Merge commit 'origin/master' into gallium-map-range

176 files changed, 4394 insertions, 3512 deletions

diff --git a/src/egl/main/Makefile b/src/egl/main/Makefile
index 89b4a200bc..5ee246dc3f 100644
--- a/src/egl/main/Makefile
+++ b/src/egl/main/Makefile
@@ -46,8 +46,6 @@ OBJECTS = $(SOURCES:.c=.o)
 # Undefined for now
 LOCAL_CFLAGS = -D_EGL_PLATFORM_X=1
 
-LIBS = -lX11
-
 
 .c.o:
 	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(LOCAL_CFLAGS) $< -o $@
@@ -63,7 +61,8 @@ library: $(TOP)/$(LIB_DIR)/libEGL.so
 $(TOP)/$(LIB_DIR)/libEGL.so: $(OBJECTS)
 	$(MKLIB) -o EGL -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major 1 -minor 0 \
-		-install $(TOP)/$(LIB_DIR) -ldl $(OBJECTS) $(LIBS)
+		-install $(TOP)/$(LIB_DIR) \
+		$(EGL_LIB_DEPS) $(OBJECTS)
 
 install:
 	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
index 556f26c0f0..533c0a15f3 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
@@ -36,7 +36,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #include <unistd.h>
 #include <sched.h>
 #endif
@@ -559,7 +559,7 @@ fenced_buffer_list_destroy(struct fenced_buffer_list *fenced_list)
    /* Wait on outstanding fences */
    while (fenced_list->numDelayed) {
       pipe_mutex_unlock(fenced_list->mutex);
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
       sched_yield();
 #endif
       _fenced_buffer_list_check_free(fenced_list, 1);
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h b/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h
index fec8db91c7..74077f8277 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr.h
@@ -200,10 +200,11 @@ pb_ondemand_manager_create(struct pb_manager *provider);
 /** 
  * Debug buffer manager to detect buffer under- and overflows.
  *
- * Band size should be a multiple of the largest alignment
+ * Under/overflow sizes should be a multiple of the largest alignment
  */
 struct pb_manager *
-pb_debug_manager_create(struct pb_manager *provider, size_t band_size); 
+pb_debug_manager_create(struct pb_manager *provider,
+                        size_t underflow_size, size_t overflow_size); 
 
 
 #ifdef __cplusplus
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
index 070bf3f517..21079b8bfd 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c
@@ -78,7 +78,8 @@ struct pb_debug_manager
 
    struct pb_manager *provider;
 
-   size_t band_size;
+   size_t underflow_size;
+   size_t overflow_size;
 };
 
 
@@ -301,7 +302,7 @@ pb_debug_manager_create_buffer(struct pb_manager *_mgr,
    if(!buf)
       return NULL;
    
-   real_size = size + 2*mgr->band_size;
+   real_size = mgr->underflow_size + size + mgr->overflow_size;
    real_desc = *desc;
    real_desc.usage |= PIPE_BUFFER_USAGE_CPU_WRITE;
    real_desc.usage |= PIPE_BUFFER_USAGE_CPU_READ;
@@ -327,7 +328,7 @@ pb_debug_manager_create_buffer(struct pb_manager *_mgr,
    buf->base.vtbl = &pb_debug_buffer_vtbl;
    buf->mgr = mgr;
 
-   buf->underflow_size = mgr->band_size;
+   buf->underflow_size = mgr->underflow_size;
    buf->overflow_size = buf->buffer->base.size - buf->underflow_size - size;
    
    pb_debug_buffer_fill(buf);
@@ -356,7 +357,8 @@ pb_debug_manager_destroy(struct pb_manager *_mgr)
 
 
 struct pb_manager *
-pb_debug_manager_create(struct pb_manager *provider, size_t band_size) 
+pb_debug_manager_create(struct pb_manager *provider, 
+                        size_t underflow_size, size_t overflow_size) 
 {
    struct pb_debug_manager *mgr;
 
@@ -371,7 +373,8 @@ pb_debug_manager_create(struct pb_manager *provider, size_t band_size)
    mgr->base.create_buffer = pb_debug_manager_create_buffer;
    mgr->base.flush = pb_debug_manager_flush;
    mgr->provider = provider;
-   mgr->band_size = band_size;
+   mgr->underflow_size = underflow_size;
+   mgr->overflow_size = overflow_size;
       
    return &mgr->base;
 }
@@ -381,7 +384,8 @@ pb_debug_manager_create(struct pb_manager *provider, size_t band_size)
 
 
 struct pb_manager *
-pb_debug_manager_create(struct pb_manager *provider, size_t band_size) 
+pb_debug_manager_create(struct pb_manager *provider, 
+                        size_t underflow_size, size_t overflow_size) 
 {
    return provider;
 }
diff --git a/src/gallium/auxiliary/rtasm/rtasm_execmem.c b/src/gallium/auxiliary/rtasm/rtasm_execmem.c
index 5acc5bcb7b..1f0923b683 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_execmem.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_execmem.c
@@ -37,8 +37,12 @@
 
 #include "rtasm_execmem.h"
 
+#if defined(PIPE_OS_BSD)
+#define MAP_ANONYMOUS MAP_ANON
+#endif
 
-#if defined(PIPE_OS_LINUX)
+
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 
 
 /*
@@ -114,7 +118,7 @@ rtasm_exec_free(void *addr)
 }
 
 
-#else /* PIPE_OS_LINUX */
+#else /* PIPE_OS_LINUX || PIPE_OS_BSD */
 
 /*
  * Just use regular memory.
@@ -134,4 +138,4 @@ rtasm_exec_free(void *addr)
 }
 
 
-#endif /* PIPE_OS_LINUX */
+#endif /* PIPE_OS_LINUX || PIPE_OS_BSD */
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index d57cb9139f..a784b7cc3c 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -245,8 +245,12 @@ iter_declaration(
       }
    }
 
-   TXT( ", " );
-   ENM( decl->Declaration.Interpolate, interpolate_names );
+   if (iter->processor.Processor == TGSI_PROCESSOR_FRAGMENT &&
+       decl->Declaration.File == TGSI_FILE_INPUT)
+   {
+      TXT( ", " );
+      ENM( decl->Declaration.Interpolate, interpolate_names );
+   }
 
    if (decl->Declaration.Centroid) {
       TXT( ", CENTROID" );
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index 76e773da91..6f1f5c2b4b 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -30,18 +30,19 @@
 #include "tgsi_info.h"
 #include "tgsi_iterate.h"
 
-#define MAX_REGISTERS 256
-
 typedef uint reg_flag;
 
 #define BITS_IN_REG_FLAG (sizeof( reg_flag ) * 8)
 
+#define MAX_REGISTERS 256
+#define MAX_REG_FLAGS ((MAX_REGISTERS + BITS_IN_REG_FLAG - 1) / BITS_IN_REG_FLAG)
+
 struct sanity_check_ctx
 {
    struct tgsi_iterate_context iter;
 
-   reg_flag regs_decl[TGSI_FILE_COUNT][MAX_REGISTERS / BITS_IN_REG_FLAG];
-   reg_flag regs_used[TGSI_FILE_COUNT][MAX_REGISTERS / BITS_IN_REG_FLAG];
+   reg_flag regs_decl[TGSI_FILE_COUNT][MAX_REG_FLAGS];
+   reg_flag regs_used[TGSI_FILE_COUNT][MAX_REG_FLAGS];
    boolean regs_ind_used[TGSI_FILE_COUNT];
    uint num_imms;
    uint num_instructions;
@@ -89,7 +90,7 @@ check_file_name(
    uint file )
 {
    if (file <= TGSI_FILE_NULL || file >= TGSI_FILE_COUNT) {
-      report_error( ctx, "Invalid register file name" );
+      report_error( ctx, "(%u): Invalid register file name", file );
       return FALSE;
    }
    return TRUE;
@@ -113,7 +114,7 @@ is_any_register_declared(
 {
    uint i;
 
-   for (i = 0; i < MAX_REGISTERS / BITS_IN_REG_FLAG; i++)
+   for (i = 0; i < MAX_REG_FLAGS; i++)
       if (ctx->regs_decl[file][i])
          return TRUE;
    return FALSE;
@@ -162,9 +163,8 @@ check_register_usage(
       ctx->regs_ind_used[file] = TRUE;
    }
    else {
-      if (index < 0 || index > MAX_REGISTERS) {
-         report_error( ctx, "%s[%i]: Invalid index %s",
-                       file_names[file], index, name );
+      if (index < 0 || index >= MAX_REGISTERS) {
+         report_error( ctx, "%s[%d]: Invalid %s index", file_names[file], index, name );
          return FALSE;
       }
 
@@ -193,15 +193,15 @@ iter_instruction(
 
    info = tgsi_get_opcode_info( inst->Instruction.Opcode );
    if (info == NULL) {
-      report_error( ctx, "Invalid instruction opcode" );
+      report_error( ctx, "(%u): Invalid instruction opcode", inst->Instruction.Opcode );
       return TRUE;
    }
 
    if (info->num_dst != inst->Instruction.NumDstRegs) {
-      report_error( ctx, "Invalid number of destination operands" );
+      report_error( ctx, "Invalid number of destination operands, should be %u", info->num_dst );
    }
    if (info->num_src != inst->Instruction.NumSrcRegs) {
-      report_error( ctx, "Invalid number of source operands" );
+      report_error( ctx, "Invalid number of source operands, should be %u", info->num_src );
    }
 
    /* Check destination and source registers' validity.
@@ -266,7 +266,7 @@ iter_declaration(
       return TRUE;
    for (i = decl->DeclarationRange.First; i <= decl->DeclarationRange.Last; i++) {
       if (is_register_declared( ctx, file, i ))
-         report_error( ctx, "The same register declared twice" );
+         report_error( ctx, "%s[%u]: The same register declared more than once", file_names[file], i );
       ctx->regs_decl[file][i / BITS_IN_REG_FLAG] |= (1 << (i % BITS_IN_REG_FLAG));
    }
 
@@ -295,7 +295,7 @@ iter_immediate(
    /* Check data type validity.
     */
    if (imm->Immediate.DataType != TGSI_IMM_FLOAT32) {
-      report_error( ctx, "Invalid immediate data type" );
+      report_error( ctx, "(%u): Invalid immediate data type", imm->Immediate.DataType );
       return TRUE;
    }
 
@@ -322,7 +322,7 @@ epilog(
 
       for (i = 0; i < MAX_REGISTERS; i++) {
          if (is_register_declared( ctx, file, i ) && !is_register_used( ctx, file, i ) && !ctx->regs_ind_used[file]) {
-            report_warning( ctx, "Register never used" );
+            report_warning( ctx, "%s[%u]: Register never used", file_names[file], i );
          }
       }
    }
diff --git a/src/gallium/auxiliary/util/u_debug_memory.c b/src/gallium/auxiliary/util/u_debug_memory.c
index 758541c282..7623cb9398 100644
--- a/src/gallium/auxiliary/util/u_debug_memory.c
+++ b/src/gallium/auxiliary/util/u_debug_memory.c
@@ -73,7 +73,9 @@ struct debug_memory_header
    const char *file;
    unsigned line;
    const char *function;
+#if DEBUG_MEMORY_STACK
    struct debug_stack_frame backtrace[DEBUG_MEMORY_STACK];
+#endif
    size_t size;
    
    unsigned magic;
@@ -140,7 +142,9 @@ debug_malloc(const char *file, unsigned line, const char *function,
    hdr->size = size;
    hdr->magic = DEBUG_MEMORY_MAGIC;
 
+#if DEBUG_MEMORY_STACK
    debug_backtrace_capture(hdr->backtrace, 0, DEBUG_MEMORY_STACK);
+#endif
 
    ftr = footer_from_header(hdr);
    ftr->magic = DEBUG_MEMORY_MAGIC;
@@ -296,7 +300,9 @@ debug_memory_end(unsigned long start_no)
 	 debug_printf("%s:%u:%s: %u bytes at %p not freed\n",
 		      hdr->file, hdr->line, hdr->function,
 		      hdr->size, ptr);
+#if DEBUG_MEMORY_STACK
 	 debug_backtrace_dump(hdr->backtrace, DEBUG_MEMORY_STACK);
+#endif
 	 total_size += hdr->size;
       }
 
diff --git a/src/gallium/auxiliary/util/u_draw_quad.c b/src/gallium/auxiliary/util/u_draw_quad.c
index f282f3d289..f0bcd75899 100644
--- a/src/gallium/auxiliary/util/u_draw_quad.c
+++ b/src/gallium/auxiliary/util/u_draw_quad.c
@@ -51,9 +51,11 @@ util_draw_vertex_buffer(struct pipe_context *pipe,
    assert(num_attribs <= PIPE_MAX_ATTRIBS);
 
    /* tell pipe about the vertex buffer */
+   memset(&vbuffer, 0, sizeof(vbuffer));
    vbuffer.buffer = vbuf;
    vbuffer.stride = num_attribs * 4 * sizeof(float);  /* vertex size */
    vbuffer.buffer_offset = offset;
+   vbuffer.max_index = num_verts - 1;
    pipe->set_vertex_buffers(pipe, 1, &vbuffer);
 
    /* tell pipe about the vertex attributes */
diff --git a/src/gallium/auxiliary/util/u_stream_stdc.c b/src/gallium/auxiliary/util/u_stream_stdc.c
index ca80bef0f3..0ead45a749 100644
--- a/src/gallium/auxiliary/util/u_stream_stdc.c
+++ b/src/gallium/auxiliary/util/u_stream_stdc.c
@@ -32,7 +32,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_SUBSYSTEM_WINDOWS_USER)
 
 #include <stdio.h>
 
diff --git a/src/gallium/auxiliary/util/u_time.c b/src/gallium/auxiliary/util/u_time.c
index dde2c74fa8..357d9360c9 100644
--- a/src/gallium/auxiliary/util/u_time.c
+++ b/src/gallium/auxiliary/util/u_time.c
@@ -35,7 +35,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #include <sys/time.h>
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
 #include <windows.h>
@@ -77,7 +77,7 @@ util_time_get_frequency(void)
 void 
 util_time_get(struct util_time *t)
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    gettimeofday(&t->tv, NULL);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
    LONGLONG temp;
@@ -102,7 +102,7 @@ util_time_add(const struct util_time *t1,
               int64_t usecs,
               struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    t2->tv.tv_sec = t1->tv.tv_sec + usecs / 1000000;
    t2->tv.tv_usec = t1->tv.tv_usec + usecs % 1000000;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
@@ -124,7 +124,7 @@ int64_t
 util_time_diff(const struct util_time *t1, 
                const struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    return (t2->tv.tv_usec - t1->tv.tv_usec) + 
           (t2->tv.tv_sec - t1->tv.tv_sec)*1000000;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
@@ -144,7 +144,7 @@ util_time_micros( void )
    
    util_time_get(&t1);
    
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    return t1.tv.tv_usec + t1.tv.tv_sec*1000000LL;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
    util_time_get_frequency();
@@ -166,7 +166,7 @@ static INLINE int
 util_time_compare(const struct util_time *t1, 
                   const struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    if (t1->tv.tv_sec < t2->tv.tv_sec)
       return -1;
    else if(t1->tv.tv_sec > t2->tv.tv_sec)
diff --git a/src/gallium/auxiliary/util/u_time.h b/src/gallium/auxiliary/util/u_time.h
index 35d97d16c7..4346ce1fa4 100644
--- a/src/gallium/auxiliary/util/u_time.h
+++ b/src/gallium/auxiliary/util/u_time.h
@@ -38,7 +38,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #include <time.h> /* timeval */
 #include <unistd.h> /* usleep */
 #endif
@@ -58,7 +58,7 @@ extern "C" {
  */
 struct util_time 
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    struct timeval tv;
 #else
    int64_t counter;
@@ -89,7 +89,7 @@ util_time_timeout(const struct util_time *start,
                   const struct util_time *end,
                   const struct util_time *curr);
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #define util_time_sleep usleep
 #else
 void
diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h
index ca03dc1511..5c3188e7f9 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.h
+++ b/src/gallium/drivers/cell/ppu/cell_context.h
@@ -127,15 +127,9 @@ struct cell_context
    struct pipe_vertex_element vertex_element[PIPE_MAX_ATTRIBS];
    uint num_vertex_elements;
 
-   struct pipe_transfer *cbuf_transfer[PIPE_MAX_COLOR_BUFS];
-   struct pipe_transfer *zsbuf_transfer;
-
    ubyte *cbuf_map[PIPE_MAX_COLOR_BUFS];
    ubyte *zsbuf_map;
 
-   struct pipe_surface *tex_surf;
-   uint *tex_map;
-
    uint dirty;
    uint dirty_textures;  /* bitmask of texture units */
    uint dirty_samplers;  /* bitmask of sampler units */
diff --git a/src/gallium/drivers/cell/ppu/cell_fence.c b/src/gallium/drivers/cell/ppu/cell_fence.c
index 867b5dcaa0..32dbf5706c 100644
--- a/src/gallium/drivers/cell/ppu/cell_fence.c
+++ b/src/gallium/drivers/cell/ppu/cell_fence.c
@@ -153,16 +153,12 @@ cell_add_fenced_textures(struct cell_context *cell)
    for (i = 0; i < cell->num_textures; i++) {
       struct cell_texture *ct = cell->texture[i];
       if (ct) {
-         uint level;
-         for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
-            if (ct->tiled_buffer[level]) {
 #if 0
-               printf("Adding texture %p buffer %p to list\n",
-                      ct, ct->tiled_buffer[level]);
+         printf("Adding texture %p buffer %p to list\n",
+                ct, ct->tiled_buffer[level]);
 #endif
-               cell_add_buffer_to_list(cell, list, ct->tiled_buffer[level]);
-            }
-         }
+         if (ct->buffer)
+            cell_add_buffer_to_list(cell, list, ct->buffer);
       }
    }
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_pipe_state.c b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
index facd9551fe..ccd0fef6e8 100644
--- a/src/gallium/drivers/cell/ppu/cell_pipe_state.c
+++ b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
@@ -285,28 +285,22 @@ cell_map_surfaces(struct cell_context *cell)
    for (i = 0; i < 1; i++) {
       struct pipe_surface *ps = cell->framebuffer.cbufs[i];
       if (ps) {
-         cell->cbuf_transfer[i] =
-            screen->get_tex_transfer(screen, ps->texture, ps->face,
-                                     ps->level, ps->zslice,
-                                     PIPE_TRANSFER_READ_WRITE,
-                                     0, 0, ps->width, ps->height);
-
-         cell->cbuf_map[i] =
-            screen->transfer_map(screen, cell->cbuf_transfer[i]);
+         struct cell_texture *ct = cell_texture(ps->texture);
+         cell->cbuf_map[i] = screen->buffer_map(screen,
+                                                ct->buffer,
+                                                (PIPE_BUFFER_USAGE_GPU_READ |
+                                                 PIPE_BUFFER_USAGE_GPU_WRITE));
       }
    }
 
    {
       struct pipe_surface *ps = cell->framebuffer.zsbuf;
       if (ps) {
-         cell->zsbuf_transfer =
-            screen->get_tex_transfer(screen, ps->texture, ps->face,
-                                     ps->level, ps->zslice,
-                                     PIPE_TRANSFER_READ_WRITE,
-                                     0, 0, ps->width, ps->height);
-
-         cell->zsbuf_map =
-            screen->transfer_map(screen, cell->zsbuf_transfer);
+         struct cell_texture *ct = cell_texture(ps->texture);
+         cell->zsbuf_map = screen->buffer_map(screen,
+                                              ct->buffer,
+                                              (PIPE_BUFFER_USAGE_GPU_READ |
+                                               PIPE_BUFFER_USAGE_GPU_WRITE));
       }
    }
 }
@@ -322,24 +316,24 @@ cell_unmap_surfaces(struct cell_context *cell)
    uint i;
 
    for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) {
-      if (cell->cbuf_transfer[i] && cell->cbuf_map[i]) {
-         /* unmap color buffer/surface [i] */
-         screen->transfer_unmap(screen, cell->cbuf_transfer[i]);
-         cell->cbuf_map[i] = NULL;
+      struct pipe_surface *ps = cell->framebuffer.cbufs[i];
+      if (ps && cell->cbuf_map[i]) {
+         struct cell_texture *ct = cell_texture(ps->texture);
+         assert(ps->texture);
+         assert(ct->buffer);
 
-         /* get rid of transfer object [i] */
-         screen->tex_transfer_release(screen, &cell->cbuf_transfer[i]);
-         assert(cell->cbuf_transfer[i] == NULL);
+         screen->buffer_unmap(screen, ct->buffer);
+         cell->cbuf_map[i] = NULL;
       }
    }
 
-   if (cell->zsbuf_transfer && cell->zsbuf_map) {
-      screen->transfer_unmap(screen, cell->zsbuf_transfer);
-      cell->zsbuf_map = NULL;
-
-      /* get rid of transfer object */
-      screen->tex_transfer_release(screen, &cell->zsbuf_transfer);
-      assert(cell->zsbuf_transfer == NULL);
+   {
+      struct pipe_surface *ps = cell->framebuffer.zsbuf;
+      if (ps && cell->zsbuf_map) {
+         struct cell_texture *ct = cell_texture(ps->texture);
+         screen->buffer_unmap(screen, ct->buffer);
+         cell->zsbuf_map = NULL;
+      }
    }
 }
 
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index ff529fe22c..9479c0898f 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_emit.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c
@@ -287,19 +287,23 @@ cell_emit_state(struct cell_context *cell)
       for (i = 0;i < CELL_MAX_SAMPLERS; i++) {
          if (cell->dirty_textures & (1 << i)) {
             STATIC_ASSERT(sizeof(struct cell_command_texture) % 16 == 0);
-            struct cell_command_texture *texture
-               =  (struct cell_command_texture *)cell_batch_alloc16(cell, sizeof(*texture));
+            struct cell_command_texture *texture =
+               (struct cell_command_texture *)
+               cell_batch_alloc16(cell, sizeof(*texture));
+
             texture->opcode[0] = CELL_CMD_STATE_TEXTURE;
             texture->unit = i;
             if (cell->texture[i]) {
+               struct cell_texture *ct = cell->texture[i];
                uint level;
                for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
-                  texture->start[level] = cell->texture[i]->tiled_mapped[level];
-                  texture->width[level] = cell->texture[i]->base.width[level];
-                  texture->height[level] = cell->texture[i]->base.height[level];
-                  texture->depth[level] = cell->texture[i]->base.depth[level];
+                  texture->start[level] = (ct->mapped +
+                                           ct->level_offset[level]);
+                  texture->width[level] = ct->base.width[level];
+                  texture->height[level] = ct->base.height[level];
+                  texture->depth[level] = ct->base.depth[level];
                }
-               texture->target = cell->texture[i]->base.target;
+               texture->target = ct->base.target;
             }
             else {
                uint level;
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index fa52e2cbea..74724393f7 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -62,7 +62,7 @@ cell_texture_layout(struct cell_texture *ct)
 
    ct->buffer_size = 0;
 
-   for ( level = 0 ; level <= pt->last_level ; level++ ) {
+   for (level = 0; level <= pt->last_level; level++) {
       unsigned size;
       unsigned w_tile, h_tile;
 
@@ -90,7 +90,7 @@ cell_texture_layout(struct cell_texture *ct)
 
       ct->buffer_size += size;
 
-      width  = minify(width);
+      width = minify(width);
       height = minify(height);
       depth = minify(depth);
    }
@@ -131,33 +131,19 @@ cell_texture_release(struct pipe_screen *screen,
    if (!*pt)
       return;
 
-   /*
-   DBG("%s %p refcount will be %d\n",
-       __FUNCTION__, (void *) *pt, (*pt)->refcount - 1);
-   */
    if (--(*pt)->refcount <= 0) {
       /* Delete this texture now.
        * But note that the underlying pipe_buffer may linger...
        */
       struct cell_texture *ct = cell_texture(*pt);
-      uint i;
 
-      /*
-      DBG("%s deleting %p\n", __FUNCTION__, (void *) ct);
-      */
+      if (ct->mapped) {
+         pipe_buffer_unmap(screen, ct->buffer);
+         ct->mapped = NULL;
+      }
 
       pipe_buffer_reference(screen, &ct->buffer, NULL);
 
-      for (i = 0; i < CELL_MAX_TEXTURE_LEVELS; i++) {
-         /* Unreference the tiled image buffer.
-          * It may not actually be deleted until a fence is hit.
-          */
-         if (ct->tiled_buffer[i]) {
-            ct->tiled_mapped[i] = NULL;
-            pipe_buffer_reference(screen, &ct->tiled_buffer[i], NULL);
-         }
-      }
-
       FREE(ct);
    }
    *pt = NULL;
@@ -294,107 +280,6 @@ untwiddle_image_uint(uint w, uint h, uint tile_size, uint *dst,
 }
 
 
-/**
- * Convert linear texture image data to tiled format for SPU usage.
- */
-static void
-cell_twiddle_texture(struct pipe_screen *screen,
-                     struct pipe_surface *surface)
-{
-#if 0 // XXX fix me
-   struct cell_texture *ct = cell_texture(surface->texture);
-   const uint level = surface->level;
-   const uint texWidth = ct->base.width[level];
-   const uint texHeight = ct->base.height[level];
-   const uint bufWidth = align(texWidth, TILE_SIZE);
-   const uint bufHeight = align(texHeight, TILE_SIZE);
-   const void *map = screen->surface_map(screen, surface, PIPE_BUFFER_USAGE_CPU_READ);
-   const uint *src = (const uint *) map;
-
-   switch (ct->base.format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
-      {
-         int numFaces = ct->base.target == PIPE_TEXTURE_CUBE ? 6 : 1;
-         int offset = bufWidth * bufHeight * 4 * surface->face;
-         uint *dst;
-
-         if (!ct->tiled_buffer[level]) {
-            /* allocate buffer for tiled data now */
-            struct pipe_winsys *ws = screen->winsys;
-            uint bytes = bufWidth * bufHeight * 4 * numFaces;
-            ct->tiled_buffer[level] =
-               ws->buffer_create(ws, 16, PIPE_BUFFER_USAGE_PIXEL, bytes);
-            /* and map it */
-            ct->tiled_mapped[level] =
-               ws->buffer_map(ws, ct->tiled_buffer[level],
-                              PIPE_BUFFER_USAGE_GPU_READ);
-         }
-         dst = (uint *) ((ubyte *) ct->tiled_mapped[level] + offset);
-
-         twiddle_image_uint(texWidth, texHeight, TILE_SIZE, dst,
-                            surface->stride, src);
-      }
-      break;
-   default:
-      printf("Cell: twiddle unsupported texture format %s\n",
-             pf_name(ct->base.format));
-   }
-
-   screen->surface_unmap(screen, surface);
-#endif
-}
-
-
-/**
- * Convert SPU tiled texture image data to linear format for app usage.
- */
-static void
-cell_untwiddle_texture(struct pipe_screen *screen,
-                     struct pipe_surface *surface)
-{
-#if 0 // XXX fix me
-   struct cell_texture *ct = cell_texture(surface->texture);
-   const uint level = surface->level;
-   const uint texWidth = ct->base.width[level];
-   const uint texHeight = ct->base.height[level];
-   const void *map = screen->surface_map(screen, surface, PIPE_BUFFER_USAGE_CPU_READ);
-   const uint *src = (const uint *) ((const ubyte *) map + surface->offset);
-
-   switch (ct->base.format) {
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
-      {
-         int numFaces = ct->base.target == PIPE_TEXTURE_CUBE ? 6 : 1;
-         int offset = surface->stride * texHeight * 4 * surface->face;
-         uint *dst;
-
-         if (!ct->untiled_data[level]) {
-            ct->untiled_data[level] =
-               align_malloc(surface->stride * texHeight * 4 * numFaces, 16);
-         }
-
-         dst = (uint *) ((ubyte *) ct->untiled_data[level] + offset);
-
-         untwiddle_image_uint(texWidth, texHeight, TILE_SIZE, dst,
-                              surface->stride, src);
-      }
-      break;
-   default:
-      {
-         ct->untiled_data[level] = NULL;
-         printf("Cell: untwiddle unsupported texture format %s\n",
-                pf_name(ct->base.format));
-      }
-   }
-
-   screen->surface_unmap(screen, surface);
-#endif
-}
-
-
 static struct pipe_surface *
 cell_get_tex_surface(struct pipe_screen *screen,
                      struct pipe_texture *pt,
@@ -409,38 +294,25 @@ cell_get_tex_surface(struct pipe_screen *screen,
       ps->refcount = 1;
       pipe_texture_reference(&ps->texture, pt);
       ps->format = pt->format;
-      //ps->block = pt->block;
       ps->width = pt->width[level];
       ps->height = pt->height[level];
-      //ps->nblocksx = pt->nblocksx[level];
-      //ps->nblocksy = pt->nblocksy[level];
-      //ps->stride = ct->stride[level];
       ps->offset = ct->level_offset[level];
-      ps->usage = usage;
-
       /* XXX may need to override usage flags (see sp_texture.c) */
-
-      pipe_texture_reference(&ps->texture, pt); 
+      ps->usage = usage;
       ps->face = face;
       ps->level = level;
       ps->zslice = zslice;
 
-      if (pt->target == PIPE_TEXTURE_CUBE || pt->target == PIPE_TEXTURE_3D) {
-#if 0 // XXX fix me
-         ps->offset += ((pt->target == PIPE_TEXTURE_CUBE) ? face : zslice) *
-            ps->nblocksy *
-            ps->stride;
-#endif
+      if (pt->target == PIPE_TEXTURE_CUBE) {
+         ps->offset += face * pt->nblocksy[level] * ct->stride[level];
+      }
+      else if (pt->target == PIPE_TEXTURE_3D) {
+         ps->offset += zslice * pt->nblocksy[level] * ct->stride[level];
       }
       else {
          assert(face == 0);
          assert(zslice == 0);
       }
-
-      if (ps->usage & PIPE_BUFFER_USAGE_CPU_READ) {
-         /* convert from tiled to linear layout */
-         cell_untwiddle_texture(screen, ps);
-      }
    }
    return ps;
 }
@@ -450,24 +322,8 @@ static void
 cell_tex_surface_release(struct pipe_screen *screen, 
                          struct pipe_surface **s)
 {
-   struct cell_texture *ct = cell_texture((*s)->texture);
-   const uint level = (*s)->level;
    struct pipe_surface *surf = *s;
 
-   if ((surf->usage & PIPE_BUFFER_USAGE_CPU_READ) && (ct->untiled_data[level]))
-   {
-      align_free(ct->untiled_data[level]);
-      ct->untiled_data[level] = NULL;
-   }
-
-   if ((ct->base.tex_usage & PIPE_TEXTURE_USAGE_SAMPLER) &&
-       (surf->usage & PIPE_BUFFER_USAGE_CPU_WRITE)) {
-      /* convert from linear to tiled layout */
-      cell_twiddle_texture(screen, surf);
-   }
-
-   /* XXX if done rendering to teximage, re-tile */
-
    if (--surf->refcount == 0) {
       pipe_texture_reference(&surf->texture, NULL);
       FREE(surf);
@@ -476,6 +332,11 @@ cell_tex_surface_release(struct pipe_screen *screen,
 }
 
 
+/**
+ * Create new pipe_transfer object.
+ * This is used by the user to put tex data into a texture (and get it
+ * back out for glGetTexImage).
+ */
 static struct pipe_transfer *
 cell_get_tex_transfer(struct pipe_screen *screen,
                       struct pipe_texture *texture,
@@ -485,14 +346,13 @@ cell_get_tex_transfer(struct pipe_screen *screen,
 {
    struct cell_texture *ct = cell_texture(texture);
    struct cell_transfer *ctrans;
-   struct pipe_transfer *pt;
 
    assert(texture);
    assert(level <= texture->last_level);
 
    ctrans = CALLOC_STRUCT(cell_transfer);
-   pt = &ctrans->base;
    if (ctrans) {
+      struct pipe_transfer *pt = &ctrans->base;
       pt->refcount = 1;
       pipe_texture_reference(&pt->texture, texture);
       pt->format = texture->format;
@@ -504,23 +364,26 @@ cell_get_tex_transfer(struct pipe_screen *screen,
       pt->nblocksx = texture->nblocksx[level];
       pt->nblocksy = texture->nblocksy[level];
       pt->stride = ct->stride[level];
-      ctrans->offset = ct->level_offset[level];
       pt->usage = usage;
       pt->face = face;
       pt->level = level;
       pt->zslice = zslice;
 
-      if (texture->target == PIPE_TEXTURE_CUBE ||
-          texture->target == PIPE_TEXTURE_3D) {
-         ctrans->offset += ((texture->target == PIPE_TEXTURE_CUBE) ? face :
-                         zslice) * pt->nblocksy * pt->stride;
+      ctrans->offset = ct->level_offset[level];
+
+      if (texture->target == PIPE_TEXTURE_CUBE) {
+         ctrans->offset += face * pt->nblocksy * pt->stride;
+      }
+      else if (texture->target == PIPE_TEXTURE_3D) {
+         ctrans->offset += zslice * pt->nblocksy * pt->stride;
       }
       else {
          assert(face == 0);
          assert(zslice == 0);
       }
+      return pt;
    }
-   return pt;
+   return NULL;
 }
 
 
@@ -542,16 +405,23 @@ cell_tex_transfer_release(struct pipe_screen *screen,
 }
 
 
+/**
+ * Return pointer to texture image data in linear layout.
+ */
 static void *
-cell_transfer_map( struct pipe_screen *screen,
-                       struct pipe_transfer *transfer )
+cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
 {
-   ubyte *map;
-   struct cell_texture *spt;
-   unsigned flags = 0;
+   struct cell_transfer *ctrans = cell_transfer(transfer);
+   struct pipe_texture *pt = transfer->texture;
+   struct cell_texture *ct = cell_texture(pt);
+   const uint level = ctrans->base.level;
+   const uint texWidth = pt->width[level];
+   const uint texHeight = pt->height[level];
+   const uint stride = ct->stride[level];
+   unsigned flags = 0x0;
+   unsigned size;
 
    assert(transfer->texture);
-   spt = cell_texture(transfer->texture);
 
    if (transfer->usage != PIPE_TRANSFER_READ) {
       flags |= PIPE_BUFFER_USAGE_CPU_WRITE;
@@ -561,89 +431,82 @@ cell_transfer_map( struct pipe_screen *screen,
       flags |= PIPE_BUFFER_USAGE_CPU_READ;
    }
 
-   map = pipe_buffer_map(screen, spt->buffer, flags);
-   if (map == NULL)
-      return NULL;
+   if (!ct->mapped) {
+      /* map now */
+      ct->mapped = pipe_buffer_map(screen, ct->buffer, flags);
+   }
 
-   /* May want to different things here depending on read/write nature
-    * of the map:
+   /*
+    * Create a buffer of ordinary memory for the linear texture.
+    * This is the memory that the user will read/write.
     */
-   if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) 
-   {
-      /* Do something to notify sharing contexts of a texture change.
-       * In cell, that would mean flushing the texture cache.
-       */
-#if 00
-      cell_screen(screen)->timestamp++;
-#endif
+   size = pt->nblocksx[level] * pt->nblocksy[level] * pt->block.size;
+
+   ctrans->map = align_malloc(size, 16);
+   if (!ctrans->map)
+      return NULL; /* out of memory */
+
+   if (transfer->usage & PIPE_TRANSFER_READ) {
+      /* need to untwiddle the texture to make a linear version */
+      const uint bpp = pf_get_size(ct->base.format);
+      if (bpp == 4) {
+         const uint *src = (uint *) (ct->mapped + ctrans->offset);
+         uint *dst = ctrans->map;
+         untwiddle_image_uint(texWidth, texHeight, TILE_SIZE,
+                              dst, stride, src);
+      }
+      else {
+         // xxx fix
+      }
    }
-   
-   return map + cell_transfer(transfer)->offset +
-      transfer->y / transfer->block.height * transfer->stride +
-      transfer->x / transfer->block.width * transfer->block.size;
+
+   return ctrans->map;
 }
 
 
+/**
+ * Called when user is done reading/writing texture data.
+ * If new data was written, this is where we convert the linear data
+ * to tiled data.
+ */
 static void
 cell_transfer_unmap(struct pipe_screen *screen,
-                       struct pipe_transfer *transfer)
+                    struct pipe_transfer *transfer)
 {
-   struct cell_texture *spt;
-
-   assert(transfer->texture);
-   spt = cell_texture(transfer->texture);
-
-   pipe_buffer_unmap( screen, spt->buffer );
-}
-
-
-static void *
-cell_surface_map(struct pipe_screen *screen,
-                 struct pipe_surface *surface,
-                 unsigned flags)
-{
-   ubyte *map;
-   struct cell_texture *ct = cell_texture(surface->texture);
-   const uint level = surface->level;
-
-   assert(ct);
-
-#if 0
-   if (flags & ~surface->usage) {
-      assert(0);
-      return NULL;
+   struct cell_transfer *ctrans = cell_transfer(transfer);
+   struct pipe_texture *pt = transfer->texture;
+   struct cell_texture *ct = cell_texture(pt);
+   const uint level = ctrans->base.level;
+   const uint texWidth = pt->width[level];
+   const uint texHeight = pt->height[level];
+   const uint stride = ct->stride[level];
+
+   if (!ct->mapped) {
+      /* map now */
+      ct->mapped = pipe_buffer_map(screen, ct->buffer,
+                                   PIPE_BUFFER_USAGE_CPU_READ);
    }
-#endif
 
-   map = pipe_buffer_map( screen, ct->buffer, flags );
-   if (map == NULL) {
-      return NULL;
-   }
-   else {
-      if ((surface->usage & PIPE_BUFFER_USAGE_CPU_READ) &&
-          (ct->untiled_data[level])) {
-         return (void *) ((ubyte *) ct->untiled_data[level] + surface->offset);
+   if (transfer->usage & PIPE_TRANSFER_WRITE) {
+      /* The user wrote new texture data into the mapped buffer.
+       * We need to convert the new linear data into the twiddled/tiled format.
+       */
+      const uint bpp = pf_get_size(ct->base.format);
+      if (bpp == 4) {
+         const uint *src = ctrans->map;
+         uint *dst = (uint *) (ct->mapped + ctrans->offset);
+         twiddle_image_uint(texWidth, texHeight, TILE_SIZE, dst, stride, src);
       }
       else {
-         return (void *) (map + surface->offset);
+         // xxx fix
       }
    }
-}
-
-
-static void
-cell_surface_unmap(struct pipe_screen *screen,
-                   struct pipe_surface *surface)
-{
-   struct cell_texture *ct = cell_texture(surface->texture);
-
-   assert(ct);
 
-   pipe_buffer_unmap( screen, ct->buffer );
+   align_free(ctrans->map);
+   ctrans->map = NULL;
 }
 
 
-
 void
 cell_init_screen_texture_funcs(struct pipe_screen *screen)
 {
@@ -655,6 +518,7 @@ cell_init_screen_texture_funcs(struct pipe_screen *screen)
 
    screen->get_tex_transfer = cell_get_tex_transfer;
    screen->tex_transfer_release = cell_tex_transfer_release;
+
    screen->transfer_map = cell_transfer_map;
    screen->transfer_unmap = cell_transfer_unmap;
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.h b/src/gallium/drivers/cell/ppu/cell_texture.h
index fc6486adbe..3ffc0bfdb5 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.h
+++ b/src/gallium/drivers/cell/ppu/cell_texture.h
@@ -43,20 +43,14 @@ struct cell_texture
    unsigned long level_offset[CELL_MAX_TEXTURE_LEVELS];
    unsigned long stride[CELL_MAX_TEXTURE_LEVELS];
 
-   /* The data is held here:
-    */
+   /** The tiled texture data is held in this buffer */
    struct pipe_buffer *buffer;
    unsigned long buffer_size;
 
-   /** Texture data in tiled layout is held here */
-   struct pipe_buffer *tiled_buffer[CELL_MAX_TEXTURE_LEVELS];
-   /** Mapped, tiled texture data */
-   void *tiled_mapped[CELL_MAX_TEXTURE_LEVELS];
-
-   struct pipe_transfer *transfer;
-
-   /** The original, linear texture data */
-   void *untiled_data[CELL_MAX_TEXTURE_LEVELS];
+   /** The buffer above, mapped.  This is the memory from which the
+    * SPUs will fetch texels.  This texture data is in the tiled layout.
+    */
+   ubyte *mapped;
 };
 
 
@@ -65,6 +59,7 @@ struct cell_transfer
    struct pipe_transfer base;
 
    unsigned long offset;
+   void *map;
 };
 
 
diff --git a/src/gallium/drivers/i915simple/i915_screen.c b/src/gallium/drivers/i915simple/i915_screen.c
index b7bd3b3b74..49471287a2 100644
--- a/src/gallium/drivers/i915simple/i915_screen.c
+++ b/src/gallium/drivers/i915simple/i915_screen.c
@@ -36,6 +36,7 @@
 #include "i915_context.h"
 #include "i915_screen.h"
 #include "i915_texture.h"
+#include "i915_winsys.h"
 
 
 static const char *
diff --git a/src/gallium/drivers/i915simple/i915_screen.h b/src/gallium/drivers/i915simple/i915_screen.h
index a371663453..5284c32595 100644
--- a/src/gallium/drivers/i915simple/i915_screen.h
+++ b/src/gallium/drivers/i915simple/i915_screen.h
@@ -75,10 +75,6 @@ i915_transfer( struct pipe_transfer *transfer )
 }
 
 
-extern struct pipe_screen *
-i915_create_screen(struct pipe_winsys *winsys, uint pci_id);
-
-
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/gallium/drivers/i915simple/i915_state_emit.c b/src/gallium/drivers/i915simple/i915_state_emit.c
index 26e03f5127..1e1fb968b4 100644
--- a/src/gallium/drivers/i915simple/i915_state_emit.c
+++ b/src/gallium/drivers/i915simple/i915_state_emit.c
@@ -214,7 +214,6 @@ i915_emit_hardware_state(struct i915_context *i915 )
 	 unsigned ctile = BUF_3D_USE_FENCE;
          struct i915_texture *tex = (struct i915_texture *)
                                     cbuf_surface->texture;
-         struct pipe_buffer *buffer = tex->buffer;
          assert(tex);
 
 	 if (tex && tex->tiled) {
@@ -238,7 +237,6 @@ i915_emit_hardware_state(struct i915_context *i915 )
 	 unsigned ztile = BUF_3D_USE_FENCE;
          struct i915_texture *tex = (struct i915_texture *)
                                     depth_surface->texture;
-         struct pipe_buffer *buffer = tex->buffer;
          assert(tex);
 
 	 if (tex && tex->tiled) {
diff --git a/src/gallium/drivers/i915simple/i915_texture.c b/src/gallium/drivers/i915simple/i915_texture.c
index 957726523f..6aead3e75e 100644
--- a/src/gallium/drivers/i915simple/i915_texture.c
+++ b/src/gallium/drivers/i915simple/i915_texture.c
@@ -42,6 +42,7 @@
 #include "i915_texture.h"
 #include "i915_debug.h"
 #include "i915_screen.h"
+#include "i915_winsys.h"
 
 /*
  * Helper function and arrays
@@ -765,3 +766,20 @@ i915_init_screen_texture_functions(struct pipe_screen *screen)
    screen->texture_blanket = i915_texture_blanket;
    screen->tex_surface_release = i915_tex_surface_release;
 }
+
+boolean i915_get_texture_buffer( struct pipe_texture *texture,
+                                 struct pipe_buffer **buf,
+                                 unsigned *stride )
+{
+   struct i915_texture *tex = (struct i915_texture *)texture;
+
+   if (!tex)
+      return FALSE;
+
+   pipe_buffer_reference(texture->screen, buf, tex->buffer);
+
+   if (stride)
+      *stride = tex->stride;
+
+   return TRUE;
+}
diff --git a/src/gallium/drivers/i915simple/i915_winsys.h b/src/gallium/drivers/i915simple/i915_winsys.h
index 81904c2a74..ff5b34f193 100644
--- a/src/gallium/drivers/i915simple/i915_winsys.h
+++ b/src/gallium/drivers/i915simple/i915_winsys.h
@@ -30,7 +30,8 @@
  * This is the interface that i915simple requires any window system
  * hosting it to implement.  This is the only include file in i915simple
  * which is public.
- * 
+ *
+ * This isn't currently true as the winsys needs i915_batchbuffer.h
  */
 
 #ifndef I915_WINSYS_H
@@ -45,10 +46,9 @@ extern "C" {
 #endif
 
    
-/* Pipe drivers are (meant to be!) independent of both GL and the
- * window system.  The window system provides a buffer manager and a
- * set of additional hooks for things like command buffer submission,
- * etc.
+/* Pipe drivers are independent of both GL and the window system.
+ * The window system provides a buffer manager and a set of additional
+ * hooks for things like command buffer submission, etc.
  *
  * There clearly has to be some agreement between the window system
  * driver and the hardware driver about the format of command buffers,
@@ -56,6 +56,7 @@ extern "C" {
  */
 
 struct i915_batchbuffer;
+struct pipe_texture;
 struct pipe_buffer;
 struct pipe_fence_handle;
 struct pipe_winsys;
@@ -64,7 +65,7 @@ struct pipe_screen;
 
 /**
  * Additional winsys interface for i915simple.
- * 
+ *
  * It is an over-simple batchbuffer mechanism.  Will want to improve the
  * performance of this, perhaps based on the cmdstream stuff.  It
  * would be pretty impossible to implement swz on top of this
@@ -110,12 +111,33 @@ struct i915_winsys {
 #define I915_BUFFER_USAGE_LIT_VERTEX  (PIPE_BUFFER_USAGE_CUSTOM << 0)
 
 
-struct pipe_context *i915_create_context( struct pipe_screen *,
-                                          struct pipe_winsys *,
-                                          struct i915_winsys * );
+/**
+ * Create i915 pipe_screen.
+ */
+struct pipe_screen *i915_create_screen( struct pipe_winsys *winsys,
+                                        uint pci_id );
+
+/**
+ * Create a i915 pipe_context.
+ */
+struct pipe_context *i915_create_context( struct pipe_screen *screen,
+                                          struct pipe_winsys *winsys,
+                                          struct i915_winsys *i915 );
+
+/**
+ * Used for the winsys to get the buffer used for a texture
+ * and also the stride used for the texture.
+ *
+ * Buffer is referenced for you so you need to unref after use.
+ *
+ * This is needed for example kms.
+ */
+boolean i915_get_texture_buffer( struct pipe_texture *texture,
+                                 struct pipe_buffer **buf,
+                                 unsigned *stride );
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif 
+#endif
diff --git a/src/gallium/drivers/nv04/Makefile b/src/gallium/drivers/nv04/Makefile
index cf9deea851..7c14bacb1d 100644
--- a/src/gallium/drivers/nv04/Makefile
+++ b/src/gallium/drivers/nv04/Makefile
@@ -15,6 +15,7 @@ C_SOURCES = \
 	nv04_state.c \
 	nv04_state_emit.c \
 	nv04_surface.c \
+	nv04_transfer.c \
 	nv04_vbo.c
 
 include ../../Makefile.template
diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c
index 993c5ef5dd..9acd613e2e 100644
--- a/src/gallium/drivers/nv04/nv04_miptree.c
+++ b/src/gallium/drivers/nv04/nv04_miptree.c
@@ -50,8 +50,6 @@ nv04_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 	mt->base = *pt;
 	mt->base.refcount = 1;
 	mt->base.screen = pscreen;
-	mt->shadow_tex = NULL;
-	mt->shadow_surface = NULL;
 
 	//mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 
@@ -111,12 +109,6 @@ nv04_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
 			FREE(mt->level[l].image_offset);
 	}
 
-	if (mt->shadow_tex) {
-		assert(mt->shadow_surface);
-		pscreen->tex_surface_release(pscreen, &mt->shadow_surface);
-		nv04_miptree_release(pscreen, &mt->shadow_tex);
-	}
-
 	FREE(mt);
 }
 
@@ -126,29 +118,26 @@ nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 			 unsigned flags)
 {
 	struct nv04_miptree *nv04mt = (struct nv04_miptree *)pt;
-	struct pipe_surface *ps;
+	struct nv04_surface *ns;
 
-	ps = CALLOC_STRUCT(pipe_surface);
-	if (!ps)
+	ns = CALLOC_STRUCT(nv04_surface);
+	if (!ns)
 		return NULL;
-	pipe_texture_reference(&ps->texture, pt);
-	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
-	ps->block = pt->block;
-	ps->nblocksx = pt->nblocksx[level];
-	ps->nblocksy = pt->nblocksy[level];
-	ps->stride = nv04mt->level[level].pitch;
-	ps->usage = flags;
-	ps->status = PIPE_SURFACE_STATUS_DEFINED;
-	ps->refcount = 1;
-	ps->face = face;
-	ps->level = level;
-	ps->zslice = zslice;
-
-	ps->offset = nv04mt->level[level].image_offset;
-
-	return ps;
+	pipe_texture_reference(&ns->base.texture, pt);
+	ns->base.format = pt->format;
+	ns->base.width = pt->width[level];
+	ns->base.height = pt->height[level];
+	ns->base.usage = flags;
+	ns->base.status = PIPE_SURFACE_STATUS_DEFINED;
+	ns->base.refcount = 1;
+	ns->base.face = face;
+	ns->base.level = level;
+	ns->base.zslice = zslice;
+	ns->pitch = nv04mt->level[level].pitch;
+
+	ns->base.offset = nv04mt->level[level].image_offset;
+
+	return &ns->base;
 }
 
 static void
diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c
index 9ef38bc244..f9f6d97426 100644
--- a/src/gallium/drivers/nv04/nv04_screen.c
+++ b/src/gallium/drivers/nv04/nv04_screen.c
@@ -119,28 +119,6 @@ nv04_screen_is_format_supported(struct pipe_screen *screen,
 	return FALSE;
 }
 
-static void *
-nv04_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
-		 unsigned flags )
-{
-	void *map;
-	struct nv04_miptree *nv04mt = (struct nv04_miptree *)surface->texture;
-
-	map = pipe_buffer_map(screen, nv04mt->buffer, flags);
-	if (!map)
-		return NULL;
-
-	return map + surface->offset;
-}
-
-static void
-nv04_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
-{
-	struct nv04_miptree *nv04mt = (struct nv04_miptree *)surface->texture;
-
-	pipe_buffer_unmap(screen, nv04mt->buffer);
-}
-
 static void
 nv04_screen_destroy(struct pipe_screen *pscreen)
 {
@@ -226,10 +204,8 @@ nv04_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 
 	screen->pipe.is_format_supported = nv04_screen_is_format_supported;
 
-	screen->pipe.surface_map = nv04_surface_map;
-	screen->pipe.surface_unmap = nv04_surface_unmap;
-
 	nv04_screen_init_miptree_functions(&screen->pipe);
+	nv04_screen_init_transfer_functions(&screen->pipe);
 	u_simple_screen_init(&screen->pipe);
 
 	return &screen->pipe;
diff --git a/src/gallium/drivers/nv04/nv04_screen.h b/src/gallium/drivers/nv04/nv04_screen.h
index 540aec907b..ee6fb6db44 100644
--- a/src/gallium/drivers/nv04/nv04_screen.h
+++ b/src/gallium/drivers/nv04/nv04_screen.h
@@ -24,4 +24,7 @@ nv04_screen(struct pipe_screen *screen)
 	return (struct nv04_screen *)screen;
 }
 
+void
+nv04_screen_init_transfer_functions(struct pipe_screen *pscreen);
+
 #endif
diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h
index 15d4685ec1..0d51439e3f 100644
--- a/src/gallium/drivers/nv04/nv04_state.h
+++ b/src/gallium/drivers/nv04/nv04_state.h
@@ -35,9 +35,6 @@ struct nv04_miptree {
 	struct pipe_buffer *buffer;
 	uint total_size;
 
-	struct pipe_texture *shadow_tex;
-	struct pipe_surface *shadow_surface;
-
 	struct {
 		uint pitch;
 		uint image_offset;
diff --git a/src/gallium/drivers/nv04/nv04_state_emit.c b/src/gallium/drivers/nv04/nv04_state_emit.c
index bd8ef1adbf..eb2c1c57c6 100644
--- a/src/gallium/drivers/nv04/nv04_state_emit.c
+++ b/src/gallium/drivers/nv04/nv04_state_emit.c
@@ -93,7 +93,7 @@ static void nv04_emit_sampler(struct nv04_context *nv04, int unit)
 static void nv04_state_emit_framebuffer(struct nv04_context* nv04)
 {
 	struct pipe_framebuffer_state* fb = nv04->framebuffer;
-	struct pipe_surface *rt, *zeta;
+	struct nv04_surface *rt, *zeta;
 	uint32_t rt_format, w, h;
 	int colour_format = 0, zeta_format = 0;
 	struct nv04_miptree *nv04mt = 0;
@@ -101,7 +101,7 @@ static void nv04_state_emit_framebuffer(struct nv04_context* nv04)
 	w = fb->cbufs[0]->width;
 	h = fb->cbufs[0]->height;
 	colour_format = fb->cbufs[0]->format;
-	rt = fb->cbufs[0];
+	rt = (struct nv04_surface *)fb->cbufs[0];
 
 	if (fb->zsbuf) {
 		if (colour_format) {
@@ -113,7 +113,7 @@ static void nv04_state_emit_framebuffer(struct nv04_context* nv04)
 		}
 
 		zeta_format = fb->zsbuf->format;
-		zeta = fb->zsbuf;
+		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
 	switch (colour_format) {
@@ -131,13 +131,13 @@ static void nv04_state_emit_framebuffer(struct nv04_context* nv04)
 	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_FORMAT, 1);
 	OUT_RING(rt_format);
 
-	nv04mt = (struct nv04_miptree *)rt->texture;
+	nv04mt = (struct nv04_miptree *)rt->base.texture;
 	/* FIXME pitches have to be aligned ! */
 	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
-	OUT_RING(rt->stride|(zeta->stride<<16));
+	OUT_RING(rt->pitch|(zeta->pitch<<16));
 	OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	if (fb->zsbuf) {
-		nv04mt = (struct nv04_miptree *)zeta->texture;
+		nv04mt = (struct nv04_miptree *)zeta->base.texture;
 		BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
 		OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	}
@@ -202,8 +202,11 @@ nv04_emit_hw_state(struct nv04_context *nv04)
 	 */
 
 	/* Render target */
+	unsigned rt_pitch = ((struct nv04_surface *)nv04->rt)->pitch;
+	unsigned zeta_pitch = ((struct nv04_surface *)nv04->zeta)->pitch;
+
 	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
-	OUT_RING(nv04->rt->stride|(nv04->zeta->stride<<16));
+	OUT_RING(rt_pitch|(zeta_pitch<<16));
 	OUT_RELOCl(nv04->rt, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	if (nv04->zeta) {
 		BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
diff --git a/src/gallium/drivers/nv04/nv04_surface_2d.c b/src/gallium/drivers/nv04/nv04_surface_2d.c
index 230cfd17dd..e8fd3166ab 100644
--- a/src/gallium/drivers/nv04/nv04_surface_2d.c
+++ b/src/gallium/drivers/nv04/nv04_surface_2d.c
@@ -101,6 +101,7 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 	struct nouveau_grobj *sifm = ctx->sifm;
 	struct nouveau_bo *src_bo = ctx->nvws->get_bo(ctx->buf(src));
 	struct nouveau_bo *dst_bo = ctx->nvws->get_bo(ctx->buf(dst));
+	const unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
 	const unsigned max_w = 1024;
 	const unsigned max_h = 1024;
 	const unsigned sub_w = w > max_w ? max_w : w;
@@ -110,6 +111,8 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 
 	/* POT or GTFO */
 	assert(!(w & (w - 1)) && !(h & (h - 1)));
+	/* That's the way she likes it */
+	assert(src_pitch == ((struct nv04_surface *)dst)->pitch);
 
 	BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1);
 	OUT_RELOCo(chan, dst_bo,
@@ -130,7 +133,7 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 	  for (cx = 0; cx < w; cx += sub_w) {
 	    BEGIN_RING(chan, swzsurf, NV04_SWIZZLED_SURFACE_OFFSET, 1);
 	    OUT_RELOCl(chan, dst_bo, dst->offset + nv04_swizzle_bits(cx, cy) *
-			     dst->block.size, NOUVEAU_BO_GART |
+			     dst->texture->block.size, NOUVEAU_BO_GART |
 			     NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
 	    BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 9);
@@ -146,11 +149,11 @@ nv04_surface_copy_swizzle(struct nv04_surface_2d *ctx,
 
 	    BEGIN_RING(chan, sifm, NV04_SCALED_IMAGE_FROM_MEMORY_SIZE, 4);
 	    OUT_RING  (chan, sub_h << 16 | sub_w);
-	    OUT_RING  (chan, src->stride |
+	    OUT_RING  (chan, src_pitch |
 			     NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER |
 			     NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE);
-	    OUT_RELOCl(chan, src_bo, src->offset + cy * src->stride +
-			     cx * src->block.size, NOUVEAU_BO_GART |
+	    OUT_RELOCl(chan, src_bo, src->offset + cy * src_pitch +
+			     cx * src->texture->block.size, NOUVEAU_BO_GART |
 			     NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
 	    OUT_RING  (chan, 0);
 	  }
@@ -168,10 +171,12 @@ nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx,
 	struct nouveau_grobj *m2mf = ctx->m2mf;
 	struct nouveau_bo *src_bo = ctx->nvws->get_bo(ctx->buf(src));
 	struct nouveau_bo *dst_bo = ctx->nvws->get_bo(ctx->buf(dst));
-	unsigned dst_offset, src_offset;
-
-	dst_offset = dst->offset + (dy * dst->stride) + (dx * dst->block.size);
-	src_offset = src->offset + (sy * src->stride) + (sx * src->block.size);
+	unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
+	unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch;
+	unsigned dst_offset = dst->offset + dy * dst_pitch +
+	                      dx * dst->texture->block.size;
+	unsigned src_offset = src->offset + sy * src_pitch +
+	                      sx * src->texture->block.size;
 
 	WAIT_RING (chan, 3 + ((h / 2047) + 1) * 9);
 	BEGIN_RING(chan, m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2);
@@ -188,16 +193,16 @@ nv04_surface_copy_m2mf(struct nv04_surface_2d *ctx,
 			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
 		OUT_RELOCl(chan, dst_bo, dst_offset,
 			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR);
-		OUT_RING  (chan, src->stride);
-		OUT_RING  (chan, dst->stride);
-		OUT_RING  (chan, w * src->block.size);
+		OUT_RING  (chan, src_pitch);
+		OUT_RING  (chan, dst_pitch);
+		OUT_RING  (chan, w * src->texture->block.size);
 		OUT_RING  (chan, count);
 		OUT_RING  (chan, 0x0101);
 		OUT_RING  (chan, 0);
 
 		h -= count;
-		src_offset += src->stride * count;
-		dst_offset += dst->stride * count;
+		src_offset += src_pitch * count;
+		dst_offset += dst_pitch * count;
 	}
 
 	return 0;
@@ -213,6 +218,8 @@ nv04_surface_copy_blit(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	struct nouveau_grobj *blit = ctx->blit;
 	struct nouveau_bo *src_bo = ctx->nvws->get_bo(ctx->buf(src));
 	struct nouveau_bo *dst_bo = ctx->nvws->get_bo(ctx->buf(dst));
+	unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
+	unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch;
 	int format;
 
 	format = nv04_surface_format(dst->format);
@@ -225,7 +232,7 @@ nv04_surface_copy_blit(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4);
 	OUT_RING  (chan, format);
-	OUT_RING  (chan, (dst->stride << 16) | src->stride);
+	OUT_RING  (chan, (dst_pitch << 16) | src_pitch);
 	OUT_RELOCl(chan, src_bo, src->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
 	OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
@@ -242,6 +249,8 @@ nv04_surface_copy(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 		  int dx, int dy, struct pipe_surface *src, int sx, int sy,
 		  int w, int h)
 {
+	unsigned src_pitch = ((struct nv04_surface *)src)->pitch;
+	unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch;
 	int src_linear = src->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR;
 	int dst_linear = dst->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR;
 
@@ -257,7 +266,8 @@ nv04_surface_copy(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	 * to NV_MEMORY_TO_MEMORY_FORMAT in this case.
 	 */
 	if ((src->offset & 63) || (dst->offset & 63) ||
-	    (src->stride & 63) || (dst->stride & 63)) {
+	    (src_pitch & 63) || (dst_pitch & 63) ||
+	    debug_get_bool_option("NOUVEAU_NO_COPYBLIT", FALSE)) {
 		nv04_surface_copy_m2mf(ctx, dst, dx, dy, src, sx, sy, w, h);
 		return;
 	}
@@ -273,6 +283,7 @@ nv04_surface_fill(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	struct nouveau_grobj *surf2d = ctx->surf2d;
 	struct nouveau_grobj *rect = ctx->rect;
 	struct nouveau_bo *dst_bo = ctx->nvws->get_bo(ctx->buf(dst));
+	unsigned dst_pitch = ((struct nv04_surface *)dst)->pitch;
 	int cs2d_format, gdirect_format;
 
 	cs2d_format = nv04_surface_format(dst->format);
@@ -287,7 +298,7 @@ nv04_surface_fill(struct nv04_surface_2d *ctx, struct pipe_surface *dst,
 	OUT_RELOCo(chan, dst_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4);
 	OUT_RING  (chan, cs2d_format);
-	OUT_RING  (chan, (dst->stride << 16) | dst->stride);
+	OUT_RING  (chan, (dst_pitch << 16) | dst_pitch);
 	OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 	OUT_RELOCl(chan, dst_bo, dst->offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
diff --git a/src/gallium/drivers/nv04/nv04_surface_2d.h b/src/gallium/drivers/nv04/nv04_surface_2d.h
index 21b8f86960..82ce7189c8 100644
--- a/src/gallium/drivers/nv04/nv04_surface_2d.h
+++ b/src/gallium/drivers/nv04/nv04_surface_2d.h
@@ -1,6 +1,11 @@
 #ifndef __NV04_SURFACE_2D_H__
 #define __NV04_SURFACE_2D_H__
 
+struct nv04_surface {
+	struct pipe_surface base;
+	unsigned pitch;
+};
+
 struct nv04_surface_2d {
 	struct nouveau_winsys *nvws;
 	struct nouveau_notifier *ntfy;
diff --git a/src/gallium/drivers/nv04/nv04_transfer.c b/src/gallium/drivers/nv04/nv04_transfer.c
new file mode 100644
index 0000000000..314d2045f7
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_transfer.c
@@ -0,0 +1,201 @@
+#include <pipe/p_state.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <nouveau/nouveau_winsys.h>
+#include "nv04_context.h"
+#include "nv04_screen.h"
+#include "nv04_state.h"
+
+struct nv04_transfer {
+	struct pipe_transfer base;
+	struct pipe_surface *surface;
+	bool direct;
+};
+
+static unsigned nv04_usage_tx_to_buf(unsigned tx_usage)
+{
+	switch (tx_usage) {
+		case PIPE_TRANSFER_READ:
+			return PIPE_BUFFER_USAGE_CPU_READ;
+		case PIPE_TRANSFER_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_WRITE;
+		case PIPE_TRANSFER_READ_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static void
+nv04_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+                             struct pipe_texture *template)
+{
+	memset(template, 0, sizeof(struct pipe_texture));
+	template->target = pt->target;
+	template->format = pt->format;
+	template->width[0] = pt->width[level];
+	template->height[0] = pt->height[level];
+	template->depth[0] = 1;
+	template->block = pt->block;
+	template->nblocksx[0] = pt->nblocksx[level];
+	template->nblocksy[0] = pt->nblocksx[level];
+	template->last_level = 0;
+	template->compressed = pt->compressed;
+	template->nr_samples = pt->nr_samples;
+
+	template->tex_usage = PIPE_TEXTURE_USAGE_DYNAMIC |
+	                      NOUVEAU_TEXTURE_USAGE_LINEAR;
+}
+
+static struct pipe_transfer *
+nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+		  unsigned face, unsigned level, unsigned zslice,
+		  enum pipe_transfer_usage usage,
+		  unsigned x, unsigned y, unsigned w, unsigned h)
+{
+	struct nv04_miptree *mt = (struct nv04_miptree *)pt;
+	struct nv04_transfer *tx;
+	struct pipe_texture tx_tex_template, *tx_tex;
+
+	tx = CALLOC_STRUCT(nv04_transfer);
+	if (!tx)
+		return NULL;
+
+	tx->base.refcount = 1;
+	pipe_texture_reference(&tx->base.texture, pt);
+	tx->base.format = pt->format;
+	tx->base.x = x;
+	tx->base.y = y;
+	tx->base.width = w;
+	tx->base.height = h;
+	tx->base.block = pt->block;
+	tx->base.nblocksx = pt->nblocksx[level];
+	tx->base.nblocksy = pt->nblocksy[level];
+	tx->base.stride = mt->level[level].pitch;
+	tx->base.usage = usage;
+	tx->base.face = face;
+	tx->base.level = level;
+	tx->base.zslice = zslice;
+
+	/* Direct access to texture */
+	if ((pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC ||
+	     debug_get_bool_option("NOUVEAU_NO_TRANSFER", TRUE/*XXX:FALSE*/)) &&
+	    pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)
+	{
+		tx->direct = true;
+		tx->surface = pscreen->get_tex_surface(pscreen, pt,
+	                                               face, level, zslice,
+	                                               nv04_usage_tx_to_buf(usage));
+		return &tx->base;
+	}
+
+	tx->direct = false;
+
+	nv04_compatible_transfer_tex(pt, level, &tx_tex_template);
+
+	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
+	if (!tx_tex)
+	{
+		FREE(tx);
+		return NULL;
+	}
+
+	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
+	                                       face, level, zslice,
+	                                       nv04_usage_tx_to_buf(usage));
+
+	pipe_texture_reference(&tx_tex, NULL);
+
+	if (!tx->surface)
+	{
+		pipe_surface_reference(&tx->surface, NULL);
+		FREE(tx);
+		return NULL;
+	}
+
+	if (usage != PIPE_TRANSFER_WRITE) {
+		struct nv04_screen *nvscreen = nv04_screen(pscreen);
+		struct pipe_surface *src;
+
+		src = pscreen->get_tex_surface(pscreen, pt,
+	                                       face, level, zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_READ);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		/* TODO: Check if SIFM can un-swizzle */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      tx->surface, 0, 0,
+		                      src, 0, 0,
+		                      src->width, src->height);
+
+		pipe_surface_reference(&src, NULL);
+	}
+
+	return &tx->base;
+}
+
+static void
+nv04_transfer_del(struct pipe_screen *pscreen, struct pipe_transfer **pptx)
+{
+	struct pipe_transfer *ptx = *pptx;
+	struct nv04_transfer *tx = (struct nv04_transfer *)ptx;
+
+	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+		struct nv04_screen *nvscreen = nv04_screen(pscreen);
+		struct pipe_surface *dst;
+
+		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
+	                                       ptx->face, ptx->level, ptx->zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      dst, 0, 0,
+		                      tx->surface, 0, 0,
+		                      dst->width, dst->height);
+
+		pipe_surface_reference(&dst, NULL);
+	}
+
+	*pptx = NULL;
+	if (--ptx->refcount)
+		return;
+
+	pipe_surface_reference(&tx->surface, NULL);
+	pipe_texture_reference(&ptx->texture, NULL);
+	FREE(ptx);
+}
+
+static void *
+nv04_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv04_transfer *tx = (struct nv04_transfer *)ptx;
+	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
+	struct nv04_miptree *mt = (struct nv04_miptree *)tx->surface->texture;
+	void *map = pipe_buffer_map(pscreen, mt->buffer,
+	                            nv04_usage_tx_to_buf(ptx->usage));
+
+	return map + ns->base.offset +
+	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+}
+
+static void
+nv04_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv04_transfer *tx = (struct nv04_transfer *)ptx;
+	struct nv04_miptree *mt = (struct nv04_miptree *)tx->surface->texture;
+
+	pipe_buffer_unmap(pscreen, mt->buffer);
+}
+
+void
+nv04_screen_init_transfer_functions(struct pipe_screen *pscreen)
+{
+	pscreen->get_tex_transfer = nv04_transfer_new;
+	pscreen->tex_transfer_release = nv04_transfer_del;
+	pscreen->transfer_map = nv04_transfer_map;
+	pscreen->transfer_unmap = nv04_transfer_unmap;
+}
diff --git a/src/gallium/drivers/nv10/Makefile b/src/gallium/drivers/nv10/Makefile
index 2b5fbd4f5a..62677f5194 100644
--- a/src/gallium/drivers/nv10/Makefile
+++ b/src/gallium/drivers/nv10/Makefile
@@ -14,6 +14,7 @@ C_SOURCES = \
 	nv10_state.c \
 	nv10_state_emit.c \
 	nv10_surface.c \
+	nv10_transfer.c \
 	nv10_vbo.c
 
 include ../../Makefile.template
diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c
index 9616135461..4747868a50 100644
--- a/src/gallium/drivers/nv10/nv10_miptree.c
+++ b/src/gallium/drivers/nv10/nv10_miptree.c
@@ -131,30 +131,31 @@ nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 			 unsigned face, unsigned level, unsigned zslice,
 			 unsigned flags)
 {
-	struct pipe_winsys *ws = screen->winsys;
 	struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt;
-	struct pipe_surface *ps;
+	struct nv04_surface *ns;
 
-	ps = CALLOC_STRUCT(pipe_surface);
-	if (!ps)
+	ns = CALLOC_STRUCT(nv04_surface);
+	if (!ns)
 		return NULL;
-	pipe_texture_reference(&ps->texture, pt);
-	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
-	ps->block = pt->block;
-	ps->nblocksx = pt->nblocksx[level];
-	ps->nblocksy = pt->nblocksy[level];
-	ps->stride = nv10mt->level[level].pitch;
-	ps->refcount = 1;
+	pipe_texture_reference(&ns->base.texture, pt);
+	ns->base.format = pt->format;
+	ns->base.width = pt->width[level];
+	ns->base.height = pt->height[level];
+	ns->base.usage = flags;
+	ns->base.status = PIPE_SURFACE_STATUS_DEFINED;
+	ns->base.refcount = 1;
+	ns->base.face = face;
+	ns->base.level = level;
+	ns->base.zslice = zslice;
+	ns->pitch = nv10mt->level[level].pitch;
 
 	if (pt->target == PIPE_TEXTURE_CUBE) {
-		ps->offset = nv10mt->level[level].image_offset[face];
+		ns->base.offset = nv10mt->level[level].image_offset[face];
 	} else {
-		ps->offset = nv10mt->level[level].image_offset[0];
+		ns->base.offset = nv10mt->level[level].image_offset[0];
 	}
 
-	return ps;
+	return &ns->base;
 }
 
 static void
diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c
index f417b06c94..6532a93c7b 100644
--- a/src/gallium/drivers/nv10/nv10_screen.c
+++ b/src/gallium/drivers/nv10/nv10_screen.c
@@ -116,30 +116,6 @@ nv10_screen_is_format_supported(struct pipe_screen *screen,
 	return FALSE;
 }
 
