STAGES];
+
+   /* True if any of the fragment shaders attached to this program use:
+    * #extension ARB_fragment_coord_conventions: enable
+    */
+   GLboolean ARB_fragment_coord_conventions_enable;
 };   
 
 
@@ -2790,9 +2798,7 @@ struct gl_pipeline_object
    /**
     * Programs used for rendering
     *
-    * There is a separate program set for each shader stage.  If
-    * GL_EXT_separate_shader_objects is not supported, each of these must point
-    * to \c NULL or to the same program.
+    * There is a separate program set for each shader stage.
     */
    struct gl_shader_program *CurrentProgram[MESA_SHADER_STAGES];
 
@@ -3521,7 +3527,6 @@ struct gl_extensions
    GLboolean ARB_point_sprite;
    GLboolean ARB_sample_shading;
    GLboolean ARB_seamless_cube_map;
-   GLboolean ARB_separate_shader_objects;
    GLboolean ARB_shader_atomic_counters;
    GLboolean ARB_shader_bit_encoding;
    GLboolean ARB_shader_image_load_store;
@@ -3547,6 +3552,7 @@ struct gl_extensions
    GLboolean ARB_texture_mirror_clamp_to_edge;
    GLboolean ARB_texture_multisample;
    GLboolean ARB_texture_non_power_of_two;
+   GLboolean ARB_texture_stencil8;
    GLboolean ARB_texture_query_levels;
    GLboolean ARB_texture_query_lod;
    GLboolean ARB_texture_rg;
@@ -3577,7 +3583,6 @@ struct gl_extensions
    GLboolean EXT_pixel_buffer_object;
    GLboolean EXT_point_parameters;
    GLboolean EXT_provoking_vertex;
-   GLboolean EXT_separate_shader_objects;
    GLboolean EXT_shader_integer_mix;
    GLboolean EXT_stencil_two_side;
    GLboolean EXT_texture3D;
@@ -3608,6 +3613,7 @@ struct gl_extensions
    GLboolean ATI_texture_env_combine3;
    GLboolean ATI_fragment_shader;
    GLboolean ATI_separate_stencil;
+   GLboolean INTEL_performance_query;
    GLboolean MESA_pack_invert;
    GLboolean MESA_ycbcr_texture;
    GLboolean NV_conditional_render;
diff --git a/mesalib/src/mesa/main/performance_monitor.c b/mesalib/src/mesa/main/performance_monitor.c
index e62f77012..21b9423e0 100644
--- a/mesalib/src/mesa/main/performance_monitor.c
+++ b/mesalib/src/mesa/main/performance_monitor.c
@@ -137,6 +137,46 @@ get_counter(const struct gl_perf_monitor_group *group_obj, GLuint id)
    return &group_obj->Counters[id];
 }
 
+/* For INTEL_performance_query, query id 0 is reserved to be invalid. We use
+ * index to Groups array + 1 as the query id. Same applies to counter id.
+ */
+static inline GLuint
+queryid_to_index(GLuint queryid)
+{
+   return queryid - 1;
+}
+
+static inline GLuint
+index_to_queryid(GLuint index)
+{
+   return index + 1;
+}
+
+static inline bool
+queryid_valid(const struct gl_context *ctx, GLuint queryid)
+{
+   return get_group(ctx, queryid_to_index(queryid)) != NULL;
+}
+
+static inline GLuint
+counterid_to_index(GLuint counterid)
+{
+   return counterid - 1;
+}
+
+static inline GLuint
+index_to_counterid(GLuint index)
+{
+   return index + 1;
+}
+
+static inline bool
+counterid_valid(const struct gl_perf_monitor_group *group_obj,
+                GLuint counterid)
+{
+   return get_counter(group_obj, counterid_to_index(counterid)) != NULL;
+}
+
 /*****************************************************************************/
 
 void GLAPIENTRY
@@ -639,3 +679,577 @@ _mesa_perf_monitor_counter_size(const struct gl_perf_monitor_counter *c)
       return 0;
    }
 }
+
+extern void GLAPIENTRY
+_mesa_GetFirstPerfQueryIdINTEL(GLuint *queryId)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   unsigned numGroups;
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If queryId pointer is equal to 0, INVALID_VALUE error is generated."
+    */
+   if (!queryId) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetFirstPerfQueryIdINTEL(queryId == NULL)");
+      return;
+   }
+
+   numGroups = ctx->PerfMonitor.NumGroups;
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If the given hardware platform doesn't support any performance
+    *    queries, then the value of 0 is returned and INVALID_OPERATION error
+    *    is raised."
+    */
+   if (numGroups == 0) {
+      *queryId = 0;
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glGetFirstPerfQueryIdINTEL(no queries supported)");
+      return;
+   }
+
+   *queryId = index_to_queryid(0);
+}
+
+extern void GLAPIENTRY
+_mesa_GetNextPerfQueryIdINTEL(GLuint queryId, GLuint *nextQueryId)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "The result is passed in location pointed by nextQueryId. If query
+    *    identified by queryId is the last query available the value of 0 is
+    *    returned. If the specified performance query identifier is invalid
+    *    then INVALID_VALUE error is generated. If nextQueryId pointer is
+    *    equal to 0, an INVALID_VALUE error is generated.  Whenever error is
+    *    generated, the value of 0 is returned."
+    */
+
+   if (!nextQueryId) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetNextPerfQueryIdINTEL(nextQueryId == NULL)");
+      return;
+   }
+
+   if (!queryid_valid(ctx, queryId)) {
+      *nextQueryId = 0;
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetNextPerfQueryIdINTEL(invalid query)");
+      return;
+   }
+
+   ++queryId;
+
+   if (!queryid_valid(ctx, queryId)) {
+      *nextQueryId = 0;
+   } else {
+      *nextQueryId = queryId;
+   }
+}
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryIdByNameINTEL(char *queryName, GLuint *queryId)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   unsigned i;
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If queryName does not reference a valid query name, an INVALID_VALUE
+    *    error is generated."
+    */
+   if (!queryName) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfQueryIdByNameINTEL(queryName == NULL)");
+      return;
+   }
+
+   /* The specification does not state that this produces an error. */
+   if (!queryId) {
+      _mesa_warning(ctx, "glGetPerfQueryIdByNameINTEL(queryId == NULL)");
+      return;
+   }
+
+   for (i = 0; i < ctx->PerfMonitor.NumGroups; ++i) {
+      const struct gl_perf_monitor_group *group_obj = get_group(ctx, i);
+      if (strcmp(group_obj->Name, queryName) == 0) {
+         *queryId = index_to_queryid(i);
+         return;
+      }
+   }
+
+   _mesa_error(ctx, GL_INVALID_VALUE,
+               "glGetPerfQueryIdByNameINTEL(invalid query name)");
+}
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryInfoINTEL(GLuint queryId,
+                            GLuint queryNameLength, char *queryName,
+                            GLuint *dataSize, GLuint *noCounters,
+                            GLuint *noActiveInstances,
+                            GLuint *capsMask)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   unsigned i;
+
+   const struct gl_perf_monitor_group *group_obj =
+      get_group(ctx, queryid_to_index(queryId));
+
+   if (group_obj == NULL) {
+      /* The GL_INTEL_performance_query spec says:
+       *
+       *    "If queryId does not reference a valid query type, an
+       *    INVALID_VALUE error is generated."
+       */
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfQueryInfoINTEL(invalid query)");
+      return;
+   }
+
+   if (queryName) {
+      strncpy(queryName, group_obj->Name, queryNameLength);
+
+      /* No specification given about whether the string needs to be
+       * zero-terminated. Zero-terminate the string always as we don't
+       * otherwise communicate the length of the returned string.
+       */
+      if (queryNameLength > 0) {
+         queryName[queryNameLength - 1] = '\0';
+      }
+   }
+
+   if (dataSize) {
+      unsigned size = 0;
+      for (i = 0; i < group_obj->NumCounters; ++i) {
+         /* What we get from the driver is group id (uint32_t) + counter id
+          * (uint32_t) + value.
+          */
+         size += 2 * sizeof(uint32_t) + _mesa_perf_monitor_counter_size(&group_obj->Counters[i]);
+      }
+      *dataSize = size;
+   }
+
+   if (noCounters) {
+      *noCounters = group_obj->NumCounters;
+   }
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "-- the actual number of already created query instances in
+    *    maxInstances location"
+    *
+    * 1) Typo in the specification, should be noActiveInstances.
+    * 2) Another typo in the specification, maxInstances parameter is not listed
+    *    in the declaration of this function in the list of new functions.
+    */
+   if (noActiveInstances) {
+      *noActiveInstances = _mesa_HashNumEntries(ctx->PerfMonitor.Monitors);
+   }
+
+   if (capsMask) {
+      /* TODO: This information not yet available in the monitor structs. For
+       * now, we hardcode SINGLE_CONTEXT, since that's what the implementation
+       * currently tries very hard to do.
+       */
+      *capsMask = GL_PERFQUERY_SINGLE_CONTEXT_INTEL;
+   }
+}
+
+extern void GLAPIENTRY
+_mesa_GetPerfCounterInfoINTEL(GLuint queryId, GLuint counterId,
+                              GLuint counterNameLength, char *counterName,
+                              GLuint counterDescLength, char *counterDesc,
+                              GLuint *counterOffset, GLuint *counterDataSize, GLuint *counterTypeEnum,
+                              GLuint *counterDataTypeEnum, GLuint64 *rawCounterMaxValue)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   const struct gl_perf_monitor_group *group_obj;
+   const struct gl_perf_monitor_counter *counter_obj;
+   unsigned counterIndex;
+   unsigned i;
+
+   group_obj = get_group(ctx, queryid_to_index(queryId));
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If the pair of queryId and counterId does not reference a valid
+    *    counter, an INVALID_VALUE error is generated."
+    */
+   if (group_obj == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfCounterInfoINTEL(invalid queryId)");
+      return;
+   }
+
+   counterIndex = counterid_to_index(counterId);
+   counter_obj = get_counter(group_obj, counterIndex);
+
+   if (counter_obj == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfCounterInfoINTEL(invalid counterId)");
+      return;
+   }
+
+   if (counterName) {
+      strncpy(counterName, counter_obj->Name, counterNameLength);
+
+      /* No specification given about whether the string needs to be
+       * zero-terminated. Zero-terminate the string always as we don't
+       * otherwise communicate the length of the returned string.
+       */
+      if (counterNameLength > 0) {
+         counterName[counterNameLength - 1] = '\0';
+      }
+   }
+
+   if (counterDesc) {
+      /* TODO: No separate description text at the moment. We pass the name
+       * again for the moment.
+       */
+      strncpy(counterDesc, counter_obj->Name, counterDescLength);
+
+      /* No specification given about whether the string needs to be
+       * zero-terminated. Zero-terminate the string always as we don't
+       * otherwise communicate the length of the returned string.
+       */
+      if (counterDescLength > 0) {
+         counterDesc[counterDescLength - 1] = '\0';
+      }
+   }
+
+   if (counterOffset) {
+      unsigned offset = 0;
+      for (i = 0; i < counterIndex; ++i) {
+         /* What we get from the driver is group id (uint32_t) + counter id
+          * (uint32_t) + value.
+          */
+         offset += 2 * sizeof(uint32_t) + _mesa_perf_monitor_counter_size(&group_obj->Counters[i]);
+      }
+      *counterOffset = 2 * sizeof(uint32_t) + offset;
+   }
+
+   if (counterDataSize) {
+      *counterDataSize = _mesa_perf_monitor_counter_size(counter_obj);
+   }
+
+   if (counterTypeEnum) {
+      /* TODO: Different counter types (semantic type, not data type) not
+       * supported as of yet.
+       */
+      *counterTypeEnum = GL_PERFQUERY_COUNTER_RAW_INTEL;
+   }
+
+   if (counterDataTypeEnum) {
+      switch (counter_obj->Type) {
+      case GL_FLOAT:
+      case GL_PERCENTAGE_AMD:
+         *counterDataTypeEnum = GL_PERFQUERY_COUNTER_DATA_FLOAT_INTEL;
+         break;
+      case GL_UNSIGNED_INT:
+         *counterDataTypeEnum = GL_PERFQUERY_COUNTER_DATA_UINT32_INTEL;
+         break;
+      case GL_UNSIGNED_INT64_AMD:
+         *counterDataTypeEnum = GL_PERFQUERY_COUNTER_DATA_UINT64_INTEL;
+         break;
+      default:
+         assert(!"Should not get here: invalid counter type");
+         return;
+      }
+   }
+
+   if (rawCounterMaxValue) {
+      /* This value is (implicitly) specified to be used only with
+       * GL_PERFQUERY_COUNTER_RAW_INTEL counters. When semantic types for
+       * counters are added, that needs to be checked.
+       */
+
+      /* The GL_INTEL_performance_query spec says:
+       *
+       *    "for some raw counters for which the maximal value is
+       *    deterministic, the maximal value of the counter in 1 second is
+       *    returned in the location pointed by rawCounterMaxValue, otherwise,
+       *    the location is written with the value of 0."
+       *
+       * The maximum value reported by the driver at the moment is not with
+       * these semantics, so write 0 always to be safe.
+       */
+      *rawCounterMaxValue = 0;
+   }
+}
+
+extern void GLAPIENTRY
+_mesa_CreatePerfQueryINTEL(GLuint queryId, GLuint *queryHandle)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLuint first;
+   GLuint group;
+   const struct gl_perf_monitor_group *group_obj;
+   struct gl_perf_monitor_object *m;
+   unsigned i;
+
+   /* This is not specified in the extension, but is the only sane thing to
+    * do.
+    */
+   if (queryHandle == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glCreatePerfQueryINTEL(queryHandle == NULL)");
+      return;
+   }
+
+   group = queryid_to_index(queryId);
+   group_obj = get_group(ctx, group);
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If queryId does not reference a valid query type, an INVALID_VALUE
+    *    error is generated."
+    */
+   if (group_obj == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glCreatePerfQueryINTEL(invalid queryId)");
+      return;
+   }
+
+   /* The query object created here is the counterpart of a `monitor' in
+    * AMD_performance_monitor. This call is equivalent to calling
+    * GenPerfMonitorsAMD and SelectPerfMonitorCountersAMD with a list of all
+    * counters in a group.
+    */
+
+   /* We keep the monitor ids contiguous */
+   first = _mesa_HashFindFreeKeyBlock(ctx->PerfMonitor.Monitors, 1);
+   if (!first) {
+      /* The GL_INTEL_performance_query spec says:
+       *
+       *    "If the query instance cannot be created due to exceeding the
+       *    number of allowed instances or driver fails query creation due to
+       *    an insufficient memory reason, an OUT_OF_MEMORY error is
+       *    generated, and the location pointed by queryHandle returns NULL."
+      */
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCreatePerfQueryINTEL");
+      return;
+   }
+
+   m = new_performance_monitor(ctx, first);
+   _mesa_HashInsert(ctx->PerfMonitor.Monitors, first, m);
+   *queryHandle = first;
+
+   ctx->Driver.ResetPerfMonitor(ctx, m);
+
+   for (i = 0; i < group_obj->NumCounters; ++i) {
+      ++m->ActiveGroups[group];
+      /* Counters are a continuous range of integers, 0 to NumCounters (excl),
+       * so i is the counter value to use here.
+       */
+      BITSET_SET(m->ActiveCounters[group], i);
+   }
+}
+
+extern void GLAPIENTRY
+_mesa_DeletePerfQueryINTEL(GLuint queryHandle)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_perf_monitor_object *m;
+
+   /* The queryHandle is the counterpart to AMD_performance_monitor's monitor
+    * id.
+    */
+   m = lookup_monitor(ctx, queryHandle);
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If a query handle doesn't reference a previously created performance
+    *    query instance, an INVALID_VALUE error is generated."
+    */
+   if (m == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glDeletePerfQueryINTEL(invalid queryHandle)");
+      return;
+   }
+
+   /* Let the driver stop the monitor if it's active. */
+   if (m->Active) {
+      ctx->Driver.ResetPerfMonitor(ctx, m);
+      m->Ended = false;
+   }
+
+   _mesa_HashRemove(ctx->PerfMonitor.Monitors, queryHandle);
+   ralloc_free(m->ActiveGroups);
+   ralloc_free(m->ActiveCounters);
+   ctx->Driver.DeletePerfMonitor(ctx, m);
+}
+
+extern void GLAPIENTRY
+_mesa_BeginPerfQueryINTEL(GLuint queryHandle)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_perf_monitor_object *m;
+
+   /* The queryHandle is the counterpart to AMD_performance_monitor's monitor
+    * id.
+    */
+
+   m = lookup_monitor(ctx, queryHandle);
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If a query handle doesn't reference a previously created performance
+    *    query instance, an INVALID_VALUE error is generated."
+    */
+   if (m == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glBeginPerfQueryINTEL(invalid queryHandle)");
+      return;
+   }
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "Note that some query types, they cannot be collected in the same
+    *    time. Therefore calls of BeginPerfQueryINTEL() cannot be nested if
+    *    they refer to queries of such different types. In such case
+    *    INVALID_OPERATION error is generated."
+    *
+    * We also generate an INVALID_OPERATION error if the driver can't begin
+    * monitoring for its own reasons, and for nesting the same query.
+    */
+   if (m->Active) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBeginPerfQueryINTEL(already active)");
+      return;
+   }
+
+   if (ctx->Driver.BeginPerfMonitor(ctx, m)) {
+      m->Active = true;
+      m->Ended = false;
+   } else {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBeginPerfQueryINTEL(driver unable to begin monitoring)");
+   }
+}
+
+extern void GLAPIENTRY
+_mesa_EndPerfQueryINTEL(GLuint queryHandle)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_perf_monitor_object *m;
+
+   /* The queryHandle is the counterpart to AMD_performance_monitor's monitor
+    * id.
+    */
+
+   m = lookup_monitor(ctx, queryHandle);
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If a performance query is not currently started, an
+    *    INVALID_OPERATION error will be generated."
+    *
+    * The specification doesn't state that an invalid handle would be an
+    * INVALID_VALUE error. Regardless, query for such a handle will not be
+    * started, so we generate an INVALID_OPERATION in that case too.
+    */
+   if (m == NULL) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glEndPerfQueryINTEL(invalid queryHandle)");
+      return;
+   }
+
+   if (!m->Active) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glEndPerfQueryINTEL(not active)");
+      return;
+   }
+
+   ctx->Driver.EndPerfMonitor(ctx, m);
+
+   m->Active = false;
+   m->Ended = true;
+}
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryDataINTEL(GLuint queryHandle, GLuint flags,
+                            GLsizei dataSize, void *data, GLuint *bytesWritten)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_perf_monitor_object *m;
+   bool result_available;
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "If bytesWritten or data pointers are NULL then an INVALID_VALUE
+    *    error is generated."
+    */
+   if (!bytesWritten || !data) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfQueryDataINTEL(bytesWritten or data is NULL)");
+      return;
+   }
+
+   /* The queryHandle is the counterpart to AMD_performance_monitor's monitor
+    * id.
+    */
+
+   m = lookup_monitor(ctx, queryHandle);
+
+   /* The specification doesn't state that an invalid handle generates an
+    * error. We could interpret that to mean the case should be handled as
+    * "measurement not ready for this query", but what should be done if
+    * `flags' equals PERFQUERY_WAIT_INTEL?
+    *
+    * To resolve this, we just generate an INVALID_VALUE from an invalid query
+    * handle.
+    */
+   if (m == NULL) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glGetPerfQueryDataINTEL(invalid queryHandle)");
+      return;
+   }
+
+   /* We need at least enough room for a single value. */
+   if (dataSize < sizeof(GLuint)) {
+      *bytesWritten = 0;
+      return;
+   }
+
+   /* The GL_INTEL_performance_query spec says:
+    *
+    *    "The call may end without returning any data if they are not ready
+    *    for reading as the measurement session is still pending (the
+    *    EndPerfQueryINTEL() command processing is not finished by
+    *    hardware). In this case location pointed by the bytesWritten
+    *    parameter will be set to 0."
+    *
+    * If EndPerfQueryINTEL() is not called at all, we follow this.
+    */
+   if (!m->Ended) {
+      *bytesWritten = 0;
+      return;
+   }
+
+   result_available = ctx->Driver.IsPerfMonitorResultAvailable(ctx, m);
+
+   if (!result_available) {
+      if (flags == GL_PERFQUERY_FLUSH_INTEL) {
+         ctx->Driver.Flush(ctx);
+      } else if (flags == GL_PERFQUERY_WAIT_INTEL) {
+         /* Assume Finish() is both enough and not too much to wait for
+          * results. If results are still not available after Finish(), the
+          * later code automatically bails out with 0 for bytesWritten.
+          */
+         ctx->Driver.Finish(ctx);
+         result_available =
+            ctx->Driver.IsPerfMonitorResultAvailable(ctx, m);
+      }
+   }
+
+   if (result_available) {
+      ctx->Driver.GetPerfMonitorResult(ctx, m, dataSize, data, (GLint*)bytesWritten);
+   } else {
+      *bytesWritten = 0;
+   }
+}
diff --git a/mesalib/src/mesa/main/performance_monitor.h b/mesalib/src/mesa/main/performance_monitor.h
index 76234e5c1..7b311e4e0 100644
--- a/mesalib/src/mesa/main/performance_monitor.h
+++ b/mesalib/src/mesa/main/performance_monitor.h
@@ -23,7 +23,8 @@
 
