Merge commit 'origin/master' into gallium-0.2

Conflicts:

	progs/glsl/Makefile

19 files changed, 320 insertions, 151 deletions

diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index 467abe4d7c..410052b3c2 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -117,6 +117,26 @@ intelDmaPrimitive(struct intel_context *intel, GLenum prim)
    intel_set_prim(intel, hw_prim[prim]);
 }
 
+static inline GLuint intel_get_vb_max(struct intel_context *intel)
+{
+   GLuint ret;
+
+   if (intel->intelScreen->no_vbo)
+      ret = intel->batch->size - 1500;
+   else
+      ret = INTEL_VB_SIZE;
+   ret /= (intel->vertex_size * 4);
+   return ret;
+}
+
+static inline GLuint intel_get_current_max(struct intel_context *intel)
+{
+
+   if (intel->intelScreen->no_vbo)
+      return intel_get_vb_max(intel);
+   else
+      return (INTEL_VB_SIZE - intel->prim.current_offset) / (intel->vertex_size * 4);
+}
 
 #define LOCAL_VARS struct intel_context *intel = intel_context(ctx)
 #define INIT( prim ) 				\
@@ -126,9 +146,8 @@ do {						\
 
 #define FLUSH() INTEL_FIREVERTICES(intel)
 
-#define GET_SUBSEQUENT_VB_MAX_VERTS() (INTEL_VB_SIZE / (intel->vertex_size * 4))
-#define GET_CURRENT_VB_MAX_VERTS() \
-   ((INTEL_VB_SIZE - intel->prim.current_offset) / (intel->vertex_size * 4))
+#define GET_SUBSEQUENT_VB_MAX_VERTS() intel_get_vb_max(intel)
+#define GET_CURRENT_VB_MAX_VERTS() intel_get_current_max(intel)
 
 #define ALLOC_VERTS(nr) intel_get_prim_space(intel, nr)
 
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index 797d6c5858..c4708dc7ab 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -61,9 +61,101 @@ static void intelRenderPrimitive(GLcontext * ctx, GLenum prim);
 static void intelRasterPrimitive(GLcontext * ctx, GLenum rprim,
                                  GLuint hwprim);
 
+static void
+intel_flush_inline_primitive(struct intel_context *intel)
+{
+   GLuint used = intel->batch->ptr - intel->prim.start_ptr;
+
+   assert(intel->prim.primitive != ~0);
+
+/*    _mesa_printf("/\n"); */
+
+   if (used < 8)
+      goto do_discard;
+
+   *(int *) intel->prim.start_ptr = (_3DPRIMITIVE |
+                                     intel->prim.primitive | (used / 4 - 2));
+
+   goto finished;
+
+ do_discard:
+   intel->batch->ptr -= used;
+
+ finished:
+   intel->prim.primitive = ~0;
+   intel->prim.start_ptr = 0;
+   intel->prim.flush = 0;
+}
+
+static void intel_start_inline(struct intel_context *intel, uint32_t prim)
+{
+   BATCH_LOCALS;
+   uint32_t batch_flags = LOOP_CLIPRECTS;
+
+   intel_wait_flips(intel);
+   intel->vtbl.emit_state(intel);
+
+   intel->no_batch_wrap = GL_TRUE;
+
+   /*_mesa_printf("%s *", __progname);*/
+
+   /* Emit a slot which will be filled with the inline primitive
+    * command later.
+    */
+   BEGIN_BATCH(2, batch_flags);
+   OUT_BATCH(0);
+
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
+
+   intel->prim.start_ptr = intel->batch->ptr;
+   intel->prim.primitive = prim;
+   intel->prim.flush = intel_flush_inline_primitive;
+
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+   intel->no_batch_wrap = GL_FALSE;
+/*    _mesa_printf(">"); */
+}
+
+static void intel_wrap_inline(struct intel_context *intel)
+{
+   GLuint prim = intel->prim.primitive;
+
+   intel_flush_inline_primitive(intel);
+   intel_batchbuffer_flush(intel->batch);
+   intel_start_inline(intel, prim);  /* ??? */
+}
+
+static GLuint *intel_extend_inline(struct intel_context *intel, GLuint dwords)
+{
+   GLuint sz = dwords * sizeof(GLuint);
+   GLuint *ptr;
+
+   assert(intel->prim.flush == intel_flush_inline_primitive);
+
+   if (intel_batchbuffer_space(intel->batch) < sz)
+      intel_wrap_inline(intel);
+
+/*    _mesa_printf("."); */
+
+   intel->vtbl.assert_not_dirty(intel);
+
+   ptr = (GLuint *) intel->batch->ptr;
+   intel->batch->ptr += sz;
+
+   return ptr;
+}
+
 /** Sets the primitive type for a primitive sequence, flushing as needed. */
 void intel_set_prim(struct intel_context *intel, uint32_t prim)
 {
+   /* if we have no VBOs */
+
+   if (intel->intelScreen->no_vbo) {
+      intel_start_inline(intel, prim);
+      return;
+   }
    if (prim != intel->prim.primitive) {
       INTEL_FIREVERTICES(intel);
       intel->prim.primitive = prim;
@@ -75,6 +167,10 @@ uint32_t *intel_get_prim_space(struct intel_context *intel, unsigned int count)
 {
    uint32_t *addr;
 
