Merge commit 'origin/master' into gallium-resources

Conflicts:
	src/gallium/drivers/llvmpipe/lp_texture.c
	src/gallium/drivers/r300/r300_context.c
	src/gallium/drivers/r300/r300_texture.c
	src/gallium/winsys/drm/radeon/core/radeon_buffer.h

38 files changed, 375 insertions, 333 deletions

diff --git a/include/GL/internal/dri_interface.h b/include/GL/internal/dri_interface.h
index 1d4e82e154..aa56eb45d7 100644
--- a/include/GL/internal/dri_interface.h
+++ b/include/GL/internal/dri_interface.h
@@ -41,7 +41,13 @@
 #define DRI_INTERFACE_H
 
 /* For archs with no drm.h */
-#if !defined(__APPLE__) && !defined(__CYGWIN__) && !defined(__GNU__)
+#if defined(__APPLE__) || defined(__CYGWIN__) || defined(__GNU__)
+#ifndef __NOT_HAVE_DRM_H
+#define __NOT_HAVE_DRM_H
+#endif
+#endif
+
+#ifndef __NOT_HAVE_DRM_H
 #include <drm.h>
 #else
 typedef unsigned int drm_context_t;
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
index 86f9266c95..0f2ae05dae 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
@@ -227,6 +227,8 @@ pb_cache_is_buffer_compat(struct pb_cache_buffer *buf,
                           pb_size size,
                           const struct pb_desc *desc)
 {
+   void *map;
+
    if(buf->base.base.size < size)
       return FALSE;
 
@@ -239,6 +241,13 @@ pb_cache_is_buffer_compat(struct pb_cache_buffer *buf,
    
    if(!pb_check_usage(desc->usage, buf->base.base.usage))
       return FALSE;
+
+   map = pb_map(buf->buffer, PIPE_BUFFER_USAGE_DONTBLOCK);
+   if (!map) {
+      return FALSE;
+   }
+
+   pb_unmap(buf->buffer);
    
    return TRUE;
 }
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index 45e6e44d33..71459abcfc 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -509,8 +509,8 @@ void
 llvmpipe_init_screen_resource_funcs(struct pipe_screen *screen)
 {
    screen->resource_create = llvmpipe_resource_create;
-   screen->resource_from_handle = llvmpipe_resource_from_handle;
    screen->resource_destroy = llvmpipe_resource_destroy;
+   screen->resource_from_handle = llvmpipe_resource_from_handle;
    screen->resource_get_handle = llvmpipe_resource_get_handle;
    screen->user_buffer_create = llvmpipe_user_buffer_create;
 
diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c
index 8afc95c9fc..5447904e9c 100644
--- a/src/gallium/drivers/nv50/nv50_clear.c
+++ b/src/gallium/drivers/nv50/nv50_clear.c
@@ -35,7 +35,10 @@ nv50_clear(struct pipe_context *pipe, unsigned buffers,
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct pipe_framebuffer_state *fb = &nv50->framebuffer;
 	unsigned mode = 0, i;
+	const unsigned dirty = nv50->dirty;
 
+	/* don't need NEW_BLEND, NV50TCL_COLOR_MASK doesn't affect CLEAR_BUFFERS */
+	nv50->dirty &= NV50_NEW_FRAMEBUFFER | NV50_NEW_SCISSOR;
 	if (!nv50_state_validate(nv50, 64))
 		return;
 
@@ -64,5 +67,6 @@ nv50_clear(struct pipe_context *pipe, unsigned buffers,
 		BEGIN_RING(chan, tesla, NV50TCL_CLEAR_BUFFERS, 1);
 		OUT_RING  (chan, (i << 6) | 0x3c);
 	}
+	nv50->dirty = dirty;
 }
 
