R6xx/R7xx: WIP r6xx-rewrite code

26 files changed, 1788 insertions, 928 deletions

diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile
index 00c56c03cd..474a3167b4 100644
--- a/src/mesa/drivers/dri/r600/Makefile
+++ b/src/mesa/drivers/dri/r600/Makefile
@@ -48,9 +48,12 @@ DRIVER_SOURCES = \
 		 r700_fragprog.c \
 		 r700_vertprog.c \
 		 r700_shader.c \
+		 r700_shaderinst.c \
 		 r700_ioctl.c \
 		 r700_oglprog.c \
 		 r700_chip.c     \
+		 r700_state.c    \
+		 r700_clear.c    \
 		 r700_render.c   \
 		 r700_tex.c      \
 		 r700_debug.c    \
diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.c b/src/mesa/drivers/dri/r600/r600_cmdbuf.c
index bc14f9a755..8550457566 100644
--- a/src/mesa/drivers/dri/r600/r600_cmdbuf.c
+++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.c
@@ -28,11 +28,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 **************************************************************************/
 
 /**
- * \file
- *
- * \author Nicolai Haehnle <prefect_@gmx.net>
+ * Mostly coppied from \radeon\radeon_cs_legacy.c
  */
 
+#include <errno.h>
+
 #include "main/glheader.h"
 #include "main/state.h"
 #include "main/imports.h"
@@ -52,11 +52,582 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_mipmap_tree.h"
 #include "radeon_reg.h"
 
-/**
- * Allocate memory for the command buffer and initialize the state atom
- * list. Note that the initial hardware state is set by r600InitState().
- */
-void r600InitCmdBuf(r600ContextPtr r600)
+struct r600_cs_manager_legacy 
 {
-	rcommonInitCmdBuf(&r600->radeon);
+    struct radeon_cs_manager    base;
+    struct radeon_context       *ctx;
+    /* hack for scratch stuff */
+    uint32_t                    pending_age;
+    uint32_t                    pending_count;
+};
+
+struct r600_cs_reloc_legacy {
+    struct radeon_cs_reloc  base;
+    uint32_t                cindices;
+    uint32_t                *indices;
+    uint32_t                *reloc_indices;
+    struct offset_modifiers offset_mod;
+};
+
+static struct radeon_cs * r600_cs_create(struct radeon_cs_manager *csm,
+                                   uint32_t ndw)
+{
+    struct radeon_cs *cs;
+
+    cs = (struct radeon_cs*)calloc(1, sizeof(struct radeon_cs));
+    if (cs == NULL) {
+        return NULL;
+    }
+    cs->csm = csm;
+    cs->ndw = (ndw + 0x3FF) & (~0x3FF);
+    cs->packets = (uint32_t*)malloc(4*cs->ndw);
+    if (cs->packets == NULL) {
+        free(cs);
+        return NULL;
+    }
+    cs->relocs_total_size = 0;
+    return cs;
+}
+
+int r600_cs_write_reloc(struct radeon_cs *cs,
+                        struct radeon_bo *bo,
+                        uint32_t read_domain,
+                        uint32_t write_domain,
+                        uint32_t flags,
+                        offset_modifiers* poffset_mod)
+{
+    struct r600_cs_reloc_legacy *relocs;
+    int i;
+
+    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
+    /* check domains */
+    if ((read_domain && write_domain) || (!read_domain && !write_domain)) {
+        /* in one CS a bo can only be in read or write domain but not
+         * in read & write domain at the same sime
+         */
+        return -EINVAL;
+    }
+    if (read_domain == RADEON_GEM_DOMAIN_CPU) {
+        return -EINVAL;
+    }
+    if (write_domain == RADEON_GEM_DOMAIN_CPU) {
+        return -EINVAL;
+    }
+    /* check if bo is already referenced */
+    for(i = 0; i < cs->crelocs; i++) {
+        uint32_t *indices;
+        uint32_t *reloc_indices;
+
+        if (relocs[i].base.bo->handle == bo->handle) {
+            /* Check domains must be in read or write. As we check already
+             * checked that in argument one of the read or write domain was
+             * set we only need to check that if previous reloc as the read
+             * domain set then the read_domain should also be set for this
+             * new relocation.
+             */
+            if (relocs[i].base.read_domain && !read_domain) {
+                return -EINVAL;
+            }
+            if (relocs[i].base.write_domain && !write_domain) {
+                return -EINVAL;
+            }
+            relocs[i].base.read_domain |= read_domain;
+            relocs[i].base.write_domain |= write_domain;
+            /* save indice */
+            relocs[i].cindices++;
+            indices = (uint32_t*)realloc(relocs[i].indices,
+                                         relocs[i].cindices * 4);
+            reloc_indices = (uint32_t*)realloc(relocs[i].reloc_indices,
+                                               relocs[i].cindices * 4);
+            if ( (indices == NULL) || (reloc_indices == NULL) ) {
+                relocs[i].cindices -= 1;
+                return -ENOMEM;
+            }
+            relocs[i].indices = indices;
+            relocs[i].reloc_indices = reloc_indices;
+            relocs[i].indices[relocs[i].cindices - 1] = cs->cdw - 1;
+            relocs[i].reloc_indices[relocs[i].cindices - 1] = cs->section_cdw;
+            cs->section_ndw += 2;
+            cs->section_cdw += 2;
+
+            relocs[i].offset_mod.shift     = poffset_mod->shift;
+            relocs[i].offset_mod.shiftbits = poffset_mod->shiftbits;
+            relocs[i].offset_mod.mask      = poffset_mod->mask;
+
+            return 0;
+        }
+    }
+    /* add bo to reloc */
+    relocs = (struct r600_cs_reloc_legacy*)
+             realloc(cs->relocs,
+                     sizeof(struct r600_cs_reloc_legacy) * (cs->crelocs + 1));
+    if (relocs == NULL) {
+        return -ENOMEM;
+    }
+    cs->relocs = relocs;
+    relocs[cs->crelocs].base.bo = bo;
+    relocs[cs->crelocs].base.read_domain = read_domain;
+    relocs[cs->crelocs].base.write_domain = write_domain;
+    relocs[cs->crelocs].base.flags = flags;
+    relocs[cs->crelocs].indices = (uint32_t*)malloc(4);
+    relocs[cs->crelocs].reloc_indices = (uint32_t*)malloc(4);
+    if ( (relocs[cs->crelocs].indices == NULL) || (relocs[cs->crelocs].reloc_indices == NULL) )
+    {
+        return -ENOMEM;
+    }
+    relocs[cs->crelocs].offset_mod.shift     = poffset_mod->shift;
+    relocs[cs->crelocs].offset_mod.shiftbits = poffset_mod->shiftbits;
+    relocs[cs->crelocs].offset_mod.mask      = poffset_mod->mask;
+
+    relocs[cs->crelocs].indices[0] = cs->cdw - 1;
+    relocs[cs->crelocs].reloc_indices[0] = cs->section_cdw;
+    cs->section_ndw += 2;
+    cs->section_cdw += 2;
+    relocs[cs->crelocs].cindices = 1;
+    cs->relocs_total_size += radeon_bo_legacy_relocs_size(bo);
+    cs->crelocs++;
+
+    radeon_bo_ref(bo);
+    return 0;
+}
+
+static int r600_cs_begin(struct radeon_cs *cs,
+                    uint32_t ndw,
+                    const char *file,
+                    const char *func,
+                    int line)
+{
+    if (cs->section) {
+        fprintf(stderr, "CS already in a section(%s,%s,%d)\n",
+                cs->section_file, cs->section_func, cs->section_line);
+        fprintf(stderr, "CS can't start section(%s,%s,%d)\n",
+                file, func, line);
+        return -EPIPE;
+    }
+
+    if (cs->cdw + ndw + 32 > cs->ndw) { /* Left 32 DWORD (8 offset+pitch) spare room for reloc indices */
+        uint32_t tmp, *ptr;
+	int num = (ndw > 0x3FF) ? ndw : 0x3FF;
+
+        tmp = (cs->cdw + 1 + num) & (~num);
+        ptr = (uint32_t*)realloc(cs->packets, 4 * tmp);
+        if (ptr == NULL) {
+            return -ENOMEM;
+        }
+        cs->packets = ptr;
+        cs->ndw = tmp;
+    }
+
+    cs->section = 1;
+    cs->section_ndw = 0; 
+    cs->section_cdw = cs->cdw + ndw; /* start of reloc indices. */
+    cs->section_file = file;
+    cs->section_func = func;
+    cs->section_line = line;
+
+    return 0;
+}
+
+static int r600_cs_end(struct radeon_cs *cs,
+                  const char *file,
+                  const char *func,
+                  int line)
+
+{
+    if (!cs->section) {
+        fprintf(stderr, "CS no section to end at (%s,%s,%d)\n",
+                file, func, line);
+        return -EPIPE;
+    }
+    cs->section = 0;
+
+    if ( (cs->section_ndw + cs->cdw) != cs->section_cdw ) 
+    {
+        fprintf(stderr, "CS section size missmatch start at (%s,%s,%d) %d vs %d\n",
+                cs->section_file, cs->section_func, cs->section_line, cs->section_ndw, cs->section_cdw);
+        fprintf(stderr, "CS section end at (%s,%s,%d)\n",
+                file, func, line);
+        return -EPIPE;
+    }
+
+    cs->cdw = cs->section_cdw;
+    return 0;
+}
+
+static int r600_cs_process_relocs(struct radeon_cs *cs, 
+                                  uint32_t * reloc_chunk,
+                                  uint32_t * length_dw_reloc_chunk) 
+{
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct r600_cs_reloc_legacy *relocs;
+    int i, j, r;
+
+    uint32_t offset_dw = 0;
+
+    csm = (struct r600_cs_manager_legacy*)cs->csm;
+    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
+restart:
+    for (i = 0; i < cs->crelocs; i++) 
+    {
+        for (j = 0; j < relocs[i].cindices; j++) 
+        {
+            uint32_t soffset, eoffset, asicoffset;
+
+            r = radeon_bo_legacy_validate(relocs[i].base.bo,
+                                           &soffset, &eoffset);
+	        if (r == -EAGAIN)
+            {
+	             goto restart;
+            }
+            if (r) 
+            {
+                fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
+                        relocs[i].base.bo, soffset, eoffset);
+                return r;
+            }
+            asicoffset = cs->packets[relocs[i].indices[j]] + soffset;
+            if (asicoffset >= eoffset) 
+            {
+	      /*                radeon_bo_debug(relocs[i].base.bo, 12); */
+                fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
+                        relocs[i].base.bo, soffset, eoffset);
+                fprintf(stderr, "above end: %p 0x%08X 0x%08X\n",
+                        relocs[i].base.bo,
+                        cs->packets[relocs[i].indices[j]],
+                        eoffset);
+                exit(0);
+                return -EINVAL;
+            }
+            /* apply offset operator */
+            switch (relocs[i].offset_mod.shift)
+            {
+            case NO_SHIFT:
+                asicoffset = asicoffset & relocs[i].offset_mod.mask;
+                break;
+            case LEFT_SHIFT:
+                asicoffset = (asicoffset << relocs[i].offset_mod.shiftbits) & relocs[i].offset_mod.mask;
+                break;
+            case RIGHT_SHIFT:
+                asicoffset = (asicoffset >> relocs[i].offset_mod.shiftbits) & relocs[i].offset_mod.mask;
+                break;
+            default:
+                break;
+            };              
+
+            /* pkt3 nop header in ib chunk */
+            cs->packets[relocs[i].reloc_indices[j]] = 0xC0001000;
+
+            /* reloc index in ib chunk */
+            cs->packets[relocs[i].reloc_indices[j] + 1] = offset_dw;
+            
+            /* asic offset in reloc chunk */ /* see alex drm r600_nomm_relocate */
+            reloc_chunk[offset_dw] = asicoffset;
+            reloc_chunk[offset_dw + 3] = 0;
+
+            offset_dw += 4;
+        }
+    }
+
+    *length_dw_reloc_chunk = offset_dw;
+
+    return 0;
+}
+
+static int r600_cs_set_age(struct radeon_cs *cs) /* -------------- */
+{
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct r600_cs_reloc_legacy *relocs;
+    int i;
+
+    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
+    for (i = 0; i < cs->crelocs; i++) {
+        radeon_bo_legacy_pending(relocs[i].base.bo, csm->pending_age);
+        radeon_bo_unref(relocs[i].base.bo);
+    }
+    return 0;
+}
+
+static int r600_cs_emit(struct radeon_cs *cs)
+{
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct drm_radeon_cs       cs_cmd;
+    struct drm_radeon_cs_chunk cs_chunk[2];
+    drm_radeon_cmd_buffer_t cmd; 
+    /* drm_r300_cmd_header_t age; */
+    uint32_t length_dw_reloc_chunk;
+    uint64_t ull;
+    uint64_t * chunk_ptrs[2];
+    uint32_t reloc_chunk[128]; 
+    int r;
+    int retry = 0;
+
+    /* TODO : put chip level things here if need. */
+    /* csm->ctx->vtbl.emit_cs_header(cs, csm->ctx); */
+
+    /* TODO : append buffer age */
+    
+    r = r600_cs_process_relocs(cs, &(reloc_chunk[0]), &length_dw_reloc_chunk);
+    if (r) {
+        return 0;
+    }
+
+    /* raw ib chunk */
+    cs_chunk[0].chunk_id   = RADEON_CHUNK_ID_IB;
+    cs_chunk[0].length_dw  = cs->cdw;
+    cs_chunk[0].chunk_data = (uint64_t)(cs->packets);
+
+    /* reloc chaunk */
+    cs_chunk[1].chunk_id   = RADEON_CHUNK_ID_RELOCS;
+    cs_chunk[1].length_dw  = length_dw_reloc_chunk;
+    cs_chunk[1].chunk_data = (uint64_t)&(reloc_chunk[0]);
+
+    chunk_ptrs[0] = (uint64_t * )&(cs_chunk[0]);
+    chunk_ptrs[1] = (uint64_t * )&(cs_chunk[1]);
+
+    cs_cmd.num_chunks = 2;
+    cs_cmd.cs_id      = 0;
+    cs_cmd.chunks     = (uint64_t)&(chunk_ptrs[0]);
+
+    /* dump_cmdbuf(cs); */
+
+    do 
+    {
+        r = drmCommandWriteRead(cs->csm->fd, DRM_RADEON_CS, &cs_cmd, sizeof(cs_cmd));
+        retry++;
+    } while (r == -EAGAIN && retry < 1000);
+
+    if (r) {
+        return r;
+    }
+
+    r600_cs_set_age(cs);
+
+    cs->csm->read_used = 0;
+    cs->csm->vram_write_used = 0;
+    cs->csm->gart_write_used = 0;
+
+    return 0;
+}
+
+static void inline r600_cs_free_reloc(void *relocs_p, int crelocs)
+{
+    struct r600_cs_reloc_legacy *relocs = relocs_p;
+    int i;
+    if (!relocs_p)
+      return;
+    for (i = 0; i < crelocs; i++)
+    {
+        free(relocs[i].indices);
+        free(relocs[i].reloc_indices);
+    }
+}
+
+static int r600_cs_destroy(struct radeon_cs *cs)
+{
+    r600_cs_free_reloc(cs->relocs, cs->crelocs);
+    free(cs->relocs);
+    free(cs->packets);
+    free(cs);
+    return 0;
+}
+
+static int r600_cs_erase(struct radeon_cs *cs)
+{
+    r600_cs_free_reloc(cs->relocs, cs->crelocs);
+    free(cs->relocs);
+    cs->relocs_total_size = 0;
+    cs->relocs = NULL;
+    cs->crelocs = 0;
+    cs->cdw = 0;
+    cs->section = 0;
+    return 0;
+}
+
+static int r600_cs_need_flush(struct radeon_cs *cs)
+{
+    /* this function used to flush when the BO usage got to
+     * a certain size, now the higher levels handle this better */
+    return 0;
+}
+
+static void r600_cs_print(struct radeon_cs *cs, FILE *file)
+{
+}
+
+static int r600_cs_check_space(struct radeon_cs *cs, struct radeon_cs_space_check *bos, int num_bo)
+{
+    struct radeon_cs_manager *csm = cs->csm;
+    int this_op_read = 0, this_op_gart_write = 0, this_op_vram_write = 0;
+    uint32_t read_domains, write_domain;
+    int i;
+    struct radeon_bo *bo;
+
+    /* check the totals for this operation */
+
+    if (num_bo == 0)
+        return 0;
+
+    /* prepare */
+    for (i = 0; i < num_bo; i++) 
+    {
+         bo = bos[i].bo;
+
+         bos[i].new_accounted = 0;
+         read_domains = bos[i].read_domains;
+         write_domain = bos[i].write_domain;
+		   
+         /* pinned bos don't count */
+         if (radeon_legacy_bo_is_static(bo))
+	     continue;
+ 
+         /* already accounted this bo */
+         if (write_domain && (write_domain == bo->space_accounted))
+	     continue;
+
+         if (read_domains && ((read_domains << 16) == bo->space_accounted))
+	     continue;
+      
+         if (bo->space_accounted == 0) 
+         {
+	         if (write_domain == RADEON_GEM_DOMAIN_VRAM)
+	             this_op_vram_write += bo->size;
+	         else if (write_domain == RADEON_GEM_DOMAIN_GTT)
+	             this_op_gart_write += bo->size;
+	         else
+	             this_op_read += bo->size;
+	         bos[i].new_accounted = (read_domains << 16) | write_domain;
+         } 
+         else 
+         {
+	        uint16_t old_read, old_write;
+	     
+	        old_read = bo->space_accounted >> 16;
+	        old_write = bo->space_accounted & 0xffff;
+
+	        if (write_domain && (old_read & write_domain)) 
+            {
+	            bos[i].new_accounted = write_domain;
+	            /* moving from read to a write domain */
+	            if (write_domain == RADEON_GEM_DOMAIN_VRAM) 
+                {
+		            this_op_read -= bo->size;
+		            this_op_vram_write += bo->size;
+	            } 
+                else if (write_domain == RADEON_GEM_DOMAIN_VRAM) 
+                {
+		            this_op_read -= bo->size;
+		            this_op_gart_write += bo->size;
+	            }
+	        } 
+            else if (read_domains & old_write) 
+            {
+	            bos[i].new_accounted = bo->space_accounted & 0xffff;
+	        } 
+            else 
+            {
+	            /* rewrite the domains */
+	            if (write_domain != old_write)
+		            fprintf(stderr,"WRITE DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, write_domain, old_write);
+	            if (read_domains != old_read)
+		            fprintf(stderr,"READ DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, read_domains, old_read);
+	            return RADEON_CS_SPACE_FLUSH;
+	        }
+         }
+	}
+	
+	if (this_op_read < 0)
+		this_op_read = 0;
+
+	/* check sizes - operation first */
+	if ((this_op_read + this_op_gart_write > csm->gart_limit) ||
+	    (this_op_vram_write > csm->vram_limit)) {
+	    return RADEON_CS_SPACE_OP_TO_BIG;
+	}
+
+	if (((csm->vram_write_used + this_op_vram_write) > csm->vram_limit) ||
+	    ((csm->read_used + csm->gart_write_used + this_op_gart_write + this_op_read) > csm->gart_limit)) {
+		return RADEON_CS_SPACE_FLUSH;
+	}
+
+	csm->gart_write_used += this_op_gart_write;
+	csm->vram_write_used += this_op_vram_write;
+	csm->read_used += this_op_read;
+	/* commit */
+	for (i = 0; i < num_bo; i++) {
+		bo = bos[i].bo;
+		bo->space_accounted = bos[i].new_accounted;
+	}
+
+	return RADEON_CS_SPACE_OK;
+}
+
+static struct radeon_cs_funcs  r600_cs_funcs = {
+    r600_cs_create,
+    r600_cs_write_reloc,
+    r600_cs_begin,
+    r600_cs_end,
+    r600_cs_emit,
+    r600_cs_destroy,
+    r600_cs_erase,
+    r600_cs_need_flush,
+    r600_cs_print,
+    r600_cs_check_space
+};
+
+struct radeon_cs_manager * r600_radeon_cs_manager_legacy_ctor(struct radeon_context *ctx)
+{
+    struct r600_cs_manager_legacy *csm;
+
+    csm = (struct r600_cs_manager_legacy*)
+          calloc(1, sizeof(struct r600_cs_manager_legacy));
+    if (csm == NULL) {
+        return NULL;
+    }
+    csm->base.funcs = &r600_cs_funcs;
+    csm->base.fd = ctx->dri.fd;
+    csm->ctx = ctx;
+    csm->pending_age = 1;
+    return (struct radeon_cs_manager*)csm;
+}
+
+void r600InitCmdBuf(r600ContextPtr r600) /* from rcommonInitCmdBuf */
+{
+    radeonContextPtr rmesa = &r600->radeon;
+	
+    GLuint size;
+	/* Initialize command buffer */
+	size = 256 * driQueryOptioni(&rmesa->optionCache,
+				     "command_buffer_size");
+	if (size < 2 * rmesa->hw.max_state_size) {
+		size = 2 * rmesa->hw.max_state_size + 65535;
+	}
+	if (size > 64 * 256)
+		size = 64 * 256;
+
+	if (rmesa->radeonScreen->kernel_mm) {
+		int fd = rmesa->radeonScreen->driScreen->fd;
+		rmesa->cmdbuf.csm = radeon_cs_manager_gem_ctor(fd);
+	} else {
+		rmesa->cmdbuf.csm = r600_radeon_cs_manager_legacy_ctor(rmesa);
+	}
+	if (rmesa->cmdbuf.csm == NULL) {
+		/* FIXME: fatal error */
+		return;
+	}
+	rmesa->cmdbuf.cs = radeon_cs_create(rmesa->cmdbuf.csm, size);
+	assert(rmesa->cmdbuf.cs != NULL);
+	rmesa->cmdbuf.size = size;
+	
+	if (!rmesa->radeonScreen->kernel_mm) {
+		radeon_cs_set_limit(rmesa->cmdbuf.cs, RADEON_GEM_DOMAIN_VRAM, rmesa->radeonScreen->texSize[0]);
+		radeon_cs_set_limit(rmesa->cmdbuf.cs, RADEON_GEM_DOMAIN_GTT, rmesa->radeonScreen->gartTextures.size);
+	} else {
+		struct drm_radeon_gem_info mminfo;
+
+		if (!drmCommandWriteRead(rmesa->dri.fd, DRM_RADEON_GEM_INFO, &mminfo, sizeof(mminfo)))
+		{
+			radeon_cs_set_limit(rmesa->cmdbuf.cs, RADEON_GEM_DOMAIN_VRAM, mminfo.vram_visible);
+			radeon_cs_set_limit(rmesa->cmdbuf.cs, RADEON_GEM_DOMAIN_GTT, mminfo.gart_size);
+		}
+	}
 }
diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.h b/src/mesa/drivers/dri/r600/r600_cmdbuf.h
index 1bab9c5821..dbb723b2b5 100644
--- a/src/mesa/drivers/dri/r600/r600_cmdbuf.h
+++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.h
@@ -37,6 +37,173 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define __R600_CMDBUF_H__
 