+   if (intel->intelScreen->no_vbo) {
+      return intel_extend_inline(intel, count * intel->vertex_size);
+   }
+
    /* Check for space in the existing VB */
    if (intel->prim.vb_bo == NULL ||
        (intel->prim.current_offset +
@@ -155,7 +251,7 @@ void intel_flush_prim(struct intel_context *intel)
 
 #if 0
    printf("emitting %d..%d=%d vertices size %d\n", offset,
-	  intel->prim.current_offset, intel->prim.count,
+	  intel->prim.current_offset, count,
 	  intel->vertex_size * 4);
 #endif
 
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index e3904be977..77980109cd 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -133,16 +133,16 @@ struct brw_context;
 #define BRW_NEW_PSP                     0x800
 #define BRW_NEW_METAOPS                 0x1000
 #define BRW_NEW_FENCE                   0x2000
-#define BRW_NEW_LOCK                    0x4000
-#define BRW_NEW_INDICES			0x8000
-#define BRW_NEW_VERTICES		0x10000
+#define BRW_NEW_INDICES			0x4000
+#define BRW_NEW_VERTICES		0x8000
 /**
  * Used for any batch entry with a relocated pointer that will be used
  * by any 3D rendering.
  */
-#define BRW_NEW_BATCH			0x8000
+#define BRW_NEW_BATCH			0x10000
 /** brw->depth_region updated */
-#define BRW_NEW_DEPTH_BUFFER		0x10000
+#define BRW_NEW_DEPTH_BUFFER		0x20000
+#define BRW_NEW_NR_SURFACES		0x40000
 
 struct brw_state_flags {
    /** State update flags signalled by mesa internals */
@@ -158,7 +158,6 @@ struct brw_state_flags {
 struct brw_vertex_program {
    struct gl_vertex_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
@@ -166,7 +165,6 @@ struct brw_vertex_program {
 struct brw_fragment_program {
    struct gl_fragment_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index c7bac7b0c5..fbf473abf6 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -184,8 +184,8 @@ static void prepare_constant_buffer(struct brw_context *brw)
     * function will also be called whenever fp or vp changes.
     */
    brw->curbe.tracked_state.dirty.mesa = (_NEW_TRANSFORM|_NEW_PROJECTION);
-   brw->curbe.tracked_state.dirty.mesa |= vp->param_state;
-   brw->curbe.tracked_state.dirty.mesa |= fp->param_state;
+   brw->curbe.tracked_state.dirty.mesa |= vp->program.Base.Parameters->StateFlags;
+   brw->curbe.tracked_state.dirty.mesa |= fp->program.Base.Parameters->StateFlags;
 
    if (sz == 0) {
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index f893dd6742..c3a26fc82e 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -49,7 +49,7 @@
 