 /**
  * \file performance_monitor.h
- * Core Mesa support for the AMD_performance_monitor extension.
+ * Core Mesa support for the AMD_performance_monitor extension and the
+ * INTEL_performance_query extension.
  */
 
 #pragma once
@@ -85,4 +86,44 @@ _mesa_GetPerfMonitorCounterDataAMD(GLuint monitor, GLenum pname,
 unsigned
 _mesa_perf_monitor_counter_size(const struct gl_perf_monitor_counter *);
 
+
+extern void GLAPIENTRY
+_mesa_GetFirstPerfQueryIdINTEL(GLuint *queryId);
+
+extern void GLAPIENTRY
+_mesa_GetNextPerfQueryIdINTEL(GLuint queryId, GLuint *nextQueryId);
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryIdByNameINTEL(char *queryName, GLuint *queryId);
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryInfoINTEL(GLuint queryId,
+                            GLuint queryNameLength, char *queryName,
+                            GLuint *dataSize, GLuint *noCounters,
+                            GLuint *noActiveInstances,
+                            GLuint *capsMask);
+
+extern void GLAPIENTRY
+_mesa_GetPerfCounterInfoINTEL(GLuint queryId, GLuint counterId,
+                              GLuint counterNameLength, char *counterName,
+                              GLuint counterDescLength, char *counterDesc,
+                              GLuint *counterOffset, GLuint *counterDataSize, GLuint *counterTypeEnum,
+                              GLuint *counterDataTypeEnum, GLuint64 *rawCounterMaxValue);
+
+extern void GLAPIENTRY
+_mesa_CreatePerfQueryINTEL(GLuint queryId, GLuint *queryHandle);
+
+extern void GLAPIENTRY
+_mesa_DeletePerfQueryINTEL(GLuint queryHandle);
+
+extern void GLAPIENTRY
+_mesa_BeginPerfQueryINTEL(GLuint queryHandle);
+
+extern void GLAPIENTRY
+_mesa_EndPerfQueryINTEL(GLuint queryHandle);
+
+extern void GLAPIENTRY
+_mesa_GetPerfQueryDataINTEL(GLuint queryHandle, GLuint flags,
+                            GLsizei dataSize, void *data, GLuint *bytesWritten);
+
 #endif
diff --git a/mesalib/src/mesa/main/pipelineobj.c b/mesalib/src/mesa/main/pipelineobj.c
index f55251e06..90c1d005f 100644
--- a/mesalib/src/mesa/main/pipelineobj.c
+++ b/mesalib/src/mesa/main/pipelineobj.c
@@ -412,8 +412,15 @@ _mesa_BindProgramPipeline(GLuint pipeline)
       newObj->EverBound = GL_TRUE;
    }
 
+   _mesa_bind_pipeline(ctx, newObj);
+}
+
+void
+_mesa_bind_pipeline(struct gl_context *ctx,
+                    struct gl_pipeline_object *pipe)
+{
    /* First bind the Pipeline to pipeline binding point */
-   _mesa_reference_pipeline_object(ctx, &ctx->Pipeline.Current, newObj);
+   _mesa_reference_pipeline_object(ctx, &ctx->Pipeline.Current, pipe);
 
    /* Section 2.11.3 (Program Objects) of the OpenGL 4.1 spec says:
     *
@@ -424,11 +431,11 @@ _mesa_BindProgramPipeline(GLuint pipeline)
     *     considered current."
     */
    if (&ctx->Shader != ctx->_Shader) {
-      if (pipeline) {
+      if (pipe != NULL) {
          /* Bound the pipeline to the current program and
           * restore the pipeline state
           */
-         _mesa_reference_pipeline_object(ctx, &ctx->_Shader, newObj);
+         _mesa_reference_pipeline_object(ctx, &ctx->_Shader, pipe);
       } else {
          /* Unbind the pipeline */
          _mesa_reference_pipeline_object(ctx, &ctx->_Shader,
diff --git a/mesalib/src/mesa/main/pipelineobj.h b/mesalib/src/mesa/main/pipelineobj.h
index ceaf4f14c..7285a78f1 100644
--- a/mesalib/src/mesa/main/pipelineobj.h
+++ b/mesalib/src/mesa/main/pipelineobj.h
@@ -59,6 +59,10 @@ _mesa_reference_pipeline_object(struct gl_context *ctx,
       _mesa_reference_pipeline_object_(ctx, ptr, obj);
 }
 
