Merge remote-tracking branch 'origin/released'

23 files changed, 840 insertions, 781 deletions

diff --git a/mesalib/src/gallium/auxiliary/util/u_blit.c b/mesalib/src/gallium/auxiliary/util/u_blit.c
index a10fd17cb..127973388 100644
--- a/mesalib/src/gallium/auxiliary/util/u_blit.c
+++ b/mesalib/src/gallium/auxiliary/util/u_blit.c
@@ -321,6 +321,26 @@ regions_overlap(int srcX0, int srcY0,
 
 
 /**
+ * Can we blit from src format to dest format with a simple copy?
+ */
+static boolean
+formats_compatible(enum pipe_format src_format,
+                   enum pipe_format dst_format)
+{
+   if (src_format == dst_format) {
+      return TRUE;
+   }
+   else {
+      const struct util_format_description *src_desc =
+         util_format_description(src_format);
+      const struct util_format_description *dst_desc =
+         util_format_description(dst_format);
+      return util_is_format_compatible(src_desc, dst_desc);
+   }
+}
+
+
+/**
  * Copy pixel block from src surface to dst surface.
  * Overlapping regions are acceptable.
  * Flipping and stretching are supported.
@@ -377,7 +397,7 @@ util_blit_pixels_writemask(struct blit_state *ctx,
     * no overlapping.
     * Filter mode should not matter since there's no stretching.
     */
-   if (dst_format == src_format &&
+   if (formats_compatible(src_format, dst_format) &&
        srcX0 < srcX1 &&
        dstX0 < dstX1 &&
        srcY0 < srcY1 &&
diff --git a/mesalib/src/gallium/auxiliary/util/u_blitter.c b/mesalib/src/gallium/auxiliary/util/u_blitter.c
index 48808ae08..5784a7ceb 100644
--- a/mesalib/src/gallium/auxiliary/util/u_blitter.c
+++ b/mesalib/src/gallium/auxiliary/util/u_blitter.c
@@ -351,12 +351,12 @@ static void blitter_unset_running_flag(struct blitter_context_priv *ctx)
 
 static void blitter_check_saved_vertex_states(struct blitter_context_priv *ctx)
 {
-   assert(ctx->base.saved_num_vertex_buffers != ~0 &&
-          ctx->base.saved_velem_state != INVALID_PTR &&
-          ctx->base.saved_vs != INVALID_PTR &&
-          (!ctx->has_geometry_shader || ctx->base.saved_gs != INVALID_PTR) &&
-          (!ctx->has_stream_out || ctx->base.saved_num_so_targets != ~0) &&
-          ctx->base.saved_rs_state != INVALID_PTR);
+   assert(ctx->base.saved_num_vertex_buffers != ~0);
+   assert(ctx->base.saved_velem_state != INVALID_PTR);
+   assert(ctx->base.saved_vs != INVALID_PTR);
+   assert(!ctx->has_geometry_shader || ctx->base.saved_gs != INVALID_PTR);
+   assert(!ctx->has_stream_out || ctx->base.saved_num_so_targets != ~0);
+   assert(ctx->base.saved_rs_state != INVALID_PTR);
 }
 
 static void blitter_restore_vertex_states(struct blitter_context_priv *ctx)
@@ -410,9 +410,9 @@ static void blitter_restore_vertex_states(struct blitter_context_priv *ctx)
 
 static void blitter_check_saved_fragment_states(struct blitter_context_priv *ctx)
 {
-   assert(ctx->base.saved_fs != INVALID_PTR &&
-          ctx->base.saved_dsa_state != INVALID_PTR &&
-          ctx->base.saved_blend_state != INVALID_PTR);
+   assert(ctx->base.saved_fs != INVALID_PTR);
+   assert(ctx->base.saved_dsa_state != INVALID_PTR);
+   assert(ctx->base.saved_blend_state != INVALID_PTR);
 }
 
 static void blitter_restore_fragment_states(struct blitter_context_priv *ctx)
@@ -453,8 +453,8 @@ static void blitter_restore_fb_state(struct blitter_context_priv *ctx)
 
 static void blitter_check_saved_textures(struct blitter_context_priv *ctx)
 {
-   assert(ctx->base.saved_num_sampler_states != ~0 &&
-          ctx->base.saved_num_sampler_views != ~0);
+   assert(ctx->base.saved_num_sampler_states != ~0);
+   assert(ctx->base.saved_num_sampler_views != ~0);
 }
 
 static void blitter_restore_textures(struct blitter_context_priv *ctx)
diff --git a/mesalib/src/gallium/auxiliary/util/u_format.h b/mesalib/src/gallium/auxiliary/util/u_format.h
index 874ea7eb1..b9ae7c190 100644
--- a/mesalib/src/gallium/auxiliary/util/u_format.h
+++ b/mesalib/src/gallium/auxiliary/util/u_format.h
@@ -578,8 +578,9 @@ boolean
 util_format_is_pure_uint(enum pipe_format format);
 
 /**
- * Whether the src format can be blitted to destation format with a simple
- * memcpy.
+ * Check if the src format can be blitted to the destination format with
+ * a simple memcpy.  For example, blitting from RGBA to RGBx is OK, but not
+ * the reverse.
  */
 boolean
 util_is_format_compatible(const struct util_format_description *src_desc,
diff --git a/mesalib/src/glsl/builtin_variables.cpp b/mesalib/src/glsl/builtin_variables.cpp
index ed6b922ca..66b3abdd8 100644
--- a/mesalib/src/glsl/builtin_variables.cpp
+++ b/mesalib/src/glsl/builtin_variables.cpp
@@ -870,13 +870,25 @@ generate_ARB_draw_instanced_variables(exec_list *instructions,
    /* gl_InstanceIDARB is only available in the vertex shader.
     */
    if (target == vertex_shader) {
-      ir_variable *const inst =
+      ir_variable *inst =
          add_variable(instructions, state->symbols,
 		      "gl_InstanceIDARB", glsl_type::int_type,
 		      ir_var_system_value, SYSTEM_VALUE_INSTANCE_ID);
 
       if (warn)
          inst->warn_extension = "GL_ARB_draw_instanced";
+
+      /* Originally ARB_draw_instanced only specified that ARB decorated name.
+       * Since no vendor actually implemented that behavior and some apps use
+       * the undecorated name, the extension now specifies that both names are
+       * available.
+       */
+      inst = add_variable(instructions, state->symbols,
+			  "gl_InstanceID", glsl_type::int_type,
+			  ir_var_system_value, SYSTEM_VALUE_INSTANCE_ID);
+
+      if (warn)
+         inst->warn_extension = "GL_ARB_draw_instanced";
    }
 }
 
diff --git a/mesalib/src/mesa/Makefile b/mesalib/src/mesa/Makefile
index 0e15d61bd..71e22b9cb 100644
--- a/mesalib/src/mesa/Makefile
+++ b/mesalib/src/mesa/Makefile
@@ -206,8 +206,6 @@ install-headers:
 install-libgl: default gl.pc install-headers
 	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
 	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
-	$(MINSTALL) $(TOP)/$(LIB_DIR)/$(GL_LIB_GLOB) \
-		$(DESTDIR)$(INSTALL_LIB_DIR)
 	$(INSTALL) -m 644 gl.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 install-osmesa: default osmesa.pc
diff --git a/mesalib/src/mesa/main/api_validate.c b/mesalib/src/mesa/main/api_validate.c
index 1ae491f25..4e94f47e3 100644
--- a/mesalib/src/mesa/main/api_validate.c
+++ b/mesalib/src/mesa/main/api_validate.c
@@ -383,6 +383,12 @@ _mesa_validate_DrawArraysInstanced(struct gl_context *ctx, GLenum mode, GLint fi
       return GL_FALSE;
    }
 
+   if (first < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+		  "glDrawArraysInstanced(start=%d)", first);
+      return GL_FALSE;
+   }
+
    if (!_mesa_valid_prim_mode(ctx, mode)) {
       _mesa_error(ctx, GL_INVALID_ENUM,
                   "glDrawArraysInstanced(mode=0x%x)", mode);
@@ -399,12 +405,6 @@ _mesa_validate_DrawArraysInstanced(struct gl_context *ctx, GLenum mode, GLint fi
    if (!check_valid_to_render(ctx, "glDrawArraysInstanced(invalid to render)"))
       return GL_FALSE;
 
-   if (ctx->CompileFlag) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glDrawArraysInstanced(display list");
-      return GL_FALSE;
-   }
-
    if (ctx->Const.CheckArrayBounds) {
       if (first + count > (GLint) ctx->Array.ArrayObj->_MaxElement)
          return GL_FALSE;
diff --git a/mesalib/src/mesa/main/bufferobj.c b/mesalib/src/mesa/main/bufferobj.c
index b6137d0f6..ada42c515 100644
--- a/mesalib/src/mesa/main/bufferobj.c
+++ b/mesalib/src/mesa/main/bufferobj.c
@@ -1172,17 +1172,17 @@ _mesa_GetBufferParameterivARB(GLenum target, GLenum pname, GLint *params)
       *params = _mesa_bufferobj_mapped(bufObj);
       return;
    case GL_BUFFER_ACCESS_FLAGS:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = bufObj->AccessFlags;
       return;
    case GL_BUFFER_MAP_OFFSET:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = (GLint) bufObj->Offset;
       return;
    case GL_BUFFER_MAP_LENGTH:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = (GLint) bufObj->Length;
       return;
@@ -1223,7 +1223,7 @@ _mesa_GetBufferParameteri64v(GLenum target, GLenum pname, GLint64 *params)
       *params = simplified_access_mode(bufObj->AccessFlags);
       return;
    case GL_BUFFER_ACCESS_FLAGS:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = bufObj->AccessFlags;
       return;
@@ -1231,12 +1231,12 @@ _mesa_GetBufferParameteri64v(GLenum target, GLenum pname, GLint64 *params)
       *params = _mesa_bufferobj_mapped(bufObj);
       return;
    case GL_BUFFER_MAP_OFFSET:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = bufObj->Offset;
       return;
    case GL_BUFFER_MAP_LENGTH:
-      if (ctx->VersionMajor < 3)
+      if (!ctx->Extensions.ARB_map_buffer_range)
          goto invalid_pname;
       *params = bufObj->Length;
       return;
diff --git a/mesalib/src/mesa/main/dlist.c b/mesalib/src/mesa/main/dlist.c
index 677debffc..3db7bebae 100644
--- a/mesalib/src/mesa/main/dlist.c
+++ b/mesalib/src/mesa/main/dlist.c
@@ -1300,6 +1300,43 @@ save_BlendEquationSeparatei(GLuint buf, GLenum modeRGB, GLenum modeA)
 }
 