 #include "r600_context.h"
+#include "r600_emit.h"
+
+#define RADEON_CP_PACKET3_NOP                       0xC0001000
+#define RADEON_CP_PACKET3_NEXT_CHAR                 0xC0001900
+#define RADEON_CP_PACKET3_PLY_NEXTSCAN              0xC0001D00
+#define RADEON_CP_PACKET3_SET_SCISSORS              0xC0001E00
+#define RADEON_CP_PACKET3_3D_RNDR_GEN_INDX_PRIM     0xC0002300
+#define RADEON_CP_PACKET3_LOAD_MICROCODE            0xC0002400
+#define RADEON_CP_PACKET3_WAIT_FOR_IDLE             0xC0002600
+#define RADEON_CP_PACKET3_3D_DRAW_VBUF              0xC0002800
+#define RADEON_CP_PACKET3_3D_DRAW_IMMD              0xC0002900
+#define RADEON_CP_PACKET3_3D_DRAW_INDX              0xC0002A00
+#define RADEON_CP_PACKET3_LOAD_PALETTE              0xC0002C00
+#define RADEON_CP_PACKET3_3D_LOAD_VBPNTR            0xC0002F00
+#define RADEON_CP_PACKET3_CNTL_PAINT                0xC0009100
+#define RADEON_CP_PACKET3_CNTL_BITBLT               0xC0009200
+#define RADEON_CP_PACKET3_CNTL_SMALLTEXT            0xC0009300
+#define RADEON_CP_PACKET3_CNTL_HOSTDATA_BLT         0xC0009400
+#define RADEON_CP_PACKET3_CNTL_POLYLINE             0xC0009500
+#define RADEON_CP_PACKET3_CNTL_POLYSCANLINES        0xC0009800
+#define RADEON_CP_PACKET3_CNTL_PAINT_MULTI          0xC0009A00
+#define RADEON_CP_PACKET3_CNTL_BITBLT_MULTI         0xC0009B00
+#define RADEON_CP_PACKET3_CNTL_TRANS_BITBLT         0xC0009C00
+
+/* r6xx/r7xx packet 3 type offsets */
+#define R600_SET_CONFIG_REG_OFFSET                  0x00008000
+#define R600_SET_CONFIG_REG_END                     0x0000ac00
+#define R600_SET_CONTEXT_REG_OFFSET                 0x00028000
+#define R600_SET_CONTEXT_REG_END                    0x00029000
+#define R600_SET_ALU_CONST_OFFSET                   0x00030000
+#define R600_SET_ALU_CONST_END                      0x00032000
+#define R600_SET_RESOURCE_OFFSET                    0x00038000
+#define R600_SET_RESOURCE_END                       0x0003c000
+#define R600_SET_SAMPLER_OFFSET                     0x0003c000
+#define R600_SET_SAMPLER_END                        0x0003cff0
+#define R600_SET_CTL_CONST_OFFSET                   0x0003cff0
+#define R600_SET_CTL_CONST_END                      0x0003e200
+#define R600_SET_LOOP_CONST_OFFSET                  0x0003e200
+#define R600_SET_LOOP_CONST_END                     0x0003e380
+#define R600_SET_BOOL_CONST_OFFSET                  0x0003e380
+#define R600_SET_BOOL_CONST_END                     0x00040000
+
+/* r6xx/r7xx packet 3 types */
+#define R600_IT_INDIRECT_BUFFER_END               0x00001700
+#define R600_IT_SET_PREDICATION                   0x00002000
+#define R600_IT_REG_RMW                           0x00002100
+#define R600_IT_COND_EXEC                         0x00002200
+#define R600_IT_PRED_EXEC                         0x00002300
+#define R600_IT_START_3D_CMDBUF                   0x00002400
+#define R600_IT_DRAW_INDEX_2                      0x00002700
+#define R600_IT_CONTEXT_CONTROL                   0x00002800
+#define R600_IT_DRAW_INDEX_IMMD_BE                0x00002900
+#define R600_IT_INDEX_TYPE                        0x00002A00
+#define R600_IT_DRAW_INDEX                        0x00002B00
+#define R600_IT_DRAW_INDEX_AUTO                   0x00002D00
+#define R600_IT_DRAW_INDEX_IMMD                   0x00002E00
+#define R600_IT_NUM_INSTANCES                     0x00002F00
+#define R600_IT_STRMOUT_BUFFER_UPDATE             0x00003400
+#define R600_IT_INDIRECT_BUFFER_MP                0x00003800
+#define R600_IT_MEM_SEMAPHORE                     0x00003900
+#define R600_IT_MPEG_INDEX                        0x00003A00
+#define R600_IT_WAIT_REG_MEM                      0x00003C00
+#define R600_IT_MEM_WRITE                         0x00003D00
+#define R600_IT_INDIRECT_BUFFER                   0x00003200
+#define R600_IT_CP_INTERRUPT                      0x00004000
+#define R600_IT_SURFACE_SYNC                      0x00004300
+#define R600_IT_ME_INITIALIZE                     0x00004400
+#define R600_IT_COND_WRITE                        0x00004500
+#define R600_IT_EVENT_WRITE                       0x00004600
+#define R600_IT_EVENT_WRITE_EOP                   0x00004700
+#define R600_IT_ONE_REG_WRITE                     0x00005700
+#define R600_IT_SET_CONFIG_REG                    0x00006800
+#define R600_IT_SET_CONTEXT_REG                   0x00006900
+#define R600_IT_SET_ALU_CONST                     0x00006A00
+#define R600_IT_SET_BOOL_CONST                    0x00006B00
+#define R600_IT_SET_LOOP_CONST                    0x00006C00
+#define R600_IT_SET_RESOURCE                      0x00006D00
+#define R600_IT_SET_SAMPLER                       0x00006E00
+#define R600_IT_SET_CTL_CONST                     0x00006F00
+#define R600_IT_SURFACE_BASE_UPDATE               0x00007300
+
+extern int r600_cs_write_reloc(struct radeon_cs *cs,
+                        struct radeon_bo *bo,
+                        uint32_t read_domain,
+                        uint32_t write_domain,
+                        uint32_t flags,
+                        offset_modifiers* poffset_mod);
+
+static inline void r600_cs_write_dword(struct radeon_cs *cs, uint32_t dword)
+{
+    cs->packets[cs->cdw++] = dword;
+}
+
+/**
+ * Write one dword to the command buffer.
+ */
+#define R600_OUT_BATCH(data) \
+	do { \
+        r600_cs_write_dword(b_l_rmesa->cmdbuf.cs, data);\
+	} while(0)
+
+/**
+ * Write n dwords from ptr to the command buffer.
+ */
+#define R600_OUT_BATCH_TABLE(ptr,n) \
+	do { \
+		int _i; \
+        for (_i=0; _i < n; _i++) {\
+            r600_cs_write_dword(b_l_rmesa->cmdbuf.cs, ptr[_i]);\
+        }\
+	} while(0)
+
+/**
+ * Write a relocated dword to the command buffer.
+ */
+#define R600_OUT_BATCH_RELOC(data, bo, offset, rd, wd, flags, offset_mod) 	\
+	do { 							\
+        if (0 && offset) {					\
+            fprintf(stderr, "(%s:%s:%d) offset : %d\n",		\
+            __FILE__, __FUNCTION__, __LINE__, offset);		\
+        }							\
+        r600_cs_write_dword(b_l_rmesa->cmdbuf.cs, offset);	\
+        r600_cs_write_reloc(b_l_rmesa->cmdbuf.cs, 		\
+                              bo, rd, wd, flags, offset_mod);		\
+	} while(0)
+
+/* R600/R700 */
+#define R600_OUT_BATCH_REGS(reg, num)					\
+do {								\
+	if ((reg) >= R600_SET_CONFIG_REG_OFFSET && (reg) < R600_SET_CONFIG_REG_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_CONFIG_REG_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_CONTEXT_REG_OFFSET && (reg) < R600_SET_CONTEXT_REG_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_CONTEXT_REG, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_CONTEXT_REG_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_ALU_CONST_OFFSET && (reg) < R600_SET_ALU_CONST_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_ALU_CONST_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_RESOURCE_OFFSET && (reg) < R600_SET_RESOURCE_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_RESOURCE_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_SAMPLER_OFFSET && (reg) < R600_SET_SAMPLER_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_SAMPLER_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_CTL_CONST_OFFSET && (reg) < R600_SET_CTL_CONST_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_CTL_CONST_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_LOOP_CONST_OFFSET && (reg) < R600_SET_LOOP_CONST_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_LOOP_CONST, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_LOOP_CONST_OFFSET) >> 2);	\
+	} else if ((reg) >= R600_SET_BOOL_CONST_OFFSET && (reg) < R600_SET_BOOL_CONST_END) { \
+		OUT_BATCH(CP_PACKET3(R600_IT_SET_BOOL_CONST, (num)));	\
+		OUT_BATCH(((reg) - R600_SET_BOOL_CONST_OFFSET) >> 2);	\
+	} else {							\
+		OUT_BATCH(CP_PACKET0((reg), (num))); \
+	}								\
+} while (0)
+
+/** Single register write to command buffer; requires 3 dwords for most things. */
+#define R600_OUT_BATCH_REGVAL(reg, val)		\
+	R600_OUT_BATCH_REGS((reg), 1);		\
+	OUT_BATCH((val))
+
+/** Continuous register range write to command buffer; requires 1 dword,
+ * expects count dwords afterwards for register contents. */
+#define R600_OUT_BATCH_REGSEQ(reg, count)	\
+	R600_OUT_BATCH_REGS((reg), (count))
 
 extern void r600InitCmdBuf(r600ContextPtr r600);
 
diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c
index 1b43f68ac2..3c8ba003d4 100644
--- a/src/mesa/drivers/dri/r600/r600_context.c
+++ b/src/mesa/drivers/dri/r600/r600_context.c
@@ -156,42 +156,6 @@ const struct dri_extension gl_20_extension[] = {
 };
 
 
-extern struct tnl_pipeline_stage _r600_render_stage;
-extern const struct tnl_pipeline_stage _r600_tcl_stage;
-
-static const struct tnl_pipeline_stage *r600_pipeline[] = {
-
-	/* Try and go straight to t&l
-	 */
-	&_r600_tcl_stage,
-
-	/* Catch any t&l fallbacks
-	 */
-	&_tnl_vertex_transform_stage,
-	&_tnl_normal_transform_stage,
-	&_tnl_lighting_stage,
-	&_tnl_fog_coordinate_stage,
-	&_tnl_texgen_stage,
-	&_tnl_texture_transform_stage,
-	&_tnl_vertex_program_stage,
-
-	/* Try again to go to tcl?
-	 *     - no good for asymmetric-twoside (do with multipass)
-	 *     - no good for asymmetric-unfilled (do with multipass)
-	 *     - good for material
-	 *     - good for texgen
-	 *     - need to manipulate a bit of state
-	 *
-	 * - worth it/not worth it?
-	 */
-
-	/* Else do them here.
-	 */
-	&_r600_render_stage,
-	&_tnl_render_stage,	/* FALLBACK  */
-	0,
-};
-
 static void r600RunPipeline(GLcontext * ctx)
 {
     _mesa_lock_context_textures(ctx);
@@ -245,31 +209,6 @@ static void r600_init_vtbl(radeonContextPtr radeon)
 }
 
 /* to be enabled */
-static void r600EmitShader(GLcontext * ctx, 
-                   struct r600_dma_region *rvb,
-			       GLvoid * data, 
-                   int sizeinDWORD) 
-{
-}
-/* to be enabled */
-static void r600FreeDmaRegion(context_t *context, 
-                              struct r600_dma_region *region)
-{
-}
-/* to be enabled */
-static void r600EmitVec(GLcontext * ctx, 
-                 struct r600_dma_region *rvb,
-			     GLvoid * data, 
-                 int size, 
-                 int stride, 
-                 int count)
-{
-}
-/* to be enabled */
-static void r600ReleaseArrays(GLcontext * ctx)
-{
-}
-/* to be enabled */
 static GLboolean r600LoadMemSurf(context_t *context,
                                GLuint     dst_offset, /* gpu addr */
                                GLuint     dst_pitch_in_pixel,                               
@@ -287,6 +226,7 @@ static GLboolean r600AllocMemSurf(context_t   *context,
                            GLuint      *prefered_heap, /* Now used RADEON_LOCAL_TEX_HEAP, return actual heap used. */
                            GLuint       totalSize)
 {
+    return GL_TRUE;
 }
 /* to be enabled */
 static int  r600FlushCmdBuffer(context_t *context)
@@ -332,6 +272,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 	 */
 	driParseConfigFiles(&r600->radeon.optionCache, &screen->optionCache,
 			    screen->driScreen->myNum, "r600");
+
 	r600->radeon.initialMaxAnisotropy = driQueryOptionf(&r600->radeon.optionCache,
 						     "def_max_anisotropy");
 
@@ -344,9 +285,10 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 
     (r600->chipobj.InitFuncs)(&functions);
     r600->chipobj.EmitShader     = r600EmitShader;
+    r600->chipobj.DeleteShader   = r600DeleteShader;
     r600->chipobj.FreeDmaRegion  = r600FreeDmaRegion;
     r600->chipobj.EmitVec        = r600EmitVec;
-    r600->chipobj.ReleaseArrays  = r600ReleaseArrays;
+    r600->chipobj.ReleaseArrays  = r600ReleaseVec;
     r600->chipobj.LoadMemSurf    = r600LoadMemSurf;
     r600->chipobj.AllocMemSurf   = r600AllocMemSurf;
     r600->chipobj.FlushCmdBuffer = r600FlushCmdBuffer;
@@ -476,8 +418,9 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
    	radeonInitSpanFuncs( ctx );
 
 	r600InitCmdBuf(r600);
+
+    (r600->chipobj.InitState)(r600->radeon.glCtx);
 #if 0 /* to be enabled */
-	r600InitState(r600);
 	if (!(screen->chip_flags & RADEON_CHIPSET_TCL))
 	        r600InitSwtcl(ctx);
 #endif
diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h
index 74e34e3684..70108b74c4 100644
--- a/src/mesa/drivers/dri/r600/r600_context.h
+++ b/src/mesa/drivers/dri/r600/r600_context.h
@@ -140,26 +140,19 @@ struct r600_swtcl_info {
    GLubyte vertex_attr_count;
 };
 
-/* to be enabled */ /* DELETE FOLLOWING 2 SRTUCTS */
-struct r600_dma_buffer 
+enum 
 {
-    int         refcount;       /* the number of retained regions in buf */
-    drmBufPtr   buf;
-    int         id;
+    NO_SHIFT    = 0,
+    LEFT_SHIFT  = 1,
+    RIGHT_SHIFT = 2,
 };
-struct r600_dma_region 
+
+typedef struct offset_modifiers
 {
-    struct r600_dma_buffer  *buf;
-    char        *address;       /* == buf->address */
-    int         start;
-    int         end;
-    int         ptr;            /* offsets from start of buf */
-
-    int         aos_offset;     /* address in GART memory */
-    int         aos_stride;     /* distance between elements, in dwords */
-    int         aos_size;       /* number of components (1-4) */
-};
-/* ----------------------- */
+    GLuint shift;
+    GLuint shiftbits;
+    GLuint mask;
+} offset_modifiers;
 
 typedef struct chip_object
 {
@@ -175,14 +168,16 @@ typedef struct chip_object
     GLuint    (*GetTexObjSize)(void);
 
     /* ------------  IN  ------------------- */
-    void      (*EmitShader)( GLcontext * ctx, 
-                             struct r600_dma_region *rvb,
+    GLboolean (*EmitShader)( GLcontext * ctx, 
+                             void ** shaderbo,
 			                 GLvoid * data, 
                              int sizeinDWORD);
+    GLboolean (*DeleteShader)(GLcontext * ctx, 
+                              void * shaderbo);
     void      (*FreeDmaRegion)( GLcontext * ctx, 
-                                struct r600_dma_region *region);
-    void      (*EmitVec)(GLcontext * ctx, 
-                         struct r600_dma_region *rvb,
+                                void * shaderbo);
+    GLboolean (*EmitVec)(GLcontext * ctx, 
+                         struct radeon_aos *aos,
 			             GLvoid * data, 
                          int size, 
                          int stride, 
diff --git a/src/mesa/drivers/dri/r600/r600_emit.c b/src/mesa/drivers/dri/r600/r600_emit.c
index 9d43873c4d..fc8f987cab 100644
--- a/src/mesa/drivers/dri/r600/r600_emit.c
+++ b/src/mesa/drivers/dri/r600/r600_emit.c
@@ -1,9 +1,8 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
+/**************************************************************************
+
+Copyright 2008, 2009 Advanced Micro Devices Inc. (AMD)
 
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
+Copyright (C) Advanced Micro Devices Inc. (AMD)  2009.  All Rights Reserved.
 