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index d7f5863fb7..1a4606d9e2 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -109,7 +109,7 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param)
 	case PIPE_CAP_TWO_SIDED_STENCIL:
 		return 1;
 	case PIPE_CAP_GLSL:
-		return 0;
+		return 1;
 	case PIPE_CAP_ANISOTROPIC_FILTER:
 		return 1;
 	case PIPE_CAP_POINT_SPRITE:
@@ -190,8 +190,6 @@ nv50_screen_destroy(struct pipe_screen *pscreen)
 		nouveau_bo_ref(NULL, &screen->tic);
 	if (screen->tsc)
 		nouveau_bo_ref(NULL, &screen->tsc);
-	if (screen->static_init)
-		so_ref(NULL, &screen->static_init);
 
 	nouveau_notifier_free(&screen->sync);
 	nouveau_grobj_free(&screen->tesla);
@@ -204,16 +202,65 @@ nv50_screen_destroy(struct pipe_screen *pscreen)
 	FREE(screen);
 }
 
+#define BGN_RELOC(ch, bo, gr, m, n, fl) \
+   OUT_RELOC(ch, bo, (n << 18) | (gr->subc << 13) | m, fl, 0, 0)
+
+void
+nv50_screen_relocs(struct nv50_screen *screen)
+{
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *tesla = screen->tesla;
+	unsigned i;
+	const unsigned rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_DUMMY;
+
+	MARK_RING (chan, 28, 26);
+
+	/* cause grobj autobind */
+	BEGIN_RING(chan, tesla, 0x0100, 1);
+	OUT_RING  (chan, 0);
+
+	BGN_RELOC (chan, screen->tic, tesla, NV50TCL_TIC_ADDRESS_HIGH, 2, rl);
+	OUT_RELOCh(chan, screen->tic, 0, rl);
+	OUT_RELOCl(chan, screen->tic, 0, rl);
+
+	BGN_RELOC (chan, screen->tsc, tesla, NV50TCL_TSC_ADDRESS_HIGH, 2, rl);
+	OUT_RELOCh(chan, screen->tsc, 0, rl);
+	OUT_RELOCl(chan, screen->tsc, 0, rl);
+
+	BGN_RELOC (chan, screen->constbuf_misc[0],
+		   tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3, rl);
+	OUT_RELOCh(chan, screen->constbuf_misc[0], 0, rl);
+	OUT_RELOCl(chan, screen->constbuf_misc[0], 0, rl);
+	OUT_RELOC (chan, screen->constbuf_misc[0],
+		   (NV50_CB_PMISC << 16) | 0x0200, rl, 0, 0);
+
+	BGN_RELOC (chan, screen->constbuf_misc[0],
+		   tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3, rl);
+	OUT_RELOCh(chan, screen->constbuf_misc[0], 0x200, rl);
+	OUT_RELOCl(chan, screen->constbuf_misc[0], 0x200, rl);
+	OUT_RELOC (chan, screen->constbuf_misc[0],
+		   (NV50_CB_AUX << 16) | 0x0200, rl, 0, 0);
+
+	for (i = 0; i < 3; ++i) {
+		BGN_RELOC (chan, screen->constbuf_parm[i],
+			   tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3, rl);
+		OUT_RELOCh(chan, screen->constbuf_parm[i], 0, rl);
+		OUT_RELOCl(chan, screen->constbuf_parm[i], 0, rl);
+		OUT_RELOC (chan, screen->constbuf_parm[i],
+			   ((NV50_CB_PVP + i) << 16) | 0x0800, rl, 0, 0);
+	}
+}
+
 struct pipe_screen *
 nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 {
 	struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen);
 	struct nouveau_channel *chan;
 	struct pipe_screen *pscreen;
-	struct nouveau_stateobj *so;
 	unsigned chipset = dev->chipset;
 	unsigned tesla_class = 0;
 	int ret, i;
+	const unsigned rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_RD;
 
 	if (!screen)
 		return NULL;
@@ -296,64 +343,58 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	}
 
 	/* Static M2MF init */
-	so = so_new(1, 3, 0);
-	so_method(so, screen->m2mf, NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 3);
-	so_data  (so, screen->sync->handle);
-	so_data  (so, chan->vram->handle);
-	so_data  (so, chan->vram->handle);
-	so_emit(chan, so);
-	so_ref (NULL, &so);
+	BEGIN_RING(chan, screen->m2mf,
+		   NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 3);
+	OUT_RING  (chan, screen->sync->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle);
 