 
+/* GL_ARB_draw_instanced. */
+static void GLAPIENTRY
+save_DrawArraysInstancedARB(GLenum mode,
+			    GLint first,
+			    GLsizei count,
+			    GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+	       "glDrawArraysInstanced() during display list compile");
+}
+
+static void GLAPIENTRY
+save_DrawElementsInstancedARB(GLenum mode,
+			      GLsizei count,
+			      GLenum type,
+			      const GLvoid *indices,
+			      GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+	       "glDrawElementsInstanced() during display list compile");
+}
+
+static void GLAPIENTRY
+save_DrawElementsInstancedBaseVertexARB(GLenum mode,
+					GLsizei count,
+					GLenum type,
+					const GLvoid *indices,
+					GLsizei primcount,
+					GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+	       "glDrawElementsInstancedBaseVertex() during display list compile");
+}
+
 static void invalidate_saved_current_state( struct gl_context *ctx )
 {
    GLint i;
@@ -9672,8 +9709,6 @@ exec_MultiModeDrawElementsIBM(const GLenum * mode,
                                   modestride));
 }
 
-
-
 /**
  * Setup the given dispatch table to point to Mesa's display list
  * building functions.
@@ -10751,6 +10786,13 @@ _mesa_save_vtxfmt_init(GLvertexformat * vfmt)
 
    vfmt->Rectf = save_Rectf;
 
+   /* GL_ARB_draw_instanced */
+   vfmt->DrawArraysInstanced = save_DrawArraysInstancedARB;
+   vfmt->DrawElementsInstanced = save_DrawElementsInstancedARB;
+
+   /* GL_ARB_draw_elements_base_vertex */
+   vfmt->DrawElementsInstancedBaseVertex = save_DrawElementsInstancedBaseVertexARB;
+
    /* The driver is required to implement these as
     * 1) They can probably do a better job.
     * 2) A lot of new mechanisms would have to be added to this module
diff --git a/mesalib/src/mesa/main/enable.c b/mesalib/src/mesa/main/enable.c
index 270b24045..515dda952 100644
--- a/mesalib/src/mesa/main/enable.c
+++ b/mesalib/src/mesa/main/enable.c
@@ -118,7 +118,7 @@ client_state(struct gl_context *ctx, GLenum cap, GLboolean state)
          CHECK_EXTENSION(NV_vertex_program, cap);
          {
             GLint n = (GLint) cap - GL_VERTEX_ATTRIB_ARRAY0_NV;
-            ASSERT(VERT_ATTRIB_GENERIC(n) < Elements(ctx->Array.ArrayObj->VertexAttrib));
+            ASSERT(VERT_ATTRIB_GENERIC(n) < Elements(arrayObj->VertexAttrib));
             var = &arrayObj->VertexAttrib[VERT_ATTRIB_GENERIC(n)].Enabled;
             flag = VERT_BIT_GENERIC(n);
          }
@@ -149,9 +149,9 @@ client_state(struct gl_context *ctx, GLenum cap, GLboolean state)
    *var = state;
 
    if (state)
-      ctx->Array.ArrayObj->_Enabled |= flag;
+      arrayObj->_Enabled |= flag;
    else
-      ctx->Array.ArrayObj->_Enabled &= ~flag;
+      arrayObj->_Enabled &= ~flag;
 
    if (ctx->Driver.Enable) {
       ctx->Driver.Enable( ctx, cap, state );
diff --git a/mesalib/src/mesa/main/light.c b/mesalib/src/mesa/main/light.c
index a16d0e998..962a3e689 100644
--- a/mesalib/src/mesa/main/light.c
+++ b/mesalib/src/mesa/main/light.c
@@ -161,11 +161,9 @@ _mesa_light(struct gl_context *ctx, GLuint lnum, GLenum pname, const GLfloat *pa
          return;
       FLUSH_VERTICES(ctx, _NEW_LIGHT);
       light->SpotCutoff = params[0];
-      light->_CosCutoffNeg = (GLfloat) (cos(light->SpotCutoff * DEG2RAD));
-      if (light->_CosCutoffNeg < 0)
+      light->_CosCutoff = (GLfloat) (cos(light->SpotCutoff * DEG2RAD));
+      if (light->_CosCutoff < 0)
          light->_CosCutoff = 0;
-      else
-         light->_CosCutoff = light->_CosCutoffNeg;
       if (light->SpotCutoff != 180.0F)
          light->_Flags |= LIGHT_SPOT;
       else
@@ -623,11 +621,6 @@ _mesa_material_bitmask( struct gl_context *ctx, GLenum face, GLenum pname,
 
 
 
-static void
-invalidate_shine_table( struct gl_context *ctx, GLuint side );
-
-
-
 /* Update derived values following a change in ctx->Light.Material
  */
 void
@@ -699,14 +692,6 @@ _mesa_update_material( struct gl_context *ctx, GLuint bitmask )
 		   mat[MAT_ATTRIB_BACK_SPECULAR]);
       }
    }
-
-   if (bitmask & MAT_BIT_FRONT_SHININESS) {
-      invalidate_shine_table( ctx, 0 );
-   }
-
-   if (bitmask & MAT_BIT_BACK_SHININESS) {
-      invalidate_shine_table( ctx, 1 );
-   }
 }
 
 
@@ -876,126 +861,6 @@ _mesa_GetMaterialiv( GLenum face, GLenum pname, GLint *params )
 
 
 
-/**********************************************************************/
-/*****                  Lighting computation                      *****/
-/**********************************************************************/
-
-
-/*
- * Notes:
- *   When two-sided lighting is enabled we compute the color (or index)
- *   for both the front and back side of the primitive.  Then, when the
- *   orientation of the facet is later learned, we can determine which
- *   color (or index) to use for rendering.
- *
- *   KW: We now know orientation in advance and only shade for
- *       the side or sides which are actually required.
- *
- * Variables:
- *   n = normal vector
- *   V = vertex position
- *   P = light source position
- *   Pe = (0,0,0,1)
- *
- * Precomputed:
- *   IF P[3]==0 THEN
- *       // light at infinity
- *       IF local_viewer THEN
- *           _VP_inf_norm = unit vector from V to P      // Precompute
- *       ELSE
- *           // eye at infinity
- *           _h_inf_norm = Normalize( VP + <0,0,1> )     // Precompute
- *       ENDIF
- *   ENDIF
- *
- * Functions:
- *   Normalize( v ) = normalized vector v
- *   Magnitude( v ) = length of vector v
- */
-
-
-
-/* Calculate a new shine table.  Doing this here saves a branch in
- * lighting, and the cost of doing it early may be partially offset
- * by keeping a MRU cache of shine tables for various shine values.
- */
-static void
-invalidate_shine_table( struct gl_context *ctx, GLuint side )
-{
-   ASSERT(side < 2);
-   if (ctx->_ShineTable[side])
-      ctx->_ShineTable[side]->refcount--;
-   ctx->_ShineTable[side] = NULL;
-}
-
-
-static void
-validate_shine_table( struct gl_context *ctx, GLuint side, GLfloat shininess )
-{
-   struct gl_shine_tab *list = ctx->_ShineTabList;
-   struct gl_shine_tab *s;
-
-   ASSERT(side < 2);
-
-   foreach(s, list)
-      if ( s->shininess == shininess )
-	 break;
-
-   if (s == list) {
-      GLint j;
-      GLfloat *m;
-
-      foreach(s, list)
-	 if (s->refcount == 0)
-	    break;
-
-      m = s->tab;
-      m[0] = 0.0;
-      if (shininess == 0.0) {
-	 for (j = 1 ; j <= SHINE_TABLE_SIZE ; j++)
-	    m[j] = 1.0;
-      }
-      else {
-	 for (j = 1 ; j < SHINE_TABLE_SIZE ; j++) {
-            GLdouble t, x = j / (GLfloat) (SHINE_TABLE_SIZE - 1);
-            if (x < 0.005) /* underflow check */
-               x = 0.005;
-            t = pow(x, shininess);
-	    if (t > 1e-20)
-	       m[j] = (GLfloat) t;
-	    else
-	       m[j] = 0.0;
-	 }
-	 m[SHINE_TABLE_SIZE] = 1.0;
-      }
-
-      s->shininess = shininess;
-   }
-
-   if (ctx->_ShineTable[side])
-      ctx->_ShineTable[side]->refcount--;
-
-   ctx->_ShineTable[side] = s;
-   move_to_tail( list, s );
-   s->refcount++;
-}
-
-
-void
-_mesa_validate_all_lighting_tables( struct gl_context *ctx )
-{
-   GLfloat shininess;
-   
-   shininess = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS][0];
-   if (!ctx->_ShineTable[0] || ctx->_ShineTable[0]->shininess != shininess)
-      validate_shine_table( ctx, 0, shininess );
-
-   shininess = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_SHININESS][0];
-   if (!ctx->_ShineTable[1] || ctx->_ShineTable[1]->shininess != shininess)
-      validate_shine_table( ctx, 1, shininess );
-}
-
-
 /**
  * Examine current lighting parameters to determine if the optimized lighting
  * function can be used.
@@ -1005,23 +870,23 @@ _mesa_validate_all_lighting_tables( struct gl_context *ctx )
 void
 _mesa_update_lighting( struct gl_context *ctx )
 {
+   GLbitfield flags = 0;
    struct gl_light *light;
    ctx->Light._NeedEyeCoords = GL_FALSE;
-   ctx->Light._Flags = 0;
 
    if (!ctx->Light.Enabled)
       return;
 
    foreach(light, &ctx->Light.EnabledList) {
-      ctx->Light._Flags |= light->_Flags;
+      flags |= light->_Flags;
    }
 
    ctx->Light._NeedVertices =
-      ((ctx->Light._Flags & (LIGHT_POSITIONAL|LIGHT_SPOT)) ||
+      ((flags & (LIGHT_POSITIONAL|LIGHT_SPOT)) ||
        ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR ||
        ctx->Light.Model.LocalViewer);
 
-   ctx->Light._NeedEyeCoords = ((ctx->Light._Flags & LIGHT_POSITIONAL) ||
+   ctx->Light._NeedEyeCoords = ((flags & LIGHT_POSITIONAL) ||
 				ctx->Light.Model.LocalViewer);
 
    /* XXX: This test is overkill & needs to be fixed both for software and
@@ -1081,9 +946,6 @@ compute_light_positions( struct gl_context *ctx )
       TRANSFORM_NORMAL( ctx->_EyeZDir, eye_z, ctx->ModelviewMatrixStack.Top->m );
    }
 
-   /* Make sure all the light tables are updated before the computation */
-   _mesa_validate_all_lighting_tables(ctx);
-
    foreach (light, &ctx->Light.EnabledList) {
 
       if (ctx->_NeedEyeCoords) {
@@ -1260,7 +1122,6 @@ init_light( struct gl_light *l, GLuint n )
    ASSIGN_3V( l->SpotDirection, 0.0, 0.0, -1.0 );
    l->SpotExponent = 0.0;
    l->SpotCutoff = 180.0;
-   l->_CosCutoffNeg = -1.0f;
    l->_CosCutoff = 0.0;		/* KW: -ve values not admitted */
    l->ConstantAttenuation = 1.0;
    l->LinearAttenuation = 0.0;
@@ -1337,17 +1198,6 @@ _mesa_init_lighting( struct gl_context *ctx )
    ctx->Light.ColorMaterialEnabled = GL_FALSE;
    ctx->Light.ClampVertexColor = GL_TRUE;
 