 Permission is hereby granted, free of charge, to any person obtaining
 a copy of this software and associated documentation files (the
@@ -27,10 +26,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 **************************************************************************/
 
-/**
- * \file
- *
- * \author Keith Whitwell <keith@tungstengraphics.com>
+/*
+ * Authors:
+ *   Richard Li <RichardZ.Li@amd.com>, <richardradeon@gmail.com>
+ *   CooperYuan <cooper.yuan@amd.com>, <cooperyuan@gmail.com>
  */
 
 #include "main/glheader.h"
@@ -48,6 +47,105 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r600_context.h"
 #include "r600_emit.h"
 
+#if defined(USE_X86_ASM)
+#define COPY_DWORDS( dst, src, nr )					\
+do {									\
+	int __tmp;							\
+	__asm__ __volatile__( "rep ; movsl"				\
+			      : "=%c" (__tmp), "=D" (dst), "=S" (__tmp)	\
+			      : "0" (nr),				\
+			        "D" ((long)dst),			\
+			        "S" ((long)src) );			\
+} while (0)
+#else
+#define COPY_DWORDS( dst, src, nr )		\
+do {						\
+   int j;					\
+   for ( j = 0 ; j < nr ; j++ )			\
+      dst[j] = ((int *)src)[j];			\
+   dst += nr;					\
+} while (0)
+#endif
+
+static void r600EmitVec4(uint32_t *out, GLvoid * data, int stride, int count)
+{
+	int i;
+
+	if (RADEON_DEBUG & DEBUG_VERTS)
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
+
+	if (stride == 4)
+		COPY_DWORDS(out, data, count);
+	else
+		for (i = 0; i < count; i++) {
+			out[0] = *(int *)data;
+			out++;
+			data += stride;
+		}
+}
+
+static void r600EmitVec8(uint32_t *out, GLvoid * data, int stride, int count)
+{
+	int i;
+
+	if (RADEON_DEBUG & DEBUG_VERTS)
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
+
+	if (stride == 8)
+		COPY_DWORDS(out, data, count * 2);
+	else
+		for (i = 0; i < count; i++) {
+			out[0] = *(int *)data;
+			out[1] = *(int *)(data + 4);
+			out += 2;
+			data += stride;
+		}
+}
+
+static void r600EmitVec12(uint32_t *out, GLvoid * data, int stride, int count)
+{
+	int i;
+
+	if (RADEON_DEBUG & DEBUG_VERTS)
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
+
+	if (stride == 12) {
+		COPY_DWORDS(out, data, count * 3);
+    }
+	else
+		for (i = 0; i < count; i++) {
+			out[0] = *(int *)data;
+			out[1] = *(int *)(data + 4);
+			out[2] = *(int *)(data + 8);
+			out += 3;
+			data += stride;
+		}
+}
+
+static void r600EmitVec16(uint32_t *out, GLvoid * data, int stride, int count)
+{
+	int i;
+
+	if (RADEON_DEBUG & DEBUG_VERTS)
+		fprintf(stderr, "%s count %d stride %d out %p data %p\n",
+			__FUNCTION__, count, stride, (void *)out, (void *)data);
+
+	if (stride == 16)
+		COPY_DWORDS(out, data, count * 4);
+	else
+		for (i = 0; i < count; i++) {
+			out[0] = *(int *)data;
+			out[1] = *(int *)(data + 4);
+			out[2] = *(int *)(data + 8);
+			out[3] = *(int *)(data + 12);
+			out += 4;
+			data += stride;
+		}
+}
+
 /* Emit vertex data to GART memory
  * Route inputs to the vertex processor
  * This function should never return R600_FALLBACK_TCL when using software tcl.
@@ -73,3 +171,111 @@ void r600EmitCacheFlush(r600ContextPtr rmesa)
 	COMMIT_BATCH();
 */
 }
+
+GLboolean r600EmitShader(GLcontext * ctx, 
+                         void ** shaderbo,
+			             GLvoid * data, 
+                         int sizeinDWORD) 
+{
+    radeonContextPtr radeonctx = RADEON_CONTEXT(ctx);
+
+    struct radeon_bo * pbo;
+    uint32_t *out;
+
+shader_again_alloc:	
+	pbo = radeon_bo_open(radeonctx->radeonScreen->bom,
+					     0, 
+                         sizeinDWORD * 4, 
+                         256, 
+                         RADEON_GEM_DOMAIN_GTT,
+					     0);
+
+	if (!pbo) 
+    {
+		rcommonFlushCmdBuf(radeonctx, __FUNCTION__);
+		goto shader_again_alloc;
+	}
+
+	radeon_validate_bo(radeonctx, pbo, RADEON_GEM_DOMAIN_GTT, 0);
+
+	if (radeon_revalidate_bos(radeonctx->glCtx) == GL_FALSE)
+    {
+	    fprintf(stderr,"failure to revalidate BOs - badness\n");
+    }
+	  
+	radeon_bo_map(pbo, 1);
+
+    radeon_bo_ref(pbo);
+
+    out = (uint32_t*)(pbo->ptr);
+
+    memcpy(out, data, sizeinDWORD * 4);
+
+    *shaderbo = (void*)pbo;
+
+    return GL_TRUE;
+}
+
+GLboolean r600DeleteShader(GLcontext * ctx, 
+                           void * shaderbo) 
+{
+    struct radeon_bo * pbo = (struct radeon_bo *)shaderbo;
+
+    radeon_bo_unmap(pbo);
+    radeon_bo_unref(pbo); /* when bo->cref <= 0, bo will be bo_free */
+
+    return GL_TRUE;
+}
+
+GLboolean r600EmitVec(GLcontext * ctx, 
+                      struct radeon_aos *aos,
+			          GLvoid * data, 
+                      int size, 
+                      int stride, 
+                      int count)
+{
+    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
+	uint32_t *out;
+
+	if (stride == 0) 
+    {
+		radeonAllocDmaRegion(rmesa, &aos->bo, &aos->offset, size * count * 4, 32);
+		aos->stride = 0;
+	} 
+    else 
+    {
+		radeonAllocDmaRegion(rmesa, &aos->bo, &aos->offset, size * count * 4, 32);
+		aos->stride = size;
+	}
+
+	aos->components = size;
+	aos->count = count;
+
+	out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
+	switch (size) {
+	case 1: r600EmitVec4(out, data, stride, count); break;
+	case 2: r600EmitVec8(out, data, stride, count); break;
+	case 3: r600EmitVec12(out, data, stride, count); break;
+	case 4: r600EmitVec16(out, data, stride, count); break;
+	default:
+		assert(0);
+		break;
+	}
+
+    return GL_TRUE;
+}
+
+void r600ReleaseVec(GLcontext * ctx)
+{
+    radeonReleaseArrays(ctx, ~0);
+}
+
+void r600FreeDmaRegion(context_t *context, 
+                       void * shaderbo)
+{
+    struct radeon_bo *pbo = (struct radeon_bo *)shaderbo;
+    if(pbo) 
+    {
+        radeon_bo_unref(pbo);
+    }
+}
diff --git a/src/mesa/drivers/dri/r600/r600_emit.h b/src/mesa/drivers/dri/r600/r600_emit.h
index 3e1cf1ff5f..11765c868c 100644
--- a/src/mesa/drivers/dri/r600/r600_emit.h
+++ b/src/mesa/drivers/dri/r600/r600_emit.h
@@ -1,41 +1,38 @@
-/*
- * Copyright (C) 2005 Vladimir Dergachev.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (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 COPYRIGHT OWNER(S) 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.
- *
- */
+/**************************************************************************
+
+Copyright 2008, 2009 Advanced Micro Devices Inc. (AMD)
+
+Copyright (C) Advanced Micro Devices Inc. (AMD)  2009.  All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (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 COPYRIGHT OWNER(S) 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.
+
+**************************************************************************/
 
 /*
  * Authors:
- *   Vladimir Dergachev <volodya@mindspring.com>
- *   Nicolai Haehnle <prefect_@gmx.net>
- *   Aapo Tahkola <aet@rasterburn.org>
- *   Ben Skeggs <darktama@iinet.net.au>
- *   Jerome Glisse <j.glisse@gmail.com>
+ *   Richard Li <RichardZ.Li@amd.com>, <richardradeon@gmail.com>
+ *   CooperYuan <cooper.yuan@amd.com>, <cooperyuan@gmail.com>
  */
 
-/* This files defines functions for accessing R600 hardware.
- */
+
 #ifndef __R600_EMIT_H__
 #define __R600_EMIT_H__
 
@@ -44,180 +41,24 @@
 #include "r600_cmdbuf.h"
 #include "radeon_reg.h"
 
-static INLINE uint32_t cmdpacket0(struct radeon_screen *rscrn,
-                                  int reg, int count)
-{
-	if (count) {
-		return CP_PACKET0(reg, count - 1);
-	}
-	return CP_PACKET2;
-}
-
-static INLINE uint32_t cmdvpu(struct radeon_screen *rscrn, int addr, int count)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-	cmd.vpu.cmd_type = R300_CMD_VPU;
-	cmd.vpu.count = count;
-	cmd.vpu.adrhi = ((unsigned int)addr & 0xFF00) >> 8;
-	cmd.vpu.adrlo = ((unsigned int)addr & 0x00FF);
-
-	return cmd.u;
-}
-
-static INLINE uint32_t cmdr500fp(struct radeon_screen *rscrn,
-                                 int addr, int count, int type, int clamp)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-	cmd.r500fp.cmd_type = R300_CMD_R500FP;
-	cmd.r500fp.count = count;
-	cmd.r500fp.adrhi_flags = ((unsigned int)addr & 0x100) >> 8;
-	cmd.r500fp.adrhi_flags |= type ? R500FP_CONSTANT_TYPE : 0;
-	cmd.r500fp.adrhi_flags |= clamp ? R500FP_CONSTANT_CLAMP : 0;
-	cmd.r500fp.adrlo = ((unsigned int)addr & 0x00FF);
-
-	return cmd.u;
-}
-
-static INLINE uint32_t cmdpacket3(struct radeon_screen *rscrn, int packet)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-	cmd.packet3.cmd_type = R300_CMD_PACKET3;
-	cmd.packet3.packet = packet;
-
-	return cmd.u;
-}
-
-static INLINE uint32_t cmdcpdelay(struct radeon_screen *rscrn,  
-                                  unsigned short count)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-
-	cmd.delay.cmd_type = R300_CMD_CP_DELAY;
-	cmd.delay.count = count;
-
-	return cmd.u;
-}
-
-static INLINE uint32_t cmdwait(struct radeon_screen *rscrn,
-                               unsigned char flags)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-	cmd.wait.cmd_type = R300_CMD_WAIT;
-	cmd.wait.flags = flags;
-
-	return cmd.u;
-}
-
-static INLINE uint32_t cmdpacify(struct radeon_screen *rscrn)
-{
-	drm_r300_cmd_header_t cmd;
-
-	cmd.u = 0;
-	cmd.header.cmd_type = R300_CMD_END3D;
-
-	return cmd.u;
-}
-
-/**
- * Write the header of a packet3 to the command buffer.
- * Outputs 2 dwords and expects (num_extra+1) additional dwords afterwards.
- */
-#define OUT_BATCH_PACKET3(packet, num_extra) do {\
-    if (!b_l_rmesa->radeonScreen->kernel_mm) {		\
-    	OUT_BATCH(cmdpacket3(b_l_rmesa->radeonScreen,\
-                  R300_CMD_PACKET3_RAW)); \
-    } else b_l_rmesa->cmdbuf.cs->section_cdw++;\
-	OUT_BATCH(CP_PACKET3((packet), (num_extra))); \
-	} while(0)
-
-/**
- * Must be sent to switch to 2d commands
- */
-void static INLINE end_3d(radeonContextPtr radeon)
-{
-	BATCH_LOCALS(radeon);
-
-	if (!radeon->radeonScreen->kernel_mm) {
-		BEGIN_BATCH_NO_AUTOSTATE(1);
-		OUT_BATCH(cmdpacify(radeon->radeonScreen));
-		END_BATCH();
-	}
-}
-
-void static INLINE cp_delay(r600ContextPtr rmesa, unsigned short count)
-{
-	BATCH_LOCALS(&rmesa->radeon);
-
-	if (!rmesa->radeon.radeonScreen->kernel_mm) {
-		BEGIN_BATCH_NO_AUTOSTATE(1);
-		OUT_BATCH(cmdcpdelay(rmesa->radeon.radeonScreen, count));
-		END_BATCH();
-	}
-}
-
-void static INLINE cp_wait(radeonContextPtr radeon, unsigned char flags)
-{
-	BATCH_LOCALS(radeon);
-	uint32_t wait_until;
-
-	if (!radeon->radeonScreen->kernel_mm) {
-		BEGIN_BATCH_NO_AUTOSTATE(1);
-		OUT_BATCH(cmdwait(radeon->radeonScreen, flags));
-		END_BATCH();
-	} else {
-		switch(flags) {
-		case R300_WAIT_2D:
-			wait_until = (1 << 14);
-			break;
-		case R300_WAIT_3D:
-			wait_until = (1 << 15);
-			break;
-		case R300_NEW_WAIT_2D_3D:
-			wait_until = (1 << 14) | (1 << 15);
-			break;
-		case R300_NEW_WAIT_2D_2D_CLEAN:
-			wait_until = (1 << 14) | (1 << 16) | (1 << 18);
-			break;
-		case R300_NEW_WAIT_3D_3D_CLEAN:
-			wait_until = (1 << 15) | (1 << 17) | (1 << 18);
-			break;
-		case R300_NEW_WAIT_2D_2D_CLEAN_3D_3D_CLEAN:
-			wait_until  = (1 << 14) | (1 << 16) | (1 << 18);
-			wait_until |= (1 << 15) | (1 << 17) | (1 << 18);
-			break;
-		default:
-			return;
-		}
-		BEGIN_BATCH_NO_AUTOSTATE(2);
-		OUT_BATCH(CP_PACKET0(RADEON_WAIT_UNTIL, 0));
-		OUT_BATCH(wait_until);
-		END_BATCH();
-	}
-}
-
-extern int r600EmitArrays(GLcontext * ctx);
-
-extern int r600PrimitiveType(r600ContextPtr rmesa, int prim);
-extern int r600NumVerts(r600ContextPtr rmesa, int num_verts, int prim);
-
-extern void r600EmitCacheFlush(r600ContextPtr rmesa);
-
-extern GLuint r600VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
-				 int *inputs, GLint * tab, GLuint nr);
-extern GLuint r600VAPInputRoute1(uint32_t * dst, int swizzle[][4], GLuint nr);
-extern GLuint r600VAPInputCntl0(GLcontext * ctx, GLuint InputsRead);
-extern GLuint r600VAPInputCntl1(GLcontext * ctx, GLuint InputsRead);
-extern GLuint r600VAPOutputCntl0(GLcontext * ctx, GLuint OutputsWritten);
-extern GLuint r600VAPOutputCntl1(GLcontext * ctx, GLuint OutputsWritten);
+extern GLboolean r600EmitShader(GLcontext * ctx, 
+                                void ** shaderbo,
+			                    GLvoid * data, 
+                                int sizeinDWORD); 
+
+extern GLboolean r600DeleteShader(GLcontext * ctx, 
+                                 void * shaderbo);
+
+extern GLboolean r600EmitVec(GLcontext * ctx, 
+                             struct radeon_aos *aos,
+			                 GLvoid * data, 
+                             int size, 
+                             int stride, 
+                             int count);
+
+extern void      r600ReleaseVec(GLcontext * ctx);
+
+extern void r600FreeDmaRegion(context_t *context, 
+                              void * shaderbo);
 
 #endif
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index d0371cdda0..e534bb11b2 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -29,15 +29,15 @@
 #include "main/glheader.h"
 
 #include "r600_context.h"
+#include "r600_cmdbuf.h"
 
 #include "r700_chip.h"
 #include "r700_state.h"
 #include "r700_tex.h"
 #include "r700_oglprog.h"
+#include "r700_fragprog.h"
+#include "r700_vertprog.h"
 #include "r700_ioctl.h"
-/* to be enable
-#include "r700_emit.h"
-*/
 
 extern const struct tnl_pipeline_stage *r700_pipeline[];
 
@@ -370,15 +370,235 @@ GLboolean r700InitChipObject(context_t *context)
     return GL_TRUE;
 }
 