 	/* Static 2D init */
-	so = so_new(4, 7, 0);
-	so_method(so, screen->eng2d, NV50_2D_DMA_NOTIFY, 4);
-	so_data  (so, screen->sync->handle);
-	so_data  (so, chan->vram->handle);
-	so_data  (so, chan->vram->handle);
-	so_data  (so, chan->vram->handle);
-	so_method(so, screen->eng2d, NV50_2D_OPERATION, 1);
-	so_data  (so, NV50_2D_OPERATION_SRCCOPY);
-	so_method(so, screen->eng2d, NV50_2D_CLIP_ENABLE, 1);
-	so_data  (so, 0);
-	so_method(so, screen->eng2d, 0x0888, 1);
-	so_data  (so, 1);
-	so_emit(chan, so);
-	so_ref(NULL, &so);
+	BEGIN_RING(chan, screen->eng2d, NV50_2D_DMA_NOTIFY, 4);
+	OUT_RING  (chan, screen->sync->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	BEGIN_RING(chan, screen->eng2d, NV50_2D_OPERATION, 1);
+	OUT_RING  (chan, NV50_2D_OPERATION_SRCCOPY);
+	BEGIN_RING(chan, screen->eng2d, NV50_2D_CLIP_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, screen->eng2d, 0x0888, 1);
+	OUT_RING  (chan, 1);
 
 	/* Static tesla init */
-	so = so_new(47, 95, 24);
-
-	so_method(so, screen->tesla, NV50TCL_COND_MODE, 1);
-	so_data  (so, NV50TCL_COND_MODE_ALWAYS);
-	so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1);
-	so_data  (so, screen->sync->handle);
-	so_method(so, screen->tesla, NV50TCL_DMA_ZETA, 11);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_COND_MODE, 1);
+	OUT_RING  (chan, NV50TCL_COND_MODE_ALWAYS);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_DMA_NOTIFY, 1);
+	OUT_RING  (chan, screen->sync->handle);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_DMA_ZETA, 11);
 	for (i = 0; i < 11; i++)
-		so_data(so, chan->vram->handle);
-	so_method(so, screen->tesla, NV50TCL_DMA_COLOR(0),
-				     NV50TCL_DMA_COLOR__SIZE);
+		OUT_RING  (chan, chan->vram->handle);
+	BEGIN_RING(chan, screen->tesla,
+		   NV50TCL_DMA_COLOR(0), NV50TCL_DMA_COLOR__SIZE);
 	for (i = 0; i < NV50TCL_DMA_COLOR__SIZE; i++)
-		so_data(so, chan->vram->handle);
-	so_method(so, screen->tesla, NV50TCL_RT_CONTROL, 1);
-	so_data  (so, 1);
+		OUT_RING  (chan, chan->vram->handle);
+
+	BEGIN_RING(chan, screen->tesla, NV50TCL_RT_CONTROL, 1);
+	OUT_RING  (chan, 1);
 
 	/* activate all 32 lanes (threads) in a warp */
-	so_method(so, screen->tesla, NV50TCL_WARP_HALVES, 1);
-	so_data  (so, 0x2);
-	so_method(so, screen->tesla, 0x1400, 1);
-	so_data  (so, 0xf);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_WARP_HALVES, 1);
+	OUT_RING  (chan, 2);
+	BEGIN_RING(chan, screen->tesla, 0x1400, 1);
+	OUT_RING  (chan, 0xf);
 
 	/* max TIC (bits 4:8) & TSC (ignored) bindings, per program type */
 	for (i = 0; i < 3; ++i) {
-		so_method(so, screen->tesla, NV50TCL_TEX_LIMITS(i), 1);
-		so_data  (so, 0x54);
+		BEGIN_RING(chan, screen->tesla, NV50TCL_TEX_LIMITS(i), 1);
+		OUT_RING  (chan, 0x54);
 	}
 
 	/* origin is top left (set to 1 for bottom left) */
-	so_method(so, screen->tesla, NV50TCL_Y_ORIGIN_BOTTOM, 1);
-	so_data  (so, 0);
-	so_method(so, screen->tesla, NV50TCL_VP_REG_ALLOC_RESULT, 1);
-	so_data  (so, 8);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_Y_ORIGIN_BOTTOM, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_VP_REG_ALLOC_RESULT, 1);
+	OUT_RING  (chan, 8);
 