-static void *
-nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
-		 unsigned flags )
-{
-	struct pipe_winsys *ws = screen->winsys;
-	void *map;
-        struct nv10_miptree *nv10mt = (struct nv10_miptree *)surface->texture;
-
-	map = ws->buffer_map(ws, nv10mt->buffer, flags);
-	if (!map)
-		return NULL;
-
-	return map + surface->offset;
-}
-
-static void
-nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
-{
-	struct pipe_winsys *ws = screen->winsys;
-        struct nv10_miptree *nv10mt = (struct nv10_miptree *)surface->texture;
-
-	ws->buffer_unmap(ws, nv10mt->buffer);
-}
-
 static void
 nv10_screen_destroy(struct pipe_screen *pscreen)
 {
@@ -215,10 +191,8 @@ nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 
 	screen->pipe.is_format_supported = nv10_screen_is_format_supported;
 
-	screen->pipe.surface_map = nv10_surface_map;
-	screen->pipe.surface_unmap = nv10_surface_unmap;
-
 	nv10_screen_init_miptree_functions(&screen->pipe);
+	nv10_screen_init_transfer_functions(&screen->pipe);
 	u_simple_screen_init(&screen->pipe);
 
 	return &screen->pipe;
diff --git a/src/gallium/drivers/nv10/nv10_screen.h b/src/gallium/drivers/nv10/nv10_screen.h
index 60102a369a..ad829ee3fd 100644
--- a/src/gallium/drivers/nv10/nv10_screen.h
+++ b/src/gallium/drivers/nv10/nv10_screen.h
@@ -21,4 +21,8 @@ nv10_screen(struct pipe_screen *screen)
 	return (struct nv10_screen *)screen;
 }
 
+
+void
+nv10_screen_init_transfer_functions(struct pipe_screen *pscreen);
+
 #endif
diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c
index 5dec618b93..d8691ef9c6 100644
--- a/src/gallium/drivers/nv10/nv10_state_emit.c
+++ b/src/gallium/drivers/nv10/nv10_state_emit.c
@@ -103,7 +103,7 @@ static void nv10_state_emit_scissor(struct nv10_context* nv10)
 static void nv10_state_emit_framebuffer(struct nv10_context* nv10)
 {
 	struct pipe_framebuffer_state* fb = nv10->framebuffer;
-	struct pipe_surface *rt, *zeta = NULL;
+	struct nv04_surface *rt, *zeta = NULL;
 	uint32_t rt_format, w, h;
 	int colour_format = 0, zeta_format = 0;
         struct nv10_miptree *nv10mt = 0;
@@ -111,7 +111,7 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10)
 	w = fb->cbufs[0]->width;
 	h = fb->cbufs[0]->height;
 	colour_format = fb->cbufs[0]->format;
-	rt = fb->cbufs[0];
+	rt = (struct nv04_surface *)fb->cbufs[0];
 
 	if (fb->zsbuf) {
 		if (colour_format) {
@@ -123,7 +123,7 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10)
 		}
 
 		zeta_format = fb->zsbuf->format;
-		zeta = fb->zsbuf;
+		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
 	rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR;
@@ -142,18 +142,18 @@ static void nv10_state_emit_framebuffer(struct nv10_context* nv10)
 
 	if (zeta) {
 		BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1);
-		OUT_RING  (rt->stride | (zeta->stride << 16));
+		OUT_RING  (rt->pitch | (zeta->pitch << 16));
 	} else {
 		BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1);
-		OUT_RING  (rt->stride | (rt->stride << 16));
+		OUT_RING  (rt->pitch | (rt->pitch << 16));
 	}
 
-	nv10mt = (struct nv10_miptree *)rt->texture;
+	nv10mt = (struct nv10_miptree *)rt->base.texture;
 	nv10->rt[0] = nv10mt->buffer;
 
 	if (zeta_format)
 	{
-		nv10mt = (struct nv10_miptree *)zeta->texture;
+		nv10mt = (struct nv10_miptree *)zeta->base.texture;
 		nv10->zeta = nv10mt->buffer;
 	}
 
diff --git a/src/gallium/drivers/nv10/nv10_transfer.c b/src/gallium/drivers/nv10/nv10_transfer.c
new file mode 100644
index 0000000000..967e2cc20c
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_transfer.c
@@ -0,0 +1,201 @@
+#include <pipe/p_state.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <nouveau/nouveau_winsys.h>
+#include "nv10_context.h"
+#include "nv10_screen.h"
+#include "nv10_state.h"
+
+struct nv10_transfer {
+	struct pipe_transfer base;
+	struct pipe_surface *surface;
+	bool direct;
+};
+
+static unsigned nv10_usage_tx_to_buf(unsigned tx_usage)
+{
+	switch (tx_usage) {
+		case PIPE_TRANSFER_READ:
+			return PIPE_BUFFER_USAGE_CPU_READ;
+		case PIPE_TRANSFER_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_WRITE;
+		case PIPE_TRANSFER_READ_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static void
+nv10_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+                             struct pipe_texture *template)
+{
+	memset(template, 0, sizeof(struct pipe_texture));
+	template->target = pt->target;
+	template->format = pt->format;
+	template->width[0] = pt->width[level];
+	template->height[0] = pt->height[level];
+	template->depth[0] = 1;
+	template->block = pt->block;
+	template->nblocksx[0] = pt->nblocksx[level];
+	template->nblocksy[0] = pt->nblocksx[level];
+	template->last_level = 0;
+	template->compressed = pt->compressed;
+	template->nr_samples = pt->nr_samples;
+
+	template->tex_usage = PIPE_TEXTURE_USAGE_DYNAMIC |
+	                      NOUVEAU_TEXTURE_USAGE_LINEAR;
+}
+
+static struct pipe_transfer *
+nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+		  unsigned face, unsigned level, unsigned zslice,
+		  enum pipe_transfer_usage usage,
+		  unsigned x, unsigned y, unsigned w, unsigned h)
+{
+	struct nv10_miptree *mt = (struct nv10_miptree *)pt;
+	struct nv10_transfer *tx;
+	struct pipe_texture tx_tex_template, *tx_tex;
+
+	tx = CALLOC_STRUCT(nv10_transfer);
+	if (!tx)
+		return NULL;
+
+	tx->base.refcount = 1;
+	pipe_texture_reference(&tx->base.texture, pt);
+	tx->base.format = pt->format;
+	tx->base.x = x;
+	tx->base.y = y;
+	tx->base.width = w;
+	tx->base.height = h;
+	tx->base.block = pt->block;
+	tx->base.nblocksx = pt->nblocksx[level];
+	tx->base.nblocksy = pt->nblocksy[level];
+	tx->base.stride = mt->level[level].pitch;
+	tx->base.usage = usage;
+	tx->base.face = face;
+	tx->base.level = level;
+	tx->base.zslice = zslice;
+
+	/* Direct access to texture */
+	if ((pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC ||
+	     debug_get_bool_option("NOUVEAU_NO_TRANSFER", TRUE/*XXX:FALSE*/)) &&
+	    pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)
+	{
+		tx->direct = true;
+		tx->surface = pscreen->get_tex_surface(pscreen, pt,
+	                                               face, level, zslice,
+	                                               nv10_usage_tx_to_buf(usage));
+		return &tx->base;
+	}
+
+	tx->direct = false;
+
+	nv10_compatible_transfer_tex(pt, level, &tx_tex_template);
+
+	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
+	if (!tx_tex)
+	{
+		FREE(tx);
+		return NULL;
+	}
+
+	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
+	                                       face, level, zslice,
+	                                       nv10_usage_tx_to_buf(usage));
+
+	pipe_texture_reference(&tx_tex, NULL);
+
+	if (!tx->surface)
+	{
+		pipe_surface_reference(&tx->surface, NULL);
+		FREE(tx);
+		return NULL;
+	}
+
+	if (usage != PIPE_TRANSFER_WRITE) {
+		struct nv10_screen *nvscreen = nv10_screen(pscreen);
+		struct pipe_surface *src;
+
+		src = pscreen->get_tex_surface(pscreen, pt,
+	                                       face, level, zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_READ);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		/* TODO: Check if SIFM can un-swizzle */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      tx->surface, 0, 0,
+		                      src, 0, 0,
+		                      src->width, src->height);
+
+		pipe_surface_reference(&src, NULL);
+	}
+
+	return &tx->base;
+}
+
+static void
+nv10_transfer_del(struct pipe_screen *pscreen, struct pipe_transfer **pptx)
+{
+	struct pipe_transfer *ptx = *pptx;
+	struct nv10_transfer *tx = (struct nv10_transfer *)ptx;
+
+	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+		struct nv10_screen *nvscreen = nv10_screen(pscreen);
+		struct pipe_surface *dst;
+
+		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
+	                                       ptx->face, ptx->level, ptx->zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      dst, 0, 0,
+		                      tx->surface, 0, 0,
+		                      dst->width, dst->height);
+
+		pipe_surface_reference(&dst, NULL);
+	}
+
+	*pptx = NULL;
+	if (--ptx->refcount)
+		return;
+
+	pipe_surface_reference(&tx->surface, NULL);
+	pipe_texture_reference(&ptx->texture, NULL);
+	FREE(ptx);
+}
+
+static void *
+nv10_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv10_transfer *tx = (struct nv10_transfer *)ptx;
+	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
+	struct nv10_miptree *mt = (struct nv10_miptree *)tx->surface->texture;
+	void *map = pipe_buffer_map(pscreen, mt->buffer,
+	                            nv10_usage_tx_to_buf(ptx->usage));
+
+	return map + ns->base.offset +
+	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+}
+
+static void
+nv10_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv10_transfer *tx = (struct nv10_transfer *)ptx;
+	struct nv10_miptree *mt = (struct nv10_miptree *)tx->surface->texture;
+
+	pipe_buffer_unmap(pscreen, mt->buffer);
+}
+
+void
+nv10_screen_init_transfer_functions(struct pipe_screen *pscreen)
+{
+	pscreen->get_tex_transfer = nv10_transfer_new;
+	pscreen->tex_transfer_release = nv10_transfer_del;
+	pscreen->transfer_map = nv10_transfer_map;
+	pscreen->transfer_unmap = nv10_transfer_unmap;
+}
diff --git a/src/gallium/drivers/nv20/Makefile b/src/gallium/drivers/nv20/Makefile
index 93e34f8e92..1305f26c59 100644
--- a/src/gallium/drivers/nv20/Makefile
+++ b/src/gallium/drivers/nv20/Makefile
@@ -14,6 +14,7 @@ C_SOURCES = \
 	nv20_state.c \
 	nv20_state_emit.c \
 	nv20_surface.c \
+	nv20_transfer.c \
 	nv20_vbo.c
 #	nv20_vertprog.c
 
diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c
index ef7e9c5428..2946240897 100644
--- a/src/gallium/drivers/nv20/nv20_miptree.c
+++ b/src/gallium/drivers/nv20/nv20_miptree.c
@@ -9,10 +9,14 @@ static void
 nv20_miptree_layout(struct nv20_miptree *nv20mt)
 {
 	struct pipe_texture *pt = &nv20mt->base;
-	boolean swizzled = FALSE;
 	uint width = pt->width[0], height = pt->height[0];
 	uint offset = 0;
 	int nr_faces, l, f;
+	uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER |
+		                           PIPE_TEXTURE_USAGE_DEPTH_STENCIL |
+		                           PIPE_TEXTURE_USAGE_RENDER_TARGET |
+		                           PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
+		                           PIPE_TEXTURE_USAGE_PRIMARY);
 
 	if (pt->target == PIPE_TEXTURE_CUBE) {
 		nr_faces = 6;
@@ -26,25 +30,31 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt)
 		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
 		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
 
-		if (swizzled)
-			nv20mt->level[l].pitch = pt->nblocksx[l] * pt->block.size;
+		if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
+			nv20mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64);
 		else
-			nv20mt->level[l].pitch = pt->nblocksx[0] * pt->block.size;
-		nv20mt->level[l].pitch = (nv20mt->level[l].pitch + 63) & ~63;
+			nv20mt->level[l].pitch = pt->width[l] * pt->block.size;
 
 		nv20mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
 		width  = MAX2(1, width  >> 1);
 		height = MAX2(1, height >> 1);
-
 	}
 
 	for (f = 0; f < nr_faces; f++) {
-		for (l = 0; l <= pt->last_level; l++) {
+		for (l = 0; l < pt->last_level; l++) {
 			nv20mt->level[l].image_offset[f] = offset;
-			offset += nv20mt->level[l].pitch * pt->height[l];
+
+			if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) &&
+			    pt->width[l + 1] > 1 && pt->height[l + 1] > 1)
+				offset += align(nv20mt->level[l].pitch * pt->height[l], 64);
+			else
+				offset += nv20mt->level[l].pitch * pt->height[l];
 		}
+
+		nv20mt->level[l].image_offset[f] = offset;
+		offset += nv20mt->level[l].pitch * pt->height[l];
 	}
 
 	nv20mt->total_size = offset;
@@ -96,7 +106,8 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
-	                     PIPE_TEXTURE_USAGE_DISPLAY_TARGET))
+	                     PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
+	                     PIPE_TEXTURE_USAGE_DEPTH_STENCIL))
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC)
@@ -107,7 +118,11 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 		case PIPE_FORMAT_A8R8G8B8_UNORM:
 		case PIPE_FORMAT_X8R8G8B8_UNORM:
 		case PIPE_FORMAT_R16_SNORM:
+		{
+			if (debug_get_bool_option("NOUVEAU_NO_SWIZZLE", FALSE))
+				mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 			break;
+		}
 		default:
 			mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 		}
@@ -152,33 +167,33 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 			 unsigned flags)
 {
 	struct nv20_miptree *nv20mt = (struct nv20_miptree *)pt;
-	struct pipe_surface *ps;
+	struct nv04_surface *ns;
 
-	ps = CALLOC_STRUCT(pipe_surface);
-	if (!ps)
+	ns = CALLOC_STRUCT(nv04_surface);
+	if (!ns)
 		return NULL;
-	pipe_texture_reference(&ps->texture, pt);
-	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
-	ps->block = pt->block;
-	ps->nblocksx = pt->nblocksx[level];
-	ps->nblocksy = pt->nblocksy[level];
-	ps->stride = nv20mt->level[level].pitch;
-	ps->usage = flags;
-	ps->status = PIPE_SURFACE_STATUS_DEFINED;
-	ps->refcount = 1;
+	pipe_texture_reference(&ns->base.texture, pt);
+	ns->base.format = pt->format;
+	ns->base.width = pt->width[level];
+	ns->base.height = pt->height[level];
+	ns->base.usage = flags;
+	ns->base.status = PIPE_SURFACE_STATUS_DEFINED;
+	ns->base.refcount = 1;
+	ns->base.face = face;
+	ns->base.level = level;
+	ns->base.zslice = zslice;
+	ns->pitch = nv20mt->level[level].pitch;
 
 	if (pt->target == PIPE_TEXTURE_CUBE) {
-		ps->offset = nv20mt->level[level].image_offset[face];
+		ns->base.offset = nv20mt->level[level].image_offset[face];
 	} else
 	if (pt->target == PIPE_TEXTURE_3D) {
-		ps->offset = nv20mt->level[level].image_offset[zslice];
+		ns->base.offset = nv20mt->level[level].image_offset[zslice];
 	} else {
-		ps->offset = nv20mt->level[level].image_offset[0];
+		ns->base.offset = nv20mt->level[level].image_offset[0];
 	}
 
-	return ps;
+	return &ns->base;
 }
 
 static void
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
index 5f2b7b4f71..7760ae27c0 100644
--- a/src/gallium/drivers/nv20/nv20_screen.c
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -116,30 +116,6 @@ nv20_screen_is_format_supported(struct pipe_screen *screen,
 	return FALSE;
 }
 
-static void *
-nv20_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
-		 unsigned flags )
-{
-	struct pipe_winsys *ws = screen->winsys;
-	void *map;
-	struct nv20_miptree *nv20mt = (struct nv20_miptree *)surface->texture;
-
-	map = ws->buffer_map(ws, nv20mt->buffer, flags);
-	if (!map)
-		return NULL;
-
-	return map + surface->offset;
-}
-
-static void
-nv20_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
-{
-	struct pipe_winsys *ws = screen->winsys;
-	struct nv20_miptree *nv20mt = (struct nv20_miptree *)surface->texture;
-
-	ws->buffer_unmap(ws, nv20mt->buffer);
-}
-
 static void
 nv20_screen_destroy(struct pipe_screen *pscreen)
 {
@@ -211,10 +187,8 @@ nv20_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 
 	screen->pipe.is_format_supported = nv20_screen_is_format_supported;
 
-	screen->pipe.surface_map = nv20_surface_map;
-	screen->pipe.surface_unmap = nv20_surface_unmap;
-
 	nv20_screen_init_miptree_functions(&screen->pipe);
+	nv20_screen_init_transfer_functions(&screen->pipe);
 	u_simple_screen_init(&screen->pipe);
 
 	return &screen->pipe;
diff --git a/src/gallium/drivers/nv20/nv20_screen.h b/src/gallium/drivers/nv20/nv20_screen.h
index bf2f2c0d9f..d9fce2bced 100644
--- a/src/gallium/drivers/nv20/nv20_screen.h
+++ b/src/gallium/drivers/nv20/nv20_screen.h
@@ -21,4 +21,8 @@ nv20_screen(struct pipe_screen *screen)
 	return (struct nv20_screen *)screen;
 }
 
+
+void
+nv20_screen_init_transfer_functions(struct pipe_screen *pscreen);
+
 #endif
diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c
index 0f4df9ca31..4042f46d05 100644
--- a/src/gallium/drivers/nv20/nv20_state_emit.c
+++ b/src/gallium/drivers/nv20/nv20_state_emit.c
@@ -109,7 +109,7 @@ static void nv20_state_emit_scissor(struct nv20_context* nv20)
 static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 {
 	struct pipe_framebuffer_state* fb = nv20->framebuffer;
-	struct pipe_surface *rt, *zeta = NULL;
+	struct nv04_surface *rt, *zeta = NULL;
 	uint32_t rt_format, w, h;
 	int colour_format = 0, zeta_format = 0;
 	struct nv20_miptree *nv20mt = 0;
@@ -117,7 +117,7 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 	w = fb->cbufs[0]->width;
 	h = fb->cbufs[0]->height;
 	colour_format = fb->cbufs[0]->format;
-	rt = fb->cbufs[0];
+	rt = (struct nv04_surface *)fb->cbufs[0];
 
 	if (fb->zsbuf) {
 		if (colour_format) {
@@ -129,7 +129,7 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 		}
 
 		zeta_format = fb->zsbuf->format;
-		zeta = fb->zsbuf;
+		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
 	rt_format = NV20TCL_RT_FORMAT_TYPE_LINEAR | 0x20;
@@ -148,18 +148,18 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 
 	if (zeta) {
 		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
-		OUT_RING  (rt->stride | (zeta->stride << 16));
+		OUT_RING  (rt->pitch | (zeta->pitch << 16));
 	} else {
 		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
-		OUT_RING  (rt->stride | (rt->stride << 16));
+		OUT_RING  (rt->pitch | (rt->pitch << 16));
 	}
 
-	nv20mt = (struct nv20_miptree *)rt->texture;
+	nv20mt = (struct nv20_miptree *)rt->base.texture;
 	nv20->rt[0] = nv20mt->buffer;
 
 	if (zeta_format)
 	{
-		nv20mt = (struct nv20_miptree *)zeta->texture;
+		nv20mt = (struct nv20_miptree *)zeta->base.texture;
 		nv20->zeta = nv20mt->buffer;
 	}
 
diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c
new file mode 100644
index 0000000000..19de09486d
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_transfer.c
@@ -0,0 +1,201 @@
+#include <pipe/p_state.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <nouveau/nouveau_winsys.h>
+#include "nv20_context.h"
+#include "nv20_screen.h"
+#include "nv20_state.h"
+
+struct nv20_transfer {
+	struct pipe_transfer base;
+	struct pipe_surface *surface;
+	bool direct;
+};
+
+static unsigned nv20_usage_tx_to_buf(unsigned tx_usage)
+{
+	switch (tx_usage) {
+		case PIPE_TRANSFER_READ:
+			return PIPE_BUFFER_USAGE_CPU_READ;
+		case PIPE_TRANSFER_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_WRITE;
+		case PIPE_TRANSFER_READ_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static void
+nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+                             struct pipe_texture *template)
+{
+	memset(template, 0, sizeof(struct pipe_texture));
+	template->target = pt->target;
+	template->format = pt->format;
+	template->width[0] = pt->width[level];
+	template->height[0] = pt->height[level];
+	template->depth[0] = 1;
+	template->block = pt->block;
+	template->nblocksx[0] = pt->nblocksx[level];
+	template->nblocksy[0] = pt->nblocksx[level];
+	template->last_level = 0;
+	template->compressed = pt->compressed;
+	template->nr_samples = pt->nr_samples;
+
+	template->tex_usage = PIPE_TEXTURE_USAGE_DYNAMIC |
+	                      NOUVEAU_TEXTURE_USAGE_LINEAR;
+}
+
+static struct pipe_transfer *
+nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+		  unsigned face, unsigned level, unsigned zslice,
+		  enum pipe_transfer_usage usage,
+		  unsigned x, unsigned y, unsigned w, unsigned h)
+{
+	struct nv20_miptree *mt = (struct nv20_miptree *)pt;
+	struct nv20_transfer *tx;
+	struct pipe_texture tx_tex_template, *tx_tex;
+
+	tx = CALLOC_STRUCT(nv20_transfer);
+	if (!tx)
+		return NULL;
+
+	tx->base.refcount = 1;
+	pipe_texture_reference(&tx->base.texture, pt);
+	tx->base.format = pt->format;
+	tx->base.x = x;
+	tx->base.y = y;
+	tx->base.width = w;
+	tx->base.height = h;
+	tx->base.block = pt->block;
+	tx->base.nblocksx = pt->nblocksx[level];
+	tx->base.nblocksy = pt->nblocksy[level];
+	tx->base.stride = mt->level[level].pitch;
+	tx->base.usage = usage;
+	tx->base.face = face;
+	tx->base.level = level;
+	tx->base.zslice = zslice;
+
+	/* Direct access to texture */
+	if ((pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC ||
+	     debug_get_bool_option("NOUVEAU_NO_TRANSFER", TRUE/*XXX:FALSE*/)) &&
+	    pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)
+	{
+		tx->direct = true;
+		tx->surface = pscreen->get_tex_surface(pscreen, pt,
+	                                               face, level, zslice,
+	                                               nv20_usage_tx_to_buf(usage));
+		return &tx->base;
+	}
+
+	tx->direct = false;
+
+	nv20_compatible_transfer_tex(pt, level, &tx_tex_template);
+
+	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
+	if (!tx_tex)
+	{
+		FREE(tx);
+		return NULL;
+	}
+
+	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
+	                                       face, level, zslice,
+	                                       nv20_usage_tx_to_buf(usage));
+
+	pipe_texture_reference(&tx_tex, NULL);
+
+	if (!tx->surface)
+	{
+		pipe_surface_reference(&tx->surface, NULL);
+		FREE(tx);
+		return NULL;
+	}
+
+	if (usage != PIPE_TRANSFER_WRITE) {
+		struct nv20_screen *nvscreen = nv20_screen(pscreen);
+		struct pipe_surface *src;
+
+		src = pscreen->get_tex_surface(pscreen, pt,
+	                                       face, level, zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_READ);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		/* TODO: Check if SIFM can un-swizzle */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      tx->surface, 0, 0,
+		                      src, 0, 0,
+		                      src->width, src->height);
+
+		pipe_surface_reference(&src, NULL);
+	}
+
+	return &tx->base;
+}
+
+static void
+nv20_transfer_del(struct pipe_screen *pscreen, struct pipe_transfer **pptx)
+{
+	struct pipe_transfer *ptx = *pptx;
+	struct nv20_transfer *tx = (struct nv20_transfer *)ptx;
+
+	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+		struct nv20_screen *nvscreen = nv20_screen(pscreen);
+		struct pipe_surface *dst;
+
+		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
+	                                       ptx->face, ptx->level, ptx->zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      dst, 0, 0,
+		                      tx->surface, 0, 0,
+		                      dst->width, dst->height);
+
+		pipe_surface_reference(&dst, NULL);
+	}
+
+	*pptx = NULL;
+	if (--ptx->refcount)
+		return;
+
+	pipe_surface_reference(&tx->surface, NULL);
+	pipe_texture_reference(&ptx->texture, NULL);
+	FREE(ptx);
+}
+
+static void *
+nv20_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv20_transfer *tx = (struct nv20_transfer *)ptx;
+	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
+	struct nv20_miptree *mt = (struct nv20_miptree *)tx->surface->texture;
+	void *map = pipe_buffer_map(pscreen, mt->buffer,
+	                            nv20_usage_tx_to_buf(ptx->usage));
+
+	return map + ns->base.offset +
+	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+}
+
+static void
+nv20_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv20_transfer *tx = (struct nv20_transfer *)ptx;
+	struct nv20_miptree *mt = (struct nv20_miptree *)tx->surface->texture;
+
+	pipe_buffer_unmap(pscreen, mt->buffer);
+}
+
+void
+nv20_screen_init_transfer_functions(struct pipe_screen *pscreen)
+{
+	pscreen->get_tex_transfer = nv20_transfer_new;
+	pscreen->tex_transfer_release = nv20_transfer_del;
+	pscreen->transfer_map = nv20_transfer_map;
+	pscreen->transfer_unmap = nv20_transfer_unmap;
+}
diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile
index 4c29e2eab3..364c80d8f3 100644
--- a/src/gallium/drivers/nv30/Makefile
+++ b/src/gallium/drivers/nv30/Makefile
@@ -22,6 +22,7 @@ C_SOURCES = \
 	nv30_state_viewport.c \
 	nv30_state_zsa.c \
 	nv30_surface.c \
+	nv30_transfer.c \
 	nv30_vbo.c \
 	nv30_vertprog.c
 
diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c
index b11ed8c24e..ec0a8b8438 100644
--- a/src/gallium/drivers/nv30/nv30_miptree.c
+++ b/src/gallium/drivers/nv30/nv30_miptree.c
@@ -69,6 +69,8 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 {
 	struct pipe_winsys *ws = pscreen->winsys;
 	struct nv30_miptree *mt;
+	unsigned buf_usage = PIPE_BUFFER_USAGE_PIXEL |
+	                     NOUVEAU_BUFFER_USAGE_TEXTURE;
 
 	mt = MALLOC(sizeof(struct nv30_miptree));
 	if (!mt)
@@ -76,8 +78,6 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 	mt->base = *pt;
 	mt->base.refcount = 1;
 	mt->base.screen = pscreen;
-	mt->shadow_tex = NULL;
-	mt->shadow_surface = NULL;
 
 	/* Swizzled textures must be POT */
 	if (pt->width[0] & (pt->width[0] - 1) ||
@@ -107,11 +107,12 @@ nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 		}
 	}
 
+	if (pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC)
+		buf_usage |= PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+
 	nv30_miptree_layout(mt);
 
-	mt->buffer = ws->buffer_create(ws, 256,
-				       PIPE_BUFFER_USAGE_PIXEL |
-				       NOUVEAU_BUFFER_USAGE_TEXTURE,
+	mt->buffer = ws->buffer_create(ws, 256, buf_usage,
 				       mt->total_size);
 	if (!mt->buffer) {
 		FREE(mt);
@@ -163,12 +164,6 @@ nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
 			FREE(mt->level[l].image_offset);
 	}
 
-	if (mt->shadow_tex) {
-		if (mt->shadow_surface)
-			pscreen->tex_surface_release(pscreen, &mt->shadow_surface);
-		nv30_miptree_release(pscreen, &mt->shadow_tex);
-	}
-
 	FREE(mt);
 }
 
@@ -178,36 +173,33 @@ nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 			 unsigned flags)
 {
 	struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt;
-	struct pipe_surface *ps;
+	struct nv04_surface *ns;
 
-	ps = CALLOC_STRUCT(pipe_surface);
-	if (!ps)
+	ns = CALLOC_STRUCT(nv04_surface);
+	if (!ns)
 		return NULL;
-	pipe_texture_reference(&ps->texture, pt);
-	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
-	ps->block = pt->block;
-	ps->nblocksx = pt->nblocksx[level];
-	ps->nblocksy = pt->nblocksy[level];
-	ps->stride = nv30mt->level[level].pitch;
-	ps->usage = flags;
-	ps->status = PIPE_SURFACE_STATUS_DEFINED;
-	ps->refcount = 1;
-	ps->face = face;
-	ps->level = level;
-	ps->zslice = zslice;
+	pipe_texture_reference(&ns->base.texture, pt);
+	ns->base.format = pt->format;
+	ns->base.width = pt->width[level];
+	ns->base.height = pt->height[level];
+	ns->base.usage = flags;
+	ns->base.status = PIPE_SURFACE_STATUS_DEFINED;
+	ns->base.refcount = 1;
+	ns->base.face = face;
+	ns->base.level = level;
+	ns->base.zslice = zslice;
+	ns->pitch = nv30mt->level[level].pitch;
 
 	if (pt->target == PIPE_TEXTURE_CUBE) {
-		ps->offset = nv30mt->level[level].image_offset[face];
+		ns->base.offset = nv30mt->level[level].image_offset[face];
 	} else
 	if (pt->target == PIPE_TEXTURE_3D) {
-		ps->offset = nv30mt->level[level].image_offset[zslice];
+		ns->base.offset = nv30mt->level[level].image_offset[zslice];
 	} else {
-		ps->offset = nv30mt->level[level].image_offset[0];
+		ns->base.offset = nv30mt->level[level].image_offset[0];
 	}
 
-	return ps;
+	return &ns->base;
 }
 
 static void
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
index c97a73f0b1..d395c5e1b7 100644
--- a/src/gallium/drivers/nv30/nv30_screen.c
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -135,82 +135,6 @@ nv30_surface_buffer(struct pipe_surface *surf)
 	return mt->buffer;
 }
 
-static void *
-nv30_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
-		 unsigned flags )
-{
-	struct pipe_winsys	*ws = screen->winsys;
-	struct pipe_surface	*surface_to_map;
-	void			*map;
-
-	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
-		struct nv30_miptree *mt = (struct nv30_miptree *)surface->texture;
-
-		if (!mt->shadow_tex) {
-			unsigned old_tex_usage = surface->texture->tex_usage;
-			surface->texture->tex_usage = NOUVEAU_TEXTURE_USAGE_LINEAR |
-			                              PIPE_TEXTURE_USAGE_DYNAMIC;
-			mt->shadow_tex = screen->texture_create(screen, surface->texture);
-			surface->texture->tex_usage = old_tex_usage;
-
-			assert(mt->shadow_tex->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR);
-		}
-
-		mt->shadow_surface = screen->get_tex_surface
-		(
-			screen, mt->shadow_tex,
-			surface->face, surface->level, surface->zslice,
-			surface->usage
-		);
-
-		surface_to_map = mt->shadow_surface;
-	}
-	else
-		surface_to_map = surface;
-
-	assert(surface_to_map);
-
-	map = ws->buffer_map(ws, nv30_surface_buffer(surface_to_map), flags);
-	if (!map)
-		return NULL;
-
-	return map + surface_to_map->offset;
-}
-
-static void
-nv30_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
-{
-	struct pipe_winsys	*ws = screen->winsys;
-	struct pipe_surface	*surface_to_unmap;
-
-	/* TODO: Copy from shadow just before push buffer is flushed instead.
-	         There are probably some programs that map/unmap excessively
-	         before rendering. */
-	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
-		struct nv30_miptree *mt = (struct nv30_miptree *)surface->texture;
-
-		assert(mt->shadow_tex);
-
-		surface_to_unmap = mt->shadow_surface;
-	}
-	else
-		surface_to_unmap = surface;
-
-	assert(surface_to_unmap);
-
-	ws->buffer_unmap(ws, nv30_surface_buffer(surface_to_unmap));
-
-	if (surface_to_unmap != surface) {
-		struct nv30_screen *nvscreen = nv30_screen(screen);
-
-		nvscreen->eng2d->copy(nvscreen->eng2d, surface, 0, 0,
-		                      surface_to_unmap, 0, 0,
-		                      surface->width, surface->height);
-
-		screen->tex_surface_release(screen, &surface_to_unmap);
-	}
-}
-
 static void
 nv30_screen_destroy(struct pipe_screen *pscreen)
 {
@@ -391,10 +315,8 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 
 	screen->pipe.is_format_supported = nv30_screen_surface_format_supported;
 
-	screen->pipe.surface_map = nv30_surface_map;
-	screen->pipe.surface_unmap = nv30_surface_unmap;
-
 	nv30_screen_init_miptree_functions(&screen->pipe);
+	nv30_screen_init_transfer_functions(&screen->pipe);
 	u_simple_screen_init(&screen->pipe);
 
 	return &screen->pipe;
diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h
index b11e470f94..8e36883975 100644
--- a/src/gallium/drivers/nv30/nv30_screen.h
+++ b/src/gallium/drivers/nv30/nv30_screen.h
@@ -34,4 +34,7 @@ nv30_screen(struct pipe_screen *screen)
 	return (struct nv30_screen *)screen;
 }
 
+void
+nv30_screen_init_transfer_functions(struct pipe_screen *pscreen);
+
 #endif
diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h
index 2023278e37..e6f23bf166 100644
--- a/src/gallium/drivers/nv30/nv30_state.h
+++ b/src/gallium/drivers/nv30/nv30_state.h
@@ -76,9 +76,6 @@ struct nv30_miptree {
 	struct pipe_buffer *buffer;
 	uint total_size;
 
-	struct pipe_texture *shadow_tex;
-	struct pipe_surface *shadow_surface;
-
 	struct {
 		uint pitch;
 		uint *image_offset;
diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c
index 77368cb205..ee0326011c 100644
--- a/src/gallium/drivers/nv30/nv30_state_fb.c
+++ b/src/gallium/drivers/nv30/nv30_state_fb.c
@@ -5,7 +5,7 @@ static boolean
 nv30_state_framebuffer_validate(struct nv30_context *nv30)
 {
 	struct pipe_framebuffer_state *fb = &nv30->framebuffer;
-	struct pipe_surface *rt[2], *zeta = NULL;
+	struct nv04_surface *rt[2], *zeta = NULL;
 	uint32_t rt_enable, rt_format;
 	int i, colour_format = 0, zeta_format = 0;
 	struct nouveau_stateobj *so = so_new(64, 10);
@@ -21,7 +21,7 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 		} else {
 			colour_format = fb->cbufs[i]->format;
 			rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i);
-			rt[i] = fb->cbufs[i];
+			rt[i] = (struct nv04_surface *)fb->cbufs[i];
 		}
 	}
 
@@ -30,13 +30,13 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 
 	if (fb->zsbuf) {
 		zeta_format = fb->zsbuf->format;
-		zeta = fb->zsbuf;
+		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
-	if (!(rt[0]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+	if (!(rt[0]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
 		assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
 		for (i = 1; i < fb->nr_cbufs; i++)
-			assert(!(rt[i]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
+			assert(!(rt[i]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
 
 		/* FIXME: NV34TCL_RT_FORMAT_LOG2_[WIDTH/HEIGHT] */
 		rt_format = NV34TCL_RT_FORMAT_TYPE_SWIZZLED |
@@ -71,44 +71,44 @@ nv30_state_framebuffer_validate(struct nv30_context *nv30)
 	}
 
 	if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) {
-		uint32_t pitch = rt[0]->stride;
+		uint32_t pitch = rt[0]->pitch;
 		if (zeta) {
-			pitch |= (zeta->stride << 16);
+			pitch |= (zeta->pitch << 16);
 		} else {
 			pitch |= (pitch << 16);
 		}
 
-		nv30mt = (struct nv30_miptree *)rt[0]->texture;
+		nv30mt = (struct nv30_miptree *)rt[0]->base.texture;
 		so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR0, 1);
 		so_reloc (so, nv30mt->buffer, 0, rt_flags | NOUVEAU_BO_OR,
 			  nv30->nvws->channel->vram->handle,
 			  nv30->nvws->channel->gart->handle);
 		so_method(so, nv30->screen->rankine, NV34TCL_COLOR0_PITCH, 2);
 		so_data  (so, pitch);
-		so_reloc (so, nv30mt->buffer, rt[0]->offset, rt_flags |
+		so_reloc (so, nv30mt->buffer, rt[0]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 	}
 
 	if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) {
-		nv30mt = (struct nv30_miptree *)rt[1]->texture;
+		nv30mt = (struct nv30_miptree *)rt[1]->base.texture;
 		so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR1, 1);
 		so_reloc (so, nv30mt->buffer, 0, rt_flags | NOUVEAU_BO_OR,
 			  nv30->nvws->channel->vram->handle,
 			  nv30->nvws->channel->gart->handle);
 		so_method(so, nv30->screen->rankine, NV34TCL_COLOR1_OFFSET, 2);
-		so_reloc (so, nv30mt->buffer, rt[1]->offset, rt_flags |
+		so_reloc (so, nv30mt->buffer, rt[1]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
-		so_data  (so, rt[1]->stride);
+		so_data  (so, rt[1]->pitch);
 	}
 
 	if (zeta_format) {
-		nv30mt = (struct nv30_miptree *)zeta->texture;
+		nv30mt = (struct nv30_miptree *)zeta->base.texture;
 		so_method(so, nv30->screen->rankine, NV34TCL_DMA_ZETA, 1);
 		so_reloc (so, nv30mt->buffer, 0, rt_flags | NOUVEAU_BO_OR,
 			  nv30->nvws->channel->vram->handle,
 			  nv30->nvws->channel->gart->handle);
 		so_method(so, nv30->screen->rankine, NV34TCL_ZETA_OFFSET, 1);
-		so_reloc (so, nv30mt->buffer, zeta->offset, rt_flags |
+		so_reloc (so, nv30mt->buffer, zeta->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 		/* TODO: allocate LMA depth buffer */
 	}
diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c
new file mode 100644
index 0000000000..df4dc4b1f6
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_transfer.c
@@ -0,0 +1,201 @@
+#include <pipe/p_state.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <nouveau/nouveau_winsys.h>
+#include "nv30_context.h"
+#include "nv30_screen.h"
+#include "nv30_state.h"
+
+struct nv30_transfer {
+	struct pipe_transfer base;
+	struct pipe_surface *surface;
+	bool direct;
+};
+
+static unsigned nv30_usage_tx_to_buf(unsigned tx_usage)
+{
+	switch (tx_usage) {
+		case PIPE_TRANSFER_READ:
+			return PIPE_BUFFER_USAGE_CPU_READ;
+		case PIPE_TRANSFER_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_WRITE;
+		case PIPE_TRANSFER_READ_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static void
+nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+                             struct pipe_texture *template)
+{
+	memset(template, 0, sizeof(struct pipe_texture));
+	template->target = pt->target;
+	template->format = pt->format;
+	template->width[0] = pt->width[level];
+	template->height[0] = pt->height[level];
+	template->depth[0] = 1;
+	template->block = pt->block;
+	template->nblocksx[0] = pt->nblocksx[level];
+	template->nblocksy[0] = pt->nblocksx[level];
+	template->last_level = 0;
+	template->compressed = pt->compressed;
+	template->nr_samples = pt->nr_samples;
+
+	template->tex_usage = PIPE_TEXTURE_USAGE_DYNAMIC |
+	                      NOUVEAU_TEXTURE_USAGE_LINEAR;
+}
+
+static struct pipe_transfer *
+nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+		  unsigned face, unsigned level, unsigned zslice,
+		  enum pipe_transfer_usage usage,
+		  unsigned x, unsigned y, unsigned w, unsigned h)
+{
+	struct nv30_miptree *mt = (struct nv30_miptree *)pt;
+	struct nv30_transfer *tx;
+	struct pipe_texture tx_tex_template, *tx_tex;
+
+	tx = CALLOC_STRUCT(nv30_transfer);
+	if (!tx)
+		return NULL;
+
+	tx->base.refcount = 1;
+	pipe_texture_reference(&tx->base.texture, pt);
+	tx->base.format = pt->format;
+	tx->base.x = x;
+	tx->base.y = y;
+	tx->base.width = w;
+	tx->base.height = h;
+	tx->base.block = pt->block;
+	tx->base.nblocksx = pt->nblocksx[level];
+	tx->base.nblocksy = pt->nblocksy[level];
+	tx->base.stride = mt->level[level].pitch;
+	tx->base.usage = usage;
+	tx->base.face = face;
+	tx->base.level = level;
+	tx->base.zslice = zslice;
+
+	/* Direct access to texture */
+	if ((pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC ||
+	     debug_get_bool_option("NOUVEAU_NO_TRANSFER", TRUE/*XXX:FALSE*/)) &&
+	    pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)
+	{
+		tx->direct = true;
+		tx->surface = pscreen->get_tex_surface(pscreen, pt,
+	                                               face, level, zslice,
+	                                               nv30_usage_tx_to_buf(usage));
+		return &tx->base;
+	}
+
+	tx->direct = false;
+
+	nv30_compatible_transfer_tex(pt, level, &tx_tex_template);
+
+	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
+	if (!tx_tex)
+	{
+		FREE(tx);
+		return NULL;
+	}
+
+	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
+	                                       face, level, zslice,
+	                                       nv30_usage_tx_to_buf(usage));
+
+	pipe_texture_reference(&tx_tex, NULL);
+
+	if (!tx->surface)
+	{
+		pipe_surface_reference(&tx->surface, NULL);
+		FREE(tx);
+		return NULL;
+	}
+
+	if (usage != PIPE_TRANSFER_WRITE) {
+		struct nv30_screen *nvscreen = nv30_screen(pscreen);
+		struct pipe_surface *src;
+
+		src = pscreen->get_tex_surface(pscreen, pt,
+	                                       face, level, zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_READ);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		/* TODO: Check if SIFM can un-swizzle */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      tx->surface, 0, 0,
+		                      src, 0, 0,
+		                      src->width, src->height);
+
+		pipe_surface_reference(&src, NULL);
+	}
+
+	return &tx->base;
+}
+
+static void
+nv30_transfer_del(struct pipe_screen *pscreen, struct pipe_transfer **pptx)
+{
+	struct pipe_transfer *ptx = *pptx;
+	struct nv30_transfer *tx = (struct nv30_transfer *)ptx;
+
+	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+		struct nv30_screen *nvscreen = nv30_screen(pscreen);
+		struct pipe_surface *dst;
+
+		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
+	                                       ptx->face, ptx->level, ptx->zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      dst, 0, 0,
+		                      tx->surface, 0, 0,
+		                      dst->width, dst->height);
+
+		pipe_surface_reference(&dst, NULL);
+	}
+
+	*pptx = NULL;
+	if (--ptx->refcount)
+		return;
+
+	pipe_surface_reference(&tx->surface, NULL);
+	pipe_texture_reference(&ptx->texture, NULL);
+	FREE(ptx);
+}
+
+static void *
+nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv30_transfer *tx = (struct nv30_transfer *)ptx;
+	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
+	struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture;
+	void *map = pipe_buffer_map(pscreen, mt->buffer,
+	                            nv30_usage_tx_to_buf(ptx->usage));
+
+	return map + ns->base.offset +
+	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+}
+
+static void
+nv30_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv30_transfer *tx = (struct nv30_transfer *)ptx;
+	struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture;
+
+	pipe_buffer_unmap(pscreen, mt->buffer);
+}
+
+void
+nv30_screen_init_transfer_functions(struct pipe_screen *pscreen)
+{
+	pscreen->get_tex_transfer = nv30_transfer_new;
+	pscreen->tex_transfer_release = nv30_transfer_del;
+	pscreen->transfer_map = nv30_transfer_map;
+	pscreen->transfer_unmap = nv30_transfer_unmap;
+}
diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile
index 8c738aefa6..0ecae2b491 100644
--- a/src/gallium/drivers/nv40/Makefile
+++ b/src/gallium/drivers/nv40/Makefile
@@ -22,6 +22,7 @@ C_SOURCES = \
 	nv40_state_viewport.c \
 	nv40_state_zsa.c \
 	nv40_surface.c \
+	nv40_transfer.c \
 	nv40_vbo.c \
 	nv40_vertprog.c
 
diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c
index e38b1e7f5c..638d279aa5 100644
--- a/src/gallium/drivers/nv40/nv40_miptree.c
+++ b/src/gallium/drivers/nv40/nv40_miptree.c
@@ -78,8 +78,6 @@ nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 	mt->base = *pt;
 	mt->base.refcount = 1;
 	mt->base.screen = pscreen;
-	mt->shadow_tex = NULL;
-	mt->shadow_surface = NULL;
 
 	/* Swizzled textures must be POT */
 	if (pt->width[0] & (pt->width[0] - 1) ||
@@ -165,12 +163,6 @@ nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
 			FREE(mt->level[l].image_offset);
 	}
 
-	if (mt->shadow_tex) {
-		if (mt->shadow_surface)
-			pscreen->tex_surface_release(pscreen, &mt->shadow_surface);
-		nv40_miptree_release(pscreen, &mt->shadow_tex);
-	}
-
 	FREE(mt);
 }
 
@@ -180,36 +172,33 @@ nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 			 unsigned flags)
 {
 	struct nv40_miptree *mt = (struct nv40_miptree *)pt;
-	struct pipe_surface *ps;
+	struct nv04_surface *ns;
 
-	ps = CALLOC_STRUCT(pipe_surface);
-	if (!ps)
+	ns = CALLOC_STRUCT(nv04_surface);
+	if (!ns)
 		return NULL;
-	pipe_texture_reference(&ps->texture, pt);
-	ps->format = pt->format;
-	ps->width = pt->width[level];
-	ps->height = pt->height[level];
-	ps->block = pt->block;
-	ps->nblocksx = pt->nblocksx[level];
-	ps->nblocksy = pt->nblocksy[level];
-	ps->stride = mt->level[level].pitch;
-	ps->usage = flags;
-	ps->status = PIPE_SURFACE_STATUS_DEFINED;
-	ps->refcount = 1;
-	ps->face = face;
-	ps->level = level;
-	ps->zslice = zslice;
+	pipe_texture_reference(&ns->base.texture, pt);
+	ns->base.format = pt->format;
+	ns->base.width = pt->width[level];
+	ns->base.height = pt->height[level];
+	ns->base.usage = flags;
+	ns->base.status = PIPE_SURFACE_STATUS_DEFINED;
+	ns->base.refcount = 1;
+	ns->base.face = face;
+	ns->base.level = level;
+	ns->base.zslice = zslice;
+	ns->pitch = mt->level[level].pitch;
 
 	if (pt->target == PIPE_TEXTURE_CUBE) {
-		ps->offset = mt->level[level].image_offset[face];
+		ns->base.offset = mt->level[level].image_offset[face];
 	} else
 	if (pt->target == PIPE_TEXTURE_3D) {
-		ps->offset = mt->level[level].image_offset[zslice];
+		ns->base.offset = mt->level[level].image_offset[zslice];
 	} else {
-		ps->offset = mt->level[level].image_offset[0];
+		ns->base.offset = mt->level[level].image_offset[0];
 	}
 
-	return ps;
+	return &ns->base;
 }
 
 static void
diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c
index 2372bc8441..0d4baefaea 100644
--- a/src/gallium/drivers/nv40/nv40_screen.c
+++ b/src/gallium/drivers/nv40/nv40_screen.c
@@ -144,81 +144,6 @@ nv40_surface_buffer(struct pipe_surface *surf)
 	return mt->buffer;
 }
 
-static void *
-nv40_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
-		 unsigned flags )
-{
-	struct pipe_winsys	*ws = screen->winsys;
-	struct pipe_surface	*surface_to_map;
-	void			*map;
-
-	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
-		struct nv40_miptree *mt = (struct nv40_miptree *)surface->texture;
-
-		if (!mt->shadow_tex) {
-			unsigned old_tex_usage = surface->texture->tex_usage;
-			surface->texture->tex_usage = NOUVEAU_TEXTURE_USAGE_LINEAR |
-			                              PIPE_TEXTURE_USAGE_DYNAMIC;
-			mt->shadow_tex = screen->texture_create(screen, surface->texture);
-			surface->texture->tex_usage = old_tex_usage;
-
-			assert(mt->shadow_tex->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR);
-		}
-
-		mt->shadow_surface = screen->get_tex_surface
-		(
-			screen, mt->shadow_tex,
-			surface->face, surface->level, surface->zslice,
-			surface->usage
-		);
-
-		surface_to_map = mt->shadow_surface;
-	}
-	else
-		surface_to_map = surface;
-
-	assert(surface_to_map);
-	map = ws->buffer_map(ws, nv40_surface_buffer(surface_to_map), flags);
-	if (!map)
-		return NULL;
-
-	return map + surface_to_map->offset;
-}
-
-static void
-nv40_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
-{
-	struct pipe_winsys	*ws = screen->winsys;
-	struct pipe_surface	*surface_to_unmap;
-
-	/* TODO: Copy from shadow just before push buffer is flushed instead.
-	         There are probably some programs that map/unmap excessively
-	         before rendering. */
-	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
-		struct nv40_miptree *mt = (struct nv40_miptree *)surface->texture;
-
-		assert(mt->shadow_tex);
-
-		surface_to_unmap = mt->shadow_surface;
-	}
-	else
-		surface_to_unmap = surface;
-
-	assert(surface_to_unmap);
-
-	ws->buffer_unmap(ws, nv40_surface_buffer(surface_to_unmap));
-
-	if (surface_to_unmap != surface) {
-		struct nv40_screen *nvscreen = nv40_screen(screen);
-
-		nvscreen->eng2d->copy(nvscreen->eng2d, surface, 0, 0,
-		                      surface_to_unmap, 0, 0,
-		                      surface->width, surface->height);
-
-		screen->tex_surface_release(screen, &surface_to_unmap);
-	}
-}
-
 static void
 nv40_screen_destroy(struct pipe_screen *pscreen)
 {
@@ -240,7 +165,7 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 {
 	struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen);
 	struct nouveau_stateobj *so;
-	unsigned curie_class;
+	unsigned curie_class = 0;
 	unsigned chipset = nvws->channel->device->chipset;
 	int ret;
 
@@ -265,8 +190,6 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 		if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f)))
 			curie_class = NV44TCL;
 		break;
-	default:
-		break;
 	}
 
 	if (!curie_class) {
@@ -372,10 +295,8 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
 
 	screen->pipe.is_format_supported = nv40_screen_surface_format_supported;
 
-	screen->pipe.surface_map = nv40_surface_map;
-	screen->pipe.surface_unmap = nv40_surface_unmap;
-
 	nv40_screen_init_miptree_functions(&screen->pipe);
+	nv40_screen_init_transfer_functions(&screen->pipe);
 	u_simple_screen_init(&screen->pipe);
 
 	return &screen->pipe;
diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h
index 4500aa0e5c..7b503bd207 100644
--- a/src/gallium/drivers/nv40/nv40_screen.h
+++ b/src/gallium/drivers/nv40/nv40_screen.h
@@ -34,4 +34,7 @@ nv40_screen(struct pipe_screen *screen)
 	return (struct nv40_screen *)screen;
 }
 
+void
+nv40_screen_init_transfer_functions(struct pipe_screen *pscreen);
+
 #endif
diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h
index 9c55903ae3..8a9d8c8fdf 100644
--- a/src/gallium/drivers/nv40/nv40_state.h
+++ b/src/gallium/drivers/nv40/nv40_state.h
@@ -79,9 +79,6 @@ struct nv40_miptree {
 	struct pipe_buffer *buffer;
 	uint total_size;
 
-	struct pipe_texture *shadow_tex;
-	struct pipe_surface *shadow_surface;
-
 	struct {
 		uint pitch;
 		uint *image_offset;
diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c
index 454abad31f..5ebd3a1a56 100644
--- a/src/gallium/drivers/nv40/nv40_state_fb.c
+++ b/src/gallium/drivers/nv40/nv40_state_fb.c
@@ -12,7 +12,7 @@ static boolean
 nv40_state_framebuffer_validate(struct nv40_context *nv40)
 {
 	struct pipe_framebuffer_state *fb = &nv40->framebuffer;
-	struct pipe_surface *rt[4], *zeta;
+	struct nv04_surface *rt[4], *zeta;
 	uint32_t rt_enable, rt_format;
 	int i, colour_format = 0, zeta_format = 0;
 	struct nouveau_stateobj *so = so_new(64, 10);
@@ -27,7 +27,7 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40)
 		} else {
 			colour_format = fb->cbufs[i]->format;
 			rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i);
-			rt[i] = fb->cbufs[i];
+			rt[i] = (struct nv04_surface *)fb->cbufs[i];
 		}
 	}
 
@@ -37,13 +37,13 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40)
 
 	if (fb->zsbuf) {
 		zeta_format = fb->zsbuf->format;
-		zeta = fb->zsbuf;
+		zeta = (struct nv04_surface *)fb->zsbuf;
 	}
 
-	if (!(rt[0]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+	if (!(rt[0]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
 		assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
 		for (i = 1; i < fb->nr_cbufs; i++)
-			assert(!(rt[i]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
+			assert(!(rt[i]->base.texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
 
 		rt_format = NV40TCL_RT_FORMAT_TYPE_SWIZZLED |
 		            log2i(fb->width) << NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT |
@@ -78,60 +78,60 @@ nv40_state_framebuffer_validate(struct nv40_context *nv40)
 
 	if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) {
 		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR0, 1);
-		so_reloc (so, nv40_surface_buffer(rt[0]), 0, rt_flags | NOUVEAU_BO_OR,
+		so_reloc (so, nv40_surface_buffer(&rt[0]->base), 0, rt_flags | NOUVEAU_BO_OR,
 			  nv40->nvws->channel->vram->handle,
 			  nv40->nvws->channel->gart->handle);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR0_PITCH, 2);
-		so_data  (so, rt[0]->stride);
-		so_reloc (so, nv40_surface_buffer(rt[0]), rt[0]->offset, rt_flags |
+		so_data  (so, rt[0]->pitch);
+		so_reloc (so, nv40_surface_buffer(&rt[0]->base), rt[0]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 	}
 
 	if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) {
 		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR1, 1);
-		so_reloc (so, nv40_surface_buffer(rt[1]), 0, rt_flags | NOUVEAU_BO_OR,
+		so_reloc (so, nv40_surface_buffer(&rt[1]->base), 0, rt_flags | NOUVEAU_BO_OR,
 			  nv40->nvws->channel->vram->handle,
 			  nv40->nvws->channel->gart->handle);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR1_OFFSET, 2);
-		so_reloc (so, nv40_surface_buffer(rt[1]), rt[1]->offset, rt_flags |
+		so_reloc (so, nv40_surface_buffer(&rt[1]->base), rt[1]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
-		so_data  (so, rt[1]->stride);
+		so_data  (so, rt[1]->pitch);
 	}
 
 	if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) {
 		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR2, 1);
-		so_reloc (so, nv40_surface_buffer(rt[2]), 0, rt_flags | NOUVEAU_BO_OR,
+		so_reloc (so, nv40_surface_buffer(&rt[2]->base), 0, rt_flags | NOUVEAU_BO_OR,
 			  nv40->nvws->channel->vram->handle,
 			  nv40->nvws->channel->gart->handle);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR2_OFFSET, 1);
-		so_reloc (so, nv40_surface_buffer(rt[2]), rt[2]->offset, rt_flags |
+		so_reloc (so, nv40_surface_buffer(&rt[2]->base), rt[2]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR2_PITCH, 1);
-		so_data  (so, rt[2]->stride);
+		so_data  (so, rt[2]->pitch);
 	}
 
 	if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) {
 		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR3, 1);
-		so_reloc (so, nv40_surface_buffer(rt[3]), 0, rt_flags | NOUVEAU_BO_OR,
+		so_reloc (so, nv40_surface_buffer(&rt[3]->base), 0, rt_flags | NOUVEAU_BO_OR,
 			  nv40->nvws->channel->vram->handle,
 			  nv40->nvws->channel->gart->handle);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR3_OFFSET, 1);
-		so_reloc (so, nv40_surface_buffer(rt[3]), rt[3]->offset, rt_flags |
+		so_reloc (so, nv40_surface_buffer(&rt[3]->base), rt[3]->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 		so_method(so, nv40->screen->curie, NV40TCL_COLOR3_PITCH, 1);
-		so_data  (so, rt[3]->stride);
+		so_data  (so, rt[3]->pitch);
 	}
 
 	if (zeta_format) {
 		so_method(so, nv40->screen->curie, NV40TCL_DMA_ZETA, 1);
-		so_reloc (so, nv40_surface_buffer(zeta), 0, rt_flags | NOUVEAU_BO_OR,
+		so_reloc (so, nv40_surface_buffer(&zeta->base), 0, rt_flags | NOUVEAU_BO_OR,
 			  nv40->nvws->channel->vram->handle,
 			  nv40->nvws->channel->gart->handle);
 		so_method(so, nv40->screen->curie, NV40TCL_ZETA_OFFSET, 1);
-		so_reloc (so, nv40_surface_buffer(zeta), zeta->offset, rt_flags |
+		so_reloc (so, nv40_surface_buffer(&zeta->base), zeta->base.offset, rt_flags |
 			  NOUVEAU_BO_LOW, 0, 0);
 		so_method(so, nv40->screen->curie, NV40TCL_ZETA_PITCH, 1);
-		so_data  (so, zeta->stride);
+		so_data  (so, zeta->pitch);
 	}
 
 	so_method(so, nv40->screen->curie, NV40TCL_RT_ENABLE, 1);
diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c
new file mode 100644
index 0000000000..b090f2238e
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_transfer.c
@@ -0,0 +1,201 @@
+#include <pipe/p_state.h>
+#include <pipe/p_defines.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <nouveau/nouveau_winsys.h>
+#include "nv40_context.h"
+#include "nv40_screen.h"
+#include "nv40_state.h"
+
+struct nv40_transfer {
+	struct pipe_transfer base;
+	struct pipe_surface *surface;
+	bool direct;
+};
+
+static unsigned nv40_usage_tx_to_buf(unsigned tx_usage)
+{
+	switch (tx_usage) {
+		case PIPE_TRANSFER_READ:
+			return PIPE_BUFFER_USAGE_CPU_READ;
+		case PIPE_TRANSFER_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_WRITE;
+		case PIPE_TRANSFER_READ_WRITE:
+			return PIPE_BUFFER_USAGE_CPU_READ_WRITE;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static void
+nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+                             struct pipe_texture *template)
+{
+	memset(template, 0, sizeof(struct pipe_texture));
+	template->target = pt->target;
+	template->format = pt->format;
+	template->width[0] = pt->width[level];
+	template->height[0] = pt->height[level];
+	template->depth[0] = 1;
+	template->block = pt->block;
+	template->nblocksx[0] = pt->nblocksx[level];
+	template->nblocksy[0] = pt->nblocksx[level];
+	template->last_level = 0;
+	template->compressed = pt->compressed;
+	template->nr_samples = pt->nr_samples;
+
+	template->tex_usage = PIPE_TEXTURE_USAGE_DYNAMIC |
+	                      NOUVEAU_TEXTURE_USAGE_LINEAR;
+}
+
+static struct pipe_transfer *
+nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+		  unsigned face, unsigned level, unsigned zslice,
+		  enum pipe_transfer_usage usage,
+		  unsigned x, unsigned y, unsigned w, unsigned h)
+{
+	struct nv40_miptree *mt = (struct nv40_miptree *)pt;
+	struct nv40_transfer *tx;
+	struct pipe_texture tx_tex_template, *tx_tex;
+
+	tx = CALLOC_STRUCT(nv40_transfer);
+	if (!tx)
+		return NULL;
+
+	tx->base.refcount = 1;
+	pipe_texture_reference(&tx->base.texture, pt);
+	tx->base.format = pt->format;
+	tx->base.x = x;
+	tx->base.y = y;
+	tx->base.width = w;
+	tx->base.height = h;
+	tx->base.block = pt->block;
+	tx->base.nblocksx = pt->nblocksx[level];
+	tx->base.nblocksy = pt->nblocksy[level];
+	tx->base.stride = mt->level[level].pitch;
+	tx->base.usage = usage;
+	tx->base.face = face;
+	tx->base.level = level;
+	tx->base.zslice = zslice;
+
+	/* Direct access to texture */
+	if ((pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC ||
+	     debug_get_bool_option("NOUVEAU_NO_TRANSFER", TRUE/*XXX:FALSE*/)) &&
+	    pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)
+	{
+		tx->direct = true;
+		tx->surface = pscreen->get_tex_surface(pscreen, pt,
+	                                               face, level, zslice,
+	                                               nv40_usage_tx_to_buf(usage));
+		return &tx->base;
+	}
+
+	tx->direct = false;
+
+	nv40_compatible_transfer_tex(pt, level, &tx_tex_template);
+
+	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
+	if (!tx_tex)
+	{
+		FREE(tx);
+		return NULL;
+	}
+
+	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
+	                                       face, level, zslice,
+	                                       nv40_usage_tx_to_buf(usage));
+
+	pipe_texture_reference(&tx_tex, NULL);
+
+	if (!tx->surface)
+	{
+		pipe_surface_reference(&tx->surface, NULL);
+		FREE(tx);
+		return NULL;
+	}
+
+	if (usage != PIPE_TRANSFER_WRITE) {
+		struct nv40_screen *nvscreen = nv40_screen(pscreen);
+		struct pipe_surface *src;
+
+		src = pscreen->get_tex_surface(pscreen, pt,
+	                                       face, level, zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_READ);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		/* TODO: Check if SIFM can un-swizzle */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      tx->surface, 0, 0,
+		                      src, 0, 0,
+		                      src->width, src->height);
+
+		pipe_surface_reference(&src, NULL);
+	}
+
+	return &tx->base;
+}
+
+static void
+nv40_transfer_del(struct pipe_screen *pscreen, struct pipe_transfer **pptx)
+{
+	struct pipe_transfer *ptx = *pptx;
+	struct nv40_transfer *tx = (struct nv40_transfer *)ptx;
+
+	if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) {
+		struct nv40_screen *nvscreen = nv40_screen(pscreen);
+		struct pipe_surface *dst;
+
+		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
+	                                       ptx->face, ptx->level, ptx->zslice,
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+
+		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
+		nvscreen->eng2d->copy(nvscreen->eng2d,
+		                      dst, 0, 0,
+		                      tx->surface, 0, 0,
+		                      dst->width, dst->height);
+
+		pipe_surface_reference(&dst, NULL);
+	}
+
+	*pptx = NULL;
+	if (--ptx->refcount)
+		return;
+
+	pipe_surface_reference(&tx->surface, NULL);
+	pipe_texture_reference(&ptx->texture, NULL);
+	FREE(ptx);
+}
+
+static void *
+nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv40_transfer *tx = (struct nv40_transfer *)ptx;
+	struct nv04_surface *ns = (struct nv04_surface *)tx->surface;
+	struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture;
+	void *map = pipe_buffer_map(pscreen, mt->buffer,
+	                            nv40_usage_tx_to_buf(ptx->usage));
+
+	return map + ns->base.offset +
+	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+}
+
+static void
+nv40_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
+{
+	struct nv40_transfer *tx = (struct nv40_transfer *)ptx;
+	struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture;
+
+	pipe_buffer_unmap(pscreen, mt->buffer);
+}
+
+void
+nv40_screen_init_transfer_functions(struct pipe_screen *pscreen)
+{
+	pscreen->get_tex_transfer = nv40_transfer_new;
+	pscreen->tex_transfer_release = nv40_transfer_del;
+	pscreen->transfer_map = nv40_transfer_map;
+	pscreen->transfer_unmap = nv40_transfer_unmap;
+}
diff --git a/src/gallium/drivers/r300/Makefile b/src/gallium/drivers/r300/Makefile
index 9b7524b523..a0fd17bd59 100644
--- a/src/gallium/drivers/r300/Makefile
+++ b/src/gallium/drivers/r300/Makefile
@@ -12,6 +12,7 @@ C_SOURCES = \
 	r300_screen.c \
 	r300_state.c \
 	r300_state_derived.c \
+	r300_state_invariant.c \
 	r300_state_shader.c \
 	r300_surface.c \
 	r300_swtcl_emit.c \
diff --git a/src/gallium/drivers/r300/r300_chipset.c b/src/gallium/drivers/r300/r300_chipset.c
index 196537a432..e01a0546b2 100644
--- a/src/gallium/drivers/r300/r300_chipset.c
+++ b/src/gallium/drivers/r300/r300_chipset.c
@@ -30,7 +30,7 @@
 void r300_parse_chipset(struct r300_capabilities* caps)
 {
     /* Reasonable defaults */
-    caps->has_tcl = TRUE;
+    caps->has_tcl = getenv("RADEON_NO_TCL") ? TRUE : FALSE;
     caps->is_r500 = FALSE;
     caps->num_vert_fpus = 4;
 
@@ -204,6 +204,13 @@ void r300_parse_chipset(struct r300_capabilities* caps)
             caps->has_tcl = FALSE;
             break;
 
+        case 0x793F:
+        case 0x7941:
+        case 0x7942:
+            caps->family = CHIP_FAMILY_RS600;
+            caps->has_tcl = FALSE;
+            break;
+
         case 0x796C:
         case 0x796D:
         case 0x796E:
diff --git a/src/gallium/drivers/r300/r300_chipset.h b/src/gallium/drivers/r300/r300_chipset.h
index a9cd372ec5..21eebeae60 100644
--- a/src/gallium/drivers/r300/r300_chipset.h
+++ b/src/gallium/drivers/r300/r300_chipset.h
@@ -64,6 +64,7 @@ enum {
     CHIP_FAMILY_RC410,
     CHIP_FAMILY_RS480,
     CHIP_FAMILY_RS482,
+    CHIP_FAMILY_RS600,
     CHIP_FAMILY_RS690,
     CHIP_FAMILY_RS740,
     CHIP_FAMILY_RV515,
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 15a8751549..a981150143 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -73,7 +73,7 @@ static boolean r300_draw_range_elements(struct pipe_context* pipe,
                                              start + count - 1, NULL);
     }
 
-    return true;
+    return TRUE;
 }
 
 static boolean r300_draw_elements(struct pipe_context* pipe,
@@ -97,6 +97,7 @@ static void r300_destroy_context(struct pipe_context* context) {
     draw_destroy(r300->draw);
 
     FREE(r300->blend_color_state);
+    FREE(r300->rs_block);
     FREE(r300->scissor_state);
     FREE(r300);
 }
@@ -126,6 +127,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     draw_set_rasterize_stage(r300->draw, r300_draw_swtcl_stage(r300));
 
     r300->blend_color_state = CALLOC_STRUCT(r300_blend_color_state);
+    r300->rs_block = CALLOC_STRUCT(r300_rs_block);
     r300->scissor_state = CALLOC_STRUCT(r300_scissor_state);
 
     r300_init_flush_functions(r300);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index aaab1dd2bc..5247901be5 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -75,6 +75,13 @@ struct r300_rs_state {
     uint32_t line_stipple_value;    /* R300_GA_LINE_STIPPLE_VALUE: 0x4260 */
 };
 
+struct r300_rs_block {
+    uint32_t ip[8]; /* R300_RS_IP_[0-7], R500_RS_IP_[0-7] */
+    uint32_t count; /* R300_RS_COUNT */
+    uint32_t inst_count; /* R300_RS_INST_COUNT */
+    uint32_t inst[8]; /* R300_RS_INST_[0-7] */
+};
+
 struct r300_sampler_state {
     uint32_t filter0;      /* R300_TX_FILTER0: 0x4400 */
     uint32_t filter1;      /* R300_TX_FILTER1: 0x4440 */
@@ -96,12 +103,13 @@ struct r300_texture_state {
 #define R300_NEW_FRAMEBUFFERS    0x0000010
 #define R300_NEW_FRAGMENT_SHADER 0x0000020
 #define R300_NEW_RASTERIZER      0x0000040
-#define R300_NEW_SAMPLER         0x0000080
-#define R300_NEW_SCISSOR         0x0008000
-#define R300_NEW_TEXTURE         0x0010000
-#define R300_NEW_VERTEX_FORMAT   0x1000000
-#define R300_NEW_VERTEX_SHADER   0x2000000
-#define R300_NEW_KITCHEN_SINK    0x3ffffff
+#define R300_NEW_RS_BLOCK        0x0000080
+#define R300_NEW_SAMPLER         0x0000100
+#define R300_NEW_SCISSOR         0x0010000
+#define R300_NEW_TEXTURE         0x0020000
+#define R300_NEW_VERTEX_FORMAT   0x2000000
+#define R300_NEW_VERTEX_SHADER   0x4000000
+#define R300_NEW_KITCHEN_SINK    0x7ffffff
 
 /* The next several objects are not pure Radeon state; they inherit from
  * various Gallium classes. */
@@ -224,6 +232,8 @@ struct r300_context {
     struct pipe_framebuffer_state framebuffer_state;
     /* Rasterizer state. */
     struct r300_rs_state* rs_state;
+    /* RS block state. */
+    struct r300_rs_block* rs_block;
     /* Sampler states. */
     struct r300_sampler_state* sampler_states[8];
     int sampler_count;
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 960f45f651..91fac62cbe 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -40,9 +40,9 @@ void r300_emit_blend_state(struct r300_context* r300,
 void r300_emit_blend_color_state(struct r300_context* r300,
                                  struct r300_blend_color_state* bc)
 {
-    struct r300_screen* r300screen =
-        (struct r300_screen*)r300->context.screen;
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
     CS_LOCALS(r300);
+
     if (r300screen->caps->is_r500) {
         BEGIN_CS(3);
         OUT_CS_REG_SEQ(R500_RB3D_CONSTANT_COLOR_AR, 2);
@@ -59,9 +59,9 @@ void r300_emit_blend_color_state(struct r300_context* r300,
 void r300_emit_dsa_state(struct r300_context* r300,
                            struct r300_dsa_state* dsa)
 {
-    struct r300_screen* r300screen =
-        (struct r300_screen*)r300->context.screen;
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
     CS_LOCALS(r300);
+
     BEGIN_CS(r300screen->caps->is_r500 ? 8 : 8);
     OUT_CS_REG(R300_FG_ALPHA_FUNC, dsa->alpha_function);
     /* XXX figure out the r300 counterpart for this */
@@ -84,6 +84,7 @@ void r300_emit_fragment_shader(struct r300_context* r300,
 {
     CS_LOCALS(r300);
     int i;
+
     BEGIN_CS(22);
 
     OUT_CS_REG(R300_US_CONFIG, MAX2(fs->indirections - 1, 0));
@@ -114,7 +115,8 @@ void r500_emit_fragment_shader(struct r300_context* r300,
                                struct r500_fragment_shader* fs)
 {
     CS_LOCALS(r300);
-    int i = 0;
+    int i;
+
     BEGIN_CS(9 + (fs->instruction_count * 6));
     OUT_CS_REG(R500_US_CONFIG, R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO);
     OUT_CS_REG(R500_US_PIXSIZE, fs->shader.stack_size);
@@ -191,9 +193,9 @@ void r300_emit_fb_state(struct r300_context* r300,
 
 void r300_emit_rs_state(struct r300_context* r300, struct r300_rs_state* rs)
 {
-    struct r300_screen* r300screen =
-        (struct r300_screen*)r300->context.screen;
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
     CS_LOCALS(r300);
+
     BEGIN_CS(13);
     OUT_CS_REG(R300_VAP_CNTL_STATUS, rs->vap_control_status);
     OUT_CS_REG_SEQ(R300_SU_POLY_OFFSET_FRONT_SCALE, 6);
@@ -208,10 +210,44 @@ void r300_emit_rs_state(struct r300_context* r300, struct r300_rs_state* rs)
     END_CS;
 }
 