+extern void
+_mesa_bind_pipeline(struct gl_context *ctx,
+                    struct gl_pipeline_object *pipe);
+
 extern GLboolean
 _mesa_validate_program_pipeline(struct gl_context * ctx, struct gl_pipeline_object *pipe, GLboolean IsBound);
 
diff --git a/mesalib/src/mesa/main/samplerobj.c b/mesalib/src/mesa/main/samplerobj.c
index 4900d5256..18a14d89a 100644
--- a/mesalib/src/mesa/main/samplerobj.c
+++ b/mesalib/src/mesa/main/samplerobj.c
@@ -51,6 +51,28 @@ _mesa_lookup_samplerobj(struct gl_context *ctx, GLuint name)
 }
 
 
+static inline void
+begin_samplerobj_lookups(struct gl_context *ctx)
+{
+   _mesa_HashLockMutex(ctx->Shared->SamplerObjects);
+}
+
+
+static inline void
+end_samplerobj_lookups(struct gl_context *ctx)
+{
+   _mesa_HashUnlockMutex(ctx->Shared->SamplerObjects);
+}
+
+
+static inline struct gl_sampler_object *
+lookup_samplerobj_locked(struct gl_context *ctx, GLuint name)
+{
+   return (struct gl_sampler_object *)
+         _mesa_HashLookupLocked(ctx->Shared->SamplerObjects, name);
+}
+
+
 /**
  * Handle reference counting.
  */
@@ -285,6 +307,105 @@ _mesa_BindSampler(GLuint unit, GLuint sampler)
 }
 
 
+void GLAPIENTRY
+_mesa_BindSamplers(GLuint first, GLsizei count, const GLuint *samplers)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   /* The ARB_multi_bind spec says:
+    *
+    *   "An INVALID_OPERATION error is generated if <first> + <count> is
+    *    greater than the number of texture image units supported by
+    *    the implementation."
+    */
+   if (first + count > ctx->Const.MaxCombinedTextureImageUnits) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBindSamplers(first=%u + count=%d > the value of "
+                  "GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS=%u)",
+                  first, count, ctx->Const.MaxCombinedTextureImageUnits);
+      return;
+   }
+
+   FLUSH_VERTICES(ctx, 0);
+
+   if (samplers) {
+      /* Note that the error semantics for multi-bind commands differ from
+       * those of other GL commands.
+       *
+       * The Issues section in the ARB_multi_bind spec says:
+       *
+       *    "(11) Typically, OpenGL specifies that if an error is generated by
+       *          a command, that command has no effect.  This is somewhat
+       *          unfortunate for multi-bind commands, because it would require
+       *          a first pass to scan the entire list of bound objects for
+       *          errors and then a second pass to actually perform the
+       *          bindings.  Should we have different error semantics?
+       *
+       *       RESOLVED:  Yes.  In this specification, when the parameters for
+       *       one of the <count> binding points are invalid, that binding
+       *       point is not updated and an error will be generated.  However,
+       *       other binding points in the same command will be updated if
+       *       their parameters are valid and no other error occurs."
+       */
+
+      begin_samplerobj_lookups(ctx);
+
+      for (i = 0; i < count; i++) {
+         const GLuint unit = first + i;
+         struct gl_sampler_object * const currentSampler =
+             ctx->Texture.Unit[unit].Sampler;
+         struct gl_sampler_object *sampObj;
+
+         if (samplers[i] != 0) {
+            if (currentSampler && currentSampler->Name == samplers[i])
+               sampObj = currentSampler;
+            else
+               sampObj = lookup_samplerobj_locked(ctx, samplers[i]);
+
+            /* The ARB_multi_bind spec says:
+             *
+             *    "An INVALID_OPERATION error is generated if any value
+             *     in <samplers> is not zero or the name of an existing
+             *     sampler object (per binding)."
+             */
+            if (!sampObj) {
+               _mesa_error(ctx, GL_INVALID_OPERATION,
+                           "glBindSamplers(samplers[%d]=%u is not zero or "
+                           "the name of an existing sampler object)",
+                           i, samplers[i]);
+               continue;
+            }
+         } else {
+            sampObj = NULL;
+         }
+
+         /* Bind the new sampler */
+         if (sampObj != currentSampler) {
+            _mesa_reference_sampler_object(ctx,
+                                           &ctx->Texture.Unit[unit].Sampler,
+                                           sampObj);
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
+
+      end_samplerobj_lookups(ctx);
+   } else {
+      /* Unbind all samplers in the range <first> through <first>+<count>-1 */
+      for (i = 0; i < count; i++) {
+         const GLuint unit = first + i;
+
+         if (ctx->Texture.Unit[unit].Sampler) {
+            _mesa_reference_sampler_object(ctx,
+                                           &ctx->Texture.Unit[unit].Sampler,
+                                           NULL);
+            ctx->NewState |= _NEW_TEXTURE;
+         }
+      }
+   }
+}
+
+
 /**
  * Check if a coordinate wrap mode is legal.
  * \return GL_TRUE if legal, GL_FALSE otherwise
diff --git a/mesalib/src/mesa/main/samplerobj.h b/mesalib/src/mesa/main/samplerobj.h
index c72b1cd8d..7d80b383d 100644
--- a/mesalib/src/mesa/main/samplerobj.h
+++ b/mesalib/src/mesa/main/samplerobj.h
@@ -81,6 +81,8 @@ _mesa_IsSampler(GLuint sampler);
 void GLAPIENTRY
 _mesa_BindSampler(GLuint unit, GLuint sampler);
 void GLAPIENTRY
+_mesa_BindSamplers(GLuint first, GLsizei count, const GLuint *samplers);
+void GLAPIENTRY
 _mesa_SamplerParameteri(GLuint sampler, GLenum pname, GLint param);
 void GLAPIENTRY
 _mesa_SamplerParameterf(GLuint sampler, GLenum pname, GLfloat param);
diff --git a/mesalib/src/mesa/main/shader_query.cpp b/mesalib/src/mesa/main/shader_query.cpp
index f66c11733..36d1d9cc0 100644
--- a/mesalib/src/mesa/main/shader_query.cpp
+++ b/mesalib/src/mesa/main/shader_query.cpp
@@ -153,6 +153,63 @@ _mesa_GetActiveAttrib(GLhandleARB program, GLuint desired_index,
    _mesa_error(ctx, GL_INVALID_VALUE, "glGetActiveAttrib(index)");
 }
 
+/* Locations associated with shader variables (array or non-array) can be
+ * queried using its base name or using the base name appended with the
+ * valid array index. For example, in case of below vertex shader, valid
+ * queries can be made to know the location of "xyz", "array", "array[0]",
+ * "array[1]", "array[2]" and "array[3]". In this example index reurned
+ * will be 0, 0, 0, 1, 2, 3 respectively.
+ *
+ * [Vertex Shader]
+ * layout(location=0) in vec4 xyz;
+ * layout(location=1) in vec4[4] array;
+ * void main()
+ * { }
+ *
+ * This requirement came up with the addition of ARB_program_interface_query
+ * to OpenGL 4.3 specification. See page 101 (page 122 of the PDF) for details.
+ *
+ * This utility function is used by:
+ * _mesa_GetAttribLocation
+ * _mesa_GetFragDataLocation
+ * _mesa_GetFragDataIndex
+ *
+ * Returns 0:
+ *    if the 'name' string matches var->name.
+ * Returns 'matched index':
+ *    if the 'name' string matches var->name appended with valid array index.
+ */
+int static inline
+get_matching_index(const ir_variable *const var, const char *name) {
+   unsigned idx = 0;
+   const char *const paren = strchr(name, '[');
+   const unsigned len = (paren != NULL) ? paren - name : strlen(name);
+
+   if (paren != NULL) {
+      if (!var->type->is_array())
+         return -1;
+
+      char *endptr;
+      idx = (unsigned) strtol(paren + 1, &endptr, 10);
+      const unsigned idx_len = endptr != (paren + 1) ? endptr - paren - 1 : 0;
+
+      /* Validate the sub string representing index in 'name' string */
+      if ((idx > 0 && paren[1] == '0') /* leading zeroes */
+          || (idx == 0 && idx_len > 1) /* all zeroes */
+          || paren[1] == ' ' /* whitespace */
+          || endptr[0] != ']' /* closing brace */
+          || endptr[1] != '\0' /* null char */
+          || idx_len == 0 /* missing index */
+          || idx >= var->type->length) /* exceeding array bound */
+         return -1;
+   }
+
+   if (strncmp(var->name, name, len) == 0 && var->name[len] == '\0')
+      return idx;
+
+   return -1;
+}
+
 GLint GLAPIENTRY
 _mesa_GetAttribLocation(GLhandleARB program, const GLcharARB * name)
 {
@@ -196,8 +253,10 @@ _mesa_GetAttribLocation(GLhandleARB program, const GLcharARB * name)
 	  || var->data.location < VERT_ATTRIB_GENERIC0)
 	 continue;
 
-      if (strcmp(var->name, name) == 0)
-	 return var->data.location - VERT_ATTRIB_GENERIC0;
+      int index = get_matching_index(var, (const char *) name);
+
+      if (index >= 0)
+         return var->data.location + index - VERT_ATTRIB_GENERIC0;
    }
 
    return -1;
@@ -358,7 +417,7 @@ _mesa_GetFragDataIndex(GLuint program, const GLchar *name)
           || var->data.location < FRAG_RESULT_DATA0)
          continue;
 
-      if (strcmp(var->name, name) == 0)
+      if (get_matching_index(var, (const char *) name) >= 0)
          return var->data.index;
    }
 
@@ -414,8 +473,10 @@ _mesa_GetFragDataLocation(GLuint program, const GLchar *name)
 	  || var->data.location < FRAG_RESULT_DATA0)
 	 continue;
 
-      if (strcmp(var->name, name) == 0)
-	 return var->data.location - FRAG_RESULT_DATA0;
+      int index = get_matching_index(var, (const char *) name);
+
+      if (index >= 0)
+         return var->data.location + index - FRAG_RESULT_DATA0;
    }
 
    return -1;
diff --git a/mesalib/src/mesa/main/shaderapi.c b/mesalib/src/mesa/main/shaderapi.c
index 1c8e6b4cf..28739daeb 100644
--- a/mesalib/src/mesa/main/shaderapi.c
+++ b/mesalib/src/mesa/main/shaderapi.c
@@ -700,9 +700,6 @@ get_programiv(struct gl_context *ctx, GLuint program, GLenum pname, GLint *param
       return;
    }
    case GL_PROGRAM_SEPARABLE:
-      if (!ctx->Extensions.ARB_separate_shader_objects)
-         break;
-
       *params = shProg->SeparateShader;
       return;
    default:
@@ -795,7 +792,7 @@ get_shader_source(struct gl_context *ctx, GLuint shader, GLsizei maxLength,
 
 /**
  * Set/replace shader source code.  A helper function used by
- * glShaderSource[ARB] and glCreateShaderProgramEXT.
+ * glShaderSource[ARB].
  */
 static void
 shader_source(struct gl_context *ctx, GLuint shader, const GLchar *source)
@@ -1548,15 +1545,14 @@ _mesa_UseProgram(GLhandleARB program)
       shProg = NULL;
    }
 
-   /* The "Dependencies on EXT_separate_shader_objects" section of the
-    * ARB_separate_shader_object spec says:
+   /* The ARB_separate_shader_object spec says:
     *
     *     "The executable code for an individual shader stage is taken from
     *     the current program for that stage.  If there is a current program
-    *     object for any shader stage or for uniform updates established by
-    *     UseProgram, UseShaderProgramEXT, or ActiveProgramEXT, the current
-    *     program for that stage (if any) is considered current.  Otherwise,
-    *     if there is a bound program pipeline object ..."
+    *     object established by UseProgram, that program is considered current
+    *     for all stages.  Otherwise, if there is a bound program pipeline
+    *     object (section 2.14.PPO), the program bound to the appropriate
+    *     stage of the pipeline object is considered current."
     */
    if (program) {
       /* Attach shader state to the binding point */
@@ -1777,9 +1773,6 @@ _mesa_ProgramParameteri(GLuint program, GLenum pname, GLint value)
       return;
 
    case GL_PROGRAM_SEPARABLE:
-      if (!ctx->Extensions.ARB_separate_shader_objects)
-         break;
-
       /* Spec imply that the behavior is the same as ARB_get_program_binary
        * Chapter 7.3 Program Objects
        */
@@ -1814,124 +1807,6 @@ _mesa_use_shader_program(struct gl_context *ctx, GLenum type,
 }
 