 	/* constant buffers for immediates and VP/FP parameters */
 	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (32 * 4) * 4,
@@ -362,6 +403,14 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		nv50_screen_destroy(pscreen);
 		return NULL;
 	}
+	BEGIN_RING(chan, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
+	OUT_RELOCh(chan, screen->constbuf_misc[0], 0, rl);
+	OUT_RELOCl(chan, screen->constbuf_misc[0], 0, rl);
+	OUT_RING  (chan, (NV50_CB_PMISC << 16) | 0x0200);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
+	OUT_RELOCh(chan, screen->constbuf_misc[0], 0x200, rl);
+	OUT_RELOCl(chan, screen->constbuf_misc[0], 0x200, rl);
+	OUT_RING  (chan, (NV50_CB_AUX << 16) | 0x0200);
 
 	for (i = 0; i < 3; i++) {
 		ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (256 * 4) * 4,
@@ -370,6 +419,10 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 			nv50_screen_destroy(pscreen);
 			return NULL;
 		}
+		BEGIN_RING(chan, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
+		OUT_RELOCh(chan, screen->constbuf_parm[i], 0, rl);
+		OUT_RELOCl(chan, screen->constbuf_parm[i], 0, rl);
+		OUT_RING  (chan, ((NV50_CB_PVP + i) << 16) | 0x0800);
 	}
 
 	if (nouveau_resource_init(&screen->immd_heap[0], 0, 128) ||
@@ -381,80 +434,16 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		return NULL;
 	}
 
-	/*
-	// map constant buffers:
-	//  B = buffer ID (maybe more than 1 byte)
-	//  N = CB index used in shader instruction
-	//  P = program type (0 = VP, 2 = GP, 3 = FP)
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x000BBNP1);
-	*/
-
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->constbuf_misc[0], 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->constbuf_misc[0], 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_PMISC << 16) | 0x00000200);
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000001 | (NV50_CB_PMISC << 12));
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000021 | (NV50_CB_PMISC << 12));
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000031 | (NV50_CB_PMISC << 12));
-
-	/* bind auxiliary constbuf to immediate data bo */
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->constbuf_misc[0], (128 * 4) * 4,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->constbuf_misc[0], (128 * 4) * 4,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_AUX << 16) | 0x00000200);
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000201 | (NV50_CB_AUX << 12));
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000221 | (NV50_CB_AUX << 12));
-
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_VERTEX], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_VERTEX], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_PVP << 16) | 0x00000800);
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000101 | (NV50_CB_PVP << 12));
-
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_GEOMETRY], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_GEOMETRY], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_PGP << 16) | 0x00000800);
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000121 | (NV50_CB_PGP << 12));
-
-	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_FRAGMENT], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_FRAGMENT], 0,
-		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, (NV50_CB_PFP << 16) | 0x00000800);
-	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
-	so_data  (so, 0x00000131 | (NV50_CB_PFP << 12));
-
 	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 3 * 32 * (8 * 4),
 			     &screen->tic);
 	if (ret) {
 		nv50_screen_destroy(pscreen);
 		return NULL;
 	}
-
-	so_method(so, screen->tesla, NV50TCL_TIC_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, 3 * 32 - 1);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_TIC_ADDRESS_HIGH, 3);
+	OUT_RELOCh(chan, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RELOCl(chan, screen->tic, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RING  (chan, 3 * 32 - 1);
 