-   /* Lighting miscellaneous */
-   ctx->_ShineTabList = MALLOC_STRUCT( gl_shine_tab );
-   make_empty_list( ctx->_ShineTabList );
-   /* Allocate 10 (arbitrary) shininess lookup tables */
-   for (i = 0 ; i < 10 ; i++) {
-      struct gl_shine_tab *s = MALLOC_STRUCT( gl_shine_tab );
-      s->shininess = -1;
-      s->refcount = 0;
-      insert_at_tail( ctx->_ShineTabList, s );
-   }
-
    /* Miscellaneous */
    ctx->Light._NeedEyeCoords = GL_FALSE;
    ctx->_NeedEyeCoords = GL_FALSE;
@@ -1362,11 +1212,4 @@ _mesa_init_lighting( struct gl_context *ctx )
 void
 _mesa_free_lighting_data( struct gl_context *ctx )
 {
-   struct gl_shine_tab *s, *tmps;
-
-   /* Free lighting shininess exponentiation table */
-   foreach_s( s, tmps, ctx->_ShineTabList ) {
-      free( s );
-   }
-   free( ctx->_ShineTabList );
 }
diff --git a/mesalib/src/mesa/main/light.h b/mesalib/src/mesa/main/light.h
index 996698793..c751d6d6b 100644
--- a/mesalib/src/mesa/main/light.h
+++ b/mesalib/src/mesa/main/light.h
@@ -87,33 +87,11 @@ extern void
 _mesa_light(struct gl_context *ctx, GLuint lnum, GLenum pname, const GLfloat *params);
 
 
-/*
- * Compute dp ^ SpecularExponent.
- * Lerp between adjacent values in the f(x) lookup table, giving a
- * continuous function, with adequate overall accuracy.  (Though still
- * pretty good compared to a straight lookup).
- */
-static inline GLfloat
-_mesa_lookup_shininess(const struct gl_context *ctx, GLuint face, GLfloat dp)
-{
-   const struct gl_shine_tab *tab = ctx->_ShineTable[face];
-   float f = dp * (SHINE_TABLE_SIZE - 1);
-   int k = (int) f;
-   if (k < 0 /* gcc may cast an overflow float value to negative int value */
-	|| k > SHINE_TABLE_SIZE - 2)
-      return powf(dp, tab->shininess);
-   else
-      return tab->tab[k] + (f - k) * (tab->tab[k+1] - tab->tab[k]);
-}
-
-
 extern GLuint _mesa_material_bitmask( struct gl_context *ctx,
                                       GLenum face, GLenum pname,
                                       GLuint legal,
                                       const char * );
 
-extern void _mesa_validate_all_lighting_tables( struct gl_context *ctx );
-
 extern void _mesa_update_lighting( struct gl_context *ctx );
 
 extern void _mesa_update_tnl_spaces( struct gl_context *ctx, GLuint new_state );
@@ -132,7 +110,6 @@ extern void _mesa_allow_light_in_model( struct gl_context *ctx, GLboolean flag )
 
 #else
 #define _mesa_update_color_material( c, r ) ((void)0)
-#define _mesa_validate_all_lighting_tables( c ) ((void)0)
 #define _mesa_material_bitmask( c, f, p, l, s ) 0
 #define _mesa_init_lighting( c ) ((void)0)
 #define _mesa_free_lighting_data( c ) ((void)0)
diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h
index 9200f3fc4..480b1cf28 100644
--- a/mesalib/src/mesa/main/mtypes.h
+++ b/mesalib/src/mesa/main/mtypes.h
@@ -77,6 +77,8 @@ struct gl_texture_object;
 struct gl_context;
 struct st_context;
 struct gl_uniform_storage;
+struct prog_instruction;
+struct gl_program_parameter_list;
 /*@}*/
 
 
@@ -86,19 +88,6 @@ struct gl_uniform_storage;
 #define PRIM_UNKNOWN             (GL_POLYGON+3)
 
 
-/**
- * Shader stages. Note that these will become 5 with tessellation.
- * These MUST have the same values as gallium's PIPE_SHADER_*
- */
-typedef enum
-{
-   MESA_SHADER_VERTEX = 0,
-   MESA_SHADER_FRAGMENT = 1,
-   MESA_SHADER_GEOMETRY = 2,
-   MESA_SHADER_TYPES = 3
-} gl_shader_type;
-
-
 
 /**
  * Indexes for vertex program attributes.
@@ -640,21 +629,6 @@ struct gl_config
 /*@}*/
 
 
-#define EXP_TABLE_SIZE 512	/**< Specular exponent lookup table sizes */
-#define SHINE_TABLE_SIZE 256	/**< Material shininess lookup table sizes */
-
-/**
- * Material shininess lookup table.
- */
-struct gl_shine_tab
-{
-   struct gl_shine_tab *next, *prev;
-   GLfloat tab[SHINE_TABLE_SIZE+1];
-   GLfloat shininess;
-   GLuint refcount;
-};
-
-
 /**
  * Light source state.
  */
@@ -670,7 +644,6 @@ struct gl_light
    GLfloat SpotDirection[4];	/**< spotlight direction in eye coordinates */
    GLfloat SpotExponent;
    GLfloat SpotCutoff;		/**< in degrees */
-   GLfloat _CosCutoffNeg;	/**< = cos(SpotCutoff) */
    GLfloat _CosCutoff;		/**< = MAX(0, cos(SpotCutoff)) */
    GLfloat ConstantAttenuation;
    GLfloat LinearAttenuation;
@@ -916,23 +889,6 @@ struct gl_fog_attrib
 };
 
 
-/**
- * \brief Layout qualifiers for gl_FragDepth.
- *
- * Extension AMD_conservative_depth allows gl_FragDepth to be redeclared with
- * a layout qualifier.
- *
- * \see enum ir_depth_layout
- */
-enum gl_frag_depth_layout {
-    FRAG_DEPTH_LAYOUT_NONE, /**< No layout is specified. */
-    FRAG_DEPTH_LAYOUT_ANY,
-    FRAG_DEPTH_LAYOUT_GREATER,
-    FRAG_DEPTH_LAYOUT_LESS,
-    FRAG_DEPTH_LAYOUT_UNCHANGED
-};
-
-
 /** 
  * Hint attribute group (GL_HINT_BIT).
  * 
@@ -995,7 +951,6 @@ struct gl_light_attrib
    /*@{*/
    GLboolean _NeedEyeCoords;		
    GLboolean _NeedVertices;		/**< Use fast shader? */
-   GLbitfield _Flags;		        /**< LIGHT_* flags, see above */
    GLfloat _BaseColor[2][3];
    /*@}*/
 };
@@ -1773,6 +1728,107 @@ struct gl_evaluators
 };
 
 
+struct gl_transform_feedback_varying_info
+{
+   char *Name;
+   GLenum Type;
+   GLint Size;
+};
+
+
+/**
+ * Per-output info vertex shaders for transform feedback.
+ */
+struct gl_transform_feedback_output
+{
+   unsigned OutputRegister;
+   unsigned OutputBuffer;
+   unsigned NumComponents;
+
+   /** offset (in DWORDs) of this output within the interleaved structure */
+   unsigned DstOffset;
+
+   /**
+    * Offset into the output register of the data to output.  For example,
+    * if NumComponents is 2 and ComponentOffset is 1, then the data to
+    * offset is in the y and z components of the output register.
+    */
+   unsigned ComponentOffset;
+};
+
+
+/** Post-link transform feedback info. */
+struct gl_transform_feedback_info
+{
+   unsigned NumOutputs;
+
+   /**
+    * Number of transform feedback buffers in use by this program.
+    */
+   unsigned NumBuffers;
+
+   struct gl_transform_feedback_output *Outputs;
+
+   /** Transform feedback varyings used for the linking of this shader program.
+    *
+    * Use for glGetTransformFeedbackVarying().
+    */
+   struct gl_transform_feedback_varying_info *Varyings;
+   GLint NumVarying;
+
+   /**
+    * Total number of components stored in each buffer.  This may be used by
+    * hardware back-ends to determine the correct stride when interleaving
+    * multiple transform feedback outputs in the same buffer.
+    */
+   unsigned BufferStride[MAX_FEEDBACK_ATTRIBS];
+};
+
+
+/**
+ * Transform feedback object state
+ */
+struct gl_transform_feedback_object
+{
+   GLuint Name;  /**< AKA the object ID */
+   GLint RefCount;
+   GLboolean Active;  /**< Is transform feedback enabled? */
+   GLboolean Paused;  /**< Is transform feedback paused? */
+   GLboolean EndedAnytime; /**< Has EndTransformFeedback been called
+                                at least once? */
+
+   /** The feedback buffers */
+   GLuint BufferNames[MAX_FEEDBACK_ATTRIBS];
+   struct gl_buffer_object *Buffers[MAX_FEEDBACK_ATTRIBS];
+
+   /** Start of feedback data in dest buffer */
+   GLintptr Offset[MAX_FEEDBACK_ATTRIBS];
+   /** Max data to put into dest buffer (in bytes) */
+   GLsizeiptr Size[MAX_FEEDBACK_ATTRIBS];
+};
+
+
+/**
+ * Context state for transform feedback.
+ */
+struct gl_transform_feedback_state
+{
+   GLenum Mode;       /**< GL_POINTS, GL_LINES or GL_TRIANGLES */
+
+   /** The general binding point (GL_TRANSFORM_FEEDBACK_BUFFER) */
+   struct gl_buffer_object *CurrentBuffer;
+
+   /** The table of all transform feedback objects */
+   struct _mesa_HashTable *Objects;
+
+   /** The current xform-fb object (GL_TRANSFORM_FEEDBACK_BINDING) */
+   struct gl_transform_feedback_object *CurrentObject;
+
+   /** The default xform-fb object (Name==0) */
+   struct gl_transform_feedback_object *DefaultObject;
+};
+
+
 /**
  * Names of the various vertex/fragment program register files, etc.
  *
@@ -1834,58 +1890,23 @@ enum glsl_interp_qualifier
 };
 
 
-/** Vertex and fragment instructions */
-struct prog_instruction;
-struct gl_program_parameter_list;
-struct gl_uniform_list;
-
-struct gl_transform_feedback_varying_info {
-   char *Name;
-   GLenum Type;
-   GLint Size;
-};
-
-struct gl_transform_feedback_output {
-   unsigned OutputRegister;
-   unsigned OutputBuffer;
-   unsigned NumComponents;
-
-   /** offset (in DWORDs) of this output within the interleaved structure */
-   unsigned DstOffset;
-
-   /**
-    * Offset into the output register of the data to output.  For example,
-    * if NumComponents is 2 and ComponentOffset is 1, then the data to
-    * offset is in the y and z components of the output register.
-    */
-   unsigned ComponentOffset;
+/**
+ * \brief Layout qualifiers for gl_FragDepth.
+ *
+ * Extension AMD_conservative_depth allows gl_FragDepth to be redeclared with
+ * a layout qualifier.
+ *
+ * \see enum ir_depth_layout
+ */
+enum gl_frag_depth_layout
+{
+   FRAG_DEPTH_LAYOUT_NONE, /**< No layout is specified. */
+   FRAG_DEPTH_LAYOUT_ANY,
+   FRAG_DEPTH_LAYOUT_GREATER,
+   FRAG_DEPTH_LAYOUT_LESS,
+   FRAG_DEPTH_LAYOUT_UNCHANGED
 };
 