 
-/**
- * For GL_EXT_separate_shader_objects
- */
-void GLAPIENTRY
-_mesa_UseShaderProgramEXT(GLenum type, GLuint program)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   struct gl_shader_program *shProg = NULL;
-
-   if (!_mesa_validate_shader_target(ctx, type)) {
-      _mesa_error(ctx, GL_INVALID_ENUM, "glUseShaderProgramEXT(type)");
-      return;
-   }
-
-   if (_mesa_is_xfb_active_and_unpaused(ctx)) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glUseShaderProgramEXT(transform feedback is active)");
-      return;
-   }
-
-   if (program) {
-      shProg = _mesa_lookup_shader_program_err(ctx, program,
-					       "glUseShaderProgramEXT");
-      if (shProg == NULL)
-	 return;
-
-      if (!shProg->LinkStatus) {
-	 _mesa_error(ctx, GL_INVALID_OPERATION,
-		     "glUseShaderProgramEXT(program not linked)");
-	 return;
-      }
-   }
-
-   /* The "Dependencies on EXT_separate_shader_objects" section of the
-    * ARB_separate_shader_object spec says:
-    *
-    *     "The executable code for an individual shader stage is taken from
-    *     the current program for that stage.  If there is a current program
-    *     object for any shader stage or for uniform updates established by
-    *     UseProgram, UseShaderProgramEXT, or ActiveProgramEXT, the current
-    *     program for that stage (if any) is considered current.  Otherwise,
-    *     if there is a bound program pipeline object ..."
-    */
-   if (program) {
-      /* Attach shader state to the binding point */
-      _mesa_reference_pipeline_object(ctx, &ctx->_Shader, &ctx->Shader);
-      /* Update the program */
-      _mesa_use_shader_program(ctx, type, shProg, ctx->_Shader);
-   } else {
-      /* Must be done first: detach the progam */
-      _mesa_use_shader_program(ctx, type, shProg, ctx->_Shader);
-
-      /* Nothing remains current */
-      if (!ctx->Shader.CurrentProgram[MESA_SHADER_VERTEX] &&
-          !ctx->Shader.CurrentProgram[MESA_SHADER_GEOMETRY] &&
-          !ctx->Shader.CurrentProgram[MESA_SHADER_FRAGMENT] &&
-          !ctx->Shader.ActiveProgram) {
-
-         /* Unattach shader_state binding point */
-         _mesa_reference_pipeline_object(ctx, &ctx->_Shader,
-                                         ctx->Pipeline.Default);
-
-         /* If a pipeline was bound, rebind it */
-         if (ctx->Pipeline.Current) {
-            _mesa_BindProgramPipeline(ctx->Pipeline.Current->Name);
-         }
-      }
-   }
-}
-
-
-/**
- * For GL_EXT_separate_shader_objects
- */
-void GLAPIENTRY
-_mesa_ActiveProgramEXT(GLuint program)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   struct gl_shader_program *shProg = (program != 0)
-      ? _mesa_lookup_shader_program_err(ctx, program, "glActiveProgramEXT")
-      : NULL;
-
-   /* The "Dependencies on EXT_separate_shader_objects" section of the
-    * ARB_separate_shader_object spec says:
-    *
-    *     "The executable code for an individual shader stage is taken from
-    *     the current program for that stage.  If there is a current program
-    *     object for any shader stage or for uniform updates established by
-    *     UseProgram, UseShaderProgramEXT, or ActiveProgramEXT, the current
-    *     program for that stage (if any) is considered current.  Otherwise,
-    *     if there is a bound program pipeline object ..."
-    */
-   if (shProg != NULL) {
-      /* Attach shader state to the binding point */
-      _mesa_reference_pipeline_object(ctx, &ctx->_Shader, &ctx->Shader);
-      _mesa_active_program(ctx, shProg, "glActiveProgramEXT");
-   } else {
-      /* Must be done first: unset the current active progam */
-      _mesa_active_program(ctx, shProg, "glActiveProgramEXT");
-
-      /* Nothing remains current */
-      if (!ctx->Shader.CurrentProgram[MESA_SHADER_VERTEX] &&
-          !ctx->Shader.CurrentProgram[MESA_SHADER_GEOMETRY] &&
-          !ctx->Shader.CurrentProgram[MESA_SHADER_FRAGMENT] &&
-          !ctx->Shader.ActiveProgram) {
-
-         /* Unattach shader_state binding point */
-         _mesa_reference_pipeline_object(ctx, &ctx->_Shader, ctx->Pipeline.Default);
-         /* If a pipeline was bound, rebind it */
-         if (ctx->Pipeline.Current) {
-            _mesa_BindProgramPipeline(ctx->Pipeline.Current->Name);
-         }
-      }
-   }
-
-   return;
-}
-
 static GLuint
 _mesa_create_shader_program(struct gl_context* ctx, GLboolean separate,
                             GLenum type, GLsizei count, const GLchar* const *strings)
@@ -2004,6 +1879,11 @@ _mesa_copy_linked_program_data(gl_shader_stage type,
       dst_gp->UsesEndPrimitive = src->Geom.UsesEndPrimitive;
    }
       break;
+   case MESA_SHADER_FRAGMENT: {
+      struct gl_fragment_program *dst_fp = (struct gl_fragment_program *) dst;
+      dst_fp->FragDepthLayout = src->FragDepthLayout;
+   }
+      break;
    case MESA_SHADER_COMPUTE: {
       struct gl_compute_program *dst_cp = (struct gl_compute_program *) dst;
       int i;
@@ -2016,23 +1896,8 @@ _mesa_copy_linked_program_data(gl_shader_stage type,
    }
 }
 
-
-/**
- * For GL_EXT_separate_shader_objects
- */
-GLuint GLAPIENTRY
-_mesa_CreateShaderProgramEXT(GLenum type, const GLchar *string)
-{
-   GET_CURRENT_CONTEXT(ctx);
-
-   return _mesa_create_shader_program(ctx, GL_FALSE, type, 1, &string);
-}
-
 /**
  * ARB_separate_shader_objects: Compile & Link Program
- *
- * Basically the same as _mesa_CreateShaderProgramEXT but with support of
- * multiple strings and sets the SeparateShader flag to true.
  */
 GLuint GLAPIENTRY
 _mesa_CreateShaderProgramv(GLenum type, GLsizei count,
diff --git a/mesalib/src/mesa/main/shaderimage.c b/mesalib/src/mesa/main/shaderimage.c
index d1e752d3d..9e62f4294 100644
--- a/mesalib/src/mesa/main/shaderimage.c
+++ b/mesalib/src/mesa/main/shaderimage.c
@@ -33,6 +33,7 @@
 #include "context.h"
 #include "texobj.h"
 #include "teximage.h"
+#include "enums.h"
 
 /*
  * Define endian-invariant aliases for some mesa formats that are
@@ -479,6 +480,148 @@ _mesa_BindImageTexture(GLuint unit, GLuint texture, GLint level,
 }
 
 void GLAPIENTRY
+_mesa_BindImageTextures(GLuint first, GLsizei count, const GLuint *textures)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   int i;
+
+   if (!ctx->Extensions.ARB_shader_image_load_store) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glBindImageTextures()");
+      return;
+   }
+
+   if (first + count > ctx->Const.MaxImageUnits) {
+      /* The ARB_multi_bind spec says:
+       *
+       *    "An INVALID_OPERATION error is generated if <first> + <count>
+       *     is greater than the number of image units supported by
+       *     the implementation."
+       */
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBindImageTextures(first=%u + count=%d > the value of "
+                  "GL_MAX_IMAGE_UNITS=%u)",
+                  first, count, ctx->Const.MaxImageUnits);
+      return;
+   }
+
+   /* Assume that at least one binding will be changed */
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |= ctx->DriverFlags.NewImageUnits;
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by
+    *          a command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require
+    *          a first pass to scan the entire list of bound objects for
+    *          errors and then a second pass to actually perform the
+    *          bindings.  Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding
+    *       point is not updated and an error will be generated.  However,
+    *       other binding points in the same command will be updated if
+    *       their parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_texture_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_image_unit *u = &ctx->ImageUnits[first + i];
+      const GLuint texture = textures ? textures[i] : 0;
+
+      if (texture != 0) {
+         struct gl_texture_object *texObj;
+         struct gl_texture_image *image;
+         mesa_format actualFormat;
+
+         if (!u->TexObj || u->TexObj->Name != texture) {
+            texObj = _mesa_lookup_texture_locked(ctx, texture);
+            if (!texObj) {
+               /* The ARB_multi_bind spec says:
+                *
+                *    "An INVALID_OPERATION error is generated if any value
+                *     in <textures> is not zero or the name of an existing
+                *     texture object (per binding)."
+                */
+               _mesa_error(ctx, GL_INVALID_OPERATION,
+                           "glBindImageTextures(textures[%d]=%u "
+                           "is not zero or the name of an existing texture "
+                           "object)", i, texture);
+               continue;
+            }
+         } else {
+            texObj = u->TexObj;
+         }
+
+         image = texObj->Image[0][0];
+
+         if (!image || image->Width == 0 || image->Height == 0 || image->Depth == 0) {
+            /* The ARB_multi_bind spec says:
+             *
+             *    "An INVALID_OPERATION error is generated if the width,
+             *     height, or depth of the level zero texture image of
+             *     any texture in <textures> is zero (per binding)."
+             */
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "glBindImageTextures(the width, height or depth "
+                        "of the level zero texture image of "
+                        "textures[%d]=%u is zero)", i, texture);
+            continue;
+         }
+
+         actualFormat = get_image_format(image->InternalFormat);
+
+         if (actualFormat == MESA_FORMAT_NONE) {
+            /* The ARB_multi_bind spec says:
+             *
+             *   "An INVALID_OPERATION error is generated if the internal
+             *    format of the level zero texture image of any texture
+             *    in <textures> is not found in table 8.33 (per binding)."
+             */
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "glBindImageTextures(the internal format %s of "
+                        "the level zero texture image of textures[%d]=%u "
+                        "is not supported)",
+                        _mesa_lookup_enum_by_nr(image->InternalFormat),
+                        i, texture);
+            continue;
+         }
+
+         /* Update the texture binding */
+         _mesa_reference_texobj(&u->TexObj, texObj);
+         u->Level = 0;
+         u->Layered = _mesa_tex_target_is_layered(texObj->Target);
+         u->Layer = 0;
+         u->Access = GL_READ_WRITE;
+         u->Format = image->InternalFormat;
+         u->_ActualFormat = actualFormat;
+         u->_Valid = validate_image_unit(ctx, u);
+      } else {
+         /* Unbind the texture from the unit */
+         _mesa_reference_texobj(&u->TexObj, NULL);
+         u->Level = 0;
+         u->Layered = GL_FALSE;
+         u->Layer = 0;
+         u->Access = GL_READ_ONLY;
+         u->Format = GL_R8;
+         u->_ActualFormat = MESA_FORMAT_R_UNORM8;
+         u->_Valid = GL_FALSE;
+      }
+
+      /* Pass the BindImageTexture call down to the device driver */
+      if (ctx->Driver.BindImageTexture)
+         ctx->Driver.BindImageTexture(ctx, u, u->TexObj, u->Level, u->Layered,
+                                      u->Layer, u->Access, u->Format);
+   }
+
+   _mesa_end_texture_lookups(ctx);
+}
+
+void GLAPIENTRY
 _mesa_MemoryBarrier(GLbitfield barriers)
 {
    GET_CURRENT_CONTEXT(ctx);
diff --git a/mesalib/src/mesa/main/shaderimage.h b/mesalib/src/mesa/main/shaderimage.h
index aaecc5d29..733ac7747 100644
--- a/mesalib/src/mesa/main/shaderimage.h
+++ b/mesalib/src/mesa/main/shaderimage.h
@@ -46,6 +46,9 @@ _mesa_BindImageTexture(GLuint unit, GLuint texture, GLint level,
                        GLenum format);
 
 void GLAPIENTRY
+_mesa_BindImageTextures(GLuint first, GLsizei count, const GLuint *textures);
+
+void GLAPIENTRY
 _mesa_MemoryBarrier(GLbitfield barriers);
 