-GLboolean r700SendContextStates(context_t *context)
+void r700SetupVTXConstans(GLcontext  * ctx, 
+                          unsigned int nStreamID,
+                          void *       pAos,
+                          unsigned int size,      /* number of elements in vector */
+                          unsigned int stride,
+                          unsigned int count)     /* number of vectors in stream */
 {
-#if 0 //to be enable
+    context_t *context = R700_CONTEXT(ctx);
+    uint32_t *dest;
+    struct radeon_aos * paos = (struct radeon_aos *)pAos;
+    offset_modifiers offset_mod = {NO_SHIFT, 0, 0xFFFFFFFF};
+
+    BATCH_LOCALS(&context->radeon);
+
+    unsigned int uSQ_VTX_CONSTANT_WORD0_0;
+    unsigned int uSQ_VTX_CONSTANT_WORD1_0;
+    unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0;
+    unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0;
+    unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0;
+
+    uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
+	uSQ_VTX_CONSTANT_WORD1_0 = count * stride - 1;
+
+	uSQ_VTX_CONSTANT_WORD2_0 |= 0 << BASE_ADDRESS_HI_shift /* TODO */
+                             |stride << SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift	
+	                         |GetSurfaceFormat(GL_FLOAT, size, NULL) << SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift /* TODO : trace back api for initial data type, not only GL_FLOAT */
+	                         |SQ_NUM_FORMAT_SCALED << SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift
+	                         |SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit;
+	
+	uSQ_VTX_CONSTANT_WORD3_0 |= 1 << MEM_REQUEST_SIZE_shift;
+	
+	uSQ_VTX_CONSTANT_WORD6_0 |= SQ_TEX_VTX_VALID_BUFFER << SQ_TEX_RESOURCE_WORD6_0__TYPE_shift;
+
+    BEGIN_BATCH_NO_AUTOSTATE(9);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); 
+    OUT_BATCH((nStreamID + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
+
+    R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD1_0, 
+                         paos->bo, 
+                         uSQ_VTX_CONSTANT_WORD1_0,					                 
+                         RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod);	
+	OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0);
+	OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0);
+	OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0);
+	OUT_BATCH(0);
+	OUT_BATCH(0);
+	OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0);
+
+    END_BATCH();
+    COMMIT_BATCH();
+    
+}
+
+int r700SetupStreams(GLcontext * ctx)
+{
+    context_t         *context = R700_CONTEXT(ctx);
+
+    BATCH_LOCALS(&context->radeon);
+
+    struct r700_vertex_program *vpc
+             = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+
+    TNLcontext *tnl = TNL_CONTEXT(ctx);
+	struct vertex_buffer *vb = &tnl->vb;
+
+    unsigned int unBit;
+	unsigned int i;
+
+    BEGIN_BATCH_NO_AUTOSTATE(6);
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1));
+	OUT_BATCH(mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX);
+    OUT_BATCH(0);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1));
+    OUT_BATCH(mmSQ_VTX_START_INST_LOC - ASIC_CTL_CONST_BASE_INDEX);
+    OUT_BATCH(0);
+    END_BATCH();
+    COMMIT_BATCH();
+
+    //context->aos_count = 0;
+	for(i=0; i<VERT_ATTRIB_MAX; i++)
+	{
+		unBit = 1 << i;
+		if(vpc->mesa_program.Base.InputsRead & unBit) 
+		{            
+            (context->chipobj.EmitVec)(ctx, 
+                                       &context->radeon.tcl.aos[i],
+				                       vb->AttribPtr[i]->data,
+				                       vb->AttribPtr[i]->size,
+				                       vb->AttribPtr[i]->stride, 
+                                       vb->Count);
+
+            /* currently aos are packed */
+            r700SetupVTXConstans(ctx, 
+                                 i,
+                                 (void*)(&context->radeon.tcl.aos[i]),
+                                 (unsigned int)vb->AttribPtr[i]->size,
+                                 (unsigned int)(vb->AttribPtr[i]->size * 4),
+                                 (unsigned int)vb->Count);
+		}
+	}
+    
+    return R600_FALLBACK_NONE;
+}
+
+inline GLboolean setRelocReg(context_t *context, unsigned int reg,
+                             void * pbo_vs, void * pbo_fs)
+{
+    BATCH_LOCALS(&context->radeon);
+    R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+
+    struct radeon_bo * pbo;
+    uint32_t voffset;
+    offset_modifiers offset_mod;
+
+    switch (reg + ASIC_CONTEXT_BASE_INDEX) 
+    {
+        case mmCB_COLOR0_BASE:
+        case mmCB_COLOR1_BASE:
+        case mmCB_COLOR2_BASE:
+        case mmCB_COLOR3_BASE:
+        case mmCB_COLOR4_BASE:
+        case mmCB_COLOR5_BASE:
+        case mmCB_COLOR6_BASE:
+        case mmCB_COLOR7_BASE:
+            {
+                GLcontext *ctx = GL_CONTEXT(context);
+                struct radeon_renderbuffer *rrb;
+
+                rrb = radeon_get_colorbuffer(&context->radeon);
+                if (!rrb || !rrb->bo) 
+                {
+                    fprintf(stderr, "no rrb\n");
+                    return GL_FALSE;
+                }
+
+                offset_mod.shift     = RIGHT_SHIFT;
+                offset_mod.shiftbits = 8;
+                offset_mod.mask      = 0x00FFFFFF;
+
+                R600_OUT_BATCH_RELOC(r700->CB_COLOR0_BASE.u32All, 
+                                     rrb->bo, 
+                                     r700->CB_COLOR0_BASE.u32All, 
+                                     0, RADEON_GEM_DOMAIN_VRAM, 0, &offset_mod);
+                return GL_TRUE;
+            }
+            break;  
+        case mmDB_DEPTH_BASE:
+            {
+                GLcontext *ctx = GL_CONTEXT(context);
+                struct radeon_renderbuffer *rrb;
+                rrb = radeon_get_depthbuffer(&context->radeon);
+
+                offset_mod.shift     = RIGHT_SHIFT;
+                offset_mod.shiftbits = 8;
+                offset_mod.mask      = 0x00FFFFFF;
+
+                R600_OUT_BATCH_RELOC(r700->DB_DEPTH_BASE.u32All, 
+                                     rrb->bo, 
+                                     r700->DB_DEPTH_BASE.u32All, 
+                                     0, RADEON_GEM_DOMAIN_VRAM, 0, &offset_mod);
+
+                return GL_TRUE;
+            }
+            break;
+        case mmSQ_PGM_START_VS:   
+            {                
+                if(NULL != pbo_vs)
+                {
+                    pbo = (struct radeon_bo *)pbo_vs;
+                }
+                else
+                {
+                    pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
+                }
+
+                offset_mod.shift     = RIGHT_SHIFT;
+                offset_mod.shiftbits = 8;
+                offset_mod.mask      = 0x00FFFFFF;           
+
+                R600_OUT_BATCH_RELOC(r700->SQ_PGM_START_VS.u32All, 
+                                     pbo, 
+                                     r700->SQ_PGM_START_VS.u32All,
+					                 RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod);
+                return GL_TRUE;
+            }
+            break;
+        case mmSQ_PGM_START_FS:            
+        case mmSQ_PGM_START_ES:            
+        case mmSQ_PGM_START_GS:            
+        case mmSQ_PGM_START_PS:	
+            {
+                if(NULL != pbo_fs)
+                {
+                    pbo = (struct radeon_bo *)pbo_fs;
+                }
+                else
+                {
+                    pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
+                }
+
+                offset_mod.shift     = RIGHT_SHIFT;
+                offset_mod.shiftbits = 8;
+                offset_mod.mask      = 0x00FFFFFF;
+
+                voffset = 0;
+                R600_OUT_BATCH_RELOC(r700->SQ_PGM_START_PS.u32All, 
+                                     pbo, 
+                                     r700->SQ_PGM_START_PS.u32All,
+					                 RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod);
+                return GL_TRUE;
+            }
+            break;
+    }
+
+    return GL_FALSE;
+}
+
+GLboolean r700SendContextStates(context_t *context, void * pbo_vs, void * pbo_fs)
+{
+    BATCH_LOCALS(&context->radeon);
+
     R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
 
     ContextState * pState = r700->pStateList;
     ContextState * pInit;
     unsigned int   toSend;
-    unsigned int   ui;
+    unsigned int   ui;    
 
     while(NULL != pState)
     {
@@ -401,17 +621,21 @@ GLboolean r700SendContextStates(context_t *context)
 
         pState = pState->pNext;
 
-        R700_CMDBUF_CHECK_SPACE(toSend + 2);
-        R700EP3(context, IT_SET_CONTEXT_REG, toSend);
-        R700E32(context, pInit->unOffset);
-
+        BEGIN_BATCH_NO_AUTOSTATE(toSend + 2);        
+        R600_OUT_BATCH_REGSEQ(((pInit->unOffset + ASIC_CONTEXT_BASE_INDEX)<<2), toSend);
         for(ui=0; ui<toSend; ui++)
         {
-            R700E32(context, *(pInit->puiValue));
+            if( GL_FALSE == setRelocReg(context, (pInit->unOffset+ui), pbo_vs, pbo_fs) )
+            {
+                /* for not reloc reg. */
+                OUT_BATCH(*(pInit->puiValue));
+            }
             pInit = pInit->pNext;
         };
+        END_BATCH();
     };
-#endif //to be enable
+    COMMIT_BATCH();
+
     return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r700_chip.h b/src/mesa/drivers/dri/r600/r700_chip.h
index f08190027c..30ca2cab84 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.h
+++ b/src/mesa/drivers/dri/r600/r700_chip.h
@@ -446,13 +446,22 @@ typedef struct _R700_CHIP_CONTEXT
     ContextState * pStateList;
 
     R700_TEXTURE_STATES texture_states;
+
+    GLboolean bEnablePerspective;
 	
 } R700_CHIP_CONTEXT;
 
 #define R700_CONTEXT_STATES(context) ((R700_CHIP_CONTEXT *)(context->chipobj.pvChipObj))
 
 extern GLboolean r700InitChipObject(context_t *context);
-extern GLboolean r700SendContextStates(context_t *context);
+extern GLboolean r700SendContextStates(context_t *context, void * pbo_vs, void * pbo_fs);
+extern int       r700SetupStreams(GLcontext * ctx);
+extern void      r700SetupVTXConstans(GLcontext  * ctx, 
+                                      unsigned int nStreamID,
+                                      void *       pAos,
+                                      unsigned int size,      /* number of elements in vector */
+                                      unsigned int stride,
+                                      unsigned int Count);    /* number of vectors in stream */
 
 #endif /* _R700_CHIP_H_ */
 
diff --git a/src/mesa/drivers/dri/r600/r700_clear.c b/src/mesa/drivers/dri/r600/r700_clear.c
index 5725360dac..743875dfe2 100644
--- a/src/mesa/drivers/dri/r600/r700_clear.c
+++ b/src/mesa/drivers/dri/r600/r700_clear.c
@@ -35,23 +35,12 @@
 #include "r600_context.h"
 #include "r700_chip.h"
 
-#if 0 /* to be enabled */
 #include "r700_shaderinst.h"
-#include "r700_emit.h"
-#endif /* to be enabled */
+#include "r600_emit.h"
 
 extern void r700InitState (GLcontext * ctx);
 
-#if 0 /* to be enabled */
-extern void r700SetupVTXConstans(GLcontext  * ctx, 
-                          unsigned int nStreamID,
-                          unsigned int aos_offset,
-                          unsigned int size,      /* number of elements in vector */
-                          unsigned int stride,
-                          unsigned int count);
-extern GLboolean r700SendContextStates(context_t *context);
 extern GLboolean r700SyncSurf(context_t *context);
-#endif /* to be enabled */
 
 static GLboolean r700ClearFast(context_t *context, GLbitfield mask)
 {
@@ -63,17 +52,15 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
 {
     GLcontext *ctx = GL_CONTEXT(context);
 
-#if 0 /* to be enabled */
+    BATCH_LOCALS(&context->radeon);
+
     R700_CHIP_CONTEXT  r700Saved;
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
-    struct r600_dma_region  aos_vs;
-    struct r600_dma_region  aos_fs;
-    struct r600_dma_region  aos_vb;
-    aos_vs.buf = NULL;
-    aos_fs.buf = NULL;
-    aos_vb.buf = NULL;
-
 
+    void * pbo_vs;
+    void * pbo_fs;
+    struct radeon_aos aos_vb;
+    
     unsigned int ui;
     GLfloat  fTemp;
     GLfloat  fVb[] = { 1.0f,  1.0f, 1.0f, 1.0f,
@@ -93,7 +80,7 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
                            0x10000000, 0x340C90,   0x10000400, 0x20340C90,
                            0x10000800, 0x40340C90, 0x90000C00, 0x60200C90};
 
-    if (context->screen->chip.type <= CHIP_TYPE_RV670)
+    if (context->radeon.radeonScreen->chip_family <= CHIP_FAMILY_RV670)
     {
         uVs[9]  = 0x541910;
         uVs[11] = 0x20541910;
@@ -116,7 +103,7 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
 
     r700InitState(ctx);
 
-    r700->CB_COLOR0_BASE.u32All = context->target.rt.gpu >> 8;
+    r700SetRenderTarget(context);
 
     /* Turn off perspective divid. */
     SETbit(r700->PA_CL_VTE_CNTL.u32All, VTX_XY_FMT_bit);
@@ -152,15 +139,19 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
     }
 
     /* Setup vb */
-    R700_CMDBUF_CHECK_SPACE(6);
-    R700EP3 (context, IT_SET_CTL_CONST, 1);
-    R700E32 (context, mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX);
-    R700E32 (context, 0);
-    R700EP3 (context, IT_SET_CTL_CONST, 1);
-    R700E32 (context, mmSQ_VTX_START_INST_LOC - ASIC_CTL_CONST_BASE_INDEX);
-    R700E32 (context, 0);
+    BEGIN_BATCH_NO_AUTOSTATE(6);
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1));
+	OUT_BATCH(mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX);
+    OUT_BATCH(0);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1));
+    OUT_BATCH(mmSQ_VTX_START_INST_LOC - ASIC_CTL_CONST_BASE_INDEX);
+    OUT_BATCH(0);
+    END_BATCH();
+    COMMIT_BATCH();
+
     (context->chipobj.EmitVec)(ctx, &aos_vb, (GLvoid *)fVb, 4, 16, 6);
-    r700SetupVTXConstans(ctx, VERT_ATTRIB_POS, (unsigned int)aos_vb.aos_offset, 4, 16, 6);
+    r700SetupVTXConstans(ctx, VERT_ATTRIB_POS, &aos_vb, 4, 16, 6);
 
     /* Setup shaders, copied from dump */
     r700->SQ_PGM_RESOURCES_PS.u32All = 0;
@@ -168,23 +159,26 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
 	SETbit(r700->SQ_PGM_RESOURCES_PS.u32All, PGM_RESOURCES__PRIME_CACHE_ON_DRAW_bit);
     SETbit(r700->SQ_PGM_RESOURCES_VS.u32All, PGM_RESOURCES__PRIME_CACHE_ON_DRAW_bit);
         /* vs */
-    (context->chipobj.EmitShader)(ctx, &aos_vs, (GLvoid *)(&uVs[0]), 28);
-    r700->SQ_PGM_START_VS.u32All     = (aos_vs.aos_offset >> 8) & 0x00FFFFFF;
+    (context->chipobj.EmitShader)(ctx, &pbo_vs, (GLvoid *)(&uVs[0]), 28);
+    r700->SQ_PGM_START_VS.u32All     = 0;
     r700->SQ_PGM_RESOURCES_VS.u32All = 0x00800004;
+
             /* vs const */ /* TODO : Set color here */
-    R700_CMDBUF_CHECK_SPACE(4 + 2);
-    R700EP3 (context, IT_SET_ALU_CONST, 4);
-    R700E32 (context, SQ_ALU_CONSTANT_VS_OFFSET * 4); 
-    R700E32 (context, *((unsigned int*)&(ctx->Color.ClearColor[0])));
-    R700E32 (context, *((unsigned int*)&(ctx->Color.ClearColor[1])));
-    R700E32 (context, *((unsigned int*)&(ctx->Color.ClearColor[2])));
-    R700E32 (context, *((unsigned int*)&(ctx->Color.ClearColor[3])));
+    BEGIN_BATCH_NO_AUTOSTATE(4 + 2);
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, 4));
+    OUT_BATCH(SQ_ALU_CONSTANT_VS_OFFSET * 4); 
+    OUT_BATCH(*((unsigned int*)&(ctx->Color.ClearColor[0])));
+    OUT_BATCH(*((unsigned int*)&(ctx->Color.ClearColor[1])));
+    OUT_BATCH(*((unsigned int*)&(ctx->Color.ClearColor[2])));
+    OUT_BATCH(*((unsigned int*)&(ctx->Color.ClearColor[3])));
+    END_BATCH();
+    COMMIT_BATCH();
 
     r700->SPI_VS_OUT_CONFIG.u32All   = 0x00000000;
 	r700->SPI_PS_IN_CONTROL_0.u32All = 0x20000001;
         /* ps */
-    (context->chipobj.EmitShader)(ctx, &aos_fs, (GLvoid *)(&uFs[0]), 12); 
-    r700->SQ_PGM_START_PS.u32All     = (aos_fs.aos_offset >> 8) & 0x00FFFFFF;
+    (context->chipobj.EmitShader)(ctx, &pbo_fs, (GLvoid *)(&uFs[0]), 12); 
+    r700->SQ_PGM_START_PS.u32All     = 0;
     r700->SQ_PGM_RESOURCES_PS.u32All = 0x00800002;
     r700->SQ_PGM_EXPORTS_PS.u32All   = 0x00000002;        
     r700->DB_SHADER_CONTROL.u32All   = 0x00000200; 
@@ -192,12 +186,12 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
     r700->CB_SHADER_CONTROL.u32All = 0x00000001;
 
     /* set a valid base address to make the command checker happy */
-    r700->SQ_PGM_START_FS.u32All     = (aos_fs.aos_offset >> 8) & 0x00FFFFFF;
-    r700->SQ_PGM_START_ES.u32All     = (aos_fs.aos_offset >> 8) & 0x00FFFFFF;
-    r700->SQ_PGM_START_GS.u32All     = (aos_fs.aos_offset >> 8) & 0x00FFFFFF;
+    r700->SQ_PGM_START_FS.u32All     = 0;
+    r700->SQ_PGM_START_ES.u32All     = 0;
+    r700->SQ_PGM_START_GS.u32All     = 0;
 
     /* Now, send the states */
-    r700SendContextStates(context);
+    r700SendContextStates(context, pbo_vs, pbo_fs);
 
     /* Draw */
     GLuint numEntires, j;
@@ -211,44 +205,47 @@ static GLboolean r700ClearWithDraw(context_t *context, GLbitfield mask)
                  + 3               /* VGT_PRIMITIVE_TYPE */
                  + numIndices + 3; /* DRAW_INDEX_IMMD */                 
                  
-    R700_CMDBUF_CHECK_SPACE(numEntires);  
+    BEGIN_BATCH_NO_AUTOSTATE(numEntires);  
 
     SETfield(VGT_INDEX_TYPE, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
 
-    R700EP3(context, IT_INDEX_TYPE, 0);
-    R700E32(context, VGT_INDEX_TYPE);
+    OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0));
+    OUT_BATCH(VGT_INDEX_TYPE);
 
     VGT_NUM_INDICES = numIndices;
 
     SETfield(VGT_PRIMITIVE_TYPE, DI_PT_TRILIST, VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask);
-    R700EP3(context, IT_SET_CONFIG_REG, 1);
-    R700E32(context, mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
-    R700E32(context, VGT_PRIMITIVE_TYPE);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
+    OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
+    OUT_BATCH(VGT_PRIMITIVE_TYPE);
 
     SETfield(VGT_DRAW_INITIATOR, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
     SETfield(VGT_DRAW_INITIATOR, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask);
 
-    R700EP3(context, IT_DRAW_INDEX_IMMD, (numIndices + 1));
-    R700E32(context, VGT_NUM_INDICES);
-    R700E32(context, VGT_DRAW_INITIATOR);
+    OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (numIndices + 1)));
+    OUT_BATCH(VGT_NUM_INDICES);
+    OUT_BATCH(VGT_DRAW_INITIATOR);
 
     for (j=0; j<numIndices; j++)
     {
-        R700E32(context, j);
+        OUT_BATCH(j);
     }
+    END_BATCH();
+    COMMIT_BATCH();
 
     (context->chipobj.FlushCmdBuffer)(context);
 
     /* TODO : keep these in context, don't load and release every time. */
-    (context->chipobj.FreeDmaRegion)(context, &aos_vs);
-    (context->chipobj.FreeDmaRegion)(context, &aos_fs);
-    (context->chipobj.FreeDmaRegion)(context, &aos_vb);
+    (context->chipobj.DeleteShader)(context, &pbo_vs);
+
+    (context->chipobj.DeleteShader)(context, &pbo_fs);
+
+    (context->chipobj.FreeDmaRegion)(context, aos_vb.bo);
 
     /* Restore chip object. */
     memcpy(r700, &r700Saved, sizeof(R700_CHIP_CONTEXT));
 
-#endif /* to be enabled */
-
     return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c
index 2b813828a5..d06af109e5 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.c
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.c
@@ -36,15 +36,11 @@
 #include "shader/prog_statevars.h"
 
 #include "r600_context.h"
+#include "r600_cmdbuf.h"
 
 #include "r700_chip.h"
 #include "r700_fragprog.h"
 
-/* to be enabled */
-#if 0
-#include "r700_emit.h"
-#endif
-
 #include "r700_debug.h"
 
 //TODO : Validate FP input with VP output.
@@ -250,9 +246,18 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
 	return GL_TRUE;
 }
 
+void * r700GetActiveFpShaderBo(GLcontext * ctx)
+{
+    struct r700_fragment_program *fp = (struct r700_fragment_program *)
+	                                   (ctx->FragmentProgram._Current);
+
+    return fp->shaderbo;
+}
+
 GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 {
     context_t *context = R700_CONTEXT(ctx);   
+    BATCH_LOCALS(&context->radeon);
     
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
 
@@ -271,14 +276,13 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 	    {
 		    Assemble( &(fp->r700Shader) );
 	    }
-/* to be enabled */
-#if 0
+
         /* Load fp to gpu */
         (context->chipobj.EmitShader)(ctx, 
-                       &(fp->shadercode), 
+                       &(fp->shaderbo), 
                        (GLvoid *)(fp->r700Shader.pProgram),
                        fp->r700Shader.uShaderBinaryDWORDSize);    			                
-#endif
+
         fp->loaded = GL_TRUE;
     }
 
@@ -288,10 +292,8 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
     /* TODO : enable this after MemUse fixed *=
     (context->chipobj.MemUse)(context, fp->shadercode.buf->id);
     */
-/* to be enabled */
-#if 0
-    r700->SQ_PGM_START_PS.u32All     = (fp->shadercode.aos_offset >> 8) & 0x00FFFFFF;
-#endif
+
+    r700->SQ_PGM_START_PS.u32All = 0; /* set from buffer obj */
 
     unNumOfReg = fp->r700Shader.nRegs + 1;
 