-/** Post-link transform feedback info. */
-struct gl_transform_feedback_info {
-   unsigned NumOutputs;
-
-   /**
-    * Number of transform feedback buffers in use by this program.
-    */
-   unsigned NumBuffers;
-
-   struct gl_transform_feedback_output *Outputs;
-
-   /** Transform feedback varyings used for the linking of this shader program.
-    *
-    * Use for glGetTransformFeedbackVarying().
-    */
-   struct gl_transform_feedback_varying_info *Varyings;
-   GLint NumVarying;
-
-   /**
-    * Total number of components stored in each buffer.  This may be used by
-    * hardware back-ends to determine the correct stride when interleaving
-    * multiple transform feedback outputs in the same buffer.
-    */
-   unsigned BufferStride[MAX_FEEDBACK_ATTRIBS];
-};
 
 /**
  * Base class for any kind of program object
@@ -2128,57 +2149,6 @@ struct gl_ati_fragment_shader_state
 };
 
 
-/**
- * Occlusion/timer query object.
- */
-struct gl_query_object
-{
-   GLenum Target;      /**< The query target, when active */
-   GLuint Id;          /**< hash table ID/name */
-   GLuint64EXT Result; /**< the counter */
-   GLboolean Active;   /**< inside Begin/EndQuery */
-   GLboolean Ready;    /**< result is ready? */
-};
-
-
-/**
- * Context state for query objects.
- */
-struct gl_query_state
-{
-   struct _mesa_HashTable *QueryObjects;
-   struct gl_query_object *CurrentOcclusionObject; /* GL_ARB_occlusion_query */
-   struct gl_query_object *CurrentTimerObject;     /* GL_EXT_timer_query */
-
-   /** GL_NV_conditional_render */
-   struct gl_query_object *CondRenderQuery;
-
-   /** GL_EXT_transform_feedback */
-   struct gl_query_object *PrimitivesGenerated;
-   struct gl_query_object *PrimitivesWritten;
-
-   /** GL_ARB_timer_query */
-   struct gl_query_object *TimeElapsed;
-
-   GLenum CondRenderMode;
-};
-
-
-/** Sync object state */
-struct gl_sync_object {
-   struct simple_node link;
-   GLenum Type;               /**< GL_SYNC_FENCE */
-   GLuint Name;               /**< Fence name */
-   GLint RefCount;            /**< Reference count */
-   GLboolean DeletePending;   /**< Object was deleted while there were still
-			       * live references (e.g., sync not yet finished)
-			       */
-   GLenum SyncCondition;
-   GLbitfield Flags;          /**< Flags passed to glFenceSync */
-   GLuint StatusFlag:1;       /**< Has the sync object been signaled? */
-};
-
-
 /** Set by #pragma directives */
 struct gl_sl_pragmas
 {
@@ -2235,6 +2205,19 @@ struct gl_shader
 
 
 /**
+ * Shader stages. Note that these will become 5 with tessellation.
+ * These MUST have the same values as gallium's PIPE_SHADER_*
+ */
+typedef enum
+{
+   MESA_SHADER_VERTEX = 0,
+   MESA_SHADER_FRAGMENT = 1,
+   MESA_SHADER_GEOMETRY = 2,
+   MESA_SHADER_TYPES = 3
+} gl_shader_type;
+
+
+/**
  * A GLSL program object.
  * Basically a linked collection of vertex and fragment shaders.
  */
@@ -2392,6 +2375,7 @@ struct gl_shader_state
    GLbitfield Flags;                    /**< Mask of GLSL_x flags */
 };
 
+
 /**
  * Compiler options for a single GLSL shaders type
  */
@@ -2424,50 +2408,58 @@ struct gl_shader_compiler_options
    struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */
 };
 
+
 /**
- * Transform feedback object state
+ * Occlusion/timer query object.
  */
-struct gl_transform_feedback_object
+struct gl_query_object
 {
-   GLuint Name;  /**< AKA the object ID */
-   GLint RefCount;
-   GLboolean Active;  /**< Is transform feedback enabled? */
-   GLboolean Paused;  /**< Is transform feedback paused? */
-   GLboolean EndedAnytime; /**< Has EndTransformFeedback been called
-                                at least once? */
-
-   /** The feedback buffers */
-   GLuint BufferNames[MAX_FEEDBACK_ATTRIBS];
-   struct gl_buffer_object *Buffers[MAX_FEEDBACK_ATTRIBS];
-
-   /** Start of feedback data in dest buffer */
-   GLintptr Offset[MAX_FEEDBACK_ATTRIBS];
-   /** Max data to put into dest buffer (in bytes) */
-   GLsizeiptr Size[MAX_FEEDBACK_ATTRIBS];
+   GLenum Target;      /**< The query target, when active */
+   GLuint Id;          /**< hash table ID/name */
+   GLuint64EXT Result; /**< the counter */
+   GLboolean Active;   /**< inside Begin/EndQuery */
+   GLboolean Ready;    /**< result is ready? */
 };
 
 
 /**
- * Context state for transform feedback.
+ * Context state for query objects.
  */
-struct gl_transform_feedback
+struct gl_query_state
 {
-   GLenum Mode;       /**< GL_POINTS, GL_LINES or GL_TRIANGLES */
+   struct _mesa_HashTable *QueryObjects;
+   struct gl_query_object *CurrentOcclusionObject; /* GL_ARB_occlusion_query */
+   struct gl_query_object *CurrentTimerObject;     /* GL_EXT_timer_query */
 
-   /** The general binding point (GL_TRANSFORM_FEEDBACK_BUFFER) */
-   struct gl_buffer_object *CurrentBuffer;
+   /** GL_NV_conditional_render */
+   struct gl_query_object *CondRenderQuery;
 
-   /** The table of all transform feedback objects */
-   struct _mesa_HashTable *Objects;
+   /** GL_EXT_transform_feedback */
+   struct gl_query_object *PrimitivesGenerated;
+   struct gl_query_object *PrimitivesWritten;
 
-   /** The current xform-fb object (GL_TRANSFORM_FEEDBACK_BINDING) */
-   struct gl_transform_feedback_object *CurrentObject;
+   /** GL_ARB_timer_query */
+   struct gl_query_object *TimeElapsed;
 
-   /** The default xform-fb object (Name==0) */
-   struct gl_transform_feedback_object *DefaultObject;
+   GLenum CondRenderMode;
 };
 
 
+/** Sync object state */
+struct gl_sync_object
+{
+   struct simple_node link;
+   GLenum Type;               /**< GL_SYNC_FENCE */
+   GLuint Name;               /**< Fence name */
+   GLint RefCount;            /**< Reference count */
+   GLboolean DeletePending;   /**< Object was deleted while there were still
+			       * live references (e.g., sync not yet finished)
+			       */
+   GLenum SyncCondition;
+   GLbitfield Flags;          /**< Flags passed to glFenceSync */
+   GLuint StatusFlag:1;       /**< Has the sync object been signaled? */
+};
+
 
 /**
  * State which can be shared by multiple contexts:
@@ -3337,7 +3329,7 @@ struct gl_context
 
    struct gl_query_state Query;  /**< occlusion, timer queries */
 
-   struct gl_transform_feedback TransformFeedback;
+   struct gl_transform_feedback_state TransformFeedback;
 
    struct gl_buffer_object *CopyReadBuffer; /**< GL_ARB_copy_buffer */
    struct gl_buffer_object *CopyWriteBuffer; /**< GL_ARB_copy_buffer */
@@ -3380,10 +3372,6 @@ struct gl_context
 
    GLuint TextureStateTimestamp; /**< detect changes to shared state */
 
-   struct gl_shine_tab *_ShineTable[2]; /**< Active shine tables */
-   struct gl_shine_tab *_ShineTabList;  /**< MRU list of inactive shine tables */
-   /**@}*/
-
    struct gl_list_extensions *ListExt; /**< driver dlist extensions */
 
    /** \name For debugging/development only */
diff --git a/mesalib/src/mesa/main/teximage.c b/mesalib/src/mesa/main/teximage.c
index 5328ae296..4fb81e62d 100644
--- a/mesalib/src/mesa/main/teximage.c
+++ b/mesalib/src/mesa/main/teximage.c
@@ -1844,23 +1844,25 @@ subtexture_error_check2( struct gl_context *ctx, GLuint dimensions,
       return GL_TRUE;
    }
    if (dimensions > 1) {
-      if (yoffset < -((GLint)destTex->Border)) {
+      GLint yBorder = (target == GL_TEXTURE_1D_ARRAY) ? 0 : destTex->Border;
+      if (yoffset < -yBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE, "glTexSubImage%dD(yoffset)",
                      dimensions);
          return GL_TRUE;
       }
-      if (yoffset + height > (GLint) (destTex->Height + destTex->Border)) {
+      if (yoffset + height > (GLint) destTex->Height + yBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE, "glTexSubImage%dD(yoffset+height)",
                      dimensions);
          return GL_TRUE;
       }
    }
    if (dimensions > 2) {
-      if (zoffset < -((GLint)destTex->Border)) {
+      GLint zBorder = (target == GL_TEXTURE_2D_ARRAY) ? 0 : destTex->Border;
+      if (zoffset < -zBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE, "glTexSubImage3D(zoffset)");
          return GL_TRUE;
       }
-      if (zoffset + depth  > (GLint) (destTex->Depth + destTex->Border)) {
+      if (zoffset + depth  > (GLint) destTex->Depth + zBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE, "glTexSubImage3D(zoffset+depth)");
          return GL_TRUE;
       }
@@ -2163,13 +2165,14 @@ copytexsubimage_error_check2( struct gl_context *ctx, GLuint dimensions,
       return GL_TRUE;
    }
    if (dimensions > 1) {
-      if (yoffset < -((GLint)teximage->Border)) {
+      GLint yBorder = (target == GL_TEXTURE_1D_ARRAY) ? 0 : teximage->Border;
+      if (yoffset < -yBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE,
                      "glCopyTexSubImage%dD(yoffset=%d)", dimensions, yoffset);
          return GL_TRUE;
       }
       /* NOTE: we're adding the border here, not subtracting! */
-      if (yoffset + height > (GLint) (teximage->Height + teximage->Border)) {
+      if (yoffset + height > (GLint) teximage->Height + yBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE,
                      "glCopyTexSubImage%dD(yoffset+height)", dimensions);
          return GL_TRUE;
@@ -2178,12 +2181,13 @@ copytexsubimage_error_check2( struct gl_context *ctx, GLuint dimensions,
 