 #endif
diff --git a/mesalib/src/mesa/main/texgetimage.c b/mesalib/src/mesa/main/texgetimage.c
index 2beb0abe6..1ac707d13 100644
--- a/mesalib/src/mesa/main/texgetimage.c
+++ b/mesalib/src/mesa/main/texgetimage.c
@@ -131,8 +131,8 @@ get_tex_depth_stencil(struct gl_context *ctx, GLuint dimensions,
    GLint img, row;
 
    assert(format == GL_DEPTH_STENCIL);
-   assert(type == GL_UNSIGNED_INT_24_8);
-   /* XXX type == GL_FLOAT_32_UNSIGNED_INT_24_8_REV is not handled yet */
+   assert(type == GL_UNSIGNED_INT_24_8 ||
+          type == GL_FLOAT_32_UNSIGNED_INT_24_8_REV);
 
    for (img = 0; img < depth; img++) {
       GLubyte *srcMap;
@@ -149,11 +149,10 @@ get_tex_depth_stencil(struct gl_context *ctx, GLuint dimensions,
             void *dest = _mesa_image_address(dimensions, &ctx->Pack, pixels,
                                              width, height, format, type,
                                              img, row, 0);
-            /* Unpack from texture's format to GL's z24_s8 layout */
-            _mesa_unpack_uint_24_8_depth_stencil_row(texImage->TexFormat,
-                                                     width,
-                                                     (const GLuint *) src,
-                                                     dest);
+            _mesa_unpack_depth_stencil_row(texImage->TexFormat,
+                                           width,
+                                           (const GLuint *) src,
+                                           type, dest);
             if (ctx->Pack.SwapBytes) {
                _mesa_swap4((GLuint *) dest, width);
             }
@@ -838,6 +837,11 @@ getteximage_error_check(struct gl_context *ctx, GLenum target, GLint level,
       _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)");
       return GL_TRUE;
    }
+   else if (_mesa_is_stencil_format(format)
+            && !ctx->Extensions.ARB_texture_stencil8) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format=GL_STENCIL_INDEX)");
+      return GL_TRUE;
+   }
    else if (_mesa_is_ycbcr_format(format)
             && !_mesa_is_ycbcr_format(baseFormat)) {
       _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)");
diff --git a/mesalib/src/mesa/main/teximage.c b/mesalib/src/mesa/main/teximage.c
index c7f301cbd..845ba8014 100644
--- a/mesalib/src/mesa/main/teximage.c
+++ b/mesalib/src/mesa/main/teximage.c
@@ -3024,7 +3024,7 @@ _mesa_choose_texture_format(struct gl_context *ctx,
    }
 
    /* choose format from scratch */
-   f = ctx->Driver.ChooseTextureFormat(ctx, texObj->Target, internalFormat,
+   f = ctx->Driver.ChooseTextureFormat(ctx, target, internalFormat,
                                        format, type);
    ASSERT(f != MESA_FORMAT_NONE);
    return f;
diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c
index 85246c8ab..2a82c2d4e 100644
--- a/mesalib/src/mesa/main/texobj.c
+++ b/mesalib/src/mesa/main/texobj.c
@@ -61,6 +61,27 @@ _mesa_lookup_texture(struct gl_context *ctx, GLuint id)
 }
 
 
+void
+_mesa_begin_texture_lookups(struct gl_context *ctx)
+{
+   _mesa_HashLockMutex(ctx->Shared->TexObjects);
+}
+
+
+void
+_mesa_end_texture_lookups(struct gl_context *ctx)
+{
+   _mesa_HashUnlockMutex(ctx->Shared->TexObjects);
+}
+
+
+struct gl_texture_object *
+_mesa_lookup_texture_locked(struct gl_context *ctx, GLuint id)
+{
+   return (struct gl_texture_object *)
+      _mesa_HashLookupLocked(ctx->Shared->TexObjects, id);
+}
+
 
 /**
  * Allocate and initialize a new texture object.  But don't put it into the
@@ -260,6 +281,7 @@ _mesa_copy_texture_object( struct gl_texture_object *dest,
                            const struct gl_texture_object *src )
 {
    dest->Target = src->Target;
+   dest->TargetIndex = src->TargetIndex;
    dest->Name = src->Name;
    dest->Priority = src->Priority;
    dest->Sampler.BorderColor.f[0] = src->Sampler.BorderColor.f[0];
@@ -1092,17 +1114,20 @@ static void
 unbind_texobj_from_texunits(struct gl_context *ctx,
                             struct gl_texture_object *texObj)
 {
-   GLuint u, tex;
+   const gl_texture_index index = texObj->TargetIndex;
+   GLuint u;
+
+   if (texObj->Target == 0)
+      return;
 
    for (u = 0; u < ctx->Texture.NumCurrentTexUsed; u++) {
       struct gl_texture_unit *unit = &ctx->Texture.Unit[u];
-      for (tex = 0; tex < NUM_TEXTURE_TARGETS; tex++) {
-         if (texObj == unit->CurrentTex[tex]) {
-            _mesa_reference_texobj(&unit->CurrentTex[tex],
-                                   ctx->Shared->DefaultTex[tex]);
-            ASSERT(unit->CurrentTex[tex]);
-            break;
-         }
+
+      if (texObj == unit->CurrentTex[index]) {
+         /* Bind the default texture for this unit/target */
+         _mesa_reference_texobj(&unit->CurrentTex[index],
+                                ctx->Shared->DefaultTex[index]);
+         unit->_BoundTextures &= ~(1 << index);
       }
    }
 }
@@ -1126,6 +1151,28 @@ unbind_texobj_from_image_units(struct gl_context *ctx,
    }
 }
 
+/**
+ * Unbinds all textures bound to the given texture image unit.
+ */
+static void
+unbind_textures_from_unit(struct gl_context *ctx, GLuint unit)
+{
+   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+
+   while (texUnit->_BoundTextures) {
+      const GLuint index = ffs(texUnit->_BoundTextures) - 1;
+      struct gl_texture_object *texObj = ctx->Shared->DefaultTex[index];
+
+      _mesa_reference_texobj(&texUnit->CurrentTex[index], texObj);
+
+      /* Pass BindTexture call to device driver */
+      if (ctx->Driver.BindTexture)
+         ctx->Driver.BindTexture(ctx, unit, 0, texObj);
+
+      texUnit->_BoundTextures &= ~(1 << index);
+      ctx->NewState |= _NEW_TEXTURE;
+   }
+}
 
 /**
  * Delete named textures.
@@ -1327,6 +1374,7 @@ _mesa_BindTexture( GLenum target, GLuint texName )
          mtx_unlock(&ctx->Shared->Mutex);
       }
       newTexObj->Target = target;
+      newTexObj->TargetIndex = targetIndex;
    }
 
    assert(valid_texture_object(newTexObj));
@@ -1357,9 +1405,118 @@ _mesa_BindTexture( GLenum target, GLuint texName )
                                          ctx->Texture.CurrentUnit + 1);
    ASSERT(texUnit->CurrentTex[targetIndex]);
 
+   if (texName != 0)
+      texUnit->_BoundTextures |= (1 << targetIndex);
+   else
+      texUnit->_BoundTextures &= ~(1 << targetIndex);
+
    /* Pass BindTexture call to device driver */
    if (ctx->Driver.BindTexture)
-      ctx->Driver.BindTexture(ctx, target, newTexObj);
+      ctx->Driver.BindTexture(ctx, ctx->Texture.CurrentUnit, target, newTexObj);
+}
+
+
+void GLAPIENTRY
+_mesa_BindTextures(GLuint first, GLsizei count, const GLuint *textures)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   /* The ARB_multi_bind spec says:
+    *
+    *     "An INVALID_OPERATION error is generated if <first> + <count>
+    *      is greater than the number of texture image units supported
+    *      by the implementation."
+    */
+   if (first + count > ctx->Const.MaxCombinedTextureImageUnits) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBindTextures(first=%u + count=%d > the value of "
+                  "GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS=%u)",
+                  first, count, ctx->Const.MaxCombinedTextureImageUnits);
+      return;
+   }
+
+   /* Flush before changing bindings */
+   FLUSH_VERTICES(ctx, 0);
+
+   ctx->Texture.NumCurrentTexUsed = MAX2(ctx->Texture.NumCurrentTexUsed,
+                                         first + count);
+
+   if (textures) {
+      /* Note that the error semantics for multi-bind commands differ from
+       * those of other GL commands.
+       *
+       * The issues section in the ARB_multi_bind spec says:
+       *
+       *    "(11) Typically, OpenGL specifies that if an error is generated by
+       *          a command, that command has no effect.  This is somewhat
+       *          unfortunate for multi-bind commands, because it would require
+       *          a first pass to scan the entire list of bound objects for
+       *          errors and then a second pass to actually perform the
+       *          bindings.  Should we have different error semantics?
+       *
+       *       RESOLVED:  Yes.  In this specification, when the parameters for
+       *       one of the <count> binding points are invalid, that binding
+       *       point is not updated and an error will be generated.  However,
+       *       other binding points in the same command will be updated if
+       *       their parameters are valid and no other error occurs."
+       */
+
+      _mesa_begin_texture_lookups(ctx);
+
+      for (i = 0; i < count; i++) {
+         if (textures[i] != 0) {
+            struct gl_texture_unit *texUnit = &ctx->Texture.Unit[first + i];
+            struct gl_texture_object *current = texUnit->_Current;
+            struct gl_texture_object *texObj;
+
+            if (current && current->Name == textures[i])
+               texObj = current;
+            else
+               texObj = _mesa_lookup_texture_locked(ctx, textures[i]);
+
+            if (texObj && texObj->Target != 0) {
+               const gl_texture_index targetIndex = texObj->TargetIndex;
+
+               if (texUnit->CurrentTex[targetIndex] != texObj) {
+                  /* Do the actual binding.  The refcount on the previously
+                   * bound texture object will be decremented.  It will be
+                   * deleted if the count hits zero.
+                   */
+                  _mesa_reference_texobj(&texUnit->CurrentTex[targetIndex],
+                                         texObj);
+
+                  texUnit->_BoundTextures |= (1 << targetIndex);
+                  ctx->NewState |= _NEW_TEXTURE;
+
+                  /* Pass the BindTexture call to the device driver */
+                  if (ctx->Driver.BindTexture)
+                     ctx->Driver.BindTexture(ctx, first + i,
+                                             texObj->Target, texObj);
+               }
+            } else {
+               /* The ARB_multi_bind spec says:
+                *
+                *     "An INVALID_OPERATION error is generated if any value
+                *      in <textures> is not zero or the name of an existing
+                *      texture object (per binding)."
+                */
+               _mesa_error(ctx, GL_INVALID_OPERATION,
+                           "glBindTextures(textures[%d]=%u is not zero "
+                           "or the name of an existing texture object)",
+                           i, textures[i]);
+            }
+         } else {
+            unbind_textures_from_unit(ctx, first + i);
+         }
+      }
+
+      _mesa_end_texture_lookups(ctx);
+   } else {
+      /* Unbind all textures in the range <first> through <first>+<count>-1 */
+      for (i = 0; i < count; i++)
+         unbind_textures_from_unit(ctx, first + i);
+   }
 }
 
 
diff --git a/mesalib/src/mesa/main/texobj.h b/mesalib/src/mesa/main/texobj.h
index a9de73fff..b1b7a3027 100644
--- a/mesalib/src/mesa/main/texobj.h
+++ b/mesalib/src/mesa/main/texobj.h
@@ -46,6 +46,15 @@
 extern struct gl_texture_object *
 _mesa_lookup_texture(struct gl_context *ctx, GLuint id);
 
+extern void
+_mesa_begin_texture_lookups(struct gl_context *ctx);
+
+extern void
+_mesa_end_texture_lookups(struct gl_context *ctx);
+
+extern struct gl_texture_object *
+_mesa_lookup_texture_locked(struct gl_context *ctx, GLuint id);
+
 extern struct gl_texture_object *
 _mesa_new_texture_object( struct gl_context *ctx, GLuint name, GLenum target );
 