@@ -337,25 +339,27 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
     if(NULL != paramList)
     {
         _mesa_load_state_parameters(ctx, paramList);
-/* to be enabled */
-#if 0
+
         unNumParamData = paramList->NumParameters * 4;
-        R700_CMDBUF_CHECK_SPACE(2 + unNumParamData);
 
-        R700EP3(context, IT_SET_ALU_CONST, unNumParamData);
+        BEGIN_BATCH_NO_AUTOSTATE(2 + unNumParamData);
+        
+        OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, unNumParamData));
+
         /* assembler map const from very beginning. */
-        R700E32(context, SQ_ALU_CONSTANT_PS_OFFSET * 4);
+        OUT_BATCH(SQ_ALU_CONSTANT_PS_OFFSET * 4);
 
         unNumParamData = paramList->NumParameters;
 
         for(ui=0; ui<unNumParamData; ui++)
         {
-            R700E32(context, *((unsigned int*)&(paramList->ParameterValues[ui][0])));
-            R700E32(context, *((unsigned int*)&(paramList->ParameterValues[ui][1])));
-            R700E32(context, *((unsigned int*)&(paramList->ParameterValues[ui][2])));
-            R700E32(context, *((unsigned int*)&(paramList->ParameterValues[ui][3])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][0])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][1])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][2])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][3])));
         }
-#endif
+        END_BATCH();
+        COMMIT_BATCH();
     }
 
     return GL_TRUE;
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.h b/src/mesa/drivers/dri/r600/r700_fragprog.h
index 649ee4822d..9c7813e908 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.h
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.h
@@ -40,10 +40,8 @@ struct r700_fragment_program
 	GLboolean translated;
     GLboolean loaded;
 	GLboolean error;
-/* to be enabled */
-#if 0
-    struct r600_dma_region shadercode;
-#endif
+
+    void * shaderbo;
 
 	GLboolean WritesDepth;
 	GLuint optimization;
@@ -60,4 +58,6 @@ extern GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
 							                 struct gl_fragment_program   *mesa_vp);
 extern GLboolean r700SetupFragmentProgram(GLcontext * ctx);
 
+extern void *    r700GetActiveFpShaderBo(GLcontext * ctx);
+
 #endif /*_R700_FRAGPROG_H_*/
diff --git a/src/mesa/drivers/dri/r600/r700_ioctl.c b/src/mesa/drivers/dri/r600/r700_ioctl.c
index 0541dc76e4..7493efab0f 100644
--- a/src/mesa/drivers/dri/r600/r700_ioctl.c
+++ b/src/mesa/drivers/dri/r600/r700_ioctl.c
@@ -37,12 +37,38 @@
 #include "radeon_lock.h"
 #include "r600_context.h"
 
+#include "r700_chip.h"
 #include "r700_ioctl.h"
 #include "r700_clear.h"
 
+static void r700Flush(GLcontext *ctx)
+{
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+    context_t *     context = R700_CONTEXT(ctx);
+
+	if (RADEON_DEBUG & DEBUG_IOCTL)
+		fprintf(stderr, "%s %d\n", __FUNCTION__, radeon->cmdbuf.cs->cdw);
+
+	/* okay if we have no cmds in the buffer &&
+	   we have no DMA flush &&
+	   we have no DMA buffer allocated.
+	   then no point flushing anything at all.
+	*/
+	if (!radeon->dma.flush && !radeon->cmdbuf.cs->cdw && !radeon->dma.current)
+		return;
+
+	if (radeon->dma.flush)
+		radeon->dma.flush( ctx );
+
+	r700SendContextStates(context, NULL, NULL);
+   
+	if (radeon->cmdbuf.cs->cdw)
+		rcommonFlushCmdBuf(radeon, __FUNCTION__);
+}
+
 void r700InitIoctlFuncs(struct dd_function_table *functions)
 {
 	functions->Clear = r700Clear;
 	functions->Finish = radeonFinish;
-	functions->Flush = radeonFlush;
+	functions->Flush = r700Flush;
 }
diff --git a/src/mesa/drivers/dri/r600/r700_oglprog.c b/src/mesa/drivers/dri/r600/r700_oglprog.c
index ee6377632b..dc8c936165 100644
--- a/src/mesa/drivers/dri/r600/r700_oglprog.c
+++ b/src/mesa/drivers/dri/r600/r700_oglprog.c
@@ -36,18 +36,16 @@
 
 #include "r700_chip.h"
 #include "r700_oglprog.h"
-
-#if 0 /* to be enabled */
 #include "r700_fragprog.h"
 #include "r700_vertprog.h"
-#endif /* to be enabled */
+
 
 static struct gl_program *r700NewProgram(GLcontext * ctx, 
                                          GLenum target,
 					                     GLuint id)
 {
 	struct gl_program *pProgram = NULL;
-#if 0 /* to be enabled */
+
     struct r700_vertex_program *vp;
 	struct r700_fragment_program *fp;
 
@@ -62,7 +60,9 @@ static struct gl_program *r700NewProgram(GLcontext * ctx,
                                              id);
         vp->translated = GL_FALSE;
         vp->loaded     = GL_FALSE;
-        vp->shadercode.buf = NULL;
+ 
+        vp->shaderbo   = NULL;
+
 	    break;
     case GL_FRAGMENT_PROGRAM_NV:
     case GL_FRAGMENT_PROGRAM_ARB:
@@ -73,18 +73,19 @@ static struct gl_program *r700NewProgram(GLcontext * ctx,
                                                id);
         fp->translated = GL_FALSE;
         fp->loaded     = GL_FALSE;
-        fp->shadercode.buf = NULL;
+
+        fp->shaderbo   = NULL;
+
 	    break;
     default:
 	    _mesa_problem(ctx, "Bad target in r700NewProgram");
     }
-#endif /* to be enabled */
+
 	return pProgram;
 }
 
 static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 {
-#if 0 /* to be enabled */
     struct r700_vertex_program   * vp;
     struct r700_fragment_program * fp;
     context_t *context = R700_CONTEXT(ctx);
@@ -95,7 +96,9 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
     case GL_VERTEX_PROGRAM_ARB:	    
         vp = (struct r700_vertex_program*)prog;
         /* Release DMA region */
-        (context->chipobj.FreeDmaRegion)(context, &(vp->shadercode));
+
+        (context->chipobj.DeleteShader)(ctx, vp->shaderbo);
+
         /* Clean up */
         Clean_Up_Assembler(&(vp->r700AsmCode));
         Clean_Up_Shader(&(vp->r700Shader));
@@ -104,7 +107,9 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
     case GL_FRAGMENT_PROGRAM_ARB:
 		fp = (struct r700_fragment_program*)prog;
         /* Release DMA region */
-        (context->chipobj.FreeDmaRegion)(context, &(fp->shadercode));
+
+        (context->chipobj.DeleteShader)(ctx, fp->shaderbo);
+
         /* Clean up */
         Clean_Up_Assembler(&(fp->r700AsmCode));
         Clean_Up_Shader(&(fp->r700Shader));
@@ -114,7 +119,6 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
     }
 
 	_mesa_delete_program(ctx, prog);
-#endif /* to be enabled */
 }
 
 static void
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index fce99ada54..f338cf823c 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -45,107 +45,64 @@
 #include "tnl/t_pipeline.h"
 
 #include "r600_context.h"
+#include "r600_cmdbuf.h"
 
 #include "r700_chip.h"
+#include "r700_tex.h"
 
-/* to be enable
 #include "r700_vertprog.h"
 #include "r700_fragprog.h"
 #include "r700_state.h"
-#include "r700_tex.h"
-#include "r700_emit.h"
-*/
 
-#if 0 // to be enable
 void r700WaitForIdle(context_t *context)
 {
-    R700EP3 (context, IT_SET_CONFIG_REG, 1);
-    R700E32 (context, mmWAIT_UNTIL - ASIC_CONFIG_BASE_INDEX);
-    R700E32 (context, 1 << 15);
+    BATCH_LOCALS(&context->radeon);
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
+    OUT_BATCH(mmWAIT_UNTIL - ASIC_CONFIG_BASE_INDEX);
+    OUT_BATCH(1 << 15);
+
+    END_BATCH();
+    COMMIT_BATCH();
 }
 
 void r700WaitForIdleClean(context_t *context)
 {
-    R700EP3 (context, IT_EVENT_WRITE, 0);
-    R700E32 (context, 0x16);
+    BATCH_LOCALS(&context->radeon);
+    BEGIN_BATCH_NO_AUTOSTATE(5);
+
+    OUT_BATCH(CP_PACKET3(R600_IT_EVENT_WRITE, 0));
+    OUT_BATCH(0x16);
 
-    R700EP3 (context, IT_SET_CONFIG_REG, 1);
-    R700E32 (context, mmWAIT_UNTIL - ASIC_CONFIG_BASE_INDEX);
-    R700E32 (context, 1 << 17); 
+    OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
+    OUT_BATCH(mmWAIT_UNTIL - ASIC_CONFIG_BASE_INDEX);
+    OUT_BATCH(1 << 17); 
+
+    END_BATCH();
+    COMMIT_BATCH();
 }
 
 static void r700Start3D(context_t *context)
 {
-    if (context->screen->chip.type <= CHIP_TYPE_RV670)
+    BATCH_LOCALS(&context->radeon);
+    if (context->radeon.radeonScreen->chip_family <= CHIP_FAMILY_RV670)
     {
-        R700EP3 (context, IT_START_3D_CMDBUF, 1);
-        R700E32 (context, 0);
+        BEGIN_BATCH_NO_AUTOSTATE(2);
+        OUT_BATCH(CP_PACKET3(R600_IT_START_3D_CMDBUF, 1));
+        OUT_BATCH(0);
+        END_BATCH();        
     }
 
-    R700EP3 (context, IT_CONTEXT_CONTROL, 1);
-    R700E32 (context, 0x80000000);
-    R700E32 (context, 0x80000000);
-    r700WaitForIdleClean(context);
-}
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    OUT_BATCH(CP_PACKET3(R600_IT_CONTEXT_CONTROL, 1));
+    OUT_BATCH(0x80000000);
+    OUT_BATCH(0x80000000);
+    END_BATCH();
 
+    COMMIT_BATCH();
 
-static int r700SetupStreams(GLcontext * ctx)
-{
-    context_t         *context = R700_CONTEXT(ctx);
-
-    struct r700_vertex_program *vpc
-             = (struct r700_vertex_program *)ctx->VertexProgram._Current;
-
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *vb = &tnl->vb;
-
-    unsigned int unBit;
-	unsigned int i;
-
-    R700_CMDBUF_CHECK_SPACE(6);
-    R700EP3 (context, IT_SET_CTL_CONST, 1);
-    R700E32 (context, mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX);
-    R700E32 (context, 0);
-
-    R700EP3 (context, IT_SET_CTL_CONST, 1);
-    R700E32 (context, mmSQ_VTX_START_INST_LOC - ASIC_CTL_CONST_BASE_INDEX);
-    R700E32 (context, 0);
-
-    context->aos_count = 0;
-	for(i=0; i<VERT_ATTRIB_MAX; i++)
-	{
-		unBit = 1 << i;
-		if(vpc->mesa_program.Base.InputsRead & unBit) 
-		{
-            (context->chipobj.EmitVec)(ctx, 
-                        &(context->aos[context->aos_count]),
-				        vb->AttribPtr[i]->data,
-				        vb->AttribPtr[i]->size,
-				        vb->AttribPtr[i]->stride, 
-                        vb->Count);
-
-            context->aos[context->aos_count].aos_size = vb->AttribPtr[i]->size;
-
-            /* currently aos are packed */
-            r700SetupVTXConstans(ctx, 
-                                 i,
-                                 (unsigned int)context->aos[context->aos_count].aos_offset,
-                                 (unsigned int)vb->AttribPtr[i]->size,
-                                 (unsigned int)(vb->AttribPtr[i]->size * 4),
-                                 (unsigned int)vb->Count);
-            /* TODO : enable this after MemUse fixed *=
-            (context->chipobj.MemUse)(context, context->aos[context->aos_count].buf->id);
-            */
-
-            context->aos_count++;
-		}
-	}
-    for(i=context->aos_count; i<VERT_ATTRIB_MAX; i++)
-    {
-        context->aos[i].buf = NULL;
-    }
-
-    return R600_FALLBACK_NONE;
+    r700WaitForIdleClean(context);
 }
 
 static GLboolean r700SetupShaders(GLcontext * ctx)
@@ -177,7 +134,7 @@ GLboolean r700SendTextureState(context_t *context)
     unsigned int i;
 
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
-
+#if 0 /* to be enabled */
     for(i=0; i<R700_TEXTURE_NUMBERUNITS; i++)
     {
         if(r700->texture_states.textures[i] != 0)
@@ -204,12 +161,15 @@ GLboolean r700SendTextureState(context_t *context)
             R700E32 (context, r700->texture_states.samplers[i]->SQ_TEX_SAMPLER2.u32All);
         }
     }
-
+#endif
     return GL_TRUE;
 }
 
 GLboolean r700SyncSurf(context_t *context)
 {
+#if 0 //to be enabled
+    BATCH_LOCALS(&context->radeon);
+
     /* TODO : too heavy? */
     unsigned int CP_COHER_CNTL   = 0;
 
@@ -221,23 +181,20 @@ GLboolean r700SyncSurf(context_t *context)
 	                |SMX_ACTION_ENA_bit;
 
 
-    R700_CMDBUF_CHECK_SPACE(5);
-    R700EP3(context, IT_SURFACE_SYNC, 3);
-    R700E32(context, CP_COHER_CNTL);
-    R700E32(context, 0xFFFFFFFF);
-    R700E32(context, 0x00000000);
-    R700E32(context, 10);
+    BEGIN_BATCH_NO_AUTOSTATE(5);
+    
+    OUT_BATCH(CP_PACKET3((IT_SURFACE_SYNC << 8), 3)));
+    OUT_BATCH(CP_COHER_CNTL);
+    OUT_BATCH(0xFFFFFFFF);
+    OUT_BATCH(0x00000000);
+    OUT_BATCH(10);
 
+    END_BATCH();
+    COMMIT_BATCH();
+#endif
     return GL_TRUE;
 }
 
-static void r700SetRenderTarget(context_t *context)
-{
-    R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
-
-    r700->CB_COLOR0_BASE.u32All = context->target.rt.gpu >> 8;
-}
-
 unsigned int r700PrimitiveType(int prim)
 {
     switch (prim & PRIM_MODE_MASK) 
@@ -279,21 +236,21 @@ unsigned int r700PrimitiveType(int prim)
     }
 }
 
-#endif // to be enable
-
 static GLboolean r700RunRender(GLcontext * ctx,
 			                   struct tnl_pipeline_stage *stage)
 {
-#if 0 // to be enable
     context_t *context = R700_CONTEXT(ctx);
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
+
+    BATCH_LOCALS(&context->radeon);
+
     unsigned int i, j;
     TNLcontext *tnl = TNL_CONTEXT(ctx);
     struct vertex_buffer *vb = &tnl->vb;
 
     struct r700_fragment_program *fp = (struct r700_fragment_program *)
 	                                   (ctx->FragmentProgram._Current);
-    if (context->screen->chip.type <= CHIP_TYPE_RV670)
+    if (context->radeon.radeonScreen->chip_family <= CHIP_FAMILY_RV670)
     {
         fp->r700AsmCode.bR6xx = 1;
     }
@@ -329,7 +286,7 @@ static GLboolean r700RunRender(GLcontext * ctx,
     r700->SQ_PGM_START_ES.u32All     = r700->SQ_PGM_START_PS.u32All;
     r700->SQ_PGM_START_GS.u32All     = r700->SQ_PGM_START_PS.u32All;
 
-    r700SendContextStates(context);
+    r700SendContextStates(context, NULL, NULL);
 
     /* richard test code */
     for (i = 0; i < vb->PrimitiveCount; i++) 
@@ -351,51 +308,55 @@ static GLboolean r700RunRender(GLcontext * ctx,
                      + numIndices + 3 /* DRAW_INDEX_IMMD */
                      + 2; /* test stamp */
                      
-        R700_CMDBUF_CHECK_SPACE(numEntires);  
+        BEGIN_BATCH_NO_AUTOSTATE(numEntires);  
 
         VGT_INDEX_TYPE |= DI_INDEX_SIZE_32_BIT << INDEX_TYPE_shift;
 
-        R700EP3(context, IT_INDEX_TYPE, 0);
-        R700E32(context, VGT_INDEX_TYPE);
+        OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0));
+        OUT_BATCH(VGT_INDEX_TYPE);
 
         VGT_NUM_INDICES = numIndices;
 
         VGT_PRIMITIVE_TYPE |= r700PrimitiveType(prim) << VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift;
-        R700EP3(context, IT_SET_CONFIG_REG, 1);
-        R700E32(context, mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
-        R700E32(context, VGT_PRIMITIVE_TYPE);
+        OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1));
+        OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX);
+        OUT_BATCH(VGT_PRIMITIVE_TYPE);
 
         VGT_DRAW_INITIATOR |= DI_SRC_SEL_IMMEDIATE << SOURCE_SELECT_shift;
         VGT_DRAW_INITIATOR |= DI_MAJOR_MODE_0 << MAJOR_MODE_shift;
 
-        R700EP3(context, IT_DRAW_INDEX_IMMD, (numIndices + 1));
-        R700E32(context, VGT_NUM_INDICES);
-        R700E32(context, VGT_DRAW_INITIATOR);
+        OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (numIndices + 1)));
+        OUT_BATCH(VGT_NUM_INDICES);
+        OUT_BATCH(VGT_DRAW_INITIATOR);
 
         for (j=0; j<numIndices; j++)
         {
-            R700E32(context, j);
+            OUT_BATCH(j);
         }
-
-        /* test stamp, write a number to mmSCRATCH4 */
-        R700EP3(context, IT_SET_CONFIG_REG, 1);
-        R700E32(context, 0x2144 - 0x2000);
-        R700E32(context, 0x12341234);
+        END_BATCH();
+        COMMIT_BATCH();
     }
 
     /* Flush render op cached for last several quads. */
-    R700_CMDBUF_CHECK_SPACE(2);
-    R700EP3 (context, IT_EVENT_WRITE, 0);
-    R700E32 (context, CACHE_FLUSH_AND_INV_EVENT);
+    BEGIN_BATCH_NO_AUTOSTATE(2);
+    OUT_BATCH(CP_PACKET3(R600_IT_EVENT_WRITE, 0));
+    OUT_BATCH(CACHE_FLUSH_AND_INV_EVENT);
+    END_BATCH();
+    COMMIT_BATCH();
 
     (context->chipobj.FlushCmdBuffer)(context);
 
-    /* free aos => TODO : cache mgr */
-    for (i = 0; i < context->aos_count; i++) 
-    {
-        (context->chipobj.FreeDmaRegion)(context, &(context->aos[i]));
-    }
-#endif // to be enable
+    (context->chipobj.ReleaseArrays)(ctx);
+
+    //richard test
+    /* test stamp, write a number to mmSCRATCH4 */
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    R600_OUT_BATCH_REGVAL((0x2144 << 2), 0x56785678);
+    END_BATCH();
+    COMMIT_BATCH();
+
+    rcommonFlushCmdBuf( &context->radeon, __FUNCTION__ );
+
     return GL_FALSE;
 }
 
@@ -412,11 +373,19 @@ static GLboolean r700RunTCLRender(GLcontext * ctx,  /*----------------------*/
 {
 	GLboolean bRet = GL_FALSE;
 
-#if 0 // to be enable
+    /* TODO : sw fallback */
+
+    /**
+    * Ensure all enabled and complete textures are uploaded along with any buffers being used.
+    */
+    if(!r700ValidateBuffers(ctx))
+    {
+        return GL_TRUE;
+    }
+
     context_t *context = R700_CONTEXT(ctx);
 
     r700UpdateShaders(ctx);
-#endif // to be enable
 
     bRet = r700RunRender(ctx, stage);
 
diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c
index 4341f33665..a31a89627c 100644
--- a/src/mesa/drivers/dri/r600/r700_state.c
+++ b/src/mesa/drivers/dri/r600/r700_state.c
@@ -52,10 +52,9 @@
 #include "r700_chip.h"
 #include "r700_state.h"
 