 #define FILE_DEBUG_FLAG DEBUG_BATCH
 
-static GLuint hw_prim[GL_POLYGON+1] = {
+static GLuint prim_to_hw_prim[GL_POLYGON+1] = {
    _3DPRIM_POINTLIST,
    _3DPRIM_LINELIST,
    _3DPRIM_LINELOOP,
@@ -103,12 +103,9 @@ static GLuint brw_set_prim(struct brw_context *brw, GLenum prim)
 	 brw->intel.reduced_primitive = reduced_prim[prim];
 	 brw->state.dirty.brw |= BRW_NEW_REDUCED_PRIMITIVE;
       }
-
-      brw_validate_state(brw);
-      brw_upload_state(brw);
    }
 
-   return hw_prim[prim];
+   return prim_to_hw_prim[prim];
 }
 
 
@@ -123,9 +120,9 @@ static GLuint trim(GLenum prim, GLuint length)
 }
 
 
-static void brw_emit_prim( struct brw_context *brw, 
-			   const struct _mesa_prim *prim )
-
+static void brw_emit_prim(struct brw_context *brw,
+			  const struct _mesa_prim *prim,
+			  uint32_t hw_prim)
 {
    struct brw_3d_primitive prim_packet;
 
@@ -136,7 +133,7 @@ static void brw_emit_prim( struct brw_context *brw,
    prim_packet.header.opcode = CMD_3D_PRIM;
    prim_packet.header.length = sizeof(prim_packet)/4 - 2;
    prim_packet.header.pad = 0;
-   prim_packet.header.topology = brw_set_prim(brw, prim->mode);
+   prim_packet.header.topology = hw_prim;
    prim_packet.header.indexed = prim->indexed;
 
    prim_packet.verts_per_instance = trim(prim->mode, prim->count);
@@ -258,11 +255,15 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    struct brw_context *brw = brw_context(ctx);
    GLboolean retval = GL_FALSE;
    GLboolean warn = GL_FALSE;
+   GLboolean first_time = GL_TRUE;
    GLuint i;
 
    if (ctx->NewState)
       _mesa_update_state( ctx );
 
+   if (check_fallbacks(brw, prim, nr_prims))
+      return GL_FALSE;
+
    brw_validate_textures( brw );
 
    /* Bind all inputs, derive varying and size information:
@@ -275,6 +276,7 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    brw->vb.min_index = min_index;
    brw->vb.max_index = max_index;
    brw->state.dirty.brw |= BRW_NEW_VERTICES;
+
    /* Have to validate state quite late.  Will rebuild tnl_program,
     * which depends on varying information.  
     * 
@@ -289,59 +291,58 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
       return GL_TRUE;
    }
 
-   /* Flush the batch if it's approaching full, so that we don't wrap while
-    * we've got validated state that needs to be in the same batch as the
-    * primitives.  This fraction is just a guess (minimal full state plus
-    * a primitive is around 512 bytes), and would be better if we had
-    * an upper bound of how much we might emit in a single
-    * brw_try_draw_prims().
-    */
-   intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
-				   LOOP_CLIPRECTS);
-   {
-      /* Set the first primitive early, ahead of validate_state:
+   for (i = 0; i < nr_prims; i++) {
+      uint32_t hw_prim;
+
+      /* Flush the batch if it's approaching full, so that we don't wrap while
+       * we've got validated state that needs to be in the same batch as the
+       * primitives.  This fraction is just a guess (minimal full state plus
+       * a primitive is around 512 bytes), and would be better if we had
+       * an upper bound of how much we might emit in a single
+       * brw_try_draw_prims().
        */
-      brw_set_prim(brw, prim[0].mode);
+      intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
+				      LOOP_CLIPRECTS);
 