+void r300_emit_rs_block_state(struct r300_context* r300,
+                              struct r300_rs_block* rs)
+{
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
+    CS_LOCALS(r300);
+    int i;
+
+    BEGIN_CS(21);
+    if (r300screen->caps->is_r500) {
+        OUT_CS_REG_SEQ(R500_RS_IP_0, 8);
+    } else {
+        OUT_CS_REG_SEQ(R300_RS_IP_0, 8);
+    }
+    for (i = 0; i < 8; i++) {
+        OUT_CS(rs->ip[i]);
+    }
+
+    OUT_CS_REG_SEQ(R300_RS_COUNT, 2);
+    OUT_CS(rs->count);
+    OUT_CS(rs->inst_count);
+
+    if (r300screen->caps->is_r500) {
+        OUT_CS_REG_SEQ(R500_RS_INST_0, 8);
+    } else {
+        OUT_CS_REG_SEQ(R300_RS_INST_0, 8);
+    }
+    for (i = 0; i < 8; i++) {
+        OUT_CS(rs->inst[i]);
+    }
+
+    END_CS;
+}
+
 void r300_emit_scissor_state(struct r300_context* r300,
                              struct r300_scissor_state* scissor)
 {
     CS_LOCALS(r300);
+
     BEGIN_CS(3);
     OUT_CS_REG_SEQ(R300_SC_SCISSORS_TL, 2);
     OUT_CS(scissor->scissor_top_left);
@@ -248,8 +284,7 @@ void r300_emit_vertex_format_state(struct r300_context* r300)
 /* Emit all dirty state. */
 void r300_emit_dirty_state(struct r300_context* r300)
 {
-    struct r300_screen* r300screen =
-        (struct r300_screen*)r300->context.screen;
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
     CS_LOCALS(r300);
 
     if (!(r300->dirty_state) && !(r300->dirty_hw)) {
diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h
index 8888b39a2f..e0da9d361e 100644
--- a/src/gallium/drivers/r300/r300_reg.h
+++ b/src/gallium/drivers/r300/r300_reg.h
@@ -732,6 +732,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R500_RS_IP_TEX_PTR_Q_SHIFT 			18
 #define R500_RS_IP_COL_PTR_SHIFT 			24
 #define R500_RS_IP_COL_FMT_SHIFT 			27
+#       define R500_RS_SEL_S(x)                         ((x) << 0)
+#       define R500_RS_SEL_T(x)                         ((x) << 6)
+#       define R500_RS_SEL_R(x)                         ((x) << 12)
+#       define R500_RS_SEL_Q(x)                         ((x) << 18)
 #	define R500_RS_COL_PTR(x)		        ((x) << 24)
 #       define R500_RS_COL_FMT(x)                       ((x) << 27)
 /* gap */
@@ -1216,14 +1220,18 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R500_RS_INST_14					0x4358
 #define R500_RS_INST_15					0x435c
 #define R500_RS_INST_TEX_ID_SHIFT			0
+#        define R500_RS_INST_TEX_ID(x)                  ((x) << 0)
 #define R500_RS_INST_TEX_CN_WRITE			(1 << 4)
 #define R500_RS_INST_TEX_ADDR_SHIFT			5
+#        define R500_RS_INST_TEX_ADDR(x)                ((x) << 0)
 #define R500_RS_INST_COL_ID_SHIFT			12
+#        define R500_RS_INST_COL_ID(x)                  ((x) << 12)
 #define R500_RS_INST_COL_CN_NO_WRITE			(0 << 16)
 #define R500_RS_INST_COL_CN_WRITE			(1 << 16)
 #define R500_RS_INST_COL_CN_WRITE_FBUFFER		(2 << 16)
 #define R500_RS_INST_COL_CN_WRITE_BACKFACE		(3 << 16)
 #define R500_RS_INST_COL_ADDR_SHIFT			18
+#        define R500_RS_INST_COL_ADDR(x)                ((x) << 18)
 #define R500_RS_INST_TEX_ADJ				(1 << 25)
 #define R500_RS_INST_W_CN				(1 << 26)
 
@@ -1240,9 +1248,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_RS_INST_7                     0x434C
 #	define R300_RS_INST_TEX_ID(x)  		((x) << 0)
 #	define R300_RS_INST_TEX_CN_WRITE 	(1 << 3)
+#	define R300_RS_INST_TEX_ADDR(x)		((x) << 6)
 #	define R300_RS_INST_TEX_ADDR_SHIFT 	6
 #	define R300_RS_INST_COL_ID(x)		((x) << 11)
 #	define R300_RS_INST_COL_CN_WRITE	(1 << 14)
+#	define R300_RS_INST_COL_ADDR(x)		((x) << 17)
 #	define R300_RS_INST_COL_ADDR_SHIFT	17
 #	define R300_RS_INST_TEX_ADJ		(1 << 22)
 #	define R300_RS_COL_BIAS_UNUSED_SHIFT    23
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 5ff9015a7b..2fcd504812 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -50,6 +50,7 @@ static const char* chip_families[] = {
     "RC410",
     "RS480",
     "RS482",
+    "RS600",
     "RS690",
     "RS740",
     "RV515",
@@ -178,8 +179,11 @@ static boolean check_tex_2d_format(enum pipe_format format, boolean is_r500)
 {
     switch (format) {
         /* Colorbuffer */
+        case PIPE_FORMAT_A4R4G4B4_UNORM:
+        case PIPE_FORMAT_R5G6B5_UNORM:
+        case PIPE_FORMAT_A1R5G5B5_UNORM:
         case PIPE_FORMAT_A8R8G8B8_UNORM:
-        /* Texture */
+        /* Colorbuffer or texture */
         case PIPE_FORMAT_I8_UNORM:
         /* Z buffer */
         case PIPE_FORMAT_Z16_UNORM:
@@ -187,29 +191,25 @@ static boolean check_tex_2d_format(enum pipe_format format, boolean is_r500)
         case PIPE_FORMAT_Z24S8_UNORM:
             return TRUE;
 
-        /* XXX Supported yet unimplemented formats: */
-        case PIPE_FORMAT_A1R5G5B5_UNORM:
-        case PIPE_FORMAT_R5G6B5_UNORM:
         /* XXX These don't even exist
         case PIPE_FORMAT_A32R32G32B32:
         case PIPE_FORMAT_A16R16G16B16: */
         /* XXX Insert YUV422 packed VYUY and YVYU here */
-        /* XXX What the deuce is UV88? (r3xx accel page 14) */
-        case PIPE_FORMAT_A4R4G4B4_UNORM:
+        /* XXX What the deuce is UV88? (r3xx accel page 14)
             debug_printf("r300: Warning: Got unimplemented format: %s in %s\n",
                 pf_name(format), __FUNCTION__);
-            return FALSE;
+            return FALSE; */
 
         /* XXX Supported yet unimplemented r5xx formats: */
         /* XXX Again, what is UV1010 this time? (r5xx accel page 148) */
         /* XXX Even more that don't exist
         case PIPE_FORMAT_A10R10G10B10_UNORM:
         case PIPE_FORMAT_A2R10G10B10_UNORM:
-        case PIPE_FORMAT_I10_UNORM: */
+        case PIPE_FORMAT_I10_UNORM:
             debug_printf(
                 "r300: Warning: Got unimplemented r500 format: %s in %s\n",
                 pf_name(format), __FUNCTION__);
-            return FALSE;
+            return FALSE; */
 
         default:
             debug_printf("r300: Warning: Got unsupported format: %s in %s\n",
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index da99a3be6b..6e64ad2dc3 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -28,89 +28,11 @@
 
 #include "r300_context.h"
 #include "r300_reg.h"
+#include "r300_state_inlines.h"
 #include "r300_state_shader.h"
 
 /* r300_state: Functions used to intialize state context by translating
- * Gallium state objects into semi-native r300 state objects.
- *
- * XXX break this file up into pieces if it gets too big! */
-
-/* Pack a float into a dword. */
-static uint32_t pack_float_32(float f)
-{
-    union {
-        float f;
-        uint32_t u;
-    } u;
-
-    u.f = f;
-    return u.u;
-}
-
-static uint32_t translate_blend_function(int blend_func) {
-    switch (blend_func) {
-        case PIPE_BLEND_ADD:
-            return R300_COMB_FCN_ADD_CLAMP;
-        case PIPE_BLEND_SUBTRACT:
-            return R300_COMB_FCN_SUB_CLAMP;
-        case PIPE_BLEND_REVERSE_SUBTRACT:
-            return R300_COMB_FCN_RSUB_CLAMP;
-        case PIPE_BLEND_MIN:
-            return R300_COMB_FCN_MIN;
-        case PIPE_BLEND_MAX:
-            return R300_COMB_FCN_MAX;
-        default:
-            debug_printf("r300: Unknown blend function %d\n", blend_func);
-            break;
-    }
-    return 0;
-}
-
-/* XXX we can also offer the D3D versions of some of these... */
-static uint32_t translate_blend_factor(int blend_fact) {
-    switch (blend_fact) {
-        case PIPE_BLENDFACTOR_ONE:
-            return R300_BLEND_GL_ONE;
-        case PIPE_BLENDFACTOR_SRC_COLOR:
-            return R300_BLEND_GL_SRC_COLOR;
-        case PIPE_BLENDFACTOR_SRC_ALPHA:
-            return R300_BLEND_GL_SRC_ALPHA;
-        case PIPE_BLENDFACTOR_DST_ALPHA:
-            return R300_BLEND_GL_DST_ALPHA;
-        case PIPE_BLENDFACTOR_DST_COLOR:
-            return R300_BLEND_GL_DST_COLOR;
-        case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
-            return R300_BLEND_GL_SRC_ALPHA_SATURATE;
-        case PIPE_BLENDFACTOR_CONST_COLOR:
-            return R300_BLEND_GL_CONST_COLOR;
-        case PIPE_BLENDFACTOR_CONST_ALPHA:
-            return R300_BLEND_GL_CONST_ALPHA;
-        /* XXX WTF are these?
-        case PIPE_BLENDFACTOR_SRC1_COLOR:
-        case PIPE_BLENDFACTOR_SRC1_ALPHA: */
-        case PIPE_BLENDFACTOR_ZERO:
-            return R300_BLEND_GL_ZERO;
-        case PIPE_BLENDFACTOR_INV_SRC_COLOR:
-            return R300_BLEND_GL_ONE_MINUS_SRC_COLOR;
-        case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-            return R300_BLEND_GL_ONE_MINUS_SRC_ALPHA;
-        case PIPE_BLENDFACTOR_INV_DST_ALPHA:
-            return R300_BLEND_GL_ONE_MINUS_DST_ALPHA;
-        case PIPE_BLENDFACTOR_INV_DST_COLOR:
-            return R300_BLEND_GL_ONE_MINUS_DST_COLOR;
-        case PIPE_BLENDFACTOR_INV_CONST_COLOR:
-            return R300_BLEND_GL_ONE_MINUS_CONST_COLOR;
-        case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
-            return R300_BLEND_GL_ONE_MINUS_CONST_ALPHA;
-        /* XXX see above
-        case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
-        case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: */
-        default:
-            debug_printf("r300: Unknown blend factor %d\n", blend_fact);
-            break;
-    }
-    return 0;
-}
+ * Gallium state objects into semi-native r300 state objects. */
 
 /* Create a new blend state based on the CSO blend state.
  *
@@ -126,16 +48,16 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
         blend->blend_control = R300_ALPHA_BLEND_ENABLE |
                 R300_SEPARATE_ALPHA_ENABLE |
                 R300_READ_ENABLE |
-                translate_blend_function(state->rgb_func) |
-                (translate_blend_factor(state->rgb_src_factor) <<
+                r300_translate_blend_function(state->rgb_func) |
+                (r300_translate_blend_factor(state->rgb_src_factor) <<
                     R300_SRC_BLEND_SHIFT) |
-                (translate_blend_factor(state->rgb_dst_factor) <<
+                (r300_translate_blend_factor(state->rgb_dst_factor) <<
                     R300_DST_BLEND_SHIFT);
         blend->alpha_blend_control =
-                translate_blend_function(state->alpha_func) |
-                (translate_blend_factor(state->alpha_src_factor) <<
+                r300_translate_blend_function(state->alpha_func) |
+                (r300_translate_blend_factor(state->alpha_src_factor) <<
                     R300_SRC_BLEND_SHIFT) |
-                (translate_blend_factor(state->alpha_dst_factor) <<
+                (r300_translate_blend_factor(state->alpha_dst_factor) <<
                     R300_DST_BLEND_SHIFT);
     }
 
@@ -178,21 +100,17 @@ static void r300_set_blend_color(struct pipe_context* pipe,
                                  const struct pipe_blend_color* color)
 {
     struct r300_context* r300 = r300_context(pipe);
-    uint32_t r, g, b, a;
     ubyte ur, ug, ub, ua;
 
-    r = util_iround(color->color[0] * 1023.0f);
-    g = util_iround(color->color[1] * 1023.0f);
-    b = util_iround(color->color[2] * 1023.0f);
-    a = util_iround(color->color[3] * 1023.0f);
-
     ur = float_to_ubyte(color->color[0]);
     ug = float_to_ubyte(color->color[1]);
     ub = float_to_ubyte(color->color[2]);
     ua = float_to_ubyte(color->color[3]);
 
-    r300->blend_color_state->blend_color = (a << 24) | (r << 16) | (g << 8) | b;
+    util_pack_color(color->color, PIPE_FORMAT_A8R8G8B8_UNORM,
+            &r300->blend_color_state->blend_color);
 
+    /* XXX this is wrong */
     r300->blend_color_state->blend_color_red_alpha = ur | (ua << 16);
     r300->blend_color_state->blend_color_green_blue = ub | (ug << 16);
 
@@ -233,82 +151,6 @@ static void
     r300->dirty_state |= R300_NEW_CONSTANTS;
 }
 
-static uint32_t translate_depth_stencil_function(int zs_func) {
-    switch (zs_func) {
-        case PIPE_FUNC_NEVER:
-            return R300_ZS_NEVER;
-        case PIPE_FUNC_LESS:
-            return R300_ZS_LESS;
-        case PIPE_FUNC_EQUAL:
-            return R300_ZS_EQUAL;
-        case PIPE_FUNC_LEQUAL:
-            return R300_ZS_LEQUAL;
-        case PIPE_FUNC_GREATER:
-            return R300_ZS_GREATER;
-        case PIPE_FUNC_NOTEQUAL:
-            return R300_ZS_NOTEQUAL;
-        case PIPE_FUNC_GEQUAL:
-            return R300_ZS_GEQUAL;
-        case PIPE_FUNC_ALWAYS:
-            return R300_ZS_ALWAYS;
-        default:
-            debug_printf("r300: Unknown depth/stencil function %d\n",
-                zs_func);
-            break;
-    }
-    return 0;
-}
-
-static uint32_t translate_stencil_op(int s_op) {
-    switch (s_op) {
-        case PIPE_STENCIL_OP_KEEP:
-            return R300_ZS_KEEP;
-        case PIPE_STENCIL_OP_ZERO:
-            return R300_ZS_ZERO;
-        case PIPE_STENCIL_OP_REPLACE:
-            return R300_ZS_REPLACE;
-        case PIPE_STENCIL_OP_INCR:
-            return R300_ZS_INCR;
-        case PIPE_STENCIL_OP_DECR:
-            return R300_ZS_DECR;
-        case PIPE_STENCIL_OP_INCR_WRAP:
-            return R300_ZS_INCR_WRAP;
-        case PIPE_STENCIL_OP_DECR_WRAP:
-            return R300_ZS_DECR_WRAP;
-        case PIPE_STENCIL_OP_INVERT:
-            return R300_ZS_INVERT;
-        default:
-            debug_printf("r300: Unknown stencil op %d", s_op);
-            break;
-    }
-    return 0;
-}
-
-static uint32_t translate_alpha_function(int alpha_func) {
-    switch (alpha_func) {
-        case PIPE_FUNC_NEVER:
-            return R300_FG_ALPHA_FUNC_NEVER;
-        case PIPE_FUNC_LESS:
-            return R300_FG_ALPHA_FUNC_LESS;
-        case PIPE_FUNC_EQUAL:
-            return R300_FG_ALPHA_FUNC_EQUAL;
-        case PIPE_FUNC_LEQUAL:
-            return R300_FG_ALPHA_FUNC_LE;
-        case PIPE_FUNC_GREATER:
-            return R300_FG_ALPHA_FUNC_GREATER;
-        case PIPE_FUNC_NOTEQUAL:
-            return R300_FG_ALPHA_FUNC_NOTEQUAL;
-        case PIPE_FUNC_GEQUAL:
-            return R300_FG_ALPHA_FUNC_GE;
-        case PIPE_FUNC_ALWAYS:
-            return R300_FG_ALPHA_FUNC_ALWAYS;
-        default:
-            debug_printf("r300: Unknown alpha function %d", alpha_func);
-            break;
-    }
-    return 0;
-}
-
 /* Create a new depth, stencil, and alpha state based on the CSO dsa state.
  *
  * This contains the depth buffer, stencil buffer, alpha test, and such.
@@ -329,7 +171,7 @@ static void*
         }
 
         dsa->z_stencil_control |=
-            (translate_depth_stencil_function(state->depth.func) <<
+            (r300_translate_depth_stencil_function(state->depth.func) <<
                 R300_Z_FUNC_SHIFT);
     }
 
@@ -337,14 +179,14 @@ static void*
     if (state->stencil[0].enabled) {
         dsa->z_buffer_control |= R300_STENCIL_ENABLE;
         dsa->z_stencil_control |=
-                (translate_depth_stencil_function(state->stencil[0].func) <<
-                    R300_S_FRONT_FUNC_SHIFT) |
-                (translate_stencil_op(state->stencil[0].fail_op) <<
-                    R300_S_FRONT_SFAIL_OP_SHIFT) |
-                (translate_stencil_op(state->stencil[0].zpass_op) <<
-                    R300_S_FRONT_ZPASS_OP_SHIFT) |
-                (translate_stencil_op(state->stencil[0].zfail_op) <<
-                    R300_S_FRONT_ZFAIL_OP_SHIFT);
+            (r300_translate_depth_stencil_function(state->stencil[0].func) <<
+                R300_S_FRONT_FUNC_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[0].fail_op) <<
+                R300_S_FRONT_SFAIL_OP_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[0].zpass_op) <<
+                R300_S_FRONT_ZPASS_OP_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[0].zfail_op) <<
+                R300_S_FRONT_ZFAIL_OP_SHIFT);
 
         dsa->stencil_ref_mask = (state->stencil[0].ref_value) |
                 (state->stencil[0].valuemask << R300_STENCILMASK_SHIFT) |
@@ -353,14 +195,14 @@ static void*
         if (state->stencil[1].enabled) {
             dsa->z_buffer_control |= R300_STENCIL_FRONT_BACK;
             dsa->z_stencil_control |=
-                (translate_depth_stencil_function(state->stencil[1].func) <<
-                    R300_S_BACK_FUNC_SHIFT) |
-                (translate_stencil_op(state->stencil[1].fail_op) <<
-                    R300_S_BACK_SFAIL_OP_SHIFT) |
-                (translate_stencil_op(state->stencil[1].zpass_op) <<
-                    R300_S_BACK_ZPASS_OP_SHIFT) |
-                (translate_stencil_op(state->stencil[1].zfail_op) <<
-                    R300_S_BACK_ZFAIL_OP_SHIFT);
+            (r300_translate_depth_stencil_function(state->stencil[1].func) <<
+                R300_S_BACK_FUNC_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[1].fail_op) <<
+                R300_S_BACK_SFAIL_OP_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[1].zpass_op) <<
+                R300_S_BACK_ZPASS_OP_SHIFT) |
+            (r300_translate_stencil_op(state->stencil[1].zfail_op) <<
+                R300_S_BACK_ZFAIL_OP_SHIFT);
 
             dsa->stencil_ref_bf = (state->stencil[1].ref_value) |
                 (state->stencil[1].valuemask << R300_STENCILMASK_SHIFT) |
@@ -370,7 +212,8 @@ static void*
 
     /* Alpha test setup. */
     if (state->alpha.enabled) {
-        dsa->alpha_function = translate_alpha_function(state->alpha.func) |
+        dsa->alpha_function =
+            r300_translate_alpha_function(state->alpha.func) |
             R300_FG_ALPHA_FUNC_ENABLE;
         dsa->alpha_reference = CLAMP(state->alpha.ref_value * 1023.0f,
                                      0, 1023);
@@ -457,7 +300,7 @@ static void r300_bind_fs_state(struct pipe_context* pipe, void* shader)
         }
     }
 
-    fs->translated = true;
+    fs->translated = TRUE;
     r300->fs = fs;
 
     r300->dirty_state |= R300_NEW_FRAGMENT_SHADER;
@@ -475,10 +318,6 @@ static void r300_set_polygon_stipple(struct pipe_context* pipe,
     /* XXX */
 }
 
-static INLINE int pack_float_16_6x(float f) {
-    return ((int)(f * 6.0) & 0xffff);
-}
-
 /* Create a new rasterizer state based on the CSO rasterizer state.
  *
  * This is a very large chunk of state, and covers most of the graphics
@@ -530,15 +369,15 @@ static void* r300_create_rs_state(struct pipe_context* pipe,
 
     if (rs->polygon_offset_enable) {
         rs->depth_offset_front = rs->depth_offset_back =
-                pack_float_32(state->offset_units);
+            fui(state->offset_units);
         rs->depth_scale_front = rs->depth_scale_back =
-                pack_float_32(state->offset_scale);
+            fui(state->offset_scale);
     }
 
     if (state->line_stipple_enable) {
         rs->line_stipple_config =
             R300_GA_LINE_STIPPLE_CONFIG_LINE_RESET_LINE |
-            (pack_float_32((float)state->line_stipple_factor) &
+            (fui((float)state->line_stipple_factor) &
                 R300_GA_LINE_STIPPLE_CONFIG_STIPPLE_SCALE_MASK);
         /* XXX this might need to be scaled up */
         rs->line_stipple_value = state->line_stipple_pattern;
@@ -567,83 +406,6 @@ static void r300_delete_rs_state(struct pipe_context* pipe, void* state)
     FREE(state);
 }
 
-static uint32_t translate_wrap(int wrap) {
-    switch (wrap) {
-        case PIPE_TEX_WRAP_REPEAT:
-            return R300_TX_REPEAT;
-        case PIPE_TEX_WRAP_CLAMP:
-            return R300_TX_CLAMP;
-        case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-            return R300_TX_CLAMP_TO_EDGE;
-        case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-            return R300_TX_CLAMP_TO_BORDER;
-        case PIPE_TEX_WRAP_MIRROR_REPEAT:
-            return R300_TX_REPEAT | R300_TX_MIRRORED;
-        case PIPE_TEX_WRAP_MIRROR_CLAMP:
-            return R300_TX_CLAMP | R300_TX_MIRRORED;
-        case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-            return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
-        case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
-            return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
-        default:
-            debug_printf("r300: Unknown texture wrap %d", wrap);
-            return 0;
-    }
-}
-
-static uint32_t translate_tex_filters(int min, int mag, int mip) {
-    uint32_t retval = 0;
-    switch (min) {
-        case PIPE_TEX_FILTER_NEAREST:
-            retval |= R300_TX_MIN_FILTER_NEAREST;
-        case PIPE_TEX_FILTER_LINEAR:
-            retval |= R300_TX_MIN_FILTER_LINEAR;
-        case PIPE_TEX_FILTER_ANISO:
-            retval |= R300_TX_MIN_FILTER_ANISO;
-        default:
-            debug_printf("r300: Unknown texture filter %d", min);
-            break;
-    }
-    switch (mag) {
-        case PIPE_TEX_FILTER_NEAREST:
-            retval |= R300_TX_MAG_FILTER_NEAREST;
-        case PIPE_TEX_FILTER_LINEAR:
-            retval |= R300_TX_MAG_FILTER_LINEAR;
-        case PIPE_TEX_FILTER_ANISO:
-            retval |= R300_TX_MAG_FILTER_ANISO;
-        default:
-            debug_printf("r300: Unknown texture filter %d", mag);
-            break;
-    }
-    switch (mip) {
-        case PIPE_TEX_MIPFILTER_NONE:
-            retval |= R300_TX_MIN_FILTER_MIP_NONE;
-        case PIPE_TEX_MIPFILTER_NEAREST:
-            retval |= R300_TX_MIN_FILTER_MIP_NEAREST;
-        case PIPE_TEX_MIPFILTER_LINEAR:
-            retval |= R300_TX_MIN_FILTER_MIP_LINEAR;
-        default:
-            debug_printf("r300: Unknown texture filter %d", mip);
-            break;
-    }
-
-    return retval;
-}
-
-static uint32_t anisotropy(float max_aniso) {
-    if (max_aniso >= 16.0f) {
-        return R300_TX_MAX_ANISO_16_TO_1;
-    } else if (max_aniso >= 8.0f) {
-        return R300_TX_MAX_ANISO_8_TO_1;
-    } else if (max_aniso >= 4.0f) {
-        return R300_TX_MAX_ANISO_4_TO_1;
-    } else if (max_aniso >= 2.0f) {
-        return R300_TX_MAX_ANISO_2_TO_1;
-    } else {
-        return R300_TX_MAX_ANISO_1_TO_1;
-    }
-}
-
 static void*
         r300_create_sampler_state(struct pipe_context* pipe,
                                   const struct pipe_sampler_state* state)
@@ -653,19 +415,19 @@ static void*
     int lod_bias;
 
     sampler->filter0 |=
-        (translate_wrap(state->wrap_s) << R300_TX_WRAP_S_SHIFT) |
-        (translate_wrap(state->wrap_t) << R300_TX_WRAP_T_SHIFT) |
-        (translate_wrap(state->wrap_r) << R300_TX_WRAP_R_SHIFT);
+        (r300_translate_wrap(state->wrap_s) << R300_TX_WRAP_S_SHIFT) |
+        (r300_translate_wrap(state->wrap_t) << R300_TX_WRAP_T_SHIFT) |
+        (r300_translate_wrap(state->wrap_r) << R300_TX_WRAP_R_SHIFT);
 
-    sampler->filter0 |= translate_tex_filters(state->min_img_filter,
-                                              state->mag_img_filter,
-                                              state->min_mip_filter);
+    sampler->filter0 |= r300_translate_tex_filters(state->min_img_filter,
+                                                   state->mag_img_filter,
+                                                   state->min_mip_filter);
 
     lod_bias = CLAMP((int)(state->lod_bias * 32), -(1 << 9), (1 << 9) - 1);
 
     sampler->filter1 |= lod_bias << R300_LOD_BIAS_SHIFT;
 
-    sampler->filter1 |= anisotropy(state->max_anisotropy);
+    sampler->filter1 |= r300_anisotropy(state->max_anisotropy);
 
     util_pack_color(state->border_color, PIPE_FORMAT_A8R8G8B8_UNORM,
                     &sampler->border_color);
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index a51904096f..548a840f25 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -57,34 +57,40 @@ static uint32_t translate_vertex_data_type(int type) {
 /* XXX this function should be able to handle vert shaders as well as draw */
 static void r300_update_vertex_layout(struct r300_context* r300)
 {
+    struct r300_vertex_format vformat;
     struct vertex_info vinfo;
-    boolean pos = false, psize = false, fog = false;
+    boolean pos = FALSE, psize = FALSE, fog = FALSE;
     int i, texs = 0, cols = 0;
+    int tab[16];
 
     struct tgsi_shader_info* info = &r300->fs->info;
+
     memset(&vinfo, 0, sizeof(vinfo));
+    for (i = 0; i < 16; i++) {
+        tab[i] = -1;
+    }
 
     assert(info->num_inputs <= 16);
 
-    /* This is rather lame. Since draw_find_vs_output doesn't return an error
-     * when it can't find an output, we have to pre-iterate and count each
-     * output ourselves. */
     for (i = 0; i < info->num_inputs; i++) {
         switch (info->input_semantic_name[i]) {
             case TGSI_SEMANTIC_POSITION:
-                pos = true;
+                pos = TRUE;
+                tab[i] = 0;
                 break;
             case TGSI_SEMANTIC_COLOR:
-                cols++;
+                tab[i] = 2 + cols++;
                 break;
             case TGSI_SEMANTIC_FOG:
-                fog = true;
+                fog = TRUE;
+                tab[i] = 6 + texs++;
                 break;
             case TGSI_SEMANTIC_PSIZE:
-                psize = true;
+                psize = TRUE;
+                tab[i] = 1;
                 break;
             case TGSI_SEMANTIC_GENERIC:
-                texs++;
+                tab[i] = 6 + texs++;
                 break;
             default:
                 debug_printf("r300: Unknown vertex input %d\n",
@@ -105,6 +111,12 @@ static void r300_update_vertex_layout(struct r300_context* r300)
 
     if (!pos) {
         debug_printf("r300: Forcing vertex position attribute emit...\n");
+        /* Make room for the position attribute
+         * at the beginning of the tab. */
+        for (i = 1; i < 16; i++) {
+            tab[i] = tab[i-1];
+        }
+        tab[0] = 0;
     }
 
     draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_POS,
@@ -125,16 +137,17 @@ static void r300_update_vertex_layout(struct r300_context* r300)
         vinfo.hwfmt[2] |= (R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT << i);
     }
 
-    if (fog) {
+    for (i = 0; i < texs; i++) {
         draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE,
-            draw_find_vs_output(r300->draw, TGSI_SEMANTIC_FOG, 0));
-        vinfo.hwfmt[2] |=
-            (R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT << cols);
+            draw_find_vs_output(r300->draw, TGSI_SEMANTIC_GENERIC, i));
+        vinfo.hwfmt[1] |= (R300_INPUT_CNTL_TC0 << i);
+        vinfo.hwfmt[3] |= (4 << (3 * i));
     }
 
-    for (i = 0; i < texs; i++) {
-        draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR,
-            draw_find_vs_output(r300->draw, TGSI_SEMANTIC_GENERIC, i));
+    if (fog) {
+        i++;
+        draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE,
+            draw_find_vs_output(r300->draw, TGSI_SEMANTIC_FOG, 0));
         vinfo.hwfmt[1] |= (R300_INPUT_CNTL_TC0 << i);
         vinfo.hwfmt[3] |= (4 << (3 * i));
     }
@@ -152,18 +165,29 @@ static void r300_update_vertex_layout(struct r300_context* r300)
          (0xf << R300_WRITE_ENA_SHIFT))
 
         for (i = 0; i < vinfo.num_attribs; i++) {
+            /* Make sure we have a proper destination for our attribute */
+            if (tab[i] == -1) {
+                debug_printf("attrib count: %d, fp input count: %d\n",
+                        vinfo.num_attribs, info->num_inputs);
+                for (i = 0; i < vinfo.num_attribs; i++) {
+                    debug_printf("attrib: offset %d, interp %d, size %d,"
+                           " tab %d\n", vinfo.attrib[i].src_index,
+                           vinfo.attrib[i].interp_mode, vinfo.attrib[i].emit,
+                           tab[i]);
+                }
+                assert(0);
+            }
+
             temp = translate_vertex_data_type(vinfo.attrib[i].emit) |
-                R300_SIGNED;
+                (tab[i] << R300_DST_VEC_LOC_SHIFT) | R300_SIGNED;
             if (i & 1) {
                 r300->vertex_info.vap_prog_stream_cntl[i >> 1] &= 0xffff0000;
                 r300->vertex_info.vap_prog_stream_cntl[i >> 1] |=
-                        (translate_vertex_data_type(vinfo.attrib[i].emit) |
-                        R300_SIGNED) << 16;
+                        temp << 16;
             } else {
                 r300->vertex_info.vap_prog_stream_cntl[i >> 1] &= 0xffff;
                 r300->vertex_info.vap_prog_stream_cntl[i >> 1] |=
-                    translate_vertex_data_type(vinfo.attrib[i].emit) |
-                    R300_SIGNED;
+                    temp;
             }
 
             r300->vertex_info.vap_prog_stream_cntl_ext[i >> 1] |=
@@ -180,8 +204,85 @@ static void r300_update_vertex_layout(struct r300_context* r300)
 /* Set up the RS block. This is the part of the chipset that actually does
  * the rasterization of vertices into fragments. This is also the part of the
  * chipset that locks up if any part of it is even slightly wrong. */
-void r300_update_rs_block(struct r300_context* r300)
+static void r300_update_rs_block(struct r300_context* r300)
 {
+    struct r300_rs_block* rs = r300->rs_block;
+    struct vertex_info* vinfo = &r300->vertex_info.vinfo;
+    int col_count = 0, fp_offset = 0, i, tex_count = 0;
+
+    memset(rs, 0, sizeof(struct r300_rs_block));
+
+    if (r300_screen(r300->context.screen)->caps->is_r500) {
+        for (i = 0; i < vinfo->num_attribs; i++) {
+            switch (vinfo->attrib[i].interp_mode) {
+                case INTERP_LINEAR:
+                    rs->ip[col_count] |=
+                        R500_RS_COL_PTR(vinfo->attrib[i].src_index) |
+                        R500_RS_COL_FMT(R300_RS_COL_FMT_RGBA);
+                    col_count++;
+                    break;
+                case INTERP_PERSPECTIVE:
+                    rs->ip[tex_count] |=
+                        R500_RS_SEL_S(vinfo->attrib[i].src_index) |
+                        R500_RS_SEL_T(vinfo->attrib[i].src_index + 1) |
+                        R500_RS_SEL_R(vinfo->attrib[i].src_index + 2) |
+                        R500_RS_SEL_Q(vinfo->attrib[i].src_index + 3);
+                    tex_count++;
+                    break;
+            }
+        }
+
+        for (i = 0; i < tex_count; i++) {
+            rs->inst[i] |= R500_RS_INST_TEX_ID(i) | R500_RS_INST_TEX_CN_WRITE |
+                R500_RS_INST_TEX_ADDR(fp_offset);
+            fp_offset++;
+        }
+
+        for (i = 0; i < col_count; i++) {
+            rs->inst[i] |= R500_RS_INST_COL_ID(i) | R500_RS_INST_COL_CN_WRITE |
+                R500_RS_INST_COL_ADDR(fp_offset);
+            fp_offset++;
+        }
+
+        rs->inst_count = MAX2(col_count, tex_count);
+    } else {
+        for (i = 0; i < vinfo->num_attribs; i++) {
+            switch (vinfo->attrib[i].interp_mode) {
+                case INTERP_LINEAR:
+                    rs->ip[col_count] |=
+                        R300_RS_COL_PTR(vinfo->attrib[i].src_index) |
+                        R300_RS_COL_FMT(R300_RS_COL_FMT_RGBA);
+                    col_count++;
+                    break;
+                case INTERP_PERSPECTIVE:
+                    rs->ip[tex_count] |=
+                        R300_RS_TEX_PTR(vinfo->attrib[i].src_index) |
+                        R300_RS_SEL_S(R300_RS_SEL_C0) |
+                        R300_RS_SEL_T(R300_RS_SEL_C1) |
+                        R300_RS_SEL_R(R300_RS_SEL_C2) |
+                        R300_RS_SEL_Q(R300_RS_SEL_C3);
+                    tex_count += 4;
+                    break;
+            }
+        }
+
+        for (i = 0; i < tex_count; i++) {
+            rs->inst[i] |= R300_RS_INST_TEX_ID(i) | R300_RS_INST_TEX_CN_WRITE |
+                R300_RS_INST_TEX_ADDR(fp_offset);
+            fp_offset++;
+        }
+
+        for (i = 0; i < col_count; i++) {
+            rs->inst[i] |= R300_RS_INST_COL_ID(i) | R300_RS_INST_COL_CN_WRITE |
+                R300_RS_INST_COL_ADDR(fp_offset);
+            fp_offset++;
+        }
+    }
+
+    rs->count = (tex_count * 4) | (col_count << R300_IC_COUNT_SHIFT) |
+        R300_HIRES_EN;
+
+    rs->inst_count = MAX2(col_count, tex_count);
 }
 
 void r300_update_derived_state(struct r300_context* r300)
diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index 005fb74ed6..630ac3b2fc 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -1,5 +1,6 @@
 /*
  * Copyright 2009 Joakim Sindholt <opensource@zhasha.com>
+ *                Corbin Simpson <MostAwesomeDude@gmail.com>
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,56 +28,319 @@
 
 #include "r300_reg.h"
 
+/* Some maths. These should probably find their way to u_math, if needed. */
+
+static INLINE int pack_float_16_6x(float f) {
+    return ((int)(f * 6.0) & 0xffff);
+}
+
+/* Blend state. */
+
+static INLINE uint32_t r300_translate_blend_function(int blend_func)
+{
+    switch (blend_func) {
+        case PIPE_BLEND_ADD:
+            return R300_COMB_FCN_ADD_CLAMP;
+        case PIPE_BLEND_SUBTRACT:
+            return R300_COMB_FCN_SUB_CLAMP;
+        case PIPE_BLEND_REVERSE_SUBTRACT:
+            return R300_COMB_FCN_RSUB_CLAMP;
+        case PIPE_BLEND_MIN:
+            return R300_COMB_FCN_MIN;
+        case PIPE_BLEND_MAX:
+            return R300_COMB_FCN_MAX;
+        default:
+            debug_printf("r300: Unknown blend function %d\n", blend_func);
+            break;
+    }
+    return 0;
+}
+
+/* XXX we can also offer the D3D versions of some of these... */
+static INLINE uint32_t r300_translate_blend_factor(int blend_fact)
+{
+    switch (blend_fact) {
+        case PIPE_BLENDFACTOR_ONE:
+            return R300_BLEND_GL_ONE;
+        case PIPE_BLENDFACTOR_SRC_COLOR:
+            return R300_BLEND_GL_SRC_COLOR;
+        case PIPE_BLENDFACTOR_SRC_ALPHA:
+            return R300_BLEND_GL_SRC_ALPHA;
+        case PIPE_BLENDFACTOR_DST_ALPHA:
+            return R300_BLEND_GL_DST_ALPHA;
+        case PIPE_BLENDFACTOR_DST_COLOR:
+            return R300_BLEND_GL_DST_COLOR;
+        case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+            return R300_BLEND_GL_SRC_ALPHA_SATURATE;
+        case PIPE_BLENDFACTOR_CONST_COLOR:
+            return R300_BLEND_GL_CONST_COLOR;
+        case PIPE_BLENDFACTOR_CONST_ALPHA:
+            return R300_BLEND_GL_CONST_ALPHA;
+        /* XXX WTF are these?
+        case PIPE_BLENDFACTOR_SRC1_COLOR:
+        case PIPE_BLENDFACTOR_SRC1_ALPHA: */
+        case PIPE_BLENDFACTOR_ZERO:
+            return R300_BLEND_GL_ZERO;
+        case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+            return R300_BLEND_GL_ONE_MINUS_SRC_COLOR;
+        case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+            return R300_BLEND_GL_ONE_MINUS_SRC_ALPHA;
+        case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+            return R300_BLEND_GL_ONE_MINUS_DST_ALPHA;
+        case PIPE_BLENDFACTOR_INV_DST_COLOR:
+            return R300_BLEND_GL_ONE_MINUS_DST_COLOR;
+        case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+            return R300_BLEND_GL_ONE_MINUS_CONST_COLOR;
+        case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+            return R300_BLEND_GL_ONE_MINUS_CONST_ALPHA;
+        /* XXX see above
+        case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+        case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: */
+        default:
+            debug_printf("r300: Unknown blend factor %d\n", blend_fact);
+            break;
+    }
+    return 0;
+}
+
+/* DSA state. */
+
+static INLINE uint32_t r300_translate_depth_stencil_function(int zs_func)
+{
+    switch (zs_func) {
+        case PIPE_FUNC_NEVER:
+            return R300_ZS_NEVER;
+        case PIPE_FUNC_LESS:
+            return R300_ZS_LESS;
+        case PIPE_FUNC_EQUAL:
+            return R300_ZS_EQUAL;
+        case PIPE_FUNC_LEQUAL:
+            return R300_ZS_LEQUAL;
+        case PIPE_FUNC_GREATER:
+            return R300_ZS_GREATER;
+        case PIPE_FUNC_NOTEQUAL:
+            return R300_ZS_NOTEQUAL;
+        case PIPE_FUNC_GEQUAL:
+            return R300_ZS_GEQUAL;
+        case PIPE_FUNC_ALWAYS:
+            return R300_ZS_ALWAYS;
+        default:
+            debug_printf("r300: Unknown depth/stencil function %d\n",
+                zs_func);
+            break;
+    }
+    return 0;
+}
+
+static INLINE uint32_t r300_translate_stencil_op(int s_op)
+{
+    switch (s_op) {
+        case PIPE_STENCIL_OP_KEEP:
+            return R300_ZS_KEEP;
+        case PIPE_STENCIL_OP_ZERO:
+            return R300_ZS_ZERO;
+        case PIPE_STENCIL_OP_REPLACE:
+            return R300_ZS_REPLACE;
+        case PIPE_STENCIL_OP_INCR:
+            return R300_ZS_INCR;
+        case PIPE_STENCIL_OP_DECR:
+            return R300_ZS_DECR;
+        case PIPE_STENCIL_OP_INCR_WRAP:
+            return R300_ZS_INCR_WRAP;
+        case PIPE_STENCIL_OP_DECR_WRAP:
+            return R300_ZS_DECR_WRAP;
+        case PIPE_STENCIL_OP_INVERT:
+            return R300_ZS_INVERT;
+        default:
+            debug_printf("r300: Unknown stencil op %d", s_op);
+            break;
+    }
+    return 0;
+}
+
+static INLINE uint32_t r300_translate_alpha_function(int alpha_func)
+{
+    switch (alpha_func) {
+        case PIPE_FUNC_NEVER:
+            return R300_FG_ALPHA_FUNC_NEVER;
+        case PIPE_FUNC_LESS:
+            return R300_FG_ALPHA_FUNC_LESS;
+        case PIPE_FUNC_EQUAL:
+            return R300_FG_ALPHA_FUNC_EQUAL;
+        case PIPE_FUNC_LEQUAL:
+            return R300_FG_ALPHA_FUNC_LE;
+        case PIPE_FUNC_GREATER:
+            return R300_FG_ALPHA_FUNC_GREATER;
+        case PIPE_FUNC_NOTEQUAL:
+            return R300_FG_ALPHA_FUNC_NOTEQUAL;
+        case PIPE_FUNC_GEQUAL:
+            return R300_FG_ALPHA_FUNC_GE;
+        case PIPE_FUNC_ALWAYS:
+            return R300_FG_ALPHA_FUNC_ALWAYS;
+        default:
+            debug_printf("r300: Unknown alpha function %d", alpha_func);
+            break;
+    }
+    return 0;
+}
+
+/* Texture sampler state. */
+
+static INLINE uint32_t r300_translate_wrap(int wrap)
+{
+    switch (wrap) {
+        case PIPE_TEX_WRAP_REPEAT:
+            return R300_TX_REPEAT;
+        case PIPE_TEX_WRAP_CLAMP:
+            return R300_TX_CLAMP;
+        case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+            return R300_TX_CLAMP_TO_EDGE;
+        case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+            return R300_TX_CLAMP_TO_BORDER;
+        case PIPE_TEX_WRAP_MIRROR_REPEAT:
+            return R300_TX_REPEAT | R300_TX_MIRRORED;
+        case PIPE_TEX_WRAP_MIRROR_CLAMP:
+            return R300_TX_CLAMP | R300_TX_MIRRORED;
+        case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+            return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
+        case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+            return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
+        default:
+            debug_printf("r300: Unknown texture wrap %d", wrap);
+            return 0;
+    }
+}
+
+static INLINE uint32_t r300_translate_tex_filters(int min, int mag, int mip)
+{
+    uint32_t retval = 0;
+    switch (min) {
+        case PIPE_TEX_FILTER_NEAREST:
+            retval |= R300_TX_MIN_FILTER_NEAREST;
+        case PIPE_TEX_FILTER_LINEAR:
+            retval |= R300_TX_MIN_FILTER_LINEAR;
+        case PIPE_TEX_FILTER_ANISO:
+            retval |= R300_TX_MIN_FILTER_ANISO;
+        default:
+            debug_printf("r300: Unknown texture filter %d", min);
+            break;
+    }
+    switch (mag) {
+        case PIPE_TEX_FILTER_NEAREST:
+            retval |= R300_TX_MAG_FILTER_NEAREST;
+        case PIPE_TEX_FILTER_LINEAR:
+            retval |= R300_TX_MAG_FILTER_LINEAR;
+        case PIPE_TEX_FILTER_ANISO:
+            retval |= R300_TX_MAG_FILTER_ANISO;
+        default:
+            debug_printf("r300: Unknown texture filter %d", mag);
+            break;
+    }
+    switch (mip) {
+        case PIPE_TEX_MIPFILTER_NONE:
+            retval |= R300_TX_MIN_FILTER_MIP_NONE;
+        case PIPE_TEX_MIPFILTER_NEAREST:
+            retval |= R300_TX_MIN_FILTER_MIP_NEAREST;
+        case PIPE_TEX_MIPFILTER_LINEAR:
+            retval |= R300_TX_MIN_FILTER_MIP_LINEAR;
+        default:
+            debug_printf("r300: Unknown texture filter %d", mip);
+            break;
+    }
+
+    return retval;
+}
+
+static INLINE uint32_t r300_anisotropy(float max_aniso)
+{
+    if (max_aniso >= 16.0f) {
+        return R300_TX_MAX_ANISO_16_TO_1;
+    } else if (max_aniso >= 8.0f) {
+        return R300_TX_MAX_ANISO_8_TO_1;
+    } else if (max_aniso >= 4.0f) {
+        return R300_TX_MAX_ANISO_4_TO_1;
+    } else if (max_aniso >= 2.0f) {
+        return R300_TX_MAX_ANISO_2_TO_1;
+    } else {
+        return R300_TX_MAX_ANISO_1_TO_1;
+    }
+}
+
+/* Buffer formats. */
+
 static INLINE uint32_t r300_translate_colorformat(enum pipe_format format)
 {
     switch (format) {
-    case PIPE_FORMAT_A8R8G8B8_UNORM:
-        return R300_COLOR_FORMAT_ARGB8888;
-    case PIPE_FORMAT_I8_UNORM:
-        return R300_COLOR_FORMAT_I8;
-    case PIPE_FORMAT_A1R5G5B5_UNORM:
-        return R300_COLOR_FORMAT_ARGB1555;
-    case PIPE_FORMAT_R5G6B5_UNORM:
-        return R300_COLOR_FORMAT_RGB565;
-    /* XXX Not in pipe_format
-    case PIPE_FORMAT_A32R32G32B32:
-        return R300_COLOR_FORMAT_ARGB32323232;
-    case PIPE_FORMAT_A16R16G16B16:
-        return R300_COLOR_FORMAT_ARGB16161616; */
-    case PIPE_FORMAT_A4R4G4B4_UNORM:
-        return R300_COLOR_FORMAT_ARGB4444;
-    /* XXX Not in pipe_format
-    case PIPE_FORMAT_A10R10G10B10_UNORM:
-        return R500_COLOR_FORMAT_ARGB10101010;
-    case PIPE_FORMAT_A2R10G10B10_UNORM:
-        return R500_COLOR_FORMAT_ARGB2101010;
-    case PIPE_FORMAT_I10_UNORM:
-        return R500_COLOR_FORMAT_I10; */
-    default:
-        debug_printf("r300: Implementation error: " \
-            "Got unsupported color format %s in %s\n",
-            pf_name(format), __FUNCTION__);
-        break;
+        /* 8-bit buffers */
+        case PIPE_FORMAT_I8_UNORM:
+            return R300_COLOR_FORMAT_I8;
+        /* 16-bit buffers */
+        case PIPE_FORMAT_R5G6B5_UNORM:
+            return R300_COLOR_FORMAT_RGB565;
+        case PIPE_FORMAT_A1R5G5B5_UNORM:
+            return R300_COLOR_FORMAT_ARGB1555;
+        case PIPE_FORMAT_A4R4G4B4_UNORM:
+            return R300_COLOR_FORMAT_ARGB4444;
+        /* 32-bit buffers */
+        case PIPE_FORMAT_A8R8G8B8_UNORM:
+            return R300_COLOR_FORMAT_ARGB8888;
+        /* XXX Not in pipe_format
+        case PIPE_FORMAT_A32R32G32B32:
+            return R300_COLOR_FORMAT_ARGB32323232;
+        case PIPE_FORMAT_A16R16G16B16:
+            return R300_COLOR_FORMAT_ARGB16161616; */
+        /* XXX Not in pipe_format
+        case PIPE_FORMAT_A10R10G10B10_UNORM:
+            return R500_COLOR_FORMAT_ARGB10101010;
+        case PIPE_FORMAT_A2R10G10B10_UNORM:
+            return R500_COLOR_FORMAT_ARGB2101010;
+        case PIPE_FORMAT_I10_UNORM:
+            return R500_COLOR_FORMAT_I10; */
+        default:
+            debug_printf("r300: Implementation error: " \
+                "Got unsupported color format %s in %s\n",
+                pf_name(format), __FUNCTION__);
+            break;
     }
-    
     return 0;
 }
 
 static INLINE uint32_t r300_translate_zsformat(enum pipe_format format)
 {
     switch (format) {
-    case PIPE_FORMAT_Z16_UNORM:
-        return R300_DEPTHFORMAT_16BIT_INT_Z;
-    /* XXX R300_DEPTHFORMAT_16BIT_13E3 anyone? */
-    case PIPE_FORMAT_Z24S8_UNORM:
-        return R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
-    default:
-        debug_printf("r300: Implementation error: " \
-            "Got unsupported ZS format %s in %s\n",
-            pf_name(format), __FUNCTION__);
-        break;
+        /* 16-bit depth, no stencil */
+        case PIPE_FORMAT_Z16_UNORM:
+            return R300_DEPTHFORMAT_16BIT_INT_Z;
+        /* 24-bit depth, 8-bit stencil */
+        case PIPE_FORMAT_Z24S8_UNORM:
+            return R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
+        default:
+            debug_printf("r300: Implementation error: " \
+                "Got unsupported ZS format %s in %s\n",
+                pf_name(format), __FUNCTION__);
+            break;
+    }
+    return 0;
+}
+
+/* Non-CSO state. (For now.) */
+
+static INLINE uint32_t r300_translate_gb_pipes(int pipe_count)
+{
+    switch (pipe_count) {
+        case 1:
+            return R300_GB_TILE_PIPE_COUNT_RV300;
+            break;
+        case 2:
+            return R300_GB_TILE_PIPE_COUNT_R300;
+            break;
+        case 3:
+            return R300_GB_TILE_PIPE_COUNT_R420_3P;
+            break;
+        case 4:
+            return R300_GB_TILE_PIPE_COUNT_R420;
+            break;
     }
-    
     return 0;
 }
 
diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
new file mode 100644
index 0000000000..7fd7aefeb7
--- /dev/null
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright 2009 Joakim Sindholt <opensource@zhasha.com>
+ *                Corbin Simpson <MostAwesomeDude@gmail.com>
+ *
+ * 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
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#include "r300_state_invariant.h"
+
+/* Calculate and emit invariant state. This is data that the 3D engine
+ * will probably want at the beginning of every CS, but it's not currently
+ * handled by any CSO setup, and in addition it doesn't really change much.
+ *
+ * Note that eventually this should be empty, but it's useful for development
+ * and general unduplication of code. */
+void r300_emit_invariant_state(struct r300_context* r300)
+{
+    struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps;
+    CS_LOCALS(r300);
+
+    BEGIN_CS(14);
+    /* Amount of time to wait for vertex fetches in PVS */
+    OUT_CS_REG(VAP_PVS_VTX_TIMEOUT_REG, 0xffff);
+    /* Various GB enables */
+    OUT_CS_REG(R300_GB_ENABLE, R300_GB_POINT_STUFF_ENABLE |
+        R300_GB_LINE_STUFF_ENABLE | R300_GB_TRIANGLE_STUFF_ENABLE);
+    /* Subpixel multisampling for AA */
+    OUT_CS_REG(R300_GB_MSPOS0, 0x66666666);
+    OUT_CS_REG(R300_GB_MSPOS1, 0x66666666);
+    /* GB tile config and pipe setup */
+    OUT_CS_REG(R300_GB_TILE_CONFIG, R300_GB_TILE_DISABLE |
+        r300_translate_gb_pipes(caps->num_frag_pipes));
+    /* Source of fog depth */
+    OUT_CS_REG(R300_GB_SELECT, R300_GB_FOG_SELECT_1_1_W);
+    /* AA enable */
+    OUT_CS_REG(R300_GB_AA_CONFIG, 0x0);
+    END_CS;
+}
diff --git a/src/gallium/drivers/r300/r300_state_invariant.h b/src/gallium/drivers/r300/r300_state_invariant.h
new file mode 100644
index 0000000000..8204bf9588
--- /dev/null
+++ b/src/gallium/drivers/r300/r300_state_invariant.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
+ *
+ * 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
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#ifndef R300_STATE_INVARIANT_H
+#define R300_STATE_INVARIANT_H
+
+#include "r300_context.h"
+#include "r300_cs.h"
+#include "r300_reg.h"
+#include "r300_state_inlines.h"
+
+void r300_emit_invariant_state(struct r300_context* r300);
+
+#endif /* R300_STATE_INVARIANT_H */
diff --git a/src/gallium/drivers/r300/r300_surface.c b/src/gallium/drivers/r300/r300_surface.c
index 49e4a96f83..288d8dea15 100644
--- a/src/gallium/drivers/r300/r300_surface.c
+++ b/src/gallium/drivers/r300/r300_surface.c
@@ -33,9 +33,10 @@ static void r300_surface_fill(struct pipe_context* pipe,
 {
     struct r300_context* r300 = r300_context(pipe);
     CS_LOCALS(r300);
-    struct r300_capabilities* caps = ((struct r300_screen*)pipe->screen)->caps;
+    struct r300_capabilities* caps = r300_screen(pipe->screen)->caps;
     struct r300_texture* tex = (struct r300_texture*)dest->texture;
     int i;
+
     float r, g, b, a;
     unsigned pixpitch = tex->stride / tex->tex.block.size;
     r = (float)((color >> 16) & 0xff) / 255.0f;
@@ -55,7 +56,14 @@ static void r300_surface_fill(struct pipe_context* pipe,
         return;
     }*/
 
-    BEGIN_CS(163 + (caps->is_r500 ? 22 : 14) + (caps->has_tcl ? 4 : 2));
+    r300_emit_invariant_state(r300);
+
+    r300_emit_blend_state(r300, &blend_clear_state);
+    r300_emit_blend_color_state(r300, &blend_color_clear_state);
+    r300_emit_dsa_state(r300, &dsa_clear_state);
+    r300_emit_rs_state(r300, &rs_clear_state);
+
+    BEGIN_CS(129 + (caps->is_r500 ? 22 : 14) + (caps->has_tcl ? 4 : 2));
     /* Flush PVS. */
     OUT_CS_REG(R300_VAP_PVS_STATE_FLUSH_REG, 0x0);
 
@@ -69,7 +77,12 @@ static void r300_surface_fill(struct pipe_context* pipe,
     OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, 0xFFFFFF);
     OUT_CS_REG(R300_VAP_VF_MIN_VTX_INDX, 0x0);
     /* XXX endian */
-    OUT_CS_REG(R300_VAP_CNTL_STATUS, R300_VC_NO_SWAP);
+    if (caps->has_tcl) {
+        OUT_CS_REG(R300_VAP_CNTL_STATUS, R300_VC_NO_SWAP);
+    } else {
+        OUT_CS_REG(R300_VAP_CNTL_STATUS, R300_VC_NO_SWAP |
+                R300_VAP_TCL_BYPASS);
+    }
     OUT_CS_REG(R300_VAP_PROG_STREAM_CNTL_0, 0x0);
     /* XXX magic number not in r300_reg */
     OUT_CS_REG(R300_VAP_PSC_SGN_NORM_CNTL, 0xAAAAAAAA);
@@ -79,18 +92,6 @@ static void r300_surface_fill(struct pipe_context* pipe,
     OUT_CS_32F(1.0);
     OUT_CS_32F(1.0);
     OUT_CS_32F(1.0);
-    /* XXX is this too long? */
-    OUT_CS_REG(VAP_PVS_VTX_TIMEOUT_REG, 0xFFFF);
-    OUT_CS_REG(R300_GB_ENABLE, R300_GB_POINT_STUFF_ENABLE |
-        R300_GB_LINE_STUFF_ENABLE | R300_GB_TRIANGLE_STUFF_ENABLE);
-    /* XXX more magic numbers */
-    OUT_CS_REG(R300_GB_MSPOS0, 0x66666666);
-    OUT_CS_REG(R300_GB_MSPOS1, 0x66666666);
-    /* XXX why doesn't classic Mesa write the number of pipes, too? */
-    OUT_CS_REG(R300_GB_TILE_CONFIG, R300_GB_TILE_ENABLE |
-        R300_GB_TILE_SIZE_16);
-    OUT_CS_REG(R300_GB_SELECT, R300_GB_FOG_SELECT_1_1_W);
-    OUT_CS_REG(R300_GB_AA_CONFIG, 0x0);
     /* XXX point tex stuffing */
     OUT_CS_REG_SEQ(R300_GA_POINT_S0, 1);
     OUT_CS_32F(0.0);
@@ -102,9 +103,6 @@ static void r300_surface_fill(struct pipe_context* pipe,
     OUT_CS_REG(R300_GA_POINT_MINMAX, 0x6 |
         (0x1800 << R300_GA_POINT_MINMAX_MAX_SHIFT));
     /* XXX this big chunk should be refactored into rs_state */
-    OUT_CS_REG(R300_GA_LINE_CNTL, 0x00030006);
-    OUT_CS_REG(R300_GA_LINE_STIPPLE_CONFIG, 0x3BAAAAAB);
-    OUT_CS_REG(R300_GA_LINE_STIPPLE_VALUE, 0x00000000);
     OUT_CS_REG(R300_GA_LINE_S0, 0x00000000);
     OUT_CS_REG(R300_GA_LINE_S1, 0x3F800000);
     OUT_CS_REG(R300_GA_ENHANCE, 0x00000002);
@@ -117,12 +115,6 @@ static void r300_surface_fill(struct pipe_context* pipe,
     OUT_CS_REG(R300_GA_FOG_SCALE, 0x3DBF1412);
     OUT_CS_REG(R300_GA_FOG_OFFSET, 0x00000000);
     OUT_CS_REG(R300_SU_TEX_WRAP, 0x00000000);
-    OUT_CS_REG(R300_SU_POLY_OFFSET_FRONT_SCALE, 0x00000000);
-    OUT_CS_REG(R300_SU_POLY_OFFSET_FRONT_OFFSET, 0x00000000);
-    OUT_CS_REG(R300_SU_POLY_OFFSET_BACK_SCALE, 0x00000000);
-    OUT_CS_REG(R300_SU_POLY_OFFSET_BACK_OFFSET, 0x00000000);
-    OUT_CS_REG(R300_SU_POLY_OFFSET_ENABLE, 0x00000000);
-    OUT_CS_REG(R300_SU_CULL_MODE, 0x00000000);
     OUT_CS_REG(R300_SU_DEPTH_SCALE, 0x4B7FFFFF);
     OUT_CS_REG(R300_SU_DEPTH_OFFSET, 0x00000000);
     OUT_CS_REG(R300_SC_HYPERZ, 0x0000001C);
@@ -190,11 +182,6 @@ static void r300_surface_fill(struct pipe_context* pipe,
             R300_PS_UCP_MODE_CLIP_AS_TRIFAN);
     }
 
-    /* The size of the point we're about to draw, in sixths of pixels */
-    OUT_CS_REG(R300_GA_POINT_SIZE,
-        ((h * 6) & R300_POINTSIZE_Y_MASK) |
-        ((w * 6) << R300_POINTSIZE_X_SHIFT));
-
     /* XXX */
     OUT_CS_REG(R300_SC_CLIP_RULE, 0xaaaa);
 
@@ -240,7 +227,7 @@ static void r300_surface_fill(struct pipe_context* pipe,
         r300_emit_fragment_shader(r300, &r300_passthrough_fragment_shader);
     }
 
-    BEGIN_CS(8 + (caps->has_tcl ? 20 : 2));
+    BEGIN_CS(7 + (caps->has_tcl ? 21 : 2));
     OUT_CS_REG_SEQ(R300_US_OUT_FMT_0, 4);
     OUT_CS(R300_C0_SEL_B | R300_C1_SEL_G | R300_C2_SEL_R | R300_C3_SEL_A);
     OUT_CS(R300_US_OUT_FMT_UNUSED);
@@ -277,9 +264,10 @@ static void r300_surface_fill(struct pipe_context* pipe,
     }
     END_CS;
 
-    r300_emit_blend_state(r300, &blend_clear_state);
-    r300_emit_blend_color_state(r300, &blend_color_clear_state);
-    r300_emit_dsa_state(r300, &dsa_clear_state);
+    /* The size of the point we're about to draw, in sixths of pixels */
+    OUT_CS_REG(R300_GA_POINT_SIZE,
+        ((h * 6) & R300_POINTSIZE_Y_MASK) |
+        ((w * 6) << R300_POINTSIZE_X_SHIFT));
 
     BEGIN_CS(24);
     /* Flush colorbuffer and blend caches. */
diff --git a/src/gallium/drivers/r300/r300_surface.h b/src/gallium/drivers/r300/r300_surface.h
index 442eac2cf2..807aad39e4 100644
--- a/src/gallium/drivers/r300/r300_surface.h
+++ b/src/gallium/drivers/r300/r300_surface.h
@@ -57,4 +57,16 @@ const struct r300_dsa_state dsa_clear_state = {
     .stencil_ref_bf = 0x0,
 };
 
+const struct r300_rs_state rs_clear_state = {
+    .line_control = 0x00030006,
+    .depth_scale_front = 0x0,
+    .depth_offset_front = 0x0,
+    .depth_scale_back = 0x0,
+    .depth_offset_back = 0x0,
+    .polygon_offset_enable = 0x0,
+    .cull_mode = 0x0,
+    .line_stipple_config = 0x3BAAAAAB,
+    .line_stipple_value = 0x0,
+};
+
 #endif /* R300_SURFACE_H */
diff --git a/src/gallium/drivers/r300/r300_swtcl_emit.c b/src/gallium/drivers/r300/r300_swtcl_emit.c
index 5b028aaf7b..f9baaade1e 100644
--- a/src/gallium/drivers/r300/r300_swtcl_emit.c
+++ b/src/gallium/drivers/r300/r300_swtcl_emit.c
@@ -27,6 +27,7 @@
 #include "r300_cs.h"
 #include "r300_context.h"
 #include "r300_reg.h"
+#include "r300_state_derived.h"
 
 /* r300_swtcl_emit: Vertex and index buffer primitive emission. No HW TCL. */
 
@@ -91,9 +92,9 @@ static boolean r300_swtcl_render_allocate_vertices(struct vbuf_render* render,
     r300render->vertex_size = vertex_size;
 
     if (r300render->vbo) {
-        return true;
+        return TRUE;
     } else {
-        return false;
+        return FALSE;
     }
 }
 
@@ -167,11 +168,11 @@ static boolean r300_swtcl_render_set_primitive(struct vbuf_render* render,
             r300render->hwprim = R300_VAP_VF_CNTL__PRIM_POLYGON;
             break;
         default:
-            return false;
+            return FALSE;
             break;
     }
 
-    return true;
+    return TRUE;
 }
 
 static void prepare_render(struct r300_swtcl_render* render)
@@ -191,12 +192,15 @@ static void prepare_render(struct r300_swtcl_render* render)
      * PACKET3 [3D_LOAD_VBPNTR]
      * COUNT   [1]
      * FORMAT  [size | stride << 8]
+     * OFFSET  [0]
      * VBPNTR  [relocated BO]
      */
     BEGIN_CS(5);
     OUT_CS(CP_PACKET3(R300_PACKET3_3D_LOAD_VBPNTR, 3));
     OUT_CS(1);
-    OUT_CS(r300->vertex_info.vinfo.size | (r300->vertex_info.vinfo.size << 8));
+    OUT_CS(r300->vertex_info.vinfo.size |
+            (r300->vertex_info.vinfo.size << 8));
+    OUT_CS(render->vbo_offset);
     OUT_CS_RELOC(render->vbo, 0, RADEON_GEM_DOMAIN_GTT, 0, 0);
     END_CS;
 }
@@ -218,7 +222,7 @@ static void r300_swtcl_render_draw_arrays(struct vbuf_render* render,
     BEGIN_CS(2);
     OUT_CS(CP_PACKET3(R300_PACKET3_3D_DRAW_VBUF_2, 0));
     OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST | (count << 16) |
-           r300render->hwprim | R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
+           r300render->hwprim);
     END_CS;
 }
 
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index edd4370663..8251a597ea 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -31,8 +31,9 @@ static void r300_setup_miptree(struct r300_texture* tex)
 {
     struct pipe_texture* base = &tex->tex;
     int stride, size, offset;
+    int i;
 
-    for (int i = 0; i <= base->last_level; i++) {
+    for (i = 0; i <= base->last_level; i++) {
         if (i > 0) {
             base->width[i] = minify(base->width[i-1]);
             base->height[i] = minify(base->height[i-1]);
diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
index d56eed80a4..eef6e5806c 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
@@ -60,6 +60,7 @@ struct softpipe_vbuf_render
    struct softpipe_context *softpipe;
    uint prim;
    uint vertex_size;
+   uint nr_vertices;
    uint vertex_buffer_size;
    void *vertex_buffer;
 };
@@ -95,12 +96,44 @@ sp_vbuf_allocate_vertices(struct vbuf_render *vbr,
    }
 
    cvbr->vertex_size = vertex_size;
+   cvbr->nr_vertices = nr_vertices;
+   
    return cvbr->vertex_buffer != NULL;
 }
 
 static void
 sp_vbuf_release_vertices(struct vbuf_render *vbr)
 {
+#if 0
+   {
+      struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
+      const struct vertex_info *info = 
+         softpipe_get_vbuf_vertex_info(cvbr->softpipe);
+      const float *vtx = (const float *) cvbr->vertex_buffer;
+      uint i, j;
+      debug_printf("%s (vtx_size = %u,  vtx_used = %u)\n",
+             __FUNCTION__, cvbr->vertex_size, cvbr->nr_vertices);
+      for (i = 0; i < cvbr->nr_vertices; i++) {
+         for (j = 0; j < info->num_attribs; j++) {
+            uint k;
+            switch (info->attrib[j].emit) {
+            case EMIT_4F:  k = 4;   break;
+            case EMIT_3F:  k = 3;   break;
+            case EMIT_2F:  k = 2;   break;
+            case EMIT_1F:  k = 1;   break;
+            default: assert(0);
+            }
+            debug_printf("Vert %u attr %u: ", i, j);
+            while (k-- > 0) {
+               debug_printf("%g ", vtx[0]);
+               vtx++;
+            }
+            debug_printf("\n");
+         }
+      }
+   }
+#endif
+
    /* keep the old allocation for next time */
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index 28a9784b16..142ce230fc 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -243,9 +243,11 @@ softpipe_get_tex_surface(struct pipe_screen *screen,
       ps->level = level;
       ps->zslice = zslice;
 
-      if (pt->target == PIPE_TEXTURE_CUBE || pt->target == PIPE_TEXTURE_3D) {
-         ps->offset += ((pt->target == PIPE_TEXTURE_CUBE) ? face : zslice) *
-            pt->nblocksy[level] * spt->stride[level];
+      if (pt->target == PIPE_TEXTURE_CUBE) {
+         ps->offset += face * pt->nblocksy[level] * spt->stride[level];
+      }
+      else if (pt->target == PIPE_TEXTURE_3D) {
+         ps->offset += zslice * pt->nblocksy[level] * spt->stride[level];
       }
       else {
          assert(face == 0);
@@ -283,14 +285,13 @@ softpipe_get_tex_transfer(struct pipe_screen *screen,
 {
    struct softpipe_texture *sptex = softpipe_texture(texture);
    struct softpipe_transfer *spt;
-   struct pipe_transfer *pt;
 
    assert(texture);
    assert(level <= texture->last_level);
 
    spt = CALLOC_STRUCT(softpipe_transfer);
-   pt = &spt->base;
    if (spt) {
+      struct pipe_transfer *pt = &spt->base;
       pt->refcount = 1;
       pipe_texture_reference(&pt->texture, texture);
       pt->format = texture->format;
@@ -302,23 +303,26 @@ softpipe_get_tex_transfer(struct pipe_screen *screen,
       pt->nblocksx = texture->nblocksx[level];
       pt->nblocksy = texture->nblocksy[level];
       pt->stride = sptex->stride[level];
-      spt->offset = sptex->level_offset[level];
       pt->usage = usage;
       pt->face = face;
       pt->level = level;
       pt->zslice = zslice;
 
-      if (texture->target == PIPE_TEXTURE_CUBE ||
-          texture->target == PIPE_TEXTURE_3D) {
-         spt->offset += ((texture->target == PIPE_TEXTURE_CUBE) ? face :
-                         zslice) * pt->nblocksy * pt->stride;
+      spt->offset = sptex->level_offset[level];
+
+      if (texture->target == PIPE_TEXTURE_CUBE) {
+         spt->offset += face * pt->nblocksy * pt->stride;
+      }
+      else if (texture->target == PIPE_TEXTURE_3D) {
+         spt->offset += zslice * pt->nblocksy * pt->stride;
       }
       else {
          assert(face == 0);
          assert(zslice == 0);
       }
+      return pt;
    }
-   return pt;
+   return NULL;
 }
 
 
diff --git a/src/gallium/drivers/trace/tr_dump.c b/src/gallium/drivers/trace/tr_dump.c
index d98cef221b..8757ac8bec 100644
--- a/src/gallium/drivers/trace/tr_dump.c
+++ b/src/gallium/drivers/trace/tr_dump.c
@@ -40,7 +40,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #include <stdlib.h>
 #endif
 
@@ -236,7 +236,7 @@ boolean trace_dump_trace_begin()
       trace_dump_writes("<?xml-stylesheet type='text/xsl' href='trace.xsl'?>\n");
       trace_dump_writes("<trace version='0.1'>\n");
       
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
       /* Linux applications rarely cleanup GL / Gallium resources so catch 
        * application exit here */ 
       atexit(trace_dump_trace_close);
diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h
index 05cbd2fc4d..7f7657031d 100644
--- a/src/gallium/include/pipe/p_config.h
+++ b/src/gallium/include/pipe/p_config.h
@@ -111,6 +111,10 @@
 #define PIPE_OS_LINUX
 #endif
 
+#if defined(__FreeBSD__)
+#define PIPE_OS_BSD
+#endif
+
 #if defined(_WIN32) || defined(WIN32)
 #define PIPE_OS_WINDOWS
 #endif
@@ -122,9 +126,9 @@
  * NOTE: There is no way to auto-detect most of these.
  */
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 #define PIPE_SUBSYSTEM_DRI
-#endif /* PIPE_OS_LINUX */
+#endif /* PIPE_OS_LINUX || PIPE_OS_BSD */
 