-#if 0 /* to be enabled */
 #include "r700_fragprog.h"
 #include "r700_vertprog.h"
-#endif /* to be enabled */
+
 
 void r700SetDefaultStates(context_t *context) //--------------------
 {
@@ -68,13 +67,13 @@ void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
 
     GLvector4f dummy_attrib[_TNL_ATTRIB_MAX];
     GLvector4f *temp_attrib[_TNL_ATTRIB_MAX];
-#if 0 /* to be enabled */
+
     struct r700_vertex_program *vp;
 	int i;
 
-    if (context->NewGLState) 
+    if (context->radeon.NewGLState) 
     {
-        context->NewGLState = 0;
+        context->radeon.NewGLState = 0;
 
         for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) 
         {
@@ -108,7 +107,6 @@ void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
     }
 
     r700UpdateStateParameters(ctx, _NEW_PROGRAM);
-#endif /* to be enabled */
 }
 
 /*
@@ -116,6 +114,7 @@ void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
  */
 void r700UpdateViewportOffset(GLcontext * ctx) //------------------
 {
+    return;
 }
 
 /**
@@ -152,7 +151,6 @@ static void r700FetchStateParameter(GLcontext * ctx,
 
 void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state) //--------------------
 {
-#if 0 /* to be enabled */
 	struct r700_fragment_program *fp;
 	struct gl_program_parameter_list *paramList;
 	GLuint i;
@@ -183,7 +181,6 @@ void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state) //------------
 						paramList->ParameterValues[i]);
 		}
 	}
-#endif /* to be enabled */
 }
 
 /**
@@ -191,7 +188,6 @@ void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state) //------------
  */
 static void r700InvalidateState(GLcontext * ctx, GLuint new_state) //-------------------
 {
-#if 0 /* to be enabled */
     context_t *context = R700_CONTEXT(ctx);
 
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
@@ -204,12 +200,16 @@ static void r700InvalidateState(GLcontext * ctx, GLuint new_state) //-----------
 
 	if (new_state & (_NEW_BUFFERS | _NEW_COLOR | _NEW_PIXEL)) 
     {
+        _mesa_update_framebuffer(ctx);
+		/* this updates the DrawBuffer's Width/Height if it's a FBO */
+		_mesa_update_draw_buffer_bounds(ctx);
+
 		r700UpdateDrawBuffer(ctx);
 	}
 
 	r700UpdateStateParameters(ctx, new_state);
 
-    if(GL_TRUE == context->bEnablePerspective)
+    if(GL_TRUE == r700->bEnablePerspective)
     {
         /* Do scale XY and Z by 1/W0 for perspective correction on pos. For orthogonal case, set both to one. */
         CLEARbit(r700->PA_CL_VTE_CNTL.u32All, VTX_XY_FMT_bit);
@@ -232,8 +232,7 @@ static void r700InvalidateState(GLcontext * ctx, GLuint new_state) //-----------
         CLEARbit(r700->SPI_PS_IN_CONTROL_0.u32All, LINEAR_GRADIENT_ENA_bit);
     }
 
-	context->NewGLState |= new_state;
-#endif /* to be enabled */
+	context->radeon.NewGLState |= new_state;
 }
 
 static void r700SetDepthState(GLcontext * ctx)
@@ -492,17 +491,11 @@ static void r700Viewport(GLcontext * ctx,
 			             GLsizei width, 
                          GLsizei height) //--------------------
 {
-#if 0 /* to be enabled */
     context_t *context = R700_CONTEXT(ctx);
 
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
 
-    context->vport_x = x; 
-    context->vport_y = y;
-    context->vport_width = width;
-    context->vport_height= height;
-
-    __DRIdrawablePrivate *dPriv = context->currentDraw;
+    __DRIdrawablePrivate *dPriv = context->radeon.dri.drawable;
 
     GLfloat xoffset = dPriv ? (GLfloat) dPriv->x : 0;
     GLfloat yoffset = dPriv ? (GLfloat) dPriv->y + dPriv->h : 0;
@@ -534,12 +527,12 @@ static void r700Viewport(GLcontext * ctx,
 	tz = v[MAT_TZ] * scale;
 
     /* TODO : Need DMA flush as well. */
-
+#if 0 /* to be enabled */
     if(context->cmdbuf.count_used > 0)
     {
 	    (context->chipobj.FlushCmdBuffer)(context);
     }
-
+#endif /* to be enabled */
     r700->PA_CL_VPORT_XSCALE.u32All  = *((unsigned int*)(&sx));
     r700->PA_CL_VPORT_XOFFSET.u32All = *((unsigned int*)(&tx));
 
@@ -548,7 +541,6 @@ static void r700Viewport(GLcontext * ctx,
 
     r700->PA_CL_VPORT_ZSCALE.u32All  = *((unsigned int*)(&sz));
     r700->PA_CL_VPORT_ZOFFSET.u32All = *((unsigned int*)(&tz));
-#endif /* to be enabled */
 }
 
 
@@ -600,60 +592,47 @@ static void r700Scissor(GLcontext* ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 	}
 }
 
-
-/**
- * Calculate initial hardware state and register state functions.
- * Assumes that the command buffer and state atoms have been
- * initialized already.
- */
-void r700InitState(GLcontext * ctx) //-------------------
+void r700SetRenderTarget(context_t *context)
 {
-#if 0 /* to be enabled */
-    context_t *context = R700_CONTEXT(ctx);
-
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
- 
-    if(context->ctx->Visual.doubleBufferMode && context->sarea->pfCurrentPage == 0) 
-    {
-		context->target.rt = context->screen->backBuffer;
-	} 
-    else 
-    {
-		context->target.rt = context->screen->frontBuffer;
-	}
 
+    struct radeon_renderbuffer *rrb;
+    unsigned int nPitchInPixel;
+
+    /* screen/window/view */
     SETfield(r700->CB_TARGET_MASK.u32All, 0xF, TARGET0_ENABLE_shift, TARGET0_ENABLE_mask);
     SETfield(r700->CB_SHADER_MASK.u32All, 0xF, OUTPUT0_ENABLE_shift, OUTPUT0_ENABLE_mask);
 
     /* screen */
     r700->PA_SC_SCREEN_SCISSOR_TL.u32All = 0x0;
-    SETfield(r700->PA_SC_SCREEN_SCISSOR_BR.u32All, context->screen->width,  
+    
+    SETfield(r700->PA_SC_SCREEN_SCISSOR_BR.u32All, ((RADEONDRIPtr)(context->radeon.radeonScreen->driScreen->pDevPriv))->width,  
              PA_SC_SCREEN_SCISSOR_BR__BR_X_shift, PA_SC_SCREEN_SCISSOR_BR__BR_X_mask);
-    SETfield(r700->PA_SC_SCREEN_SCISSOR_BR.u32All, context->screen->height, 
+    SETfield(r700->PA_SC_SCREEN_SCISSOR_BR.u32All, ((RADEONDRIPtr)(context->radeon.radeonScreen->driScreen->pDevPriv))->height, 
              PA_SC_SCREEN_SCISSOR_BR__BR_Y_shift, PA_SC_SCREEN_SCISSOR_BR__BR_Y_mask);
 
     /* window */
     SETbit(r700->PA_SC_WINDOW_SCISSOR_TL.u32All, WINDOW_OFFSET_DISABLE_bit);
-    SETfield(r700->PA_SC_WINDOW_SCISSOR_TL.u32All, context->currentDraw->x, 
+    SETfield(r700->PA_SC_WINDOW_SCISSOR_TL.u32All, context->radeon.dri.drawable->x, 
              PA_SC_WINDOW_SCISSOR_TL__TL_X_shift, PA_SC_WINDOW_SCISSOR_TL__TL_X_mask);
-    SETfield(r700->PA_SC_WINDOW_SCISSOR_TL.u32All, context->currentDraw->y, 
+    SETfield(r700->PA_SC_WINDOW_SCISSOR_TL.u32All, context->radeon.dri.drawable->y, 
              PA_SC_WINDOW_SCISSOR_TL__TL_Y_shift, PA_SC_WINDOW_SCISSOR_TL__TL_Y_mask);
 
-	SETfield(r700->PA_SC_WINDOW_SCISSOR_BR.u32All, context->currentDraw->x + context->currentDraw->w, 
+	SETfield(r700->PA_SC_WINDOW_SCISSOR_BR.u32All, context->radeon.dri.drawable->x + context->radeon.dri.drawable->w, 
              PA_SC_WINDOW_SCISSOR_BR__BR_X_shift, PA_SC_WINDOW_SCISSOR_BR__BR_X_mask);
-    SETfield(r700->PA_SC_WINDOW_SCISSOR_BR.u32All, context->currentDraw->y + context->currentDraw->h, 
+    SETfield(r700->PA_SC_WINDOW_SCISSOR_BR.u32All, context->radeon.dri.drawable->y + context->radeon.dri.drawable->h, 
              PA_SC_WINDOW_SCISSOR_BR__BR_Y_shift, PA_SC_WINDOW_SCISSOR_BR__BR_Y_mask);
 
     /* 4 clip rectangles */ /* TODO : set these clip rects according to context->currentDraw->numClipRects */
 	r700->PA_SC_CLIPRECT_RULE.u32All = 0x0000FFFF;
 
-    SETfield(r700->PA_SC_CLIPRECT_0_TL.u32All, context->currentDraw->x, 
+    SETfield(r700->PA_SC_CLIPRECT_0_TL.u32All, context->radeon.dri.drawable->x, 
              PA_SC_CLIPRECT_0_TL__TL_X_shift, PA_SC_CLIPRECT_0_TL__TL_X_mask);
-    SETfield(r700->PA_SC_CLIPRECT_0_TL.u32All, context->currentDraw->y, 
+    SETfield(r700->PA_SC_CLIPRECT_0_TL.u32All, context->radeon.dri.drawable->y, 
              PA_SC_CLIPRECT_0_TL__TL_Y_shift, PA_SC_CLIPRECT_0_TL__TL_Y_mask);
-	SETfield(r700->PA_SC_CLIPRECT_0_BR.u32All, context->currentDraw->x + context->currentDraw->w, 
+	SETfield(r700->PA_SC_CLIPRECT_0_BR.u32All, context->radeon.dri.drawable->x + context->radeon.dri.drawable->w, 
              PA_SC_CLIPRECT_0_BR__BR_X_shift, PA_SC_CLIPRECT_0_BR__BR_X_mask);
-    SETfield(r700->PA_SC_CLIPRECT_0_BR.u32All, context->currentDraw->y + context->currentDraw->h, 
+    SETfield(r700->PA_SC_CLIPRECT_0_BR.u32All, context->radeon.dri.drawable->y + context->radeon.dri.drawable->h, 
              PA_SC_CLIPRECT_0_BR__BR_Y_shift, PA_SC_CLIPRECT_0_BR__BR_Y_mask);
 
     r700->PA_SC_CLIPRECT_1_TL.u32All = r700->PA_SC_CLIPRECT_0_TL.u32All;
@@ -665,42 +644,138 @@ void r700InitState(GLcontext * ctx) //-------------------
 
     /* more....2d clip */
     SETbit(r700->PA_SC_GENERIC_SCISSOR_TL.u32All, WINDOW_OFFSET_DISABLE_bit);
-    SETfield(r700->PA_SC_GENERIC_SCISSOR_TL.u32All, context->currentDraw->x, 
+    SETfield(r700->PA_SC_GENERIC_SCISSOR_TL.u32All, context->radeon.dri.drawable->x, 
              PA_SC_GENERIC_SCISSOR_TL__TL_X_shift, PA_SC_GENERIC_SCISSOR_TL__TL_X_mask);
-    SETfield(r700->PA_SC_GENERIC_SCISSOR_TL.u32All, context->currentDraw->y, 
+    SETfield(r700->PA_SC_GENERIC_SCISSOR_TL.u32All, context->radeon.dri.drawable->y, 
              PA_SC_GENERIC_SCISSOR_TL__TL_Y_shift, PA_SC_GENERIC_SCISSOR_TL__TL_Y_mask);
-    SETfield(r700->PA_SC_GENERIC_SCISSOR_BR.u32All, context->currentDraw->x + context->currentDraw->w, 
+    SETfield(r700->PA_SC_GENERIC_SCISSOR_BR.u32All, context->radeon.dri.drawable->x + context->radeon.dri.drawable->w, 
              PA_SC_GENERIC_SCISSOR_BR__BR_X_shift, PA_SC_GENERIC_SCISSOR_BR__BR_X_mask);
-    SETfield(r700->PA_SC_GENERIC_SCISSOR_BR.u32All, context->currentDraw->y + context->currentDraw->h, 
+    SETfield(r700->PA_SC_GENERIC_SCISSOR_BR.u32All, context->radeon.dri.drawable->y + context->radeon.dri.drawable->h, 
              PA_SC_GENERIC_SCISSOR_BR__BR_Y_shift, PA_SC_GENERIC_SCISSOR_BR__BR_Y_mask);
 
     SETbit(r700->PA_SC_VPORT_SCISSOR_0_TL.u32All, WINDOW_OFFSET_DISABLE_bit);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_0_TL.u32All, context->currentDraw->x, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_0_TL.u32All, context->radeon.dri.drawable->x, 
              PA_SC_VPORT_SCISSOR_0_TL__TL_X_shift, PA_SC_VPORT_SCISSOR_0_TL__TL_X_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_0_TL.u32All, context->currentDraw->y, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_0_TL.u32All, context->radeon.dri.drawable->y, 
              PA_SC_VPORT_SCISSOR_0_TL__TL_Y_shift, PA_SC_VPORT_SCISSOR_0_TL__TL_Y_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_0_BR.u32All, context->currentDraw->x + context->currentDraw->w, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_0_BR.u32All, context->radeon.dri.drawable->x + context->radeon.dri.drawable->w, 
              PA_SC_VPORT_SCISSOR_0_BR__BR_X_shift, PA_SC_VPORT_SCISSOR_0_BR__BR_X_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_0_BR.u32All, context->currentDraw->y + context->currentDraw->h, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_0_BR.u32All, context->radeon.dri.drawable->y + context->radeon.dri.drawable->h, 
              PA_SC_VPORT_SCISSOR_0_BR__BR_Y_shift, PA_SC_VPORT_SCISSOR_0_BR__BR_Y_mask);
     
     SETbit(r700->PA_SC_VPORT_SCISSOR_1_TL.u32All, WINDOW_OFFSET_DISABLE_bit);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_1_TL.u32All, context->currentDraw->x, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_1_TL.u32All, context->radeon.dri.drawable->x, 
              PA_SC_VPORT_SCISSOR_0_TL__TL_X_shift, PA_SC_VPORT_SCISSOR_0_TL__TL_X_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_1_TL.u32All, context->currentDraw->y, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_1_TL.u32All, context->radeon.dri.drawable->y, 
              PA_SC_VPORT_SCISSOR_0_TL__TL_Y_shift, PA_SC_VPORT_SCISSOR_0_TL__TL_Y_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_1_BR.u32All, context->currentDraw->x + context->currentDraw->w, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_1_BR.u32All, context->radeon.dri.drawable->x + context->radeon.dri.drawable->w, 
              PA_SC_VPORT_SCISSOR_0_BR__BR_X_shift, PA_SC_VPORT_SCISSOR_0_BR__BR_X_mask);
-    SETfield(r700->PA_SC_VPORT_SCISSOR_1_BR.u32All, context->currentDraw->y + context->currentDraw->h, 
+    SETfield(r700->PA_SC_VPORT_SCISSOR_1_BR.u32All, context->radeon.dri.drawable->y + context->radeon.dri.drawable->h, 
              PA_SC_VPORT_SCISSOR_0_BR__BR_Y_shift, PA_SC_VPORT_SCISSOR_0_BR__BR_Y_mask);
 
     /* setup viewport */
-    r700Viewport(ctx, 
+    r700Viewport(GL_CONTEXT(context), 
                  0,
                  0,
-			     context->currentDraw->w,
-                 context->currentDraw->h);
+			     context->radeon.dri.drawable->w,
+                 context->radeon.dri.drawable->h);
+
+    rrb = radeon_get_colorbuffer(&context->radeon);
+	if (!rrb || !rrb->bo) {
+		fprintf(stderr, "no rrb\n");
+		return;
+	}
+
+    /* color buffer */ 
+    r700->CB_COLOR0_BASE.u32All = context->radeon.state.color.draw_offset;
     
+    nPitchInPixel = rrb->pitch/rrb->cpp;
+    SETfield(r700->CB_COLOR0_SIZE.u32All, (nPitchInPixel/8)-1,
+             PITCH_TILE_MAX_shift, PITCH_TILE_MAX_mask);
+    SETfield(r700->CB_COLOR0_SIZE.u32All, ( (nPitchInPixel * rrb->base.Height)/64 )-1,
+             SLICE_TILE_MAX_shift, SLICE_TILE_MAX_mask);
+    r700->CB_COLOR0_BASE.u32All = 0;
+    SETfield(r700->CB_COLOR0_INFO.u32All, ENDIAN_NONE, ENDIAN_shift, ENDIAN_mask);
+    SETfield(r700->CB_COLOR0_INFO.u32All, ARRAY_LINEAR_GENERAL, 
+             CB_COLOR0_INFO__ARRAY_MODE_shift, CB_COLOR0_INFO__ARRAY_MODE_mask); 
+    if(4 == rrb->cpp)
+    {
+        SETfield(r700->CB_COLOR0_INFO.u32All, COLOR_8_8_8_8,
+                 CB_COLOR0_INFO__FORMAT_shift, CB_COLOR0_INFO__FORMAT_mask);
+        SETfield(r700->CB_COLOR0_INFO.u32All, SWAP_ALT, COMP_SWAP_shift, COMP_SWAP_mask);
+    }
+    else
+    {
+        SETfield(r700->CB_COLOR0_INFO.u32All, COLOR_5_6_5,
+                 CB_COLOR0_INFO__FORMAT_shift, CB_COLOR0_INFO__FORMAT_mask);
+        SETfield(r700->CB_COLOR0_INFO.u32All, SWAP_ALT_REV, 
+                 COMP_SWAP_shift, COMP_SWAP_mask);        
+    } 
+    SETbit(r700->CB_COLOR0_INFO.u32All, SOURCE_FORMAT_bit);
+    SETbit(r700->CB_COLOR0_INFO.u32All, BLEND_CLAMP_bit);
+    SETfield(r700->CB_COLOR0_INFO.u32All, NUMBER_UNORM, NUMBER_TYPE_shift, NUMBER_TYPE_mask);
+
+    /* depth buf */ 
+	r700->DB_DEPTH_SIZE.u32All = 0;
+    r700->DB_DEPTH_BASE.u32All = 0;
+    r700->DB_DEPTH_INFO.u32All = 0;
+
+    r700->DB_DEPTH_CONTROL.u32All   = 0;
+    r700->DB_DEPTH_CLEAR.u32All     = 0x3F800000;
+    r700->DB_DEPTH_VIEW.u32All      = 0;
+    r700->DB_RENDER_CONTROL.u32All  = 0;
+    r700->DB_RENDER_OVERRIDE.u32All = 0;
+    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIZ_ENABLE_shift, FORCE_HIZ_ENABLE_mask); 
+    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE0_shift, FORCE_HIS_ENABLE0_mask); 
+    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask);
+
+    rrb = radeon_get_depthbuffer(&context->radeon);
+	if (!rrb)
+		return;
+
+    nPitchInPixel = rrb->pitch/rrb->cpp;
+
+    SETfield(r700->DB_DEPTH_SIZE.u32All, (nPitchInPixel/8)-1,
+             PITCH_TILE_MAX_shift, PITCH_TILE_MAX_mask);
+    SETfield(r700->DB_DEPTH_SIZE.u32All, ( (nPitchInPixel * rrb->base.Height)/64 )-1,
+             SLICE_TILE_MAX_shift, SLICE_TILE_MAX_mask); /* size in pixel / 64 - 1 */
+
+    if(4 == rrb->cpp) 
+    {
+        switch (GL_CONTEXT(context)->Visual.depthBits) 
+        {
+        case 16:           
+        case 24:
+            SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_8_24, 
+                     DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask);       
+            break;
+        default:
+            fprintf(stderr, "Error: Unsupported depth %d... exiting\n",
+                GL_CONTEXT(context)->Visual.depthBits);
+            _mesa_exit(-1);
+        }
+    }
+    else
+    {
+        SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_16, 
+                     DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask);          
+    } 
+    SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_2D_TILED_THIN1, 
+             DB_DEPTH_INFO__ARRAY_MODE_shift, DB_DEPTH_INFO__ARRAY_MODE_mask); 
+    /* r700->DB_PREFETCH_LIMIT.bits.DEPTH_HEIGHT_TILE_MAX = (context->currentDraw->h >> 3) - 1; */ /* z buffer sie may much bigger than what need, so use actual used h. */     
+}
+
+/**
+ * Calculate initial hardware state and register state functions.
+ * Assumes that the command buffer and state atoms have been
+ * initialized already.
+ */
+void r700InitState(GLcontext * ctx) //-------------------
+{
+    context_t *context = R700_CONTEXT(ctx);
+
+    R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
+
     /* Turn off vgt reuse */
     r700->VGT_REUSE_OFF.u32All = 0;
     SETbit(r700->VGT_REUSE_OFF.u32All, REUSE_OFF_bit);
@@ -740,7 +815,7 @@ void r700InitState(GLcontext * ctx) //-------------------
              POLYMODE_BACK_PTYPE_shift, POLYMODE_BACK_PTYPE_mask); 
 