 	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, 3 * 32 * (8 * 4),
 			     &screen->tsc);
@@ -462,37 +451,50 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		nv50_screen_destroy(pscreen);
 		return NULL;
 	}
-
-	so_method(so, screen->tesla, NV50TCL_TSC_ADDRESS_HIGH, 3);
-	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
-		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
-	so_data  (so, 0x00000000); /* ignored if TSC_LINKED (0x1234) = 1 */
-
+	BEGIN_RING(chan, screen->tesla, NV50TCL_TSC_ADDRESS_HIGH, 3);
+	OUT_RELOCh(chan, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RELOCl(chan, screen->tsc, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RING  (chan, 0); /* ignored if TSC_LINKED (0x1234) == 1 */
+
+	/* map constant buffers:
+	 *  B = buffer ID (maybe more than 1 byte)
+	 *  N = CB index used in shader instruction
+	 *  P = program type (0 = VP, 2 = GP, 3 = FP)
+	 * SET_PROGRAM_CB = 0x000BBNP1
+	 */
+	BEGIN_RING_NI(chan, screen->tesla, NV50TCL_SET_PROGRAM_CB, 8);
+	/* bind immediate buffer */
+	OUT_RING  (chan, 0x001 | (NV50_CB_PMISC << 12));
+	OUT_RING  (chan, 0x021 | (NV50_CB_PMISC << 12));
+	OUT_RING  (chan, 0x031 | (NV50_CB_PMISC << 12));
+	/* bind auxiliary constbuf to immediate data bo */
+	OUT_RING  (chan, 0x201 | (NV50_CB_AUX << 12));
+	OUT_RING  (chan, 0x221 | (NV50_CB_AUX << 12));
+	/* bind parameter buffers */
+	OUT_RING  (chan, 0x101 | (NV50_CB_PVP << 12));
+	OUT_RING  (chan, 0x121 | (NV50_CB_PGP << 12));
+	OUT_RING  (chan, 0x131 | (NV50_CB_PFP << 12));
 
 	/* Vertex array limits - max them out */
 	for (i = 0; i < 16; i++) {
-		so_method(so, screen->tesla, NV50TCL_VERTEX_ARRAY_LIMIT_HIGH(i), 2);
-		so_data  (so, 0x000000ff);
-		so_data  (so, 0xffffffff);
+		BEGIN_RING(chan, screen->tesla,
+			   NV50TCL_VERTEX_ARRAY_LIMIT_HIGH(i), 2);
+		OUT_RING  (chan, 0x000000ff);
+		OUT_RING  (chan, 0xffffffff);
 	}
 
-	so_method(so, screen->tesla, NV50TCL_DEPTH_RANGE_NEAR(0), 2);
-	so_data  (so, fui(0.0));
-	so_data  (so, fui(1.0));
+	BEGIN_RING(chan, screen->tesla, NV50TCL_DEPTH_RANGE_NEAR(0), 2);
+	OUT_RINGf (chan, 0.0f);
+	OUT_RINGf (chan, 1.0f);
 
 	/* no dynamic combination of TIC & TSC entries => only BIND_TIC used */
-	so_method(so, screen->tesla, NV50TCL_LINKED_TSC, 1);
-	so_data  (so, 1);
+	BEGIN_RING(chan, screen->tesla, NV50TCL_LINKED_TSC, 1);
+	OUT_RING  (chan, 1);
 
-	so_method(so, screen->tesla, NV50TCL_EDGEFLAG_ENABLE, 1);
-	so_data  (so, 1); /* default edgeflag to TRUE */
+	BEGIN_RING(chan, screen->tesla, NV50TCL_EDGEFLAG_ENABLE, 1);
+	OUT_RING  (chan, 1); /* default edgeflag to TRUE */
 
-	so_emit(chan, so);
-	so_ref (so, &screen->static_init);
-	so_ref (NULL, &so);
-	nouveau_pushbuf_flush(chan, 0);
+	FIRE_RING (chan);
 
 	screen->force_push = debug_get_bool_option("NV50_ALWAYS_PUSH", FALSE);
 	return pscreen;
diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h
index 22fe7e46bb..092333a3b1 100644
--- a/src/gallium/drivers/nv50/nv50_screen.h
+++ b/src/gallium/drivers/nv50/nv50_screen.h
@@ -27,8 +27,6 @@ struct nv50_screen {
 	struct nouveau_bo *tic;
 	struct nouveau_bo *tsc;
 
-	struct nouveau_stateobj *static_init;
-
 	boolean force_push;
 };
 
@@ -38,4 +36,6 @@ nv50_screen(struct pipe_screen *screen)
 	return (struct nv50_screen *)screen;
 }
 