 #if defined(PIPE_OS_WINDOWS)
 #if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
diff --git a/src/gallium/include/pipe/p_thread.h b/src/gallium/include/pipe/p_thread.h
index 8af3cd958b..e59b999b9a 100644
--- a/src/gallium/include/pipe/p_thread.h
+++ b/src/gallium/include/pipe/p_thread.h
@@ -38,7 +38,7 @@
 #include "pipe/p_compiler.h"
 
 
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
 
 #include <pthread.h> /* POSIX threads headers */
 #include <stdio.h> /* for perror() */
@@ -210,7 +210,7 @@ typedef unsigned pipe_condvar;
  */
 
 typedef struct {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    pthread_key_t key;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    DWORD key;
@@ -225,7 +225,7 @@ typedef struct {
 static INLINE void
 pipe_tsd_init(pipe_tsd *tsd)
 {
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    if (pthread_key_create(&tsd->key, NULL/*free*/) != 0) {
       perror("pthread_key_create(): failed to allocate key for thread specific data");
       exit(-1);
@@ -242,7 +242,7 @@ pipe_tsd_get(pipe_tsd *tsd)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    return pthread_getspecific(tsd->key);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    assert(0);
@@ -259,7 +259,7 @@ pipe_tsd_set(pipe_tsd *tsd, void *value)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
    if (pthread_setspecific(tsd->key, value) != 0) {
       perror("pthread_set_specific() failed");
       exit(-1);
diff --git a/src/gallium/include/state_tracker/drm_api.h b/src/gallium/include/state_tracker/drm_api.h
index 54480fa047..91a727be66 100644
--- a/src/gallium/include/state_tracker/drm_api.h
+++ b/src/gallium/include/state_tracker/drm_api.h
@@ -2,14 +2,18 @@
 #ifndef _DRM_API_H_
 #define _DRM_API_H_
 
+#include "pipe/p_compiler.h"
+
 struct pipe_screen;
 struct pipe_winsys;
+struct pipe_buffer;
 struct pipe_context;
+struct pipe_texture;
 
 struct drm_api
 {
 	/**
-	 * Special buffer function
+	 * Special buffer functions
 	 */
 	/*@{*/
 	struct pipe_screen*  (*create_screen)(int drmFB, int pciID);
@@ -17,11 +21,21 @@ struct drm_api
 	/*@}*/
 
 	/**
-	 * Special buffer function
+	 * Special buffer functions
 	 */
 	/*@{*/
-	struct pipe_buffer* (*buffer_from_handle)(struct pipe_winsys *winsys, const char *name, unsigned handle);
-	unsigned (*handle_from_buffer)(struct pipe_winsys *winsys, struct pipe_buffer *buffer);
+	boolean (*buffer_from_texture)(struct pipe_texture *texture,
+                                       struct pipe_buffer **buffer,
+                                       unsigned *stride);
+	struct pipe_buffer* (*buffer_from_handle)(struct pipe_screen *screen,
+                                                  const char *name,
+                                                  unsigned handle);
+	boolean (*handle_from_buffer)(struct pipe_screen *screen,
+                                      struct pipe_buffer *buffer,
+                                      unsigned *handle);
+	boolean (*global_handle_from_buffer)(struct pipe_screen *screen,
+                                             struct pipe_buffer *buffer,
+                                             unsigned *handle);
 	/*@}*/
 };
 
diff --git a/src/gallium/state_trackers/egl/Makefile b/src/gallium/state_trackers/egl/Makefile
index ea4cec0bb8..692a3c8b76 100644
--- a/src/gallium/state_trackers/egl/Makefile
+++ b/src/gallium/state_trackers/egl/Makefile
@@ -6,14 +6,15 @@ TOP        = ../../../..
 
 include ${TOP}/configs/current
 
-CFLAGS += -g -Wall -Werror-implicit-function-declaration -fPIC \
+CFLAGS := \
           -I${GALLIUMDIR}/include \
           -I${GALLIUMDIR}/auxiliary \
           -I${TOP}/src/mesa/drivers/dri/common \
           -I${TOP}/src/mesa \
           -I$(TOP)/include \
           -I$(TOP)/src/egl/main \
-          ${LIBDRM_CFLAGS}
+          ${LIBDRM_CFLAGS} \
+          ${CFLAGS}
 
 #############################################
 
diff --git a/src/gallium/state_trackers/egl/egl_surface.c b/src/gallium/state_trackers/egl/egl_surface.c
index 281dff9f8a..aa3c32d41c 100644
--- a/src/gallium/state_trackers/egl/egl_surface.c
+++ b/src/gallium/state_trackers/egl/egl_surface.c
@@ -132,7 +132,7 @@ drm_create_texture(_EGLDriver *drv,
 	scrn->front.width = w;
 	scrn->front.height = h;
 	scrn->front.pitch = pitch;
-	scrn->front.handle = drm_api_hocks.handle_from_buffer(dev->winsys, scrn->buffer);
+	drm_api_hocks.handle_from_buffer(dev->winsys, scrn->buffer, &scrn->front.handle);
 	if (0)
 		goto err_handle;
 
diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.c b/src/gallium/state_trackers/wgl/icd/stw_icd.c
index 8ae6aa1f3e..faf7f2f410 100644
--- a/src/gallium/state_trackers/wgl/icd/stw_icd.c
+++ b/src/gallium/state_trackers/wgl/icd/stw_icd.c
@@ -36,6 +36,7 @@
 #include "shared/stw_public.h"
 #include "icd/stw_icd.h"
 
+#define DBG 0
 
 static GLCLTPROCTABLE cpt;
 static boolean cpt_initialized = FALSE;
@@ -81,7 +82,8 @@ DrvDescribeLayerPlane(
    UINT nBytes,
    LPLAYERPLANEDESCRIPTOR plpd )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG) 
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return FALSE;
 }
@@ -97,8 +99,9 @@ DrvDescribePixelFormat(
 
    r = stw_pixelformat_describe( hdc, iPixelFormat, cjpfd, ppfd );
 
-   debug_printf( "%s( %p, %d, %u, %p ) = %d\n",
-                 __FUNCTION__, hdc, iPixelFormat, cjpfd, ppfd, r );
+   if (DBG)
+      debug_printf( "%s( %p, %d, %u, %p ) = %d\n",
+                    __FUNCTION__, hdc, iPixelFormat, cjpfd, ppfd, r );
 
    return r;
 }
@@ -111,7 +114,8 @@ DrvGetLayerPaletteEntries(
    INT cEntries,
    COLORREF *pcr )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG)
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return 0;
 }
@@ -124,7 +128,8 @@ DrvGetProcAddress(
 
    r = stw_get_proc_address( lpszProc );
 
-   debug_printf( "%s( \", __FUNCTION__%s\" ) = %p\n", lpszProc, r );
+   if (DBG)
+      debug_printf( "%s( \", __FUNCTION__%s\" ) = %p\n", lpszProc, r );
 
    return r;
 }
@@ -135,7 +140,8 @@ DrvRealizeLayerPalette(
    INT iLayerPlane,
    BOOL bRealize )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG)
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return FALSE;
 }
@@ -152,7 +158,8 @@ DrvSetCallbackProcs(
    INT nProcs,
    PROC *pProcs )
 {
-   debug_printf( "%s( %d, %p )\n", __FUNCTION__, nProcs, pProcs );
+   if (DBG)
+      debug_printf( "%s( %d, %p )\n", __FUNCTION__, nProcs, pProcs );
 
    return;
 }
@@ -510,8 +517,9 @@ DrvSetContext(
    DHGLRC dhglrc,
    PFN_SETPROCTABLE pfnSetProcTable )
 {
-   debug_printf( "%s( 0x%p, %u, 0x%p )\n", 
-                 __FUNCTION__, hdc, dhglrc, pfnSetProcTable );
+   if (DBG)
+      debug_printf( "%s( 0x%p, %u, 0x%p )\n", 
+                    __FUNCTION__, hdc, dhglrc, pfnSetProcTable );
 
    /* Although WGL allows different dispatch entrypoints per 
     */
@@ -534,7 +542,8 @@ DrvSetLayerPaletteEntries(
    INT cEntries,
    CONST COLORREF *pcr )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG)
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return 0;
 }
@@ -548,7 +557,8 @@ DrvSetPixelFormat(
 
    r = stw_pixelformat_set( hdc, iPixelFormat );
 
-   debug_printf( "%s( %p, %d ) = %s\n", __FUNCTION__, hdc, iPixelFormat, r ? "TRUE" : "FALSE" );
+   if (DBG)
+      debug_printf( "%s( %p, %d ) = %s\n", __FUNCTION__, hdc, iPixelFormat, r ? "TRUE" : "FALSE" );
 
    return r;
 }
@@ -558,7 +568,8 @@ DrvShareLists(
    DHGLRC dhglrc1,
    DHGLRC dhglrc2 )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG)
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return FALSE;
 }
@@ -567,7 +578,8 @@ BOOL APIENTRY
 DrvSwapBuffers(
    HDC hdc )
 {
-   debug_printf( "%s( %p )\n", __FUNCTION__, hdc );
+   if (DBG)
+      debug_printf( "%s( %p )\n", __FUNCTION__, hdc );
 
    return stw_swap_buffers( hdc );
 }
@@ -577,7 +589,8 @@ DrvSwapLayerBuffers(
    HDC hdc,
    UINT fuPlanes )
 {
-   debug_printf( "%s\n", __FUNCTION__ );
+   if (DBG)
+      debug_printf( "%s\n", __FUNCTION__ );
 
    return FALSE;
 }
@@ -586,7 +599,8 @@ BOOL APIENTRY
 DrvValidateVersion(
    ULONG ulVersion )
 {
-   debug_printf( "%s( %u )\n", __FUNCTION__, ulVersion );
+   if (DBG)
+      debug_printf( "%s( %u )\n", __FUNCTION__, ulVersion );
 
    /* TODO: get the expected version from the winsys */
    
diff --git a/src/gallium/winsys/drm/intel/gem/intel_be_api.c b/src/gallium/winsys/drm/intel/gem/intel_be_api.c
index 6cffed5134..3f71c25441 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_be_api.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_be_api.c
@@ -1,12 +1,15 @@
 
 #include "intel_be_api.h"
+#include "i915simple/i915_winsys.h"
 
 struct drm_api drm_api_hocks =
 {
 	/* intel_be_context.c */
 	.create_context = intel_be_create_context,
-	/* intel_be_screen.c */
+	/* intel_be_device.c */
 	.create_screen = intel_be_create_screen,
+	.buffer_from_texture = i915_get_texture_buffer,
 	.buffer_from_handle = intel_be_buffer_from_handle,
 	.handle_from_buffer = intel_be_handle_from_buffer,
+	.global_handle_from_buffer = intel_be_global_handle_from_buffer,
 };
diff --git a/src/gallium/winsys/drm/intel/gem/intel_be_device.c b/src/gallium/winsys/drm/intel/gem/intel_be_device.c
index a2163a1e6d..595de44726 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_be_device.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_be_device.c
@@ -9,7 +9,7 @@
 
 #include "intel_be_fence.h"
 
-#include "i915simple/i915_screen.h"
+#include "i915simple/i915_winsys.h"
 
 #include "intel_be_api.h"
 
@@ -70,6 +70,8 @@ intel_be_buffer_create(struct pipe_winsys *winsys,
 	buffer->base.alignment = alignment;
 	buffer->base.usage = usage;
 	buffer->base.size = size;
+	buffer->flinked = FALSE;
+	buffer->flink = 0;
 
 	if (usage & (PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_CONSTANT)) {
 		/* Local buffer */
@@ -133,10 +135,10 @@ err:
 }
 
 struct pipe_buffer *
-intel_be_buffer_from_handle(struct pipe_winsys *winsys,
+intel_be_buffer_from_handle(struct pipe_screen *screen,
                             const char* name, unsigned handle)
 {
-	struct intel_be_device *dev = intel_be_device(winsys);
+	struct intel_be_device *dev = intel_be_device(screen->winsys);
 	struct intel_be_buffer *buffer = CALLOC_STRUCT(intel_be_buffer);
 
 	if (!buffer)
@@ -162,14 +164,39 @@ err:
 	return NULL;
 }
 
-unsigned
-intel_be_handle_from_buffer(struct pipe_winsys *winsys,
-                            struct pipe_buffer *buf)
+boolean
+intel_be_handle_from_buffer(struct pipe_screen *screen,
+                            struct pipe_buffer *buffer,
+                            unsigned *handle)
 {
-	drm_intel_bo *bo = intel_bo(buf);
-	return bo->handle;
+	drm_intel_bo *bo;
+
+	if (!buffer)
+		return FALSE;
+
+	*handle = intel_bo(buffer)->handle;
+	return TRUE;
 }
 
+boolean
+intel_be_global_handle_from_buffer(struct pipe_screen *screen,
+				   struct pipe_buffer *buffer,
+				   unsigned *handle)
+{
+	struct intel_be_buffer *buf = intel_be_buffer(buffer);
+
+	if (!buffer)
+		return FALSE;
+
+	if (!buf->flinked) {
+		if (drm_intel_bo_flink(intel_bo(buffer), &buf->flink))
+			return FALSE;
+		buf->flinked = TRUE;
+	}
+
+	*handle = buf->flink;
+	return TRUE;
+}
 /*
  * Fence
  */
diff --git a/src/gallium/winsys/drm/intel/gem/intel_be_device.h b/src/gallium/winsys/drm/intel/gem/intel_be_device.h
index c4837e65fa..47d2176cb4 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_be_device.h
+++ b/src/gallium/winsys/drm/intel/gem/intel_be_device.h
@@ -44,6 +44,8 @@ intel_be_init_device(struct intel_be_device *device, int fd, unsigned id);
 struct intel_be_buffer {
 	struct pipe_buffer base;
 	drm_intel_bo *bo;
+	boolean flinked;
+	unsigned flink;
 };
 
 /**
@@ -52,7 +54,7 @@ struct intel_be_buffer {
  * Takes a reference.
  */
 struct pipe_buffer *
-intel_be_buffer_from_handle(struct pipe_winsys *winsys,
+intel_be_buffer_from_handle(struct pipe_screen *screen,
                             const char* name, unsigned handle);
 
 /**
@@ -60,9 +62,20 @@ intel_be_buffer_from_handle(struct pipe_winsys *winsys,
  *
  * If buffer is destroyed handle may become invalid.
  */
-unsigned
-intel_be_handle_from_buffer(struct pipe_winsys *winsys,
-                            struct pipe_buffer *buffer);
+boolean
+intel_be_handle_from_buffer(struct pipe_screen *screen,
+                            struct pipe_buffer *buffer,
+                            unsigned *handle);
+
+/**
+ * Gets the global handle from a buffer.
+ *
+ * If buffer is destroyed handle may become invalid.
+ */
+boolean
+intel_be_global_handle_from_buffer(struct pipe_screen *screen,
+                                   struct pipe_buffer *buffer,
+                                   unsigned *handle);
 
 static INLINE struct intel_be_buffer *
 intel_be_buffer(struct pipe_buffer *buf)
diff --git a/src/gallium/winsys/drm/radeon/Makefile b/src/gallium/winsys/drm/radeon/Makefile
index dca1e3233a..784686ffb6 100644
--- a/src/gallium/winsys/drm/radeon/Makefile
+++ b/src/gallium/winsys/drm/radeon/Makefile
@@ -13,6 +13,7 @@ PIPE_DRIVERS = \
 DRIVER_SOURCES = \
 	radeon_buffer.c \
 	radeon_context.c \
+	radeon_drm.c \
 	radeon_r300.c \
 	radeon_screen.c \
 	radeon_winsys_softpipe.c
diff --git a/src/gallium/winsys/drm/radeon/radeon_drm.c b/src/gallium/winsys/drm/radeon/radeon_drm.c
new file mode 100644
index 0000000000..839b0bf0bd
--- /dev/null
+++ b/src/gallium/winsys/drm/radeon/radeon_drm.c
@@ -0,0 +1,30 @@
+/* 
+ * Copyright © 2009 Corbin Simpson
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
+ * AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ */
+/*
+ * Authors:
+ *      Corbin Simpson <MostAwesomeDude@gmail.com>
+ */
+#include "radeon_drm.h"
diff --git a/src/gallium/winsys/drm/radeon/radeon_drm.h b/src/gallium/winsys/drm/radeon/radeon_drm.h
new file mode 100644
index 0000000000..8ccbc81895
--- /dev/null
+++ b/src/gallium/winsys/drm/radeon/radeon_drm.h
@@ -0,0 +1,35 @@
+/* 
+ * Copyright © 2009 Corbin Simpson
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
+ * AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ */
+/*
+ * Authors:
+ *      Corbin Simpson <MostAwesomeDude@gmail.com>
+ */
+#ifndef RADEON_DRM_H
+#define RADEON_DRM_H
+
+#include "state_tracker/drm_api.h"
+
+#endif
diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c
index 40d683234f..acfb8ec4ea 100644
--- a/src/gallium/winsys/g3dvl/xsp_winsys.c
+++ b/src/gallium/winsys/g3dvl/xsp_winsys.c
@@ -1,9 +1,10 @@
 #include "vl_winsys.h"
 #include <X11/Xutil.h>
-#include <pipe/p_winsys.h>
+#include <pipe/internal/p_winsys_screen.h>
 #include <pipe/p_state.h>
 #include <pipe/p_inlines.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <softpipe/sp_winsys.h>
 
 /* pipe_winsys implementation */
@@ -96,12 +97,6 @@ static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buff
 	free(xsp_buf);
 }
 
-/* Borrowed from Mesa's xm_winsys */
-static unsigned int round_up(unsigned n, unsigned multiple)
-{
-   return (n + multiple - 1) & ~(multiple - 1);
-}
-
 static struct pipe_buffer* xsp_surface_buffer_create
 (
 	struct pipe_winsys *pws,
@@ -119,11 +114,11 @@ static struct pipe_buffer* xsp_surface_buffer_create
 	pf_get_block(format, &block);
 	nblocksx = pf_get_nblocksx(&block, width);
 	nblocksy = pf_get_nblocksy(&block, height);
-	*stride = round_up(nblocksx * block.size, ALIGNMENT);
+	*stride = align(nblocksx * block.size, ALIGNMENT);
 
-	return winsys->buffer_create(winsys, ALIGNMENT,
-				     usage,
-				     *stride * nblocksy);
+	return pws->buffer_create(pws, ALIGNMENT,
+				  usage,
+				  *stride * nblocksy);
 }
 
 static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence)
diff --git a/src/gallium/winsys/xlib/xlib_cell.c b/src/gallium/winsys/xlib/xlib_cell.c
index c87564f4dc..40bcdfe42a 100644
--- a/src/gallium/winsys/xlib/xlib_cell.c
+++ b/src/gallium/winsys/xlib/xlib_cell.c
@@ -222,7 +222,8 @@ xm_flush_frontbuffer(struct pipe_winsys *pws,
     * This function copies that XImage to the actual X Window.
     */
    XMesaContext xmctx = (XMesaContext) context_private;
-   xlib_cell_display_surface(xmctx->xm_buffer, surf);
+   if (xmctx)
+      xlib_cell_display_surface(xmctx->xm_buffer, surf);
 }
 
 
diff --git a/src/glx/x11/glxext.c b/src/glx/x11/glxext.c
index be6edf9b19..b296b7c651 100644
--- a/src/glx/x11/glxext.c
+++ b/src/glx/x11/glxext.c
@@ -56,19 +56,6 @@
 void __glXDumpDrawBuffer(__GLXcontext * ctx);
 #endif
 
-#ifdef USE_SPARC_ASM
-static void _glx_mesa_init_sparc_glapi_relocs(void);
-static int _mesa_sparc_needs_init = 1;
-#define INIT_MESA_SPARC do {               \
-   if (_mesa_sparc_needs_init) {           \
-      _glx_mesa_init_sparc_glapi_relocs(); \
-      _mesa_sparc_needs_init = 0;          \
-   }                                       \
-  } while(0)
-#else
-#define INIT_MESA_SPARC do { } while(0)
-#endif
-
 /*
 ** You can set this cell to 1 to force the gl drawing stuff to be
 ** one command per packet
@@ -670,7 +657,6 @@ __glXInitialize(Display * dpy)
    }
 #endif
 
-   INIT_MESA_SPARC;
    /* The one and only long long lock */
    __glXLock();
 
@@ -785,7 +771,6 @@ __glXSetupForCommand(Display * dpy)
 
       if (gc->currentDpy == dpy) {
          /* Use opcode from gc because its right */
-         INIT_MESA_SPARC;
          return gc->majorOpcode;
       }
       else {
@@ -979,74 +964,3 @@ __glXDumpDrawBuffer(__GLXcontext * ctx)
    }
 }
 #endif
-
-#ifdef  USE_SPARC_ASM
-/*
- * This is where our dispatch table's bounds are.
- * And the static mesa_init is taken directly from
- * Mesa's 'sparc.c' initializer.
- *
- * We need something like this here, because this version
- * of openGL/glx never initializes a Mesa context, and so
- * the address of the dispatch table pointer never gets stuffed
- * into the dispatch jump table otherwise.
- *
- * It matters only on SPARC, and only if you are using assembler
- * code instead of C-code indirect dispatch.
- *
- * -- FEM, 04.xii.03
- */
-extern unsigned int _mesa_sparc_glapi_begin;
-extern unsigned int _mesa_sparc_glapi_end;
-extern void __glapi_sparc_icache_flush(unsigned int *);
-
-static void
-_glx_mesa_init_sparc_glapi_relocs(void)
-{
-   unsigned int *insn_ptr, *end_ptr;
-   unsigned long disp_addr;
-
-   insn_ptr = &_mesa_sparc_glapi_begin;
-   end_ptr = &_mesa_sparc_glapi_end;
-   disp_addr = (unsigned long) &_glapi_Dispatch;
-
-   /*
-    * Verbatim from Mesa sparc.c.  It's needed because there doesn't
-    * seem to be a better way to do this:
-    *
-    * UNCONDITIONAL_JUMP ( (*_glapi_Dispatch) + entry_offset )
-    *
-    * This code is patching in the ADDRESS of the pointer to the
-    * dispatch table.  Hence, it must be called exactly once, because
-    * that address is not going to change.
-    *
-    * What it points to can change, but Mesa (and hence, we) assume
-    * that there is only one pointer.
-    *
-    */
-   while (insn_ptr < end_ptr) {
-#if ( defined(__sparc_v9__) && ( !defined(__linux__) || defined(__linux_64__) ) )
-/*
-	This code patches for 64-bit addresses.  This had better
-	not happen for Sparc/Linux, no matter what architecture we
-	are building for.  So, don't do this.
-
-        The 'defined(__linux_64__)' is used here as a placeholder for
-        when we do do 64-bit usermode on sparc linux.
-	*/
-      insn_ptr[0] |= (disp_addr >> (32 + 10));
-      insn_ptr[1] |= ((disp_addr & 0xffffffff) >> 10);
-      __glapi_sparc_icache_flush(&insn_ptr[0]);
-      insn_ptr[2] |= ((disp_addr >> 32) & ((1 << 10) - 1));
-      insn_ptr[3] |= (disp_addr & ((1 << 10) - 1));
-      __glapi_sparc_icache_flush(&insn_ptr[2]);
-      insn_ptr += 11;
-#else
-      insn_ptr[0] |= (disp_addr >> 10);
-      insn_ptr[1] |= (disp_addr & ((1 << 10) - 1));
-      __glapi_sparc_icache_flush(&insn_ptr[0]);
-      insn_ptr += 5;
-#endif
-   }
-}
-#endif /* sparc ASM in use */
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index f091d600c3..52f09a4b1b 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -180,9 +180,9 @@ get_result_vector(struct i915_fragment_program *p,
    switch (inst->DstReg.File) {
    case PROGRAM_OUTPUT:
       switch (inst->DstReg.Index) {
-      case FRAG_RESULT_COLR:
+      case FRAG_RESULT_COLOR:
          return UREG(REG_TYPE_OC, 0);
-      case FRAG_RESULT_DEPR:
+      case FRAG_RESULT_DEPTH:
          p->depth_written = 1;
          return UREG(REG_TYPE_OD, 0);
       default:
diff --git a/src/mesa/drivers/dri/i965/brw_clip_line.c b/src/mesa/drivers/dri/i965/brw_clip_line.c
index c45d48dff8..d830e49e50 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_line.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_line.c
@@ -181,34 +181,54 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
 	 brw_DP4(p, vec4(c->reg.dp1), deref_4f(vtx1, c->offset[VERT_RESULT_HPOS]), c->reg.plane_equation);
 	 is_negative = brw_IF(p, BRW_EXECUTE_1);
 	 {
-	    brw_ADD(p, c->reg.t, c->reg.dp1, negate(c->reg.dp0));
-	    brw_math_invert(p, c->reg.t, c->reg.t);
-	    brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp1);
-
-	    brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t1 );
-	    brw_MOV(p, c->reg.t1, c->reg.t);
-	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+             /*
+              * Both can be negative on GM965/G965 due to RHW workaround
+              * if so, this object should be rejected.
+              */
+             if (!BRW_IS_G4X(p->brw)) {
+                 brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_LE, c->reg.dp0, brw_imm_f(0.0));
+                 is_neg2 = brw_IF(p, BRW_EXECUTE_1);
+                 {
+                     brw_clip_kill_thread(c);
+                 }
+                 brw_ENDIF(p, is_neg2);
+             }
+
+             brw_ADD(p, c->reg.t, c->reg.dp1, negate(c->reg.dp0));
+             brw_math_invert(p, c->reg.t, c->reg.t);
+             brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp1);
+
+             brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t1 );
+             brw_MOV(p, c->reg.t1, c->reg.t);
+             brw_set_predicate_control(p, BRW_PREDICATE_NONE);
 	 } 
 	 is_negative = brw_ELSE(p, is_negative);
 	 {
-	    /* Coming back in.  We know that both cannot be negative
-	     * because the line would have been culled in that case.
-	     */
+             /* Coming back in.  We know that both cannot be negative
+              * because the line would have been culled in that case.
+              */
+
+             /* If both are positive, do nothing */
+             /* Only on GM965/G965 */
+             if (!BRW_IS_G4X(p->brw)) {
+                 brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.dp0, brw_imm_f(0.0));
+                 is_neg2 = brw_IF(p, BRW_EXECUTE_1);
+             }
 
-	    /* If both are positive, do nothing */
-             brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.dp0, brw_imm_f(0.0));
-             is_neg2 = brw_IF(p, BRW_EXECUTE_1);
              {
-		brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
-		brw_math_invert(p, c->reg.t, c->reg.t);
-		brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
-
-		brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
-		brw_MOV(p, c->reg.t0, c->reg.t);
-		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-	     }
-	     brw_ENDIF(p, is_neg2);
-	 }
+                 brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
+                 brw_math_invert(p, c->reg.t, c->reg.t);
+                 brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
+
+                 brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
+                 brw_MOV(p, c->reg.t0, c->reg.t);
+                 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+             }
+
+             if (!BRW_IS_G4X(p->brw)) {
+                 brw_ENDIF(p, is_neg2);
+             }
+         }
 	 brw_ENDIF(p, is_negative);	 
       }
       brw_ENDIF(p, plane_active);
diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c b/src/mesa/drivers/dri/i965/brw_clip_tri.c
index 1dbba37fe7..7fd37bd05f 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_tri.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c
@@ -455,6 +455,8 @@ static void brw_clip_test( struct brw_clip_compile *c )
     struct brw_indirect vt2 = brw_indirect(2, 0);
 
     struct brw_compile *p = &c->func;
+    struct brw_instruction *is_outside;
+    struct brw_reg tmp0 = c->reg.loopcount; /* handy temporary */
 
     brw_MOV(p, get_addr_reg(vt0), brw_address(c->reg.vertex[0]));
     brw_MOV(p, get_addr_reg(vt1), brw_address(c->reg.vertex[1]));
@@ -462,53 +464,87 @@ static void brw_clip_test( struct brw_clip_compile *c )
     brw_MOV(p, v0, deref_4f(vt0, c->offset[VERT_RESULT_HPOS]));
     brw_MOV(p, v1, deref_4f(vt1, c->offset[VERT_RESULT_HPOS]));
     brw_MOV(p, v2, deref_4f(vt2, c->offset[VERT_RESULT_HPOS]));
+    brw_AND(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(~0x3f));
 
     /* test nearz, xmin, ymin plane */
-    brw_CMP(p, t1, BRW_CONDITIONAL_LE, negate(v0), get_element(v0, 3)); 
+    /* clip.xyz < -clip.w */
+    brw_CMP(p, t1, BRW_CONDITIONAL_L, v0, negate(get_element(v0, 3))); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t2, BRW_CONDITIONAL_LE, negate(v1), get_element(v1, 3)); 
+    brw_CMP(p, t2, BRW_CONDITIONAL_L, v1, negate(get_element(v1, 3))); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t3, BRW_CONDITIONAL_LE, negate(v2), get_element(v2, 3)); 
+    brw_CMP(p, t3, BRW_CONDITIONAL_L, v2, negate(get_element(v2, 3))); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    /* All vertices are outside of a plane, rejected */
+    brw_AND(p, t, t1, t2);
+    brw_AND(p, t, t, t3);
+    brw_OR(p, tmp0, get_element(t, 0), get_element(t, 1));
+    brw_OR(p, tmp0, tmp0, get_element(t, 2));
+    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+    brw_AND(p, brw_null_reg(), tmp0, brw_imm_ud(0x1));
+    is_outside = brw_IF(p, BRW_EXECUTE_1);
+    {
+        brw_clip_kill_thread(c);
+    }
+    brw_ENDIF(p, is_outside);
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    /* some vertices are inside a plane, some are outside,need to clip */
     brw_XOR(p, t, t1, t2);
     brw_XOR(p, t1, t2, t3);
     brw_OR(p, t, t, t1);
-
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 0), brw_imm_ud(0));
+    brw_AND(p, t, t, brw_imm_ud(0x1));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 0), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<5)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 1), brw_imm_ud(0));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 1), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<3)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 2), brw_imm_ud(0));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 2), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<1)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
 
     /* test farz, xmax, ymax plane */
-    brw_CMP(p, t1, BRW_CONDITIONAL_L, v0, get_element(v0, 3)); 
+    /* clip.xyz > clip.w */
+    brw_CMP(p, t1, BRW_CONDITIONAL_G, v0, get_element(v0, 3)); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t2, BRW_CONDITIONAL_L, v1, get_element(v1, 3)); 
+    brw_CMP(p, t2, BRW_CONDITIONAL_G, v1, get_element(v1, 3)); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, t3, BRW_CONDITIONAL_L, v2, get_element(v2, 3)); 
+    brw_CMP(p, t3, BRW_CONDITIONAL_G, v2, get_element(v2, 3)); 
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
 
+    /* All vertices are outside of a plane, rejected */
+    brw_AND(p, t, t1, t2);
+    brw_AND(p, t, t, t3);
+    brw_OR(p, tmp0, get_element(t, 0), get_element(t, 1));
+    brw_OR(p, tmp0, tmp0, get_element(t, 2));
+    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+    brw_AND(p, brw_null_reg(), tmp0, brw_imm_ud(0x1));
+    is_outside = brw_IF(p, BRW_EXECUTE_1);
+    {
+        brw_clip_kill_thread(c);
+    }
+    brw_ENDIF(p, is_outside);
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    /* some vertices are inside a plane, some are outside,need to clip */
     brw_XOR(p, t, t1, t2);
     brw_XOR(p, t1, t2, t3);
     brw_OR(p, t, t, t1);
-
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 0), brw_imm_ud(0));
+    brw_AND(p, t, t, brw_imm_ud(0x1));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 0), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<4)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 1), brw_imm_ud(0));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 1), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<2)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
-	    get_element(t, 2), brw_imm_ud(0));
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ,
+            get_element(t, 2), brw_imm_ud(0));
     brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<0)));
     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
 
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 9106b748ad..a43fc40ac9 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -418,8 +418,8 @@ struct brw_context
       struct brw_tracked_state **atoms;
       GLuint nr_atoms;
 
-      GLuint nr_draw_regions;
-      struct intel_region *draw_regions[MAX_DRAW_BUFFERS];
+      GLuint nr_color_regions;
+      struct intel_region *color_regions[MAX_DRAW_BUFFERS];
       struct intel_region *depth_region;
 
       /**
@@ -670,7 +670,9 @@ void brwInitFragProgFuncs( struct dd_function_table *functions );
  */
 void brw_upload_urb_fence(struct brw_context *brw);
 
-void brw_upload_constant_buffer_state(struct brw_context *brw);
+/* brw_curbe.c
+ */
+void brw_upload_cs_urb_state(struct brw_context *brw);
 
 
 /*======================================================================
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index 4eaaa5f871..4ec0cfb816 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -138,24 +138,24 @@ const struct brw_tracked_state brw_curbe_offsets = {
  * fixed-function hardware in a double-buffering scheme to avoid a
  * pipeline stall each time the contents of the curbe is changed.
  */
-void brw_upload_constant_buffer_state(struct brw_context *brw)
+void brw_upload_cs_urb_state(struct brw_context *brw)
 {
-   struct brw_constant_buffer_state cbs; 
-   memset(&cbs, 0, sizeof(cbs));
+   struct brw_cs_urb_state cs_urb;
+   memset(&cs_urb, 0, sizeof(cs_urb));
 
    /* It appears that this is the state packet for the CS unit, ie. the
     * urb entries detailed here are housed in the CS range from the
     * URB_FENCE command.
     */
-   cbs.header.opcode = CMD_CONST_BUFFER_STATE;
-   cbs.header.length = sizeof(cbs)/4 - 2;
+   cs_urb.header.opcode = CMD_CS_URB_STATE;
+   cs_urb.header.length = sizeof(cs_urb)/4 - 2;
 
    /* BRW_NEW_URB_FENCE */
-   cbs.bits0.nr_urb_entries = brw->urb.nr_cs_entries;
-   cbs.bits0.urb_entry_size = brw->urb.csize - 1;
+   cs_urb.bits0.nr_urb_entries = brw->urb.nr_cs_entries;
+   cs_urb.bits0.urb_entry_size = brw->urb.csize - 1;
 
    assert(brw->urb.nr_cs_entries);
-   BRW_CACHED_BATCH_STRUCT(brw, &cbs);
+   BRW_CACHED_BATCH_STRUCT(brw, &cs_urb);
 }
 
 static GLfloat fixed_plane[6][4] = {
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 39c32255f8..590b064c7e 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -734,7 +734,7 @@
 
 
 #define CMD_URB_FENCE                 0x6000
-#define CMD_CONST_BUFFER_STATE        0x6001
+#define CMD_CS_URB_STATE              0x6001
 #define CMD_CONST_BUFFER              0x6002
 
 #define CMD_STATE_BASE_ADDRESS        0x6101
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 3ca953da37..7ab1ece263 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -194,6 +194,9 @@ static GLboolean check_fallbacks( struct brw_context *brw,
    GLcontext *ctx = &brw->intel.ctx;
    GLuint i;
 
+   /* If we don't require strict OpenGL conformance, never 
+    * use fallbacks.
+    */
    if (!brw->intel.strict_conformance)
       return GL_FALSE;
 
@@ -230,13 +233,31 @@ static GLboolean check_fallbacks( struct brw_context *brw,
       }
    }
 
-
    if (ctx->Point.SmoothFlag) {
       for (i = 0; i < nr_prims; i++)
 	 if (prim[i].mode == GL_POINTS) 
 	    return GL_TRUE;
    }
+
+   /* BRW hardware doesn't handle GL_CLAMP texturing correctly;
+    * brw_wm_sampler_state:translate_wrap_mode() treats GL_CLAMP
+    * as GL_CLAMP_TO_EDGE instead.  If we're using GL_CLAMP, and
+    * we want strict conformance, force the fallback.
+    * Right now, we only do this for 2D textures.
+    */
+   {
+      int u;
+      for (u = 0; u < ctx->Const.MaxTextureCoordUnits; u++) {
+         if (ctx->Texture.Unit[u].Enabled) {
+            if (ctx->Texture.Unit[u].CurrentTex[TEXTURE_2D_INDEX]->WrapS == GL_CLAMP ||
+                ctx->Texture.Unit[u].CurrentTex[TEXTURE_2D_INDEX]->WrapT == GL_CLAMP) {
+                   return GL_TRUE;
+            }
+         }
+      }
+   }
       
+   /* Nothing stopping us from the fast path now */
    return GL_FALSE;
 }
 
@@ -261,11 +282,18 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    if (ctx->NewState)
       _mesa_update_state( ctx );
 
+   /* We have to validate the textures *before* checking for fallbacks;
+    * otherwise, the software fallback won't be able to rely on the
+    * texture state, the firstLevel and lastLevel fields won't be
+    * set in the intel texture object (they'll both be 0), and the 
+    * software fallback will segfault if it attempts to access any
+    * texture level other than level 0.
+    */
+   brw_validate_textures( brw );
+
    if (check_fallbacks(brw, prim, nr_prims))
       return GL_FALSE;
 
-   brw_validate_textures( brw );
-
    /* Bind all inputs, derive varying and size information:
     */
    brw_merge_inputs( brw, arrays );
diff --git a/src/mesa/drivers/dri/i965/brw_misc_state.c b/src/mesa/drivers/dri/i965/brw_misc_state.c
index 9dcdad7b4e..40cb7a341d 100644
--- a/src/mesa/drivers/dri/i965/brw_misc_state.c
+++ b/src/mesa/drivers/dri/i965/brw_misc_state.c
@@ -178,7 +178,7 @@ static void upload_psp_urb_cbs(struct brw_context *brw )
 {
    upload_pipelined_state_pointers(brw);
    brw_upload_urb_fence(brw);
-   brw_upload_constant_buffer_state(brw);
+   brw_upload_cs_urb_state(brw);
 }
 
 const struct brw_tracked_state brw_psp_urb_cbs = {
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index df839c5b30..81b0a45998 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -52,7 +52,6 @@ const struct brw_tracked_state brw_cc_vp;
 const struct brw_tracked_state brw_check_fallback;
 const struct brw_tracked_state brw_clip_prog;
 const struct brw_tracked_state brw_clip_unit;
-const struct brw_tracked_state brw_constant_buffer_state;
 const struct brw_tracked_state brw_constant_buffer;
 const struct brw_tracked_state brw_curbe_offsets;
 const struct brw_tracked_state brw_invarient_state;
diff --git a/src/mesa/drivers/dri/i965/brw_structs.h b/src/mesa/drivers/dri/i965/brw_structs.h
index d97ff27f0a..89e2981203 100644
--- a/src/mesa/drivers/dri/i965/brw_structs.h
+++ b/src/mesa/drivers/dri/i965/brw_structs.h
@@ -439,7 +439,7 @@ struct brw_urb_fence
    } bits1;
 };
 
-struct brw_constant_buffer_state /* previously brw_command_streamer */
+struct brw_cs_urb_state
 {
    struct header header;
 
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 15ceac6ba1..fc9f6cc80c 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -71,9 +71,9 @@ static void brw_destroy_context( struct intel_context *intel )
 
    brw_FrameBufferTexDestroy( brw );
 
-   for (i = 0; i < brw->state.nr_draw_regions; i++)
-      intel_region_release(&brw->state.draw_regions[i]);
-   brw->state.nr_draw_regions = 0;
+   for (i = 0; i < brw->state.nr_color_regions; i++)
+      intel_region_release(&brw->state.color_regions[i]);
+   brw->state.nr_color_regions = 0;
    intel_region_release(&brw->state.depth_region);
 
    dri_bo_release(&brw->curbe.curbe_bo);
@@ -105,25 +105,25 @@ static void brw_destroy_context( struct intel_context *intel )
  * called from intelDrawBuffer()
  */
 static void brw_set_draw_region( struct intel_context *intel, 
-                                 struct intel_region *draw_regions[],
+                                 struct intel_region *color_regions[],
                                  struct intel_region *depth_region,
-                                 GLuint num_regions)
+                                 GLuint num_color_regions)
 {
    struct brw_context *brw = brw_context(&intel->ctx);
-   int i;
+   GLuint i;
 
    /* release old color/depth regions */
    if (brw->state.depth_region != depth_region)
       brw->state.dirty.brw |= BRW_NEW_DEPTH_BUFFER;
-   for (i = 0; i < brw->state.nr_draw_regions; i++)
-       intel_region_release(&brw->state.draw_regions[i]);
+   for (i = 0; i < brw->state.nr_color_regions; i++)
+       intel_region_release(&brw->state.color_regions[i]);
    intel_region_release(&brw->state.depth_region);
 
    /* reference new color/depth regions */
-   for (i = 0; i < num_regions; i++)
-       intel_region_reference(&brw->state.draw_regions[i], draw_regions[i]);
+   for (i = 0; i < num_color_regions; i++)
+       intel_region_reference(&brw->state.color_regions[i], color_regions[i]);
    intel_region_reference(&brw->state.depth_region, depth_region);
-   brw->state.nr_draw_regions = num_regions;
+   brw->state.nr_color_regions = num_color_regions;
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 10f38e02a4..06a6f3f0f4 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -198,7 +198,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
        ctx->Color.AlphaEnabled)
       lookup |= IZ_PS_KILL_ALPHATEST_BIT;
 
-   if (fp->program.Base.OutputsWritten & (1<<FRAG_RESULT_DEPR))
+   if (fp->program.Base.OutputsWritten & (1<<FRAG_RESULT_DEPTH))
       lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
 
    /* _NEW_DEPTH */
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index 37766b7c3a..7e88ef0053 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -289,8 +289,7 @@ static struct prog_src_register get_pixel_w( struct brw_wm_compile *c )
       struct prog_dst_register pixel_w = get_temp(c);
       struct prog_src_register deltas = get_delta_xy(c);
       struct prog_src_register interp_wpos = src_reg(PROGRAM_PAYLOAD, FRAG_ATTRIB_WPOS);
-      
-      
+
       /* deltas.xyw = DELTAS2 deltas.xy, payload.interp_wpos.x
        */
       emit_op(c,
@@ -510,7 +509,6 @@ static void precalc_dst( struct brw_wm_compile *c,
 	      src_undef());
    }
 
-
    if (dst.WriteMask & WRITEMASK_XZ) {
       struct prog_instruction *swz;
       GLuint z = GET_SWZ(src0.Swizzle, Z);
@@ -563,7 +561,6 @@ static void precalc_lit( struct brw_wm_compile *c,
       swz->SrcReg[0].NegateBase = 0;
    }
 
-
    if (dst.WriteMask & WRITEMASK_YZ) {
       emit_op(c,
 	      OPCODE_LIT,
@@ -865,42 +862,41 @@ static void precalc_txp( struct brw_wm_compile *c,
 static void emit_fb_write( struct brw_wm_compile *c )
 {
    struct prog_src_register payload_r0_depth = src_reg(PROGRAM_PAYLOAD, PAYLOAD_DEPTH);
-   struct prog_src_register outdepth = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DEPR);
+   struct prog_src_register outdepth = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DEPTH);
    struct prog_src_register outcolor;
    GLuint i;
 
    struct prog_instruction *inst, *last_inst;
    struct brw_context *brw = c->func.brw;
 
-   /* inst->Sampler is not used by backend, 
-      use it for fb write target and eot */
-
-   if (brw->state.nr_draw_regions > 1) {
-       for (i = 0 ; i < brw->state.nr_draw_regions; i++) {
-	   outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0 + i);
-	   last_inst = inst = emit_op(c,
-		   WM_FB_WRITE, dst_mask(dst_undef(),0), 0,
-		   outcolor, payload_r0_depth, outdepth);
-	   inst->Sampler = (i<<1);
-	   if (c->fp_fragcolor_emitted) {
-	       outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
-	       last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
-		       0, outcolor, payload_r0_depth, outdepth);
-	       inst->Sampler = (i<<1);
-	   }
-       }
-       last_inst->Sampler |= 1; //eot
+   /* The inst->Aux field is used for FB write target and the EOT marker */
+
+   if (brw->state.nr_color_regions > 1) {
+      for (i = 0 ; i < brw->state.nr_color_regions; i++) {
+         outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0 + i);
+         last_inst = inst = emit_op(c,
+                                    WM_FB_WRITE, dst_mask(dst_undef(),0), 0,
+                                    outcolor, payload_r0_depth, outdepth);
+         inst->Aux = (i<<1);
+         if (c->fp_fragcolor_emitted) {
+            outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
+            last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
+                                       0, outcolor, payload_r0_depth, outdepth);
+            inst->Aux = (i<<1);
+         }
+      }
+      last_inst->Aux |= 1; //eot
    }
    else {
       /* if gl_FragData[0] is written, use it, else use gl_FragColor */
       if (c->fp->program.Base.OutputsWritten & (1 << FRAG_RESULT_DATA0))
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0);
       else 
-         outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
+         outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
 
-       inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
-	       0, outcolor, payload_r0_depth, outdepth);
-       inst->Sampler = 1|(0<<1);
+      inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
+                     0, outcolor, payload_r0_depth, outdepth);
+      inst->Aux = 1|(0<<1);
    }
 }
 
@@ -931,9 +927,9 @@ static void validate_dst_regs( struct brw_wm_compile *c,
 			       const struct prog_instruction *inst )
 {
    if (inst->DstReg.File == PROGRAM_OUTPUT) {
-       GLuint idx = inst->DstReg.Index;
-       if (idx == FRAG_RESULT_COLR)
-	   c->fp_fragcolor_emitted = 1;
+      GLuint idx = inst->DstReg.Index;
+      if (idx == FRAG_RESULT_COLOR)
+         c->fp_fragcolor_emitted = 1;
    }
 }
 
@@ -954,7 +950,6 @@ static void print_insns( const struct prog_instruction *insn,
       }
       else 
 	 _mesa_printf("UNKNOWN\n");
-	   
    }
 }
 
@@ -1079,9 +1074,9 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
    }
 
    if (INTEL_DEBUG & DEBUG_WM) {
-	   _mesa_printf("pass_fp:\n");
-	   print_insns( c->prog_instructions, c->nr_fp_insns );
-	   _mesa_printf("\n");
+      _mesa_printf("pass_fp:\n");
+      print_insns( c->prog_instructions, c->nr_fp_insns );
+      _mesa_printf("\n");
    }
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index 121cdc040e..8e404bcdc3 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -426,8 +426,8 @@ static void emit_fb_write(struct brw_wm_compile *c,
        nr += 2;
     }
 
-    target = inst->Sampler >> 1;
-    eot = inst->Sampler & 1;
+    target = inst->Aux >> 1;
+    eot = inst->Aux & 1;
     fire_fb_write(c, 0, nr, target, eot);
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
index 783545de56..2debd0678a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
@@ -350,8 +350,8 @@ translate_insn(struct brw_wm_compile *c,
    out->tex_unit = inst->TexSrcUnit;
    out->tex_idx = inst->TexSrcTarget;
    out->tex_shadow = inst->TexShadow;
-   out->eot = inst->Sampler & 1;
-   out->target = inst->Sampler>>1;
+   out->eot = inst->Aux & 1;
+   out->target = inst->Aux >> 1;
 
    /* Args:
     */
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 3c3b3473d6..f7ea756fb0 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -103,7 +103,7 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
 
    /* as far as we can tell */
    key->computes_depth =
-      (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) != 0;
+      (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) != 0;
 
    /* _NEW_COLOR */
    key->uses_kill = fp->UsesKill || ctx->Color.AlphaEnabled;
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 654adcd131..e402fdf0ee 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -454,13 +454,13 @@ static void prepare_wm_surfaces(struct brw_context *brw )
    GLuint i;
    int old_nr_surfaces;
 
-   if (brw->state.nr_draw_regions  > 1) {
-      for (i = 0; i < brw->state.nr_draw_regions; i++) {
-         brw_update_region_surface(brw, brw->state.draw_regions[i], i,
+   if (brw->state.nr_color_regions  > 1) {
+      for (i = 0; i < brw->state.nr_color_regions; i++) {
+         brw_update_region_surface(brw, brw->state.color_regions[i], i,
 				   GL_FALSE);
       }
-   }else {
-      brw_update_region_surface(brw, brw->state.draw_regions[0], 0, GL_TRUE);
+   } else {
+      brw_update_region_surface(brw, brw->state.color_regions[0], 0, GL_TRUE);
    }
 
    old_nr_surfaces = brw->wm.nr_surfaces;
diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
index 60d7bb3770..b7c7eeb368 100644
--- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
@@ -35,9 +35,6 @@
 #include "intel_batchbuffer.h"
 #include "intel_regions.h"
 
-static GLboolean intel_bufferobj_unmap(GLcontext * ctx,
-				       GLenum target,
-				       struct gl_buffer_object *obj);
 
 /** Allocates a new dri_bo to store the data for the buffer object. */
 static void
@@ -103,12 +100,7 @@ intel_bufferobj_free(GLcontext * ctx, struct gl_buffer_object *obj)
    struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
 
    assert(intel_obj);
-
-   /* Buffer objects are automatically unmapped when deleting according
-    * to the spec.
-    */
-   if (obj->Pointer)
-      intel_bufferobj_unmap(ctx, 0, obj);
+   assert(!obj->Pointer); /* Mesa should have unmapped it */
 
    if (intel_obj->region) {
       intel_bufferobj_release_region(intel, intel_obj);
@@ -141,11 +133,7 @@ intel_bufferobj_data(GLcontext * ctx,
    intel_obj->Base.Size = size;
    intel_obj->Base.Usage = usage;
 
-   /* Buffer objects are automatically unmapped when creating new data buffers
-    * according to the spec.
-    */
-   if (obj->Pointer)
-      intel_bufferobj_unmap(ctx, 0, obj);
+   assert(!obj->Pointer); /* Mesa should have unmapped it */
 
    if (intel_obj->region)
       intel_bufferobj_release_region(intel, intel_obj);
diff --git a/src/mesa/drivers/dri/intel/intel_chipset.h b/src/mesa/drivers/dri/intel/intel_chipset.h
index d1b4941601..4593d90df3 100644
--- a/src/mesa/drivers/dri/intel/intel_chipset.h
+++ b/src/mesa/drivers/dri/intel/intel_chipset.h
@@ -46,6 +46,13 @@
 #define PCI_CHIP_G33_G			0x29C2
 #define PCI_CHIP_Q33_G			0x29D2
 
+#define PCI_CHIP_IGD_GM			0xA011
+#define PCI_CHIP_IGD_G			0xA001
+
+#define IS_IGDGM(devid)	(devid == PCI_CHIP_IGD_GM)
+#define IS_IGDG(devid)	(devid == PCI_CHIP_IGD_G)
+#define IS_IGD(devid) (IS_IGDG(devid) || IS_IGDGM(devid))
+
 #define PCI_CHIP_I965_G			0x29A2
 #define PCI_CHIP_I965_Q			0x2992
 #define PCI_CHIP_I965_G_1		0x2982
@@ -66,7 +73,7 @@
 				 devid == PCI_CHIP_I945_GME || \
 				 devid == PCI_CHIP_I965_GM || \
 				 devid == PCI_CHIP_I965_GME || \
-				 devid == PCI_CHIP_GM45_GM)
+				 devid == PCI_CHIP_GM45_GM || IS_IGD(devid))
 
 #define IS_G45(devid)           (devid == PCI_CHIP_IGD_E_G || \
                                  devid == PCI_CHIP_Q45_G || \
@@ -84,7 +91,7 @@
 				 devid == PCI_CHIP_I945_GME || \
 				 devid == PCI_CHIP_G33_G || \
 				 devid == PCI_CHIP_Q33_G || \
-				 devid == PCI_CHIP_Q35_G)
+				 devid == PCI_CHIP_Q35_G || IS_IGD(devid))
 
 #define IS_965(devid)		(devid == PCI_CHIP_I965_G || \
 				 devid == PCI_CHIP_I965_Q || \
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 1aa173db18..b3f6fc02df 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -123,6 +123,10 @@ intelGetString(GLcontext * ctx, GLenum name)
       case PCI_CHIP_Q33_G:
 	 chipset = "Intel(R) Q33";
 	 break;
+      case PCI_CHIP_IGD_GM:
+      case PCI_CHIP_IGD_G:
+	 chipset = "Intel(R) IGD";
+	 break;
       case PCI_CHIP_I965_Q:
 	 chipset = "Intel(R) 965Q";
 	 break;
diff --git a/src/mesa/drivers/dri/intel/intel_depthtmp.h b/src/mesa/drivers/dri/intel/intel_depthtmp.h
new file mode 100644
index 0000000000..16d7708453
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_depthtmp.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright © 2009 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+/**
+ * Wrapper around the depthtmp.h macrofest to generate spans code for
+ * all the tiling styles.
+ */
+
+#define VALUE_TYPE INTEL_VALUE_TYPE
+#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(NO_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(NO_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel##x)
+#include "depthtmp.h"
+
+#define VALUE_TYPE INTEL_VALUE_TYPE
+#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(X_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(X_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel_XTile_##x)
+#include "depthtmp.h"
+
+#define VALUE_TYPE INTEL_VALUE_TYPE
+#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(Y_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(Y_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel_YTile_##x)
+#include "depthtmp.h"
+
+#undef INTEL_VALUE_TYPE
+#undef INTEL_WRITE_DEPTH
+#undef INTEL_READ_DEPTH
+#undef INTEL_TAG
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c
index c9fb90f3b2..787c290022 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.c
+++ b/src/mesa/drivers/dri/intel/intel_fbo.c
@@ -323,6 +323,15 @@ intel_create_renderbuffer(GLenum intFormat)
       irb->Base.BlueBits = 5;
       irb->Base.DataType = GL_UNSIGNED_BYTE;
       break;
+   case GL_RGB8:
+      irb->Base._ActualFormat = GL_RGB8;
+      irb->Base._BaseFormat = GL_RGB;
+      irb->Base.RedBits = 8;
+      irb->Base.GreenBits = 8;
+      irb->Base.BlueBits = 8;
+      irb->Base.AlphaBits = 0;
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
+      break;
    case GL_RGBA8:
       irb->Base._ActualFormat = GL_RGBA8;
       irb->Base._BaseFormat = GL_RGBA;
@@ -603,19 +612,16 @@ static void
 intel_finish_render_texture(GLcontext * ctx,
                             struct gl_renderbuffer_attachment *att)
 {
-   struct intel_renderbuffer *irb = intel_renderbuffer(att->Renderbuffer);
-
-   DBG("End render texture (tid %x) tex %u\n", _glthread_GetID(), att->Texture->Name);
-
-   if (irb) {
-      /* just release the region */
-      intel_region_release(&irb->region);
-   }
-   else if (att->Renderbuffer) {
-      /* software fallback */
-      _mesa_finish_render_texture(ctx, att);
-      /* XXX FBO: Need to unmap the buffer (or in intelSpanRenderStart???) */
-   }
+   /* no-op
+    * Previously we released the renderbuffer's intel_region but
+    * that's not necessary and actually caused problems when trying
+    * to do a glRead/CopyPixels from the renderbuffer later.
+    * The region will be released later if the texture is replaced
+    * or the renderbuffer deleted.
+    *
+    * The intention of this driver hook is more of a "done rendering
+    * to texture, please re-twiddle/etc if necessary".
+    */
 }
 
 
@@ -625,6 +631,7 @@ intel_finish_render_texture(GLcontext * ctx,
 static void
 intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
 {
+   struct intel_context *intel = intel_context(ctx);
    const struct intel_renderbuffer *depthRb =
       intel_get_renderbuffer(fb, BUFFER_DEPTH);
    const struct intel_renderbuffer *stencilRb =
@@ -636,6 +643,34 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
        */
       fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
    }
+
+   /* check that texture color buffers are a format we can render into */
+   {
+      const struct gl_texture_format *supportedFormat;
+      GLuint i;
+
+      /* The texture format we can render into seems to depend on the
+       * screen depth.  There currently seems to be a problem when
+       * rendering into a rgb565 texture when the screen is abgr8888.
+       */
+      if (intel->front_region->cpp == 4)
+         supportedFormat = &_mesa_texformat_argb8888;
+      else 
+         supportedFormat = &_mesa_texformat_rgb565;
+
+      for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
+         const struct gl_texture_object *texObj =
+            fb->Attachment[BUFFER_COLOR0 + i].Texture;
+         if (texObj) {
+            const struct gl_texture_image *texImg =
+               texObj->Image[0][texObj->BaseLevel];
+            if (texImg && texImg->TexFormat != supportedFormat) {
+               fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+               break;
+            }
+         }
+      }
+   }
 }
 
 
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index 749920502a..0aa5b8c02c 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -466,7 +466,8 @@ intel_recreate_static(struct intel_context *intel,
    else
       region->cpp = intel->ctx.Visual.rgbBits / 8;
    region->pitch = intelScreen->pitch;
-   region->height = intelScreen->height;     /* needed? */
+   region->width = intelScreen->width;
+   region->height = intelScreen->height;
 
    if (region->buffer != NULL) {
       dri_bo_unreference(region->buffer);
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index d6110b8163..09eba13aab 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -323,7 +323,7 @@ intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
    else {
       GLboolean swStencil = (mesaVis->stencilBits > 0 &&
                              mesaVis->depthBits != 24);
-      GLenum rgbFormat = (mesaVis->redBits == 5 ? GL_RGB5 : GL_RGBA8);
+      GLenum rgbFormat;
 
       struct intel_framebuffer *intel_fb = CALLOC_STRUCT(intel_framebuffer);
 
@@ -332,6 +332,13 @@ intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
 
       _mesa_initialize_framebuffer(&intel_fb->Base, mesaVis);
 
+      if (mesaVis->redBits == 5)
+	 rgbFormat = GL_RGB5;
+      else if (mesaVis->alphaBits == 0)
+	 rgbFormat = GL_RGB8;
+      else
+	 rgbFormat = GL_RGBA8;
+
       /* setup the hardware-based renderbuffers */
       intel_fb->color_rb[0] = intel_create_renderbuffer(rgbFormat);
       _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT,
@@ -758,13 +765,34 @@ __DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
    fb_format[2] = GL_BGRA;
    fb_type[2] = GL_UNSIGNED_INT_8_8_8_8_REV;
 
+   depth_bits[0] = 0;
+   stencil_bits[0] = 0;
+
    for (color = 0; color < ARRAY_SIZE(fb_format); color++) {
       __DRIconfig **new_configs;
+      int depth_factor;
 
+      /* With DRI2 right now, GetBuffers always returns a depth/stencil buffer
+       * with the same cpp as the drawable.  So we can't support depth cpp !=
+       * color cpp currently.
+       */
+      if (fb_type[color] == GL_UNSIGNED_SHORT_5_6_5) {
+	 depth_bits[1] = 16;
+	 stencil_bits[1] = 0;
+
+	 depth_factor = 2;
+      } else {
+	 depth_bits[1] = 24;
+	 stencil_bits[1] = 0;
+	 depth_bits[2] = 24;
+	 stencil_bits[2] = 8;
+
+	 depth_factor = 3;
+      }
       new_configs = driCreateConfigs(fb_format[color], fb_type[color],
 				     depth_bits,
 				     stencil_bits,
-				     ARRAY_SIZE(depth_bits),
+				     depth_factor,
 				     back_buffer_modes,
 				     ARRAY_SIZE(back_buffer_modes),
 				     msaa_samples_array,
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
index bdd2fd9e85..c3a873f1ab 100644
--- a/src/mesa/drivers/dri/intel/intel_span.c
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -131,6 +131,18 @@ pwrite_8(struct intel_renderbuffer *irb, uint32_t offset, uint8_t val)
    dri_bo_subdata(irb->region->buffer, offset, 1, &val);
 }
 
+static uint32_t
+z24s8_to_s8z24(uint32_t val)
+{
+   return (val << 24) | (val >> 8);
+}
+
+static uint32_t
+s8z24_to_z24s8(uint32_t val)
+{
+   return (val >> 24) | (val << 8);
+}
+
 static uint32_t no_tile_swizzle(struct intel_renderbuffer *irb,
 				int x, int y)
 {
@@ -293,101 +305,29 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 #define X_TILE(_X, _Y) x_tile_swizzle(irb, (_X) + x_off, (_Y) + y_off)
 #define Y_TILE(_X, _Y) y_tile_swizzle(irb, (_X) + x_off, (_Y) + y_off)
 
-/* 16 bit, RGB565 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x)    intel##x##_RGB565
-#define TAG2(x,y) intel##x##_RGB565##y
-#define GET_VALUE(X, Y) pread_16(irb, NO_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_16(irb, NO_TILE(X, Y), V)
-#include "spantmp2.h"
-
-/* 32 bit, ARGB8888 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel##x##_ARGB8888
-#define TAG2(x,y) intel##x##_ARGB8888##y
-#define GET_VALUE(X, Y) pread_32(irb, NO_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_32(irb, NO_TILE(X, Y), V)
-#include "spantmp2.h"
-
-/* 32 bit, xRGB8888 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel##x##_xRGB8888
-#define TAG2(x,y) intel##x##_xRGB8888##y
-#define GET_VALUE(X, Y) pread_xrgb8888(irb, NO_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, NO_TILE(X, Y), V)
-#include "spantmp2.h"
-
-/* 16 bit RGB565 color tile spanline and pixel functions
- */
-
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x)    intel_XTile_##x##_RGB565
-#define TAG2(x,y) intel_XTile_##x##_RGB565##y
-#define GET_VALUE(X, Y) pread_16(irb, X_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_16(irb, X_TILE(X, Y), V)
-#include "spantmp2.h"
-
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x)    intel_YTile_##x##_RGB565
-#define TAG2(x,y) intel_YTile_##x##_RGB565##y
-#define GET_VALUE(X, Y) pread_16(irb, Y_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_16(irb, Y_TILE(X, Y), V)
-#include "spantmp2.h"
-
-/* 32 bit ARGB888 color tile spanline and pixel functions
- */
-
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel_XTile_##x##_ARGB8888
-#define TAG2(x,y) intel_XTile_##x##_ARGB8888##y
-#define GET_VALUE(X, Y) pread_32(irb, X_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_32(irb, X_TILE(X, Y), V)
-#include "spantmp2.h"
-
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel_YTile_##x##_ARGB8888
-#define TAG2(x,y) intel_YTile_##x##_ARGB8888##y
-#define GET_VALUE(X, Y) pread_32(irb, Y_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_32(irb, Y_TILE(X, Y), V)
-#include "spantmp2.h"
-
-/* 32 bit xRGB888 color tile spanline and pixel functions
- */
-
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel_XTile_##x##_xRGB8888
-#define TAG2(x,y) intel_XTile_##x##_xRGB8888##y
-#define GET_VALUE(X, Y) pread_xrgb8888(irb, X_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, X_TILE(X, Y), V)
-#include "spantmp2.h"
-
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel_YTile_##x##_xRGB8888
-#define TAG2(x,y) intel_YTile_##x##_xRGB8888##y
-#define GET_VALUE(X, Y) pread_xrgb8888(irb, Y_TILE(X, Y))
-#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, Y_TILE(X, Y), V)
-#include "spantmp2.h"
+/* r5g6b5 color span and pixel functions */
+#define INTEL_PIXEL_FMT GL_RGB
+#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
+#define INTEL_READ_VALUE(offset) pread_16(irb, offset)
+#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v)
+#define INTEL_TAG(x) x##_RGB565
+#include "intel_spantmp.h"
+
+/* a8r8g8b8 color span and pixel functions */
+#define INTEL_PIXEL_FMT GL_BGRA
+#define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+#define INTEL_READ_VALUE(offset) pread_32(irb, offset)
+#define INTEL_WRITE_VALUE(offset, v) pwrite_32(irb, offset, v)
+#define INTEL_TAG(x) x##_ARGB8888
+#include "intel_spantmp.h"
+
+/* x8r8g8b8 color span and pixel functions */
+#define INTEL_PIXEL_FMT GL_BGRA
+#define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+#define INTEL_READ_VALUE(offset) pread_xrgb8888(irb, offset)
+#define INTEL_WRITE_VALUE(offset, v) pwrite_xrgb8888(irb, offset, v)
+#define INTEL_TAG(x) x##_xRGB8888
+#include "intel_spantmp.h"
 
 #define LOCAL_DEPTH_VARS						\
    struct intel_context *intel = intel_context(ctx);			\
@@ -402,98 +342,26 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 
 #define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS
 
-/**
- ** 16-bit depthbuffer functions.
- **/
-#define VALUE_TYPE GLushort
-#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, NO_TILE(_x, _y), d)
-#define READ_DEPTH(d, _x, _y) d = pread_16(irb, NO_TILE(_x, _y))
-#define TAG(x) intel##x##_z16
-#include "depthtmp.h"
+/* z16 depthbuffer functions. */
+#define INTEL_VALUE_TYPE GLushort
+#define INTEL_WRITE_DEPTH(offset, d) pwrite_16(irb, offset, d)
+#define INTEL_READ_DEPTH(offset) pread_16(irb, offset)
+#define INTEL_TAG(name) name##_z16
+#include "intel_depthtmp.h"
 
+/* z24 depthbuffer functions. */
+#define INTEL_VALUE_TYPE GLuint
+#define INTEL_WRITE_DEPTH(offset, d) pwrite_32(irb, offset, d)
+#define INTEL_READ_DEPTH(offset) pread_32(irb, offset)
+#define INTEL_TAG(name) name##_z24
+#include "intel_depthtmp.h"
 
-/**
- ** 16-bit x tile depthbuffer functions.
- **/
-#define VALUE_TYPE GLushort
-#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, X_TILE(_x, _y), d)
-#define READ_DEPTH(d, _x, _y) d = pread_16(irb, X_TILE(_x, _y))
-#define TAG(x) intel_XTile_##x##_z16
-#include "depthtmp.h"
-
-/**
- ** 16-bit y tile depthbuffer functions.
- **/
-#define VALUE_TYPE GLushort
-#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, Y_TILE(_x, _y), d)
-#define READ_DEPTH(d, _x, _y) d = pread_16(irb, Y_TILE(_x, _y))
-#define TAG(x) intel_YTile_##x##_z16
-#include "depthtmp.h"
-
-
-/**
- ** 24/8-bit interleaved depth/stencil functions
- ** Note: we're actually reading back combined depth+stencil values.
- ** The wrappers in main/depthstencil.c are used to extract the depth
- ** and stencil values.
- **/
-#define VALUE_TYPE GLuint
-
-/* Change ZZZS -> SZZZ */
-#define WRITE_DEPTH(_x, _y, d)					\
-   pwrite_32(irb, NO_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
-
-/* Change SZZZ -> ZZZS */
-#define READ_DEPTH( d, _x, _y ) {				\
-   GLuint tmp = pread_32(irb, NO_TILE(_x, _y));			\
-   d = (tmp << 8) | (tmp >> 24);				\
-}
-
-#define TAG(x) intel##x##_z24_s8
-#include "depthtmp.h"
-
-
-/**
- ** 24/8-bit x-tile interleaved depth/stencil functions
- ** Note: we're actually reading back combined depth+stencil values.
- ** The wrappers in main/depthstencil.c are used to extract the depth
- ** and stencil values.
- **/
-#define VALUE_TYPE GLuint
-
-/* Change ZZZS -> SZZZ */
-#define WRITE_DEPTH(_x, _y, d)					\
-   pwrite_32(irb, X_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
-
-/* Change SZZZ -> ZZZS */
-#define READ_DEPTH( d, _x, _y ) {				\
-   GLuint tmp = pread_32(irb, X_TILE(_x, _y));		\
-   d = (tmp << 8) | (tmp >> 24);				\
-}
-
-#define TAG(x) intel_XTile_##x##_z24_s8
-#include "depthtmp.h"
-
-/**
- ** 24/8-bit y-tile interleaved depth/stencil functions
- ** Note: we're actually reading back combined depth+stencil values.
- ** The wrappers in main/depthstencil.c are used to extract the depth
- ** and stencil values.
- **/
-#define VALUE_TYPE GLuint
-
-/* Change ZZZS -> SZZZ */
-#define WRITE_DEPTH(_x, _y, d)					\
-   pwrite_32(irb, Y_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
-
-/* Change SZZZ -> ZZZS */
-#define READ_DEPTH( d, _x, _y ) {				\
-   GLuint tmp = pread_32(irb, Y_TILE(_x, _y));			\
-   d = (tmp << 8) | (tmp >> 24);				\
-}
-
-#define TAG(x) intel_YTile_##x##_z24_s8
-#include "depthtmp.h"
+/* z24s8 depthbuffer functions. */
+#define INTEL_VALUE_TYPE GLuint
+#define INTEL_WRITE_DEPTH(offset, d) pwrite_32(irb, offset, z24s8_to_s8z24(d))
+#define INTEL_READ_DEPTH(offset) s8z24_to_z24s8(pread_32(irb, offset))
+#define INTEL_TAG(name) name##_z24_s8
+#include "intel_depthtmp.h"
 
 
 /**
@@ -752,8 +620,21 @@ intel_set_span_functions(struct intel_context *intel,
 	 break;
       }
    }
-   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24 ||        /* XXX FBO remove */
-            rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
+   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24) {
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitDepthPointers_z24(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitDepthPointers_z24(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitDepthPointers_z24(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
       switch (tiling) {
       case I915_TILING_NONE:
       default:
diff --git a/src/mesa/drivers/dri/intel/intel_spantmp.h b/src/mesa/drivers/dri/intel/intel_spantmp.h
new file mode 100644
index 0000000000..ead0b1c168
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_spantmp.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright © 2009 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+/**
+ * Wrapper around the spantmp.h macrofest to generate spans code for
+ * all the tiling styles.
+ */
+
+#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
+#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
+#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(NO_TILE(_x, _y), v)
+#define GET_VALUE(_x, _y) INTEL_READ_VALUE(NO_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel##x)
+#define TAG2(x, y) INTEL_TAG(intel##x)##y
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
+#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
+#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(X_TILE(_x, _y), v)
+#define GET_VALUE(_x, _y) INTEL_READ_VALUE(X_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel_XTile_##x)
+#define TAG2(x, y) INTEL_TAG(intel_XTile_##x)##y
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
+#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
+#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(X_TILE(_x, _y), v)
+#define GET_VALUE(_x, _y) INTEL_READ_VALUE(X_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel_YTile_##x)
+#define TAG2(x, y) INTEL_TAG(intel_YTile_##x)##y
+#include "spantmp2.h"
+
+#undef INTEL_PIXEL_FMT
+#undef INTEL_PIXEL_TYPE
+#undef INTEL_WRITE_VALUE
+#undef INTEL_READ_VALUE
+#undef INTEL_TAG
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
index 2715a540d0..2d57c67848 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_format.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -3,11 +3,15 @@
 #include "main/texformat.h"
 #include "main/enums.h"
 
-/* It works out that this function is fine for all the supported
+
+/**
+ * Choose hardware texture format given the user's glTexImage parameters.
+ *
+ * It works out that this function is fine for all the supported
  * hardware.  However, there is still a need to map the formats onto
  * hardware descriptors.
- */
-/* Note that the i915 can actually support many more formats than
+ *
+ * Note that the i915 can actually support many more formats than
  * these if we take the step of simply swizzling the colors
  * immediately after sampling...
  */
@@ -18,6 +22,11 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    struct intel_context *intel = intel_context(ctx);
    const GLboolean do32bpt = (intel->ctx.Visual.rgbBits >= 24);
 