    /* check z offset */
    if (dimensions > 2) {
-      if (zoffset < -((GLint)teximage->Border)) {
+      GLint zBorder = (target == GL_TEXTURE_2D_ARRAY) ? 0 : teximage->Border;
+      if (zoffset < -zBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE,
                      "glCopyTexSubImage%dD(zoffset)", dimensions);
          return GL_TRUE;
       }
-      if (zoffset > (GLint) (teximage->Depth + teximage->Border)) {
+      if (zoffset > (GLint) teximage->Depth + zBorder) {
          _mesa_error(ctx, GL_INVALID_VALUE,
                      "glCopyTexSubImage%dD(zoffset+depth)", dimensions);
          return GL_TRUE;
@@ -2759,10 +2763,12 @@ texsubimage(struct gl_context *ctx, GLuint dims, GLenum target, GLint level,
          /* If we have a border, offset=-1 is legal.  Bias by border width. */
          switch (dims) {
          case 3:
-            zoffset += texImage->Border;
+            if (target != GL_TEXTURE_2D_ARRAY)
+               zoffset += texImage->Border;
             /* fall-through */
          case 2:
-            yoffset += texImage->Border;
+            if (target != GL_TEXTURE_1D_ARRAY)
+               yoffset += texImage->Border;
             /* fall-through */
          case 1:
             xoffset += texImage->Border;
@@ -3034,10 +3040,12 @@ copytexsubimage(struct gl_context *ctx, GLuint dims, GLenum target, GLint level,
          /* If we have a border, offset=-1 is legal.  Bias by border width. */
          switch (dims) {
          case 3:
-            zoffset += texImage->Border;
+            if (target != GL_TEXTURE_2D_ARRAY)
+               zoffset += texImage->Border;
             /* fall-through */
          case 2:
-            yoffset += texImage->Border;
+            if (target != GL_TEXTURE_1D_ARRAY)
+               yoffset += texImage->Border;
             /* fall-through */
          case 1:
             xoffset += texImage->Border;
diff --git a/mesalib/src/mesa/main/varray.c b/mesalib/src/mesa/main/varray.c
index 39d3a27e0..a402c7b22 100644
--- a/mesalib/src/mesa/main/varray.c
+++ b/mesalib/src/mesa/main/varray.c
@@ -1117,8 +1117,9 @@ _mesa_PrimitiveRestartIndex(GLuint index)
 void GLAPIENTRY
 _mesa_VertexAttribDivisor(GLuint index, GLuint divisor)
 {
+   struct gl_client_array *array;
    GET_CURRENT_CONTEXT(ctx);
-   ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    if (!ctx->Extensions.ARB_instanced_arrays) {
       _mesa_error(ctx, GL_INVALID_OPERATION, "glVertexAttribDivisor()");
@@ -1133,7 +1134,12 @@ _mesa_VertexAttribDivisor(GLuint index, GLuint divisor)
 
    ASSERT(VERT_ATTRIB_GENERIC(index) < Elements(ctx->Array.ArrayObj->VertexAttrib));
 
-   ctx->Array.ArrayObj->VertexAttrib[VERT_ATTRIB_GENERIC(index)].InstanceDivisor = divisor;
+   array = &ctx->Array.ArrayObj->VertexAttrib[VERT_ATTRIB_GENERIC(index)];
+   if (array->InstanceDivisor != divisor) {
+      FLUSH_VERTICES(ctx, _NEW_ARRAY);
+      array->InstanceDivisor = divisor;
+      ctx->Array.NewState |= VERT_BIT(VERT_ATTRIB_GENERIC(index));
+   }
 }
 
 
diff --git a/mesalib/src/mesa/main/vtxfmt.c b/mesalib/src/mesa/main/vtxfmt.c
index f3cca937d..6fb016b9e 100644
--- a/mesalib/src/mesa/main/vtxfmt.c
+++ b/mesalib/src/mesa/main/vtxfmt.c
@@ -41,7 +41,7 @@
 #if FEATURE_beginend
 
 /**
- * Use the per-vertex functions found in <vfmt> to initialoze the given
+ * Use the per-vertex functions found in <vfmt> to initialize the given
  * API dispatch table.
  */
 static void
diff --git a/mesalib/src/mesa/state_tracker/st_draw.c b/mesalib/src/mesa/state_tracker/st_draw.c
index c4b2caccd..eb187519f 100644
--- a/mesalib/src/mesa/state_tracker/st_draw.c
+++ b/mesalib/src/mesa/state_tracker/st_draw.c
@@ -905,7 +905,6 @@ st_validate_varrays(struct gl_context *ctx,
    unsigned num_vbuffers, num_velements;
    GLuint attr;
    unsigned i;
-   unsigned old_num_user_attribs;
 
    /* must get these after state validation! */
    vp = st->vp;
@@ -914,7 +913,9 @@ st_validate_varrays(struct gl_context *ctx,
    memset(velements, 0, sizeof(struct pipe_vertex_element) * vpv->num_inputs);
 
    /* Unreference any user vertex buffers. */
-   old_num_user_attribs = st->num_user_attribs;
+   for (i = 0; i < st->num_user_attribs; i++) {
+      pipe_resource_reference(&st->user_attrib[i].buffer, NULL);
+   }
    st->num_user_attribs = 0;
 
    /*
@@ -953,10 +954,6 @@ st_validate_varrays(struct gl_context *ctx,
       assert(!vbuffer[attr].buffer);
    }
 
-   for (i = old_num_user_attribs; i < st->num_user_attribs; i++) {
-      pipe_resource_reference(&st->user_attrib[i].buffer, NULL);
-   }
-
    return GL_TRUE;
 }
 
@@ -1016,9 +1013,6 @@ st_draw_vbo(struct gl_context *ctx,
    if (st->dirty.st) {
       GLboolean vertDataEdgeFlags;
 
-      /* sanity check for pointer arithmetic below */
-      assert(sizeof(arrays[0]->Ptr[0]) == 1);
-
       vertDataEdgeFlags = arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj &&
                           arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj->Name;
       if (vertDataEdgeFlags != st->vertdata_edgeflags) {
diff --git a/mesalib/src/mesa/tnl/t_context.c b/mesalib/src/mesa/tnl/t_context.c
index 1ded44ccf..dbda7a5fc 100644
--- a/mesalib/src/mesa/tnl/t_context.c
+++ b/mesalib/src/mesa/tnl/t_context.c
@@ -46,6 +46,7 @@ GLboolean
 _tnl_CreateContext( struct gl_context *ctx )
 {
    TNLcontext *tnl;
+   GLuint i;
 
    /* Create the TNLcontext structure
     */
@@ -76,10 +77,21 @@ _tnl_CreateContext( struct gl_context *ctx )
     */
    tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
    tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-   tnl->Driver.NotifyMaterialChange = _mesa_validate_all_lighting_tables;
+   tnl->Driver.NotifyMaterialChange = _tnl_validate_shine_tables;
 
    tnl->nr_blocks = 0;
 
+   /* Lighting miscellaneous */
+   tnl->_ShineTabList = MALLOC_STRUCT( tnl_shine_tab );
+   make_empty_list( tnl->_ShineTabList );
+   /* Allocate 10 (arbitrary) shininess lookup tables */
+   for (i = 0 ; i < 10 ; i++) {
+      struct tnl_shine_tab *s = MALLOC_STRUCT( tnl_shine_tab );
+      s->shininess = -1;
+      s->refcount = 0;
+      insert_at_tail( tnl->_ShineTabList, s );
+   }
+
    /* plug in the VBO drawing function */
    vbo_set_draw_func(ctx, _tnl_vbo_draw_prims);
 
@@ -93,8 +105,15 @@ _tnl_CreateContext( struct gl_context *ctx )
 void
 _tnl_DestroyContext( struct gl_context *ctx )
 {
+   struct tnl_shine_tab *s, *tmps;
    TNLcontext *tnl = TNL_CONTEXT(ctx);
 
+   /* Free lighting shininess exponentiation table */
+   foreach_s( s, tmps, tnl->_ShineTabList ) {
+      free( s );
+   }
+   free( tnl->_ShineTabList );
+
    _tnl_destroy_pipeline( ctx );
 
    FREE(tnl);
@@ -151,8 +170,7 @@ _tnl_InvalidateState( struct gl_context *ctx, GLuint new_state )
    if (ctx->RenderMode == GL_FEEDBACK)
       tnl->render_inputs_bitset |= BITFIELD64_BIT(_TNL_ATTRIB_TEX0);
 
-   if (ctx->Point._Attenuated ||
-       (ctx->VertexProgram._Enabled && ctx->VertexProgram.PointSizeEnabled))
+   if (ctx->Point._Attenuated || ctx->VertexProgram.PointSizeEnabled)
       tnl->render_inputs_bitset |= BITFIELD64_BIT(_TNL_ATTRIB_POINTSIZE);
 
    /* check for varying vars which are written by the vertex program */
diff --git a/mesalib/src/mesa/tnl/t_context.h b/mesalib/src/mesa/tnl/t_context.h
index c5a902082..8f18ddeb1 100644
--- a/mesalib/src/mesa/tnl/t_context.h
+++ b/mesalib/src/mesa/tnl/t_context.h
@@ -384,6 +384,20 @@ struct tnl_clipspace
 };
 
 
+#define SHINE_TABLE_SIZE 256	/**< Material shininess lookup table sizes */
+
+/**
+ * Material shininess lookup table.
+ */
+struct tnl_shine_tab
+{
+   struct tnl_shine_tab *next, *prev;
+   GLfloat tab[SHINE_TABLE_SIZE+1];
+   GLfloat shininess;
+   GLuint refcount;
+};
+
+
 struct tnl_device_driver
 {
    /***
@@ -518,6 +532,10 @@ typedef struct
    GLuint nr_blocks;
 
    GLuint CurInstance;
+
+   struct tnl_shine_tab *_ShineTable[2]; /**< Active shine tables */
+   struct tnl_shine_tab *_ShineTabList;  /**< MRU list of inactive shine tables */
+   /**@}*/
 } TNLcontext;
 
 
diff --git a/mesalib/src/mesa/tnl/t_rasterpos.c b/mesalib/src/mesa/tnl/t_rasterpos.c
index 17611cd21..50b5fcb4c 100644
--- a/mesalib/src/mesa/tnl/t_rasterpos.c
+++ b/mesalib/src/mesa/tnl/t_rasterpos.c
@@ -123,8 +123,6 @@ shade_rastpos(struct gl_context *ctx,
    const struct gl_light *light;
    GLfloat diffuseColor[4], specularColor[4];  /* for RGB mode only */
 
-   _mesa_validate_all_lighting_tables( ctx );
-
    COPY_3V(diffuseColor, base[0]);
    diffuseColor[3] = CLAMP( 
       ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_DIFFUSE][3], 0.0F, 1.0F );
@@ -214,7 +212,11 @@ shade_rastpos(struct gl_context *ctx,
 	 n_dot_h = DOT3(normal, h);
 
 	 if (n_dot_h > 0.0F) {
-	    GLfloat spec_coef = _mesa_lookup_shininess(ctx, 0, n_dot_h);
+	    GLfloat shine;
+	    GLfloat spec_coef;
+
+	    shine = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS][0];
+	    spec_coef = powf(n_dot_h, shine);
 
 	    if (spec_coef > 1.0e-10) {
                if (ctx->Light.Model.ColorControl==GL_SEPARATE_SPECULAR_COLOR) {
diff --git a/mesalib/src/mesa/tnl/t_vb_light.c b/mesalib/src/mesa/tnl/t_vb_light.c
index 0760e6630..39467fac7 100644
--- a/mesalib/src/mesa/tnl/t_vb_light.c
+++ b/mesalib/src/mesa/tnl/t_vb_light.c
@@ -1,343 +1,473 @@
-/*

- * Mesa 3-D graphics library

- * Version:  6.5

- *

- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.