+extern void nv50_screen_relocs(struct nv50_screen *);
+
 #endif
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index 356bdcd74b..565cf2f8b3 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -436,7 +436,7 @@ nv50_state_validate(struct nv50_context *nv50, unsigned wait_dwords)
 	so_emit_reloc_markers(chan, nv50->state.hw[3]); /* vp */
 	so_emit_reloc_markers(chan, nv50->state.hw[4]); /* fp */
 	so_emit_reloc_markers(chan, nv50->state.hw[17]); /* vb */
-	so_emit_reloc_markers(chan, nv50->screen->static_init);
+	nv50_screen_relocs(nv50->screen);
 
 	/* No idea.. */
 	BEGIN_RING(chan, tesla, 0x142c, 1);
diff --git a/src/gallium/drivers/r300/r300_blit.c b/src/gallium/drivers/r300/r300_blit.c
index f99f9dffaa..bcdf950df9 100644
--- a/src/gallium/drivers/r300/r300_blit.c
+++ b/src/gallium/drivers/r300/r300_blit.c
@@ -149,6 +149,9 @@ void r300_surface_copy(struct pipe_context* pipe,
             case 4:
                 new_format = PIPE_FORMAT_B8G8R8A8_UNORM;
                 break;
+            case 8:
+                new_format = PIPE_FORMAT_R16G16B16A16_UNORM;
+                break;
             default:
                 debug_printf("r300: surface_copy: Unhandled format: %s. Falling back to software.\n"
                              "r300: surface_copy: Software fallback doesn't work for tiled textures.\n",
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 261117a366..e805a84290 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -33,10 +33,11 @@
 #include "r300_query.h"
 #include "r300_render.h"
 #include "r300_screen.h"
+#include "r300_screen_buffer.h"
 #include "r300_state_invariant.h"
 #include "r300_texture.h"
-
-#include "radeon_winsys.h"
+#include "r300_transfer.h"
+#include "r300_winsys.h"
 
 static void r300_destroy_context(struct pipe_context* context)
 {
@@ -177,7 +178,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     /* Open up the OQ BO. */
     r300->oqbo = pipe_buffer_create(screen, 4096,
-            PIPE_BUFFER_USAGE_VERTEX, 4096);
+            PIPE_BUFFER_USAGE_PIXEL, 4096);
     make_empty_list(&r300->query_list);
 
     r300_init_flush_functions(r300);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index ca15d87d1f..bcf87b3866 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -31,6 +31,7 @@
 #include "util/u_inlines.h"
 #include "util/u_transfer.h"
 
+#include "r300_defines.h"
 #include "r300_screen.h"
 
 struct u_upload_mgr;
@@ -136,8 +137,6 @@ struct r300_texture_format_state {
     uint32_t format2; /* R300_TX_FORMAT2: 0x4500 */
 };
 
-#define R300_MAX_TEXTURE_LEVELS 13
-
 struct r300_texture_fb_state {
     /* Colorbuffer. */
     uint32_t colorpitch[R300_MAX_TEXTURE_LEVELS]; /* R300_RB3D_COLORPITCH[0-3]*/
@@ -196,12 +195,6 @@ struct r300_ztop_state {
     uint32_t z_buffer_top;      /* R300_ZB_ZTOP: 0x4f14 */
 };
 
-#define R300_NEW_FRAGMENT_SHADER 0x00000020
-#define R300_NEW_FRAGMENT_SHADER_CONSTANTS    0x00000040
-#define R300_NEW_VERTEX_SHADER_CONSTANTS    0x10000000
-#define R300_NEW_QUERY           0x40000000
-#define R300_NEW_KITCHEN_SINK    0x7fffffff
-
 /* The next several objects are not pure Radeon state; they inherit from
  * various Gallium classes. */
 
@@ -239,12 +232,6 @@ struct r300_query {
     struct r300_query* next;
 };
 
-enum r300_buffer_tiling {
-    R300_BUFFER_LINEAR = 0,
-    R300_BUFFER_TILED,
-    R300_BUFFER_SQUARETILED
-};
-
 struct r300_texture {
     /* Parent class */
     struct u_resource b;
@@ -444,4 +431,3 @@ static INLINE void CTX_DBG(struct r300_context * ctx, unsigned flags,
 #define DBG     CTX_DBG
 