@@ -177,6 +186,10 @@ _mesa_BindTexture( GLenum target, GLuint texture );
 
 
 extern void GLAPIENTRY
+_mesa_BindTextures( GLuint first, GLsizei count, const GLuint *textures );
+
+
+extern void GLAPIENTRY
 _mesa_PrioritizeTextures( GLsizei n, const GLuint *textures,
                           const GLclampf *priorities );
 
diff --git a/mesalib/src/mesa/main/texstate.c b/mesalib/src/mesa/main/texstate.c
index 91b290691..19508cf39 100644
--- a/mesalib/src/mesa/main/texstate.c
+++ b/mesalib/src/mesa/main/texstate.c
@@ -113,6 +113,7 @@ _mesa_copy_texture_state( const struct gl_context *src, struct gl_context *dst )
                   MAX2(dst->Texture.NumCurrentTexUsed, u + 1);
             }
          }
+         dst->Texture.Unit[u]._BoundTextures = src->Texture.Unit[u]._BoundTextures;
          _mesa_unlock_context_textures(dst);
       }
    }
@@ -877,6 +878,8 @@ init_texture_unit( struct gl_context *ctx, GLuint unit )
       _mesa_reference_texobj(&texUnit->CurrentTex[tex],
                              ctx->Shared->DefaultTex[tex]);
    }
+
+   texUnit->_BoundTextures = 0;
 }
 
 
diff --git a/mesalib/src/mesa/main/texstore.c b/mesalib/src/mesa/main/texstore.c
index d9096abf3..764214669 100644
--- a/mesalib/src/mesa/main/texstore.c
+++ b/mesalib/src/mesa/main/texstore.c
@@ -2473,74 +2473,71 @@ _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
    const GLint srcRowStride
       = _mesa_image_row_stride(srcPacking, srcWidth, srcFormat, srcType);
    GLint img, row;
+   GLuint *depth = malloc(srcWidth * sizeof(GLuint));
+   GLubyte *stencil = malloc(srcWidth * sizeof(GLubyte));
 
    ASSERT(dstFormat == MESA_FORMAT_S8_UINT_Z24_UNORM);
    ASSERT(srcFormat == GL_DEPTH_STENCIL_EXT ||
           srcFormat == GL_DEPTH_COMPONENT ||
           srcFormat == GL_STENCIL_INDEX);
-   ASSERT(srcFormat != GL_DEPTH_STENCIL_EXT || srcType == GL_UNSIGNED_INT_24_8_EXT);
+   ASSERT(srcFormat != GL_DEPTH_STENCIL_EXT ||
+          srcType == GL_UNSIGNED_INT_24_8_EXT ||
+          srcType == GL_FLOAT_32_UNSIGNED_INT_24_8_REV);
 
-   if (srcFormat == GL_DEPTH_COMPONENT ||
-       srcFormat == GL_STENCIL_INDEX) {
-      GLuint *depth = malloc(srcWidth * sizeof(GLuint));
-      GLubyte *stencil = malloc(srcWidth * sizeof(GLubyte));
+   if (!depth || !stencil) {
+      free(depth);
+      free(stencil);
+      return GL_FALSE;
+   }
 
-      if (!depth || !stencil) {
-         free(depth);
-         free(stencil);
-         return GL_FALSE;
-      }
+   /* In case we only upload depth we need to preserve the stencil */
+   for (img = 0; img < srcDepth; img++) {
+      GLuint *dstRow = (GLuint *) dstSlices[img];
+      const GLubyte *src
+         = (const GLubyte *) _mesa_image_address(dims, srcPacking, srcAddr,
+               srcWidth, srcHeight,
+               srcFormat, srcType,
+               img, 0, 0);
+      for (row = 0; row < srcHeight; row++) {
+         GLint i;
+         GLboolean keepdepth = GL_FALSE, keepstencil = GL_FALSE;
 
-      /* In case we only upload depth we need to preserve the stencil */
-      for (img = 0; img < srcDepth; img++) {
-	 GLuint *dstRow = (GLuint *) dstSlices[img];
-         const GLubyte *src
-            = (const GLubyte *) _mesa_image_address(dims, srcPacking, srcAddr,
-                  srcWidth, srcHeight,
-                  srcFormat, srcType,
-                  img, 0, 0);
-         for (row = 0; row < srcHeight; row++) {
-            GLint i;
-	    GLboolean keepdepth = GL_FALSE, keepstencil = GL_FALSE;
+         if (srcFormat == GL_DEPTH_COMPONENT) { /* preserve stencil */
+            keepstencil = GL_TRUE;
+         }
+         else if (srcFormat == GL_STENCIL_INDEX) { /* preserve depth */
+            keepdepth = GL_TRUE;
+         }
 
-	    if (srcFormat == GL_DEPTH_COMPONENT) { /* preserve stencil */
-	       keepstencil = GL_TRUE;
-	    }
-            else if (srcFormat == GL_STENCIL_INDEX) { /* preserve depth */
-	       keepdepth = GL_TRUE;
-	    }
+         if (keepdepth == GL_FALSE)
+            /* the 24 depth bits will be in the low position: */
+            _mesa_unpack_depth_span(ctx, srcWidth,
+                                    GL_UNSIGNED_INT, /* dst type */
+                                    keepstencil ? depth : dstRow, /* dst addr */
+                                    depthScale,
+                                    srcType, src, srcPacking);
 
-	    if (keepdepth == GL_FALSE)
-	       /* the 24 depth bits will be in the low position: */
-	       _mesa_unpack_depth_span(ctx, srcWidth,
-				       GL_UNSIGNED_INT, /* dst type */
-				       keepstencil ? depth : dstRow, /* dst addr */
-				       depthScale,
-				       srcType, src, srcPacking);
-
-	    if (keepstencil == GL_FALSE)
-	       /* get the 8-bit stencil values */
-	       _mesa_unpack_stencil_span(ctx, srcWidth,
-					 GL_UNSIGNED_BYTE, /* dst type */
-					 stencil, /* dst addr */
-					 srcType, src, srcPacking,
-					 ctx->_ImageTransferState);
-
-	    for (i = 0; i < srcWidth; i++) {
-	       if (keepstencil)
-		  dstRow[i] = depth[i] << 8 | (dstRow[i] & 0x000000FF);
-	       else
-		  dstRow[i] = (dstRow[i] & 0xFFFFFF00) | (stencil[i] & 0xFF);
-	    }
+         if (keepstencil == GL_FALSE)
+            /* get the 8-bit stencil values */
+            _mesa_unpack_stencil_span(ctx, srcWidth,
+                                      GL_UNSIGNED_BYTE, /* dst type */
+                                      stencil, /* dst addr */
+                                      srcType, src, srcPacking,
+                                      ctx->_ImageTransferState);
 
-            src += srcRowStride;
-            dstRow += dstRowStride / sizeof(GLuint);
+         for (i = 0; i < srcWidth; i++) {
+            if (keepstencil)
+               dstRow[i] = depth[i] << 8 | (dstRow[i] & 0x000000FF);
+            else
+               dstRow[i] = (dstRow[i] & 0xFFFFFF00) | (stencil[i] & 0xFF);
          }
+         src += srcRowStride;
+         dstRow += dstRowStride / sizeof(GLuint);
       }
-
-      free(depth);
-      free(stencil);
    }
+
+   free(depth);
+   free(stencil);
    return GL_TRUE;
 }
 
@@ -2563,7 +2560,8 @@ _mesa_texstore_s8_z24(TEXSTORE_PARAMS)
           srcFormat == GL_DEPTH_COMPONENT ||
           srcFormat == GL_STENCIL_INDEX);
    ASSERT(srcFormat != GL_DEPTH_STENCIL_EXT ||
-          srcType == GL_UNSIGNED_INT_24_8_EXT);
+          srcType == GL_UNSIGNED_INT_24_8_EXT ||
+          srcType == GL_FLOAT_32_UNSIGNED_INT_24_8_REV);
 
    depth = malloc(srcWidth * sizeof(GLuint));
    stencil = malloc(srcWidth * sizeof(GLubyte));
@@ -3417,49 +3415,47 @@ _mesa_texstore_r11_g11_b10f(TEXSTORE_PARAMS)
 static GLboolean
 _mesa_texstore_z32f_x24s8(TEXSTORE_PARAMS)
 {
+   GLint img, row;
+   const GLint srcRowStride
+      = _mesa_image_row_stride(srcPacking, srcWidth, srcFormat, srcType)
+      / sizeof(uint64_t);
+
    ASSERT(dstFormat == MESA_FORMAT_Z32_FLOAT_S8X24_UINT);
    ASSERT(srcFormat == GL_DEPTH_STENCIL ||
           srcFormat == GL_DEPTH_COMPONENT ||
           srcFormat == GL_STENCIL_INDEX);
    ASSERT(srcFormat != GL_DEPTH_STENCIL ||
+          srcType == GL_UNSIGNED_INT_24_8 ||
           srcType == GL_FLOAT_32_UNSIGNED_INT_24_8_REV);
 
-   if (srcFormat == GL_DEPTH_COMPONENT ||
-       srcFormat == GL_STENCIL_INDEX) {
-      GLint img, row;
-      const GLint srcRowStride
-         = _mesa_image_row_stride(srcPacking, srcWidth, srcFormat, srcType)
-         / sizeof(uint64_t);
+   /* In case we only upload depth we need to preserve the stencil */
+   for (img = 0; img < srcDepth; img++) {
+      uint64_t *dstRow = (uint64_t *) dstSlices[img];
+      const uint64_t *src
+         = (const uint64_t *) _mesa_image_address(dims, srcPacking, srcAddr,
+               srcWidth, srcHeight,
+               srcFormat, srcType,
+               img, 0, 0);
+      for (row = 0; row < srcHeight; row++) {
+         /* The unpack functions with:
+          *    dstType = GL_FLOAT_32_UNSIGNED_INT_24_8_REV
+          * only write their own dword, so the other dword (stencil
+          * or depth) is preserved. */
+         if (srcFormat != GL_STENCIL_INDEX)
+            _mesa_unpack_depth_span(ctx, srcWidth,
+                                    GL_FLOAT_32_UNSIGNED_INT_24_8_REV, /* dst type */
+                                    dstRow, /* dst addr */
+                                    ~0U, srcType, src, srcPacking);
 
-      /* In case we only upload depth we need to preserve the stencil */
-      for (img = 0; img < srcDepth; img++) {
-         uint64_t *dstRow = (uint64_t *) dstSlices[img];
-         const uint64_t *src
-            = (const uint64_t *) _mesa_image_address(dims, srcPacking, srcAddr,
-                  srcWidth, srcHeight,
-                  srcFormat, srcType,
-                  img, 0, 0);
-         for (row = 0; row < srcHeight; row++) {
-            /* The unpack functions with:
-             *    dstType = GL_FLOAT_32_UNSIGNED_INT_24_8_REV
-             * only write their own dword, so the other dword (stencil
-             * or depth) is preserved. */
-            if (srcFormat != GL_STENCIL_INDEX)
-               _mesa_unpack_depth_span(ctx, srcWidth,
-                                       GL_FLOAT_32_UNSIGNED_INT_24_8_REV, /* dst type */
-                                       dstRow, /* dst addr */
-                                       ~0U, srcType, src, srcPacking);
-
-            if (srcFormat != GL_DEPTH_COMPONENT)
-               _mesa_unpack_stencil_span(ctx, srcWidth,
-                                         GL_FLOAT_32_UNSIGNED_INT_24_8_REV, /* dst type */
-                                         dstRow, /* dst addr */
-                                         srcType, src, srcPacking,
-                                         ctx->_ImageTransferState);
+         if (srcFormat != GL_DEPTH_COMPONENT)
+            _mesa_unpack_stencil_span(ctx, srcWidth,
+                                      GL_FLOAT_32_UNSIGNED_INT_24_8_REV, /* dst type */
+                                      dstRow, /* dst addr */
+                                      srcType, src, srcPacking,
+                                      ctx->_ImageTransferState);
 
-            src += srcRowStride;
-            dstRow += dstRowStride / sizeof(uint64_t);
-         }
+         src += srcRowStride;
+         dstRow += dstRowStride / sizeof(uint64_t);
       }
    }
    return GL_TRUE;
diff --git a/mesalib/src/mesa/main/transformfeedback.c b/mesalib/src/mesa/main/transformfeedback.c
index abeb25e87..a73746335 100644
--- a/mesalib/src/mesa/main/transformfeedback.c
+++ b/mesalib/src/mesa/main/transformfeedback.c
@@ -32,7 +32,6 @@
 
 
 #include "buffers.h"
-#include "bufferobj.h"
 #include "context.h"
 #include "hash.h"
 #include "macros.h"
@@ -533,14 +532,7 @@ bind_buffer_range(struct gl_context *ctx, GLuint index,
                                  bufObj);
 
    /* The per-attribute binding point */
-   _mesa_reference_buffer_object(ctx,
-                                 &obj->Buffers[index],
-                                 bufObj);
-
-   obj->BufferNames[index] = bufObj->Name;
-
-   obj->Offset[index] = offset;
-   obj->RequestedSize[index] = size;
+   _mesa_set_transform_feedback_binding(ctx, obj, index, bufObj, offset, size);
 }
 
 
diff --git a/mesalib/src/mesa/main/transformfeedback.h b/mesalib/src/mesa/main/transformfeedback.h
index 7aecd66a7..87f4080e0 100644
--- a/mesalib/src/mesa/main/transformfeedback.h
+++ b/mesalib/src/mesa/main/transformfeedback.h
@@ -27,6 +27,7 @@
 #define TRANSFORM_FEEDBACK_H
 