+#if 0
+   printf("%s intFmt=0x%x format=0x%x type=0x%x\n",
+          __FUNCTION__, internalFormat, format, type);
+#endif
+
    switch (internalFormat) {
    case 4:
    case GL_RGBA:
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.c b/src/mesa/drivers/dri/r300/r300_fragprog.c
index 8d030c63fb..32182bb667 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.c
@@ -356,8 +356,8 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler)
 
 static void nqssadce_init(struct nqssadce_state* s)
 {
-	s->Outputs[FRAG_RESULT_COLR].Sourced = WRITEMASK_XYZW;
-	s->Outputs[FRAG_RESULT_DEPR].Sourced = WRITEMASK_W;
+	s->Outputs[FRAG_RESULT_COLOR].Sourced = WRITEMASK_XYZW;
+	s->Outputs[FRAG_RESULT_DEPTH].Sourced = WRITEMASK_W;
 }
 
 
diff --git a/src/mesa/drivers/dri/r300/r500_fragprog.c b/src/mesa/drivers/dri/r300/r500_fragprog.c
index 926ddd5964..07a2a7b17c 100644
--- a/src/mesa/drivers/dri/r300/r500_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r500_fragprog.c
@@ -298,8 +298,8 @@ static void insert_WPOS_trailer(struct r500_fragment_program_compiler *compiler)
 
 static void nqssadce_init(struct nqssadce_state* s)
 {
-	s->Outputs[FRAG_RESULT_COLR].Sourced = WRITEMASK_XYZW;
-	s->Outputs[FRAG_RESULT_DEPR].Sourced = WRITEMASK_W;
+	s->Outputs[FRAG_RESULT_COLOR].Sourced = WRITEMASK_XYZW;
+	s->Outputs[FRAG_RESULT_DEPTH].Sourced = WRITEMASK_W;
 }
 
 static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
diff --git a/src/mesa/drivers/dri/r300/radeon_nqssadce.c b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
index 97ce016c99..a083c3d243 100644
--- a/src/mesa/drivers/dri/r300/radeon_nqssadce.c
+++ b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
@@ -191,7 +191,7 @@ static void process_instruction(struct nqssadce_state* s)
 
 	if (inst->Opcode != OPCODE_KIL) {
 		if (s->Descr->RewriteDepthOut) {
-			if (inst->DstReg.File == PROGRAM_OUTPUT && inst->DstReg.Index == FRAG_RESULT_DEPR)
+			if (inst->DstReg.File == PROGRAM_OUTPUT && inst->DstReg.Index == FRAG_RESULT_DEPTH)
 				rewrite_depth_out(inst);
 		}
 
diff --git a/src/mesa/drivers/dri/r300/radeon_program_pair.c b/src/mesa/drivers/dri/r300/radeon_program_pair.c
index 58bc0d5843..365e7c1722 100644
--- a/src/mesa/drivers/dri/r300/radeon_program_pair.c
+++ b/src/mesa/drivers/dri/r300/radeon_program_pair.c
@@ -778,10 +778,10 @@ static void fill_dest_into_pair(struct pair_state *s, struct radeon_pair_instruc
 	struct prog_instruction *inst = s->Program->Instructions + ip;
 
 	if (inst->DstReg.File == PROGRAM_OUTPUT) {
-		if (inst->DstReg.Index == FRAG_RESULT_COLR) {
+		if (inst->DstReg.Index == FRAG_RESULT_COLOR) {
 			pair->RGB.OutputWriteMask |= inst->DstReg.WriteMask & WRITEMASK_XYZ;
 			pair->Alpha.OutputWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
-		} else if (inst->DstReg.Index == FRAG_RESULT_DEPR) {
+		} else if (inst->DstReg.Index == FRAG_RESULT_DEPTH) {
 			pair->Alpha.DepthWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
 		}
 	} else {
diff --git a/src/mesa/drivers/dri/radeon/radeon_chipset.h b/src/mesa/drivers/dri/radeon/radeon_chipset.h
index 55a73eab20..f6bd1eb83f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_chipset.h
+++ b/src/mesa/drivers/dri/radeon/radeon_chipset.h
@@ -247,6 +247,9 @@
 #define PCI_CHIP_RS350_7835		0x7835
 #define PCI_CHIP_RS690_791E             0x791E
 #define PCI_CHIP_RS690_791F             0x791F
+#define PCI_CHIP_RS600_793F             0x793F
+#define PCI_CHIP_RS600_7941             0x7941
+#define PCI_CHIP_RS600_7942             0x7942
 #define PCI_CHIP_RS740_796C             0x796C
 #define PCI_CHIP_RS740_796D             0x796D
 #define PCI_CHIP_RS740_796E             0x796E
@@ -270,6 +273,7 @@ enum {
    CHIP_FAMILY_R420,
    CHIP_FAMILY_RV410,
    CHIP_FAMILY_RS400,
+   CHIP_FAMILY_RS600,
    CHIP_FAMILY_RS690,
    CHIP_FAMILY_RS740,
    CHIP_FAMILY_RV515,
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 56c22faafa..f7cba5016b 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -685,6 +685,12 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
       screen->chip_family = CHIP_FAMILY_RS400;
       break;
 
+   case PCI_CHIP_RS600_793F:
+   case PCI_CHIP_RS600_7941:
+   case PCI_CHIP_RS600_7942:
+      screen->chip_family = CHIP_FAMILY_RS600;
+      break;
+
    case PCI_CHIP_RS690_791E:
    case PCI_CHIP_RS690_791F:
       screen->chip_family = CHIP_FAMILY_RS690;
@@ -843,7 +849,7 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
    ret = radeonGetParam( sPriv->fd, RADEON_PARAM_FB_LOCATION,
                          &temp);
    if (ret) {
-       if (screen->chip_family < CHIP_FAMILY_RS690)
+       if (screen->chip_family < CHIP_FAMILY_RS600)
 	   screen->fbLocation      = ( INREG( RADEON_MC_FB_LOCATION ) & 0xffff) << 16;
        else {
            FREE( screen );
@@ -854,7 +860,7 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
        screen->fbLocation = (temp & 0xffff) << 16;
    }
 
-   if (screen->chip_family >= CHIP_FAMILY_RV515) {
+   if (screen->chip_family >= CHIP_FAMILY_R300) {
        ret = radeonGetParam( sPriv->fd, RADEON_PARAM_NUM_GB_PIPES,
 			     &temp);
        if (ret) {
diff --git a/src/mesa/glapi/glX_proto_recv.py b/src/mesa/glapi/glX_proto_recv.py
index 923c1958f0..f423c6db08 100644
--- a/src/mesa/glapi/glX_proto_recv.py
+++ b/src/mesa/glapi/glX_proto_recv.py
@@ -277,7 +277,7 @@ class PrintGlxDispatchFunctions(glX_proto_common.glx_print_proto):
 			if param.is_image():
 				offset = f.offset_of( param.name )
 
-				print '%s    %s const %s = (%s) %s(pc + %s);' % (indent, type_string, param.name, type_string, cond, offset)
+				print '%s    %s const %s = (%s) (%s(pc + %s));' % (indent, type_string, param.name, type_string, cond, offset)
 				
 				if param.depth:
 					print '%s    __GLXpixel3DHeader * const hdr = (__GLXpixel3DHeader *)(pc);' % (indent)
diff --git a/src/mesa/glapi/gl_SPARC_asm.py b/src/mesa/glapi/gl_SPARC_asm.py
index 178c00a227..33e752df38 100644
--- a/src/mesa/glapi/gl_SPARC_asm.py
+++ b/src/mesa/glapi/gl_SPARC_asm.py
@@ -39,80 +39,202 @@ class PrintGenericStubs(gl_XML.gl_print_base):
 
 
 	def printRealHeader(self):
-		print '#include "glapioffsets.h"'
+		print '#include "glapi/glapioffsets.h"'
 		print ''
 		print '#ifdef __arch64__'
-		print '#  define GL_STUB(fn,off)\t\t\t\t\\'
-		print 'fn:\t\t\t\t\t\\'
-		print '\tsethi\t%hi(0xDEADBEEF), %g4 ;\t\t\t\\'
-		print '\tsethi\t%hi(0xDEADBEEF), %g1 ;\t\t\t\\'
-		print '\tor\t%g4, %lo(0xDEADBEEF), %g4 ;\t\t\\'
-		print '\tor\t%g1, %lo(0xDEADBEEF), %g1 ;\t\t\\'
-		print '\tsllx\t%g4, 32, %g4 ;\t\t\t\t\\'
-		print '\tldx\t[%g1 + %g4], %g1 ;\t\t\t\\'
-		print '\tsethi\t%hi(8 * off), %g4 ;\t\t\t\\'
-		print '\tor\t%g4, %lo(8 * off), %g4 ;\t\t\\'
-		print '\tldx\t[%g1 + %g4], %g5 ;\t\t\t\\'
-		print '\tjmpl\t%g5, %g0 ;\t\t\t\t\\'
-		print '\tnop'
+		print '#define GL_OFF(N)\t((N) * 8)'
+		print '#define GL_LL\t\tldx'
+		print '#define GL_TIE_LD(SYM)\t%tie_ldx(SYM)'
+		print '#define GL_STACK_SIZE\t128'
 		print '#else'
-		print '#  define GL_STUB(fn,off)\t\t\t\t\\'
-		print 'fn:\t\t\t\t\t\\'
-		print '\tsethi\t%hi(0xDEADBEEF), %g1 ;\t\t\t\\'
-		print '\tld\t[%g1 + %lo(0xDEADBEEF)], %g1 ;\t\t\\'
-		print '\tld\t[%g1 + (4 * off)], %g5 ;\t\t\\'
-		print '\tjmpl\t%g5, %g0 ;\t\t\t\t\\'
-		print '\tnop'
+		print '#define GL_OFF(N)\t((N) * 4)'
+		print '#define GL_LL\t\tld'
+		print '#define GL_TIE_LD(SYM)\t%tie_ld(SYM)'
+		print '#define GL_STACK_SIZE\t64'
 		print '#endif'
 		print ''
-		print '#define GL_STUB_ALIAS(fn,alias) fn = alias'
+		print '#define GLOBL_FN(x) .globl x ; .type x, @function'
+		print '#define HIDDEN(x) .hidden x'
 		print ''
-		print '.text'
-		print '.align 32'
-		print '\t\t.globl __glapi_sparc_icache_flush ; .type __glapi_sparc_icache_flush,#function'
+		print '\t.register %g2, #scratch'
+		print '\t.register %g3, #scratch'
+		print ''
+		print '\t.text'
+		print ''
+		print '\tGLOBL_FN(__glapi_sparc_icache_flush)'
+		print '\tHIDDEN(__glapi_sparc_icache_flush)'
+		print '\t.type\t__glapi_sparc_icache_flush, @function'
 		print '__glapi_sparc_icache_flush: /* %o0 = insn_addr */'
 		print '\tflush\t%o0'
 		print '\tretl'
-		print '\tnop'
+		print '\t nop'
+		print ''
+		print '\t.align\t32'
+		print ''
+		print '\t.type\t__glapi_sparc_get_pc, @function'
+		print '__glapi_sparc_get_pc:'
+		print '\tretl'
+		print '\t add\t%o7, %g2, %g2'
+		print '\t.size\t__glapi_sparc_get_pc, .-__glapi_sparc_get_pc'
+		print ''
+		print '#ifdef GLX_USE_TLS'
+		print ''
+		print '\tGLOBL_FN(__glapi_sparc_get_dispatch)'
+		print '\tHIDDEN(__glapi_sparc_get_dispatch)'
+		print '__glapi_sparc_get_dispatch:'
+		print '\tmov\t%o7, %g1'
+		print '\tsethi\t%hi(_GLOBAL_OFFSET_TABLE_-4), %g2'
+		print '\tcall\t__glapi_sparc_get_pc'
+		print '\tadd\t%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2'
+		print '\tmov\t%g1, %o7'
+		print '\tsethi\t%tie_hi22(_glapi_tls_Dispatch), %g1'
+		print '\tadd\t%g1, %tie_lo10(_glapi_tls_Dispatch), %g1'
+		print '\tGL_LL\t[%g2 + %g1], %g2, GL_TIE_LD(_glapi_tls_Dispatch)'
+		print '\tretl'
+		print '\t mov\t%g2, %o0'
+		print ''
+		print '\t.data'
+		print '\t.align\t32'
+		print ''
+		print '\t/* --> sethi %hi(_glapi_tls_Dispatch), %g1 */'
+		print '\t/* --> or %g1, %lo(_glapi_tls_Dispatch), %g1 */'
+		print '\tGLOBL_FN(__glapi_sparc_tls_stub)'
+		print '\tHIDDEN(__glapi_sparc_tls_stub)'
+		print '__glapi_sparc_tls_stub: /* Call offset in %g3 */'
+		print '\tmov\t%o7, %g1'
+		print '\tsethi\t%hi(_GLOBAL_OFFSET_TABLE_-4), %g2'
+		print '\tcall\t__glapi_sparc_get_pc'
+		print '\tadd\t%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2'
+		print '\tmov\t%g1, %o7'
+		print '\tsrl\t%g3, 10, %g3'
+		print '\tsethi\t%tie_hi22(_glapi_tls_Dispatch), %g1'
+		print '\tadd\t%g1, %tie_lo10(_glapi_tls_Dispatch), %g1'
+		print '\tGL_LL\t[%g2 + %g1], %g2, GL_TIE_LD(_glapi_tls_Dispatch)'
+		print '\tGL_LL\t[%g7+%g2], %g1'
+		print '\tGL_LL\t[%g1 + %g3], %g1'
+		print '\tjmp\t%g1'
+		print '\t nop'
+		print '\t.size\t__glapi_sparc_tls_stub, .-__glapi_sparc_tls_stub'
+		print ''
+		print '#define GL_STUB(fn, off)\t\t\t\t\\'
+		print '\tGLOBL_FN(fn);\t\t\t\t\t\\'
+		print 'fn:\tba\t__glapi_sparc_tls_stub;\t\t\t\\'
+		print '\t sethi\tGL_OFF(off), %g3;\t\t\t\\'
+		print '\t.size\tfn,.-fn;'
+		print ''
+		print '#elif defined(PTHREADS)'
+		print ''
+		print '\t/* 64-bit 0x00 --> sethi %hh(_glapi_Dispatch), %g1 */'
+		print '\t/* 64-bit 0x04 --> sethi %lm(_glapi_Dispatch), %g2 */'
+		print '\t/* 64-bit 0x08 --> or %g1, %hm(_glapi_Dispatch), %g1 */'
+		print '\t/* 64-bit 0x0c --> sllx %g1, 32, %g1 */'
+		print '\t/* 64-bit 0x10 --> add %g1, %g2, %g1 */'
+		print '\t/* 64-bit 0x14 --> ldx [%g1 + %lo(_glapi_Dispatch)], %g1 */'
+		print ''
+		print '\t/* 32-bit 0x00 --> sethi %hi(_glapi_Dispatch), %g1 */'
+		print '\t/* 32-bit 0x04 --> ld [%g1 + %lo(_glapi_Dispatch)], %g1 */'
+		print ''
+		print '\t.data'
+		print '\t.align\t32'
+		print ''
+		print '\tGLOBL_FN(__glapi_sparc_pthread_stub)'
+		print '\tHIDDEN(__glapi_sparc_pthread_stub)'
+		print '__glapi_sparc_pthread_stub: /* Call offset in %g3 */'
+		print '\tmov\t%o7, %g1'
+		print '\tsethi\t%hi(_GLOBAL_OFFSET_TABLE_-4), %g2'
+		print '\tcall\t__glapi_sparc_get_pc'
+		print '\t add\t%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2'
+		print '\tmov\t%g1, %o7'
+		print '\tsethi\t%hi(_glapi_Dispatch), %g1'
+		print '\tor\t%g1, %lo(_glapi_Dispatch), %g1'
+		print '\tsrl\t%g3, 10, %g3'
+		print '\tGL_LL\t[%g2+%g1], %g2'
+		print '\tGL_LL\t[%g2], %g1'
+		print '\tcmp\t%g1, 0'
+		print '\tbe\t2f'
+		print '\t nop'
+		print '1:\tGL_LL\t[%g1 + %g3], %g1'
+		print '\tjmp\t%g1'
+		print '\t nop'
+		print '2:\tsave\t%sp, GL_STACK_SIZE, %sp'
+		print '\tmov\t%g3, %l0'
+		print '\tcall\t_glapi_get_dispatch'
+		print '\t nop'
+		print '\tmov\t%o0, %g1'
+		print '\tmov\t%l0, %g3'
+		print '\tba\t1b'
+		print '\t restore %g0, %g0, %g0'
+		print '\t.size\t__glapi_sparc_pthread_stub, .-__glapi_sparc_pthread_stub'
+		print ''
+		print '#define GL_STUB(fn, off)\t\t\t\\'
+		print '\tGLOBL_FN(fn);\t\t\t\t\\'
+		print 'fn:\tba\t__glapi_sparc_pthread_stub;\t\\'
+		print '\t sethi\tGL_OFF(off), %g3;\t\t\\'
+		print '\t.size\tfn,.-fn;'
+		print ''
+		print '#else /* Non-threaded version. */'
 		print ''
-		print '.data'
-		print '.align 64'
+		print '\t.type	__glapi_sparc_nothread_stub, @function'
+		print '__glapi_sparc_nothread_stub: /* Call offset in %g3 */'
+		print '\tmov\t%o7, %g1'
+		print '\tsethi\t%hi(_GLOBAL_OFFSET_TABLE_-4), %g2'
+		print '\tcall\t__glapi_sparc_get_pc'
+		print '\t add\t%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2'
+		print '\tmov\t%g1, %o7'
+		print '\tsrl\t%g3, 10, %g3'
+		print '\tsethi\t%hi(_glapi_Dispatch), %g1'
+		print '\tor\t%g1, %lo(_glapi_Dispatch), %g1'
+		print '\tGL_LL\t[%g2+%g1], %g2'
+		print '\tGL_LL\t[%g2], %g1'
+		print '\tGL_LL\t[%g1 + %g3], %g1'
+		print '\tjmp\t%g1'
+		print '\t nop'
+		print '\t.size\t__glapi_sparc_nothread_stub, .-__glapi_sparc_nothread_stub'
+		print ''
+		print '#define GL_STUB(fn, off)\t\t\t\\'
+		print '\tGLOBL_FN(fn);\t\t\t\t\\'
+		print 'fn:\tba\t__glapi_sparc_nothread_stub;\t\\'
+		print '\t sethi\tGL_OFF(off), %g3;\t\t\\'
+		print '\t.size\tfn,.-fn;'
+		print ''
+		print '#endif'
+		print ''
+		print '#define GL_STUB_ALIAS(fn, alias)		\\'
+		print '	.globl	fn;				\\'
+		print '	.set	fn, alias'
+		print ''
+		print '\t.text'
+		print '\t.align\t32'
+		print ''
+		print '\t.globl\tgl_dispatch_functions_start'
+		print '\tHIDDEN(gl_dispatch_functions_start)'
+		print 'gl_dispatch_functions_start:'
 		print ''
 		return
 
+	def printRealFooter(self):
+		print ''
+		print '\t.globl\tgl_dispatch_functions_end'
+		print '\tHIDDEN(gl_dispatch_functions_end)'
+		print 'gl_dispatch_functions_end:'
+		return
 
 	def printBody(self, api):
 		for f in api.functionIterateByOffset():
-			if f.is_static_entry_point(f.name):
-				name = f.name
-			else:
-				name = "_dispatch_stub_%u" % (f.offset)
+			name = f.dispatch_name()
 
-			print '\t\t.globl gl%s ; .type gl%s,#function' % (name, name)
+			print '\tGL_STUB(gl%s, _gloffset_%s)' % (name, f.name)
 
-		print '\t\t.globl _mesa_sparc_glapi_begin ; .type _mesa_sparc_glapi_begin,#function'
-		print '_mesa_sparc_glapi_begin:'
-		print ''
+			if not f.is_static_entry_point(f.name):
+				print '\tHIDDEN(gl%s)' % (name)
 
 		for f in api.functionIterateByOffset():
-			if f.is_static_entry_point(f.name):
-				name = f.name
-			else:
-				name = "_dispatch_stub_%u" % (f.offset)
-
-			print '\tGL_STUB(gl%s, _gloffset_%s)' % (name, name)
+			name = f.dispatch_name()
 
-		print ''
-		print '\t\t.globl _mesa_sparc_glapi_end ; .type _mesa_sparc_glapi_end,#function'
-		print '_mesa_sparc_glapi_end:'
-		print ''
-
-
-		for f in api.functionIterateByOffset():
-			for n in f.entry_points:
-				if n != f.name:
-					if f.is_static_entry_point(n):
-						text = '\t.globl gl%s ; .type gl%s,#function ; gl%s = gl%s' % (n, n, n, f.name)
+			if f.is_static_entry_point(f.name):
+				for n in f.entry_points:
+					if n != f.name:
+						text = '\tGL_STUB_ALIAS(gl%s, gl%s)' % (n, f.name)
 
 						if f.has_different_protocol(n):
 							print '#ifndef GLX_INDIRECT_RENDERING'
diff --git a/src/mesa/glapi/glapi.c b/src/mesa/glapi/glapi.c
index e20618abc1..2b105d0f17 100644
--- a/src/mesa/glapi/glapi.c
+++ b/src/mesa/glapi/glapi.c
@@ -318,6 +318,14 @@ extern const GLubyte gl_dispatch_functions_start[];
 # endif
 #endif
 
+#ifdef USE_SPARC_ASM
+#ifdef GLX_USE_TLS
+extern unsigned int __glapi_sparc_tls_stub;
+#else
+extern unsigned int __glapi_sparc_pthread_stub;
+#endif
+#endif
+
 #if !defined(DISPATCH_FUNCTION_SIZE) && !defined(XFree86Server) && !defined(XGLServer)
 # define NEED_FUNCTION_POINTER
 #endif
@@ -344,6 +352,129 @@ init_glapi_relocs( void )
 	curr_func += DISPATCH_FUNCTION_SIZE;
     }
 #endif
+#ifdef USE_SPARC_ASM
+    extern void __glapi_sparc_icache_flush(unsigned int *);
+    static const unsigned int template[] = {
+#ifdef GLX_USE_TLS
+	0x05000000, /* sethi %hi(_glapi_tls_Dispatch), %g2 */
+	0x8730e00a, /* srl %g3, 10, %g3 */
+	0x8410a000, /* or %g2, %lo(_glapi_tls_Dispatch), %g2 */
+#ifdef __arch64__
+	0xc259c002, /* ldx [%g7 + %g2], %g1 */
+	0xc2584003, /* ldx [%g1 + %g3], %g1 */
+#else
+	0xc201c002, /* ld [%g7 + %g2], %g1 */
+	0xc2004003, /* ld [%g1 + %g3], %g1 */
+#endif
+	0x81c04000, /* jmp %g1 */
+	0x01000000, /* nop  */
+#else
+#ifdef __arch64__
+	0x03000000, /* 64-bit 0x00 --> sethi %hh(_glapi_Dispatch), %g1 */
+	0x05000000, /* 64-bit 0x04 --> sethi %lm(_glapi_Dispatch), %g2 */
+	0x82106000, /* 64-bit 0x08 --> or %g1, %hm(_glapi_Dispatch), %g1 */
+	0x8730e00a, /* 64-bit 0x0c --> srl %g3, 10, %g3 */
+	0x83287020, /* 64-bit 0x10 --> sllx %g1, 32, %g1 */
+	0x82004002, /* 64-bit 0x14 --> add %g1, %g2, %g1 */
+	0xc2586000, /* 64-bit 0x18 --> ldx [%g1 + %lo(_glapi_Dispatch)], %g1 */
+#else
+	0x03000000, /* 32-bit 0x00 --> sethi %hi(_glapi_Dispatch), %g1 */
+	0x8730e00a, /* 32-bit 0x04 --> srl %g3, 10, %g3 */
+	0xc2006000, /* 32-bit 0x08 --> ld [%g1 + %lo(_glapi_Dispatch)], %g1 */
+#endif
+	0x80a06000, /*             --> cmp %g1, 0 */
+	0x02800005, /*             --> be +4*5 */
+	0x01000000, /*             -->  nop  */
+#ifdef __arch64__
+	0xc2584003, /* 64-bit      --> ldx [%g1 + %g3], %g1 */
+#else
+	0xc2004003, /* 32-bit      --> ld [%g1 + %g3], %g1 */
+#endif
+	0x81c04000, /*             --> jmp %g1 */
+	0x01000000, /*             --> nop  */
+#ifdef __arch64__
+	0x9de3bf80, /* 64-bit      --> save  %sp, -128, %sp */
+#else
+	0x9de3bfc0, /* 32-bit      --> save  %sp, -64, %sp */
+#endif
+	0xa0100003, /*             --> mov  %g3, %l0 */
+	0x40000000, /*             --> call _glapi_get_dispatch */
+	0x01000000, /*             -->  nop */
+	0x82100008, /*             --> mov %o0, %g1 */
+	0x86100010, /*             --> mov %l0, %g3 */
+	0x10bffff7, /*             --> ba -4*9 */
+	0x81e80000, /*             -->  restore  */
+#endif
+    };
+#ifdef GLX_USE_TLS
+    extern unsigned long __glapi_sparc_get_dispatch(void);
+    unsigned int *code = &__glapi_sparc_tls_stub;
+    unsigned long dispatch = __glapi_sparc_get_dispatch();
+#else
+    unsigned int *code = &__glapi_sparc_pthread_stub;
+    unsigned long dispatch = (unsigned long) &_glapi_Dispatch;
+    unsigned long call_dest = (unsigned long ) &_glapi_get_dispatch;
+    int idx;
+#endif
+
+#if defined(GLX_USE_TLS)
+    code[0] = template[0] | (dispatch >> 10);
+    code[1] = template[1];
+    __glapi_sparc_icache_flush(&code[0]);
+    code[2] = template[2] | (dispatch & 0x3ff);
+    code[3] = template[3];
+    __glapi_sparc_icache_flush(&code[2]);
+    code[4] = template[4];
+    code[5] = template[5];
+    __glapi_sparc_icache_flush(&code[4]);
+    code[6] = template[6];
+    __glapi_sparc_icache_flush(&code[6]);
+#else
+#if defined(__arch64__)
+    code[0] = template[0] | (dispatch >> (32 + 10));
+    code[1] = template[1] | ((dispatch & 0xffffffff) >> 10);
+    __glapi_sparc_icache_flush(&code[0]);
+    code[2] = template[2] | ((dispatch >> 32) & 0x3ff);
+    code[3] = template[3];
+    __glapi_sparc_icache_flush(&code[2]);
+    code[4] = template[4];
+    code[5] = template[5];
+    __glapi_sparc_icache_flush(&code[4]);
+    code[6] = template[6] | (dispatch & 0x3ff);
+    idx = 7;
+#else
+    code[0] = template[0] | (dispatch >> 10);
+    code[1] = template[1];
+    __glapi_sparc_icache_flush(&code[0]);
+    code[2] = template[2] | (dispatch & 0x3ff);
+    idx = 3;
+#endif
+    code[idx + 0] = template[idx + 0];
+    __glapi_sparc_icache_flush(&code[idx - 1]);
+    code[idx + 1] = template[idx + 1];
+    code[idx + 2] = template[idx + 2];
+    __glapi_sparc_icache_flush(&code[idx + 1]);
+    code[idx + 3] = template[idx + 3];
+    code[idx + 4] = template[idx + 4];
+    __glapi_sparc_icache_flush(&code[idx + 3]);
+    code[idx + 5] = template[idx + 5];
+    code[idx + 6] = template[idx + 6];
+    __glapi_sparc_icache_flush(&code[idx + 5]);
+    code[idx + 7] = template[idx + 7];
+    code[idx + 8] = template[idx + 8] |
+	    (((call_dest - ((unsigned long) &code[idx + 8]))
+	      >> 2) & 0x3fffffff);
+    __glapi_sparc_icache_flush(&code[idx + 7]);
+    code[idx + 9] = template[idx + 9];
+    code[idx + 10] = template[idx + 10];
+    __glapi_sparc_icache_flush(&code[idx + 9]);
+    code[idx + 11] = template[idx + 11];
+    code[idx + 12] = template[idx + 12];
+    __glapi_sparc_icache_flush(&code[idx + 11]);
+    code[idx + 13] = template[idx + 13];
+    __glapi_sparc_icache_flush(&code[idx + 13]);
+#endif
+#endif
 }
 #endif /* defined(PTHREADS) || defined(GLX_USE_TLS) */
 
diff --git a/src/mesa/glapi/glapi_getproc.c b/src/mesa/glapi/glapi_getproc.c
index 4adc1aece0..764e2ef98e 100644
--- a/src/mesa/glapi/glapi_getproc.c
+++ b/src/mesa/glapi/glapi_getproc.c
@@ -259,53 +259,30 @@ generate_entrypoint(GLuint functionOffset)
    }
 
    return (_glapi_proc) code;
-#elif defined(USE_SPARC_ASM)
-
-#ifdef __arch64__
-   static const unsigned int insn_template[] = {
-	   0x05000000,	/* sethi	%uhi(_glapi_Dispatch), %g2	*/
-	   0x03000000,	/* sethi	%hi(_glapi_Dispatch), %g1	*/
-	   0x8410a000,	/* or		%g2, %ulo(_glapi_Dispatch), %g2	*/
-	   0x82106000,	/* or		%g1, %lo(_glapi_Dispatch), %g1	*/
-	   0x8528b020,	/* sllx		%g2, 32, %g2			*/
-	   0xc2584002,	/* ldx		[%g1 + %g2], %g1		*/
-	   0x05000000,	/* sethi	%hi(8 * glapioffset), %g2	*/
-	   0x8410a000,	/* or		%g2, %lo(8 * glapioffset), %g2	*/
-	   0xc6584002,	/* ldx		[%g1 + %g2], %g3		*/
-	   0x81c0c000,	/* jmpl		%g3, %g0			*/
-	   0x01000000	/*  nop						*/
+#elif defined(USE_SPARC_ASM) && (defined(PTHREADS) || defined(GLX_USE_TLS))
+   static const unsigned int template[] = {
+      0x07000000, /* sethi %hi(0), %g3 */
+      0x8210000f, /* mov  %o7, %g1 */
+      0x40000000, /* call */
+      0x9e100001, /* mov  %g1, %o7 */
    };
+#ifdef GLX_USE_TLS
+   extern unsigned int __glapi_sparc_tls_stub;
+   unsigned long call_dest = (unsigned long ) &__glapi_sparc_tls_stub;
 #else
-   static const unsigned int insn_template[] = {
-	   0x03000000,	/* sethi	%hi(_glapi_Dispatch), %g1	  */
-	   0xc2006000,	/* ld		[%g1 + %lo(_glapi_Dispatch)], %g1 */
-	   0xc6006000,	/* ld		[%g1 + %lo(4*glapioffset)], %g3	  */
-	   0x81c0c000,	/* jmpl		%g3, %g0			  */
-	   0x01000000	/*  nop						  */
-   };
-#endif /* __arch64__ */
-   unsigned int *code = (unsigned int *) malloc(sizeof(insn_template));
-   unsigned long glapi_addr = (unsigned long) &_glapi_Dispatch;
+   extern unsigned int __glapi_sparc_pthread_stub;
+   unsigned long call_dest = (unsigned long ) &__glapi_sparc_pthread_stub;
+#endif
+   unsigned int *code = (unsigned int *) malloc(sizeof(template));
    if (code) {
-      memcpy(code, insn_template, sizeof(insn_template));
-
-#ifdef __arch64__
-      code[0] |= (glapi_addr >> (32 + 10));
-      code[1] |= ((glapi_addr & 0xffffffff) >> 10);
+      code[0] = template[0] | (functionOffset & 0x3fffff);
+      code[1] = template[1];
       __glapi_sparc_icache_flush(&code[0]);
-      code[2] |= ((glapi_addr >> 32) & ((1 << 10) - 1));
-      code[3] |= (glapi_addr & ((1 << 10) - 1));
+      code[2] = template[2] |
+         (((call_dest - ((unsigned long) &code[2]))
+	   >> 2) & 0x3fffffff);
+      code[3] = template[3];
       __glapi_sparc_icache_flush(&code[2]);
-      code[6] |= ((functionOffset * 8) >> 10);
-      code[7] |= ((functionOffset * 8) & ((1 << 10) - 1));
-      __glapi_sparc_icache_flush(&code[6]);
-#else
-      code[0] |= (glapi_addr >> 10);
-      code[1] |= (glapi_addr & ((1 << 10) - 1));
-      __glapi_sparc_icache_flush(&code[0]);
-      code[2] |= (functionOffset * 4);
-      __glapi_sparc_icache_flush(&code[2]);
-#endif /* __arch64__ */
    }
    return (_glapi_proc) code;
 #else
@@ -337,21 +314,10 @@ fill_in_entrypoint_offset(_glapi_proc entrypoint, GLuint offset)
 #endif
 
 #elif defined(USE_SPARC_ASM)
-
-   /* XXX this hasn't been tested! */
    unsigned int *code = (unsigned int *) entrypoint;
-#ifdef __arch64__
-   code[6] = 0x05000000;  /* sethi	%hi(8 * glapioffset), %g2	*/
-   code[7] = 0x8410a000;  /* or		%g2, %lo(8 * glapioffset), %g2	*/
-   code[6] |= ((offset * 8) >> 10);
-   code[7] |= ((offset * 8) & ((1 << 10) - 1));
-   __glapi_sparc_icache_flush(&code[6]);
-#else /* __arch64__ */
-   code[2] = 0xc6006000;  /* ld		[%g1 + %lo(4*glapioffset)], %g3	  */
-   code[2] |= (offset * 4);
-   __glapi_sparc_icache_flush(&code[2]);
-#endif /* __arch64__ */
-
+   code[0] &= ~0x3fffff;
+   code[0] |= (offset * sizeof(void *)) & 0x3fffff;
+   __glapi_sparc_icache_flush(&code[0]);
 #else
 
    /* an unimplemented architecture */
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index 996033a2d8..de27917709 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -59,6 +59,102 @@
 
 
 /**
+ * glEnable()/glDisable() attribute group (GL_ENABLE_BIT).
+ */
+struct gl_enable_attrib
+{
+   GLboolean AlphaTest;
+   GLboolean AutoNormal;
+   GLboolean Blend;
+   GLbitfield ClipPlanes;
+   GLboolean ColorMaterial;
+   GLboolean ColorTable[COLORTABLE_MAX];
+   GLboolean Convolution1D;
+   GLboolean Convolution2D;
+   GLboolean Separable2D;
+   GLboolean CullFace;
+   GLboolean DepthTest;
+   GLboolean Dither;
+   GLboolean Fog;
+   GLboolean Histogram;
+   GLboolean Light[MAX_LIGHTS];
+   GLboolean Lighting;
+   GLboolean LineSmooth;
+   GLboolean LineStipple;
+   GLboolean IndexLogicOp;
+   GLboolean ColorLogicOp;
+
+   GLboolean Map1Color4;
+   GLboolean Map1Index;
+   GLboolean Map1Normal;
+   GLboolean Map1TextureCoord1;
+   GLboolean Map1TextureCoord2;
+   GLboolean Map1TextureCoord3;
+   GLboolean Map1TextureCoord4;
+   GLboolean Map1Vertex3;
+   GLboolean Map1Vertex4;
+   GLboolean Map1Attrib[16];  /* GL_NV_vertex_program */
+   GLboolean Map2Color4;
+   GLboolean Map2Index;
+   GLboolean Map2Normal;
+   GLboolean Map2TextureCoord1;
+   GLboolean Map2TextureCoord2;
+   GLboolean Map2TextureCoord3;
+   GLboolean Map2TextureCoord4;
+   GLboolean Map2Vertex3;
+   GLboolean Map2Vertex4;
+   GLboolean Map2Attrib[16];  /* GL_NV_vertex_program */
+
+   GLboolean MinMax;
+   GLboolean Normalize;
+   GLboolean PixelTexture;
+   GLboolean PointSmooth;
+   GLboolean PolygonOffsetPoint;
+   GLboolean PolygonOffsetLine;
+   GLboolean PolygonOffsetFill;
+   GLboolean PolygonSmooth;
+   GLboolean PolygonStipple;
+   GLboolean RescaleNormals;
+   GLboolean Scissor;
+   GLboolean Stencil;
+   GLboolean StencilTwoSide;          /* GL_EXT_stencil_two_side */
+   GLboolean MultisampleEnabled;      /* GL_ARB_multisample */
+   GLboolean SampleAlphaToCoverage;   /* GL_ARB_multisample */
+   GLboolean SampleAlphaToOne;        /* GL_ARB_multisample */
+   GLboolean SampleCoverage;          /* GL_ARB_multisample */
+   GLboolean SampleCoverageInvert;    /* GL_ARB_multisample */
+   GLboolean RasterPositionUnclipped; /* GL_IBM_rasterpos_clip */
+
+   GLbitfield Texture[MAX_TEXTURE_UNITS];
+   GLbitfield TexGen[MAX_TEXTURE_UNITS];
+
+   /* SGI_texture_color_table */
+   GLboolean TextureColorTable[MAX_TEXTURE_UNITS];
+
+   /* GL_ARB_vertex_program / GL_NV_vertex_program */
+   GLboolean VertexProgram;
+   GLboolean VertexProgramPointSize;
+   GLboolean VertexProgramTwoSide;
+
+   /* GL_ARB_point_sprite / GL_NV_point_sprite */
+   GLboolean PointSprite;
+   GLboolean FragmentShaderATI;
+};
+
+
+/**
+ * Node for the attribute stack.
+ */
+struct gl_attrib_node
+{
+   GLbitfield kind;
+   void *data;
+   struct gl_attrib_node *next;
+};
+
+
+
+/**
  * Special struct for saving/restoring texture state (GL_TEXTURE_BIT)
  */
 struct texture_state
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 016543da01..c8d160baa9 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.2
+ * Version:  7.5
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -25,7 +26,7 @@
 
 /**
  * \file bufferobj.c
- * \brief Functions for the GL_ARB_vertex_buffer_object extension.
+ * \brief Functions for the GL_ARB_vertex/pixel_buffer_object extensions.
  * \author Brian Paul, Ian Romanick
  */
 
@@ -144,8 +145,7 @@ buffer_object_subdata_range_good( GLcontext * ctx, GLenum target,
 /**
  * Allocate and initialize a new buffer object.
  * 
- * This function is intended to be called via
- * \c dd_function_table::NewBufferObject.
+ * Default callback for the \c dd_function_table::NewBufferObject() hook.
  */
 struct gl_buffer_object *
 _mesa_new_buffer_object( GLcontext *ctx, GLuint name, GLenum target )
@@ -163,8 +163,7 @@ _mesa_new_buffer_object( GLcontext *ctx, GLuint name, GLenum target )
 /**
  * Delete a buffer object.
  * 
- * This function is intended to be called via
- * \c dd_function_table::DeleteBuffer.
+ * Default callback for the \c dd_function_table::DeleteBuffer() hook.
  */
 void
 _mesa_delete_buffer_object( GLcontext *ctx, struct gl_buffer_object *bufObj )
@@ -271,9 +270,8 @@ _mesa_initialize_buffer_object( struct gl_buffer_object *obj,
  * previously stored in the buffer object is lost.  If \c data is \c NULL,
  * memory will be allocated, but no copy will occur.
  *
- * This function is intended to be called via
- * \c dd_function_table::BufferData.  This function need not set GL error
- * codes.  The input parameters will have been tested before calling.
+ * This is the default callback for \c dd_function_table::BufferData()
+ * Note that all GL error checking will have been done already.
  *
  * \param ctx     GL context.
  * \param target  Buffer object target on which to operate.
@@ -312,9 +310,8 @@ _mesa_buffer_data( GLcontext *ctx, GLenum target, GLsizeiptrARB size,
  * specified by \c size + \c offset extends beyond the end of the buffer or
  * if \c data is \c NULL, no copy is performed.
  *
- * This function is intended to be called by
- * \c dd_function_table::BufferSubData.  This function need not set GL error
- * codes.  The input parameters will have been tested before calling.
+ * This is the default callback for \c dd_function_table::BufferSubData()
+ * Note that all GL error checking will have been done already.
  *
  * \param ctx     GL context.
  * \param target  Buffer object target on which to operate.
@@ -346,15 +343,14 @@ _mesa_buffer_subdata( GLcontext *ctx, GLenum target, GLintptrARB offset,
  * specified by \c size + \c offset extends beyond the end of the buffer or
  * if \c data is \c NULL, no copy is performed.
  *
- * This function is intended to be called by
- * \c dd_function_table::BufferGetSubData.  This function need not set GL error
- * codes.  The input parameters will have been tested before calling.
+ * This is the default callback for \c dd_function_table::GetBufferSubData()
+ * Note that all GL error checking will have been done already.
  *
  * \param ctx     GL context.
  * \param target  Buffer object target on which to operate.
- * \param offset  Offset of the first byte to be modified.
+ * \param offset  Offset of the first byte to be fetched.
  * \param size    Size, in bytes, of the data range.
- * \param data    Pointer to the data to store in the buffer object.
+ * \param data    Destination for data
  * \param bufObj  Object to be used.
  *
  * \sa glBufferGetSubDataARB, dd_function_table::GetBufferSubData.
@@ -373,9 +369,7 @@ _mesa_buffer_get_subdata( GLcontext *ctx, GLenum target, GLintptrARB offset,
 
 
 /**
- * Fallback function called via ctx->Driver.MapBuffer().
- * Hardware drivers that really implement buffer objects should never use
- * this function.
+ * Default callback for \c dd_function_tabel::MapBuffer().
  *
  * The function parameters will have been already tested for errors.
  *
@@ -407,9 +401,7 @@ _mesa_buffer_map( GLcontext *ctx, GLenum target, GLenum access,
 
 
 /**
- * Fallback function called via ctx->Driver.MapBuffer().
- * Hardware drivers that really implement buffer objects should never use
- * function.
+ * Default callback for \c dd_function_table::MapBuffer().
  *
  * The input parameters will have been already tested for errors.
  *
@@ -446,6 +438,7 @@ _mesa_init_buffer_objects( GLcontext *ctx )
    ctx->Array.ElementArrayBufferObj = ctx->Array.NullBufferObj;
 }
 
+
 /**
  * Bind the specified target to buffer for the specified context.
  */
@@ -796,11 +789,18 @@ _mesa_DeleteBuffersARB(GLsizei n, const GLuint *ids)
    for (i = 0; i < n; i++) {
       struct gl_buffer_object *bufObj = _mesa_lookup_bufferobj(ctx, ids[i]);
       if (bufObj) {
-         /* unbind any vertex pointers bound to this buffer */
          GLuint j;
 
          ASSERT(bufObj->Name == ids[i]);
 
+         if (bufObj->Pointer) {
+            /* if mapped, unmap it now */
+            ctx->Driver.UnmapBuffer(ctx, 0, bufObj);
+            bufObj->Access = DEFAULT_ACCESS;
+            bufObj->Pointer = NULL;
+         }
+
+         /* unbind any vertex pointers bound to this buffer */
          unbind(ctx, &ctx->Array.ArrayObj->Vertex.BufferObj, bufObj);
          unbind(ctx, &ctx->Array.ArrayObj->Normal.BufferObj, bufObj);
          unbind(ctx, &ctx->Array.ArrayObj->Color.BufferObj, bufObj);
@@ -822,6 +822,7 @@ _mesa_DeleteBuffersARB(GLsizei n, const GLuint *ids)
             _mesa_BindBufferARB( GL_ELEMENT_ARRAY_BUFFER_ARB, 0 );
          }
 
+         /* unbind any pixel pack/unpack pointers bound to this buffer */
          if (ctx->Pack.BufferObj == bufObj) {
             _mesa_BindBufferARB( GL_PIXEL_PACK_BUFFER_EXT, 0 );
          }
@@ -951,8 +952,10 @@ _mesa_BufferDataARB(GLenum target, GLsizeiptrARB size,
    }
    
    if (bufObj->Pointer) {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "glBufferDataARB(buffer is mapped)" );
-      return;
+      /* Unmap the existing buffer.  We'll replace it now.  Not an error. */
+      ctx->Driver.UnmapBuffer(ctx, target, bufObj);
+      bufObj->Access = DEFAULT_ACCESS;
+      bufObj->Pointer = NULL;
    }  
 
    ASSERT(ctx->Driver.BufferData);
@@ -1068,10 +1071,7 @@ _mesa_UnmapBufferARB(GLenum target)
       return GL_FALSE;
    }
 
-   if (ctx->Driver.UnmapBuffer) {
-      status = ctx->Driver.UnmapBuffer( ctx, target, bufObj );
-   }
-
+   status = ctx->Driver.UnmapBuffer( ctx, target, bufObj );
    bufObj->Access = DEFAULT_ACCESS;
    bufObj->Pointer = NULL;
 
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index a94bbf3fdf..97ee925c0b 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -351,6 +351,36 @@ _mesa_destroy_visual( GLvisual *vis )
 /**********************************************************************/
 /*@{*/
 
+
+/**
+ * This is lame.  gdb only seems to recognize enum types that are
+ * actually used somewhere.  We want to be able to print/use enum
+ * values such as TEXTURE_2D_INDEX in gdb.  But we don't actually use
+ * the gl_texture_index type anywhere.  Thus, this lame function.
+ */
+static void
+dummy_enum_func(void)
+{
+   gl_buffer_index bi;
+   gl_colortable_index ci;
+   gl_face_index fi;
+   gl_frag_attrib fa;
+   gl_frag_result fr;
+   gl_texture_index ti;
+   gl_vert_attrib va;
+   gl_vert_result vr;
+
+   (void) bi;
+   (void) ci;
+   (void) fi;
+   (void) fa;
+   (void) fr;
+   (void) ti;
+   (void) va;
+   (void) vr;
+}
+
+
 /**
  * One-time initialization mutex lock.
  *
@@ -390,9 +420,6 @@ one_time_init( GLcontext *ctx )
          _mesa_ubyte_to_float_color_tab[i] = (float) i / 255.0F;
       }
 
-#ifdef USE_SPARC_ASM
-      _mesa_init_sparc_glapi_relocs();
-#endif
       if (_mesa_getenv("MESA_DEBUG")) {
          _glapi_noop_enable_warnings(GL_TRUE);
          _glapi_set_warning_func( (_glapi_warning_func) _mesa_warning );
@@ -409,6 +436,8 @@ one_time_init( GLcontext *ctx )
       alreadyCalled = GL_TRUE;
    }
    _glthread_UNLOCK_MUTEX(OneTimeLock);
+
+   dummy_enum_func();
 }
 
 
diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c
index 13cfa0e756..e9de0c097a 100644
--- a/src/mesa/main/drawpix.c
+++ b/src/mesa/main/drawpix.c
@@ -103,7 +103,7 @@ _mesa_DrawPixels( GLsizei width, GLsizei height,
    else if (ctx->RenderMode == GL_FEEDBACK) {
       /* Feedback the current raster pos info */
       FLUSH_CURRENT( ctx, 0 );
-      FEEDBACK_TOKEN( ctx, (GLfloat) (GLint) GL_DRAW_PIXEL_TOKEN );
+      _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_DRAW_PIXEL_TOKEN );
       _mesa_feedback_vertex( ctx,
                              ctx->Current.RasterPos,
                              ctx->Current.RasterColor,
@@ -166,7 +166,7 @@ _mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height,
    }
    else if (ctx->RenderMode == GL_FEEDBACK) {
       FLUSH_CURRENT( ctx, 0 );
-      FEEDBACK_TOKEN( ctx, (GLfloat) (GLint) GL_COPY_PIXEL_TOKEN );
+      _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_COPY_PIXEL_TOKEN );
       _mesa_feedback_vertex( ctx, 
                              ctx->Current.RasterPos,
                              ctx->Current.RasterColor,
@@ -243,7 +243,7 @@ _mesa_Bitmap( GLsizei width, GLsizei height,
 #if _HAVE_FULL_GL
    else if (ctx->RenderMode == GL_FEEDBACK) {
       FLUSH_CURRENT(ctx, 0);
-      FEEDBACK_TOKEN( ctx, (GLfloat) (GLint) GL_BITMAP_TOKEN );
+      _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_BITMAP_TOKEN );
       _mesa_feedback_vertex( ctx,
                              ctx->Current.RasterPos,
                              ctx->Current.RasterColor,
@@ -311,7 +311,7 @@ _mesa_DrawDepthPixelsMESA( GLsizei width, GLsizei height,
    else if (ctx->RenderMode == GL_FEEDBACK) {
       /* Feedback the current raster pos info */
       FLUSH_CURRENT( ctx, 0 );
-      FEEDBACK_TOKEN( ctx, (GLfloat) (GLint) GL_DRAW_PIXEL_TOKEN );
+      _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_DRAW_PIXEL_TOKEN );
       _mesa_feedback_vertex( ctx,
                              ctx->Current.RasterPos,
                              ctx->Current.RasterColor,
diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c
index 23b3fb68fb..151e29053a 100644
--- a/src/mesa/main/fbobject.c
+++ b/src/mesa/main/fbobject.c
@@ -1229,19 +1229,26 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
    ASSERT(newFb != &DummyFramebuffer);
 
    /*
-    * XXX check if re-binding same buffer and skip some of this code.
+    * OK, now bind the new Draw/Read framebuffers, if they're changing.
     */
 
    if (bindReadBuf) {
-      _mesa_reference_framebuffer(&ctx->ReadBuffer, newFbread);
+      if (ctx->ReadBuffer == newFbread)
+         bindReadBuf = GL_FALSE; /* no change */
+      else
+         _mesa_reference_framebuffer(&ctx->ReadBuffer, newFbread);
    }
 
    if (bindDrawBuf) {
       /* check if old FB had any texture attachments */
-      check_end_texture_render(ctx, ctx->DrawBuffer);
+      if (ctx->DrawBuffer->Name != 0) {
+         check_end_texture_render(ctx, ctx->DrawBuffer);
+      }
 
-      /* check if time to delete this framebuffer */
-      _mesa_reference_framebuffer(&ctx->DrawBuffer, newFb);
+      if (ctx->DrawBuffer == newFb)
+         bindDrawBuf = GL_FALSE; /* no change */
+      else
+         _mesa_reference_framebuffer(&ctx->DrawBuffer, newFb);
 
       if (newFb->Name != 0) {
          /* check if newly bound framebuffer has any texture attachments */
@@ -1249,7 +1256,7 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
       }
    }
 
-   if (ctx->Driver.BindFramebuffer) {
+   if ((bindDrawBuf || bindReadBuf) && ctx->Driver.BindFramebuffer) {
       ctx->Driver.BindFramebuffer(ctx, target, newFb, newFbread);
    }
 }
diff --git a/src/mesa/main/feedback.c b/src/mesa/main/feedback.c
index 48c2ccbff3..beab535b15 100644
--- a/src/mesa/main/feedback.c
+++ b/src/mesa/main/feedback.c
@@ -1,13 +1,9 @@
-/**
- * \file feedback.c
- * Selection and feedback modes functions.
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  5.1
+ * Version:  7.5
  *
- * Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,6 +23,11 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+/**
+ * \file feedback.c
+ * Selection and feedback modes functions.
+ */
+
 
 #include "glheader.h"
 #include "colormac.h"
@@ -110,60 +111,49 @@ _mesa_PassThrough( GLfloat token )
 
    if (ctx->RenderMode==GL_FEEDBACK) {
       FLUSH_VERTICES(ctx, 0);
-      FEEDBACK_TOKEN( ctx, (GLfloat) (GLint) GL_PASS_THROUGH_TOKEN );
-      FEEDBACK_TOKEN( ctx, token );
+      _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_PASS_THROUGH_TOKEN );
+      _mesa_feedback_token( ctx, token );
    }
 }
 
 
-
-/*
+/**
  * Put a vertex into the feedback buffer.
  */
-void _mesa_feedback_vertex( GLcontext *ctx,
-                            const GLfloat win[4],
-                            const GLfloat color[4],
-                            GLfloat index,
-                            const GLfloat texcoord[4] )
+void
+_mesa_feedback_vertex(GLcontext *ctx,
+                      const GLfloat win[4],
+                      const GLfloat color[4],
+                      GLfloat index,
+                      const GLfloat texcoord[4])
 {
-#if 0
-   {
-      /* snap window x, y to fractional pixel position */
-      const GLint snapMask = ~((FIXED_ONE / (1 << SUB_PIXEL_BITS)) - 1);
-      GLfixed x, y;
-      x = FloatToFixed(win[0]) & snapMask;
-      y = FloatToFixed(win[1]) & snapMask;
-      FEEDBACK_TOKEN(ctx, FixedToFloat(x));
-      FEEDBACK_TOKEN(ctx, FixedToFloat(y) );
-   }
-#else
-   FEEDBACK_TOKEN( ctx, win[0] );
-   FEEDBACK_TOKEN( ctx, win[1] );
-#endif
+   _mesa_feedback_token( ctx, win[0] );
+   _mesa_feedback_token( ctx, win[1] );
    if (ctx->Feedback._Mask & FB_3D) {
-      FEEDBACK_TOKEN( ctx, win[2] );
+      _mesa_feedback_token( ctx, win[2] );
    }
    if (ctx->Feedback._Mask & FB_4D) {
-      FEEDBACK_TOKEN( ctx, win[3] );
+      _mesa_feedback_token( ctx, win[3] );
    }
    if (ctx->Feedback._Mask & FB_INDEX) {
-      FEEDBACK_TOKEN( ctx, (GLfloat) index );
+      _mesa_feedback_token( ctx, (GLfloat) index );
    }
    if (ctx->Feedback._Mask & FB_COLOR) {
-      FEEDBACK_TOKEN( ctx, color[0] );
-      FEEDBACK_TOKEN( ctx, color[1] );
-      FEEDBACK_TOKEN( ctx, color[2] );
-      FEEDBACK_TOKEN( ctx, color[3] );
+      _mesa_feedback_token( ctx, color[0] );
+      _mesa_feedback_token( ctx, color[1] );
+      _mesa_feedback_token( ctx, color[2] );
+      _mesa_feedback_token( ctx, color[3] );
    }
    if (ctx->Feedback._Mask & FB_TEXTURE) {
-      FEEDBACK_TOKEN( ctx, texcoord[0] );
-      FEEDBACK_TOKEN( ctx, texcoord[1] );
-      FEEDBACK_TOKEN( ctx, texcoord[2] );
-      FEEDBACK_TOKEN( ctx, texcoord[3] );
+      _mesa_feedback_token( ctx, texcoord[0] );
+      _mesa_feedback_token( ctx, texcoord[1] );
+      _mesa_feedback_token( ctx, texcoord[2] );
+      _mesa_feedback_token( ctx, texcoord[3] );
    }
 }
 
-#endif
+
+#endif /* _HAVE_FULL_GL */
 
 
 /**********************************************************************/
@@ -213,11 +203,14 @@ _mesa_SelectBuffer( GLsizei size, GLuint *buffer )
  * Verifies there is free space in the buffer to write the value and
  * increments the pointer.
  */
-#define WRITE_RECORD( CTX, V )					\
-	if (CTX->Select.BufferCount < CTX->Select.BufferSize) {	\
-	   CTX->Select.Buffer[CTX->Select.BufferCount] = (V);	\
-	}							\
-	CTX->Select.BufferCount++;
+static INLINE void
+write_record(GLcontext *ctx, GLuint value)
+{
+   if (ctx->Select.BufferCount < ctx->Select.BufferSize) {
+      ctx->Select.Buffer[ctx->Select.BufferCount] = value;
+   }
+   ctx->Select.BufferCount++;
+}
 
 
 /**
@@ -229,7 +222,8 @@ _mesa_SelectBuffer( GLsizei size, GLuint *buffer )
  * Sets gl_selection::HitFlag and updates gl_selection::HitMinZ and
  * gl_selection::HitMaxZ.
  */
-void _mesa_update_hitflag( GLcontext *ctx, GLfloat z )
+void
+_mesa_update_hitflag(GLcontext *ctx, GLfloat z)
 {
    ctx->Select.HitFlag = GL_TRUE;
    if (z < ctx->Select.HitMinZ) {
@@ -252,7 +246,8 @@ void _mesa_update_hitflag( GLcontext *ctx, GLfloat z )
  *
  * \sa gl_selection.
  */
-static void write_hit_record( GLcontext *ctx )
+static void
+write_hit_record(GLcontext *ctx)
 {
    GLuint i;
    GLuint zmin, zmax, zscale = (~0u);
@@ -264,11 +259,11 @@ static void write_hit_record( GLcontext *ctx )
    zmin = (GLuint) ((GLfloat) zscale * ctx->Select.HitMinZ);
    zmax = (GLuint) ((GLfloat) zscale * ctx->Select.HitMaxZ);
 
-   WRITE_RECORD( ctx, ctx->Select.NameStackDepth );
-   WRITE_RECORD( ctx, zmin );
-   WRITE_RECORD( ctx, zmax );
+   write_record( ctx, ctx->Select.NameStackDepth );
+   write_record( ctx, zmin );
+   write_record( ctx, zmax );
    for (i = 0; i < ctx->Select.NameStackDepth; i++) {
-      WRITE_RECORD( ctx, ctx->Select.NameStack[i] );
+      write_record( ctx, ctx->Select.NameStack[i] );
    }
 
    ctx->Select.Hits++;
diff --git a/src/mesa/main/feedback.h b/src/mesa/main/feedback.h
index 6c448ad631..72c2acd5ed 100644
--- a/src/mesa/main/feedback.h
+++ b/src/mesa/main/feedback.h
@@ -1,13 +1,9 @@
-/**
- * \file feedback.h
- * Selection and feedback modes functions.
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  3.5
+ * Version:  7.5
  *
- * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,7 +23,6 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
-
 #ifndef FEEDBACK_H
 #define FEEDBACK_H
 
@@ -35,23 +30,29 @@
 #include "mtypes.h"
 
 
-#define FEEDBACK_TOKEN( CTX, T )				\
-	if (CTX->Feedback.Count < CTX->Feedback.BufferSize) {	\
-	   CTX->Feedback.Buffer[CTX->Feedback.Count] = (GLfloat) (T); \
-	}							\
-	CTX->Feedback.Count++;
+extern void
+_mesa_init_feedback( GLcontext *ctx );
 
+extern void
+_mesa_feedback_vertex( GLcontext *ctx,
+                       const GLfloat win[4],
+                       const GLfloat color[4],
+                       GLfloat index,
+                       const GLfloat texcoord[4] );
 
-extern void _mesa_init_feedback( GLcontext * ctx );
 
-extern void _mesa_feedback_vertex( GLcontext *ctx,
-                                const GLfloat win[4],
-                                const GLfloat color[4],
-				GLfloat index,
-                                const GLfloat texcoord[4] );
+static INLINE void
+_mesa_feedback_token( GLcontext *ctx, GLfloat token )
+{
+   if (ctx->Feedback.Count < ctx->Feedback.BufferSize) {
+      ctx->Feedback.Buffer[ctx->Feedback.Count] = token;
+   }
+   ctx->Feedback.Count++;
+}
 
 
-extern void _mesa_update_hitflag( GLcontext *ctx, GLfloat z );
+extern void
+_mesa_update_hitflag( GLcontext *ctx, GLfloat z );
 
 
 extern void GLAPIENTRY
diff --git a/src/mesa/main/glheader.h b/src/mesa/main/glheader.h
index 482a36de93..ad095321e3 100644
--- a/src/mesa/main/glheader.h
+++ b/src/mesa/main/glheader.h
@@ -52,7 +52,7 @@
 #define GL_GLEXT_PROTOTYPES
 #include "GL/gl.h"
 #include "GL/glext.h"
-#include <GL/internal/glcore.h>
+#include "GL/internal/glcore.h"
 
 
 #ifndef GL_FIXED
diff --git a/src/mesa/main/imports.h b/src/mesa/main/imports.h
index 4192f037c0..7b61e22e93 100644
--- a/src/mesa/main/imports.h
+++ b/src/mesa/main/imports.h
@@ -256,15 +256,7 @@ static INLINE int GET_FLOAT_BITS( float x )
 /***
  *** IROUND: return (as an integer) float rounded to nearest integer
  ***/
-#if defined(USE_SPARC_ASM) && defined(__GNUC__) && defined(__sparc__)
-static INLINE int iround(float f)
-{
-   int r;
-   __asm__ ("fstoi %1, %0" : "=f" (r) : "f" (f));
-   return r;
-}
-#define IROUND(x)  iround(x)
-#elif defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__) && \
+#if defined(USE_X86_ASM) && defined(__GNUC__) && defined(__i386__) && \
 			(!(defined(__BEOS__) || defined(__HAIKU__))  || \
 			(__GNUC__ > 2 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 95)))
 static INLINE int iround(float f)
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index e80d3db043..f17b9e1e71 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -38,7 +38,6 @@
 #include "main/config.h"
 #include "main/compiler.h"
 #include "main/mfeatures.h"
-#include "main/bitset.h"
 #include "glapi/glapi.h"
 #include "math/m_matrix.h"	/* GLmatrix */
 
@@ -79,42 +78,11 @@
 
 
 /**
- * Fixed point data type.
- */
-typedef int GLfixed;
-/*
- * Fixed point arithmetic macros
- */
-#ifndef FIXED_FRAC_BITS
-#define FIXED_FRAC_BITS 11
-#endif
-
-#define FIXED_SHIFT     FIXED_FRAC_BITS
-#define FIXED_ONE       (1 << FIXED_SHIFT)
-#define FIXED_HALF      (1 << (FIXED_SHIFT-1))
-#define FIXED_FRAC_MASK (FIXED_ONE - 1)
-#define FIXED_INT_MASK  (~FIXED_FRAC_MASK)
-#define FIXED_EPSILON   1
-#define FIXED_SCALE     ((float) FIXED_ONE)
-#define FIXED_DBL_SCALE ((double) FIXED_ONE)
-#define FloatToFixed(X) (IROUND((X) * FIXED_SCALE))
-#define FixedToDouble(X) ((X) * (1.0 / FIXED_DBL_SCALE))
-#define IntToFixed(I)   ((I) << FIXED_SHIFT)
-#define FixedToInt(X)   ((X) >> FIXED_SHIFT)
-#define FixedToUns(X)   (((unsigned int)(X)) >> FIXED_SHIFT)
-#define FixedCeil(X)    (((X) + FIXED_ONE - FIXED_EPSILON) & FIXED_INT_MASK)
-#define FixedFloor(X)   ((X) & FIXED_INT_MASK)
-#define FixedToFloat(X) ((X) * (1.0F / FIXED_SCALE))
-#define PosFloatToFixed(X)      FloatToFixed(X)
-#define SignedFloatToFixed(X)   FloatToFixed(X)
-
-
-
-/**
  * \name Some forward type declarations
  */
 /*@{*/
 struct _mesa_HashTable;
+struct gl_attrib_node;
 struct gl_pixelstore_attrib;
 struct gl_program_cache;
 struct gl_texture_format;
@@ -135,7 +103,7 @@ typedef struct gl_framebuffer GLframebuffer;
  * In GL_ARB_vertex_shader / OpenGL 2.0 the aliasing is disallowed (the
  * generic attributes are distinct/separate).
  */
-enum
+typedef enum
 {
    VERT_ATTRIB_POS = 0,
    VERT_ATTRIB_WEIGHT = 1,
@@ -171,7 +139,7 @@ enum
    VERT_ATTRIB_GENERIC14 = 30,
    VERT_ATTRIB_GENERIC15 = 31,
    VERT_ATTRIB_MAX = 32
-};
+} gl_vert_attrib;
 
 /**
  * Bitflags for vertex attributes.
@@ -219,8 +187,7 @@ enum
 /**
  * Indexes for vertex program result attributes
  */
-/*@{*/
-enum
+typedef enum
 {
    VERT_RESULT_HPOS = 0,
    VERT_RESULT_COL0 = 1,
@@ -240,14 +207,13 @@ enum
    VERT_RESULT_EDGE = 15,
    VERT_RESULT_VAR0 = 16  /**< shader varying */,
    VERT_RESULT_MAX = (VERT_RESULT_VAR0 + MAX_VARYING)
-};
-/*@}*/
+} gl_vert_result;
 
 
 /**
  * Indexes for fragment program input attributes.
  */
-enum
+typedef enum
 {
    FRAG_ATTRIB_WPOS = 0,
    FRAG_ATTRIB_COL0 = 1,
@@ -263,7 +229,7 @@ enum
    FRAG_ATTRIB_TEX7 = 11,
    FRAG_ATTRIB_VAR0 = 12,  /**< shader varying */
    FRAG_ATTRIB_MAX = (FRAG_ATTRIB_VAR0 + MAX_VARYING)
-};
+} gl_frag_attrib;
 
 /**
  * Bitflags for fragment program input attributes.
@@ -300,41 +266,44 @@ enum
 /**
  * Fragment program results
  */
-enum
+typedef enum
 {
-   FRAG_RESULT_COLR = 0,
-   FRAG_RESULT_COLH = 1,
-   FRAG_RESULT_DEPR = 2,
-   FRAG_RESULT_DATA0 = 3,
+   FRAG_RESULT_DEPTH = 0,
+   FRAG_RESULT_COLOR = 1,
+   FRAG_RESULT_DATA0 = 2,
    FRAG_RESULT_MAX = (FRAG_RESULT_DATA0 + MAX_DRAW_BUFFERS)
-};
+} gl_frag_result;
 
 
 /**
  * Indexes for all renderbuffers
  */
-enum {
-   BUFFER_FRONT_LEFT  = 0,  /* the four standard color buffers */
-   BUFFER_BACK_LEFT   = 1,
-   BUFFER_FRONT_RIGHT = 2,
-   BUFFER_BACK_RIGHT  = 3,
-   BUFFER_AUX0        = 4,  /* optional aux buffer */
-   BUFFER_AUX1        = 5,
-   BUFFER_AUX2        = 6,
-   BUFFER_AUX3        = 7,
-   BUFFER_DEPTH       = 8,
-   BUFFER_STENCIL     = 9,
-   BUFFER_ACCUM       = 10,
-   BUFFER_COLOR0      = 11, /* generic renderbuffers */
-   BUFFER_COLOR1      = 12,
-   BUFFER_COLOR2      = 13,
-   BUFFER_COLOR3      = 14,
-   BUFFER_COLOR4      = 15,
-   BUFFER_COLOR5      = 16,
-   BUFFER_COLOR6      = 17,
-   BUFFER_COLOR7      = 18,
-   BUFFER_COUNT       = 19
-};
+typedef enum
+{
+   /* the four standard color buffers */
+   BUFFER_FRONT_LEFT,
+   BUFFER_BACK_LEFT,
+   BUFFER_FRONT_RIGHT,
+   BUFFER_BACK_RIGHT,
+   /* optional aux buffers */
+   BUFFER_AUX0,
+   BUFFER_AUX1,
+   BUFFER_AUX2,
+   BUFFER_AUX3,
+   BUFFER_DEPTH,
+   BUFFER_STENCIL,
+   BUFFER_ACCUM,
+   /* generic renderbuffers */
+   BUFFER_COLOR0,
+   BUFFER_COLOR1,
+   BUFFER_COLOR2,
+   BUFFER_COLOR3,
+   BUFFER_COLOR4,
+   BUFFER_COLOR5,
+   BUFFER_COLOR6,
+   BUFFER_COLOR7,
+   BUFFER_COUNT
+} gl_buffer_index;
 
 /**
  * Bit flags for all renderbuffers
@@ -381,12 +350,13 @@ enum {
 
 
 /** The pixel transfer path has three color tables: */
-/*@{*/
-#define COLORTABLE_PRECONVOLUTION  0
-#define COLORTABLE_POSTCONVOLUTION 1
-#define COLORTABLE_POSTCOLORMATRIX 2
-#define COLORTABLE_MAX 3
-/*@}*/
+typedef enum
+{
+   COLORTABLE_PRECONVOLUTION,
+   COLORTABLE_POSTCONVOLUTION,
+   COLORTABLE_POSTCOLORMATRIX,
+   COLORTABLE_MAX
+} gl_colortable_index;
 
 
 /**
@@ -623,9 +593,7 @@ struct gl_current_attrib
     * \note Index and Edgeflag current values are stored as floats in the 
     * SIX and SEVEN attribute slots.
     */
-   /*@{*/
    GLfloat Attrib[VERT_ATTRIB_MAX][4];	/**< Position, color, texcoords, etc */
-   /*@}*/
 
    /**
     * \name Current raster position attributes (always valid).
@@ -658,84 +626,6 @@ struct gl_depthbuffer_attrib
 
 
 /**
- * glEnable()/glDisable() attribute group (GL_ENABLE_BIT).
- */
-struct gl_enable_attrib
-{
-   GLboolean AlphaTest;
-   GLboolean AutoNormal;
-   GLboolean Blend;
-   GLbitfield ClipPlanes;
-   GLboolean ColorMaterial;
-   GLboolean ColorTable[COLORTABLE_MAX];
-   GLboolean Convolution1D;
-   GLboolean Convolution2D;
-   GLboolean Separable2D;
-   GLboolean CullFace;
-   GLboolean DepthTest;
-   GLboolean Dither;
-   GLboolean Fog;
-   GLboolean Histogram;
-   GLboolean Light[MAX_LIGHTS];
-   GLboolean Lighting;
-   GLboolean LineSmooth;
-   GLboolean LineStipple;
-   GLboolean IndexLogicOp;
-   GLboolean ColorLogicOp;
-   GLboolean Map1Color4;
-   GLboolean Map1Index;
-   GLboolean Map1Normal;
-   GLboolean Map1TextureCoord1;
-   GLboolean Map1TextureCoord2;
-   GLboolean Map1TextureCoord3;
-   GLboolean Map1TextureCoord4;
-   GLboolean Map1Vertex3;
-   GLboolean Map1Vertex4;
-   GLboolean Map1Attrib[16];  /* GL_NV_vertex_program */
-   GLboolean Map2Color4;
-   GLboolean Map2Index;
-   GLboolean Map2Normal;
-   GLboolean Map2TextureCoord1;
-   GLboolean Map2TextureCoord2;
-   GLboolean Map2TextureCoord3;
-   GLboolean Map2TextureCoord4;
-   GLboolean Map2Vertex3;
-   GLboolean Map2Vertex4;
-   GLboolean Map2Attrib[16];  /* GL_NV_vertex_program */
-   GLboolean MinMax;
-   GLboolean Normalize;
-   GLboolean PixelTexture;
-   GLboolean PointSmooth;
-   GLboolean PolygonOffsetPoint;
-   GLboolean PolygonOffsetLine;
-   GLboolean PolygonOffsetFill;
-   GLboolean PolygonSmooth;
-   GLboolean PolygonStipple;
-   GLboolean RescaleNormals;
-   GLboolean Scissor;
-   GLboolean Stencil;
-   GLboolean StencilTwoSide;          /* GL_EXT_stencil_two_side */
-   GLboolean MultisampleEnabled;      /* GL_ARB_multisample */
-   GLboolean SampleAlphaToCoverage;   /* GL_ARB_multisample */
-   GLboolean SampleAlphaToOne;        /* GL_ARB_multisample */
-   GLboolean SampleCoverage;          /* GL_ARB_multisample */
-   GLboolean SampleCoverageInvert;    /* GL_ARB_multisample */
-   GLboolean RasterPositionUnclipped; /* GL_IBM_rasterpos_clip */
-   GLuint Texture[MAX_TEXTURE_UNITS];
-   GLuint TexGen[MAX_TEXTURE_UNITS];
-   /* SGI_texture_color_table */
-   GLboolean TextureColorTable[MAX_TEXTURE_UNITS];
-   /* GL_ARB_vertex_program / GL_NV_vertex_program */
-   GLboolean VertexProgram;
-   GLboolean VertexProgramPointSize;
-   GLboolean VertexProgramTwoSide;
-   /* GL_ARB_point_sprite / GL_NV_point_sprite */
-   GLboolean PointSprite;
-   GLboolean FragmentShaderATI;
-};
-
-
-/**
  * Evaluator attribute group (GL_EVAL_BIT).
  */
 struct gl_eval_attrib
@@ -822,11 +712,11 @@ struct gl_hint_attrib
  */
 struct gl_histogram_attrib
 {
-   GLuint Width;				/**< number of table entries */
-   GLint Format;				/**< GL_ALPHA, GL_RGB, etc */
-   GLuint Count[HISTOGRAM_TABLE_SIZE][4];	/**< the histogram */
-   GLboolean Sink;				/**< terminate image transfer? */
-   GLubyte RedSize;				/**< Bits per counter */
+   GLuint Width;                            /**< number of table entries */
+   GLint Format;                            /**< GL_ALPHA, GL_RGB, etc */
+   GLuint Count[HISTOGRAM_TABLE_SIZE][4];   /**< the histogram */
+   GLboolean Sink;                          /**< terminate image transfer? */
+   GLubyte RedSize;                         /**< Bits per counter */
    GLubyte GreenSize;
    GLubyte BlueSize;
    GLubyte AlphaSize;
@@ -1005,13 +895,15 @@ struct gl_pixel_attrib
    /*--- Begin Pixel Transfer State ---*/
    /* Fields are in the order in which they're applied... */
 