- *

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

- * copy of this software and associated documentation files (the "Software"),

- * to deal in the Software without restriction, including without limitation

- * the rights to use, copy, modify, merge, publish, distribute, sublicense,

- * and/or sell copies of the Software, and to permit persons to whom the

- * Software is furnished to do so, subject to the following conditions:

- *

- * The above copyright notice and this permission notice shall be included

- * in all copies or substantial portions of the Software.

- *

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

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

- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

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

- */

-

-

-

-#include "main/glheader.h"

-#include "main/colormac.h"

-#include "main/light.h"

-#include "main/macros.h"

-#include "main/imports.h"

-#include "main/simple_list.h"

-#include "main/mtypes.h"

-

-#include "math/m_translate.h"

-

-#include "t_context.h"

-#include "t_pipeline.h"

-

-#define LIGHT_TWOSIDE       0x1

-#define LIGHT_MATERIAL      0x2

-#define MAX_LIGHT_FUNC      0x4

-

-typedef void (*light_func)( struct gl_context *ctx,

-			    struct vertex_buffer *VB,

-			    struct tnl_pipeline_stage *stage,

-			    GLvector4f *input );

-

-/**

- * Information for updating current material attributes from vertex color,

- * for GL_COLOR_MATERIAL.

- */

-struct material_cursor {

-   const GLfloat *ptr;    /* points to src vertex color (in VB array) */

-   GLuint stride;         /* stride to next vertex color (bytes) */

-   GLfloat *current;      /* points to material attribute to update */

-   GLuint size;           /* vertex/color size: 1, 2, 3 or 4 */

-};

-

-/**

- * Data private to this pipeline stage.

- */

-struct light_stage_data {

-   GLvector4f Input;

-   GLvector4f LitColor[2];

-   GLvector4f LitSecondary[2];

-   light_func *light_func_tab;

-

-   struct material_cursor mat[MAT_ATTRIB_MAX];

-   GLuint mat_count;

-   GLuint mat_bitmask;

-};

-

-

-#define LIGHT_STAGE_DATA(stage) ((struct light_stage_data *)(stage->privatePtr))

-

-

-

-/**

- * In the case of colormaterial, the effected material attributes

- * should already have been bound to point to the incoming color data,

- * prior to running the pipeline.

- * This function copies the vertex's color to the material attributes

- * which are tracking glColor.

- * It's called per-vertex in the lighting loop.

- */

-static void

-update_materials(struct gl_context *ctx, struct light_stage_data *store)

-{

-   GLuint i;

-

-   for (i = 0 ; i < store->mat_count ; i++) {

-      /* update the material */

-      COPY_CLEAN_4V(store->mat[i].current, store->mat[i].size, store->mat[i].ptr);

-      /* increment src vertex color pointer */

-      STRIDE_F(store->mat[i].ptr, store->mat[i].stride);

-   }

-      

-   /* recompute derived light/material values */

-   _mesa_update_material( ctx, store->mat_bitmask );

-   /* XXX we should only call this if we're tracking/changing the specular

-    * exponent.

-    */

-   _mesa_validate_all_lighting_tables( ctx );

-}

-

-

-/**

- * Prepare things prior to running the lighting stage.

- * Return number of material attributes which will track vertex color.

- */

-static GLuint

-prepare_materials(struct gl_context *ctx,

-                  struct vertex_buffer *VB, struct light_stage_data *store)

-{

-   GLuint i;

-   

-   store->mat_count = 0;

-   store->mat_bitmask = 0;

-

-   /* Examine the ColorMaterialBitmask to determine which materials

-    * track vertex color.  Override the material attribute's pointer

-    * with the color pointer for each one.

-    */

-   if (ctx->Light.ColorMaterialEnabled) {

-      const GLuint bitmask = ctx->Light.ColorMaterialBitmask;

-      for (i = 0 ; i < MAT_ATTRIB_MAX ; i++)

-	 if (bitmask & (1<<i))

-	    VB->AttribPtr[_TNL_ATTRIB_MAT_FRONT_AMBIENT + i] = VB->AttribPtr[_TNL_ATTRIB_COLOR0];

-   }

-

-   /* Now, for each material attribute that's tracking vertex color, save

-    * some values (ptr, stride, size, current) that we'll need in

-    * update_materials(), above, that'll actually copy the vertex color to

-    * the material attribute(s).

-    */

-   for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {

-      if (VB->AttribPtr[i]->stride) {

-	 const GLuint j = store->mat_count++;

-	 const GLuint attr = i - _TNL_ATTRIB_MAT_FRONT_AMBIENT;

-	 store->mat[j].ptr    = VB->AttribPtr[i]->start;

-	 store->mat[j].stride = VB->AttribPtr[i]->stride;

-	 store->mat[j].size   = VB->AttribPtr[i]->size;

-	 store->mat[j].current = ctx->Light.Material.Attrib[attr];

-	 store->mat_bitmask |= (1<<attr);

-      }

-   }

-

-   /* FIXME: Is this already done?

-    */

-   _mesa_update_material( ctx, ~0 );

-   _mesa_validate_all_lighting_tables( ctx );

-

-   return store->mat_count;

-}

-

-/* Tables for all the shading functions.

- */

-static light_func _tnl_light_tab[MAX_LIGHT_FUNC];

-static light_func _tnl_light_fast_tab[MAX_LIGHT_FUNC];

-static light_func _tnl_light_fast_single_tab[MAX_LIGHT_FUNC];

-static light_func _tnl_light_spec_tab[MAX_LIGHT_FUNC];

-

-#define TAG(x)           x

-#define IDX              (0)

-#include "t_vb_lighttmp.h"

-

-#define TAG(x)           x##_twoside

-#define IDX              (LIGHT_TWOSIDE)

-#include "t_vb_lighttmp.h"

-

-#define TAG(x)           x##_material

-#define IDX              (LIGHT_MATERIAL)

-#include "t_vb_lighttmp.h"

-

-#define TAG(x)           x##_twoside_material

-#define IDX              (LIGHT_TWOSIDE|LIGHT_MATERIAL)

-#include "t_vb_lighttmp.h"

-

-

-static void init_lighting_tables( void )

-{

-   static int done;

-

-   if (!done) {

-      init_light_tab();

-      init_light_tab_twoside();

-      init_light_tab_material();

-      init_light_tab_twoside_material();

-      done = 1;

-   }

-}

-

-

-static GLboolean run_lighting( struct gl_context *ctx, 

-			       struct tnl_pipeline_stage *stage )

-{

-   struct light_stage_data *store = LIGHT_STAGE_DATA(stage);

-   TNLcontext *tnl = TNL_CONTEXT(ctx);

-   struct vertex_buffer *VB = &tnl->vb;

-   GLvector4f *input = ctx->_NeedEyeCoords ? VB->EyePtr : VB->AttribPtr[_TNL_ATTRIB_POS];

-   GLuint idx;

-

-   if (!ctx->Light.Enabled || ctx->VertexProgram._Current)

-      return GL_TRUE;

-

-   /* Make sure we can talk about position x,y and z:

-    */

-   if (input->size <= 2 && input == VB->AttribPtr[_TNL_ATTRIB_POS]) {

-

-      _math_trans_4f( store->Input.data,

-		      VB->AttribPtr[_TNL_ATTRIB_POS]->data,

-		      VB->AttribPtr[_TNL_ATTRIB_POS]->stride,

-		      GL_FLOAT,

-		      VB->AttribPtr[_TNL_ATTRIB_POS]->size,

-		      0,

-		      VB->Count );

-

-      if (input->size <= 2) {

-	 /* Clean z.

-	  */

-	 _mesa_vector4f_clean_elem(&store->Input, VB->Count, 2);

-      }

-	 

-      if (input->size <= 1) {

-	 /* Clean y.

-	  */

-	 _mesa_vector4f_clean_elem(&store->Input, VB->Count, 1);

-      }

-

-      input = &store->Input;

-   }

-   

-   idx = 0;

-

-   if (prepare_materials( ctx, VB, store ))

-      idx |= LIGHT_MATERIAL;

-

-   if (ctx->Light.Model.TwoSide)

-      idx |= LIGHT_TWOSIDE;

-

-   /* The individual functions know about replaying side-effects

-    * vs. full re-execution. 

-    */

-   store->light_func_tab[idx]( ctx, VB, stage, input );

-

-   return GL_TRUE;

-}

-

-

-/* Called in place of do_lighting when the light table may have changed.

- */

-static void validate_lighting( struct gl_context *ctx,

-					struct tnl_pipeline_stage *stage )

-{

-   light_func *tab;

-

-   if (!ctx->Light.Enabled || ctx->VertexProgram._Current)

-      return;

-

-   if (ctx->Light._NeedVertices) {

-      if (ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)

-	 tab = _tnl_light_spec_tab;

-      else

-	 tab = _tnl_light_tab;

-   }

-   else {

-      if (ctx->Light.EnabledList.next == ctx->Light.EnabledList.prev)

-	 tab = _tnl_light_fast_single_tab;

-      else

-	 tab = _tnl_light_fast_tab;

-   }

-

-

-   LIGHT_STAGE_DATA(stage)->light_func_tab = tab;

-

-   /* This and the above should only be done on _NEW_LIGHT:

-    */

-   TNL_CONTEXT(ctx)->Driver.NotifyMaterialChange( ctx );

-}

-

-

-

-/* Called the first time stage->run is called.  In effect, don't

- * allocate data until the first time the stage is run.

- */

-static GLboolean init_lighting( struct gl_context *ctx,

-				struct tnl_pipeline_stage *stage )

-{

-   TNLcontext *tnl = TNL_CONTEXT(ctx);

-   struct light_stage_data *store;

-   GLuint size = tnl->vb.Size;

-

-   stage->privatePtr = MALLOC(sizeof(*store));

-   store = LIGHT_STAGE_DATA(stage);

-   if (!store)

-      return GL_FALSE;

-

-   /* Do onetime init.