-      brw_validate_state( brw );
+      hw_prim = brw_set_prim(brw, prim[i].mode);
 
-      /* Various fallback checks:
-       */
-      if (brw->intel.Fallback) 
-	 goto out;
+      if (first_time || (brw->state.dirty.brw & BRW_NEW_PRIMITIVE)) {
+	 first_time = GL_FALSE;
 
-      if (check_fallbacks( brw, prim, nr_prims ))
-	 goto out;
+	 /* Various fallback checks:  */
+	 if (brw->intel.Fallback)
+	    goto out;
 
-      /* Check that we can fit our state in with our existing batchbuffer, or
-       * flush otherwise.
-       */
-      if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
-					  brw->state.validated_bo_count)) {
-	 static GLboolean warned;
-	 intel_batchbuffer_flush(intel->batch);
-
-	 /* Validate the state after we flushed the batch (which would have
-	  * changed the set of dirty state).  If we still fail to
-	  * check_aperture, warn of what's happening, but attempt to continue
-	  * on since it may succeed anyway, and the user would probably rather
-	  * see a failure and a warning than a fallback.
-	  */
 	 brw_validate_state(brw);
-	 if (!warned &&
-	     dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+
+	 /* Check that we can fit our state in with our existing batchbuffer, or
+	  * flush otherwise.
+	  */
+	 if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
 					     brw->state.validated_bo_count)) {
-	    warn = GL_TRUE;
-	    warned = GL_TRUE;
+	    static GLboolean warned;
+	    intel_batchbuffer_flush(intel->batch);
+
+	    /* Validate the state after we flushed the batch (which would have
+	     * changed the set of dirty state).  If we still fail to
+	     * check_aperture, warn of what's happening, but attempt to continue
+	     * on since it may succeed anyway, and the user would probably rather
+	     * see a failure and a warning than a fallback.
+	     */
+	    brw_validate_state(brw);
+	    if (!warned &&
+		dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+						brw->state.validated_bo_count)) {
+	       warn = GL_TRUE;
+	       warned = GL_TRUE;
+	    }
 	 }
-      }
-
-      brw_upload_state(brw);
 
-      for (i = 0; i < nr_prims; i++) {
-	 brw_emit_prim(brw, &prim[i]);
+	 brw_upload_state(brw);
       }
 
+      brw_emit_prim(brw, &prim[i], hw_prim);
+
       retval = GL_TRUE;
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_program.c b/src/mesa/drivers/dri/i965/brw_program.c
index c38610b24e..a18dee85e8 100644
--- a/src/mesa/drivers/dri/i965/brw_program.c
+++ b/src/mesa/drivers/dri/i965/brw_program.c
@@ -117,7 +117,6 @@ static void brwProgramStringNotify( GLcontext *ctx,
       if (p == fp)
 	 brw->state.dirty.brw |= BRW_NEW_FRAGMENT_PROGRAM;
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
    }
    else if (target == GL_VERTEX_PROGRAM_ARB) {
       struct brw_context *brw = brw_context(ctx);
@@ -129,7 +128,6 @@ static void brwProgramStringNotify( GLcontext *ctx,
 	 _mesa_insert_mvp_code(ctx, &p->program);
       }
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
 
       /* Also tell tnl about it:
        */
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index 7a642bd2a8..5124535f2a 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -181,6 +181,118 @@ brw_clear_validated_bos(struct brw_context *brw)
    brw->state.validated_bo_count = 0;
 }
 