 #include <stdbool.h>
+#include "bufferobj.h"
 #include "compiler.h"
 #include "glheader.h"
 #include "mtypes.h"
@@ -127,4 +128,17 @@ extern bool
 _mesa_transform_feedback_is_using_program(struct gl_context *ctx,
                                           struct gl_shader_program *shProg);
 
+static inline void
+_mesa_set_transform_feedback_binding(struct gl_context *ctx,
+                                     struct gl_transform_feedback_object *tfObj, GLuint index,
+                                     struct gl_buffer_object *bufObj,
+                                     GLintptr offset, GLsizeiptr size)
+{
+   _mesa_reference_buffer_object(ctx, &tfObj->Buffers[index], bufObj);
+
+   tfObj->BufferNames[index]   = bufObj->Name;
+   tfObj->Offset[index]        = offset;
+   tfObj->RequestedSize[index] = size;
+}
+
 #endif /* TRANSFORM_FEEDBACK_H */
diff --git a/mesalib/src/mesa/main/uniforms.h b/mesalib/src/mesa/main/uniforms.h
index c8b555cbe..10518dcbb 100644
--- a/mesalib/src/mesa/main/uniforms.h
+++ b/mesalib/src/mesa/main/uniforms.h
@@ -319,7 +319,7 @@ struct gl_builtin_uniform_element {
 
 struct gl_builtin_uniform_desc {
    const char *name;
-   struct gl_builtin_uniform_element *elements;
+   const struct gl_builtin_uniform_element *elements;
    unsigned int num_elements;
 };
 
diff --git a/mesalib/src/mesa/main/varray.c b/mesalib/src/mesa/main/varray.c
index 66a3ef119..46956efb5 100644
--- a/mesalib/src/mesa/main/varray.c
+++ b/mesalib/src/mesa/main/varray.c
@@ -784,13 +784,7 @@ static const GLfloat *
 get_current_attrib(struct gl_context *ctx, GLuint index, const char *function)
 {
    if (index == 0) {
-      /* In OpenGL 3.1 attribute 0 becomes non-magic, just like in OpenGL ES
-       * 2.0.  Note that we cannot just check for API_OPENGL_CORE here because
-       * that will erroneously allow this usage in a 3.0 forward-compatible
-       * context too.
-       */
-      if ((ctx->API != API_OPENGL_CORE || ctx->Version < 31)
-          && ctx->API != API_OPENGLES2) {
+      if (_mesa_attr_zero_aliases_vertex(ctx)) {
 	 _mesa_error(ctx, GL_INVALID_OPERATION, "%s(index==0)", function);
 	 return NULL;
       }
@@ -1446,6 +1440,126 @@ _mesa_BindVertexBuffer(GLuint bindingIndex, GLuint buffer, GLintptr offset,
 
 
 void GLAPIENTRY
+_mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
+                        const GLintptr *offsets, const GLsizei *strides)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_vertex_array_object * const vao = ctx->Array.VAO;
+   GLuint i;
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   /* The ARB_vertex_attrib_binding spec says:
+    *
+    *    "An INVALID_OPERATION error is generated if no
+    *     vertex array object is bound."
+    */
+   if (ctx->API == API_OPENGL_CORE &&
+       ctx->Array.VAO == ctx->Array.DefaultVAO) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBindVertexBuffers(No array object bound)");
+      return;
+   }
+
+   /* The ARB_multi_bind spec says:
+    *
+    *    "An INVALID_OPERATION error is generated if <first> + <count>
+    *     is greater than the value of MAX_VERTEX_ATTRIB_BINDINGS."
+    */
+   if (first + count > ctx->Const.MaxVertexAttribBindings) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glBindVertexBuffers(first=%u + count=%d > the value of "
+                  "GL_MAX_VERTEX_ATTRIB_BINDINGS=%u)",
+                  first, count, ctx->Const.MaxVertexAttribBindings);
+      return;
+   }
+
+   if (!buffers) {
+      /**
+       * The ARB_multi_bind spec says:
+       *
+       *    "If <buffers> is NULL, each affected vertex buffer binding point
+       *     from <first> through <first>+<count>-1 will be reset to have no
+       *     bound buffer object.  In this case, the offsets and strides
+       *     associated with the binding points are set to default values,
+       *     ignoring <offsets> and <strides>."
+       */
+      struct gl_buffer_object *vbo = ctx->Shared->NullBufferObj;
+
+      for (i = 0; i < count; i++)
+         bind_vertex_buffer(ctx, VERT_ATTRIB_GENERIC(first + i), vbo, 0, 16);
+
+      return;
+   }
+
+   /* Note that the error semantics for multi-bind commands differ from
+    * those of other GL commands.
+    *
+    * The Issues section in the ARB_multi_bind spec says:
+    *
+    *    "(11) Typically, OpenGL specifies that if an error is generated by
+    *          a command, that command has no effect.  This is somewhat
+    *          unfortunate for multi-bind commands, because it would require
+    *          a first pass to scan the entire list of bound objects for
+    *          errors and then a second pass to actually perform the
+    *          bindings.  Should we have different error semantics?
+    *
+    *       RESOLVED:  Yes.  In this specification, when the parameters for
+    *       one of the <count> binding points are invalid, that binding
+    *       point is not updated and an error will be generated.  However,
+    *       other binding points in the same command will be updated if
+    *       their parameters are valid and no other error occurs."
+    */
+
+   _mesa_begin_bufferobj_lookups(ctx);
+
+   for (i = 0; i < count; i++) {
+      struct gl_buffer_object *vbo;
+
+      /* The ARB_multi_bind spec says:
+       *
+       *    "An INVALID_VALUE error is generated if any value in
+       *     <offsets> or <strides> is negative (per binding)."
+       */
+      if (offsets[i] < 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindVertexBuffer(offsets[%u]=%lldd < 0)",
+                     i, (long long int) offsets[i]);
+         continue;
+      }
+
+      if (strides[i] < 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glBindVertexBuffer(strides[%u]=%lld < 0)",
+                     i, (long long int) strides[i]);
+         continue;
+      }
+
+      if (buffers[i]) {
+         struct gl_vertex_buffer_binding *binding =
+            &vao->VertexBinding[VERT_ATTRIB_GENERIC(first + i)];
+
+         if (buffers[i] == binding->BufferObj->Name)
+            vbo = binding->BufferObj;
+         else
+            vbo = _mesa_multi_bind_lookup_bufferobj(ctx, buffers, i,
+                                                    "glBindVertexBuffers");
+
+         if (!vbo)
+            continue;
+      } else {
+         vbo = ctx->Shared->NullBufferObj;
+      }
+
+      bind_vertex_buffer(ctx, VERT_ATTRIB_GENERIC(first + i), vbo,
+                         offsets[i], strides[i]);
+   }
+
+   _mesa_end_bufferobj_lookups(ctx);
+}
+
+
+void GLAPIENTRY
 _mesa_VertexAttribFormat(GLuint attribIndex, GLint size, GLenum type,
                          GLboolean normalized, GLuint relativeOffset)
 {
diff --git a/mesalib/src/mesa/main/varray.h b/mesalib/src/mesa/main/varray.h
index bc820ed23..f94ebac99 100644
--- a/mesalib/src/mesa/main/varray.h
+++ b/mesalib/src/mesa/main/varray.h
@@ -106,6 +106,22 @@ _mesa_update_client_array(struct gl_context *ctx,
    _mesa_reference_buffer_object(ctx, &dst->BufferObj, binding->BufferObj);
 }
 
+static inline bool
+_mesa_attr_zero_aliases_vertex(struct gl_context *ctx)
+{
+   const bool is_forward_compatible_context =
+      ctx->Const.ContextFlags & GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT;
+
+   /* In OpenGL 3.1 attribute 0 becomes non-magic, just like in OpenGL ES
+    * 2.0.  Note that we cannot just check for API_OPENGL_COMPAT here because
+    * that will erroneously allow this usage in a 3.0 forward-compatible
+    * context too.
+    */
+   return (ctx->API == API_OPENGLES
+           || (ctx->API == API_OPENGL_COMPAT
+               && !is_forward_compatible_context));
+}
+
 extern void GLAPIENTRY
 _mesa_VertexPointer(GLint size, GLenum type, GLsizei stride,
                     const GLvoid *ptr);
@@ -291,6 +307,10 @@ _mesa_BindVertexBuffer(GLuint bindingIndex, GLuint buffer, GLintptr offset,
                        GLsizei stride);
 
 extern void GLAPIENTRY
+_mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
+                        const GLintptr *offsets, const GLsizei *strides);
+
+extern void GLAPIENTRY
 _mesa_VertexAttribFormat(GLuint attribIndex, GLint size, GLenum type,
                          GLboolean normalized, GLuint relativeOffset);
 
diff --git a/mesalib/src/mesa/state_tracker/st_atom_texture.c b/mesalib/src/mesa/state_tracker/st_atom_texture.c
index afc6d9dab..928a4ffd2 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_texture.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_texture.c
@@ -271,6 +271,10 @@ st_get_texture_sampler_view_from_stobj(struct st_context *st,
 
    sv = st_texture_get_sampler_view(st, stObj);
 
+   if (stObj->base.StencilSampling &&
+       util_format_is_depth_and_stencil(format))
+      format = util_format_stencil_only(format);
+
    /* if sampler view has changed dereference it */
    if (*sv) {
       if (check_sampler_swizzle(stObj, *sv) ||
diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c
index a245fdf18..12ba82df3 100644
--- a/mesalib/src/mesa/state_tracker/st_extensions.c
+++ b/mesalib/src/mesa/state_tracker/st_extensions.c
@@ -507,6 +507,11 @@ void st_init_extensions(struct st_context *st)
 
       { { o(OES_compressed_ETC1_RGB8_texture) },
         { PIPE_FORMAT_ETC1_RGB8 } },
+
+      { { o(ARB_stencil_texturing) },
+        { PIPE_FORMAT_X24S8_UINT,
+          PIPE_FORMAT_S8X24_UINT },
+        GL_TRUE }, /* at least one format must be supported */
    };
 
    /* Required: vertex fetch support. */
@@ -560,13 +565,6 @@ void st_init_extensions(struct st_context *st)
    ctx->Extensions.EXT_point_parameters = GL_TRUE;
    ctx->Extensions.EXT_provoking_vertex = GL_TRUE;
 
-   /* IMPORTANT:
-    *    Don't enable EXT_separate_shader_objects. It disallows a certain
-    *    optimization in the GLSL compiler and therefore is considered
-    *    harmful.
-    */
-   ctx->Extensions.EXT_separate_shader_objects = GL_FALSE;
-
    ctx->Extensions.EXT_texture_env_dot3 = GL_TRUE;
    ctx->Extensions.EXT_vertex_array_bgra = GL_TRUE;
 
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 d1c3856b5..bdee1f4eb 100644
--- a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -87,8 +87,7 @@ extern "C" {
  */
 #define MAX_ARRAYS        256
 
-/* if we support a native gallium TG4 with the ability to take 4 texoffsets then bump this */
-#define MAX_GLSL_TEXTURE_OFFSET 1
+#define MAX_GLSL_TEXTURE_OFFSET 4
 
 class st_src_reg;
 class st_dst_reg;
@@ -2728,12 +2727,13 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
 void
 glsl_to_tgsi_visitor::visit(ir_texture *ir)
 {
-   st_src_reg result_src, coord, cube_sc, lod_info, projector, dx, dy, offset, sample_index, component;
+   st_src_reg result_src, coord, cube_sc, lod_info, projector, dx, dy, offset[MAX_GLSL_TEXTURE_OFFSET], sample_index, component;
    st_dst_reg result_dst, coord_dst, cube_sc_dst;
    glsl_to_tgsi_instruction *inst = NULL;
    unsigned opcode = TGSI_OPCODE_NOP;
    const glsl_type *sampler_type = ir->sampler->type;
    bool is_cube_array = false;
+   unsigned i;
 