 #endif /* R300_CONTEXT_H */
-
diff --git a/src/gallium/drivers/r300/r300_cs.h b/src/gallium/drivers/r300/r300_cs.h
index ad07efbffd..456b2ec7b9 100644
--- a/src/gallium/drivers/r300/r300_cs.h
+++ b/src/gallium/drivers/r300/r300_cs.h
@@ -26,8 +26,7 @@
 #include "util/u_math.h"
 
 #include "r300_reg.h"
-
-#include "radeon_winsys.h"
+#include "r300_winsys.h"
 
 /* Yes, I know macros are ugly. However, they are much prettier than the code
  * that they neatly hide away, and don't have the cost of function setup,so
diff --git a/src/gallium/drivers/r300/r300_defines.h b/src/gallium/drivers/r300/r300_defines.h
new file mode 100644
index 0000000000..2e04876de9
--- /dev/null
+++ b/src/gallium/drivers/r300/r300_defines.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright 2010 Marek Olšák <maraeo@gmail.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE. */
+
+#ifndef R300_DEFINES_H
+#define R300_DEFINES_H
+
+#include "pipe/p_defines.h"
+
+#define R300_MAX_TEXTURE_LEVELS         13
+#define R300_MAX_DRAW_VBO_SIZE          (1024 * 1024)
+
+#define R300_TEXTURE_USAGE_TRANSFER     PIPE_TEXTURE_USAGE_CUSTOM
+
+/* Non-atom dirty state flags. */
+#define R300_NEW_FRAGMENT_SHADER                0x00000020
+#define R300_NEW_FRAGMENT_SHADER_CONSTANTS      0x00000040
+#define R300_NEW_VERTEX_SHADER_CONSTANTS        0x10000000
+#define R300_NEW_QUERY                          0x40000000
+#define R300_NEW_KITCHEN_SINK                   0x7fffffff
+
+/* Tiling flags. */
+enum r300_buffer_tiling {
+    R300_BUFFER_LINEAR = 0,
+    R300_BUFFER_TILED,
+    R300_BUFFER_SQUARETILED
+};
+
+#endif
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 65a4a6388d..fa641ae8d0 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -41,9 +41,6 @@
 #include "r300_render.h"
 #include "r300_state_derived.h"
 
-/* r300_render: Vertex and index buffer primitive emission. */
-#define R300_MAX_VBO_SIZE  (1024 * 1024)
-
 /* XXX The DRM rejects VAP_ALT_NUM_VERTICES.. */
 //#define ENABLE_ALT_NUM_VERTS
 
@@ -709,9 +706,9 @@ static boolean r300_render_allocate_vertices(struct vbuf_render* render,
         r300render->vbo = pipe_buffer_create(screen,
                                              64,
                                              PIPE_BUFFER_USAGE_VERTEX,
-                                             R300_MAX_VBO_SIZE);
+                                             R300_MAX_DRAW_VBO_SIZE);
         r300render->vbo_offset = 0;
-        r300render->vbo_size = R300_MAX_VBO_SIZE;
+        r300render->vbo_size = R300_MAX_DRAW_VBO_SIZE;
     }
 
     r300render->vertex_size = vertex_size;
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index ec7c65f468..c0620972f1 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -26,10 +26,8 @@
 
 #include "r300_context.h"
 #include "r300_texture.h"
-
-#include "radeon_winsys.h"
-
 #include "r300_screen_buffer.h"
+#include "r300_winsys.h"
 
 /* Return the identifier behind whom the brave coders responsible for this
  * amalgamation of code, sweat, and duct tape, routinely obscure their names.
diff --git a/src/gallium/drivers/r300/r300_screen.h b/src/gallium/drivers/r300/r300_screen.h
index 9972a7da07..f49aae715a 100644
--- a/src/gallium/drivers/r300/r300_screen.h
+++ b/src/gallium/drivers/r300/r300_screen.h
@@ -28,10 +28,6 @@
 
 #include "r300_chipset.h"
 