-   /* Scale & Bias (index shift, offset) */
+   /** Scale & Bias (index shift, offset) */
+   /*@{*/
    GLfloat RedBias, RedScale;
    GLfloat GreenBias, GreenScale;
    GLfloat BlueBias, BlueScale;
    GLfloat AlphaBias, AlphaScale;
    GLfloat DepthBias, DepthScale;
    GLint IndexShift, IndexOffset;
+   /*@}*/
 
    /* Pixel Maps */
    /* Note: actual pixel maps are not part of this attrib group */
@@ -1027,7 +919,7 @@ struct gl_pixel_attrib
    GLboolean Convolution1DEnabled;
    GLboolean Convolution2DEnabled;
    GLboolean Separable2DEnabled;
-   GLfloat ConvolutionBorderColor[3][4];
+   GLfloat ConvolutionBorderColor[3][4];  /**< RGBA */
    GLenum ConvolutionBorderMode[3];
    GLfloat ConvolutionFilterScale[3][4];  /**< RGBA */
    GLfloat ConvolutionFilterBias[3][4];   /**< RGBA */
@@ -1144,7 +1036,8 @@ struct gl_stencil_attrib
  * target target enums, such as GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP, etc.
  * Note: the order is from highest priority to lowest priority.
  */
-enum {
+typedef enum
+{
    TEXTURE_2D_ARRAY_INDEX,
    TEXTURE_1D_ARRAY_INDEX,
    TEXTURE_CUBE_INDEX,
@@ -1153,7 +1046,7 @@ enum {
    TEXTURE_2D_INDEX,
    TEXTURE_1D_INDEX,
    NUM_TEXTURE_TARGETS
-};
+} gl_texture_index;
 
 
 /**
@@ -1383,15 +1276,16 @@ struct gl_texture_image
 /**
  * Indexes for cube map faces.
  */
-/*@{*/
-#define FACE_POS_X   0
-#define FACE_NEG_X   1
-#define FACE_POS_Y   2
-#define FACE_NEG_Y   3
-#define FACE_POS_Z   4
-#define FACE_NEG_Z   5
-#define MAX_FACES    6
-/*@}*/
+typedef enum
+{
+   FACE_POS_X = 0,
+   FACE_NEG_X = 1,
+   FACE_POS_Y = 2,
+   FACE_NEG_Y = 3,
+   FACE_POS_Z = 4,
+   FACE_NEG_Z = 5,
+   MAX_FACES = 6
+} gl_face_index;
 
 
 /**
@@ -1538,24 +1432,12 @@ struct gl_texture_unit
 };
 
 
-
 /**
  * Texture attribute group (GL_TEXTURE_BIT).
  */
 struct gl_texture_attrib
 {
-   /**
-    * name multitexture 
-    */
-   /**@{*/
-   GLuint CurrentUnit;   /**< Active texture unit [0, MaxTextureImageUnits-1] */
-   GLbitfield _EnabledUnits;  /**< one bit set for each really-enabled unit */
-   GLbitfield _EnabledCoordUnits;   /**< one bit per enabled coordinate unit */
-   GLbitfield _GenFlags;            /**< for texgen */
-   GLbitfield _TexGenEnabled;  /**< Mask of ENABLE_TEXGEN flags */
-   GLbitfield _TexMatEnabled;  /**< Mask of ENABLE_TEXMAT flags */
-   /**@}*/
-
+   GLuint CurrentUnit;   /**< GL_ACTIVE_TEXTURE */
    struct gl_texture_unit Unit[MAX_TEXTURE_UNITS];
 
    struct gl_texture_object *ProxyTex[NUM_TEXTURE_TARGETS];
@@ -1563,6 +1445,15 @@ struct gl_texture_attrib
    /** GL_EXT_shared_texture_palette */
    GLboolean SharedPalette;
    struct gl_color_table Palette;
+
+   /** Per-unit flags */
+   /*@{*/
+   GLbitfield _EnabledUnits;  /**< one bit set for each really-enabled unit */
+   GLbitfield _EnabledCoordUnits;   /**< one bit per enabled coordinate unit */
+   GLbitfield _GenFlags;            /**< for texgen */
+   GLbitfield _TexGenEnabled;  /**< Mask of ENABLE_TEXGEN flags */
+   GLbitfield _TexMatEnabled;  /**< Mask of ENABLE_TEXMAT flags */
+   /*@}*/
 };
 
 
@@ -1598,17 +1489,6 @@ struct gl_viewport_attrib
 
 
 /**
- * Node for the attribute stack.
- */
-struct gl_attrib_node
-{
-   GLbitfield kind;
-   void *data;
-   struct gl_attrib_node *next;
-};
-
-
-/**
  * GL_ARB_vertex/pixel_buffer_object buffer object
  */
 struct gl_buffer_object
@@ -1624,7 +1504,6 @@ struct gl_buffer_object
 };
 
 
-
 /**
  * Client pixel packing/unpacking attributes
  */
@@ -1634,8 +1513,8 @@ struct gl_pixelstore_attrib
    GLint RowLength;
    GLint SkipPixels;
    GLint SkipRows;
-   GLint ImageHeight;     /**< for GL_EXT_texture3D */
-   GLint SkipImages;      /**< for GL_EXT_texture3D */
+   GLint ImageHeight;
+   GLint SkipImages;
    GLboolean SwapBytes;
    GLboolean LsbFirst;
    GLboolean ClientStorage; /**< GL_APPLE_client_storage */
@@ -1644,7 +1523,6 @@ struct gl_pixelstore_attrib
 };
 
 
-
 /**
  * Client vertex array attributes
  */
@@ -1724,7 +1602,7 @@ struct gl_array_attrib
 struct gl_feedback
 {
    GLenum Type;
-   GLbitfield _Mask;		/* FB_* bits */
+   GLbitfield _Mask;    /**< FB_* bits */
    GLfloat *Buffer;
    GLuint BufferSize;
    GLuint Count;
@@ -1948,14 +1826,14 @@ struct gl_program_state
  */
 struct gl_vertex_program_state
 {
-   GLboolean Enabled;               /**< User-set GL_VERTEX_PROGRAM_ARB/NV flag */
-   GLboolean _Enabled;              /**< Enabled and _valid_ user program? */
-   GLboolean PointSizeEnabled;      /**< GL_VERTEX_PROGRAM_POINT_SIZE_ARB/NV */
-   GLboolean TwoSideEnabled;        /**< GL_VERTEX_PROGRAM_TWO_SIDE_ARB/NV */
+   GLboolean Enabled;            /**< User-set GL_VERTEX_PROGRAM_ARB/NV flag */
+   GLboolean _Enabled;           /**< Enabled and _valid_ user program? */
+   GLboolean PointSizeEnabled;   /**< GL_VERTEX_PROGRAM_POINT_SIZE_ARB/NV */
+   GLboolean TwoSideEnabled;     /**< GL_VERTEX_PROGRAM_TWO_SIDE_ARB/NV */
    struct gl_vertex_program *Current;  /**< User-bound vertex program */
 
-   /** Currently enabled and valid vertex program (including internal programs,
-    * user-defined vertex programs and GLSL vertex shaders).
+   /** Currently enabled and valid vertex program (including internal
+    * programs, user-defined vertex programs and GLSL vertex shaders).
     * This is the program we must use when rendering.
     */
    struct gl_vertex_program *_Current;
@@ -1995,8 +1873,8 @@ struct gl_fragment_program_state
    GLboolean _Enabled;    /**< Enabled and _valid_ user program? */
    struct gl_fragment_program *Current;  /**< User-bound fragment program */
 
-   /** Currently enabled and valid fragment program (including internal programs,
-    * user-defined fragment programs and GLSL fragment shaders).
+   /** Currently enabled and valid fragment program (including internal
+    * programs, user-defined fragment programs and GLSL fragment shaders).
     * This is the program we must use when rendering.
     */
    struct gl_fragment_program *_Current;
@@ -2339,9 +2217,8 @@ struct gl_renderbuffer
 
 
 /**
- * A renderbuffer attachment point points to either a texture object
- * (and specifies a mipmap level, cube face or 3D texture slice) or
- * points to a renderbuffer.
+ * A renderbuffer attachment points to either a texture object (and specifies
+ * a mipmap level, cube face or 3D texture slice) or points to a renderbuffer.
  */
 struct gl_renderbuffer_attachment
 {
@@ -2373,14 +2250,16 @@ struct gl_renderbuffer_attachment
  */
 struct gl_framebuffer
 {
-   _glthread_Mutex Mutex;		   /**< for thread safety */
-   GLuint Name;      /* if zero, this is a window system framebuffer */
+   _glthread_Mutex Mutex;  /**< for thread safety */
+   GLuint Name;            /**< if zero, this is a window system framebuffer */
    GLint RefCount;
    GLboolean DeletePending;
 
-   GLvisual Visual;	/**< The framebuffer's visual.
-                             Immutable if this is a window system buffer.
-                             Computed from attachments if user-made FBO. */
+   /**
+    * The framebuffer's visual. Immutable if this is a window system buffer.
+    * Computed from attachments if user-made FBO.
+    */
+   GLvisual Visual;
 
    GLboolean Initialized;
 
@@ -2399,9 +2278,10 @@ struct gl_framebuffer
    GLfloat _MRD;	/**< minimum resolvable difference in Z values */
    /*@}*/
 
-   GLenum _Status; /* One of the GL_FRAMEBUFFER_(IN)COMPLETE_* tokens */
+   /** One of the GL_FRAMEBUFFER_(IN)COMPLETE_* tokens */
+   GLenum _Status;
 
-   /* Array of all renderbuffer attachments, indexed by BUFFER_* tokens. */
+   /** Array of all renderbuffer attachments, indexed by BUFFER_* tokens. */
    struct gl_renderbuffer_attachment Attachment[BUFFER_COUNT];
 
    /* In unextended OpenGL these vars are part of the GL_COLOR_BUFFER
@@ -2463,52 +2343,58 @@ struct gl_program_constants
  */
 struct gl_constants
 {
-   GLint MaxTextureLevels;		/**< Maximum number of allowed mipmap levels. */ 
-   GLint Max3DTextureLevels;		/**< Maximum number of allowed mipmap levels for 3D texture targets. */
-   GLint MaxCubeTextureLevels;          /**< Maximum number of allowed mipmap levels for GL_ARB_texture_cube_map */
-   GLint MaxArrayTextureLayers;         /**< Maximum number of layers in an array texture. */
-   GLint MaxTextureRectSize;            /* GL_NV_texture_rectangle */
+   GLint MaxTextureLevels;      /**< Max mipmap levels. */ 
+   GLint Max3DTextureLevels;    /**< Max mipmap levels for 3D textures */
+   GLint MaxCubeTextureLevels;  /**< Max mipmap levels for cube textures */
+   GLint MaxArrayTextureLayers; /**< Max layers in array textures */
+   GLint MaxTextureRectSize;    /**< Max rectangle texture size, in pixes */
    GLuint MaxTextureCoordUnits;
    GLuint MaxTextureImageUnits;
-   GLuint MaxTextureUnits;              /**< = MIN(CoordUnits, ImageUnits) */
-   GLfloat MaxTextureMaxAnisotropy;     /**< GL_EXT_texture_filter_anisotropic */
-   GLfloat MaxTextureLodBias;           /**< GL_EXT_texture_lod_bias */
+   GLuint MaxVertexTextureImageUnits;
+   GLuint MaxTextureUnits;           /**< = MIN(CoordUnits, ImageUnits) */
+   GLfloat MaxTextureMaxAnisotropy;  /**< GL_EXT_texture_filter_anisotropic */
+   GLfloat MaxTextureLodBias;        /**< GL_EXT_texture_lod_bias */
+
    GLuint MaxArrayLockSize;
+
    GLint SubPixelBits;
-   GLfloat MinPointSize, MaxPointSize;		/* aliased */
-   GLfloat MinPointSizeAA, MaxPointSizeAA;	/* antialiased */
+
+   GLfloat MinPointSize, MaxPointSize;	     /**< aliased */
+   GLfloat MinPointSizeAA, MaxPointSizeAA;   /**< antialiased */
    GLfloat PointSizeGranularity;
-   GLfloat MinLineWidth, MaxLineWidth;		/* aliased */
-   GLfloat MinLineWidthAA, MaxLineWidthAA;	/* antialiased */
+   GLfloat MinLineWidth, MaxLineWidth;       /**< aliased */
+   GLfloat MinLineWidthAA, MaxLineWidthAA;   /**< antialiased */
    GLfloat LineWidthGranularity;
+
    GLuint MaxColorTableSize;
    GLuint MaxConvolutionWidth;
    GLuint MaxConvolutionHeight;
+
    GLuint MaxClipPlanes;
    GLuint MaxLights;
-   GLfloat MaxShininess;			/* GL_NV_light_max_exponent */
-   GLfloat MaxSpotExponent;			/* GL_NV_light_max_exponent */
+   GLfloat MaxShininess;                     /**< GL_NV_light_max_exponent */
+   GLfloat MaxSpotExponent;                  /**< GL_NV_light_max_exponent */
+
    GLuint MaxViewportWidth, MaxViewportHeight;
-   struct gl_program_constants VertexProgram;    /* GL_ARB_vertex_program */
-   struct gl_program_constants FragmentProgram;  /* GL_ARB_fragment_program */
-   /* shared by vertex and fragment program: */
+
+   struct gl_program_constants VertexProgram;   /**< GL_ARB_vertex_program */
+   struct gl_program_constants FragmentProgram; /**< GL_ARB_fragment_program */
    GLuint MaxProgramMatrices;
    GLuint MaxProgramMatrixStackDepth;
-   /* vertex array / buffer object bounds checking */
+
+   /** vertex array / buffer object bounds checking */
    GLboolean CheckArrayBounds;
-   /* GL_ARB_draw_buffers */
-   GLuint MaxDrawBuffers;
-   /* GL_OES_read_format */
-   GLenum ColorReadFormat;
-   GLenum ColorReadType;
-   /* GL_EXT_framebuffer_object */
-   GLuint MaxColorAttachments;
-   GLuint MaxRenderbufferSize;
-   /* GL_ARB_vertex_shader */
-   GLuint MaxVertexTextureImageUnits;
-   GLuint MaxVarying;  /**< Number of float[4] vectors */
-   /* GL_ARB_framebuffer_object */
-   GLuint MaxSamples;
+
+   GLuint MaxDrawBuffers;    /**< GL_ARB_draw_buffers */
+
+   GLenum ColorReadFormat;   /**< GL_OES_read_format */
+   GLenum ColorReadType;     /**< GL_OES_read_format */
+
+   GLuint MaxColorAttachments;   /**< GL_EXT_framebuffer_object */
+   GLuint MaxRenderbufferSize;   /**< GL_EXT_framebuffer_object */
+   GLuint MaxSamples;            /**< GL_ARB_framebuffer_object */
+
+   GLuint MaxVarying;  /**< Number of float[4] varying parameters */
 };
 
 
@@ -2518,12 +2404,6 @@ struct gl_constants
  */
 struct gl_extensions
 {
-   /**
-    * \name Flags to quickly test if certain extensions are available.
-    * 
-    * Not every extension needs to have such a flag, but it's encouraged.
-    */
-   /*@{*/
    GLboolean dummy;  /* don't remove this! */
    GLboolean ARB_depth_texture;
    GLboolean ARB_draw_buffers;
@@ -2641,8 +2521,7 @@ struct gl_extensions
    GLboolean SGIS_texture_lod;
    GLboolean TDFX_texture_compression_FXT1;
    GLboolean S3_s3tc;
-   /*@}*/
-   /* The extension string */
+   /** The extension string */
    const GLubyte *String;
 };
 
@@ -2662,7 +2541,6 @@ struct gl_matrix_stack
 
 /**
  * \name Bits for image transfer operations 
- *
  * \sa __GLcontextRec::ImageTransferState.
  */
 /*@{*/
@@ -2677,7 +2555,7 @@ struct gl_matrix_stack
 #define IMAGE_POST_COLOR_MATRIX_COLOR_TABLE_BIT   0x100
 #define IMAGE_HISTOGRAM_BIT                       0x200
 #define IMAGE_MIN_MAX_BIT                         0x400
-#define IMAGE_CLAMP_BIT                           0x800 /* extra */
+#define IMAGE_CLAMP_BIT                           0x800
 
 
 /** Pixel Transfer ops up to convolution */
@@ -3040,21 +2918,15 @@ struct __GLcontextRec
 
    struct gl_color_table ColorTable[COLORTABLE_MAX];
    struct gl_color_table ProxyColorTable[COLORTABLE_MAX];
-#if 0
-   struct gl_color_table PostConvolutionColorTable;
-   struct gl_color_table ProxyPostConvolutionColorTable;
-   struct gl_color_table PostColorMatrixColorTable;
-   struct gl_color_table ProxyPostColorMatrixColorTable;
-#endif
-
-   struct gl_program_state Program;        /**< for vertex or fragment progs */
-   struct gl_vertex_program_state VertexProgram;   /**< GL_ARB/NV_vertex_program */
-   struct gl_fragment_program_state FragmentProgram;  /**< GL_ARB/NV_vertex_program */
-   struct gl_ati_fragment_shader_state ATIFragmentShader;  /**< GL_ATI_fragment_shader */
 
-   struct gl_query_state Query;  /**< GL_ARB_occlusion_query */
+   struct gl_program_state Program;  /**< general program state */
+   struct gl_vertex_program_state VertexProgram;
+   struct gl_fragment_program_state FragmentProgram;
+   struct gl_ati_fragment_shader_state ATIFragmentShader;
 
    struct gl_shader_state Shader; /**< GLSL shader object state */
+
+   struct gl_query_state Query;  /**< occlusion, timer queries */
    /*@}*/
 
 #if FEATURE_EXT_framebuffer_object
@@ -3078,9 +2950,9 @@ struct __GLcontextRec
    GLfloat _ModelViewInvScale;
    GLboolean _NeedEyeCoords;
    GLboolean _ForceEyeCoords; 
-   GLenum _CurrentProgram;    /* currently executing program */
+   GLenum _CurrentProgram;    /**< currently executing program */
 
-   GLuint TextureStateTimestamp; /* detect changes to shared state */
+   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 */
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index d2a9e35dd5..51c13a563d 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -1098,7 +1098,7 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit)
        rgb_shift)
       dest = get_temp( p );
    else
-      dest = make_ureg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
+      dest = make_ureg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
 
    /* Emit the RGB and A combine ops
     */
@@ -1278,7 +1278,7 @@ create_new_program(GLcontext *ctx, struct state_key *key,
    p.program->Base.Parameters = _mesa_new_parameter_list();
 
    p.program->Base.InputsRead = 0;
-   p.program->Base.OutputsWritten = 1 << FRAG_RESULT_COLR;
+   p.program->Base.OutputsWritten = 1 << FRAG_RESULT_COLOR;
 
    for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
       p.src_texture[unit] = undef;
@@ -1313,7 +1313,7 @@ create_new_program(GLcontext *ctx, struct state_key *key,
    }
 
    cf = get_source( &p, SRC_PREVIOUS, 0 );
-   out = make_ureg( PROGRAM_OUTPUT, FRAG_RESULT_COLR );
+   out = make_ureg( PROGRAM_OUTPUT, FRAG_RESULT_COLOR );
 
    if (key->separate_specular) {
       /* Emit specular add.
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index 3e6b09baa1..e25c9e732c 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -550,7 +550,6 @@ update_texture_state( GLcontext *ctx )
 
       texUnit->_Current = NULL;
       texUnit->_ReallyEnabled = 0x0;
-      texUnit->_GenFlags = 0x0;
 
       /* Get the bitmask of texture target enables.
        * enableBits will be a mask of the TEXTURE_*_BIT flags indicating
@@ -609,6 +608,8 @@ update_texture_state( GLcontext *ctx )
    for (unit = 0; unit < ctx->Const.MaxTextureCoordUnits; unit++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 
+      texUnit->_GenFlags = 0x0;
+
       if (!(ctx->Texture._EnabledCoordUnits & (1 << unit)))
 	 continue;
 
diff --git a/src/mesa/main/version.h b/src/mesa/main/version.h
index 9287e8515c..f5bf6e2c85 100644
--- a/src/mesa/main/version.h
+++ b/src/mesa/main/version.h
@@ -1,10 +1,6 @@
 /*
  * Mesa 3-D graphics library
-<<<<<<< HEAD:src/mesa/main/version.h
  * Version:  7.5
-=======
- * Version:  7.4
->>>>>>> origin/gallium-0.2:src/mesa/main/version.h
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
diff --git a/src/mesa/math/m_debug_util.h b/src/mesa/math/m_debug_util.h
index 7abe6f2565..2e67db8e55 100644
--- a/src/mesa/math/m_debug_util.h
+++ b/src/mesa/math/m_debug_util.h
@@ -231,8 +231,8 @@ extern char *mesa_profile;
 #define  BEGIN_RACE(x)                                                        \
 x = LONG_MAX;                                                                 \
 for (cycle_i = 0; cycle_i <10; cycle_i++) {                                   \
-   register long cycle_tmp1 asm("l0");					      \
-   register long cycle_tmp2 asm("l1");					      \
+   register long cycle_tmp1 __asm__("l0");				      \
+   register long cycle_tmp2 __asm__("l1");				      \
    /* rd %tick, %l0 */							      \
    __asm__ __volatile__ (".word 0xa1410000" : "=r" (cycle_tmp1));  /*  save timestamp   */
 
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index 62edb7f595..ccc0318a53 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -1688,7 +1688,7 @@ parse_result_binding(GLcontext *ctx, const GLubyte **inst,
              */
             parse_output_color_num(ctx, inst, Program, &out_color);
             ASSERT(out_color < MAX_DRAW_BUFFERS);
-            *outputReg = FRAG_RESULT_COLR;
+            *outputReg = FRAG_RESULT_COLOR;
          }
          else {
             /* for vtx programs, this is VERTEX_RESULT_POSITION */
@@ -1699,7 +1699,7 @@ parse_result_binding(GLcontext *ctx, const GLubyte **inst,
       case FRAGMENT_RESULT_DEPTH:
          if (Program->Base.Target == GL_FRAGMENT_PROGRAM_ARB) {
             /* for frag programs, this is FRAGMENT_RESULT_DEPTH */
-            *outputReg = FRAG_RESULT_DEPR;
+            *outputReg = FRAG_RESULT_DEPTH;
          }
          else {
             /* for vtx programs, this is VERTEX_RESULT_COLOR */
diff --git a/src/mesa/shader/nvfragparse.c b/src/mesa/shader/nvfragparse.c
index 20e4781372..37418ffb6e 100644
--- a/src/mesa/shader/nvfragparse.c
+++ b/src/mesa/shader/nvfragparse.c
@@ -384,18 +384,12 @@ static const char *InputRegisters[MAX_NV_FRAGMENT_PROGRAM_INPUTS + 1] = {
    "TEX0", "TEX1", "TEX2", "TEX3", "TEX4", "TEX5", "TEX6", "TEX7", NULL
 };
 
+
 static const char *OutputRegisters[MAX_NV_FRAGMENT_PROGRAM_OUTPUTS + 1] = {
-   "COLR", "COLH",
-   /* These are only allows for register combiners */
-   /*
-   "TEX0", "TEX1", "TEX2", "TEX3",
-   */
-   "DEPR", NULL
+   "DEPR", "COLR", "DATA0", NULL
 };
 
 
-
-
 /**********************************************************************/
 
 /**
@@ -828,7 +822,6 @@ static GLboolean
 Parse_OutputReg(struct parse_state *parseState, GLint *outputRegNum)
 {
    GLubyte token[100];
-   GLint j;
 
    /* Match "o[" */
    if (!Parse_String(parseState, "o["))
@@ -839,19 +832,19 @@ Parse_OutputReg(struct parse_state *parseState, GLint *outputRegNum)
       RETURN_ERROR;
 
    /* try to match an output register name */
-   for (j = 0; OutputRegisters[j]; j++) {
-      if (_mesa_strcmp((const char *) token, OutputRegisters[j]) == 0) {
-         static GLuint bothColors = (1 << FRAG_RESULT_COLR) | (1 << FRAG_RESULT_COLH);
-         *outputRegNum = j;
-         parseState->outputsWritten |= (1 << j);
-         if ((parseState->outputsWritten & bothColors) == bothColors) {
-            RETURN_ERROR1("Illegal to write to both o[COLR] and o[COLH]");
-         }
-         break;
-      }
+   if (_mesa_strcmp((char *) token, "COLR") == 0 ||
+       _mesa_strcmp((char *) token, "COLH") == 0) {
+      /* note that we don't distinguish between COLR and COLH */
+      *outputRegNum = FRAG_RESULT_COLOR;
+      parseState->outputsWritten |= (1 << FRAG_RESULT_COLOR);
    }
-   if (!OutputRegisters[j])
+   else if (_mesa_strcmp((char *) token, "DEPR") == 0) {
+      *outputRegNum = FRAG_RESULT_DEPTH;
+      parseState->outputsWritten |= (1 << FRAG_RESULT_DEPTH);
+   }
+   else {
       RETURN_ERROR1("Invalid output register name");
+   }
 
    /* Match ']' */
    if (!Parse_String(parseState, "]"))
@@ -1826,10 +1819,3 @@ _mesa_nv_fragment_input_register_name(GLuint i)
    return InputRegisters[i];
 }
 
-
-const char *
-_mesa_nv_fragment_output_register_name(GLuint i)
-{
-   ASSERT(i < MAX_NV_FRAGMENT_PROGRAM_OUTPUTS);
-   return OutputRegisters[i];
-}
diff --git a/src/mesa/shader/nvfragparse.h b/src/mesa/shader/nvfragparse.h
index de45cf543d..ac97921080 100644
--- a/src/mesa/shader/nvfragparse.h
+++ b/src/mesa/shader/nvfragparse.h
@@ -44,9 +44,4 @@ _mesa_print_nv_fragment_program(const struct gl_fragment_program *program);
 extern const char *
 _mesa_nv_fragment_input_register_name(GLuint i);
 
-
-extern const char *
-_mesa_nv_fragment_output_register_name(GLuint i);
-
-
 #endif
diff --git a/src/mesa/shader/prog_debug.c b/src/mesa/shader/prog_debug.c
index 7bcb2ef734..35ce37d5ce 100644
--- a/src/mesa/shader/prog_debug.c
+++ b/src/mesa/shader/prog_debug.c
@@ -222,20 +222,16 @@ _mesa_GetProgramRegisterfvMESA(GLenum target,
                         "glGetProgramRegisterfvMESA(registerName)");
             return;
          }
-         else if (_mesa_strcmp(reg, "o[COLR]") == 0) {
+         else if (_mesa_strcmp(reg, "o[COLR]") == 0 ||
+                  _mesa_strcmp(reg, "o[COLH]") == 0) {
             /* Fragment output color */
             ctx->Driver.GetProgramRegister(ctx, PROGRAM_OUTPUT,
-                                           FRAG_RESULT_COLR, v);
-         }
-         else if (_mesa_strcmp(reg, "o[COLH]") == 0) {
-            /* Fragment output color */
-            ctx->Driver.GetProgramRegister(ctx, PROGRAM_OUTPUT,
-                                           FRAG_RESULT_COLH, v);
+                                           FRAG_RESULT_COLOR, v);
          }
          else if (_mesa_strcmp(reg, "o[DEPR]") == 0) {
             /* Fragment output depth */
             ctx->Driver.GetProgramRegister(ctx, PROGRAM_OUTPUT,
-                                           FRAG_RESULT_DEPR, v);
+                                           FRAG_RESULT_DEPTH, v);
          }
          else {
             /* try user-defined identifiers */
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index e3bb7ac01d..3808644550 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -410,20 +410,15 @@ struct prog_instruction
     */
    GLint BranchTarget;
 
-#if 01 /* XXX just use this for i965 driver for now! */
-   /**
-    * For TEX instructions in shaders, the sampler to use for the
-    * texture lookup.
-    */
-   GLint Sampler;
-#endif
-
    /** for debugging purposes */
    const char *Comment;
 
    /** Arbitrary data.  Used for OPCODE_PRINT and some drivers */
    void *Data;
 
+   /** for driver use (try to remove someday) */
+   GLint Aux;
+
    /* XXX obsolete - remove someday */
 #if FEATURE_MESA_program_debug
    GLshort StringPos;
diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c
index 80be51c3c5..2747480834 100644
--- a/src/mesa/shader/prog_print.c
+++ b/src/mesa/shader/prog_print.c
@@ -86,6 +86,9 @@ file_string(enum register_file f, gl_prog_print_mode mode)
 static const char *
 arb_input_attrib_string(GLint index, GLenum progType)
 {
+   /*
+    * These strings should match the VERT_ATTRIB_x and FRAG_ATTRIB_x tokens.
+    */
    const char *vertAttribs[] = {
       "vertex.position",
       "vertex.weight",
@@ -160,6 +163,9 @@ arb_input_attrib_string(GLint index, GLenum progType)
 static const char *
 arb_output_attrib_string(GLint index, GLenum progType)
 {
+   /*
+    * These strings should match the VERT_RESULT_x and FRAG_RESULT_x tokens.
+    */
    const char *vertResults[] = {
       "result.position",
       "result.color.primary",
@@ -184,7 +190,12 @@ arb_output_attrib_string(GLint index, GLenum progType)
    };
    const char *fragResults[] = {
       "result.color",
-      "result.depth"
+      "result.color(half)",
+      "result.depth",
+      "result.color[0]",
+      "result.color[1]",
+      "result.color[2]",
+      "result.color[3]"
    };
 
    if (progType == GL_VERTEX_PROGRAM_ARB) {
diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c
index 00655f0288..2e5632710e 100644
--- a/src/mesa/shader/program.c
+++ b/src/mesa/shader/program.c
@@ -738,7 +738,7 @@ _mesa_combine_programs(GLcontext *ctx,
       /* Connect color outputs of fprogA to color inputs of fprogB, via a
        * new temporary register.
        */
-      if ((progA->OutputsWritten & (1 << FRAG_RESULT_COLR)) &&
+      if ((progA->OutputsWritten & (1 << FRAG_RESULT_COLOR)) &&
           (progB_inputsRead & FRAG_BIT_COL0)) {
          GLint tempReg = _mesa_find_free_register(newProg, PROGRAM_TEMPORARY);
          if (tempReg < 0) {
@@ -748,7 +748,7 @@ _mesa_combine_programs(GLcontext *ctx,
          }
          /* replace writes to result.color[0] with tempReg */
          replace_registers(newInst, lenA,
-                           PROGRAM_OUTPUT, FRAG_RESULT_COLR,
+                           PROGRAM_OUTPUT, FRAG_RESULT_COLOR,
                            PROGRAM_TEMPORARY, tempReg);
          /* replace reads from the input color with tempReg */
          replace_registers(newInst + lenA, lenB,
@@ -758,7 +758,7 @@ _mesa_combine_programs(GLcontext *ctx,
 
       /* compute combined program's InputsRead */
       inputsB = progB_inputsRead;
-      if (progA->OutputsWritten & (1 << FRAG_RESULT_COLR)) {
+      if (progA->OutputsWritten & (1 << FRAG_RESULT_COLOR)) {
          inputsB &= ~(1 << FRAG_ATTRIB_COL0);
       }
       newProg->InputsRead = progA->InputsRead | inputsB;
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index 56f1eb832e..05d9bdf7ec 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -171,7 +171,7 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
       if (inst->Opcode == OPCODE_END)
          break;
       if (inst->DstReg.File == PROGRAM_OUTPUT &&
-          inst->DstReg.Index == FRAG_RESULT_COLR) {
+          inst->DstReg.Index == FRAG_RESULT_COLOR) {
          /* change the instruction to write to colorTemp w/ clamping */
          inst->DstReg.File = PROGRAM_TEMPORARY;
          inst->DstReg.Index = colorTemp;
@@ -249,7 +249,7 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
    /* LRP result.color.xyz, fogFactorTemp.xxxx, colorTemp, fogColorRef; */
    inst->Opcode = OPCODE_LRP;
    inst->DstReg.File = PROGRAM_OUTPUT;
-   inst->DstReg.Index = FRAG_RESULT_COLR;
+   inst->DstReg.Index = FRAG_RESULT_COLOR;
    inst->DstReg.WriteMask = WRITEMASK_XYZ;
    inst->SrcReg[0].File = PROGRAM_TEMPORARY;
    inst->SrcReg[0].Index = fogFactorTemp;
@@ -264,7 +264,7 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
    /* MOV result.color.w, colorTemp.x;  # copy alpha */
    inst->Opcode = OPCODE_MOV;
    inst->DstReg.File = PROGRAM_OUTPUT;
-   inst->DstReg.Index = FRAG_RESULT_COLR;
+   inst->DstReg.Index = FRAG_RESULT_COLOR;
    inst->DstReg.WriteMask = WRITEMASK_W;
    inst->SrcReg[0].File = PROGRAM_TEMPORARY;
    inst->SrcReg[0].Index = colorTemp;
diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c
index c0f4c79e13..9858a0f7fd 100644
--- a/src/mesa/shader/slang/slang_builtin.c
+++ b/src/mesa/shader/slang/slang_builtin.c
@@ -494,7 +494,7 @@ emit_statevars(const char *name, int array_len,
    }
    else if (type->type == SLANG_SPEC_STRUCT) {
       const slang_variable_scope *fields = type->_struct->fields;
-      GLuint i, pos;
+      GLuint i, pos = 0;
       for (i = 0; i < fields->num_variables; i++) {
          const slang_variable *var = fields->variables[i];
          GLint p = emit_statevars(var->a_name, 0, &var->type.specifier,
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index 5e2ce0ce3a..ee24e2e138 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -450,8 +450,8 @@ _slang_output_index(const char *name, GLenum target)
       { NULL, 0 }
    };
    static const struct output_info fragOutputs[] = {
-      { "gl_FragColor", FRAG_RESULT_COLR },
-      { "gl_FragDepth", FRAG_RESULT_DEPR },
+      { "gl_FragColor", FRAG_RESULT_COLOR },
+      { "gl_FragDepth", FRAG_RESULT_DEPTH },
       { "gl_FragData", FRAG_RESULT_DATA0 },
       { NULL, 0 }
    };
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index bcabe71bfa..f98434892b 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -686,7 +686,7 @@ _slang_link(GLcontext *ctx,
    /* check that gl_FragColor and gl_FragData are not both written to */
    if (shProg->FragmentProgram) {
       GLbitfield outputsWritten = shProg->FragmentProgram->Base.OutputsWritten;
-      if ((outputsWritten & ((1 << FRAG_RESULT_COLR))) &&
+      if ((outputsWritten & ((1 << FRAG_RESULT_COLOR))) &&
           (outputsWritten >= (1 << FRAG_RESULT_DATA0))) {
          link_error(shProg, "Fragment program cannot write both gl_FragColor"
                     " and gl_FragData[].\n");
diff --git a/src/mesa/sources.mak b/src/mesa/sources.mak
index a76c7b7441..357661456a 100644
--- a/src/mesa/sources.mak
+++ b/src/mesa/sources.mak
@@ -308,9 +308,7 @@ COMMON_DRIVER_SOURCES =			\
 	drivers/common/driverfuncs.c
 
 
-
-### All the core C sources
-
+# Sources for building non-Gallium drivers
 MESA_SOURCES = \
 	$(MAIN_SOURCES)		\
 	$(MATH_XFORM_SOURCES)	\
@@ -323,11 +321,7 @@ MESA_SOURCES = \
 	$(ASM_C_SOURCES)	\
 	$(SLANG_SOURCES)
 
-ALL_SOURCES = \
-	$(MESA_SOURCES)		\
-	$(GLAPI_SOURCES)	\
-	$(MESA_ASM_SOURCES)
-
+# Sources for building Gallium drivers
 MESA_GALLIUM_SOURCES = \
 	$(MAIN_SOURCES)		\
 	$(MATH_SOURCES)		\
@@ -336,6 +330,13 @@ MESA_GALLIUM_SOURCES = \
 	$(SHADER_SOURCES)	\
 	$(SLANG_SOURCES)
 
+# All the core C sources, for dependency checking
+ALL_SOURCES = \
+	$(MESA_SOURCES)		\
+	$(GLAPI_SOURCES)	\
+	$(MESA_ASM_SOURCES)	\
+	$(STATETRACKER_SOURCES)
+
 
 ### Object files
 
diff --git a/src/mesa/sparc/clip.S b/src/mesa/sparc/clip.S
index 58c228ed15..208843c606 100644
--- a/src/mesa/sparc/clip.S
+++ b/src/mesa/sparc/clip.S
@@ -25,12 +25,8 @@
 #define VEC_SIZE_3   	7
 #define VEC_SIZE_4   	15
 
-#if defined(SVR4) || defined(__SVR4) || defined(__svr4__)
-        /* Solaris requires this for 64-bit. */
         .register %g2, #scratch
         .register %g3, #scratch
-        .register %g7, #scratch
-#endif
 
 	.text
 	.align		64
@@ -78,7 +74,7 @@ _mesa_sparc_cliptest_points4:
 	add		%g1, 0x4, %g1
 
 	ld		[%i0 + V4F_STRIDE], %l1
-	ld		[%i0 + V4F_COUNT], %g7
+	ld		[%i0 + V4F_COUNT], %l3
 	LDPTR		[%i0 + V4F_START], %i0
 	LDPTR		[%i1 + V4F_START], %i5
 	ldub		[%i3], %g2
@@ -91,12 +87,12 @@ _mesa_sparc_cliptest_points4:
 	st		%g3, [%i1 + V4F_FLAGS]
 	mov		3, %g3
 	st		%g3, [%i1 + V4F_SIZE]
-	st		%g7, [%i1 + V4F_COUNT]
+	st		%l3, [%i1 + V4F_COUNT]
 	clr		%l2
 	clr		%l0
 
 	/* l0:	i
-	 * g7:	count
+	 * l3:	count
 	 * l1:	stride
 	 * l2:	c
 	 * g2:	(tmpAndMask << 8) | tmpOrMask
@@ -153,12 +149,12 @@ _mesa_sparc_cliptest_points4:
 3:	add		%i5, 0x10, %i5		! IEU1
 	add		%l0, 1, %l0		! IEU0	Group
 	add		%i2, 1, %i2		! IEU0	Group
-	cmp		%l0, %g7		! IEU1	Group
+	cmp		%l0, %l3		! IEU1	Group
 	bne		1b			! CTI
 	 add		%i0, %l1, %i0		! IEU0	Group
 	stb		%g2, [%i3]		! LSU
 	srl		%g2, 8, %g3		! IEU0	Group
-	cmp		%l2, %g7		! IEU1	Group
+	cmp		%l2, %l3		! IEU1	Group
 	bl,a		1f			! CTI
 	 clr		%g3			! IEU0
 1:	stb		%g3, [%i4]		! LSU	Group
@@ -174,25 +170,18 @@ _mesa_sparc_cliptest_points4_np:
 	add		%g1, 0x4, %g1
 
 	ld		[%i0 + V4F_STRIDE], %l1
-	ld		[%i0 + V4F_COUNT], %g7
+	ld		[%i0 + V4F_COUNT], %l3
 	LDPTR		[%i0 + V4F_START], %i0
-	LDPTR		[%i1 + V4F_START], %i5
 	ldub		[%i3], %g2
 	ldub		[%i4], %g3
 	sll		%g3, 8, %g3
 	or		%g2, %g3, %g2
 
-	ld		[%i1 + V4F_FLAGS], %g3
-	or		%g3, VEC_SIZE_4, %g3
-	st		%g3, [%i1 + V4F_FLAGS]
-	mov		3, %g3
-	st		%g3, [%i1 + V4F_SIZE]
-	st		%g7, [%i1 + V4F_COUNT]
 	clr		%l2
 	clr		%l0
 
 	/* l0:	i
-	 * g7:	count
+	 * l3:	count
 	 * l1:	stride
 	 * l2:	c
 	 * g2:	(tmpAndMask << 8) | tmpOrMask
@@ -230,12 +219,12 @@ _mesa_sparc_cliptest_points4_np:
 	and		%g2, %g4, %g2		! IEU0	Group
 2:	add		%l0, 1, %l0		! IEU0	Group
 	add		%i2, 1, %i2		! IEU0	Group
-	cmp		%l0, %g7		! IEU1	Group
+	cmp		%l0, %l3		! IEU1	Group
 	bne		1b			! CTI
 	 add		%i0, %l1, %i0		! IEU0	Group
 	stb		%g2, [%i3]		! LSU
 	srl		%g2, 8, %g3		! IEU0	Group
-	cmp		%l2, %g7		! IEU1	Group
+	cmp		%l2, %l3		! IEU1	Group
 	bl,a		1f			! CTI
 	 clr		%g3			! IEU0
 1:	stb		%g3, [%i4]		! LSU	Group
diff --git a/src/mesa/sparc/glapi_sparc.S b/src/mesa/sparc/glapi_sparc.S
index 7d215939b4..bf70fceef5 100644
--- a/src/mesa/sparc/glapi_sparc.S
+++ b/src/mesa/sparc/glapi_sparc.S
@@ -26,818 +26,176 @@
  * SOFTWARE.
  */
 
-#include "glapioffsets.h"
+#include "glapi/glapioffsets.h"
 
 #ifdef __arch64__
-#  define GL_STUB(fn,off)				\
-fn:					\
-	sethi	%hi(0xDEADBEEF), %g4 ;			\
-	sethi	%hi(0xDEADBEEF), %g1 ;			\
-	or	%g4, %lo(0xDEADBEEF), %g4 ;		\
-	or	%g1, %lo(0xDEADBEEF), %g1 ;		\
-	sllx	%g4, 32, %g4 ;				\
-	ldx	[%g1 + %g4], %g1 ;			\
-	sethi	%hi(8 * off), %g4 ;			\
-	or	%g4, %lo(8 * off), %g4 ;		\
-	ldx	[%g1 + %g4], %g5 ;			\
-	jmpl	%g5, %g0 ;				\
-	nop
+#define GL_OFF(N)	((N) * 8)
+#define GL_LL		ldx
+#define GL_TIE_LD(SYM)	%tie_ldx(SYM)
+#define GL_STACK_SIZE	128
 #else
-#  define GL_STUB(fn,off)				\
-fn:					\
-	sethi	%hi(0xDEADBEEF), %g1 ;			\
-	ld	[%g1 + %lo(0xDEADBEEF)], %g1 ;		\
-	ld	[%g1 + (4 * off)], %g5 ;		\
-	jmpl	%g5, %g0 ;				\
-	nop
+#define GL_OFF(N)	((N) * 4)
+#define GL_LL		ld
+#define GL_TIE_LD(SYM)	%tie_ld(SYM)
+#define GL_STACK_SIZE	64
 #endif
 
-#define GL_STUB_ALIAS(fn,alias) fn = alias
+#define GLOBL_FN(x) .globl x ; .type x, @function
+#define HIDDEN(x) .hidden x
 
-.text
-.align 32
-		.globl __glapi_sparc_icache_flush ; .type __glapi_sparc_icache_flush,#function
+	.register %g2, #scratch
+	.register %g3, #scratch
+
+	.text
+
+	GLOBL_FN(__glapi_sparc_icache_flush)
+	HIDDEN(__glapi_sparc_icache_flush)
+	.type	__glapi_sparc_icache_flush, @function
 __glapi_sparc_icache_flush: /* %o0 = insn_addr */
 	flush	%o0
 	retl
-	nop
+	 nop
+
+	.align	32
+
+	.type	__glapi_sparc_get_pc, @function
+__glapi_sparc_get_pc:
+	retl
+	 add	%o7, %g2, %g2
+	.size	__glapi_sparc_get_pc, .-__glapi_sparc_get_pc
+
+#ifdef GLX_USE_TLS
+
+	GLOBL_FN(__glapi_sparc_get_dispatch)
+	HIDDEN(__glapi_sparc_get_dispatch)
+__glapi_sparc_get_dispatch:
+	mov	%o7, %g1
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %g2
+	call	__glapi_sparc_get_pc
+	add	%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2
+	mov	%g1, %o7
+	sethi	%tie_hi22(_glapi_tls_Dispatch), %g1
+	add	%g1, %tie_lo10(_glapi_tls_Dispatch), %g1
+	GL_LL	[%g2 + %g1], %g2, GL_TIE_LD(_glapi_tls_Dispatch)
+	retl
+	 mov	%g2, %o0
+
+	.data
+	.align	32
+
+	/* --> sethi %hi(_glapi_tls_Dispatch), %g1 */
+	/* --> or %g1, %lo(_glapi_tls_Dispatch), %g1 */
+	GLOBL_FN(__glapi_sparc_tls_stub)
+	HIDDEN(__glapi_sparc_tls_stub)
+__glapi_sparc_tls_stub: /* Call offset in %g3 */
+	mov	%o7, %g1
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %g2
+	call	__glapi_sparc_get_pc
+	add	%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2
+	mov	%g1, %o7
+	srl	%g3, 10, %g3
+	sethi	%tie_hi22(_glapi_tls_Dispatch), %g1
+	add	%g1, %tie_lo10(_glapi_tls_Dispatch), %g1
+	GL_LL	[%g2 + %g1], %g2, GL_TIE_LD(_glapi_tls_Dispatch)
+	GL_LL	[%g7+%g2], %g1
+	GL_LL	[%g1 + %g3], %g1
+	jmp	%g1
+	 nop
+	.size	__glapi_sparc_tls_stub, .-__glapi_sparc_tls_stub
+
+#define GL_STUB(fn, off)				\
+	GLOBL_FN(fn);					\
+fn:	ba	__glapi_sparc_tls_stub;			\
+	 sethi	GL_OFF(off), %g3;			\
+	.size	fn,.-fn;
+
+#elif defined(PTHREADS)
+
+	/* 64-bit 0x00 --> sethi %hh(_glapi_Dispatch), %g1 */
+	/* 64-bit 0x04 --> sethi %lm(_glapi_Dispatch), %g2 */
+	/* 64-bit 0x08 --> or %g1, %hm(_glapi_Dispatch), %g1 */
+	/* 64-bit 0x0c --> sllx %g1, 32, %g1 */
+	/* 64-bit 0x10 --> add %g1, %g2, %g1 */
+	/* 64-bit 0x14 --> ldx [%g1 + %lo(_glapi_Dispatch)], %g1 */
 
-.data
-.align 64
+	/* 32-bit 0x00 --> sethi %hi(_glapi_Dispatch), %g1 */
+	/* 32-bit 0x04 --> ld [%g1 + %lo(_glapi_Dispatch)], %g1 */
 