     /* Do scale XY and Z by 1/W0. */
-    context->bEnablePerspective = GL_TRUE;
+    r700->bEnablePerspective = GL_TRUE;
     CLEARbit(r700->PA_CL_VTE_CNTL.u32All, VTX_XY_FMT_bit);
     CLEARbit(r700->PA_CL_VTE_CNTL.u32All, VTX_Z_FMT_bit);
     SETbit(r700->PA_CL_VTE_CNTL.u32All, VTX_W0_FMT_bit);
@@ -811,35 +886,8 @@ void r700InitState(GLcontext * ctx) //-------------------
 
     /* depth buf */ 
 	r700->DB_DEPTH_SIZE.u32All = 0;
-	SETfield(r700->DB_DEPTH_SIZE.u32All, (context->screen->depthBuffer.pitch/8)-1,
-             PITCH_TILE_MAX_shift, PITCH_TILE_MAX_mask);
-    SETfield(r700->DB_DEPTH_SIZE.u32All, ( (context->screen->depthBuffer.size / context->screen->cpp)/64 )-1,
-             SLICE_TILE_MAX_shift, SLICE_TILE_MAX_mask); /* size in pixel / 64 - 1 */
-    r700->DB_DEPTH_BASE.u32All = context->screen->depthBuffer.gpu >> 8;
+    r700->DB_DEPTH_BASE.u32All = 0;
     r700->DB_DEPTH_INFO.u32All = 0;
-    if(4 == context->screen->cpp) /* TODO : in scrren create, gives z its own format alloc. */
-    {
-        switch (ctx->Visual.depthBits) 
-        {
-        case 16:           
-        case 24:
-            SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_8_24, 
-                     DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask);       
-            break;
-        default:
-            fprintf(stderr, "Error: Unsupported depth %d... exiting\n",
-                ctx->Visual.depthBits);
-            _mesa_exit(-1);
-        }
-    }
-    else
-    {
-        SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_16, 
-                     DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask);          
-    } 
-    SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_2D_TILED_THIN1, 
-             DB_DEPTH_INFO__ARRAY_MODE_shift, DB_DEPTH_INFO__ARRAY_MODE_mask); 
-    /* r700->DB_PREFETCH_LIMIT.bits.DEPTH_HEIGHT_TILE_MAX = (context->currentDraw->h >> 3) - 1; */ /* z buffer sie may much bigger than what need, so use actual used h. */
     r700->DB_DEPTH_CONTROL.u32All   = 0;
     r700->DB_DEPTH_CLEAR.u32All     = 0x3F800000;
     r700->DB_DEPTH_VIEW.u32All      = 0;
@@ -850,27 +898,9 @@ void r700InitState(GLcontext * ctx) //-------------------
     SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask); 
     
     /* color buffer */ 
-    SETfield(r700->CB_COLOR0_SIZE.u32All, (context->screen->frontBuffer.pitch/8)-1,
-             PITCH_TILE_MAX_shift, PITCH_TILE_MAX_mask);
-    SETfield(r700->CB_COLOR0_SIZE.u32All, ( (context->screen->frontBuffer.size / context->screen->cpp)/64 )-1,
-             SLICE_TILE_MAX_shift, SLICE_TILE_MAX_mask);
-    r700->CB_COLOR0_BASE.u32All = context->screen->frontBuffer.gpu >> 8;
-    SETfield(r700->CB_COLOR0_INFO.u32All, ENDIAN_NONE, ENDIAN_shift, ENDIAN_mask);
-    SETfield(r700->CB_COLOR0_INFO.u32All, ARRAY_LINEAR_GENERAL, 
-             CB_COLOR0_INFO__ARRAY_MODE_shift, CB_COLOR0_INFO__ARRAY_MODE_mask); 
-    if(4 == context->screen->cpp)
-    {
-        SETfield(r700->CB_COLOR0_INFO.u32All, COLOR_8_8_8_8,
-                 CB_COLOR0_INFO__FORMAT_shift, CB_COLOR0_INFO__FORMAT_mask);
-        SETfield(r700->CB_COLOR0_INFO.u32All, SWAP_ALT, COMP_SWAP_shift, COMP_SWAP_mask);
-    }
-    else
-    {
-        SETfield(r700->CB_COLOR0_INFO.u32All, COLOR_5_6_5,
-                 CB_COLOR0_INFO__FORMAT_shift, CB_COLOR0_INFO__FORMAT_mask);
-        SETfield(r700->CB_COLOR0_INFO.u32All, SWAP_ALT_REV, 
-                 COMP_SWAP_shift, COMP_SWAP_mask);        
-    } 
+    r700->CB_COLOR0_SIZE.u32All = 0;
+    r700->CB_COLOR0_BASE.u32All = 0;
+    r700->CB_COLOR0_INFO.u32All = 0;
     SETbit(r700->CB_COLOR0_INFO.u32All, SOURCE_FORMAT_bit);
     SETbit(r700->CB_COLOR0_INFO.u32All, BLEND_CLAMP_bit);
     SETfield(r700->CB_COLOR0_INFO.u32All, NUMBER_UNORM, NUMBER_TYPE_shift, NUMBER_TYPE_mask);
@@ -879,8 +909,7 @@ void r700InitState(GLcontext * ctx) //-------------------
     r700->CB_COLOR0_FRAG.u32All   = 0;
     r700->CB_COLOR0_MASK.u32All   = 0;
 
-	r700->PA_SC_VPORT_ZMAX_0.u32All = 0x3F800000;
-#endif /* to be enabled */
+    r700->PA_SC_VPORT_ZMAX_0.u32All = 0x3F800000;
 }
 
 void r700InitStateFuncs(struct dd_function_table *functions) //-----------------
diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h
index b9542f4dd2..a75c7f0188 100644
--- a/src/mesa/drivers/dri/r600/r700_state.h
+++ b/src/mesa/drivers/dri/r600/r700_state.h
@@ -42,6 +42,7 @@ extern void r700UpdateDrawBuffer (GLcontext * ctx);
 extern void r700InitState (GLcontext * ctx);
 extern void r700InitStateFuncs (struct dd_function_table *functions);
 
+extern void r700SetRenderTarget(context_t *context);
 extern void r700SetDefaultStates(context_t * context);
 
 #endif	/* _R600_SCREEN_H */
diff --git a/src/mesa/drivers/dri/r600/r700_tex.c b/src/mesa/drivers/dri/r600/r700_tex.c
index 75300d2c5f..85352022bc 100644
--- a/src/mesa/drivers/dri/r600/r700_tex.c
+++ b/src/mesa/drivers/dri/r600/r700_tex.c
@@ -42,9 +42,7 @@
 #include "r600_context.h"
 #include "r700_chip.h"
 
-#if 0 /* to be enabled */
 #include "r700_state.h"
-#endif /* to be enabled */
 
 #include "r700_tex.h"
 
@@ -72,7 +70,6 @@ void r700SetTexOffset(__DRIcontext *pDRICtx, GLint texname,
 {
 }
 
-#if 0 /* to be enabled */
 static GLboolean r700GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_format)
 {
     r700TexObjPtr t = (r700TexObjPtr) tObj->DriverData;
@@ -548,7 +545,6 @@ static void compute_tex_image_offset(
 
     *curOffset += size;
 }
-#endif /* to be enabled */
  
 void r700DestroyTexObj(context_t context, r700TexObjPtr t) 
 {
@@ -563,9 +559,9 @@ void r700DestroyTexObj(context_t context, r700TexObjPtr t)
     //}
 }
  
-#if 0 /* to be enabled */
 static void r700SetTexImages(context_t *context, struct gl_texture_object *tObj)
 {
+#if 0 /* to be enabled */
     r700TexObjPtr t = (r700TexObjPtr) tObj->DriverData;
     const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
     GLint curOffset;
@@ -668,6 +664,7 @@ static void r700SetTexImages(context_t *context, struct gl_texture_object *tObj)
              TEX_WIDTH_shift, TEX_WIDTH_mask);
     SETfield(t->texture_state.SQ_TEX_RESOURCE1.u32All, tObj->Image[0][t->base.firstLevel]->Height - 1,
              TEX_HEIGHT_shift, TEX_HEIGHT_mask);
+#endif  /* to be enabled */
 }
 
 static void r700UploadSubImage(context_t    *context, 
@@ -677,6 +674,7 @@ static void r700UploadSubImage(context_t    *context,
                                GLint         y,                                
 			                   GLuint        face)
 {
+#if 0 /* to be enabled */
     struct gl_texture_image *texImage = NULL;
     GLuint offset;
     GLint imageWidth, imageHeight;
@@ -739,10 +737,12 @@ static void r700UploadSubImage(context_t    *context,
                          t->src_hight_in_pexel[face][hwlevel], /* height */
                          t->byte_per_texel, /* byte_per_pixel */
                          pSrc);  /* source data */
+#endif /* to be enabled */
 }
 
 int r700UploadTexImages(GLcontext * ctx, struct gl_texture_object *tObj, GLuint face)
 {
+#if 0 /* to be enabled */
     context_t *context = R700_CONTEXT(ctx);
     r700TexObjPtr t    = (r700TexObjPtr) tObj->DriverData; 
 
@@ -823,7 +823,7 @@ int r700UploadTexImages(GLcontext * ctx, struct gl_texture_object *tObj, GLuint
         SETfield(t->texture_state.SQ_TEX_RESOURCE4.u32All, t->base.firstLevel, BASE_LEVEL_shift, BASE_LEVEL_mask);
         SETfield(t->texture_state.SQ_TEX_RESOURCE5.u32All, t->base.lastLevel, LAST_LEVEL_shift, LAST_LEVEL_mask);        
     }
-
+#endif /* to be enabled */
 	return 0;
 }
  
@@ -889,8 +889,6 @@ static const struct gl_texture_format *r700Choose8888TexFormat(GLenum srcFormat,
     return gtfRet;
 }
 
-#endif /* to be enabled */
-
 static r700TexObjPtr r700AllocTexObj(struct gl_texture_object *texObj)
 {
 	r700TexObjPtr t;
@@ -1527,6 +1525,16 @@ void r700UpdateTextureState(context_t * context)
 #endif /* to be enabled */
 }
 
+/**
+ * Ensure all enabled and complete textures are uploaded along with any buffers being used.
+ */
+GLboolean r700ValidateBuffers(GLcontext * ctx)
+{
+	/* TODO */
+
+	return radeon_revalidate_bos(ctx);
+}
+
 void r700InitTextureFuncs(struct dd_function_table *functions) 
 {
 	/* Note: we only plug in the functions we implement in the driver
diff --git a/src/mesa/drivers/dri/r600/r700_tex.h b/src/mesa/drivers/dri/r600/r700_tex.h
index 13290578a1..e322bbc47d 100644
--- a/src/mesa/drivers/dri/r600/r700_tex.h
+++ b/src/mesa/drivers/dri/r600/r700_tex.h
@@ -97,6 +97,8 @@ extern void r700SetTexOffset(__DRIcontext *pDRICtx,
 
 extern void r700DestroyTexObj(context_t rmesa, r700TexObjPtr t);
 
+extern GLboolean r700ValidateBuffers(GLcontext * ctx);
+
 extern void r700InitTextureFuncs(struct dd_function_table *functions);
 
 #endif /* __r700_TEX_H__ */
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index 22a12a00d1..f0dfbe20f0 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -39,15 +39,12 @@
 #include "shader/prog_statevars.h"
 
 #include "r600_context.h"
+#include "r600_cmdbuf.h"
 
 #include "r700_chip.h"
 #include "r700_debug.h"
 #include "r700_vertprog.h"
 
-#if 0 /* to be enabled */
-#include "r700_emit.h"
-#endif
-
 unsigned int Map_Vertex_Output(r700_AssemblerBase       *pAsm, 
 					           struct gl_vertex_program *mesa_vp,
 					           unsigned int unStart)
@@ -297,11 +294,10 @@ GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp,
 
 void r700SelectVertexShader(GLcontext *ctx)
 {
-#if 0 /* to be enabled */
     context_t *context = R700_CONTEXT(ctx);
     struct r700_vertex_program *vpc
              = (struct r700_vertex_program *)ctx->VertexProgram._Current;
-    if (context->screen->chip.type <= CHIP_TYPE_RV670)
+    if (context->radeon.radeonScreen->chip_family <= CHIP_FAMILY_RV670)
     {
         vpc->r700AsmCode.bR6xx = 1;
     }
@@ -327,56 +323,22 @@ void r700SelectVertexShader(GLcontext *ctx)
         r700TranslateVertexShader(vpc,
 		    					  &(vpc->mesa_program) );
     }
-#endif /* to be enabled */
 }
 
-void r700SetupVTXConstans(GLcontext  * ctx, 
-                          unsigned int nStreamID,
-                          unsigned int aos_offset,
-                          unsigned int size,      /* number of elements in vector */
-                          unsigned int stride,
-                          unsigned int count)     /* number of vectors in stream */
+void * r700GetActiveVpShaderBo(GLcontext * ctx)
 {
-    context_t *context = R700_CONTEXT(ctx);
-    uint32_t *dest;
-
-    unsigned int uSQ_VTX_CONSTANT_WORD0_0;
-    unsigned int uSQ_VTX_CONSTANT_WORD1_0;
-    unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0;
-    unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0;
-    unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0;
-
-    uSQ_VTX_CONSTANT_WORD0_0 = aos_offset;
-	uSQ_VTX_CONSTANT_WORD1_0 = count * stride - 1;
-
-	uSQ_VTX_CONSTANT_WORD2_0 |= 0 << BASE_ADDRESS_HI_shift /* TODO */
-                             |stride << SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift	
-	                         |GetSurfaceFormat(GL_FLOAT, size, NULL) << SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift /* TODO : trace back api for initial data type, not only GL_FLOAT */
-	                         |SQ_NUM_FORMAT_SCALED << SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift
-	                         |SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit;
-	
-	uSQ_VTX_CONSTANT_WORD3_0 |= 1 << MEM_REQUEST_SIZE_shift;
-	
-	uSQ_VTX_CONSTANT_WORD6_0 |= SQ_TEX_VTX_VALID_BUFFER << SQ_TEX_RESOURCE_WORD6_0__TYPE_shift;
-#if 0 /* to be enabled */
-    R700_CMDBUF_CHECK_SPACE(9);
-    R700EP3 (context, IT_SET_RESOURCE, 7);
-    R700E32 (context, (nStreamID + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
-    
-	R700E32 (context, uSQ_VTX_CONSTANT_WORD0_0);
-	R700E32 (context, uSQ_VTX_CONSTANT_WORD1_0);
-	R700E32 (context, uSQ_VTX_CONSTANT_WORD2_0);
-	R700E32 (context, uSQ_VTX_CONSTANT_WORD3_0);
-	R700E32 (context, 0);
-	R700E32 (context, 0);
-	R700E32 (context, uSQ_VTX_CONSTANT_WORD6_0);
-#endif /* to be enabled */
+    struct r700_vertex_program *vp
+             = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+
+    return vp->shaderbo;
 }
 
 GLboolean r700SetupVertexProgram(GLcontext * ctx)
 {
     context_t *context = R700_CONTEXT(ctx);
 
+    BATCH_LOCALS(&context->radeon);
+
     R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(context->chipobj.pvChipObj);
 
     struct r700_vertex_program *vp
@@ -396,7 +358,7 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
 
         /* Load vp to gpu */
         (context->chipobj.EmitShader)(ctx, 
-                       &(vp->shadercode), 
+                       &(vp->shaderbo), 
                        (GLvoid *)(vp->r700Shader.pProgram),
                        vp->r700Shader.uShaderBinaryDWORDSize); 
 
@@ -410,7 +372,7 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
     (context->chipobj.MemUse)(context, vp->shadercode.buf->id);
     */
 
-    r700->SQ_PGM_START_VS.u32All            = (vp->shadercode.aos_offset >> 8) & 0x00FFFFFF;
+    r700->SQ_PGM_START_VS.u32All = 0; /* set from buffer object. */ 
     
     SETfield(r700->SQ_PGM_RESOURCES_VS.u32All, vp->r700Shader.nRegs + 1,
              NUM_GPRS_shift, NUM_GPRS_mask);
@@ -438,24 +400,26 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
     if(NULL != paramList)
     {
         _mesa_load_state_parameters(ctx, paramList);
-#if 0 /* to be enabled */
+
         unNumParamData = paramList->NumParameters * 4;
 
-        R700_CMDBUF_CHECK_SPACE(unNumParamData + 2);
-        R700EP3 (context, IT_SET_ALU_CONST, unNumParamData);
+        BEGIN_BATCH_NO_AUTOSTATE(unNumParamData + 2); 
+        
+        OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, unNumParamData));
         /* assembler map const from very beginning. */
-        R700E32 (context, SQ_ALU_CONSTANT_VS_OFFSET * 4);
+        OUT_BATCH(SQ_ALU_CONSTANT_VS_OFFSET * 4);
 
         unNumParamData = paramList->NumParameters;
 
         for(ui=0; ui<unNumParamData; ui++)
         {
-            R700E32 (context, *((unsigned int*)&(paramList->ParameterValues[ui][0])));
-            R700E32 (context, *((unsigned int*)&(paramList->ParameterValues[ui][1])));
-            R700E32 (context, *((unsigned int*)&(paramList->ParameterValues[ui][2])));
-            R700E32 (context, *((unsigned int*)&(paramList->ParameterValues[ui][3])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][0])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][1])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][2])));
+            OUT_BATCH(*((unsigned int*)&(paramList->ParameterValues[ui][3])));
         }
-#endif /* to be enabled */
+        END_BATCH();
+        COMMIT_BATCH();
     }
 
     return GL_TRUE;
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h
index 0822335c10..6a9726a3d0 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.h
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.h
@@ -55,7 +55,8 @@ struct r700_vertex_program
 	
 	/* ... */
 
-    struct r600_dma_region shadercode;
+    void * shaderbo;
+
 	ArrayDesc              aos_desc[VERT_ATTRIB_MAX];
 };
 