+struct dirty_bit_map {
+   uint32_t bit;
+   char *name;
+   uint32_t count;
+};
+
+#define DEFINE_BIT(name) {name, #name, 0}
+
+static struct dirty_bit_map mesa_bits[] = {
+   DEFINE_BIT(_NEW_MODELVIEW),
+   DEFINE_BIT(_NEW_PROJECTION),
+   DEFINE_BIT(_NEW_TEXTURE_MATRIX),
+   DEFINE_BIT(_NEW_COLOR_MATRIX),
+   DEFINE_BIT(_NEW_ACCUM),
+   DEFINE_BIT(_NEW_COLOR),
+   DEFINE_BIT(_NEW_DEPTH),
+   DEFINE_BIT(_NEW_EVAL),
+   DEFINE_BIT(_NEW_FOG),
+   DEFINE_BIT(_NEW_HINT),
+   DEFINE_BIT(_NEW_LIGHT),
+   DEFINE_BIT(_NEW_LINE),
+   DEFINE_BIT(_NEW_PIXEL),
+   DEFINE_BIT(_NEW_POINT),
+   DEFINE_BIT(_NEW_POLYGON),
+   DEFINE_BIT(_NEW_POLYGONSTIPPLE),
+   DEFINE_BIT(_NEW_SCISSOR),
+   DEFINE_BIT(_NEW_STENCIL),
+   DEFINE_BIT(_NEW_TEXTURE),
+   DEFINE_BIT(_NEW_TRANSFORM),
+   DEFINE_BIT(_NEW_VIEWPORT),
+   DEFINE_BIT(_NEW_PACKUNPACK),
+   DEFINE_BIT(_NEW_ARRAY),
+   DEFINE_BIT(_NEW_RENDERMODE),
+   DEFINE_BIT(_NEW_BUFFERS),
+   DEFINE_BIT(_NEW_MULTISAMPLE),
+   DEFINE_BIT(_NEW_TRACK_MATRIX),
+   DEFINE_BIT(_NEW_PROGRAM),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map brw_bits[] = {
+   DEFINE_BIT(BRW_NEW_URB_FENCE),
+   DEFINE_BIT(BRW_NEW_FRAGMENT_PROGRAM),
+   DEFINE_BIT(BRW_NEW_VERTEX_PROGRAM),
+   DEFINE_BIT(BRW_NEW_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_CURBE_OFFSETS),
+   DEFINE_BIT(BRW_NEW_REDUCED_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_CONTEXT),
+   DEFINE_BIT(BRW_NEW_WM_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_INPUT_VARYING),
+   DEFINE_BIT(BRW_NEW_PSP),
+   DEFINE_BIT(BRW_NEW_METAOPS),
+   DEFINE_BIT(BRW_NEW_FENCE),
+   DEFINE_BIT(BRW_NEW_INDICES),
+   DEFINE_BIT(BRW_NEW_VERTICES),
+   DEFINE_BIT(BRW_NEW_BATCH),
+   DEFINE_BIT(BRW_NEW_DEPTH_BUFFER),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map cache_bits[] = {
+   DEFINE_BIT(CACHE_NEW_CC_VP),
+   DEFINE_BIT(CACHE_NEW_CC_UNIT),
+   DEFINE_BIT(CACHE_NEW_WM_PROG),
+   DEFINE_BIT(CACHE_NEW_SAMPLER_DEFAULT_COLOR),
+   DEFINE_BIT(CACHE_NEW_SAMPLER),
+   DEFINE_BIT(CACHE_NEW_WM_UNIT),
+   DEFINE_BIT(CACHE_NEW_SF_PROG),
+   DEFINE_BIT(CACHE_NEW_SF_VP),
+   DEFINE_BIT(CACHE_NEW_SF_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_PROG),
+   DEFINE_BIT(CACHE_NEW_GS_UNIT),
+   DEFINE_BIT(CACHE_NEW_GS_PROG),
+   DEFINE_BIT(CACHE_NEW_CLIP_VP),
+   DEFINE_BIT(CACHE_NEW_CLIP_UNIT),
+   DEFINE_BIT(CACHE_NEW_CLIP_PROG),
+   DEFINE_BIT(CACHE_NEW_SURFACE),
+   DEFINE_BIT(CACHE_NEW_SURF_BIND),
+   {0, 0, 0}
+};
+
+
+static void
+brw_update_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      if (bit_map[i].bit & bits)
+	 bit_map[i].count++;
+   }
+}
+
+static void
+brw_print_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      fprintf(stderr, "0x%08x: %12d (%s)\n",
+	      bit_map[i].bit, bit_map[i].count, bit_map[i].name);
+   }
+}
+
 /***********************************************************************
  * Emit all state:
  */
@@ -246,6 +358,7 @@ void brw_upload_state(struct brw_context *brw)
 {
    struct brw_state_flags *state = &brw->state.dirty;
    int i;
+   static int dirty_count = 0;
 
    brw_clear_validated_bos(brw);
 