-		.globl glNewList ; .type glNewList,#function
-		.globl glEndList ; .type glEndList,#function
-		.globl glCallList ; .type glCallList,#function
-		.globl glCallLists ; .type glCallLists,#function
-		.globl glDeleteLists ; .type glDeleteLists,#function
-		.globl glGenLists ; .type glGenLists,#function
-		.globl glListBase ; .type glListBase,#function
-		.globl glBegin ; .type glBegin,#function
-		.globl glBitmap ; .type glBitmap,#function
-		.globl glColor3b ; .type glColor3b,#function
-		.globl glColor3bv ; .type glColor3bv,#function
-		.globl glColor3d ; .type glColor3d,#function
-		.globl glColor3dv ; .type glColor3dv,#function
-		.globl glColor3f ; .type glColor3f,#function
-		.globl glColor3fv ; .type glColor3fv,#function
-		.globl glColor3i ; .type glColor3i,#function
-		.globl glColor3iv ; .type glColor3iv,#function
-		.globl glColor3s ; .type glColor3s,#function
-		.globl glColor3sv ; .type glColor3sv,#function
-		.globl glColor3ub ; .type glColor3ub,#function
-		.globl glColor3ubv ; .type glColor3ubv,#function
-		.globl glColor3ui ; .type glColor3ui,#function
-		.globl glColor3uiv ; .type glColor3uiv,#function
-		.globl glColor3us ; .type glColor3us,#function
-		.globl glColor3usv ; .type glColor3usv,#function
-		.globl glColor4b ; .type glColor4b,#function
-		.globl glColor4bv ; .type glColor4bv,#function
-		.globl glColor4d ; .type glColor4d,#function
-		.globl glColor4dv ; .type glColor4dv,#function
-		.globl glColor4f ; .type glColor4f,#function
-		.globl glColor4fv ; .type glColor4fv,#function
-		.globl glColor4i ; .type glColor4i,#function
-		.globl glColor4iv ; .type glColor4iv,#function
-		.globl glColor4s ; .type glColor4s,#function
-		.globl glColor4sv ; .type glColor4sv,#function
-		.globl glColor4ub ; .type glColor4ub,#function
-		.globl glColor4ubv ; .type glColor4ubv,#function
-		.globl glColor4ui ; .type glColor4ui,#function
-		.globl glColor4uiv ; .type glColor4uiv,#function
-		.globl glColor4us ; .type glColor4us,#function
-		.globl glColor4usv ; .type glColor4usv,#function
-		.globl glEdgeFlag ; .type glEdgeFlag,#function
-		.globl glEdgeFlagv ; .type glEdgeFlagv,#function
-		.globl glEnd ; .type glEnd,#function
-		.globl glIndexd ; .type glIndexd,#function
-		.globl glIndexdv ; .type glIndexdv,#function
-		.globl glIndexf ; .type glIndexf,#function
-		.globl glIndexfv ; .type glIndexfv,#function
-		.globl glIndexi ; .type glIndexi,#function
-		.globl glIndexiv ; .type glIndexiv,#function
-		.globl glIndexs ; .type glIndexs,#function
-		.globl glIndexsv ; .type glIndexsv,#function
-		.globl glNormal3b ; .type glNormal3b,#function
-		.globl glNormal3bv ; .type glNormal3bv,#function
-		.globl glNormal3d ; .type glNormal3d,#function
-		.globl glNormal3dv ; .type glNormal3dv,#function
-		.globl glNormal3f ; .type glNormal3f,#function
-		.globl glNormal3fv ; .type glNormal3fv,#function
-		.globl glNormal3i ; .type glNormal3i,#function
-		.globl glNormal3iv ; .type glNormal3iv,#function
-		.globl glNormal3s ; .type glNormal3s,#function
-		.globl glNormal3sv ; .type glNormal3sv,#function
-		.globl glRasterPos2d ; .type glRasterPos2d,#function
-		.globl glRasterPos2dv ; .type glRasterPos2dv,#function
-		.globl glRasterPos2f ; .type glRasterPos2f,#function
-		.globl glRasterPos2fv ; .type glRasterPos2fv,#function
-		.globl glRasterPos2i ; .type glRasterPos2i,#function
-		.globl glRasterPos2iv ; .type glRasterPos2iv,#function
-		.globl glRasterPos2s ; .type glRasterPos2s,#function
-		.globl glRasterPos2sv ; .type glRasterPos2sv,#function
-		.globl glRasterPos3d ; .type glRasterPos3d,#function
-		.globl glRasterPos3dv ; .type glRasterPos3dv,#function
-		.globl glRasterPos3f ; .type glRasterPos3f,#function
-		.globl glRasterPos3fv ; .type glRasterPos3fv,#function
-		.globl glRasterPos3i ; .type glRasterPos3i,#function
-		.globl glRasterPos3iv ; .type glRasterPos3iv,#function
-		.globl glRasterPos3s ; .type glRasterPos3s,#function
-		.globl glRasterPos3sv ; .type glRasterPos3sv,#function
-		.globl glRasterPos4d ; .type glRasterPos4d,#function
-		.globl glRasterPos4dv ; .type glRasterPos4dv,#function
-		.globl glRasterPos4f ; .type glRasterPos4f,#function
-		.globl glRasterPos4fv ; .type glRasterPos4fv,#function
-		.globl glRasterPos4i ; .type glRasterPos4i,#function
-		.globl glRasterPos4iv ; .type glRasterPos4iv,#function
-		.globl glRasterPos4s ; .type glRasterPos4s,#function
-		.globl glRasterPos4sv ; .type glRasterPos4sv,#function
-		.globl glRectd ; .type glRectd,#function
-		.globl glRectdv ; .type glRectdv,#function
-		.globl glRectf ; .type glRectf,#function
-		.globl glRectfv ; .type glRectfv,#function
-		.globl glRecti ; .type glRecti,#function
-		.globl glRectiv ; .type glRectiv,#function
-		.globl glRects ; .type glRects,#function
-		.globl glRectsv ; .type glRectsv,#function
-		.globl glTexCoord1d ; .type glTexCoord1d,#function
-		.globl glTexCoord1dv ; .type glTexCoord1dv,#function
-		.globl glTexCoord1f ; .type glTexCoord1f,#function
-		.globl glTexCoord1fv ; .type glTexCoord1fv,#function
-		.globl glTexCoord1i ; .type glTexCoord1i,#function
-		.globl glTexCoord1iv ; .type glTexCoord1iv,#function
-		.globl glTexCoord1s ; .type glTexCoord1s,#function
-		.globl glTexCoord1sv ; .type glTexCoord1sv,#function
-		.globl glTexCoord2d ; .type glTexCoord2d,#function
-		.globl glTexCoord2dv ; .type glTexCoord2dv,#function
-		.globl glTexCoord2f ; .type glTexCoord2f,#function
-		.globl glTexCoord2fv ; .type glTexCoord2fv,#function
-		.globl glTexCoord2i ; .type glTexCoord2i,#function
-		.globl glTexCoord2iv ; .type glTexCoord2iv,#function
-		.globl glTexCoord2s ; .type glTexCoord2s,#function
-		.globl glTexCoord2sv ; .type glTexCoord2sv,#function
-		.globl glTexCoord3d ; .type glTexCoord3d,#function
-		.globl glTexCoord3dv ; .type glTexCoord3dv,#function
-		.globl glTexCoord3f ; .type glTexCoord3f,#function
-		.globl glTexCoord3fv ; .type glTexCoord3fv,#function
-		.globl glTexCoord3i ; .type glTexCoord3i,#function
-		.globl glTexCoord3iv ; .type glTexCoord3iv,#function
-		.globl glTexCoord3s ; .type glTexCoord3s,#function
-		.globl glTexCoord3sv ; .type glTexCoord3sv,#function
-		.globl glTexCoord4d ; .type glTexCoord4d,#function
-		.globl glTexCoord4dv ; .type glTexCoord4dv,#function
-		.globl glTexCoord4f ; .type glTexCoord4f,#function
-		.globl glTexCoord4fv ; .type glTexCoord4fv,#function
-		.globl glTexCoord4i ; .type glTexCoord4i,#function
-		.globl glTexCoord4iv ; .type glTexCoord4iv,#function
-		.globl glTexCoord4s ; .type glTexCoord4s,#function
-		.globl glTexCoord4sv ; .type glTexCoord4sv,#function
-		.globl glVertex2d ; .type glVertex2d,#function
-		.globl glVertex2dv ; .type glVertex2dv,#function
-		.globl glVertex2f ; .type glVertex2f,#function
-		.globl glVertex2fv ; .type glVertex2fv,#function
-		.globl glVertex2i ; .type glVertex2i,#function
-		.globl glVertex2iv ; .type glVertex2iv,#function
-		.globl glVertex2s ; .type glVertex2s,#function
-		.globl glVertex2sv ; .type glVertex2sv,#function
-		.globl glVertex3d ; .type glVertex3d,#function
-		.globl glVertex3dv ; .type glVertex3dv,#function
-		.globl glVertex3f ; .type glVertex3f,#function
-		.globl glVertex3fv ; .type glVertex3fv,#function
-		.globl glVertex3i ; .type glVertex3i,#function
-		.globl glVertex3iv ; .type glVertex3iv,#function
-		.globl glVertex3s ; .type glVertex3s,#function
-		.globl glVertex3sv ; .type glVertex3sv,#function
-		.globl glVertex4d ; .type glVertex4d,#function
-		.globl glVertex4dv ; .type glVertex4dv,#function
-		.globl glVertex4f ; .type glVertex4f,#function
-		.globl glVertex4fv ; .type glVertex4fv,#function
-		.globl glVertex4i ; .type glVertex4i,#function
-		.globl glVertex4iv ; .type glVertex4iv,#function
-		.globl glVertex4s ; .type glVertex4s,#function
-		.globl glVertex4sv ; .type glVertex4sv,#function
-		.globl glClipPlane ; .type glClipPlane,#function
-		.globl glColorMaterial ; .type glColorMaterial,#function
-		.globl glCullFace ; .type glCullFace,#function
-		.globl glFogf ; .type glFogf,#function
-		.globl glFogfv ; .type glFogfv,#function
-		.globl glFogi ; .type glFogi,#function
-		.globl glFogiv ; .type glFogiv,#function
-		.globl glFrontFace ; .type glFrontFace,#function
-		.globl glHint ; .type glHint,#function
-		.globl glLightf ; .type glLightf,#function
-		.globl glLightfv ; .type glLightfv,#function
-		.globl glLighti ; .type glLighti,#function
-		.globl glLightiv ; .type glLightiv,#function
-		.globl glLightModelf ; .type glLightModelf,#function
-		.globl glLightModelfv ; .type glLightModelfv,#function
-		.globl glLightModeli ; .type glLightModeli,#function
-		.globl glLightModeliv ; .type glLightModeliv,#function
-		.globl glLineStipple ; .type glLineStipple,#function
-		.globl glLineWidth ; .type glLineWidth,#function
-		.globl glMaterialf ; .type glMaterialf,#function
-		.globl glMaterialfv ; .type glMaterialfv,#function
-		.globl glMateriali ; .type glMateriali,#function
-		.globl glMaterialiv ; .type glMaterialiv,#function
-		.globl glPointSize ; .type glPointSize,#function
-		.globl glPolygonMode ; .type glPolygonMode,#function
-		.globl glPolygonStipple ; .type glPolygonStipple,#function
-		.globl glScissor ; .type glScissor,#function
-		.globl glShadeModel ; .type glShadeModel,#function
-		.globl glTexParameterf ; .type glTexParameterf,#function
-		.globl glTexParameterfv ; .type glTexParameterfv,#function
-		.globl glTexParameteri ; .type glTexParameteri,#function
-		.globl glTexParameteriv ; .type glTexParameteriv,#function
-		.globl glTexImage1D ; .type glTexImage1D,#function
-		.globl glTexImage2D ; .type glTexImage2D,#function
-		.globl glTexEnvf ; .type glTexEnvf,#function
-		.globl glTexEnvfv ; .type glTexEnvfv,#function
-		.globl glTexEnvi ; .type glTexEnvi,#function
-		.globl glTexEnviv ; .type glTexEnviv,#function
-		.globl glTexGend ; .type glTexGend,#function
-		.globl glTexGendv ; .type glTexGendv,#function
-		.globl glTexGenf ; .type glTexGenf,#function
-		.globl glTexGenfv ; .type glTexGenfv,#function
-		.globl glTexGeni ; .type glTexGeni,#function
-		.globl glTexGeniv ; .type glTexGeniv,#function
-		.globl glFeedbackBuffer ; .type glFeedbackBuffer,#function
-		.globl glSelectBuffer ; .type glSelectBuffer,#function
-		.globl glRenderMode ; .type glRenderMode,#function
-		.globl glInitNames ; .type glInitNames,#function
-		.globl glLoadName ; .type glLoadName,#function
-		.globl glPassThrough ; .type glPassThrough,#function
-		.globl glPopName ; .type glPopName,#function
-		.globl glPushName ; .type glPushName,#function
-		.globl glDrawBuffer ; .type glDrawBuffer,#function
-		.globl glClear ; .type glClear,#function
-		.globl glClearAccum ; .type glClearAccum,#function
-		.globl glClearIndex ; .type glClearIndex,#function
-		.globl glClearColor ; .type glClearColor,#function
-		.globl glClearStencil ; .type glClearStencil,#function
-		.globl glClearDepth ; .type glClearDepth,#function
-		.globl glStencilMask ; .type glStencilMask,#function
-		.globl glColorMask ; .type glColorMask,#function
-		.globl glDepthMask ; .type glDepthMask,#function
-		.globl glIndexMask ; .type glIndexMask,#function
-		.globl glAccum ; .type glAccum,#function
-		.globl glDisable ; .type glDisable,#function
-		.globl glEnable ; .type glEnable,#function
-		.globl glFinish ; .type glFinish,#function
-		.globl glFlush ; .type glFlush,#function
-		.globl glPopAttrib ; .type glPopAttrib,#function
-		.globl glPushAttrib ; .type glPushAttrib,#function
-		.globl glMap1d ; .type glMap1d,#function
-		.globl glMap1f ; .type glMap1f,#function
-		.globl glMap2d ; .type glMap2d,#function
-		.globl glMap2f ; .type glMap2f,#function
-		.globl glMapGrid1d ; .type glMapGrid1d,#function
-		.globl glMapGrid1f ; .type glMapGrid1f,#function
-		.globl glMapGrid2d ; .type glMapGrid2d,#function
-		.globl glMapGrid2f ; .type glMapGrid2f,#function
-		.globl glEvalCoord1d ; .type glEvalCoord1d,#function
-		.globl glEvalCoord1dv ; .type glEvalCoord1dv,#function
-		.globl glEvalCoord1f ; .type glEvalCoord1f,#function
-		.globl glEvalCoord1fv ; .type glEvalCoord1fv,#function
-		.globl glEvalCoord2d ; .type glEvalCoord2d,#function
-		.globl glEvalCoord2dv ; .type glEvalCoord2dv,#function
-		.globl glEvalCoord2f ; .type glEvalCoord2f,#function
-		.globl glEvalCoord2fv ; .type glEvalCoord2fv,#function
-		.globl glEvalMesh1 ; .type glEvalMesh1,#function
-		.globl glEvalPoint1 ; .type glEvalPoint1,#function
-		.globl glEvalMesh2 ; .type glEvalMesh2,#function
-		.globl glEvalPoint2 ; .type glEvalPoint2,#function
-		.globl glAlphaFunc ; .type glAlphaFunc,#function
-		.globl glBlendFunc ; .type glBlendFunc,#function
-		.globl glLogicOp ; .type glLogicOp,#function
-		.globl glStencilFunc ; .type glStencilFunc,#function
-		.globl glStencilOp ; .type glStencilOp,#function
-		.globl glDepthFunc ; .type glDepthFunc,#function
-		.globl glPixelZoom ; .type glPixelZoom,#function
-		.globl glPixelTransferf ; .type glPixelTransferf,#function
-		.globl glPixelTransferi ; .type glPixelTransferi,#function
-		.globl glPixelStoref ; .type glPixelStoref,#function
-		.globl glPixelStorei ; .type glPixelStorei,#function
-		.globl glPixelMapfv ; .type glPixelMapfv,#function
-		.globl glPixelMapuiv ; .type glPixelMapuiv,#function
-		.globl glPixelMapusv ; .type glPixelMapusv,#function
-		.globl glReadBuffer ; .type glReadBuffer,#function
-		.globl glCopyPixels ; .type glCopyPixels,#function
-		.globl glReadPixels ; .type glReadPixels,#function
-		.globl glDrawPixels ; .type glDrawPixels,#function
-		.globl glGetBooleanv ; .type glGetBooleanv,#function
-		.globl glGetClipPlane ; .type glGetClipPlane,#function
-		.globl glGetDoublev ; .type glGetDoublev,#function
-		.globl glGetError ; .type glGetError,#function
-		.globl glGetFloatv ; .type glGetFloatv,#function
-		.globl glGetIntegerv ; .type glGetIntegerv,#function
-		.globl glGetLightfv ; .type glGetLightfv,#function
-		.globl glGetLightiv ; .type glGetLightiv,#function
-		.globl glGetMapdv ; .type glGetMapdv,#function
-		.globl glGetMapfv ; .type glGetMapfv,#function
-		.globl glGetMapiv ; .type glGetMapiv,#function
-		.globl glGetMaterialfv ; .type glGetMaterialfv,#function
-		.globl glGetMaterialiv ; .type glGetMaterialiv,#function
-		.globl glGetPixelMapfv ; .type glGetPixelMapfv,#function
-		.globl glGetPixelMapuiv ; .type glGetPixelMapuiv,#function
-		.globl glGetPixelMapusv ; .type glGetPixelMapusv,#function
-		.globl glGetPolygonStipple ; .type glGetPolygonStipple,#function
-		.globl glGetString ; .type glGetString,#function
-		.globl glGetTexEnvfv ; .type glGetTexEnvfv,#function
-		.globl glGetTexEnviv ; .type glGetTexEnviv,#function
-		.globl glGetTexGendv ; .type glGetTexGendv,#function
-		.globl glGetTexGenfv ; .type glGetTexGenfv,#function
-		.globl glGetTexGeniv ; .type glGetTexGeniv,#function
-		.globl glGetTexImage ; .type glGetTexImage,#function
-		.globl glGetTexParameterfv ; .type glGetTexParameterfv,#function
-		.globl glGetTexParameteriv ; .type glGetTexParameteriv,#function
-		.globl glGetTexLevelParameterfv ; .type glGetTexLevelParameterfv,#function
-		.globl glGetTexLevelParameteriv ; .type glGetTexLevelParameteriv,#function
-		.globl glIsEnabled ; .type glIsEnabled,#function
-		.globl glIsList ; .type glIsList,#function
-		.globl glDepthRange ; .type glDepthRange,#function
-		.globl glFrustum ; .type glFrustum,#function
-		.globl glLoadIdentity ; .type glLoadIdentity,#function
-		.globl glLoadMatrixf ; .type glLoadMatrixf,#function
-		.globl glLoadMatrixd ; .type glLoadMatrixd,#function
-		.globl glMatrixMode ; .type glMatrixMode,#function
-		.globl glMultMatrixf ; .type glMultMatrixf,#function
-		.globl glMultMatrixd ; .type glMultMatrixd,#function
-		.globl glOrtho ; .type glOrtho,#function
-		.globl glPopMatrix ; .type glPopMatrix,#function
-		.globl glPushMatrix ; .type glPushMatrix,#function
-		.globl glRotated ; .type glRotated,#function
-		.globl glRotatef ; .type glRotatef,#function
-		.globl glScaled ; .type glScaled,#function
-		.globl glScalef ; .type glScalef,#function
-		.globl glTranslated ; .type glTranslated,#function
-		.globl glTranslatef ; .type glTranslatef,#function
-		.globl glViewport ; .type glViewport,#function
-		.globl glArrayElement ; .type glArrayElement,#function
-		.globl glBindTexture ; .type glBindTexture,#function
-		.globl glColorPointer ; .type glColorPointer,#function
-		.globl glDisableClientState ; .type glDisableClientState,#function
-		.globl glDrawArrays ; .type glDrawArrays,#function
-		.globl glDrawElements ; .type glDrawElements,#function
-		.globl glEdgeFlagPointer ; .type glEdgeFlagPointer,#function
-		.globl glEnableClientState ; .type glEnableClientState,#function
-		.globl glIndexPointer ; .type glIndexPointer,#function
-		.globl glIndexub ; .type glIndexub,#function
-		.globl glIndexubv ; .type glIndexubv,#function
-		.globl glInterleavedArrays ; .type glInterleavedArrays,#function
-		.globl glNormalPointer ; .type glNormalPointer,#function
-		.globl glPolygonOffset ; .type glPolygonOffset,#function
-		.globl glTexCoordPointer ; .type glTexCoordPointer,#function
-		.globl glVertexPointer ; .type glVertexPointer,#function
-		.globl glAreTexturesResident ; .type glAreTexturesResident,#function
-		.globl glCopyTexImage1D ; .type glCopyTexImage1D,#function
-		.globl glCopyTexImage2D ; .type glCopyTexImage2D,#function
-		.globl glCopyTexSubImage1D ; .type glCopyTexSubImage1D,#function
-		.globl glCopyTexSubImage2D ; .type glCopyTexSubImage2D,#function
-		.globl glDeleteTextures ; .type glDeleteTextures,#function
-		.globl glGenTextures ; .type glGenTextures,#function
-		.globl glGetPointerv ; .type glGetPointerv,#function
-		.globl glIsTexture ; .type glIsTexture,#function
-		.globl glPrioritizeTextures ; .type glPrioritizeTextures,#function
-		.globl glTexSubImage1D ; .type glTexSubImage1D,#function
-		.globl glTexSubImage2D ; .type glTexSubImage2D,#function
-		.globl glPopClientAttrib ; .type glPopClientAttrib,#function
-		.globl glPushClientAttrib ; .type glPushClientAttrib,#function
-		.globl glBlendColor ; .type glBlendColor,#function
-		.globl glBlendEquation ; .type glBlendEquation,#function
-		.globl glDrawRangeElements ; .type glDrawRangeElements,#function
-		.globl glColorTable ; .type glColorTable,#function
-		.globl glColorTableParameterfv ; .type glColorTableParameterfv,#function
-		.globl glColorTableParameteriv ; .type glColorTableParameteriv,#function
-		.globl glCopyColorTable ; .type glCopyColorTable,#function
-		.globl glGetColorTable ; .type glGetColorTable,#function
-		.globl glGetColorTableParameterfv ; .type glGetColorTableParameterfv,#function
-		.globl glGetColorTableParameteriv ; .type glGetColorTableParameteriv,#function
-		.globl glColorSubTable ; .type glColorSubTable,#function
-		.globl glCopyColorSubTable ; .type glCopyColorSubTable,#function
-		.globl glConvolutionFilter1D ; .type glConvolutionFilter1D,#function
-		.globl glConvolutionFilter2D ; .type glConvolutionFilter2D,#function
-		.globl glConvolutionParameterf ; .type glConvolutionParameterf,#function
-		.globl glConvolutionParameterfv ; .type glConvolutionParameterfv,#function
-		.globl glConvolutionParameteri ; .type glConvolutionParameteri,#function
-		.globl glConvolutionParameteriv ; .type glConvolutionParameteriv,#function
-		.globl glCopyConvolutionFilter1D ; .type glCopyConvolutionFilter1D,#function
-		.globl glCopyConvolutionFilter2D ; .type glCopyConvolutionFilter2D,#function
-		.globl glGetConvolutionFilter ; .type glGetConvolutionFilter,#function
-		.globl glGetConvolutionParameterfv ; .type glGetConvolutionParameterfv,#function
-		.globl glGetConvolutionParameteriv ; .type glGetConvolutionParameteriv,#function
-		.globl glGetSeparableFilter ; .type glGetSeparableFilter,#function
-		.globl glSeparableFilter2D ; .type glSeparableFilter2D,#function
-		.globl glGetHistogram ; .type glGetHistogram,#function
-		.globl glGetHistogramParameterfv ; .type glGetHistogramParameterfv,#function
-		.globl glGetHistogramParameteriv ; .type glGetHistogramParameteriv,#function
-		.globl glGetMinmax ; .type glGetMinmax,#function
-		.globl glGetMinmaxParameterfv ; .type glGetMinmaxParameterfv,#function
-		.globl glGetMinmaxParameteriv ; .type glGetMinmaxParameteriv,#function
-		.globl glHistogram ; .type glHistogram,#function
-		.globl glMinmax ; .type glMinmax,#function
-		.globl glResetHistogram ; .type glResetHistogram,#function
-		.globl glResetMinmax ; .type glResetMinmax,#function
-		.globl glTexImage3D ; .type glTexImage3D,#function
-		.globl glTexSubImage3D ; .type glTexSubImage3D,#function
-		.globl glCopyTexSubImage3D ; .type glCopyTexSubImage3D,#function
-		.globl glActiveTextureARB ; .type glActiveTextureARB,#function
-		.globl glClientActiveTextureARB ; .type glClientActiveTextureARB,#function
-		.globl glMultiTexCoord1dARB ; .type glMultiTexCoord1dARB,#function
-		.globl glMultiTexCoord1dvARB ; .type glMultiTexCoord1dvARB,#function
-		.globl glMultiTexCoord1fARB ; .type glMultiTexCoord1fARB,#function
-		.globl glMultiTexCoord1fvARB ; .type glMultiTexCoord1fvARB,#function
-		.globl glMultiTexCoord1iARB ; .type glMultiTexCoord1iARB,#function
-		.globl glMultiTexCoord1ivARB ; .type glMultiTexCoord1ivARB,#function
-		.globl glMultiTexCoord1sARB ; .type glMultiTexCoord1sARB,#function
-		.globl glMultiTexCoord1svARB ; .type glMultiTexCoord1svARB,#function
-		.globl glMultiTexCoord2dARB ; .type glMultiTexCoord2dARB,#function
-		.globl glMultiTexCoord2dvARB ; .type glMultiTexCoord2dvARB,#function
-		.globl glMultiTexCoord2fARB ; .type glMultiTexCoord2fARB,#function
-		.globl glMultiTexCoord2fvARB ; .type glMultiTexCoord2fvARB,#function
-		.globl glMultiTexCoord2iARB ; .type glMultiTexCoord2iARB,#function
-		.globl glMultiTexCoord2ivARB ; .type glMultiTexCoord2ivARB,#function
-		.globl glMultiTexCoord2sARB ; .type glMultiTexCoord2sARB,#function
-		.globl glMultiTexCoord2svARB ; .type glMultiTexCoord2svARB,#function
-		.globl glMultiTexCoord3dARB ; .type glMultiTexCoord3dARB,#function
-		.globl glMultiTexCoord3dvARB ; .type glMultiTexCoord3dvARB,#function
-		.globl glMultiTexCoord3fARB ; .type glMultiTexCoord3fARB,#function
-		.globl glMultiTexCoord3fvARB ; .type glMultiTexCoord3fvARB,#function
-		.globl glMultiTexCoord3iARB ; .type glMultiTexCoord3iARB,#function
-		.globl glMultiTexCoord3ivARB ; .type glMultiTexCoord3ivARB,#function
-		.globl glMultiTexCoord3sARB ; .type glMultiTexCoord3sARB,#function
-		.globl glMultiTexCoord3svARB ; .type glMultiTexCoord3svARB,#function
-		.globl glMultiTexCoord4dARB ; .type glMultiTexCoord4dARB,#function
-		.globl glMultiTexCoord4dvARB ; .type glMultiTexCoord4dvARB,#function
-		.globl glMultiTexCoord4fARB ; .type glMultiTexCoord4fARB,#function
-		.globl glMultiTexCoord4fvARB ; .type glMultiTexCoord4fvARB,#function
-		.globl glMultiTexCoord4iARB ; .type glMultiTexCoord4iARB,#function
-		.globl glMultiTexCoord4ivARB ; .type glMultiTexCoord4ivARB,#function
-		.globl glMultiTexCoord4sARB ; .type glMultiTexCoord4sARB,#function
-		.globl glMultiTexCoord4svARB ; .type glMultiTexCoord4svARB,#function
-		.globl glAttachShader ; .type glAttachShader,#function
-		.globl glCreateProgram ; .type glCreateProgram,#function
-		.globl glCreateShader ; .type glCreateShader,#function
-		.globl glDeleteProgram ; .type glDeleteProgram,#function
-		.globl glDeleteShader ; .type glDeleteShader,#function
-		.globl glDetachShader ; .type glDetachShader,#function
-		.globl glGetAttachedShaders ; .type glGetAttachedShaders,#function
-		.globl glGetProgramInfoLog ; .type glGetProgramInfoLog,#function
-		.globl glGetProgramiv ; .type glGetProgramiv,#function
-		.globl glGetShaderInfoLog ; .type glGetShaderInfoLog,#function
-		.globl glGetShaderiv ; .type glGetShaderiv,#function
-		.globl glIsProgram ; .type glIsProgram,#function
-		.globl glIsShader ; .type glIsShader,#function
-		.globl glStencilFuncSeparate ; .type glStencilFuncSeparate,#function
-		.globl glStencilMaskSeparate ; .type glStencilMaskSeparate,#function
-		.globl glStencilOpSeparate ; .type glStencilOpSeparate,#function
-		.globl glUniformMatrix2x3fv ; .type glUniformMatrix2x3fv,#function
-		.globl glUniformMatrix2x4fv ; .type glUniformMatrix2x4fv,#function
-		.globl glUniformMatrix3x2fv ; .type glUniformMatrix3x2fv,#function
-		.globl glUniformMatrix3x4fv ; .type glUniformMatrix3x4fv,#function
-		.globl glUniformMatrix4x2fv ; .type glUniformMatrix4x2fv,#function
-		.globl glUniformMatrix4x3fv ; .type glUniformMatrix4x3fv,#function
-		.globl glLoadTransposeMatrixdARB ; .type glLoadTransposeMatrixdARB,#function
-		.globl glLoadTransposeMatrixfARB ; .type glLoadTransposeMatrixfARB,#function
-		.globl glMultTransposeMatrixdARB ; .type glMultTransposeMatrixdARB,#function
-		.globl glMultTransposeMatrixfARB ; .type glMultTransposeMatrixfARB,#function
-		.globl glSampleCoverageARB ; .type glSampleCoverageARB,#function
-		.globl glCompressedTexImage1DARB ; .type glCompressedTexImage1DARB,#function
-		.globl glCompressedTexImage2DARB ; .type glCompressedTexImage2DARB,#function
-		.globl glCompressedTexImage3DARB ; .type glCompressedTexImage3DARB,#function
-		.globl glCompressedTexSubImage1DARB ; .type glCompressedTexSubImage1DARB,#function
-		.globl glCompressedTexSubImage2DARB ; .type glCompressedTexSubImage2DARB,#function
-		.globl glCompressedTexSubImage3DARB ; .type glCompressedTexSubImage3DARB,#function
-		.globl glGetCompressedTexImageARB ; .type glGetCompressedTexImageARB,#function
-		.globl glDisableVertexAttribArrayARB ; .type glDisableVertexAttribArrayARB,#function
-		.globl glEnableVertexAttribArrayARB ; .type glEnableVertexAttribArrayARB,#function
-		.globl glGetProgramEnvParameterdvARB ; .type glGetProgramEnvParameterdvARB,#function
-		.globl glGetProgramEnvParameterfvARB ; .type glGetProgramEnvParameterfvARB,#function
-		.globl glGetProgramLocalParameterdvARB ; .type glGetProgramLocalParameterdvARB,#function
-		.globl glGetProgramLocalParameterfvARB ; .type glGetProgramLocalParameterfvARB,#function
-		.globl glGetProgramStringARB ; .type glGetProgramStringARB,#function
-		.globl glGetProgramivARB ; .type glGetProgramivARB,#function
-		.globl glGetVertexAttribdvARB ; .type glGetVertexAttribdvARB,#function
-		.globl glGetVertexAttribfvARB ; .type glGetVertexAttribfvARB,#function
-		.globl glGetVertexAttribivARB ; .type glGetVertexAttribivARB,#function
-		.globl glProgramEnvParameter4dARB ; .type glProgramEnvParameter4dARB,#function
-		.globl glProgramEnvParameter4dvARB ; .type glProgramEnvParameter4dvARB,#function
-		.globl glProgramEnvParameter4fARB ; .type glProgramEnvParameter4fARB,#function
-		.globl glProgramEnvParameter4fvARB ; .type glProgramEnvParameter4fvARB,#function
-		.globl glProgramLocalParameter4dARB ; .type glProgramLocalParameter4dARB,#function
-		.globl glProgramLocalParameter4dvARB ; .type glProgramLocalParameter4dvARB,#function
-		.globl glProgramLocalParameter4fARB ; .type glProgramLocalParameter4fARB,#function
-		.globl glProgramLocalParameter4fvARB ; .type glProgramLocalParameter4fvARB,#function
-		.globl glProgramStringARB ; .type glProgramStringARB,#function
-		.globl glVertexAttrib1dARB ; .type glVertexAttrib1dARB,#function
-		.globl glVertexAttrib1dvARB ; .type glVertexAttrib1dvARB,#function
-		.globl glVertexAttrib1fARB ; .type glVertexAttrib1fARB,#function
-		.globl glVertexAttrib1fvARB ; .type glVertexAttrib1fvARB,#function
-		.globl glVertexAttrib1sARB ; .type glVertexAttrib1sARB,#function
-		.globl glVertexAttrib1svARB ; .type glVertexAttrib1svARB,#function
-		.globl glVertexAttrib2dARB ; .type glVertexAttrib2dARB,#function
-		.globl glVertexAttrib2dvARB ; .type glVertexAttrib2dvARB,#function
-		.globl glVertexAttrib2fARB ; .type glVertexAttrib2fARB,#function
-		.globl glVertexAttrib2fvARB ; .type glVertexAttrib2fvARB,#function
-		.globl glVertexAttrib2sARB ; .type glVertexAttrib2sARB,#function
-		.globl glVertexAttrib2svARB ; .type glVertexAttrib2svARB,#function
-		.globl glVertexAttrib3dARB ; .type glVertexAttrib3dARB,#function
-		.globl glVertexAttrib3dvARB ; .type glVertexAttrib3dvARB,#function
-		.globl glVertexAttrib3fARB ; .type glVertexAttrib3fARB,#function
-		.globl glVertexAttrib3fvARB ; .type glVertexAttrib3fvARB,#function
-		.globl glVertexAttrib3sARB ; .type glVertexAttrib3sARB,#function
-		.globl glVertexAttrib3svARB ; .type glVertexAttrib3svARB,#function
-		.globl glVertexAttrib4NbvARB ; .type glVertexAttrib4NbvARB,#function
-		.globl glVertexAttrib4NivARB ; .type glVertexAttrib4NivARB,#function
-		.globl glVertexAttrib4NsvARB ; .type glVertexAttrib4NsvARB,#function
-		.globl glVertexAttrib4NubARB ; .type glVertexAttrib4NubARB,#function
-		.globl glVertexAttrib4NubvARB ; .type glVertexAttrib4NubvARB,#function
-		.globl glVertexAttrib4NuivARB ; .type glVertexAttrib4NuivARB,#function
-		.globl glVertexAttrib4NusvARB ; .type glVertexAttrib4NusvARB,#function
-		.globl glVertexAttrib4bvARB ; .type glVertexAttrib4bvARB,#function
-		.globl glVertexAttrib4dARB ; .type glVertexAttrib4dARB,#function
-		.globl glVertexAttrib4dvARB ; .type glVertexAttrib4dvARB,#function
-		.globl glVertexAttrib4fARB ; .type glVertexAttrib4fARB,#function
-		.globl glVertexAttrib4fvARB ; .type glVertexAttrib4fvARB,#function
-		.globl glVertexAttrib4ivARB ; .type glVertexAttrib4ivARB,#function
-		.globl glVertexAttrib4sARB ; .type glVertexAttrib4sARB,#function
-		.globl glVertexAttrib4svARB ; .type glVertexAttrib4svARB,#function
-		.globl glVertexAttrib4ubvARB ; .type glVertexAttrib4ubvARB,#function
-		.globl glVertexAttrib4uivARB ; .type glVertexAttrib4uivARB,#function
-		.globl glVertexAttrib4usvARB ; .type glVertexAttrib4usvARB,#function
-		.globl glVertexAttribPointerARB ; .type glVertexAttribPointerARB,#function
-		.globl glBindBufferARB ; .type glBindBufferARB,#function
-		.globl glBufferDataARB ; .type glBufferDataARB,#function
-		.globl glBufferSubDataARB ; .type glBufferSubDataARB,#function
-		.globl glDeleteBuffersARB ; .type glDeleteBuffersARB,#function
-		.globl glGenBuffersARB ; .type glGenBuffersARB,#function
-		.globl glGetBufferParameterivARB ; .type glGetBufferParameterivARB,#function
-		.globl glGetBufferPointervARB ; .type glGetBufferPointervARB,#function
-		.globl glGetBufferSubDataARB ; .type glGetBufferSubDataARB,#function
-		.globl glIsBufferARB ; .type glIsBufferARB,#function
-		.globl glMapBufferARB ; .type glMapBufferARB,#function
-		.globl glUnmapBufferARB ; .type glUnmapBufferARB,#function
-		.globl glBeginQueryARB ; .type glBeginQueryARB,#function
-		.globl glDeleteQueriesARB ; .type glDeleteQueriesARB,#function
-		.globl glEndQueryARB ; .type glEndQueryARB,#function
-		.globl glGenQueriesARB ; .type glGenQueriesARB,#function
-		.globl glGetQueryObjectivARB ; .type glGetQueryObjectivARB,#function
-		.globl glGetQueryObjectuivARB ; .type glGetQueryObjectuivARB,#function
-		.globl glGetQueryivARB ; .type glGetQueryivARB,#function
-		.globl glIsQueryARB ; .type glIsQueryARB,#function
-		.globl glAttachObjectARB ; .type glAttachObjectARB,#function
-		.globl glCompileShaderARB ; .type glCompileShaderARB,#function
-		.globl glCreateProgramObjectARB ; .type glCreateProgramObjectARB,#function
-		.globl glCreateShaderObjectARB ; .type glCreateShaderObjectARB,#function
-		.globl glDeleteObjectARB ; .type glDeleteObjectARB,#function
-		.globl glDetachObjectARB ; .type glDetachObjectARB,#function
-		.globl glGetActiveUniformARB ; .type glGetActiveUniformARB,#function
-		.globl glGetAttachedObjectsARB ; .type glGetAttachedObjectsARB,#function
-		.globl glGetHandleARB ; .type glGetHandleARB,#function
-		.globl glGetInfoLogARB ; .type glGetInfoLogARB,#function
-		.globl glGetObjectParameterfvARB ; .type glGetObjectParameterfvARB,#function
-		.globl glGetObjectParameterivARB ; .type glGetObjectParameterivARB,#function
-		.globl glGetShaderSourceARB ; .type glGetShaderSourceARB,#function
-		.globl glGetUniformLocationARB ; .type glGetUniformLocationARB,#function
-		.globl glGetUniformfvARB ; .type glGetUniformfvARB,#function
-		.globl glGetUniformivARB ; .type glGetUniformivARB,#function
-		.globl glLinkProgramARB ; .type glLinkProgramARB,#function
-		.globl glShaderSourceARB ; .type glShaderSourceARB,#function
-		.globl glUniform1fARB ; .type glUniform1fARB,#function
-		.globl glUniform1fvARB ; .type glUniform1fvARB,#function
-		.globl glUniform1iARB ; .type glUniform1iARB,#function
-		.globl glUniform1ivARB ; .type glUniform1ivARB,#function
-		.globl glUniform2fARB ; .type glUniform2fARB,#function
-		.globl glUniform2fvARB ; .type glUniform2fvARB,#function
-		.globl glUniform2iARB ; .type glUniform2iARB,#function
-		.globl glUniform2ivARB ; .type glUniform2ivARB,#function
-		.globl glUniform3fARB ; .type glUniform3fARB,#function
-		.globl glUniform3fvARB ; .type glUniform3fvARB,#function
-		.globl glUniform3iARB ; .type glUniform3iARB,#function
-		.globl glUniform3ivARB ; .type glUniform3ivARB,#function
-		.globl glUniform4fARB ; .type glUniform4fARB,#function
-		.globl glUniform4fvARB ; .type glUniform4fvARB,#function
-		.globl glUniform4iARB ; .type glUniform4iARB,#function
-		.globl glUniform4ivARB ; .type glUniform4ivARB,#function
-		.globl glUniformMatrix2fvARB ; .type glUniformMatrix2fvARB,#function
-		.globl glUniformMatrix3fvARB ; .type glUniformMatrix3fvARB,#function
-		.globl glUniformMatrix4fvARB ; .type glUniformMatrix4fvARB,#function
-		.globl glUseProgramObjectARB ; .type glUseProgramObjectARB,#function
-		.globl glValidateProgramARB ; .type glValidateProgramARB,#function
-		.globl glBindAttribLocationARB ; .type glBindAttribLocationARB,#function
-		.globl glGetActiveAttribARB ; .type glGetActiveAttribARB,#function
-		.globl glGetAttribLocationARB ; .type glGetAttribLocationARB,#function
-		.globl glDrawBuffersARB ; .type glDrawBuffersARB,#function
-		.globl glRenderbufferStorageMultisample ; .type glRenderbufferStorageMultisample,#function
-		.globl glPolygonOffsetEXT ; .type glPolygonOffsetEXT,#function
-		.globl gl_dispatch_stub_563 ; .type gl_dispatch_stub_563,#function
-		.globl gl_dispatch_stub_564 ; .type gl_dispatch_stub_564,#function
-		.globl gl_dispatch_stub_565 ; .type gl_dispatch_stub_565,#function
-		.globl gl_dispatch_stub_566 ; .type gl_dispatch_stub_566,#function
-		.globl gl_dispatch_stub_567 ; .type gl_dispatch_stub_567,#function
-		.globl gl_dispatch_stub_568 ; .type gl_dispatch_stub_568,#function
-		.globl gl_dispatch_stub_569 ; .type gl_dispatch_stub_569,#function
-		.globl gl_dispatch_stub_570 ; .type gl_dispatch_stub_570,#function
-		.globl glColorPointerEXT ; .type glColorPointerEXT,#function
-		.globl glEdgeFlagPointerEXT ; .type glEdgeFlagPointerEXT,#function
-		.globl glIndexPointerEXT ; .type glIndexPointerEXT,#function
-		.globl glNormalPointerEXT ; .type glNormalPointerEXT,#function
-		.globl glTexCoordPointerEXT ; .type glTexCoordPointerEXT,#function
-		.globl glVertexPointerEXT ; .type glVertexPointerEXT,#function
-		.globl glPointParameterfEXT ; .type glPointParameterfEXT,#function
-		.globl glPointParameterfvEXT ; .type glPointParameterfvEXT,#function
-		.globl glLockArraysEXT ; .type glLockArraysEXT,#function
-		.globl glUnlockArraysEXT ; .type glUnlockArraysEXT,#function
-		.globl gl_dispatch_stub_581 ; .type gl_dispatch_stub_581,#function
-		.globl gl_dispatch_stub_582 ; .type gl_dispatch_stub_582,#function
-		.globl glSecondaryColor3bEXT ; .type glSecondaryColor3bEXT,#function
-		.globl glSecondaryColor3bvEXT ; .type glSecondaryColor3bvEXT,#function
-		.globl glSecondaryColor3dEXT ; .type glSecondaryColor3dEXT,#function
-		.globl glSecondaryColor3dvEXT ; .type glSecondaryColor3dvEXT,#function
-		.globl glSecondaryColor3fEXT ; .type glSecondaryColor3fEXT,#function
-		.globl glSecondaryColor3fvEXT ; .type glSecondaryColor3fvEXT,#function
-		.globl glSecondaryColor3iEXT ; .type glSecondaryColor3iEXT,#function
-		.globl glSecondaryColor3ivEXT ; .type glSecondaryColor3ivEXT,#function
-		.globl glSecondaryColor3sEXT ; .type glSecondaryColor3sEXT,#function
-		.globl glSecondaryColor3svEXT ; .type glSecondaryColor3svEXT,#function
-		.globl glSecondaryColor3ubEXT ; .type glSecondaryColor3ubEXT,#function
-		.globl glSecondaryColor3ubvEXT ; .type glSecondaryColor3ubvEXT,#function
-		.globl glSecondaryColor3uiEXT ; .type glSecondaryColor3uiEXT,#function
-		.globl glSecondaryColor3uivEXT ; .type glSecondaryColor3uivEXT,#function
-		.globl glSecondaryColor3usEXT ; .type glSecondaryColor3usEXT,#function
-		.globl glSecondaryColor3usvEXT ; .type glSecondaryColor3usvEXT,#function
-		.globl glSecondaryColorPointerEXT ; .type glSecondaryColorPointerEXT,#function
-		.globl glMultiDrawArraysEXT ; .type glMultiDrawArraysEXT,#function
-		.globl glMultiDrawElementsEXT ; .type glMultiDrawElementsEXT,#function
-		.globl glFogCoordPointerEXT ; .type glFogCoordPointerEXT,#function
-		.globl glFogCoorddEXT ; .type glFogCoorddEXT,#function
-		.globl glFogCoorddvEXT ; .type glFogCoorddvEXT,#function
-		.globl glFogCoordfEXT ; .type glFogCoordfEXT,#function
-		.globl glFogCoordfvEXT ; .type glFogCoordfvEXT,#function
-		.globl gl_dispatch_stub_607 ; .type gl_dispatch_stub_607,#function
-		.globl glBlendFuncSeparateEXT ; .type glBlendFuncSeparateEXT,#function
-		.globl glFlushVertexArrayRangeNV ; .type glFlushVertexArrayRangeNV,#function
-		.globl glVertexArrayRangeNV ; .type glVertexArrayRangeNV,#function
-		.globl glCombinerInputNV ; .type glCombinerInputNV,#function
-		.globl glCombinerOutputNV ; .type glCombinerOutputNV,#function
-		.globl glCombinerParameterfNV ; .type glCombinerParameterfNV,#function
-		.globl glCombinerParameterfvNV ; .type glCombinerParameterfvNV,#function
-		.globl glCombinerParameteriNV ; .type glCombinerParameteriNV,#function
-		.globl glCombinerParameterivNV ; .type glCombinerParameterivNV,#function
-		.globl glFinalCombinerInputNV ; .type glFinalCombinerInputNV,#function
-		.globl glGetCombinerInputParameterfvNV ; .type glGetCombinerInputParameterfvNV,#function
-		.globl glGetCombinerInputParameterivNV ; .type glGetCombinerInputParameterivNV,#function
-		.globl glGetCombinerOutputParameterfvNV ; .type glGetCombinerOutputParameterfvNV,#function
-		.globl glGetCombinerOutputParameterivNV ; .type glGetCombinerOutputParameterivNV,#function
-		.globl glGetFinalCombinerInputParameterfvNV ; .type glGetFinalCombinerInputParameterfvNV,#function
-		.globl glGetFinalCombinerInputParameterivNV ; .type glGetFinalCombinerInputParameterivNV,#function
-		.globl glResizeBuffersMESA ; .type glResizeBuffersMESA,#function
-		.globl glWindowPos2dMESA ; .type glWindowPos2dMESA,#function
-		.globl glWindowPos2dvMESA ; .type glWindowPos2dvMESA,#function
-		.globl glWindowPos2fMESA ; .type glWindowPos2fMESA,#function
-		.globl glWindowPos2fvMESA ; .type glWindowPos2fvMESA,#function
-		.globl glWindowPos2iMESA ; .type glWindowPos2iMESA,#function
-		.globl glWindowPos2ivMESA ; .type glWindowPos2ivMESA,#function
-		.globl glWindowPos2sMESA ; .type glWindowPos2sMESA,#function
-		.globl glWindowPos2svMESA ; .type glWindowPos2svMESA,#function
-		.globl glWindowPos3dMESA ; .type glWindowPos3dMESA,#function
-		.globl glWindowPos3dvMESA ; .type glWindowPos3dvMESA,#function
-		.globl glWindowPos3fMESA ; .type glWindowPos3fMESA,#function
-		.globl glWindowPos3fvMESA ; .type glWindowPos3fvMESA,#function
-		.globl glWindowPos3iMESA ; .type glWindowPos3iMESA,#function
-		.globl glWindowPos3ivMESA ; .type glWindowPos3ivMESA,#function
-		.globl glWindowPos3sMESA ; .type glWindowPos3sMESA,#function
-		.globl glWindowPos3svMESA ; .type glWindowPos3svMESA,#function
-		.globl glWindowPos4dMESA ; .type glWindowPos4dMESA,#function
-		.globl glWindowPos4dvMESA ; .type glWindowPos4dvMESA,#function
-		.globl glWindowPos4fMESA ; .type glWindowPos4fMESA,#function
-		.globl glWindowPos4fvMESA ; .type glWindowPos4fvMESA,#function
-		.globl glWindowPos4iMESA ; .type glWindowPos4iMESA,#function
-		.globl glWindowPos4ivMESA ; .type glWindowPos4ivMESA,#function
-		.globl glWindowPos4sMESA ; .type glWindowPos4sMESA,#function
-		.globl glWindowPos4svMESA ; .type glWindowPos4svMESA,#function
-		.globl gl_dispatch_stub_649 ; .type gl_dispatch_stub_649,#function
-		.globl gl_dispatch_stub_650 ; .type gl_dispatch_stub_650,#function
-		.globl gl_dispatch_stub_651 ; .type gl_dispatch_stub_651,#function
-		.globl gl_dispatch_stub_652 ; .type gl_dispatch_stub_652,#function
-		.globl gl_dispatch_stub_653 ; .type gl_dispatch_stub_653,#function
-		.globl gl_dispatch_stub_654 ; .type gl_dispatch_stub_654,#function
-		.globl gl_dispatch_stub_655 ; .type gl_dispatch_stub_655,#function
-		.globl gl_dispatch_stub_656 ; .type gl_dispatch_stub_656,#function
-		.globl gl_dispatch_stub_657 ; .type gl_dispatch_stub_657,#function
-		.globl glAreProgramsResidentNV ; .type glAreProgramsResidentNV,#function
-		.globl glBindProgramNV ; .type glBindProgramNV,#function
-		.globl glDeleteProgramsNV ; .type glDeleteProgramsNV,#function
-		.globl glExecuteProgramNV ; .type glExecuteProgramNV,#function
-		.globl glGenProgramsNV ; .type glGenProgramsNV,#function
-		.globl glGetProgramParameterdvNV ; .type glGetProgramParameterdvNV,#function
-		.globl glGetProgramParameterfvNV ; .type glGetProgramParameterfvNV,#function
-		.globl glGetProgramStringNV ; .type glGetProgramStringNV,#function
-		.globl glGetProgramivNV ; .type glGetProgramivNV,#function
-		.globl glGetTrackMatrixivNV ; .type glGetTrackMatrixivNV,#function
-		.globl glGetVertexAttribPointervNV ; .type glGetVertexAttribPointervNV,#function
-		.globl glGetVertexAttribdvNV ; .type glGetVertexAttribdvNV,#function
-		.globl glGetVertexAttribfvNV ; .type glGetVertexAttribfvNV,#function
-		.globl glGetVertexAttribivNV ; .type glGetVertexAttribivNV,#function
-		.globl glIsProgramNV ; .type glIsProgramNV,#function
-		.globl glLoadProgramNV ; .type glLoadProgramNV,#function
-		.globl glProgramParameters4dvNV ; .type glProgramParameters4dvNV,#function
-		.globl glProgramParameters4fvNV ; .type glProgramParameters4fvNV,#function
-		.globl glRequestResidentProgramsNV ; .type glRequestResidentProgramsNV,#function
-		.globl glTrackMatrixNV ; .type glTrackMatrixNV,#function
-		.globl glVertexAttrib1dNV ; .type glVertexAttrib1dNV,#function
-		.globl glVertexAttrib1dvNV ; .type glVertexAttrib1dvNV,#function
-		.globl glVertexAttrib1fNV ; .type glVertexAttrib1fNV,#function
-		.globl glVertexAttrib1fvNV ; .type glVertexAttrib1fvNV,#function
-		.globl glVertexAttrib1sNV ; .type glVertexAttrib1sNV,#function
-		.globl glVertexAttrib1svNV ; .type glVertexAttrib1svNV,#function
-		.globl glVertexAttrib2dNV ; .type glVertexAttrib2dNV,#function
-		.globl glVertexAttrib2dvNV ; .type glVertexAttrib2dvNV,#function
-		.globl glVertexAttrib2fNV ; .type glVertexAttrib2fNV,#function
-		.globl glVertexAttrib2fvNV ; .type glVertexAttrib2fvNV,#function
-		.globl glVertexAttrib2sNV ; .type glVertexAttrib2sNV,#function
-		.globl glVertexAttrib2svNV ; .type glVertexAttrib2svNV,#function
-		.globl glVertexAttrib3dNV ; .type glVertexAttrib3dNV,#function
-		.globl glVertexAttrib3dvNV ; .type glVertexAttrib3dvNV,#function
-		.globl glVertexAttrib3fNV ; .type glVertexAttrib3fNV,#function
-		.globl glVertexAttrib3fvNV ; .type glVertexAttrib3fvNV,#function
-		.globl glVertexAttrib3sNV ; .type glVertexAttrib3sNV,#function
-		.globl glVertexAttrib3svNV ; .type glVertexAttrib3svNV,#function
-		.globl glVertexAttrib4dNV ; .type glVertexAttrib4dNV,#function
-		.globl glVertexAttrib4dvNV ; .type glVertexAttrib4dvNV,#function
-		.globl glVertexAttrib4fNV ; .type glVertexAttrib4fNV,#function
-		.globl glVertexAttrib4fvNV ; .type glVertexAttrib4fvNV,#function
-		.globl glVertexAttrib4sNV ; .type glVertexAttrib4sNV,#function
-		.globl glVertexAttrib4svNV ; .type glVertexAttrib4svNV,#function
-		.globl glVertexAttrib4ubNV ; .type glVertexAttrib4ubNV,#function
-		.globl glVertexAttrib4ubvNV ; .type glVertexAttrib4ubvNV,#function
-		.globl glVertexAttribPointerNV ; .type glVertexAttribPointerNV,#function
-		.globl glVertexAttribs1dvNV ; .type glVertexAttribs1dvNV,#function
-		.globl glVertexAttribs1fvNV ; .type glVertexAttribs1fvNV,#function
-		.globl glVertexAttribs1svNV ; .type glVertexAttribs1svNV,#function
-		.globl glVertexAttribs2dvNV ; .type glVertexAttribs2dvNV,#function
-		.globl glVertexAttribs2fvNV ; .type glVertexAttribs2fvNV,#function
-		.globl glVertexAttribs2svNV ; .type glVertexAttribs2svNV,#function
-		.globl glVertexAttribs3dvNV ; .type glVertexAttribs3dvNV,#function
-		.globl glVertexAttribs3fvNV ; .type glVertexAttribs3fvNV,#function
-		.globl glVertexAttribs3svNV ; .type glVertexAttribs3svNV,#function
-		.globl glVertexAttribs4dvNV ; .type glVertexAttribs4dvNV,#function
-		.globl glVertexAttribs4fvNV ; .type glVertexAttribs4fvNV,#function
-		.globl glVertexAttribs4svNV ; .type glVertexAttribs4svNV,#function
-		.globl glVertexAttribs4ubvNV ; .type glVertexAttribs4ubvNV,#function
-		.globl glAlphaFragmentOp1ATI ; .type glAlphaFragmentOp1ATI,#function
-		.globl glAlphaFragmentOp2ATI ; .type glAlphaFragmentOp2ATI,#function
-		.globl glAlphaFragmentOp3ATI ; .type glAlphaFragmentOp3ATI,#function
-		.globl glBeginFragmentShaderATI ; .type glBeginFragmentShaderATI,#function
-		.globl glBindFragmentShaderATI ; .type glBindFragmentShaderATI,#function
-		.globl glColorFragmentOp1ATI ; .type glColorFragmentOp1ATI,#function
-		.globl glColorFragmentOp2ATI ; .type glColorFragmentOp2ATI,#function
-		.globl glColorFragmentOp3ATI ; .type glColorFragmentOp3ATI,#function
-		.globl glDeleteFragmentShaderATI ; .type glDeleteFragmentShaderATI,#function
-		.globl glEndFragmentShaderATI ; .type glEndFragmentShaderATI,#function
-		.globl glGenFragmentShadersATI ; .type glGenFragmentShadersATI,#function
-		.globl glPassTexCoordATI ; .type glPassTexCoordATI,#function
-		.globl glSampleMapATI ; .type glSampleMapATI,#function
-		.globl glSetFragmentShaderConstantATI ; .type glSetFragmentShaderConstantATI,#function
-		.globl glPointParameteriNV ; .type glPointParameteriNV,#function
-		.globl glPointParameterivNV ; .type glPointParameterivNV,#function
-		.globl gl_dispatch_stub_734 ; .type gl_dispatch_stub_734,#function
-		.globl gl_dispatch_stub_735 ; .type gl_dispatch_stub_735,#function
-		.globl gl_dispatch_stub_736 ; .type gl_dispatch_stub_736,#function
-		.globl gl_dispatch_stub_737 ; .type gl_dispatch_stub_737,#function
-		.globl gl_dispatch_stub_738 ; .type gl_dispatch_stub_738,#function
-		.globl glGetProgramNamedParameterdvNV ; .type glGetProgramNamedParameterdvNV,#function
-		.globl glGetProgramNamedParameterfvNV ; .type glGetProgramNamedParameterfvNV,#function
-		.globl glProgramNamedParameter4dNV ; .type glProgramNamedParameter4dNV,#function
-		.globl glProgramNamedParameter4dvNV ; .type glProgramNamedParameter4dvNV,#function
-		.globl glProgramNamedParameter4fNV ; .type glProgramNamedParameter4fNV,#function
-		.globl glProgramNamedParameter4fvNV ; .type glProgramNamedParameter4fvNV,#function
-		.globl gl_dispatch_stub_745 ; .type gl_dispatch_stub_745,#function
-		.globl gl_dispatch_stub_746 ; .type gl_dispatch_stub_746,#function
-		.globl glBindFramebufferEXT ; .type glBindFramebufferEXT,#function
-		.globl glBindRenderbufferEXT ; .type glBindRenderbufferEXT,#function
-		.globl glCheckFramebufferStatusEXT ; .type glCheckFramebufferStatusEXT,#function
-		.globl glDeleteFramebuffersEXT ; .type glDeleteFramebuffersEXT,#function
-		.globl glDeleteRenderbuffersEXT ; .type glDeleteRenderbuffersEXT,#function
-		.globl glFramebufferRenderbufferEXT ; .type glFramebufferRenderbufferEXT,#function
-		.globl glFramebufferTexture1DEXT ; .type glFramebufferTexture1DEXT,#function
-		.globl glFramebufferTexture2DEXT ; .type glFramebufferTexture2DEXT,#function
-		.globl glFramebufferTexture3DEXT ; .type glFramebufferTexture3DEXT,#function
-		.globl glGenFramebuffersEXT ; .type glGenFramebuffersEXT,#function
-		.globl glGenRenderbuffersEXT ; .type glGenRenderbuffersEXT,#function
-		.globl glGenerateMipmapEXT ; .type glGenerateMipmapEXT,#function
-		.globl glGetFramebufferAttachmentParameterivEXT ; .type glGetFramebufferAttachmentParameterivEXT,#function
-		.globl glGetRenderbufferParameterivEXT ; .type glGetRenderbufferParameterivEXT,#function
-		.globl glIsFramebufferEXT ; .type glIsFramebufferEXT,#function
-		.globl glIsRenderbufferEXT ; .type glIsRenderbufferEXT,#function
-		.globl glRenderbufferStorageEXT ; .type glRenderbufferStorageEXT,#function
-		.globl gl_dispatch_stub_764 ; .type gl_dispatch_stub_764,#function
-		.globl glFramebufferTextureLayerEXT ; .type glFramebufferTextureLayerEXT,#function
-		.globl gl_dispatch_stub_766 ; .type gl_dispatch_stub_766,#function
-		.globl gl_dispatch_stub_767 ; .type gl_dispatch_stub_767,#function
-		.globl gl_dispatch_stub_768 ; .type gl_dispatch_stub_768,#function
-		.globl gl_dispatch_stub_769 ; .type gl_dispatch_stub_769,#function
-		.globl gl_dispatch_stub_770 ; .type gl_dispatch_stub_770,#function
-		.globl _mesa_sparc_glapi_begin ; .type _mesa_sparc_glapi_begin,#function
-_mesa_sparc_glapi_begin:
+	.data
+	.align	32
+
+	GLOBL_FN(__glapi_sparc_pthread_stub)
+	HIDDEN(__glapi_sparc_pthread_stub)
+__glapi_sparc_pthread_stub: /* Call offset in %g3 */
+	mov	%o7, %g1
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %g2
+	call	__glapi_sparc_get_pc
+	 add	%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2
+	mov	%g1, %o7
+	sethi	%hi(_glapi_Dispatch), %g1
+	or	%g1, %lo(_glapi_Dispatch), %g1
+	srl	%g3, 10, %g3
+	GL_LL	[%g2+%g1], %g2
+	GL_LL	[%g2], %g1
+	cmp	%g1, 0
+	be	2f
+	 nop
+1:	GL_LL	[%g1 + %g3], %g1
+	jmp	%g1
+	 nop
+2:	save	%sp, GL_STACK_SIZE, %sp
+	mov	%g3, %l0
+	call	_glapi_get_dispatch
+	 nop
+	mov	%o0, %g1
+	mov	%l0, %g3
+	ba	1b
+	 restore %g0, %g0, %g0
+	.size	__glapi_sparc_pthread_stub, .-__glapi_sparc_pthread_stub
+
+#define GL_STUB(fn, off)			\
+	GLOBL_FN(fn);				\
+fn:	ba	__glapi_sparc_pthread_stub;	\
+	 sethi	GL_OFF(off), %g3;		\
+	.size	fn,.-fn;
+
+#else /* Non-threaded version. */
+
+	.type	__glapi_sparc_nothread_stub, @function
+__glapi_sparc_nothread_stub: /* Call offset in %g3 */
+	mov	%o7, %g1
+	sethi	%hi(_GLOBAL_OFFSET_TABLE_-4), %g2
+	call	__glapi_sparc_get_pc
+	 add	%g2, %lo(_GLOBAL_OFFSET_TABLE_+4), %g2
+	mov	%g1, %o7
+	srl	%g3, 10, %g3
+	sethi	%hi(_glapi_Dispatch), %g1
+	or	%g1, %lo(_glapi_Dispatch), %g1
+	GL_LL	[%g2+%g1], %g2
+	GL_LL	[%g2], %g1
+	GL_LL	[%g1 + %g3], %g1
+	jmp	%g1
+	 nop
+	.size	__glapi_sparc_nothread_stub, .-__glapi_sparc_nothread_stub
+
+#define GL_STUB(fn, off)			\
+	GLOBL_FN(fn);				\
+fn:	ba	__glapi_sparc_nothread_stub;	\
+	 sethi	GL_OFF(off), %g3;		\
+	.size	fn,.-fn;
+
+#endif
+
+#define GL_STUB_ALIAS(fn, alias)		\
+	.globl	fn;				\
+	.set	fn, alias
+
+	.text
+	.align	32
+
+	.globl	gl_dispatch_functions_start
+	HIDDEN(gl_dispatch_functions_start)
+gl_dispatch_functions_start:
 