@@ -82,11 +83,6 @@ extern void r700SelectVertexShader(GLcontext *ctx);
 
 extern GLboolean r700SetupVertexProgram(GLcontext * ctx);
 
-extern void      r700SetupVTXConstans(GLcontext  * ctx, 
-                                      unsigned int nStreamID,
-                                      unsigned int aos_offset,
-                                      unsigned int size,      /* number of elements in vector */
-                                      unsigned int stride,
-                                      unsigned int Count);    /* number of vectors in stream */
+extern void *    r700GetActiveVpShaderBo(GLcontext * ctx);
 
 #endif /* _R700_VERTPROG_H_ */
diff --git a/src/mesa/drivers/dri/radeon/radeon_cmdbuf.h b/src/mesa/drivers/dri/radeon/radeon_cmdbuf.h
index 851b488433..abb023c7de 100644
--- a/src/mesa/drivers/dri/radeon/radeon_cmdbuf.h
+++ b/src/mesa/drivers/dri/radeon/radeon_cmdbuf.h
@@ -16,85 +16,7 @@ void rcommonBeginBatch(radeonContextPtr rmesa,
 		       const char *function,
 		       int line);
 
-#define RADEON_CP_PACKET3_NOP                       0xC0001000
-#define RADEON_CP_PACKET3_NEXT_CHAR                 0xC0001900
-#define RADEON_CP_PACKET3_PLY_NEXTSCAN              0xC0001D00
-#define RADEON_CP_PACKET3_SET_SCISSORS              0xC0001E00
-#define RADEON_CP_PACKET3_3D_RNDR_GEN_INDX_PRIM     0xC0002300
-#define RADEON_CP_PACKET3_LOAD_MICROCODE            0xC0002400
-#define RADEON_CP_PACKET3_WAIT_FOR_IDLE             0xC0002600
-#define RADEON_CP_PACKET3_3D_DRAW_VBUF              0xC0002800
-#define RADEON_CP_PACKET3_3D_DRAW_IMMD              0xC0002900
-#define RADEON_CP_PACKET3_3D_DRAW_INDX              0xC0002A00
-#define RADEON_CP_PACKET3_LOAD_PALETTE              0xC0002C00
-#define RADEON_CP_PACKET3_3D_LOAD_VBPNTR            0xC0002F00
-#define RADEON_CP_PACKET3_CNTL_PAINT                0xC0009100
-#define RADEON_CP_PACKET3_CNTL_BITBLT               0xC0009200
-#define RADEON_CP_PACKET3_CNTL_SMALLTEXT            0xC0009300
-#define RADEON_CP_PACKET3_CNTL_HOSTDATA_BLT         0xC0009400
-#define RADEON_CP_PACKET3_CNTL_POLYLINE             0xC0009500
-#define RADEON_CP_PACKET3_CNTL_POLYSCANLINES        0xC0009800
-#define RADEON_CP_PACKET3_CNTL_PAINT_MULTI          0xC0009A00
-#define RADEON_CP_PACKET3_CNTL_BITBLT_MULTI         0xC0009B00
-#define RADEON_CP_PACKET3_CNTL_TRANS_BITBLT         0xC0009C00
-
-/* r6xx/r7xx packet 3 type offsets */
-#define R600_SET_CONFIG_REG_OFFSET                  0x00008000
-#define R600_SET_CONFIG_REG_END                     0x0000ac00
-#define R600_SET_CONTEXT_REG_OFFSET                 0x00028000
-#define R600_SET_CONTEXT_REG_END                    0x00029000
-#define R600_SET_ALU_CONST_OFFSET                   0x00030000
-#define R600_SET_ALU_CONST_END                      0x00032000
-#define R600_SET_RESOURCE_OFFSET                    0x00038000
-#define R600_SET_RESOURCE_END                       0x0003c000
-#define R600_SET_SAMPLER_OFFSET                     0x0003c000
-#define R600_SET_SAMPLER_END                        0x0003cff0
-#define R600_SET_CTL_CONST_OFFSET                   0x0003cff0
-#define R600_SET_CTL_CONST_END                      0x0003e200
-#define R600_SET_LOOP_CONST_OFFSET                  0x0003e200
-#define R600_SET_LOOP_CONST_END                     0x0003e380
-#define R600_SET_BOOL_CONST_OFFSET                  0x0003e380
-#define R600_SET_BOOL_CONST_END                     0x00040000
-
-/* r6xx/r7xx packet 3 types */
-#define R600_IT_INDIRECT_BUFFER_END               0x00001700
-#define R600_IT_SET_PREDICATION                   0x00002000
-#define R600_IT_REG_RMW                           0x00002100
-#define R600_IT_COND_EXEC                         0x00002200
-#define R600_IT_PRED_EXEC                         0x00002300
-#define R600_IT_START_3D_CMDBUF                   0x00002400
-#define R600_IT_DRAW_INDEX_2                      0x00002700
-#define R600_IT_CONTEXT_CONTROL                   0x00002800
-#define R600_IT_DRAW_INDEX_IMMD_BE                0x00002900
-#define R600_IT_INDEX_TYPE                        0x00002A00
-#define R600_IT_DRAW_INDEX                        0x00002B00
-#define R600_IT_DRAW_INDEX_AUTO                   0x00002D00
-#define R600_IT_DRAW_INDEX_IMMD                   0x00002E00
-#define R600_IT_NUM_INSTANCES                     0x00002F00
-#define R600_IT_STRMOUT_BUFFER_UPDATE             0x00003400
-#define R600_IT_INDIRECT_BUFFER_MP                0x00003800
-#define R600_IT_MEM_SEMAPHORE                     0x00003900
-#define R600_IT_MPEG_INDEX                        0x00003A00
-#define R600_IT_WAIT_REG_MEM                      0x00003C00
-#define R600_IT_MEM_WRITE                         0x00003D00
-#define R600_IT_INDIRECT_BUFFER                   0x00003200
-#define R600_IT_CP_INTERRUPT                      0x00004000
-#define R600_IT_SURFACE_SYNC                      0x00004300
-#define R600_IT_ME_INITIALIZE                     0x00004400
-#define R600_IT_COND_WRITE                        0x00004500
-#define R600_IT_EVENT_WRITE                       0x00004600
-#define R600_IT_EVENT_WRITE_EOP                   0x00004700
-#define R600_IT_ONE_REG_WRITE                     0x00005700
-#define R600_IT_SET_CONFIG_REG                    0x00006800
-#define R600_IT_SET_CONTEXT_REG                   0x00006900
-#define R600_IT_SET_ALU_CONST                     0x00006A00
-#define R600_IT_SET_BOOL_CONST                    0x00006B00
-#define R600_IT_SET_LOOP_CONST                    0x00006C00
-#define R600_IT_SET_RESOURCE                      0x00006D00
-#define R600_IT_SET_SAMPLER                       0x00006E00
-#define R600_IT_SET_CTL_CONST                     0x00006F00
-#define R600_IT_SURFACE_BASE_UPDATE               0x00007300
-
+/* +r6/r7 : code here moved */
 
 #define CP_PACKET2  (2 << 30)
 #define CP_PACKET0(reg, n)	(RADEON_CP_PACKET0 | ((n)<<16) | ((reg)>>2))
@@ -188,54 +110,14 @@ void rcommonBeginBatch(radeonContextPtr rmesa,
 #define OUT_BATCH_FLOAT32(f) \
 	OUT_BATCH(radeonPackFloat32((f)))
 
-/* R600/R700 */
-#define R600_OUT_BATCH_REGS(reg, num)					\
-do {								\
-	if ((reg) >= R600_SET_CONFIG_REG_OFFSET && (reg) < R600_SET_CONFIG_REG_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_CONFIG_REG_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_CONTEXT_REG_OFFSET && (reg) < R600_SET_CONTEXT_REG_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_CONTEXT_REG, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_CONTEXT_REG_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_ALU_CONST_OFFSET && (reg) < R600_SET_ALU_CONST_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_ALU_CONST_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_RESOURCE_OFFSET && (reg) < R600_SET_RESOURCE_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_RESOURCE_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_SAMPLER_OFFSET && (reg) < R600_SET_SAMPLER_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_SAMPLER_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_CTL_CONST_OFFSET && (reg) < R600_SET_CTL_CONST_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_CTL_CONST_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_LOOP_CONST_OFFSET && (reg) < R600_SET_LOOP_CONST_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_LOOP_CONST, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_LOOP_CONST_OFFSET) >> 2);	\
-	} else if ((reg) >= R600_SET_BOOL_CONST_OFFSET && (reg) < R600_SET_BOOL_CONST_END) { \
-		OUT_BATCH(CP_PACKET3(R600_IT_SET_BOOL_CONST, (num)));	\
-		OUT_BATCH(((reg) - R600_SET_BOOL_CONST_OFFSET) >> 2);	\
-	} else {							\
-		OUT_BATCH(cmdpacket0(b_l_rmesa->radeonScreen, (reg), (num))); \
-	}								\
-} while (0)
-
-/** Single register write to command buffer; requires 3 dwords for most things. */
-#define R600_OUT_BATCH_REGVAL(reg, val)		\
-	R600_OUT_BATCH_REGS((reg), 1);		\
-	OUT_BATCH((val))
-
-/** Continuous register range write to command buffer; requires 1 dword,
- * expects count dwords afterwards for register contents. */
-#define R600_OUT_BATCH_REGSEQ(reg, count)	\
-	R600_OUT_BATCH_REGS((reg), (count))
+/* +r6/r7 : code here moved */
 
 /* Fire the buffered vertices no matter what.
  */
 static INLINE void radeon_firevertices(radeonContextPtr radeon)
 {
    if (radeon->cmdbuf.cs->cdw || radeon->dma.flush )
-      radeonFlush(radeon->glCtx);
+      radeon->glCtx->Driver.Flush(radeon->glCtx); /* +r6/r7 */
 }
 
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c
index 756c09fff3..691615938f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common.c
@@ -914,7 +914,7 @@ again:
 	if (ret == RADEON_CS_SPACE_OP_TO_BIG)
 		return GL_FALSE;
 	if (ret == RADEON_CS_SPACE_FLUSH) {
-		radeonFlush(ctx);
+		ctx->Driver.Flush(ctx); /* +r6/r7 */
 		if (flushed)
 			return GL_FALSE;
 		flushed = 1;
@@ -1014,7 +1014,7 @@ void radeonFinish(GLcontext * ctx)
 	struct gl_framebuffer *fb = ctx->DrawBuffer;
 	int i;
 
-	radeonFlush(ctx);
+	ctx->Driver.Flush(ctx); /* +r6/r7 */
 
 	if (radeon->radeonScreen->kernel_mm) {
 		for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c b/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
index b47b095cf2..b33417e93e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
+++ b/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
@@ -216,22 +216,28 @@ static int cs_process_relocs(struct radeon_cs *cs)
 
     csm = (struct cs_manager_legacy*)cs->csm;
     relocs = (struct cs_reloc_legacy *)cs->relocs;
- restart:
-    for (i = 0; i < cs->crelocs; i++) {
-        for (j = 0; j < relocs[i].cindices; j++) {
+restart:
+    for (i = 0; i < cs->crelocs; i++) 
+    {
+        for (j = 0; j < relocs[i].cindices; j++) 
+        {
             uint32_t soffset, eoffset;
 
             r = radeon_bo_legacy_validate(relocs[i].base.bo,
                                            &soffset, &eoffset);
-	    if (r == -EAGAIN)
-	      goto restart;
-            if (r) {
+	        if (r == -EAGAIN)
+            {
+	             goto restart;
+            }
+            if (r) 
+            {
                 fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
                         relocs[i].base.bo, soffset, eoffset);
                 return r;
             }
             cs->packets[relocs[i].indices[j]] += soffset;
-            if (cs->packets[relocs[i].indices[j]] >= eoffset) {
+            if (cs->packets[relocs[i].indices[j]] >= eoffset) 
+            {
 	      /*                radeon_bo_debug(relocs[i].base.bo, 12); */
                 fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
                         relocs[i].base.bo, soffset, eoffset);
@@ -280,7 +286,8 @@ static int cs_emit(struct radeon_cs *cs)
     csm->ctx->vtbl.emit_cs_header(cs, csm->ctx);
 
     /* append buffer age */
-    if (IS_R300_CLASS(csm->ctx->radeonScreen)) {
+    if ( IS_R300_CLASS(csm->ctx->radeonScreen) )
+    { 
       age.scratch.cmd_type = R300_CMD_SCRATCH;
       /* Scratch register 2 corresponds to what radeonGetAge polls */
       csm->pending_age = 0;
@@ -315,7 +322,8 @@ static int cs_emit(struct radeon_cs *cs)
     if (r) {
         return r;
     }
-    if (!IS_R300_CLASS(csm->ctx->radeonScreen)) {
+    if ((!IS_R300_CLASS(csm->ctx->radeonScreen)) &&
+        (!IS_R600_CLASS(csm->ctx->radeonScreen))) { /* +r6/r7 : No irq for r6/r7 yet. */
 	drm_radeon_irq_emit_t emit_cmd;
 	emit_cmd.irq_seq = &csm->pending_age;
 	r = drmCommandWrite(cs->csm->fd, DRM_RADEON_IRQ_EMIT, &emit_cmd, sizeof(emit_cmd));
@@ -387,59 +395,71 @@ static int cs_check_space(struct radeon_cs *cs, struct radeon_cs_space_check *bo
         return 0;
 
     /* prepare */
-    for (i = 0; i < num_bo; i++) {
-      bo = bos[i].bo;
-
-      bos[i].new_accounted = 0;
-      read_domains = bos[i].read_domains;
-      write_domain = bos[i].write_domain;
-		
-      /* pinned bos don't count */
-      if (radeon_legacy_bo_is_static(bo))
-	  continue;
+    for (i = 0; i < num_bo; i++) 
+    {
+         bo = bos[i].bo;
+
+         bos[i].new_accounted = 0;
+         read_domains = bos[i].read_domains;
+         write_domain = bos[i].write_domain;
+		   
+         /* pinned bos don't count */
+         if (radeon_legacy_bo_is_static(bo))
+	     continue;
  
-      /* already accounted this bo */
-      if (write_domain && (write_domain == bo->space_accounted))
-	  continue;
+         /* already accounted this bo */
+         if (write_domain && (write_domain == bo->space_accounted))
+	     continue;
 
-      if (read_domains && ((read_domains << 16) == bo->space_accounted))
-	  continue;
+         if (read_domains && ((read_domains << 16) == bo->space_accounted))
+	     continue;
       
-      if (bo->space_accounted == 0) {
-	  if (write_domain == RADEON_GEM_DOMAIN_VRAM)
-	      this_op_vram_write += bo->size;
-	  else if (write_domain == RADEON_GEM_DOMAIN_GTT)
-	      this_op_gart_write += bo->size;
-	  else
-	      this_op_read += bo->size;
-	  bos[i].new_accounted = (read_domains << 16) | write_domain;
-      } else {
-	  uint16_t old_read, old_write;
-	  
-	  old_read = bo->space_accounted >> 16;
-	  old_write = bo->space_accounted & 0xffff;
-
-	  if (write_domain && (old_read & write_domain)) {
-	      bos[i].new_accounted = write_domain;
-	      /* moving from read to a write domain */
-	      if (write_domain == RADEON_GEM_DOMAIN_VRAM) {
-		  this_op_read -= bo->size;
-		  this_op_vram_write += bo->size;
-	      } else if (write_domain == RADEON_GEM_DOMAIN_VRAM) {
-		  this_op_read -= bo->size;
-		  this_op_gart_write += bo->size;
-	      }
-	  } else if (read_domains & old_write) {
-	      bos[i].new_accounted = bo->space_accounted & 0xffff;
-	  } else {
-	      /* rewrite the domains */
-	      if (write_domain != old_write)
-		  fprintf(stderr,"WRITE DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, write_domain, old_write);
-	      if (read_domains != old_read)
-		  fprintf(stderr,"READ DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, read_domains, old_read);
-	      return RADEON_CS_SPACE_FLUSH;
-	  }
-      }
+         if (bo->space_accounted == 0) 
+         {
+	         if (write_domain == RADEON_GEM_DOMAIN_VRAM)
+	             this_op_vram_write += bo->size;
+	         else if (write_domain == RADEON_GEM_DOMAIN_GTT)
+	             this_op_gart_write += bo->size;
+	         else
+	             this_op_read += bo->size;
+	         bos[i].new_accounted = (read_domains << 16) | write_domain;
+         } 
+         else 
+         {
+	        uint16_t old_read, old_write;
+	     
+	        old_read = bo->space_accounted >> 16;
+	        old_write = bo->space_accounted & 0xffff;
+
+	        if (write_domain && (old_read & write_domain)) 
+            {
+	            bos[i].new_accounted = write_domain;
+	            /* moving from read to a write domain */
+	            if (write_domain == RADEON_GEM_DOMAIN_VRAM) 
+                {
+		            this_op_read -= bo->size;
+		            this_op_vram_write += bo->size;
+	            } 
+                else if (write_domain == RADEON_GEM_DOMAIN_VRAM) 
+                {
+		            this_op_read -= bo->size;
+		            this_op_gart_write += bo->size;
+	            }
+	        } 
+            else if (read_domains & old_write) 
+            {
+	            bos[i].new_accounted = bo->space_accounted & 0xffff;
+	        } 
+            else 
+            {
+	            /* rewrite the domains */
+	            if (write_domain != old_write)
+		            fprintf(stderr,"WRITE DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, write_domain, old_write);
+	            if (read_domains != old_read)
+		            fprintf(stderr,"READ DOMAIN RELOC FAILURE 0x%x %d %d\n", bo->handle, read_domains, old_read);
+	            return RADEON_CS_SPACE_FLUSH;
+	        }
+         }
 	}
 	
 	if (this_op_read < 0)
diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c
index f62ca7f9eb..b2585eae77 100644
--- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
+++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
@@ -165,7 +165,7 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
       return GL_FALSE;
    }
 
-  radeonFlush(ctx);
+  ctx->Driver.Flush(ctx); /* +r6/r7 */
 
   if (rrb->bo)
     radeon_bo_unref(rrb->bo);
@@ -371,7 +371,7 @@ radeon_framebuffer_renderbuffer(GLcontext * ctx,
                                GLenum attachment, struct gl_renderbuffer *rb)
 {
 
-   radeonFlush(ctx);
+   ctx->Driver.Flush(ctx); /* +r6/r7 */
 
    _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
    radeon_draw_buffer(ctx, fb);
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index b75db5b80c..be3b816b86 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -61,8 +61,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r300_tex.h"
 #elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
 #include "r600_context.h"
-#include "r700_fragprog.h"
-#include "r700_tex.h"
+//#include "r700_driconf.h" /* +r6/r7 */
+#include "r700_tex.h"     /* +r6/r7 */
 #endif
 
 #include "utils.h"
@@ -405,13 +405,13 @@ static const __DRItexBufferExtension r300TexBufferExtension = {
 #if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600)
 static const __DRItexOffsetExtension r600texOffsetExtension = {
     { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
-   r700SetTexOffset,
+   r700SetTexOffset, /* +r6/r7 */
 };
 
 static const __DRItexBufferExtension r600TexBufferExtension = {
     { __DRI_TEX_BUFFER, __DRI_TEX_BUFFER_VERSION },
-   r700SetTexBuffer,
-   r700SetTexBuffer2,
+   r700SetTexBuffer,  /* +r6/r7 */
+   r700SetTexBuffer2, /* +r6/r7 */
 };
 #endif
 
@@ -941,7 +941,6 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
 #if DO_DEBUG && RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
 	RADEON_DEBUG = driParseDebugString(getenv("RADEON_DEBUG"), debug_control);
 #endif
-
    /* parse information in __driConfigOptions */
    driParseOptionInfo (&screen->optionCache,
 		       __driConfigOptions, __driNConfigOptions);