-    */

-   init_lighting_tables();

-

-   _mesa_vector4f_alloc( &store->Input, 0, size, 32 );

-   _mesa_vector4f_alloc( &store->LitColor[0], 0, size, 32 );

-   _mesa_vector4f_alloc( &store->LitColor[1], 0, size, 32 );

-   _mesa_vector4f_alloc( &store->LitSecondary[0], 0, size, 32 );

-   _mesa_vector4f_alloc( &store->LitSecondary[1], 0, size, 32 );

-

-   store->LitColor[0].size = 4;

-   store->LitColor[1].size = 4;

-   store->LitSecondary[0].size = 3;

-   store->LitSecondary[1].size = 3;

-

-   return GL_TRUE;

-}

-

-

-

-

-static void dtr( struct tnl_pipeline_stage *stage )

-{

-   struct light_stage_data *store = LIGHT_STAGE_DATA(stage);

-

-   if (store) {

-      _mesa_vector4f_free( &store->Input );

-      _mesa_vector4f_free( &store->LitColor[0] );

-      _mesa_vector4f_free( &store->LitColor[1] );

-      _mesa_vector4f_free( &store->LitSecondary[0] );

-      _mesa_vector4f_free( &store->LitSecondary[1] );

-      FREE( store );

-      stage->privatePtr = NULL;

-   }

-}

-

-const struct tnl_pipeline_stage _tnl_lighting_stage =

-{

-   "lighting",			/* name */

-   NULL,			/* private_data */

-   init_lighting,

-   dtr,				/* destroy */

-   validate_lighting,

-   run_lighting

-};

+/*
+ * Mesa 3-D graphics library
+ * Version:  6.5
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+
+#include "main/glheader.h"
+#include "main/colormac.h"
+#include "main/light.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/mtypes.h"
+
+#include "math/m_translate.h"
+
+#include "t_context.h"
+#include "t_pipeline.h"
+#include "tnl.h"
+
+#define LIGHT_TWOSIDE       0x1
+#define LIGHT_MATERIAL      0x2
+#define MAX_LIGHT_FUNC      0x4
+
+typedef void (*light_func)( struct gl_context *ctx,
+			    struct vertex_buffer *VB,
+			    struct tnl_pipeline_stage *stage,
+			    GLvector4f *input );
+
+/**
+ * Information for updating current material attributes from vertex color,
+ * for GL_COLOR_MATERIAL.
+ */
+struct material_cursor {
+   const GLfloat *ptr;    /* points to src vertex color (in VB array) */
+   GLuint stride;         /* stride to next vertex color (bytes) */
+   GLfloat *current;      /* points to material attribute to update */
+   GLuint size;           /* vertex/color size: 1, 2, 3 or 4 */
+};
+
+/**
+ * Data private to this pipeline stage.
+ */
+struct light_stage_data {
+   GLvector4f Input;
+   GLvector4f LitColor[2];
+   GLvector4f LitSecondary[2];
+   light_func *light_func_tab;
+
+   struct material_cursor mat[MAT_ATTRIB_MAX];
+   GLuint mat_count;
+   GLuint mat_bitmask;
+};
+
+
+#define LIGHT_STAGE_DATA(stage) ((struct light_stage_data *)(stage->privatePtr))
+
+
+
+/**********************************************************************/
+/*****                  Lighting computation                      *****/
+/**********************************************************************/
+
+
+/*
+ * Notes:
+ *   When two-sided lighting is enabled we compute the color (or index)
+ *   for both the front and back side of the primitive.  Then, when the
+ *   orientation of the facet is later learned, we can determine which
+ *   color (or index) to use for rendering.
+ *
+ *   KW: We now know orientation in advance and only shade for
+ *       the side or sides which are actually required.
+ *
+ * Variables:
+ *   n = normal vector
+ *   V = vertex position
+ *   P = light source position
+ *   Pe = (0,0,0,1)
+ *
+ * Precomputed:
+ *   IF P[3]==0 THEN
+ *       // light at infinity
+ *       IF local_viewer THEN
+ *           _VP_inf_norm = unit vector from V to P      // Precompute
+ *       ELSE
+ *           // eye at infinity
+ *           _h_inf_norm = Normalize( VP + <0,0,1> )     // Precompute
+ *       ENDIF
+ *   ENDIF
+ *
+ * Functions:
+ *   Normalize( v ) = normalized vector v
+ *   Magnitude( v ) = length of vector v
+ */
+
+
+
+static void
+validate_shine_table( struct gl_context *ctx, GLuint side, GLfloat shininess )
+{
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   struct tnl_shine_tab *list = tnl->_ShineTabList;
+   struct tnl_shine_tab *s;
+
+   ASSERT(side < 2);
+
+   foreach(s, list)
+      if ( s->shininess == shininess )
+	 break;
+
+   if (s == list) {
+      GLint j;
+      GLfloat *m;
+
+      foreach(s, list)
+	 if (s->refcount == 0)
+	    break;
+
+      m = s->tab;
+      m[0] = 0.0;
+      if (shininess == 0.0) {
+	 for (j = 1 ; j <= SHINE_TABLE_SIZE ; j++)
+	    m[j] = 1.0;
+      }
+      else {
+	 for (j = 1 ; j < SHINE_TABLE_SIZE ; j++) {
+            GLdouble t, x = j / (GLfloat) (SHINE_TABLE_SIZE - 1);
+            if (x < 0.005) /* underflow check */
+               x = 0.005;
+            t = pow(x, shininess);
+	    if (t > 1e-20)
+	       m[j] = (GLfloat) t;
+	    else
+	       m[j] = 0.0;
+	 }
+	 m[SHINE_TABLE_SIZE] = 1.0;
+      }
+
+      s->shininess = shininess;
+   }
+
+   if (tnl->_ShineTable[side])
+      tnl->_ShineTable[side]->refcount--;
+
+   tnl->_ShineTable[side] = s;
+   move_to_tail( list, s );
+   s->refcount++;
+}
+
+
+void
+_tnl_validate_shine_tables( struct gl_context *ctx )
+{
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   GLfloat shininess;
+   
+   shininess = ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS][0];
+   if (!tnl->_ShineTable[0] || tnl->_ShineTable[0]->shininess != shininess)
+      validate_shine_table( ctx, 0, shininess );
+
+   shininess = ctx->Light.Material.Attrib[MAT_ATTRIB_BACK_SHININESS][0];
+   if (!tnl->_ShineTable[1] || tnl->_ShineTable[1]->shininess != shininess)
+      validate_shine_table( ctx, 1, shininess );
+}
+
+
+/**
+ * In the case of colormaterial, the effected material attributes
+ * should already have been bound to point to the incoming color data,
+ * prior to running the pipeline.
+ * This function copies the vertex's color to the material attributes
+ * which are tracking glColor.
+ * It's called per-vertex in the lighting loop.
+ */
+static void
+update_materials(struct gl_context *ctx, struct light_stage_data *store)
+{
+   GLuint i;
+
+   for (i = 0 ; i < store->mat_count ; i++) {
+      /* update the material */
+      COPY_CLEAN_4V(store->mat[i].current, store->mat[i].size, store->mat[i].ptr);
+      /* increment src vertex color pointer */
+      STRIDE_F(store->mat[i].ptr, store->mat[i].stride);
+   }
+      
+   /* recompute derived light/material values */
+   _mesa_update_material( ctx, store->mat_bitmask );
+   /* XXX we should only call this if we're tracking/changing the specular
+    * exponent.
+    */
+   _tnl_validate_shine_tables( ctx );
+}
+
+
+/**
+ * Prepare things prior to running the lighting stage.
+ * Return number of material attributes which will track vertex color.
+ */
+static GLuint
+prepare_materials(struct gl_context *ctx,
+                  struct vertex_buffer *VB, struct light_stage_data *store)
+{
+   GLuint i;
+   
+   store->mat_count = 0;
+   store->mat_bitmask = 0;
+
+   /* Examine the ColorMaterialBitmask to determine which materials
+    * track vertex color.  Override the material attribute's pointer
+    * with the color pointer for each one.
+    */
+   if (ctx->Light.ColorMaterialEnabled) {
+      const GLuint bitmask = ctx->Light.ColorMaterialBitmask;
+      for (i = 0 ; i < MAT_ATTRIB_MAX ; i++)
+	 if (bitmask & (1<<i))
+	    VB->AttribPtr[_TNL_ATTRIB_MAT_FRONT_AMBIENT + i] = VB->AttribPtr[_TNL_ATTRIB_COLOR0];
+   }
+
+   /* Now, for each material attribute that's tracking vertex color, save
+    * some values (ptr, stride, size, current) that we'll need in
+    * update_materials(), above, that'll actually copy the vertex color to
+    * the material attribute(s).
+    */
+   for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {
+      if (VB->AttribPtr[i]->stride) {
+	 const GLuint j = store->mat_count++;
+	 const GLuint attr = i - _TNL_ATTRIB_MAT_FRONT_AMBIENT;
+	 store->mat[j].ptr    = VB->AttribPtr[i]->start;
+	 store->mat[j].stride = VB->AttribPtr[i]->stride;
+	 store->mat[j].size   = VB->AttribPtr[i]->size;
+	 store->mat[j].current = ctx->Light.Material.Attrib[attr];
+	 store->mat_bitmask |= (1<<attr);
+      }
+   }
+
+   /* FIXME: Is this already done?
+    */
+   _mesa_update_material( ctx, ~0 );
+
+   _tnl_validate_shine_tables( ctx );
+
+   return store->mat_count;
+}
+
+/*
+ * Compute dp ^ SpecularExponent.
+ * Lerp between adjacent values in the f(x) lookup table, giving a
+ * continuous function, with adequate overall accuracy.  (Though still
+ * pretty good compared to a straight lookup).
+ */
+static inline GLfloat
+lookup_shininess(const struct gl_context *ctx, GLuint face, GLfloat dp)
+{
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   const struct tnl_shine_tab *tab = tnl->_ShineTable[face];
+   float f = dp * (SHINE_TABLE_SIZE - 1);
+   int k = (int) f;
+   if (k < 0 /* gcc may cast an overflow float value to negative int value */
+	|| k > SHINE_TABLE_SIZE - 2)
+      return powf(dp, tab->shininess);
+   else
+      return tab->tab[k] + (f - k) * (tab->tab[k+1] - tab->tab[k]);
+}
+
+/* Tables for all the shading functions.
+ */
+static light_func _tnl_light_tab[MAX_LIGHT_FUNC];
+static light_func _tnl_light_fast_tab[MAX_LIGHT_FUNC];
+static light_func _tnl_light_fast_single_tab[MAX_LIGHT_FUNC];
+static light_func _tnl_light_spec_tab[MAX_LIGHT_FUNC];
+
+#define TAG(x)           x
+#define IDX              (0)
+#include "t_vb_lighttmp.h"
+
+#define TAG(x)           x##_twoside
+#define IDX              (LIGHT_TWOSIDE)
+#include "t_vb_lighttmp.h"
+
+#define TAG(x)           x##_material
+#define IDX              (LIGHT_MATERIAL)
+#include "t_vb_lighttmp.h"
+
+#define TAG(x)           x##_twoside_material
+#define IDX              (LIGHT_TWOSIDE|LIGHT_MATERIAL)
+#include "t_vb_lighttmp.h"
+
+
+static void init_lighting_tables( void )
+{
+   static int done;
+
+   if (!done) {
+      init_light_tab();
+      init_light_tab_twoside();
+      init_light_tab_material();
+      init_light_tab_twoside_material();
+      done = 1;
+   }
+}
+
+
+static GLboolean run_lighting( struct gl_context *ctx, 
+			       struct tnl_pipeline_stage *stage )
+{
+   struct light_stage_data *store = LIGHT_STAGE_DATA(stage);
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   struct vertex_buffer *VB = &tnl->vb;
+   GLvector4f *input = ctx->_NeedEyeCoords ? VB->EyePtr : VB->AttribPtr[_TNL_ATTRIB_POS];
+   GLuint idx;
+
+   if (!ctx->Light.Enabled || ctx->VertexProgram._Current)
+      return GL_TRUE;
+
+   /* Make sure we can talk about position x,y and z:
+    */
+   if (input->size <= 2 && input == VB->AttribPtr[_TNL_ATTRIB_POS]) {
+
+      _math_trans_4f( store->Input.data,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
+		      GL_FLOAT,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->size,
+		      0,
+		      VB->Count );
+
+      if (input->size <= 2) {
+	 /* Clean z.
+	  */
+	 _mesa_vector4f_clean_elem(&store->Input, VB->Count, 2);
+      }
+	 
+      if (input->size <= 1) {
+	 /* Clean y.
+	  */
+	 _mesa_vector4f_clean_elem(&store->Input, VB->Count, 1);
+      }
+
+      input = &store->Input;
+   }
+   
+   idx = 0;
+
+   if (prepare_materials( ctx, VB, store ))
+      idx |= LIGHT_MATERIAL;
+
+   if (ctx->Light.Model.TwoSide)
+      idx |= LIGHT_TWOSIDE;
+
+   /* The individual functions know about replaying side-effects
+    * vs. full re-execution. 
+    */
+   store->light_func_tab[idx]( ctx, VB, stage, input );
+
+   return GL_TRUE;
+}
+
+
+/* Called in place of do_lighting when the light table may have changed.
+ */
+static void validate_lighting( struct gl_context *ctx,
+					struct tnl_pipeline_stage *stage )
+{
+   light_func *tab;
+
+   if (!ctx->Light.Enabled || ctx->VertexProgram._Current)
+      return;
+
+   if (ctx->Light._NeedVertices) {
+      if (ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)
+	 tab = _tnl_light_spec_tab;
+      else
+	 tab = _tnl_light_tab;
+   }
+   else {
+      if (ctx->Light.EnabledList.next == ctx->Light.EnabledList.prev)
+	 tab = _tnl_light_fast_single_tab;
+      else
+	 tab = _tnl_light_fast_tab;
+   }
+
+
+   LIGHT_STAGE_DATA(stage)->light_func_tab = tab;
+
+   /* This and the above should only be done on _NEW_LIGHT:
+    */
+   TNL_CONTEXT(ctx)->Driver.NotifyMaterialChange( ctx );
+}
+
+
+
+/* Called the first time stage->run is called.  In effect, don't
+ * allocate data until the first time the stage is run.
+ */
+static GLboolean init_lighting( struct gl_context *ctx,
+				struct tnl_pipeline_stage *stage )
+{
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   struct light_stage_data *store;
+   GLuint size = tnl->vb.Size;
+
+   stage->privatePtr = MALLOC(sizeof(*store));
+   store = LIGHT_STAGE_DATA(stage);
+   if (!store)
+      return GL_FALSE;
+
+   /* Do onetime init.
+    */
+   init_lighting_tables();
+
+   _mesa_vector4f_alloc( &store->Input, 0, size, 32 );
+   _mesa_vector4f_alloc( &store->LitColor[0], 0, size, 32 );
+   _mesa_vector4f_alloc( &store->LitColor[1], 0, size, 32 );
+   _mesa_vector4f_alloc( &store->LitSecondary[0], 0, size, 32 );
+   _mesa_vector4f_alloc( &store->LitSecondary[1], 0, size, 32 );
+
+   store->LitColor[0].size = 4;
+   store->LitColor[1].size = 4;
+   store->LitSecondary[0].size = 3;
+   store->LitSecondary[1].size = 3;
+
+   return GL_TRUE;
+}
+
+
+
+
+static void dtr( struct tnl_pipeline_stage *stage )
+{
+   struct light_stage_data *store = LIGHT_STAGE_DATA(stage);
+
+   if (store) {
+      _mesa_vector4f_free( &store->Input );
+      _mesa_vector4f_free( &store->LitColor[0] );
+      _mesa_vector4f_free( &store->LitColor[1] );
+      _mesa_vector4f_free( &store->LitSecondary[0] );
+      _mesa_vector4f_free( &store->LitSecondary[1] );
+      FREE( store );
+      stage->privatePtr = NULL;
+   }
+}
+
+const struct tnl_pipeline_stage _tnl_lighting_stage =
+{
+   "lighting",			/* name */
+   NULL,			/* private_data */
+   init_lighting,
+   dtr,				/* destroy */
+   validate_lighting,
+   run_lighting
+};
diff --git a/mesalib/src/mesa/tnl/t_vb_lighttmp.h b/mesalib/src/mesa/tnl/t_vb_lighttmp.h
index 1041a24e7..1944e5ddc 100644
--- a/mesalib/src/mesa/tnl/t_vb_lighttmp.h
+++ b/mesalib/src/mesa/tnl/t_vb_lighttmp.h
@@ -204,7 +204,7 @@ static void TAG(light_rgba_spec)( struct gl_context *ctx,
 	 n_dot_h = correction * DOT3(normal, h);
 