 	GL_STUB(glNewList, _gloffset_NewList)
 	GL_STUB(glEndList, _gloffset_EndList)
@@ -1402,14 +760,22 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glDrawBuffersARB, _gloffset_DrawBuffersARB)
 	GL_STUB(glRenderbufferStorageMultisample, _gloffset_RenderbufferStorageMultisample)
 	GL_STUB(glPolygonOffsetEXT, _gloffset_PolygonOffsetEXT)
-	GL_STUB(gl_dispatch_stub_563, _gloffset__dispatch_stub_563)
-	GL_STUB(gl_dispatch_stub_564, _gloffset__dispatch_stub_564)
-	GL_STUB(gl_dispatch_stub_565, _gloffset__dispatch_stub_565)
-	GL_STUB(gl_dispatch_stub_566, _gloffset__dispatch_stub_566)
-	GL_STUB(gl_dispatch_stub_567, _gloffset__dispatch_stub_567)
-	GL_STUB(gl_dispatch_stub_568, _gloffset__dispatch_stub_568)
-	GL_STUB(gl_dispatch_stub_569, _gloffset__dispatch_stub_569)
-	GL_STUB(gl_dispatch_stub_570, _gloffset__dispatch_stub_570)
+	GL_STUB(gl_dispatch_stub_563, _gloffset_GetPixelTexGenParameterfvSGIS)
+	HIDDEN(gl_dispatch_stub_563)
+	GL_STUB(gl_dispatch_stub_564, _gloffset_GetPixelTexGenParameterivSGIS)
+	HIDDEN(gl_dispatch_stub_564)
+	GL_STUB(gl_dispatch_stub_565, _gloffset_PixelTexGenParameterfSGIS)
+	HIDDEN(gl_dispatch_stub_565)
+	GL_STUB(gl_dispatch_stub_566, _gloffset_PixelTexGenParameterfvSGIS)
+	HIDDEN(gl_dispatch_stub_566)
+	GL_STUB(gl_dispatch_stub_567, _gloffset_PixelTexGenParameteriSGIS)
+	HIDDEN(gl_dispatch_stub_567)
+	GL_STUB(gl_dispatch_stub_568, _gloffset_PixelTexGenParameterivSGIS)
+	HIDDEN(gl_dispatch_stub_568)
+	GL_STUB(gl_dispatch_stub_569, _gloffset_SampleMaskSGIS)
+	HIDDEN(gl_dispatch_stub_569)
+	GL_STUB(gl_dispatch_stub_570, _gloffset_SamplePatternSGIS)
+	HIDDEN(gl_dispatch_stub_570)
 	GL_STUB(glColorPointerEXT, _gloffset_ColorPointerEXT)
 	GL_STUB(glEdgeFlagPointerEXT, _gloffset_EdgeFlagPointerEXT)
 	GL_STUB(glIndexPointerEXT, _gloffset_IndexPointerEXT)
@@ -1420,8 +786,10 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glPointParameterfvEXT, _gloffset_PointParameterfvEXT)
 	GL_STUB(glLockArraysEXT, _gloffset_LockArraysEXT)
 	GL_STUB(glUnlockArraysEXT, _gloffset_UnlockArraysEXT)
-	GL_STUB(gl_dispatch_stub_581, _gloffset__dispatch_stub_581)
-	GL_STUB(gl_dispatch_stub_582, _gloffset__dispatch_stub_582)
+	GL_STUB(gl_dispatch_stub_581, _gloffset_CullParameterdvEXT)
+	HIDDEN(gl_dispatch_stub_581)
+	GL_STUB(gl_dispatch_stub_582, _gloffset_CullParameterfvEXT)
+	HIDDEN(gl_dispatch_stub_582)
 	GL_STUB(glSecondaryColor3bEXT, _gloffset_SecondaryColor3bEXT)
 	GL_STUB(glSecondaryColor3bvEXT, _gloffset_SecondaryColor3bvEXT)
 	GL_STUB(glSecondaryColor3dEXT, _gloffset_SecondaryColor3dEXT)
@@ -1446,7 +814,8 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glFogCoorddvEXT, _gloffset_FogCoorddvEXT)
 	GL_STUB(glFogCoordfEXT, _gloffset_FogCoordfEXT)
 	GL_STUB(glFogCoordfvEXT, _gloffset_FogCoordfvEXT)
-	GL_STUB(gl_dispatch_stub_607, _gloffset__dispatch_stub_607)
+	GL_STUB(gl_dispatch_stub_607, _gloffset_PixelTexGenSGIX)
+	HIDDEN(gl_dispatch_stub_607)
 	GL_STUB(glBlendFuncSeparateEXT, _gloffset_BlendFuncSeparateEXT)
 	GL_STUB(glFlushVertexArrayRangeNV, _gloffset_FlushVertexArrayRangeNV)
 	GL_STUB(glVertexArrayRangeNV, _gloffset_VertexArrayRangeNV)
@@ -1488,15 +857,24 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glWindowPos4ivMESA, _gloffset_WindowPos4ivMESA)
 	GL_STUB(glWindowPos4sMESA, _gloffset_WindowPos4sMESA)
 	GL_STUB(glWindowPos4svMESA, _gloffset_WindowPos4svMESA)
-	GL_STUB(gl_dispatch_stub_649, _gloffset__dispatch_stub_649)
-	GL_STUB(gl_dispatch_stub_650, _gloffset__dispatch_stub_650)
-	GL_STUB(gl_dispatch_stub_651, _gloffset__dispatch_stub_651)
-	GL_STUB(gl_dispatch_stub_652, _gloffset__dispatch_stub_652)
-	GL_STUB(gl_dispatch_stub_653, _gloffset__dispatch_stub_653)
-	GL_STUB(gl_dispatch_stub_654, _gloffset__dispatch_stub_654)
-	GL_STUB(gl_dispatch_stub_655, _gloffset__dispatch_stub_655)
-	GL_STUB(gl_dispatch_stub_656, _gloffset__dispatch_stub_656)
-	GL_STUB(gl_dispatch_stub_657, _gloffset__dispatch_stub_657)
+	GL_STUB(gl_dispatch_stub_649, _gloffset_MultiModeDrawArraysIBM)
+	HIDDEN(gl_dispatch_stub_649)
+	GL_STUB(gl_dispatch_stub_650, _gloffset_MultiModeDrawElementsIBM)
+	HIDDEN(gl_dispatch_stub_650)
+	GL_STUB(gl_dispatch_stub_651, _gloffset_DeleteFencesNV)
+	HIDDEN(gl_dispatch_stub_651)
+	GL_STUB(gl_dispatch_stub_652, _gloffset_FinishFenceNV)
+	HIDDEN(gl_dispatch_stub_652)
+	GL_STUB(gl_dispatch_stub_653, _gloffset_GenFencesNV)
+	HIDDEN(gl_dispatch_stub_653)
+	GL_STUB(gl_dispatch_stub_654, _gloffset_GetFenceivNV)
+	HIDDEN(gl_dispatch_stub_654)
+	GL_STUB(gl_dispatch_stub_655, _gloffset_IsFenceNV)
+	HIDDEN(gl_dispatch_stub_655)
+	GL_STUB(gl_dispatch_stub_656, _gloffset_SetFenceNV)
+	HIDDEN(gl_dispatch_stub_656)
+	GL_STUB(gl_dispatch_stub_657, _gloffset_TestFenceNV)
+	HIDDEN(gl_dispatch_stub_657)
 	GL_STUB(glAreProgramsResidentNV, _gloffset_AreProgramsResidentNV)
 	GL_STUB(glBindProgramNV, _gloffset_BindProgramNV)
 	GL_STUB(glDeleteProgramsNV, _gloffset_DeleteProgramsNV)
@@ -1573,19 +951,26 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glSetFragmentShaderConstantATI, _gloffset_SetFragmentShaderConstantATI)
 	GL_STUB(glPointParameteriNV, _gloffset_PointParameteriNV)
 	GL_STUB(glPointParameterivNV, _gloffset_PointParameterivNV)
-	GL_STUB(gl_dispatch_stub_734, _gloffset__dispatch_stub_734)
-	GL_STUB(gl_dispatch_stub_735, _gloffset__dispatch_stub_735)
-	GL_STUB(gl_dispatch_stub_736, _gloffset__dispatch_stub_736)
-	GL_STUB(gl_dispatch_stub_737, _gloffset__dispatch_stub_737)
-	GL_STUB(gl_dispatch_stub_738, _gloffset__dispatch_stub_738)
+	GL_STUB(gl_dispatch_stub_734, _gloffset_ActiveStencilFaceEXT)
+	HIDDEN(gl_dispatch_stub_734)
+	GL_STUB(gl_dispatch_stub_735, _gloffset_BindVertexArrayAPPLE)
+	HIDDEN(gl_dispatch_stub_735)
+	GL_STUB(gl_dispatch_stub_736, _gloffset_DeleteVertexArraysAPPLE)
+	HIDDEN(gl_dispatch_stub_736)
+	GL_STUB(gl_dispatch_stub_737, _gloffset_GenVertexArraysAPPLE)
+	HIDDEN(gl_dispatch_stub_737)
+	GL_STUB(gl_dispatch_stub_738, _gloffset_IsVertexArrayAPPLE)
+	HIDDEN(gl_dispatch_stub_738)
 	GL_STUB(glGetProgramNamedParameterdvNV, _gloffset_GetProgramNamedParameterdvNV)
 	GL_STUB(glGetProgramNamedParameterfvNV, _gloffset_GetProgramNamedParameterfvNV)
 	GL_STUB(glProgramNamedParameter4dNV, _gloffset_ProgramNamedParameter4dNV)
 	GL_STUB(glProgramNamedParameter4dvNV, _gloffset_ProgramNamedParameter4dvNV)
 	GL_STUB(glProgramNamedParameter4fNV, _gloffset_ProgramNamedParameter4fNV)
 	GL_STUB(glProgramNamedParameter4fvNV, _gloffset_ProgramNamedParameter4fvNV)
-	GL_STUB(gl_dispatch_stub_745, _gloffset__dispatch_stub_745)
-	GL_STUB(gl_dispatch_stub_746, _gloffset__dispatch_stub_746)
+	GL_STUB(gl_dispatch_stub_745, _gloffset_DepthBoundsEXT)
+	HIDDEN(gl_dispatch_stub_745)
+	GL_STUB(gl_dispatch_stub_746, _gloffset_BlendEquationSeparateEXT)
+	HIDDEN(gl_dispatch_stub_746)
 	GL_STUB(glBindFramebufferEXT, _gloffset_BindFramebufferEXT)
 	GL_STUB(glBindRenderbufferEXT, _gloffset_BindRenderbufferEXT)
 	GL_STUB(glCheckFramebufferStatusEXT, _gloffset_CheckFramebufferStatusEXT)
@@ -1603,287 +988,353 @@ _mesa_sparc_glapi_begin:
 	GL_STUB(glIsFramebufferEXT, _gloffset_IsFramebufferEXT)
 	GL_STUB(glIsRenderbufferEXT, _gloffset_IsRenderbufferEXT)
 	GL_STUB(glRenderbufferStorageEXT, _gloffset_RenderbufferStorageEXT)
-	GL_STUB(gl_dispatch_stub_764, _gloffset__dispatch_stub_764)
+	GL_STUB(gl_dispatch_stub_764, _gloffset_BlitFramebufferEXT)
+	HIDDEN(gl_dispatch_stub_764)
 	GL_STUB(glFramebufferTextureLayerEXT, _gloffset_FramebufferTextureLayerEXT)
-	GL_STUB(gl_dispatch_stub_766, _gloffset__dispatch_stub_766)
-	GL_STUB(gl_dispatch_stub_767, _gloffset__dispatch_stub_767)
-	GL_STUB(gl_dispatch_stub_768, _gloffset__dispatch_stub_768)
-	GL_STUB(gl_dispatch_stub_769, _gloffset__dispatch_stub_769)
-	GL_STUB(gl_dispatch_stub_770, _gloffset__dispatch_stub_770)
-
-		.globl _mesa_sparc_glapi_end ; .type _mesa_sparc_glapi_end,#function
-_mesa_sparc_glapi_end:
-
-	.globl glArrayElementEXT ; .type glArrayElementEXT,#function ; glArrayElementEXT = glArrayElement
-	.globl glBindTextureEXT ; .type glBindTextureEXT,#function ; glBindTextureEXT = glBindTexture
-	.globl glDrawArraysEXT ; .type glDrawArraysEXT,#function ; glDrawArraysEXT = glDrawArrays
+	GL_STUB(gl_dispatch_stub_766, _gloffset_StencilFuncSeparateATI)
+	HIDDEN(gl_dispatch_stub_766)
+	GL_STUB(gl_dispatch_stub_767, _gloffset_ProgramEnvParameters4fvEXT)
+	HIDDEN(gl_dispatch_stub_767)
+	GL_STUB(gl_dispatch_stub_768, _gloffset_ProgramLocalParameters4fvEXT)
+	HIDDEN(gl_dispatch_stub_768)
+	GL_STUB(gl_dispatch_stub_769, _gloffset_GetQueryObjecti64vEXT)
+	HIDDEN(gl_dispatch_stub_769)
+	GL_STUB(gl_dispatch_stub_770, _gloffset_GetQueryObjectui64vEXT)
+	HIDDEN(gl_dispatch_stub_770)
+	GL_STUB_ALIAS(glArrayElementEXT, glArrayElement)
+	GL_STUB_ALIAS(glBindTextureEXT, glBindTexture)
+	GL_STUB_ALIAS(glDrawArraysEXT, glDrawArrays)
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glAreTexturesResidentEXT, glAreTexturesResident)
+#endif
+	GL_STUB_ALIAS(glCopyTexImage1DEXT, glCopyTexImage1D)
+	GL_STUB_ALIAS(glCopyTexImage2DEXT, glCopyTexImage2D)
+	GL_STUB_ALIAS(glCopyTexSubImage1DEXT, glCopyTexSubImage1D)
+	GL_STUB_ALIAS(glCopyTexSubImage2DEXT, glCopyTexSubImage2D)
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glDeleteTexturesEXT, glDeleteTextures)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGenTexturesEXT, glGenTextures)
+#endif
+	GL_STUB_ALIAS(glGetPointervEXT, glGetPointerv)
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glIsTextureEXT, glIsTexture)
+#endif
+	GL_STUB_ALIAS(glPrioritizeTexturesEXT, glPrioritizeTextures)
+	GL_STUB_ALIAS(glTexSubImage1DEXT, glTexSubImage1D)
+	GL_STUB_ALIAS(glTexSubImage2DEXT, glTexSubImage2D)
+	GL_STUB_ALIAS(glBlendColorEXT, glBlendColor)
+	GL_STUB_ALIAS(glBlendEquationEXT, glBlendEquation)
+	GL_STUB_ALIAS(glDrawRangeElementsEXT, glDrawRangeElements)
+	GL_STUB_ALIAS(glColorTableSGI, glColorTable)
+	GL_STUB_ALIAS(glColorTableEXT, glColorTable)
+	GL_STUB_ALIAS(glColorTableParameterfvSGI, glColorTableParameterfv)
+	GL_STUB_ALIAS(glColorTableParameterivSGI, glColorTableParameteriv)
+	GL_STUB_ALIAS(glCopyColorTableSGI, glCopyColorTable)
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetColorTableSGI, glGetColorTable)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetColorTableEXT, glGetColorTable)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetColorTableParameterfvSGI, glGetColorTableParameterfv)
+#endif
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glAreTexturesResidentEXT ; .type glAreTexturesResidentEXT,#function ; glAreTexturesResidentEXT = glAreTexturesResident
+	GL_STUB_ALIAS(glGetColorTableParameterfvEXT, glGetColorTableParameterfv)
 #endif
-	.globl glCopyTexImage1DEXT ; .type glCopyTexImage1DEXT,#function ; glCopyTexImage1DEXT = glCopyTexImage1D
-	.globl glCopyTexImage2DEXT ; .type glCopyTexImage2DEXT,#function ; glCopyTexImage2DEXT = glCopyTexImage2D
-	.globl glCopyTexSubImage1DEXT ; .type glCopyTexSubImage1DEXT,#function ; glCopyTexSubImage1DEXT = glCopyTexSubImage1D
-	.globl glCopyTexSubImage2DEXT ; .type glCopyTexSubImage2DEXT,#function ; glCopyTexSubImage2DEXT = glCopyTexSubImage2D
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glDeleteTexturesEXT ; .type glDeleteTexturesEXT,#function ; glDeleteTexturesEXT = glDeleteTextures
+	GL_STUB_ALIAS(glGetColorTableParameterivSGI, glGetColorTableParameteriv)
 #endif
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glGenTexturesEXT ; .type glGenTexturesEXT,#function ; glGenTexturesEXT = glGenTextures
+	GL_STUB_ALIAS(glGetColorTableParameterivEXT, glGetColorTableParameteriv)
 #endif
-	.globl glGetPointervEXT ; .type glGetPointervEXT,#function ; glGetPointervEXT = glGetPointerv
+	GL_STUB_ALIAS(glColorSubTableEXT, glColorSubTable)
+	GL_STUB_ALIAS(glCopyColorSubTableEXT, glCopyColorSubTable)
+	GL_STUB_ALIAS(glConvolutionFilter1DEXT, glConvolutionFilter1D)
+	GL_STUB_ALIAS(glConvolutionFilter2DEXT, glConvolutionFilter2D)
+	GL_STUB_ALIAS(glConvolutionParameterfEXT, glConvolutionParameterf)
+	GL_STUB_ALIAS(glConvolutionParameterfvEXT, glConvolutionParameterfv)
+	GL_STUB_ALIAS(glConvolutionParameteriEXT, glConvolutionParameteri)
+	GL_STUB_ALIAS(glConvolutionParameterivEXT, glConvolutionParameteriv)
+	GL_STUB_ALIAS(glCopyConvolutionFilter1DEXT, glCopyConvolutionFilter1D)
+	GL_STUB_ALIAS(glCopyConvolutionFilter2DEXT, glCopyConvolutionFilter2D)
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glIsTextureEXT ; .type glIsTextureEXT,#function ; glIsTextureEXT = glIsTexture
+	GL_STUB_ALIAS(glGetConvolutionFilterEXT, glGetConvolutionFilter)
 #endif
-	.globl glPrioritizeTexturesEXT ; .type glPrioritizeTexturesEXT,#function ; glPrioritizeTexturesEXT = glPrioritizeTextures
-	.globl glTexSubImage1DEXT ; .type glTexSubImage1DEXT,#function ; glTexSubImage1DEXT = glTexSubImage1D
-	.globl glTexSubImage2DEXT ; .type glTexSubImage2DEXT,#function ; glTexSubImage2DEXT = glTexSubImage2D
-	.globl glBlendColorEXT ; .type glBlendColorEXT,#function ; glBlendColorEXT = glBlendColor
-	.globl glBlendEquationEXT ; .type glBlendEquationEXT,#function ; glBlendEquationEXT = glBlendEquation
-	.globl glDrawRangeElementsEXT ; .type glDrawRangeElementsEXT,#function ; glDrawRangeElementsEXT = glDrawRangeElements
-	.globl glColorTableEXT ; .type glColorTableEXT,#function ; glColorTableEXT = glColorTable
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glGetColorTableEXT ; .type glGetColorTableEXT,#function ; glGetColorTableEXT = glGetColorTable
+	GL_STUB_ALIAS(glGetConvolutionParameterfvEXT, glGetConvolutionParameterfv)
 #endif
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glGetColorTableParameterfvEXT ; .type glGetColorTableParameterfvEXT,#function ; glGetColorTableParameterfvEXT = glGetColorTableParameterfv
+	GL_STUB_ALIAS(glGetConvolutionParameterivEXT, glGetConvolutionParameteriv)
 #endif
 #ifndef GLX_INDIRECT_RENDERING
-	.globl glGetColorTableParameterivEXT ; .type glGetColorTableParameterivEXT,#function ; glGetColorTableParameterivEXT = glGetColorTableParameteriv
+	GL_STUB_ALIAS(glGetSeparableFilterEXT, glGetSeparableFilter)
 #endif
-	.globl glTexImage3DEXT ; .type glTexImage3DEXT,#function ; glTexImage3DEXT = glTexImage3D
-	.globl glTexSubImage3DEXT ; .type glTexSubImage3DEXT,#function ; glTexSubImage3DEXT = glTexSubImage3D
-	.globl glCopyTexSubImage3DEXT ; .type glCopyTexSubImage3DEXT,#function ; glCopyTexSubImage3DEXT = glCopyTexSubImage3D
-	.globl glActiveTexture ; .type glActiveTexture,#function ; glActiveTexture = glActiveTextureARB
-	.globl glClientActiveTexture ; .type glClientActiveTexture,#function ; glClientActiveTexture = glClientActiveTextureARB
-	.globl glMultiTexCoord1d ; .type glMultiTexCoord1d,#function ; glMultiTexCoord1d = glMultiTexCoord1dARB
-	.globl glMultiTexCoord1dv ; .type glMultiTexCoord1dv,#function ; glMultiTexCoord1dv = glMultiTexCoord1dvARB
-	.globl glMultiTexCoord1f ; .type glMultiTexCoord1f,#function ; glMultiTexCoord1f = glMultiTexCoord1fARB
-	.globl glMultiTexCoord1fv ; .type glMultiTexCoord1fv,#function ; glMultiTexCoord1fv = glMultiTexCoord1fvARB
-	.globl glMultiTexCoord1i ; .type glMultiTexCoord1i,#function ; glMultiTexCoord1i = glMultiTexCoord1iARB
-	.globl glMultiTexCoord1iv ; .type glMultiTexCoord1iv,#function ; glMultiTexCoord1iv = glMultiTexCoord1ivARB
-	.globl glMultiTexCoord1s ; .type glMultiTexCoord1s,#function ; glMultiTexCoord1s = glMultiTexCoord1sARB
-	.globl glMultiTexCoord1sv ; .type glMultiTexCoord1sv,#function ; glMultiTexCoord1sv = glMultiTexCoord1svARB
-	.globl glMultiTexCoord2d ; .type glMultiTexCoord2d,#function ; glMultiTexCoord2d = glMultiTexCoord2dARB
-	.globl glMultiTexCoord2dv ; .type glMultiTexCoord2dv,#function ; glMultiTexCoord2dv = glMultiTexCoord2dvARB
-	.globl glMultiTexCoord2f ; .type glMultiTexCoord2f,#function ; glMultiTexCoord2f = glMultiTexCoord2fARB
-	.globl glMultiTexCoord2fv ; .type glMultiTexCoord2fv,#function ; glMultiTexCoord2fv = glMultiTexCoord2fvARB
-	.globl glMultiTexCoord2i ; .type glMultiTexCoord2i,#function ; glMultiTexCoord2i = glMultiTexCoord2iARB
-	.globl glMultiTexCoord2iv ; .type glMultiTexCoord2iv,#function ; glMultiTexCoord2iv = glMultiTexCoord2ivARB
-	.globl glMultiTexCoord2s ; .type glMultiTexCoord2s,#function ; glMultiTexCoord2s = glMultiTexCoord2sARB
-	.globl glMultiTexCoord2sv ; .type glMultiTexCoord2sv,#function ; glMultiTexCoord2sv = glMultiTexCoord2svARB
-	.globl glMultiTexCoord3d ; .type glMultiTexCoord3d,#function ; glMultiTexCoord3d = glMultiTexCoord3dARB
-	.globl glMultiTexCoord3dv ; .type glMultiTexCoord3dv,#function ; glMultiTexCoord3dv = glMultiTexCoord3dvARB
-	.globl glMultiTexCoord3f ; .type glMultiTexCoord3f,#function ; glMultiTexCoord3f = glMultiTexCoord3fARB
-	.globl glMultiTexCoord3fv ; .type glMultiTexCoord3fv,#function ; glMultiTexCoord3fv = glMultiTexCoord3fvARB
-	.globl glMultiTexCoord3i ; .type glMultiTexCoord3i,#function ; glMultiTexCoord3i = glMultiTexCoord3iARB
-	.globl glMultiTexCoord3iv ; .type glMultiTexCoord3iv,#function ; glMultiTexCoord3iv = glMultiTexCoord3ivARB
-	.globl glMultiTexCoord3s ; .type glMultiTexCoord3s,#function ; glMultiTexCoord3s = glMultiTexCoord3sARB
-	.globl glMultiTexCoord3sv ; .type glMultiTexCoord3sv,#function ; glMultiTexCoord3sv = glMultiTexCoord3svARB
-	.globl glMultiTexCoord4d ; .type glMultiTexCoord4d,#function ; glMultiTexCoord4d = glMultiTexCoord4dARB
-	.globl glMultiTexCoord4dv ; .type glMultiTexCoord4dv,#function ; glMultiTexCoord4dv = glMultiTexCoord4dvARB
-	.globl glMultiTexCoord4f ; .type glMultiTexCoord4f,#function ; glMultiTexCoord4f = glMultiTexCoord4fARB
-	.globl glMultiTexCoord4fv ; .type glMultiTexCoord4fv,#function ; glMultiTexCoord4fv = glMultiTexCoord4fvARB
-	.globl glMultiTexCoord4i ; .type glMultiTexCoord4i,#function ; glMultiTexCoord4i = glMultiTexCoord4iARB
-	.globl glMultiTexCoord4iv ; .type glMultiTexCoord4iv,#function ; glMultiTexCoord4iv = glMultiTexCoord4ivARB
-	.globl glMultiTexCoord4s ; .type glMultiTexCoord4s,#function ; glMultiTexCoord4s = glMultiTexCoord4sARB
-	.globl glMultiTexCoord4sv ; .type glMultiTexCoord4sv,#function ; glMultiTexCoord4sv = glMultiTexCoord4svARB
-	.globl glLoadTransposeMatrixd ; .type glLoadTransposeMatrixd,#function ; glLoadTransposeMatrixd = glLoadTransposeMatrixdARB
-	.globl glLoadTransposeMatrixf ; .type glLoadTransposeMatrixf,#function ; glLoadTransposeMatrixf = glLoadTransposeMatrixfARB
-	.globl glMultTransposeMatrixd ; .type glMultTransposeMatrixd,#function ; glMultTransposeMatrixd = glMultTransposeMatrixdARB
-	.globl glMultTransposeMatrixf ; .type glMultTransposeMatrixf,#function ; glMultTransposeMatrixf = glMultTransposeMatrixfARB
-	.globl glSampleCoverage ; .type glSampleCoverage,#function ; glSampleCoverage = glSampleCoverageARB
-	.globl glCompressedTexImage1D ; .type glCompressedTexImage1D,#function ; glCompressedTexImage1D = glCompressedTexImage1DARB
-	.globl glCompressedTexImage2D ; .type glCompressedTexImage2D,#function ; glCompressedTexImage2D = glCompressedTexImage2DARB
-	.globl glCompressedTexImage3D ; .type glCompressedTexImage3D,#function ; glCompressedTexImage3D = glCompressedTexImage3DARB
-	.globl glCompressedTexSubImage1D ; .type glCompressedTexSubImage1D,#function ; glCompressedTexSubImage1D = glCompressedTexSubImage1DARB
-	.globl glCompressedTexSubImage2D ; .type glCompressedTexSubImage2D,#function ; glCompressedTexSubImage2D = glCompressedTexSubImage2DARB
-	.globl glCompressedTexSubImage3D ; .type glCompressedTexSubImage3D,#function ; glCompressedTexSubImage3D = glCompressedTexSubImage3DARB
-	.globl glGetCompressedTexImage ; .type glGetCompressedTexImage,#function ; glGetCompressedTexImage = glGetCompressedTexImageARB
-	.globl glDisableVertexAttribArray ; .type glDisableVertexAttribArray,#function ; glDisableVertexAttribArray = glDisableVertexAttribArrayARB
-	.globl glEnableVertexAttribArray ; .type glEnableVertexAttribArray,#function ; glEnableVertexAttribArray = glEnableVertexAttribArrayARB
-	.globl glGetVertexAttribdv ; .type glGetVertexAttribdv,#function ; glGetVertexAttribdv = glGetVertexAttribdvARB
-	.globl glGetVertexAttribfv ; .type glGetVertexAttribfv,#function ; glGetVertexAttribfv = glGetVertexAttribfvARB
-	.globl glGetVertexAttribiv ; .type glGetVertexAttribiv,#function ; glGetVertexAttribiv = glGetVertexAttribivARB
-	.globl glProgramParameter4dNV ; .type glProgramParameter4dNV,#function ; glProgramParameter4dNV = glProgramEnvParameter4dARB
-	.globl glProgramParameter4dvNV ; .type glProgramParameter4dvNV,#function ; glProgramParameter4dvNV = glProgramEnvParameter4dvARB
-	.globl glProgramParameter4fNV ; .type glProgramParameter4fNV,#function ; glProgramParameter4fNV = glProgramEnvParameter4fARB
-	.globl glProgramParameter4fvNV ; .type glProgramParameter4fvNV,#function ; glProgramParameter4fvNV = glProgramEnvParameter4fvARB
-	.globl glVertexAttrib1d ; .type glVertexAttrib1d,#function ; glVertexAttrib1d = glVertexAttrib1dARB
-	.globl glVertexAttrib1dv ; .type glVertexAttrib1dv,#function ; glVertexAttrib1dv = glVertexAttrib1dvARB
-	.globl glVertexAttrib1f ; .type glVertexAttrib1f,#function ; glVertexAttrib1f = glVertexAttrib1fARB
-	.globl glVertexAttrib1fv ; .type glVertexAttrib1fv,#function ; glVertexAttrib1fv = glVertexAttrib1fvARB
-	.globl glVertexAttrib1s ; .type glVertexAttrib1s,#function ; glVertexAttrib1s = glVertexAttrib1sARB
-	.globl glVertexAttrib1sv ; .type glVertexAttrib1sv,#function ; glVertexAttrib1sv = glVertexAttrib1svARB
-	.globl glVertexAttrib2d ; .type glVertexAttrib2d,#function ; glVertexAttrib2d = glVertexAttrib2dARB
-	.globl glVertexAttrib2dv ; .type glVertexAttrib2dv,#function ; glVertexAttrib2dv = glVertexAttrib2dvARB
-	.globl glVertexAttrib2f ; .type glVertexAttrib2f,#function ; glVertexAttrib2f = glVertexAttrib2fARB
-	.globl glVertexAttrib2fv ; .type glVertexAttrib2fv,#function ; glVertexAttrib2fv = glVertexAttrib2fvARB
-	.globl glVertexAttrib2s ; .type glVertexAttrib2s,#function ; glVertexAttrib2s = glVertexAttrib2sARB
-	.globl glVertexAttrib2sv ; .type glVertexAttrib2sv,#function ; glVertexAttrib2sv = glVertexAttrib2svARB
-	.globl glVertexAttrib3d ; .type glVertexAttrib3d,#function ; glVertexAttrib3d = glVertexAttrib3dARB
-	.globl glVertexAttrib3dv ; .type glVertexAttrib3dv,#function ; glVertexAttrib3dv = glVertexAttrib3dvARB
-	.globl glVertexAttrib3f ; .type glVertexAttrib3f,#function ; glVertexAttrib3f = glVertexAttrib3fARB
-	.globl glVertexAttrib3fv ; .type glVertexAttrib3fv,#function ; glVertexAttrib3fv = glVertexAttrib3fvARB
-	.globl glVertexAttrib3s ; .type glVertexAttrib3s,#function ; glVertexAttrib3s = glVertexAttrib3sARB
-	.globl glVertexAttrib3sv ; .type glVertexAttrib3sv,#function ; glVertexAttrib3sv = glVertexAttrib3svARB
-	.globl glVertexAttrib4Nbv ; .type glVertexAttrib4Nbv,#function ; glVertexAttrib4Nbv = glVertexAttrib4NbvARB
-	.globl glVertexAttrib4Niv ; .type glVertexAttrib4Niv,#function ; glVertexAttrib4Niv = glVertexAttrib4NivARB
-	.globl glVertexAttrib4Nsv ; .type glVertexAttrib4Nsv,#function ; glVertexAttrib4Nsv = glVertexAttrib4NsvARB
-	.globl glVertexAttrib4Nub ; .type glVertexAttrib4Nub,#function ; glVertexAttrib4Nub = glVertexAttrib4NubARB
-	.globl glVertexAttrib4Nubv ; .type glVertexAttrib4Nubv,#function ; glVertexAttrib4Nubv = glVertexAttrib4NubvARB
-	.globl glVertexAttrib4Nuiv ; .type glVertexAttrib4Nuiv,#function ; glVertexAttrib4Nuiv = glVertexAttrib4NuivARB
-	.globl glVertexAttrib4Nusv ; .type glVertexAttrib4Nusv,#function ; glVertexAttrib4Nusv = glVertexAttrib4NusvARB
-	.globl glVertexAttrib4bv ; .type glVertexAttrib4bv,#function ; glVertexAttrib4bv = glVertexAttrib4bvARB
-	.globl glVertexAttrib4d ; .type glVertexAttrib4d,#function ; glVertexAttrib4d = glVertexAttrib4dARB
-	.globl glVertexAttrib4dv ; .type glVertexAttrib4dv,#function ; glVertexAttrib4dv = glVertexAttrib4dvARB
-	.globl glVertexAttrib4f ; .type glVertexAttrib4f,#function ; glVertexAttrib4f = glVertexAttrib4fARB
-	.globl glVertexAttrib4fv ; .type glVertexAttrib4fv,#function ; glVertexAttrib4fv = glVertexAttrib4fvARB
-	.globl glVertexAttrib4iv ; .type glVertexAttrib4iv,#function ; glVertexAttrib4iv = glVertexAttrib4ivARB
-	.globl glVertexAttrib4s ; .type glVertexAttrib4s,#function ; glVertexAttrib4s = glVertexAttrib4sARB
-	.globl glVertexAttrib4sv ; .type glVertexAttrib4sv,#function ; glVertexAttrib4sv = glVertexAttrib4svARB
-	.globl glVertexAttrib4ubv ; .type glVertexAttrib4ubv,#function ; glVertexAttrib4ubv = glVertexAttrib4ubvARB
-	.globl glVertexAttrib4uiv ; .type glVertexAttrib4uiv,#function ; glVertexAttrib4uiv = glVertexAttrib4uivARB
-	.globl glVertexAttrib4usv ; .type glVertexAttrib4usv,#function ; glVertexAttrib4usv = glVertexAttrib4usvARB
-	.globl glVertexAttribPointer ; .type glVertexAttribPointer,#function ; glVertexAttribPointer = glVertexAttribPointerARB
-	.globl glBindBuffer ; .type glBindBuffer,#function ; glBindBuffer = glBindBufferARB
-	.globl glBufferData ; .type glBufferData,#function ; glBufferData = glBufferDataARB
-	.globl glBufferSubData ; .type glBufferSubData,#function ; glBufferSubData = glBufferSubDataARB
-	.globl glDeleteBuffers ; .type glDeleteBuffers,#function ; glDeleteBuffers = glDeleteBuffersARB
-	.globl glGenBuffers ; .type glGenBuffers,#function ; glGenBuffers = glGenBuffersARB
-	.globl glGetBufferParameteriv ; .type glGetBufferParameteriv,#function ; glGetBufferParameteriv = glGetBufferParameterivARB
-	.globl glGetBufferPointerv ; .type glGetBufferPointerv,#function ; glGetBufferPointerv = glGetBufferPointervARB
-	.globl glGetBufferSubData ; .type glGetBufferSubData,#function ; glGetBufferSubData = glGetBufferSubDataARB
-	.globl glIsBuffer ; .type glIsBuffer,#function ; glIsBuffer = glIsBufferARB
-	.globl glMapBuffer ; .type glMapBuffer,#function ; glMapBuffer = glMapBufferARB
-	.globl glUnmapBuffer ; .type glUnmapBuffer,#function ; glUnmapBuffer = glUnmapBufferARB
-	.globl glBeginQuery ; .type glBeginQuery,#function ; glBeginQuery = glBeginQueryARB
-	.globl glDeleteQueries ; .type glDeleteQueries,#function ; glDeleteQueries = glDeleteQueriesARB
-	.globl glEndQuery ; .type glEndQuery,#function ; glEndQuery = glEndQueryARB
-	.globl glGenQueries ; .type glGenQueries,#function ; glGenQueries = glGenQueriesARB
-	.globl glGetQueryObjectiv ; .type glGetQueryObjectiv,#function ; glGetQueryObjectiv = glGetQueryObjectivARB
-	.globl glGetQueryObjectuiv ; .type glGetQueryObjectuiv,#function ; glGetQueryObjectuiv = glGetQueryObjectuivARB
-	.globl glGetQueryiv ; .type glGetQueryiv,#function ; glGetQueryiv = glGetQueryivARB
-	.globl glIsQuery ; .type glIsQuery,#function ; glIsQuery = glIsQueryARB
-	.globl glCompileShader ; .type glCompileShader,#function ; glCompileShader = glCompileShaderARB
-	.globl glGetActiveUniform ; .type glGetActiveUniform,#function ; glGetActiveUniform = glGetActiveUniformARB
-	.globl glGetShaderSource ; .type glGetShaderSource,#function ; glGetShaderSource = glGetShaderSourceARB
-	.globl glGetUniformLocation ; .type glGetUniformLocation,#function ; glGetUniformLocation = glGetUniformLocationARB
-	.globl glGetUniformfv ; .type glGetUniformfv,#function ; glGetUniformfv = glGetUniformfvARB
-	.globl glGetUniformiv ; .type glGetUniformiv,#function ; glGetUniformiv = glGetUniformivARB
-	.globl glLinkProgram ; .type glLinkProgram,#function ; glLinkProgram = glLinkProgramARB
-	.globl glShaderSource ; .type glShaderSource,#function ; glShaderSource = glShaderSourceARB
-	.globl glUniform1f ; .type glUniform1f,#function ; glUniform1f = glUniform1fARB
-	.globl glUniform1fv ; .type glUniform1fv,#function ; glUniform1fv = glUniform1fvARB
-	.globl glUniform1i ; .type glUniform1i,#function ; glUniform1i = glUniform1iARB
-	.globl glUniform1iv ; .type glUniform1iv,#function ; glUniform1iv = glUniform1ivARB
-	.globl glUniform2f ; .type glUniform2f,#function ; glUniform2f = glUniform2fARB
-	.globl glUniform2fv ; .type glUniform2fv,#function ; glUniform2fv = glUniform2fvARB
-	.globl glUniform2i ; .type glUniform2i,#function ; glUniform2i = glUniform2iARB
-	.globl glUniform2iv ; .type glUniform2iv,#function ; glUniform2iv = glUniform2ivARB
-	.globl glUniform3f ; .type glUniform3f,#function ; glUniform3f = glUniform3fARB
-	.globl glUniform3fv ; .type glUniform3fv,#function ; glUniform3fv = glUniform3fvARB
-	.globl glUniform3i ; .type glUniform3i,#function ; glUniform3i = glUniform3iARB
-	.globl glUniform3iv ; .type glUniform3iv,#function ; glUniform3iv = glUniform3ivARB
-	.globl glUniform4f ; .type glUniform4f,#function ; glUniform4f = glUniform4fARB
-	.globl glUniform4fv ; .type glUniform4fv,#function ; glUniform4fv = glUniform4fvARB
-	.globl glUniform4i ; .type glUniform4i,#function ; glUniform4i = glUniform4iARB
-	.globl glUniform4iv ; .type glUniform4iv,#function ; glUniform4iv = glUniform4ivARB
-	.globl glUniformMatrix2fv ; .type glUniformMatrix2fv,#function ; glUniformMatrix2fv = glUniformMatrix2fvARB
-	.globl glUniformMatrix3fv ; .type glUniformMatrix3fv,#function ; glUniformMatrix3fv = glUniformMatrix3fvARB
-	.globl glUniformMatrix4fv ; .type glUniformMatrix4fv,#function ; glUniformMatrix4fv = glUniformMatrix4fvARB
-	.globl glUseProgram ; .type glUseProgram,#function ; glUseProgram = glUseProgramObjectARB
-	.globl glValidateProgram ; .type glValidateProgram,#function ; glValidateProgram = glValidateProgramARB
-	.globl glBindAttribLocation ; .type glBindAttribLocation,#function ; glBindAttribLocation = glBindAttribLocationARB
-	.globl glGetActiveAttrib ; .type glGetActiveAttrib,#function ; glGetActiveAttrib = glGetActiveAttribARB
-	.globl glGetAttribLocation ; .type glGetAttribLocation,#function ; glGetAttribLocation = glGetAttribLocationARB
-	.globl glDrawBuffers ; .type glDrawBuffers,#function ; glDrawBuffers = glDrawBuffersARB
-	.globl glDrawBuffersATI ; .type glDrawBuffersATI,#function ; glDrawBuffersATI = glDrawBuffersARB
-	.globl glPointParameterf ; .type glPointParameterf,#function ; glPointParameterf = glPointParameterfEXT
-	.globl glPointParameterfARB ; .type glPointParameterfARB,#function ; glPointParameterfARB = glPointParameterfEXT
-	.globl glPointParameterfv ; .type glPointParameterfv,#function ; glPointParameterfv = glPointParameterfvEXT
-	.globl glPointParameterfvARB ; .type glPointParameterfvARB,#function ; glPointParameterfvARB = glPointParameterfvEXT
-	.globl glSecondaryColor3b ; .type glSecondaryColor3b,#function ; glSecondaryColor3b = glSecondaryColor3bEXT
-	.globl glSecondaryColor3bv ; .type glSecondaryColor3bv,#function ; glSecondaryColor3bv = glSecondaryColor3bvEXT
-	.globl glSecondaryColor3d ; .type glSecondaryColor3d,#function ; glSecondaryColor3d = glSecondaryColor3dEXT
-	.globl glSecondaryColor3dv ; .type glSecondaryColor3dv,#function ; glSecondaryColor3dv = glSecondaryColor3dvEXT
-	.globl glSecondaryColor3f ; .type glSecondaryColor3f,#function ; glSecondaryColor3f = glSecondaryColor3fEXT
-	.globl glSecondaryColor3fv ; .type glSecondaryColor3fv,#function ; glSecondaryColor3fv = glSecondaryColor3fvEXT
-	.globl glSecondaryColor3i ; .type glSecondaryColor3i,#function ; glSecondaryColor3i = glSecondaryColor3iEXT
-	.globl glSecondaryColor3iv ; .type glSecondaryColor3iv,#function ; glSecondaryColor3iv = glSecondaryColor3ivEXT
-	.globl glSecondaryColor3s ; .type glSecondaryColor3s,#function ; glSecondaryColor3s = glSecondaryColor3sEXT
-	.globl glSecondaryColor3sv ; .type glSecondaryColor3sv,#function ; glSecondaryColor3sv = glSecondaryColor3svEXT
-	.globl glSecondaryColor3ub ; .type glSecondaryColor3ub,#function ; glSecondaryColor3ub = glSecondaryColor3ubEXT
-	.globl glSecondaryColor3ubv ; .type glSecondaryColor3ubv,#function ; glSecondaryColor3ubv = glSecondaryColor3ubvEXT
-	.globl glSecondaryColor3ui ; .type glSecondaryColor3ui,#function ; glSecondaryColor3ui = glSecondaryColor3uiEXT
-	.globl glSecondaryColor3uiv ; .type glSecondaryColor3uiv,#function ; glSecondaryColor3uiv = glSecondaryColor3uivEXT
-	.globl glSecondaryColor3us ; .type glSecondaryColor3us,#function ; glSecondaryColor3us = glSecondaryColor3usEXT
-	.globl glSecondaryColor3usv ; .type glSecondaryColor3usv,#function ; glSecondaryColor3usv = glSecondaryColor3usvEXT
-	.globl glSecondaryColorPointer ; .type glSecondaryColorPointer,#function ; glSecondaryColorPointer = glSecondaryColorPointerEXT
-	.globl glMultiDrawArrays ; .type glMultiDrawArrays,#function ; glMultiDrawArrays = glMultiDrawArraysEXT
-	.globl glMultiDrawElements ; .type glMultiDrawElements,#function ; glMultiDrawElements = glMultiDrawElementsEXT
-	.globl glFogCoordPointer ; .type glFogCoordPointer,#function ; glFogCoordPointer = glFogCoordPointerEXT
-	.globl glFogCoordd ; .type glFogCoordd,#function ; glFogCoordd = glFogCoorddEXT
-	.globl glFogCoorddv ; .type glFogCoorddv,#function ; glFogCoorddv = glFogCoorddvEXT
-	.globl glFogCoordf ; .type glFogCoordf,#function ; glFogCoordf = glFogCoordfEXT
-	.globl glFogCoordfv ; .type glFogCoordfv,#function ; glFogCoordfv = glFogCoordfvEXT
-	.globl glBlendFuncSeparate ; .type glBlendFuncSeparate,#function ; glBlendFuncSeparate = glBlendFuncSeparateEXT
-	.globl glWindowPos2d ; .type glWindowPos2d,#function ; glWindowPos2d = glWindowPos2dMESA
-	.globl glWindowPos2dARB ; .type glWindowPos2dARB,#function ; glWindowPos2dARB = glWindowPos2dMESA
-	.globl glWindowPos2dv ; .type glWindowPos2dv,#function ; glWindowPos2dv = glWindowPos2dvMESA
-	.globl glWindowPos2dvARB ; .type glWindowPos2dvARB,#function ; glWindowPos2dvARB = glWindowPos2dvMESA
-	.globl glWindowPos2f ; .type glWindowPos2f,#function ; glWindowPos2f = glWindowPos2fMESA
-	.globl glWindowPos2fARB ; .type glWindowPos2fARB,#function ; glWindowPos2fARB = glWindowPos2fMESA
-	.globl glWindowPos2fv ; .type glWindowPos2fv,#function ; glWindowPos2fv = glWindowPos2fvMESA
-	.globl glWindowPos2fvARB ; .type glWindowPos2fvARB,#function ; glWindowPos2fvARB = glWindowPos2fvMESA
-	.globl glWindowPos2i ; .type glWindowPos2i,#function ; glWindowPos2i = glWindowPos2iMESA
-	.globl glWindowPos2iARB ; .type glWindowPos2iARB,#function ; glWindowPos2iARB = glWindowPos2iMESA
-	.globl glWindowPos2iv ; .type glWindowPos2iv,#function ; glWindowPos2iv = glWindowPos2ivMESA
-	.globl glWindowPos2ivARB ; .type glWindowPos2ivARB,#function ; glWindowPos2ivARB = glWindowPos2ivMESA
-	.globl glWindowPos2s ; .type glWindowPos2s,#function ; glWindowPos2s = glWindowPos2sMESA
-	.globl glWindowPos2sARB ; .type glWindowPos2sARB,#function ; glWindowPos2sARB = glWindowPos2sMESA
-	.globl glWindowPos2sv ; .type glWindowPos2sv,#function ; glWindowPos2sv = glWindowPos2svMESA
-	.globl glWindowPos2svARB ; .type glWindowPos2svARB,#function ; glWindowPos2svARB = glWindowPos2svMESA
-	.globl glWindowPos3d ; .type glWindowPos3d,#function ; glWindowPos3d = glWindowPos3dMESA
-	.globl glWindowPos3dARB ; .type glWindowPos3dARB,#function ; glWindowPos3dARB = glWindowPos3dMESA
-	.globl glWindowPos3dv ; .type glWindowPos3dv,#function ; glWindowPos3dv = glWindowPos3dvMESA
-	.globl glWindowPos3dvARB ; .type glWindowPos3dvARB,#function ; glWindowPos3dvARB = glWindowPos3dvMESA
-	.globl glWindowPos3f ; .type glWindowPos3f,#function ; glWindowPos3f = glWindowPos3fMESA
-	.globl glWindowPos3fARB ; .type glWindowPos3fARB,#function ; glWindowPos3fARB = glWindowPos3fMESA
-	.globl glWindowPos3fv ; .type glWindowPos3fv,#function ; glWindowPos3fv = glWindowPos3fvMESA
-	.globl glWindowPos3fvARB ; .type glWindowPos3fvARB,#function ; glWindowPos3fvARB = glWindowPos3fvMESA
-	.globl glWindowPos3i ; .type glWindowPos3i,#function ; glWindowPos3i = glWindowPos3iMESA
-	.globl glWindowPos3iARB ; .type glWindowPos3iARB,#function ; glWindowPos3iARB = glWindowPos3iMESA
-	.globl glWindowPos3iv ; .type glWindowPos3iv,#function ; glWindowPos3iv = glWindowPos3ivMESA
-	.globl glWindowPos3ivARB ; .type glWindowPos3ivARB,#function ; glWindowPos3ivARB = glWindowPos3ivMESA
-	.globl glWindowPos3s ; .type glWindowPos3s,#function ; glWindowPos3s = glWindowPos3sMESA
-	.globl glWindowPos3sARB ; .type glWindowPos3sARB,#function ; glWindowPos3sARB = glWindowPos3sMESA
-	.globl glWindowPos3sv ; .type glWindowPos3sv,#function ; glWindowPos3sv = glWindowPos3svMESA
-	.globl glWindowPos3svARB ; .type glWindowPos3svARB,#function ; glWindowPos3svARB = glWindowPos3svMESA
-	.globl glBindProgramARB ; .type glBindProgramARB,#function ; glBindProgramARB = glBindProgramNV
-	.globl glDeleteProgramsARB ; .type glDeleteProgramsARB,#function ; glDeleteProgramsARB = glDeleteProgramsNV
-	.globl glGenProgramsARB ; .type glGenProgramsARB,#function ; glGenProgramsARB = glGenProgramsNV
-	.globl glGetVertexAttribPointerv ; .type glGetVertexAttribPointerv,#function ; glGetVertexAttribPointerv = glGetVertexAttribPointervNV
-	.globl glGetVertexAttribPointervARB ; .type glGetVertexAttribPointervARB,#function ; glGetVertexAttribPointervARB = glGetVertexAttribPointervNV
-	.globl glIsProgramARB ; .type glIsProgramARB,#function ; glIsProgramARB = glIsProgramNV
-	.globl glPointParameteri ; .type glPointParameteri,#function ; glPointParameteri = glPointParameteriNV
-	.globl glPointParameteriv ; .type glPointParameteriv,#function ; glPointParameteriv = glPointParameterivNV
-	.globl glBlendEquationSeparate ; .type glBlendEquationSeparate,#function ; glBlendEquationSeparate = glBlendEquationSeparateEXT
-	.globl glBindFramebuffer ; .type glBindFramebuffer,#function ; glBindFramebuffer = glBindFramebufferEXT
-	.globl glBindRenderbuffer ; .type glBindRenderbuffer,#function ; glBindRenderbuffer = glBindRenderbufferEXT
-	.globl glCheckFramebufferStatus ; .type glCheckFramebufferStatus,#function ; glCheckFramebufferStatus = glCheckFramebufferStatusEXT
-	.globl glDeleteFramebuffers ; .type glDeleteFramebuffers,#function ; glDeleteFramebuffers = glDeleteFramebuffersEXT
-	.globl glDeleteRenderbuffers ; .type glDeleteRenderbuffers,#function ; glDeleteRenderbuffers = glDeleteRenderbuffersEXT
-	.globl glFramebufferRenderbuffer ; .type glFramebufferRenderbuffer,#function ; glFramebufferRenderbuffer = glFramebufferRenderbufferEXT
-	.globl glFramebufferTexture1D ; .type glFramebufferTexture1D,#function ; glFramebufferTexture1D = glFramebufferTexture1DEXT
-	.globl glFramebufferTexture2D ; .type glFramebufferTexture2D,#function ; glFramebufferTexture2D = glFramebufferTexture2DEXT
-	.globl glFramebufferTexture3D ; .type glFramebufferTexture3D,#function ; glFramebufferTexture3D = glFramebufferTexture3DEXT
-	.globl glGenFramebuffers ; .type glGenFramebuffers,#function ; glGenFramebuffers = glGenFramebuffersEXT
-	.globl glGenRenderbuffers ; .type glGenRenderbuffers,#function ; glGenRenderbuffers = glGenRenderbuffersEXT
-	.globl glGenerateMipmap ; .type glGenerateMipmap,#function ; glGenerateMipmap = glGenerateMipmapEXT
-	.globl glGetFramebufferAttachmentParameteriv ; .type glGetFramebufferAttachmentParameteriv,#function ; glGetFramebufferAttachmentParameteriv = glGetFramebufferAttachmentParameterivEXT
-	.globl glGetRenderbufferParameteriv ; .type glGetRenderbufferParameteriv,#function ; glGetRenderbufferParameteriv = glGetRenderbufferParameterivEXT
-	.globl glIsFramebuffer ; .type glIsFramebuffer,#function ; glIsFramebuffer = glIsFramebufferEXT
-	.globl glIsRenderbuffer ; .type glIsRenderbuffer,#function ; glIsRenderbuffer = glIsRenderbufferEXT
-	.globl glRenderbufferStorage ; .type glRenderbufferStorage,#function ; glRenderbufferStorage = glRenderbufferStorageEXT
-	.globl glBlitFramebuffer ; .type glBlitFramebuffer,#function ; glBlitFramebuffer = glBlitFramebufferEXT
-	.globl glFramebufferTextureLayer ; .type glFramebufferTextureLayer,#function ; glFramebufferTextureLayer = glFramebufferTextureLayerEXT
+	GL_STUB_ALIAS(glSeparableFilter2DEXT, glSeparableFilter2D)
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetHistogramEXT, glGetHistogram)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetHistogramParameterfvEXT, glGetHistogramParameterfv)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetHistogramParameterivEXT, glGetHistogramParameteriv)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetMinmaxEXT, glGetMinmax)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetMinmaxParameterfvEXT, glGetMinmaxParameterfv)
+#endif
+#ifndef GLX_INDIRECT_RENDERING
+	GL_STUB_ALIAS(glGetMinmaxParameterivEXT, glGetMinmaxParameteriv)
+#endif
+	GL_STUB_ALIAS(glHistogramEXT, glHistogram)
+	GL_STUB_ALIAS(glMinmaxEXT, glMinmax)
+	GL_STUB_ALIAS(glResetHistogramEXT, glResetHistogram)
+	GL_STUB_ALIAS(glResetMinmaxEXT, glResetMinmax)
+	GL_STUB_ALIAS(glTexImage3DEXT, glTexImage3D)
+	GL_STUB_ALIAS(glTexSubImage3DEXT, glTexSubImage3D)
+	GL_STUB_ALIAS(glCopyTexSubImage3DEXT, glCopyTexSubImage3D)
+	GL_STUB_ALIAS(glActiveTexture, glActiveTextureARB)
+	GL_STUB_ALIAS(glClientActiveTexture, glClientActiveTextureARB)
+	GL_STUB_ALIAS(glMultiTexCoord1d, glMultiTexCoord1dARB)
+	GL_STUB_ALIAS(glMultiTexCoord1dv, glMultiTexCoord1dvARB)
+	GL_STUB_ALIAS(glMultiTexCoord1f, glMultiTexCoord1fARB)
+	GL_STUB_ALIAS(glMultiTexCoord1fv, glMultiTexCoord1fvARB)
+	GL_STUB_ALIAS(glMultiTexCoord1i, glMultiTexCoord1iARB)
+	GL_STUB_ALIAS(glMultiTexCoord1iv, glMultiTexCoord1ivARB)
+	GL_STUB_ALIAS(glMultiTexCoord1s, glMultiTexCoord1sARB)
+	GL_STUB_ALIAS(glMultiTexCoord1sv, glMultiTexCoord1svARB)
+	GL_STUB_ALIAS(glMultiTexCoord2d, glMultiTexCoord2dARB)
+	GL_STUB_ALIAS(glMultiTexCoord2dv, glMultiTexCoord2dvARB)
+	GL_STUB_ALIAS(glMultiTexCoord2f, glMultiTexCoord2fARB)
+	GL_STUB_ALIAS(glMultiTexCoord2fv, glMultiTexCoord2fvARB)
+	GL_STUB_ALIAS(glMultiTexCoord2i, glMultiTexCoord2iARB)
+	GL_STUB_ALIAS(glMultiTexCoord2iv, glMultiTexCoord2ivARB)
+	GL_STUB_ALIAS(glMultiTexCoord2s, glMultiTexCoord2sARB)
+	GL_STUB_ALIAS(glMultiTexCoord2sv, glMultiTexCoord2svARB)
+	GL_STUB_ALIAS(glMultiTexCoord3d, glMultiTexCoord3dARB)
+	GL_STUB_ALIAS(glMultiTexCoord3dv, glMultiTexCoord3dvARB)
+	GL_STUB_ALIAS(glMultiTexCoord3f, glMultiTexCoord3fARB)
+	GL_STUB_ALIAS(glMultiTexCoord3fv, glMultiTexCoord3fvARB)
+	GL_STUB_ALIAS(glMultiTexCoord3i, glMultiTexCoord3iARB)
+	GL_STUB_ALIAS(glMultiTexCoord3iv, glMultiTexCoord3ivARB)
+	GL_STUB_ALIAS(glMultiTexCoord3s, glMultiTexCoord3sARB)
+	GL_STUB_ALIAS(glMultiTexCoord3sv, glMultiTexCoord3svARB)
+	GL_STUB_ALIAS(glMultiTexCoord4d, glMultiTexCoord4dARB)
+	GL_STUB_ALIAS(glMultiTexCoord4dv, glMultiTexCoord4dvARB)
+	GL_STUB_ALIAS(glMultiTexCoord4f, glMultiTexCoord4fARB)
+	GL_STUB_ALIAS(glMultiTexCoord4fv, glMultiTexCoord4fvARB)
+	GL_STUB_ALIAS(glMultiTexCoord4i, glMultiTexCoord4iARB)
+	GL_STUB_ALIAS(glMultiTexCoord4iv, glMultiTexCoord4ivARB)
+	GL_STUB_ALIAS(glMultiTexCoord4s, glMultiTexCoord4sARB)
+	GL_STUB_ALIAS(glMultiTexCoord4sv, glMultiTexCoord4svARB)
+	GL_STUB_ALIAS(glStencilOpSeparateATI, glStencilOpSeparate)
+	GL_STUB_ALIAS(glLoadTransposeMatrixd, glLoadTransposeMatrixdARB)
+	GL_STUB_ALIAS(glLoadTransposeMatrixf, glLoadTransposeMatrixfARB)
+	GL_STUB_ALIAS(glMultTransposeMatrixd, glMultTransposeMatrixdARB)
+	GL_STUB_ALIAS(glMultTransposeMatrixf, glMultTransposeMatrixfARB)
+	GL_STUB_ALIAS(glSampleCoverage, glSampleCoverageARB)
+	GL_STUB_ALIAS(glCompressedTexImage1D, glCompressedTexImage1DARB)
+	GL_STUB_ALIAS(glCompressedTexImage2D, glCompressedTexImage2DARB)
+	GL_STUB_ALIAS(glCompressedTexImage3D, glCompressedTexImage3DARB)
+	GL_STUB_ALIAS(glCompressedTexSubImage1D, glCompressedTexSubImage1DARB)
+	GL_STUB_ALIAS(glCompressedTexSubImage2D, glCompressedTexSubImage2DARB)
+	GL_STUB_ALIAS(glCompressedTexSubImage3D, glCompressedTexSubImage3DARB)
+	GL_STUB_ALIAS(glGetCompressedTexImage, glGetCompressedTexImageARB)
+	GL_STUB_ALIAS(glDisableVertexAttribArray, glDisableVertexAttribArrayARB)
+	GL_STUB_ALIAS(glEnableVertexAttribArray, glEnableVertexAttribArrayARB)
+	GL_STUB_ALIAS(glGetVertexAttribdv, glGetVertexAttribdvARB)
+	GL_STUB_ALIAS(glGetVertexAttribfv, glGetVertexAttribfvARB)
+	GL_STUB_ALIAS(glGetVertexAttribiv, glGetVertexAttribivARB)
+	GL_STUB_ALIAS(glProgramParameter4dNV, glProgramEnvParameter4dARB)
+	GL_STUB_ALIAS(glProgramParameter4dvNV, glProgramEnvParameter4dvARB)
+	GL_STUB_ALIAS(glProgramParameter4fNV, glProgramEnvParameter4fARB)
+	GL_STUB_ALIAS(glProgramParameter4fvNV, glProgramEnvParameter4fvARB)
+	GL_STUB_ALIAS(glVertexAttrib1d, glVertexAttrib1dARB)
+	GL_STUB_ALIAS(glVertexAttrib1dv, glVertexAttrib1dvARB)
+	GL_STUB_ALIAS(glVertexAttrib1f, glVertexAttrib1fARB)
+	GL_STUB_ALIAS(glVertexAttrib1fv, glVertexAttrib1fvARB)
+	GL_STUB_ALIAS(glVertexAttrib1s, glVertexAttrib1sARB)
+	GL_STUB_ALIAS(glVertexAttrib1sv, glVertexAttrib1svARB)
+	GL_STUB_ALIAS(glVertexAttrib2d, glVertexAttrib2dARB)
+	GL_STUB_ALIAS(glVertexAttrib2dv, glVertexAttrib2dvARB)
+	GL_STUB_ALIAS(glVertexAttrib2f, glVertexAttrib2fARB)
+	GL_STUB_ALIAS(glVertexAttrib2fv, glVertexAttrib2fvARB)
+	GL_STUB_ALIAS(glVertexAttrib2s, glVertexAttrib2sARB)
+	GL_STUB_ALIAS(glVertexAttrib2sv, glVertexAttrib2svARB)
+	GL_STUB_ALIAS(glVertexAttrib3d, glVertexAttrib3dARB)
+	GL_STUB_ALIAS(glVertexAttrib3dv, glVertexAttrib3dvARB)
+	GL_STUB_ALIAS(glVertexAttrib3f, glVertexAttrib3fARB)
+	GL_STUB_ALIAS(glVertexAttrib3fv, glVertexAttrib3fvARB)
+	GL_STUB_ALIAS(glVertexAttrib3s, glVertexAttrib3sARB)
+	GL_STUB_ALIAS(glVertexAttrib3sv, glVertexAttrib3svARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nbv, glVertexAttrib4NbvARB)
+	GL_STUB_ALIAS(glVertexAttrib4Niv, glVertexAttrib4NivARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nsv, glVertexAttrib4NsvARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nub, glVertexAttrib4NubARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nubv, glVertexAttrib4NubvARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nuiv, glVertexAttrib4NuivARB)
+	GL_STUB_ALIAS(glVertexAttrib4Nusv, glVertexAttrib4NusvARB)
+	GL_STUB_ALIAS(glVertexAttrib4bv, glVertexAttrib4bvARB)
+	GL_STUB_ALIAS(glVertexAttrib4d, glVertexAttrib4dARB)
+	GL_STUB_ALIAS(glVertexAttrib4dv, glVertexAttrib4dvARB)
+	GL_STUB_ALIAS(glVertexAttrib4f, glVertexAttrib4fARB)
+	GL_STUB_ALIAS(glVertexAttrib4fv, glVertexAttrib4fvARB)
+	GL_STUB_ALIAS(glVertexAttrib4iv, glVertexAttrib4ivARB)
+	GL_STUB_ALIAS(glVertexAttrib4s, glVertexAttrib4sARB)
+	GL_STUB_ALIAS(glVertexAttrib4sv, glVertexAttrib4svARB)
+	GL_STUB_ALIAS(glVertexAttrib4ubv, glVertexAttrib4ubvARB)
+	GL_STUB_ALIAS(glVertexAttrib4uiv, glVertexAttrib4uivARB)
+	GL_STUB_ALIAS(glVertexAttrib4usv, glVertexAttrib4usvARB)
+	GL_STUB_ALIAS(glVertexAttribPointer, glVertexAttribPointerARB)
+	GL_STUB_ALIAS(glBindBuffer, glBindBufferARB)
+	GL_STUB_ALIAS(glBufferData, glBufferDataARB)
+	GL_STUB_ALIAS(glBufferSubData, glBufferSubDataARB)
+	GL_STUB_ALIAS(glDeleteBuffers, glDeleteBuffersARB)
+	GL_STUB_ALIAS(glGenBuffers, glGenBuffersARB)
+	GL_STUB_ALIAS(glGetBufferParameteriv, glGetBufferParameterivARB)
+	GL_STUB_ALIAS(glGetBufferPointerv, glGetBufferPointervARB)
+	GL_STUB_ALIAS(glGetBufferSubData, glGetBufferSubDataARB)
+	GL_STUB_ALIAS(glIsBuffer, glIsBufferARB)
+	GL_STUB_ALIAS(glMapBuffer, glMapBufferARB)
+	GL_STUB_ALIAS(glUnmapBuffer, glUnmapBufferARB)
+	GL_STUB_ALIAS(glBeginQuery, glBeginQueryARB)
+	GL_STUB_ALIAS(glDeleteQueries, glDeleteQueriesARB)
+	GL_STUB_ALIAS(glEndQuery, glEndQueryARB)
+	GL_STUB_ALIAS(glGenQueries, glGenQueriesARB)
+	GL_STUB_ALIAS(glGetQueryObjectiv, glGetQueryObjectivARB)
+	GL_STUB_ALIAS(glGetQueryObjectuiv, glGetQueryObjectuivARB)
+	GL_STUB_ALIAS(glGetQueryiv, glGetQueryivARB)
+	GL_STUB_ALIAS(glIsQuery, glIsQueryARB)
+	GL_STUB_ALIAS(glCompileShader, glCompileShaderARB)
+	GL_STUB_ALIAS(glGetActiveUniform, glGetActiveUniformARB)
+	GL_STUB_ALIAS(glGetShaderSource, glGetShaderSourceARB)
+	GL_STUB_ALIAS(glGetUniformLocation, glGetUniformLocationARB)
+	GL_STUB_ALIAS(glGetUniformfv, glGetUniformfvARB)
+	GL_STUB_ALIAS(glGetUniformiv, glGetUniformivARB)
+	GL_STUB_ALIAS(glLinkProgram, glLinkProgramARB)
+	GL_STUB_ALIAS(glShaderSource, glShaderSourceARB)
+	GL_STUB_ALIAS(glUniform1f, glUniform1fARB)
+	GL_STUB_ALIAS(glUniform1fv, glUniform1fvARB)
+	GL_STUB_ALIAS(glUniform1i, glUniform1iARB)
+	GL_STUB_ALIAS(glUniform1iv, glUniform1ivARB)
+	GL_STUB_ALIAS(glUniform2f, glUniform2fARB)
+	GL_STUB_ALIAS(glUniform2fv, glUniform2fvARB)
+	GL_STUB_ALIAS(glUniform2i, glUniform2iARB)
+	GL_STUB_ALIAS(glUniform2iv, glUniform2ivARB)
+	GL_STUB_ALIAS(glUniform3f, glUniform3fARB)
+	GL_STUB_ALIAS(glUniform3fv, glUniform3fvARB)
+	GL_STUB_ALIAS(glUniform3i, glUniform3iARB)
+	GL_STUB_ALIAS(glUniform3iv, glUniform3ivARB)
+	GL_STUB_ALIAS(glUniform4f, glUniform4fARB)
+	GL_STUB_ALIAS(glUniform4fv, glUniform4fvARB)
+	GL_STUB_ALIAS(glUniform4i, glUniform4iARB)
+	GL_STUB_ALIAS(glUniform4iv, glUniform4ivARB)
+	GL_STUB_ALIAS(glUniformMatrix2fv, glUniformMatrix2fvARB)
+	GL_STUB_ALIAS(glUniformMatrix3fv, glUniformMatrix3fvARB)
+	GL_STUB_ALIAS(glUniformMatrix4fv, glUniformMatrix4fvARB)
+	GL_STUB_ALIAS(glUseProgram, glUseProgramObjectARB)
+	GL_STUB_ALIAS(glValidateProgram, glValidateProgramARB)
+	GL_STUB_ALIAS(glBindAttribLocation, glBindAttribLocationARB)
+	GL_STUB_ALIAS(glGetActiveAttrib, glGetActiveAttribARB)
+	GL_STUB_ALIAS(glGetAttribLocation, glGetAttribLocationARB)
+	GL_STUB_ALIAS(glDrawBuffers, glDrawBuffersARB)
+	GL_STUB_ALIAS(glDrawBuffersATI, glDrawBuffersARB)
+	GL_STUB_ALIAS(glPointParameterf, glPointParameterfEXT)
+	GL_STUB_ALIAS(glPointParameterfARB, glPointParameterfEXT)
+	GL_STUB_ALIAS(glPointParameterfSGIS, glPointParameterfEXT)
+	GL_STUB_ALIAS(glPointParameterfv, glPointParameterfvEXT)
+	GL_STUB_ALIAS(glPointParameterfvARB, glPointParameterfvEXT)
+	GL_STUB_ALIAS(glPointParameterfvSGIS, glPointParameterfvEXT)
+	GL_STUB_ALIAS(glSecondaryColor3b, glSecondaryColor3bEXT)
+	GL_STUB_ALIAS(glSecondaryColor3bv, glSecondaryColor3bvEXT)
+	GL_STUB_ALIAS(glSecondaryColor3d, glSecondaryColor3dEXT)
+	GL_STUB_ALIAS(glSecondaryColor3dv, glSecondaryColor3dvEXT)
+	GL_STUB_ALIAS(glSecondaryColor3f, glSecondaryColor3fEXT)
+	GL_STUB_ALIAS(glSecondaryColor3fv, glSecondaryColor3fvEXT)
+	GL_STUB_ALIAS(glSecondaryColor3i, glSecondaryColor3iEXT)
+	GL_STUB_ALIAS(glSecondaryColor3iv, glSecondaryColor3ivEXT)
+	GL_STUB_ALIAS(glSecondaryColor3s, glSecondaryColor3sEXT)
+	GL_STUB_ALIAS(glSecondaryColor3sv, glSecondaryColor3svEXT)
+	GL_STUB_ALIAS(glSecondaryColor3ub, glSecondaryColor3ubEXT)
+	GL_STUB_ALIAS(glSecondaryColor3ubv, glSecondaryColor3ubvEXT)
+	GL_STUB_ALIAS(glSecondaryColor3ui, glSecondaryColor3uiEXT)
+	GL_STUB_ALIAS(glSecondaryColor3uiv, glSecondaryColor3uivEXT)
+	GL_STUB_ALIAS(glSecondaryColor3us, glSecondaryColor3usEXT)
+	GL_STUB_ALIAS(glSecondaryColor3usv, glSecondaryColor3usvEXT)
+	GL_STUB_ALIAS(glSecondaryColorPointer, glSecondaryColorPointerEXT)
+	GL_STUB_ALIAS(glMultiDrawArrays, glMultiDrawArraysEXT)
+	GL_STUB_ALIAS(glMultiDrawElements, glMultiDrawElementsEXT)
+	GL_STUB_ALIAS(glFogCoordPointer, glFogCoordPointerEXT)
+	GL_STUB_ALIAS(glFogCoordd, glFogCoorddEXT)
+	GL_STUB_ALIAS(glFogCoorddv, glFogCoorddvEXT)
+	GL_STUB_ALIAS(glFogCoordf, glFogCoordfEXT)
+	GL_STUB_ALIAS(glFogCoordfv, glFogCoordfvEXT)
+	GL_STUB_ALIAS(glBlendFuncSeparate, glBlendFuncSeparateEXT)
+	GL_STUB_ALIAS(glBlendFuncSeparateINGR, glBlendFuncSeparateEXT)
+	GL_STUB_ALIAS(glWindowPos2d, glWindowPos2dMESA)
+	GL_STUB_ALIAS(glWindowPos2dARB, glWindowPos2dMESA)
+	GL_STUB_ALIAS(glWindowPos2dv, glWindowPos2dvMESA)
+	GL_STUB_ALIAS(glWindowPos2dvARB, glWindowPos2dvMESA)
+	GL_STUB_ALIAS(glWindowPos2f, glWindowPos2fMESA)
+	GL_STUB_ALIAS(glWindowPos2fARB, glWindowPos2fMESA)
+	GL_STUB_ALIAS(glWindowPos2fv, glWindowPos2fvMESA)
+	GL_STUB_ALIAS(glWindowPos2fvARB, glWindowPos2fvMESA)
+	GL_STUB_ALIAS(glWindowPos2i, glWindowPos2iMESA)
+	GL_STUB_ALIAS(glWindowPos2iARB, glWindowPos2iMESA)
+	GL_STUB_ALIAS(glWindowPos2iv, glWindowPos2ivMESA)
+	GL_STUB_ALIAS(glWindowPos2ivARB, glWindowPos2ivMESA)
+	GL_STUB_ALIAS(glWindowPos2s, glWindowPos2sMESA)
+	GL_STUB_ALIAS(glWindowPos2sARB, glWindowPos2sMESA)
+	GL_STUB_ALIAS(glWindowPos2sv, glWindowPos2svMESA)
+	GL_STUB_ALIAS(glWindowPos2svARB, glWindowPos2svMESA)
+	GL_STUB_ALIAS(glWindowPos3d, glWindowPos3dMESA)
+	GL_STUB_ALIAS(glWindowPos3dARB, glWindowPos3dMESA)
+	GL_STUB_ALIAS(glWindowPos3dv, glWindowPos3dvMESA)
+	GL_STUB_ALIAS(glWindowPos3dvARB, glWindowPos3dvMESA)
+	GL_STUB_ALIAS(glWindowPos3f, glWindowPos3fMESA)
+	GL_STUB_ALIAS(glWindowPos3fARB, glWindowPos3fMESA)
+	GL_STUB_ALIAS(glWindowPos3fv, glWindowPos3fvMESA)
+	GL_STUB_ALIAS(glWindowPos3fvARB, glWindowPos3fvMESA)
+	GL_STUB_ALIAS(glWindowPos3i, glWindowPos3iMESA)
+	GL_STUB_ALIAS(glWindowPos3iARB, glWindowPos3iMESA)
+	GL_STUB_ALIAS(glWindowPos3iv, glWindowPos3ivMESA)
+	GL_STUB_ALIAS(glWindowPos3ivARB, glWindowPos3ivMESA)
+	GL_STUB_ALIAS(glWindowPos3s, glWindowPos3sMESA)
+	GL_STUB_ALIAS(glWindowPos3sARB, glWindowPos3sMESA)
+	GL_STUB_ALIAS(glWindowPos3sv, glWindowPos3svMESA)
+	GL_STUB_ALIAS(glWindowPos3svARB, glWindowPos3svMESA)
+	GL_STUB_ALIAS(glBindProgramARB, glBindProgramNV)
+	GL_STUB_ALIAS(glDeleteProgramsARB, glDeleteProgramsNV)
+	GL_STUB_ALIAS(glGenProgramsARB, glGenProgramsNV)
+	GL_STUB_ALIAS(glGetVertexAttribPointerv, glGetVertexAttribPointervNV)
+	GL_STUB_ALIAS(glGetVertexAttribPointervARB, glGetVertexAttribPointervNV)
+	GL_STUB_ALIAS(glIsProgramARB, glIsProgramNV)
+	GL_STUB_ALIAS(glPointParameteri, glPointParameteriNV)
+	GL_STUB_ALIAS(glPointParameteriv, glPointParameterivNV)
+	GL_STUB_ALIAS(glBindFramebuffer, glBindFramebufferEXT)
+	GL_STUB_ALIAS(glBindRenderbuffer, glBindRenderbufferEXT)
+	GL_STUB_ALIAS(glCheckFramebufferStatus, glCheckFramebufferStatusEXT)
+	GL_STUB_ALIAS(glDeleteFramebuffers, glDeleteFramebuffersEXT)
+	GL_STUB_ALIAS(glDeleteRenderbuffers, glDeleteRenderbuffersEXT)
+	GL_STUB_ALIAS(glFramebufferRenderbuffer, glFramebufferRenderbufferEXT)
+	GL_STUB_ALIAS(glFramebufferTexture1D, glFramebufferTexture1DEXT)
+	GL_STUB_ALIAS(glFramebufferTexture2D, glFramebufferTexture2DEXT)
+	GL_STUB_ALIAS(glFramebufferTexture3D, glFramebufferTexture3DEXT)
+	GL_STUB_ALIAS(glGenFramebuffers, glGenFramebuffersEXT)
+	GL_STUB_ALIAS(glGenRenderbuffers, glGenRenderbuffersEXT)
+	GL_STUB_ALIAS(glGenerateMipmap, glGenerateMipmapEXT)
+	GL_STUB_ALIAS(glGetFramebufferAttachmentParameteriv, glGetFramebufferAttachmentParameterivEXT)
+	GL_STUB_ALIAS(glGetRenderbufferParameteriv, glGetRenderbufferParameterivEXT)
+	GL_STUB_ALIAS(glIsFramebuffer, glIsFramebufferEXT)
+	GL_STUB_ALIAS(glIsRenderbuffer, glIsRenderbufferEXT)
+	GL_STUB_ALIAS(glRenderbufferStorage, glRenderbufferStorageEXT)
+	GL_STUB_ALIAS(glFramebufferTextureLayer, glFramebufferTextureLayerEXT)
+
+	.globl	gl_dispatch_functions_end
+	HIDDEN(gl_dispatch_functions_end)
+gl_dispatch_functions_end:
diff --git a/src/mesa/sparc/norm.S b/src/mesa/sparc/norm.S
index 44950a10a5..117d36fa22 100644
--- a/src/mesa/sparc/norm.S
+++ b/src/mesa/sparc/norm.S
@@ -1,11 +1,8 @@
 
 #include "sparc_matrix.h"
 
-#if defined(SVR4) || defined(__SVR4) || defined(__svr4__)
-	/* Solaris requires this for 64-bit. */
 	.register %g2, #scratch
 	.register %g3, #scratch
-#endif
 
 	.text
 
@@ -98,7 +95,7 @@ _mesa_sparc_transform_normalize_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 	ba	7f
 	 nop
@@ -154,7 +151,7 @@ _mesa_sparc_transform_normalize_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	5b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -225,7 +222,7 @@ _mesa_sparc_transform_normalize_normals_no_rot:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 	ba	7f
 	 nop
@@ -263,7 +260,7 @@ _mesa_sparc_transform_normalize_normals_no_rot:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	5b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -314,7 +311,7 @@ _mesa_sparc_transform_rescale_normals_no_rot:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -379,7 +376,7 @@ _mesa_sparc_transform_rescale_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -421,7 +418,7 @@ _mesa_sparc_transform_normals_no_rot:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -471,7 +468,7 @@ _mesa_sparc_transform_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -529,7 +526,7 @@ _mesa_sparc_normalize_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 	ba	7f
 	 nop
@@ -557,7 +554,7 @@ _mesa_sparc_normalize_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	5b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
@@ -602,7 +599,7 @@ _mesa_sparc_rescale_normals:
 
 	cmp	%o4, %g1			! continue if (i < count)
 	bl	1b
-	 add	%g3, 0x0c, %g3			! advance out vector pointer
+	 add	%g3, 0x10, %g3			! advance out vector pointer
 
 7:	retl
 	 nop
diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c
index 84e8ac6723..d2286a2c83 100644
--- a/src/mesa/sparc/sparc.c
+++ b/src/mesa/sparc/sparc.c
@@ -113,8 +113,6 @@ void _mesa_init_all_sparc_transform_asm(void)
    _mesa_clip_tab[4] = _mesa_sparc_cliptest_points4;
    _mesa_clip_np_tab[4] = _mesa_sparc_cliptest_points4_np;
 
-#if 0
-   /* disable these too.  See bug 673938 */
    _mesa_normal_tab[NORM_TRANSFORM | NORM_NORMALIZE] =
 	   _mesa_sparc_transform_normalize_normals;
    _mesa_normal_tab[NORM_TRANSFORM_NO_ROT | NORM_NORMALIZE] =
@@ -131,7 +129,6 @@ void _mesa_init_all_sparc_transform_asm(void)
 	   _mesa_sparc_normalize_normals;
    _mesa_normal_tab[NORM_RESCALE] =
 	   _mesa_sparc_rescale_normals;
-#endif
 
 #ifdef DEBUG_MATH
    _math_test_all_transform_functions("sparc");
@@ -140,38 +137,4 @@ void _mesa_init_all_sparc_transform_asm(void)
 #endif
 }
 
-extern unsigned int _mesa_sparc_glapi_begin;
-extern unsigned int _mesa_sparc_glapi_end;
-extern void __glapi_sparc_icache_flush(unsigned int *);
-
 #endif /* USE_SPARC_ASM */
-
-
-void _mesa_init_sparc_glapi_relocs(void)
-{
-#ifdef USE_SPARC_ASM
-	unsigned int *insn_ptr, *end_ptr;
-	unsigned long disp_addr;
-
-	insn_ptr = &_mesa_sparc_glapi_begin;
-	end_ptr = &_mesa_sparc_glapi_end;
-	disp_addr = (unsigned long) &_glapi_Dispatch;
-
-	while (insn_ptr < end_ptr) {
-#ifdef __arch64__
-		insn_ptr[0] |= (disp_addr >> (32 + 10));
-		insn_ptr[1] |= ((disp_addr & 0xffffffff) >> 10);
-		__glapi_sparc_icache_flush(&insn_ptr[0]);
-		insn_ptr[2] |= ((disp_addr >> 32) & ((1 << 10) - 1));
-		insn_ptr[3] |= (disp_addr & ((1 << 10) - 1));
-		__glapi_sparc_icache_flush(&insn_ptr[2]);
-		insn_ptr += 11;
-#else
-		insn_ptr[0] |= (disp_addr >> 10);
-		insn_ptr[1] |= (disp_addr & ((1 << 10) - 1));
-		__glapi_sparc_icache_flush(&insn_ptr[0]);
-		insn_ptr += 5;
-#endif
-	}
-#endif /* USE_SPARC_ASM */
-}
diff --git a/src/mesa/sparc/sparc.h b/src/mesa/sparc/sparc.h
index a98e4d0e40..b9ea336856 100644
--- a/src/mesa/sparc/sparc.h
+++ b/src/mesa/sparc/sparc.h
@@ -32,6 +32,5 @@
 #define SPARC_H
 
 extern void _mesa_init_all_sparc_transform_asm(void);
-extern void _mesa_init_sparc_glapi_relocs(void);
 
 #endif /* !(SPARC_H) */
diff --git a/src/mesa/state_tracker/st_atom_depth.c b/src/mesa/state_tracker/st_atom_depth.c
index a6156df7ae..4ea62dda18 100644
--- a/src/mesa/state_tracker/st_atom_depth.c
+++ b/src/mesa/state_tracker/st_atom_depth.c
@@ -106,7 +106,7 @@ update_depth_stencil_alpha(struct st_context *st)
        ctx->Query.CurrentOcclusionObject->Active)
       dsa->depth.occlusion_count = 1;
 
-   if (ctx->Stencil.Enabled && ctx->Visual.stencilBits > 0) {
+   if (ctx->Stencil.Enabled && ctx->DrawBuffer->Visual.stencilBits > 0) {
       dsa->stencil[0].enabled = 1;
       dsa->stencil[0].func = st_compare_func_to_pipe(ctx->Stencil.Function[0]);
       dsa->stencil[0].fail_op = gl_stencil_op_to_pipe(ctx->Stencil.FailFunc[0]);
diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c
index ce872458e3..05b69c9d00 100644
--- a/src/mesa/state_tracker/st_atom_pixeltransfer.c
+++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c
@@ -212,7 +212,7 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key)
    inst[ic].TexSrcTarget = TEXTURE_2D_INDEX;
    ic++;
    fp->Base.InputsRead = (1 << FRAG_ATTRIB_TEX0);
-   fp->Base.OutputsWritten = (1 << FRAG_RESULT_COLR);
+   fp->Base.OutputsWritten = (1 << FRAG_RESULT_COLOR);
    fp->Base.SamplersUsed = 0x1;  /* sampler 0 (bit 0) is used */
 
    if (key->scaleAndBias) {
@@ -400,7 +400,7 @@ get_pixel_transfer_program(GLcontext *ctx, const struct state_key *key)
    {
       struct prog_instruction *last = &inst[ic - 1];
       last->DstReg.File = PROGRAM_OUTPUT;
-      last->DstReg.Index = FRAG_RESULT_COLR;
+      last->DstReg.Index = FRAG_RESULT_COLOR;
    }
 
    /* END; */
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index 66d7cbf8e6..f55a5e713f 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -142,6 +142,10 @@ make_bitmap_fragment_program(GLcontext *ctx, GLuint samplerIndex)
    /* KIL if -tmp0 < 0 # texel=0 -> keep / texel=0 -> discard */
    p->Instructions[ic].Opcode = OPCODE_KIL;
    p->Instructions[ic].SrcReg[0].File = PROGRAM_TEMPORARY;
+
+   if (ctx->st->bitmap.tex_format == PIPE_FORMAT_L8_UNORM)
+      p->Instructions[ic].SrcReg[0].Swizzle = SWIZZLE_XXXX;
+
    p->Instructions[ic].SrcReg[0].Index = 0;
    p->Instructions[ic].SrcReg[0].NegateBase = NEGATE_XYZW;
    ic++;
@@ -157,7 +161,11 @@ make_bitmap_fragment_program(GLcontext *ctx, GLuint samplerIndex)
 
    stfp = (struct st_fragment_program *) p;
    stfp->Base.UsesKill = GL_TRUE;
-   st_translate_fragment_program(ctx->st, stfp, NULL);
+
+   /* No need to send this incomplete program down to hardware:
+    *
+    * st_translate_fragment_program(ctx->st, stfp, NULL);
+    */
 
    return stfp;
 }
@@ -786,6 +794,10 @@ st_init_bitmap(struct st_context *st)
                                    PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
       st->bitmap.tex_format = PIPE_FORMAT_I8_UNORM;
    }
+   else if (screen->is_format_supported(screen, PIPE_FORMAT_L8_UNORM, PIPE_TEXTURE_2D, 
+                                        PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
+      st->bitmap.tex_format = PIPE_FORMAT_L8_UNORM;
+   }
    else {
       /* XXX support more formats */
       assert(0);
diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index c73d0080ca..cc7a9e7890 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -75,7 +75,7 @@ is_passthrough_program(const struct gl_fragment_program *prog)
       if (inst[0].Opcode == OPCODE_MOV &&
           inst[1].Opcode == OPCODE_END &&
           inst[0].DstReg.File == PROGRAM_OUTPUT &&
-          inst[0].DstReg.Index == FRAG_RESULT_COLR &&
+          inst[0].DstReg.Index == FRAG_RESULT_COLOR &&
           inst[0].DstReg.WriteMask == WRITEMASK_XYZW &&
           inst[0].SrcReg[0].File == PROGRAM_INPUT &&
           inst[0].SrcReg[0].Index == FRAG_ATTRIB_COL0 &&
@@ -158,7 +158,7 @@ combined_drawpix_fragment_program(GLcontext *ctx)
 
 /**
  * Create fragment shader that does a TEX() instruction to get a Z
- * value, then writes to FRAG_RESULT_DEPR.
+ * value, then writes to FRAG_RESULT_DEPTH.
  * Pass fragment color through as-is.
  */
 static struct st_fragment_program *
@@ -191,7 +191,7 @@ make_fragment_shader_z(struct st_context *st)
    /* TEX result.depth, fragment.texcoord[0], texture[0], 2D; */
    p->Instructions[ic].Opcode = OPCODE_TEX;
    p->Instructions[ic].DstReg.File = PROGRAM_OUTPUT;
-   p->Instructions[ic].DstReg.Index = FRAG_RESULT_DEPR;
+   p->Instructions[ic].DstReg.Index = FRAG_RESULT_DEPTH;
    p->Instructions[ic].DstReg.WriteMask = WRITEMASK_Z;
    p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT;
    p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_TEX0;
@@ -202,7 +202,7 @@ make_fragment_shader_z(struct st_context *st)
    /* MOV result.color, fragment.color */
    p->Instructions[ic].Opcode = OPCODE_MOV;
    p->Instructions[ic].DstReg.File = PROGRAM_OUTPUT;
-   p->Instructions[ic].DstReg.Index = FRAG_RESULT_COLR;
+   p->Instructions[ic].DstReg.Index = FRAG_RESULT_COLOR;
    p->Instructions[ic].SrcReg[0].File = PROGRAM_INPUT;
    p->Instructions[ic].SrcReg[0].Index = FRAG_ATTRIB_COL0;
    ic++;
@@ -213,7 +213,7 @@ make_fragment_shader_z(struct st_context *st)
    assert(ic == p->NumInstructions);
 
    p->InputsRead = FRAG_BIT_TEX0 | FRAG_BIT_COL0;
-   p->OutputsWritten = (1 << FRAG_RESULT_COLR) | (1 << FRAG_RESULT_DEPR);
+   p->OutputsWritten = (1 << FRAG_RESULT_COLOR) | (1 << FRAG_RESULT_DEPTH);
    p->SamplersUsed = 0x1;  /* sampler 0 (bit 0) is used */
 
    st->drawpix.z_shader = (struct st_fragment_program *) p;
diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c
index 5cc6960c06..15bd6fee05 100644
--- a/src/mesa/state_tracker/st_cb_fbo.c
+++ b/src/mesa/state_tracker/st_cb_fbo.c
@@ -439,6 +439,25 @@ st_finish_render_texture(GLcontext *ctx,
 }
 
 
+/**
+ * Check that the framebuffer configuration is valid in terms of what
+ * the driver can support.
+ *
+ * For Gallium we only supports combined Z+stencil, not separate buffers.
+ */
+static void
+st_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
+{
+   const struct gl_renderbuffer *depthRb =
+      fb->Attachment[BUFFER_DEPTH].Renderbuffer;
+   const struct gl_renderbuffer *stencilRb =
+      fb->Attachment[BUFFER_STENCIL].Renderbuffer;
+
+   if (stencilRb && depthRb && stencilRb != depthRb) {
+      fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+   }
+}
+
 
 void st_init_fbo_functions(struct dd_function_table *functions)
 {
@@ -448,6 +467,7 @@ void st_init_fbo_functions(struct dd_function_table *functions)
    functions->FramebufferRenderbuffer = st_framebuffer_renderbuffer;
    functions->RenderTexture = st_render_texture;
    functions->FinishRenderTexture = st_finish_render_texture;
+   functions->ValidateFramebuffer = st_validate_framebuffer;
    /* no longer needed by core Mesa, drivers handle resizes...
    functions->ResizeBuffers = st_resize_buffers;
    */
diff --git a/src/mesa/state_tracker/st_cb_feedback.c b/src/mesa/state_tracker/st_cb_feedback.c
index 86d3542262..93f7145219 100644
--- a/src/mesa/state_tracker/st_cb_feedback.c
+++ b/src/mesa/state_tracker/st_cb_feedback.c
@@ -122,8 +122,8 @@ feedback_tri( struct draw_stage *stage, struct prim_header *prim )
 {
    struct feedback_stage *fs = feedback_stage(stage);
    struct draw_context *draw = stage->draw;
-   FEEDBACK_TOKEN(fs->ctx, (GLfloat) GL_POLYGON_TOKEN);
-   FEEDBACK_TOKEN(fs->ctx, (GLfloat) 3); /* three vertices */
+   _mesa_feedback_token(fs->ctx, (GLfloat) GL_POLYGON_TOKEN);
+   _mesa_feedback_token(fs->ctx, (GLfloat) 3); /* three vertices */
    feedback_vertex(fs->ctx, draw, prim->v[0]);
    feedback_vertex(fs->ctx, draw, prim->v[1]);
    feedback_vertex(fs->ctx, draw, prim->v[2]);
@@ -136,11 +136,11 @@ feedback_line( struct draw_stage *stage, struct prim_header *prim )
    struct feedback_stage *fs = feedback_stage(stage);
    struct draw_context *draw = stage->draw;
    if (fs->reset_stipple_counter) {
-      FEEDBACK_TOKEN(fs->ctx, (GLfloat) GL_LINE_RESET_TOKEN);
+      _mesa_feedback_token(fs->ctx, (GLfloat) GL_LINE_RESET_TOKEN);
       fs->reset_stipple_counter = GL_FALSE;
    }
    else {
-      FEEDBACK_TOKEN(fs->ctx, (GLfloat) GL_LINE_TOKEN);
+      _mesa_feedback_token(fs->ctx, (GLfloat) GL_LINE_TOKEN);
    }
    feedback_vertex(fs->ctx, draw, prim->v[0]);
    feedback_vertex(fs->ctx, draw, prim->v[1]);
@@ -152,7 +152,7 @@ feedback_point( struct draw_stage *stage, struct prim_header *prim )
 {
    struct feedback_stage *fs = feedback_stage(stage);
    struct draw_context *draw = stage->draw;
-   FEEDBACK_TOKEN(fs->ctx, (GLfloat) GL_POINT_TOKEN);
+   _mesa_feedback_token(fs->ctx, (GLfloat) GL_POINT_TOKEN);
    feedback_vertex(fs->ctx, draw, prim->v[0]);
 }
 
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index d2535b6a2f..f825204915 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -466,23 +466,23 @@ st_translate_fragment_program(struct st_context *st,
       GLbitfield outputsWritten = stfp->Base.Base.OutputsWritten;
 
       /* if z is written, emit that first */
-      if (outputsWritten & (1 << FRAG_RESULT_DEPR)) {
+      if (outputsWritten & (1 << FRAG_RESULT_DEPTH)) {
          fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_POSITION;
          fs_output_semantic_index[fs_num_outputs] = 0;
-         outputMapping[FRAG_RESULT_DEPR] = fs_num_outputs;
+         outputMapping[FRAG_RESULT_DEPTH] = fs_num_outputs;
          fs_num_outputs++;
-         outputsWritten &= ~(1 << FRAG_RESULT_DEPR);
+         outputsWritten &= ~(1 << FRAG_RESULT_DEPTH);
       }
 
       /* handle remaning outputs (color) */
       for (attr = 0; attr < FRAG_RESULT_MAX; attr++) {
          if (outputsWritten & (1 << attr)) {
             switch (attr) {
-            case FRAG_RESULT_DEPR:
+            case FRAG_RESULT_DEPTH:
                /* handled above */
                assert(0);
                break;
-            case FRAG_RESULT_COLR:
+            case FRAG_RESULT_COLOR:
                fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR;
                fs_output_semantic_index[fs_num_outputs] = numColors;
                outputMapping[attr] = fs_num_outputs;
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index 297940adbd..719e6b8296 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.c
@@ -206,7 +206,7 @@ _swrast_update_deferred_texture(GLcontext *ctx)
    else {
       const struct gl_fragment_program *fprog
          = ctx->FragmentProgram._Current;
-      if (fprog && (fprog->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR))) {
+      if (fprog && (fprog->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH))) {
          /* Z comes from fragment program/shader */
          swrast->_DeferredTexture = GL_FALSE;
       }
diff --git a/src/mesa/swrast/s_context.h b/src/mesa/swrast/s_context.h
index cdd6fa5048..6e8d080704 100644
--- a/src/mesa/swrast/s_context.h
+++ b/src/mesa/swrast/s_context.h
@@ -289,6 +289,34 @@ swrast_render_finish(GLcontext *ctx)
 
 
 /*
+ * Fixed point arithmetic macros
+ */
+#ifndef FIXED_FRAC_BITS
+#define FIXED_FRAC_BITS 11
+#endif
+
+#define FIXED_SHIFT     FIXED_FRAC_BITS
+#define FIXED_ONE       (1 << FIXED_SHIFT)
+#define FIXED_HALF      (1 << (FIXED_SHIFT-1))
+#define FIXED_FRAC_MASK (FIXED_ONE - 1)
+#define FIXED_INT_MASK  (~FIXED_FRAC_MASK)
+#define FIXED_EPSILON   1
+#define FIXED_SCALE     ((float) FIXED_ONE)
+#define FIXED_DBL_SCALE ((double) FIXED_ONE)
+#define FloatToFixed(X) (IROUND((X) * FIXED_SCALE))
+#define FixedToDouble(X) ((X) * (1.0 / FIXED_DBL_SCALE))
+#define IntToFixed(I)   ((I) << FIXED_SHIFT)
+#define FixedToInt(X)   ((X) >> FIXED_SHIFT)
+#define FixedToUns(X)   (((unsigned int)(X)) >> FIXED_SHIFT)
+#define FixedCeil(X)    (((X) + FIXED_ONE - FIXED_EPSILON) & FIXED_INT_MASK)
+#define FixedFloor(X)   ((X) & FIXED_INT_MASK)
+#define FixedToFloat(X) ((X) * (1.0F / FIXED_SCALE))
+#define PosFloatToFixed(X)      FloatToFixed(X)
+#define SignedFloatToFixed(X)   FloatToFixed(X)
+
+
+
+/*
  * XXX these macros are just bandages for now in order to make
  * CHAN_BITS==32 compile cleanly.
  * These should probably go elsewhere at some point.
diff --git a/src/mesa/swrast/s_feedback.c b/src/mesa/swrast/s_feedback.c
index aa79531277..7bb914b658 100644
--- a/src/mesa/swrast/s_feedback.c
+++ b/src/mesa/swrast/s_feedback.c
@@ -59,8 +59,8 @@ _swrast_feedback_triangle(GLcontext *ctx, const SWvertex *v0,
                           const SWvertex *v1, const SWvertex *v2)
 {
    if (_swrast_culltriangle(ctx, v0, v1, v2)) {
-      FEEDBACK_TOKEN(ctx, (GLfloat) (GLint) GL_POLYGON_TOKEN);
-      FEEDBACK_TOKEN(ctx, (GLfloat) 3); /* three vertices */
+      _mesa_feedback_token(ctx, (GLfloat) (GLint) GL_POLYGON_TOKEN);
+      _mesa_feedback_token(ctx, (GLfloat) 3); /* three vertices */
 
       if (ctx->Light.ShadeModel == GL_SMOOTH) {
          feedback_vertex(ctx, v0, v0);
@@ -86,7 +86,7 @@ _swrast_feedback_line(GLcontext *ctx, const SWvertex *v0,
    if (swrast->StippleCounter == 0)
       token = GL_LINE_RESET_TOKEN;
 
-   FEEDBACK_TOKEN(ctx, (GLfloat) (GLint) token);
+   _mesa_feedback_token(ctx, (GLfloat) (GLint) token);
 
    if (ctx->Light.ShadeModel == GL_SMOOTH) {
       feedback_vertex(ctx, v0, v0);
@@ -104,7 +104,7 @@ _swrast_feedback_line(GLcontext *ctx, const SWvertex *v0,
 void
 _swrast_feedback_point(GLcontext *ctx, const SWvertex *v)
 {
-   FEEDBACK_TOKEN(ctx, (GLfloat) (GLint) GL_POINT_TOKEN);
+   _mesa_feedback_token(ctx, (GLfloat) (GLint) GL_POINT_TOKEN);
    feedback_vertex(ctx, v, v);
 }
 
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index c6601f5593..3735f041d6 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -202,9 +202,9 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
          if (_mesa_execute_program(ctx, &program->Base, machine)) {
 
             /* Store result color */
-            if (outputsWritten & (1 << FRAG_RESULT_COLR)) {
+            if (outputsWritten & (1 << FRAG_RESULT_COLOR)) {
                COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
-                       machine->Outputs[FRAG_RESULT_COLR]);
+                       machine->Outputs[FRAG_RESULT_COLOR]);
             }
             else {
                /* Multiple drawbuffers / render targets
@@ -221,8 +221,8 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
             }
 
             /* Store result depth/z */
-            if (outputsWritten & (1 << FRAG_RESULT_DEPR)) {
-               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPR][2];
+            if (outputsWritten & (1 << FRAG_RESULT_DEPTH)) {
+               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
                if (depth <= 0.0)
                   span->array->z[i] = 0;
                else if (depth >= 1.0)
@@ -259,12 +259,12 @@ _swrast_exec_fragment_program( GLcontext *ctx, SWspan *span )
 
    run_program(ctx, span, 0, span->end);
 
-   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_COLR)) {
+   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_COLOR)) {
       span->interpMask &= ~SPAN_RGBA;
       span->arrayMask |= SPAN_RGBA;
    }
 
-   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) {
+   if (program->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH)) {
       span->interpMask &= ~SPAN_Z;
       span->arrayMask |= SPAN_Z;
    }
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index 047f7991e6..015f8a05c3 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -75,6 +75,12 @@ typedef struct {
 } SWvertex;
 
 
+/**
+ * Fixed point data type.
+ */
+typedef int GLfixed;
+
+
 #define FRAG_ATTRIB_CI FRAG_ATTRIB_COL0
 
 
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index f0d31fdac3..f69b122046 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -149,13 +149,10 @@ _tnl_InvalidateState( GLcontext *ctx, GLuint new_state )
       /* fixed-function fog */
       RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_FOG );
    }
-   else if (ctx->FragmentProgram._Current) {
-      struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
-      if (fp) {
-         if (fp->FogOption != GL_NONE || (fp->Base.InputsRead & FRAG_BIT_FOGC)) {
-            /* fragment program needs fog coord */
-            RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_FOG );
-         }
+   else if (fp) {
+      if (fp->FogOption != GL_NONE || (fp->Base.InputsRead & FRAG_BIT_FOGC)) {
+         /* fragment program needs fog coord */
+         RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_FOG );
       }
    }
 
diff --git a/src/mesa/tnl/t_context.h b/src/mesa/tnl/t_context.h
index 0a6ce04614..c19eb3df3c 100644
--- a/src/mesa/tnl/t_context.h
+++ b/src/mesa/tnl/t_context.h
@@ -50,6 +50,7 @@
 #define _T_CONTEXT_H
 
 #include "main/glheader.h"
+#include "main/bitset.h"
 #include "main/mtypes.h"
 
 #include "math/m_matrix.h"
diff --git a/src/mesa/tnl_dd/t_dd_vb.c b/src/mesa/tnl_dd/t_dd_vb.c
index 3cedd90119..b3937c29a0 100644
--- a/src/mesa/tnl_dd/t_dd_vb.c
+++ b/src/mesa/tnl_dd/t_dd_vb.c
@@ -188,6 +188,9 @@ void TAG(translate_vertex)(GLcontext *ctx,
 }
 
 
+/* prototype to silence warning */
+void TAG(print_vertex)( GLcontext *ctx, const VERTEX *v );
+
 
 void TAG(print_vertex)( GLcontext *ctx, const VERTEX *v )
 {