@@ -301,6 +414,18 @@ void brw_upload_state(struct brw_context *brw)
       }
    }
 
+   if (INTEL_DEBUG & DEBUG_STATE) {
+      brw_update_dirty_count(mesa_bits, state->mesa);
+      brw_update_dirty_count(brw_bits, state->brw);
+      brw_update_dirty_count(cache_bits, state->cache);
+      if (dirty_count++ % 1000 == 0) {
+	 brw_print_dirty_count(mesa_bits, state->mesa);
+	 brw_print_dirty_count(brw_bits, state->brw);
+	 brw_print_dirty_count(cache_bits, state->cache);
+	 fprintf(stderr, "\n");
+      }
+   }
+
    if (!brw->intel.Fallback)
       memset(state, 0, sizeof(*state));
 }
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 2d4c81274e..f7293ef423 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -171,8 +171,6 @@ static void brw_note_unlock( struct intel_context *intel )
    struct brw_context *brw = brw_context(&intel->ctx);
 
    brw_state_cache_check_size(brw);
-
-   brw_context(&intel->ctx)->state.dirty.brw |= BRW_NEW_LOCK;
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 361312c2ca..bad76793af 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -230,12 +230,6 @@ static void brw_wm_populate_key( struct brw_context *brw,
 	 lookup |= IZ_STENCIL_WRITE_ENABLE_BIT;
    }
 
-   /* XXX: when should this be disabled?
-    */
-   if (1)
-      lookup |= IZ_EARLY_DEPTH_TEST_BIT;
-
-   
    line_aa = AA_NEVER;
 
    /* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index 896390c17b..ded079695e 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -49,8 +49,7 @@
 #define IZ_DEPTH_TEST_ENABLE_BIT    0x8
 #define IZ_STENCIL_WRITE_ENABLE_BIT 0x10
 #define IZ_STENCIL_TEST_ENABLE_BIT  0x20
-#define IZ_EARLY_DEPTH_TEST_BIT     0x40
-#define IZ_BIT_MAX                  0x80
+#define IZ_BIT_MAX                  0x40
 
 #define AA_NEVER     0
 #define AA_SOMETIMES 1
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index d3e926c328..7f7b957cbe 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -426,10 +426,6 @@ static struct prog_src_register search_or_add_param5(struct brw_wm_compile *c,
 
    idx = _mesa_add_state_reference( paramList, tokens );
 
-   /* Recalculate state dependency: 
-    */
-   c->fp->param_state = paramList->StateFlags;
-
    return src_reg(PROGRAM_STATE_VAR, idx);
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_iz.c b/src/mesa/drivers/dri/i965/brw_wm_iz.c
index 0bb5d5ba83..bd60ac9b31 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_iz.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_iz.c
@@ -52,70 +52,6 @@ const struct {
 {
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index fd461618bc..5302405eda 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -88,7 +88,7 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
    /* BRW_NEW_CURBE_OFFSETS */
    key->curbe_offset = brw->curbe.wm_start;
 
-   /* CACHE_NEW_SURFACE */
+   /* BRW_NEW_NR_SURFACEs */
    key->nr_surfaces = brw->wm.nr_surfaces;
 
    /* CACHE_NEW_SAMPLER */
@@ -281,10 +281,9 @@ const struct brw_tracked_state brw_wm_unit = {
 