 	 if (n_dot_h > 0.0F) {
-	    GLfloat spec_coef = _mesa_lookup_shininess(ctx, side, n_dot_h);
+	    GLfloat spec_coef = lookup_shininess(ctx, side, n_dot_h);
 	    if (spec_coef > 1.0e-10) {
 	       spec_coef *= attenuation;
 	       ACC_SCALE_SCALAR_3V( spec[side], spec_coef,
@@ -383,7 +383,7 @@ static void TAG(light_rgba)( struct gl_context *ctx,
 	    n_dot_h = correction * DOT3(normal, h);
 
 	    if (n_dot_h > 0.0F) {
-	       GLfloat spec_coef = _mesa_lookup_shininess(ctx, side, n_dot_h);
+	       GLfloat spec_coef = lookup_shininess(ctx, side, n_dot_h);
 	       ACC_SCALE_SCALAR_3V( contrib, spec_coef,
 				    light->_MatSpecular[side]);
 	    }
@@ -483,7 +483,7 @@ static void TAG(light_fast_rgba_single)( struct gl_context *ctx,
          COPY_3V(sum, base[1]);
          ACC_SCALE_SCALAR_3V(sum, -n_dot_VP, light->_MatDiffuse[1]);
          if (n_dot_h > 0.0F) {
-            GLfloat spec = _mesa_lookup_shininess(ctx, 1, n_dot_h);
+            GLfloat spec = lookup_shininess(ctx, 1, n_dot_h);
             ACC_SCALE_SCALAR_3V(sum, spec, light->_MatSpecular[1]);
          }
          COPY_3V(Bcolor[j], sum );
@@ -497,7 +497,7 @@ static void TAG(light_fast_rgba_single)( struct gl_context *ctx,
 	 COPY_3V(sum, base[0]);
 	 ACC_SCALE_SCALAR_3V(sum, n_dot_VP, light->_MatDiffuse[0]);
 	 if (n_dot_h > 0.0F) {
-            GLfloat spec = _mesa_lookup_shininess(ctx, 0, n_dot_h);
+            GLfloat spec = lookup_shininess(ctx, 0, n_dot_h);
 	    ACC_SCALE_SCALAR_3V(sum, spec, light->_MatSpecular[0]);
 	 }
 	 COPY_3V(Fcolor[j], sum );
@@ -589,7 +589,7 @@ static void TAG(light_fast_rgba)( struct gl_context *ctx,
 	    ACC_SCALE_SCALAR_3V(sum[0], n_dot_VP, light->_MatDiffuse[0]);
 	    n_dot_h = DOT3(normal, light->_h_inf_norm);
 	    if (n_dot_h > 0.0F) {
-               spec = _mesa_lookup_shininess(ctx, 0, n_dot_h);
+               spec = lookup_shininess(ctx, 0, n_dot_h);
 	       ACC_SCALE_SCALAR_3V( sum[0], spec, light->_MatSpecular[0]);
 	    }
 	 }
@@ -598,7 +598,7 @@ static void TAG(light_fast_rgba)( struct gl_context *ctx,
 	    ACC_SCALE_SCALAR_3V(sum[1], -n_dot_VP, light->_MatDiffuse[1]);
 	    n_dot_h = -DOT3(normal, light->_h_inf_norm);
 	    if (n_dot_h > 0.0F) {
-               spec = _mesa_lookup_shininess(ctx, 1, n_dot_h);
+               spec = lookup_shininess(ctx, 1, n_dot_h);
 	       ACC_SCALE_SCALAR_3V( sum[1], spec, light->_MatSpecular[1]);
 	    }
 	 }
diff --git a/mesalib/src/mesa/tnl/tnl.h b/mesalib/src/mesa/tnl/tnl.h
index d3889811e..434bd7fcd 100644
--- a/mesalib/src/mesa/tnl/tnl.h
+++ b/mesalib/src/mesa/tnl/tnl.h
@@ -101,4 +101,7 @@ _mesa_load_tracked_matrices(struct gl_context *ctx);
 extern void
 _tnl_RasterPos(struct gl_context *ctx, const GLfloat vObj[4]);
 
+extern void
+_tnl_validate_shine_tables( struct gl_context *ctx );
+
 #endif
diff --git a/mesalib/src/mesa/x86/gen_matypes.c b/mesalib/src/mesa/x86/gen_matypes.c
index 7af82b4ae..4fe99e799 100644
--- a/mesalib/src/mesa/x86/gen_matypes.c
+++ b/mesalib/src/mesa/x86/gen_matypes.c
@@ -103,7 +103,6 @@ int main( int argc, char **argv )
    OFFSET( "CTX_LIGHT_COLOR_MAT_ENABLED ", struct gl_context, Light.ColorMaterialEnabled );
    OFFSET( "CTX_LIGHT_ENABLED_LIST      ", struct gl_context, Light.EnabledList );
    OFFSET( "CTX_LIGHT_NEED_VERTS        ", struct gl_context, Light._NeedVertices );
-   OFFSET( "CTX_LIGHT_FLAGS             ", struct gl_context, Light._Flags );
    OFFSET( "CTX_LIGHT_BASE_COLOR        ", struct gl_context, Light._BaseColor );