    /* if we are a cube array sampler */
    if ((sampler_type->sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE &&
@@ -2771,7 +2771,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       opcode = (is_cube_array && ir->shadow_comparitor) ? TGSI_OPCODE_TEX2 : TGSI_OPCODE_TEX; 
       if (ir->offset) {
          ir->offset->accept(this);
-         offset = this->result;
+         offset[0] = this->result;
       }
       break;
    case ir_txb:
@@ -2780,7 +2780,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       lod_info = this->result;
       if (ir->offset) {
          ir->offset->accept(this);
-         offset = this->result;
+         offset[0] = this->result;
       }
       break;
    case ir_txl:
@@ -2789,7 +2789,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       lod_info = this->result;
       if (ir->offset) {
          ir->offset->accept(this);
-         offset = this->result;
+         offset[0] = this->result;
       }
       break;
    case ir_txd:
@@ -2800,7 +2800,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       dy = this->result;
       if (ir->offset) {
          ir->offset->accept(this);
-         offset = this->result;
+         offset[0] = this->result;
       }
       break;
    case ir_txs:
@@ -2814,7 +2814,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       lod_info = this->result;
       if (ir->offset) {
          ir->offset->accept(this);
-         offset = this->result;
+         offset[0] = this->result;
       }
       break;
    case ir_txf_ms:
@@ -2828,9 +2828,17 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
       component = this->result;
       if (ir->offset) {
          ir->offset->accept(this);
-         /* this should have been lowered */
-         assert(ir->offset->type->base_type != GLSL_TYPE_ARRAY);
-         offset = this->result;
+         if (ir->offset->type->base_type == GLSL_TYPE_ARRAY) {
+            const glsl_type *elt_type = ir->offset->type->fields.array;
+            for (i = 0; i < ir->offset->type->length; i++) {
+               offset[i] = this->result;
+               offset[i].index += i * type_size(elt_type);
+               offset[i].type = elt_type->base_type;
+               offset[i].swizzle = swizzle_for_size(elt_type->vector_elements);
+            }
+         } else {
+            offset[0] = this->result;
+         }
       }
       break;
    case ir_lod:
@@ -2960,8 +2968,9 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
         					   this->prog);
 
    if (ir->offset) {
-      inst->tex_offset_num_offset = 1;
-      inst->tex_offsets[0] = offset;
+      for (i = 0; i < MAX_GLSL_TEXTURE_OFFSET && offset[i].file != PROGRAM_UNDEFINED; i++)
+         inst->tex_offsets[i] = offset[i];
+      inst->tex_offset_num_offset = i;
    }
 
    switch (sampler_type->sampler_dimensionality) {
@@ -4479,6 +4488,8 @@ translate_tex_offset(struct st_translate *t,
 {
    struct tgsi_texture_offset offset;
    struct ureg_src imm_src;
+   struct ureg_dst dst;
+   int array;
 
    switch (in_offset->file) {
    case PROGRAM_IMMEDIATE:
@@ -4500,6 +4511,20 @@ translate_tex_offset(struct st_translate *t,
       offset.SwizzleZ = GET_SWZ(in_offset->swizzle, 2);
       offset.Padding = 0;
       break;
+   case PROGRAM_ARRAY:
+      array = in_offset->index >> 16;
+
+      assert(array >= 0);
+      assert(array < (int) Elements(t->arrays));
+
+      dst = t->arrays[array];
+      offset.File = dst.File;
+      offset.Index = dst.Index + (in_offset->index & 0xFFFF) - 0x8000;
+      offset.SwizzleX = GET_SWZ(in_offset->swizzle, 0);
+      offset.SwizzleY = GET_SWZ(in_offset->swizzle, 1);
+      offset.SwizzleZ = GET_SWZ(in_offset->swizzle, 2);
+      offset.Padding = 0;
+      break;
    default:
       break;
    }
@@ -5350,6 +5375,7 @@ st_new_shader_program(struct gl_context *ctx, GLuint name)
 GLboolean
 st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
 {
+   struct pipe_screen *pscreen = ctx->st->pipe->screen;
    assert(prog->LinkStatus);
 
    for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
@@ -5388,7 +5414,8 @@ st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
          lower_packing_builtins(ir, lower_inst);
       }
 
-      lower_offset_arrays(ir);
+      if (!pscreen->get_param(pscreen, PIPE_CAP_TEXTURE_GATHER_OFFSETS))
+         lower_offset_arrays(ir);
       do_mat_op_to_vec(ir);
       lower_instructions(ir,
                          MOD_TO_FRACT |
@@ -5396,6 +5423,8 @@ st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
                          EXP_TO_EXP2 |
                          LOG_TO_LOG2 |
                          LDEXP_TO_ARITH |
+                         CARRY_TO_ARITH |
+                         BORROW_TO_ARITH |
                          (options->EmitNoPow ? POW_TO_EXP2 : 0) |
                          (!ctx->Const.NativeIntegers ? INT_DIV_TO_MUL_RCP : 0));
 
diff --git a/mesalib/src/mesa/tnl/t_vertex.c b/mesalib/src/mesa/tnl/t_vertex.c
index b3deac024..421bae2b8 100644
--- a/mesalib/src/mesa/tnl/t_vertex.c
+++ b/mesalib/src/mesa/tnl/t_vertex.c
@@ -83,12 +83,22 @@ void _tnl_register_fastpath( struct tnl_clipspace *vtx,
    struct tnl_clipspace_fastpath *fastpath = CALLOC_STRUCT(tnl_clipspace_fastpath);
    GLuint i;
 
+   if (fastpath == NULL) {
+      _mesa_error_no_memory(__func__);
+      return;
+   }
+
    fastpath->vertex_size = vtx->vertex_size;
    fastpath->attr_count = vtx->attr_count;
    fastpath->match_strides = match_strides;
    fastpath->func = vtx->emit;
-   fastpath->attr =
-      malloc(vtx->attr_count * sizeof(fastpath->attr[0]));
+   fastpath->attr = malloc(vtx->attr_count * sizeof(fastpath->attr[0]));
+
+   if (fastpath->attr == NULL) {
+      free(fastpath);
+      _mesa_error_no_memory(__func__);
+      return;
+   }
 
    for (i = 0; i < vtx->attr_count; i++) {
       fastpath->attr[i].format = vtx->attr[i].format;
diff --git a/mesalib/src/mesa/vbo/vbo_attrib_tmp.h b/mesalib/src/mesa/vbo/vbo_attrib_tmp.h
index a2ea6d0c8..ec66934fc 100644
--- a/mesalib/src/mesa/vbo/vbo_attrib_tmp.h
+++ b/mesalib/src/mesa/vbo/vbo_attrib_tmp.h
@@ -26,6 +26,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 **************************************************************************/
 
 #include "util/u_format_r11g11b10f.h"
+#include "main/varray.h"
 
 /* float */
 #define ATTR1FV( A, V ) ATTR( A, 1, GL_FLOAT, (V)[0], 0, 0, 1 )
@@ -499,7 +500,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib1fARB)(GLuint index, GLfloat x)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR1F(0, x);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR1F(VBO_ATTRIB_GENERIC0 + index, x);
@@ -511,7 +512,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib1fvARB)(GLuint index, const GLfloat * v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR1FV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR1FV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -523,7 +524,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib2fARB)(GLuint index, GLfloat x, GLfloat y)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2F(0, x, y);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2F(VBO_ATTRIB_GENERIC0 + index, x, y);
@@ -535,7 +536,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib2fvARB)(GLuint index, const GLfloat * v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2FV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2FV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -547,7 +548,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib3fARB)(GLuint index, GLfloat x, GLfloat y, GLfloat z)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3F(0, x, y, z);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3F(VBO_ATTRIB_GENERIC0 + index, x, y, z);
@@ -559,7 +560,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib3fvARB)(GLuint index, const GLfloat * v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3FV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3FV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -571,7 +572,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib4fARB)(GLuint index, GLfloat x, GLfloat y, GLfloat z, GLfloat w)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4F(0, x, y, z, w);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4F(VBO_ATTRIB_GENERIC0 + index, x, y, z, w);
@@ -583,7 +584,7 @@ static void GLAPIENTRY
 TAG(VertexAttrib4fvARB)(GLuint index, const GLfloat * v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4FV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4FV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -600,7 +601,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI1i)(GLuint index, GLint x)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR1I(0, x);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR1I(VBO_ATTRIB_GENERIC0 + index, x);
@@ -612,7 +613,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI2i)(GLuint index, GLint x, GLint y)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2I(0, x, y);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2I(VBO_ATTRIB_GENERIC0 + index, x, y);
@@ -624,7 +625,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI3i)(GLuint index, GLint x, GLint y, GLint z)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3I(0, x, y, z);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3I(VBO_ATTRIB_GENERIC0 + index, x, y, z);
@@ -636,7 +637,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI4i)(GLuint index, GLint x, GLint y, GLint z, GLint w)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4I(0, x, y, z, w);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4I(VBO_ATTRIB_GENERIC0 + index, x, y, z, w);
@@ -648,7 +649,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI2iv)(GLuint index, const GLint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2IV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2IV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -660,7 +661,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI3iv)(GLuint index, const GLint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3IV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3IV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -672,7 +673,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI4iv)(GLuint index, const GLint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4IV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4IV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -689,7 +690,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI1ui)(GLuint index, GLuint x)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR1UI(0, x);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR1UI(VBO_ATTRIB_GENERIC0 + index, x);
@@ -701,7 +702,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI2ui)(GLuint index, GLuint x, GLuint y)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2UI(0, x, y);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2UI(VBO_ATTRIB_GENERIC0 + index, x, y);
@@ -713,7 +714,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI3ui)(GLuint index, GLuint x, GLuint y, GLuint z)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3UI(0, x, y, z);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3UI(VBO_ATTRIB_GENERIC0 + index, x, y, z);
@@ -725,7 +726,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI4ui)(GLuint index, GLuint x, GLuint y, GLuint z, GLuint w)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4UI(0, x, y, z, w);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4UI(VBO_ATTRIB_GENERIC0 + index, x, y, z, w);
@@ -737,7 +738,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI2uiv)(GLuint index, const GLuint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR2UIV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR2UIV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -749,7 +750,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI3uiv)(GLuint index, const GLuint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR3UIV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR3UIV(VBO_ATTRIB_GENERIC0 + index, v);
@@ -761,7 +762,7 @@ static void GLAPIENTRY
 TAG(VertexAttribI4uiv)(GLuint index, const GLuint *v)
 {
    GET_CURRENT_CONTEXT(ctx);
-   if (index == 0)
+   if (index == 0 && _mesa_attr_zero_aliases_vertex(ctx))
       ATTR4UIV(0, v);
    else if (index < MAX_VERTEX_GENERIC_ATTRIBS)
       ATTR4UIV(VBO_ATTRIB_GENERIC0 + index, v);
diff --git a/mesalib/src/mesa/vbo/vbo_rebase.c b/mesalib/src/mesa/vbo/vbo_rebase.c
index f3fe5f757..82a0b8e2b 100644
--- a/mesalib/src/mesa/vbo/vbo_rebase.c
+++ b/mesalib/src/mesa/vbo/vbo_rebase.c
@@ -58,9 +58,14 @@ static void *rebase_##TYPE( const void *ptr,			\
 			  GLuint count, 			\
 			  TYPE min_index )			\
 {								\
-   const TYPE *in = (TYPE *)ptr;				\
-   TYPE *tmp_indices = malloc(count * sizeof(TYPE));	\
    GLuint i;							\
+   const TYPE *in = (TYPE *)ptr;				\
+   TYPE *tmp_indices = malloc(count * sizeof(TYPE));		\
+								\
+   if (tmp_indices == NULL) {                                   \
+      _mesa_error_no_memory(__func__);                          \
+      return NULL;                                              \
+   }                                                            \
 								\
    for (i = 0; i < count; i++)  				\
       tmp_indices[i] = in[i] - min_index;			\
@@ -148,6 +153,11 @@ void vbo_rebase_prims( struct gl_context *ctx,
        */
       tmp_prims = malloc(sizeof(*prim) * nr_prims);
 
+      if (tmp_prims == NULL) {
+         _mesa_error_no_memory(__func__);
+         return;
+      }
+
       for (i = 0; i < nr_prims; i++) {
 	 tmp_prims[i] = prim[i];
 	 tmp_prims[i].basevertex -= min_index;
@@ -186,6 +196,10 @@ void vbo_rebase_prims( struct gl_context *ctx,
       if (map_ib) 
 	 ctx->Driver.UnmapBuffer(ctx, ib->obj, MAP_INTERNAL);
 
+      if (tmp_indices == NULL) {
+         return;
+      }
+
       tmp_ib.obj = ctx->Shared->NullBufferObj;
       tmp_ib.ptr = tmp_indices;
       tmp_ib.count = ib->count;
@@ -198,6 +212,11 @@ void vbo_rebase_prims( struct gl_context *ctx,
        */
       tmp_prims = malloc(sizeof(*prim) * nr_prims);
 
+      if (tmp_prims == NULL) {
+         _mesa_error_no_memory(__func__);
+         return;
+      }
+
       for (i = 0; i < nr_prims; i++) {
 	 /* If this fails, it could indicate an application error:
 	  */