       .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
 	      BRW_NEW_CURBE_OFFSETS |
-	      BRW_NEW_LOCK),
+	      BRW_NEW_NR_SURFACES),
 
-      .cache = (CACHE_NEW_SURFACE | 
-		CACHE_NEW_WM_PROG | 
+      .cache = (CACHE_NEW_WM_PROG |
 		CACHE_NEW_SAMPLER)
    },
    .prepare = upload_wm_unit,
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index 3790b50c97..63e14cc390 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -438,6 +438,7 @@ static void prepare_wm_surfaces(struct brw_context *brw )
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = &brw->intel;
    GLuint i;
+   int old_nr_surfaces;
 
    if (brw->state.nr_draw_regions  > 1) {
       for (i = 0; i < brw->state.nr_draw_regions; i++) {
@@ -448,6 +449,7 @@ static void prepare_wm_surfaces(struct brw_context *brw )
       brw_update_region_surface(brw, brw->state.draw_regions[0], 0, GL_TRUE);
    }
 
+   old_nr_surfaces = brw->wm.nr_surfaces;
    brw->wm.nr_surfaces = MAX_DRAW_BUFFERS;
 
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
@@ -473,6 +475,9 @@ static void prepare_wm_surfaces(struct brw_context *brw )
 
    dri_bo_unreference(brw->wm.bind_bo);
    brw->wm.bind_bo = brw_wm_get_binding_table(brw);
+
+   if (brw->wm.nr_surfaces != old_nr_surfaces)
+      brw->state.dirty.brw |= BRW_NEW_NR_SURFACES;
 }
 
 
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index 60028f090a..ee43ed7e83 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -184,6 +184,7 @@ struct intel_context
       GLuint id;
       uint32_t primitive;	/**< Current hardware primitive type */
       void (*flush) (struct intel_context *);
+      GLubyte *start_ptr; /**< for i8xx */
       dri_bo *vb_bo;
       uint8_t *vb;
       unsigned int start_offset; /**< Byte offset of primitive sequence */
diff --git a/src/mesa/drivers/dri/intel/intel_reg.h b/src/mesa/drivers/dri/intel/intel_reg.h
index 68d8a05b83..57ac8f0cc1 100644
--- a/src/mesa/drivers/dri/intel/intel_reg.h
+++ b/src/mesa/drivers/dri/intel/intel_reg.h
@@ -71,14 +71,14 @@
 /** @{
  * 915 definitions
  */
-#define S0_VB_OFFSET_MASK		0xffffffc
+#define S0_VB_OFFSET_MASK		0xffffffc0
 #define S0_AUTO_CACHE_INV_DISABLE	(1<<0)
 /** @} */
 
 /** @{
  * 830 definitions
  */
-#define S0_VB_OFFSET_MASK_830		0xffffff8
+#define S0_VB_OFFSET_MASK_830		0xffffff80
 #define S0_VB_PITCH_SHIFT_830		1
 #define S0_VB_ENABLE_830		(1<<0)
 /** @} */
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index cf09fad745..61b55b97b5 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -461,6 +461,7 @@ intelCreateContext(const __GLcontextModes * mesaVis,
 				  sharedContextPrivate);
       }
    } else {
+      intelScreen->no_vbo = GL_TRUE;
       return i830CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
    }
 #else
diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index fc913da5bf..91f0d6d1ae 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -77,6 +77,7 @@ typedef struct
 
    GLboolean no_hw;
 
+   GLboolean no_vbo;
    int ttm;
    dri_bufmgr *bufmgr;
 
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 36446efde7..dd932aebc9 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -98,7 +98,9 @@ do_copy_texsubimage(struct intel_context *intel,
       get_teximage_source(intel, internalFormat);
 
    if (!intelImage->mt || !src) {
-      DBG("%s fail %p %p\n", __FUNCTION__, intelImage->mt, src);
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "%s fail %p %p\n",
+		 __FUNCTION__, intelImage->mt, src);
       return GL_FALSE;
    }