-#define R300_TEXTURE_USAGE_TRANSFER PIPE_TEXTURE_USAGE_CUSTOM
-
-struct radeon_winsys;
-
 struct r300_screen {
     /* Parent class */
     struct pipe_screen screen;
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 429d371f64..4759ccb5ec 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -38,8 +38,7 @@
 #include "r300_state_inlines.h"
 #include "r300_fs.h"
 #include "r300_vs.h"
-
-#include "radeon_winsys.h"
+#include "r300_winsys.h"
 
 /* r300_state: Functions used to intialize state context by translating
  * Gallium state objects into semi-native r300 state objects. */
@@ -528,8 +527,8 @@ static void r300_fb_update_tiling_flags(struct r300_context *r300,
         if (tex) {
             r300->rws->buffer_set_tiling(r300->rws, tex->buffer,
                                             tex->pitch[0],
-                                            tex->microtile != 0,
-                                            tex->macrotile != 0);
+                                            tex->microtile,
+                                            tex->macrotile);
         }
     }
     if (old_state->zsbuf &&
@@ -540,8 +539,8 @@ static void r300_fb_update_tiling_flags(struct r300_context *r300,
         if (tex) {
             r300->rws->buffer_set_tiling(r300->rws, tex->buffer,
                                             tex->pitch[0],
-                                            tex->microtile != 0,
-                                            tex->macrotile != 0);
+                                            tex->microtile,
+                                            tex->macrotile);
         }
     }
 
@@ -552,8 +551,8 @@ static void r300_fb_update_tiling_flags(struct r300_context *r300,
 
         r300->rws->buffer_set_tiling(r300->rws, tex->buffer,
                                         tex->pitch[level],
-                                        tex->microtile != 0,
-                                        tex->mip_macrotile[level] != 0);
+                                        tex->microtile,
+                                        tex->mip_macrotile[level]);
     }
     if (new_state->zsbuf) {
         tex = (struct r300_texture*)new_state->zsbuf->texture;
@@ -561,8 +560,8 @@ static void r300_fb_update_tiling_flags(struct r300_context *r300,
 
         r300->rws->buffer_set_tiling(r300->rws, tex->buffer,
                                         tex->pitch[level],
-                                        tex->microtile != 0,
-                                        tex->mip_macrotile[level] != 0);
+                                        tex->microtile,
+                                        tex->mip_macrotile[level]);
     }
 }
 
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index 1e3052e51c..f4040b2431 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -32,8 +32,7 @@
 #include "r300_transfer.h"
 #include "r300_screen.h"
 #include "r300_state_inlines.h"
-
-#include "radeon_winsys.h"
+#include "r300_winsys.h"
 
 #define TILE_WIDTH 0
 #define TILE_HEIGHT 1
@@ -47,6 +46,18 @@ static const unsigned microblock_table[5][3][2] = {
     {{ 2, 1}, {0, 0}, {0, 0}}  /* 128 bits per pixel */
 };
 
+/* Return true for non-compressed and non-YUV formats. */
+static boolean r300_format_is_plain(enum pipe_format format)
+{
+    const struct util_format_description *desc = util_format_description(format);
+
+    if (!format) {
+        return FALSE;
+    }
+
+    return desc->layout == UTIL_FORMAT_LAYOUT_PLAIN;
+}
+
 /* Translate a pipe_format into a useful texture format for sampling.
  *
  * Some special formats are translated directly using R300_EASY_TX_FORMAT,
@@ -641,7 +652,7 @@ unsigned r300_texture_get_stride(struct r300_screen* screen,
 
     width = u_minify(tex->b.b.width0, level);
 
-    if (!util_format_is_compressed(tex->b.b.format)) {
+    if (r300_format_is_plain(tex->b.b.format)) {
         tile_width = r300_texture_get_tile_size(tex, TILE_WIDTH,
                                                 tex->mip_macrotile[level]);
         width = align(width, tile_width);
@@ -659,7 +670,7 @@ static unsigned r300_texture_get_nblocksy(struct r300_texture* tex,
 
     height = u_minify(tex->b.b.height0, level);
 
-    if (!util_format_is_compressed(tex->b.b.format)) {
+    if (r300_format_is_plain(tex->b.b.format)) {
         tile_height = r300_texture_get_tile_size(tex, TILE_HEIGHT,
                                                  tex->mip_macrotile[level]);
         height = align(height, tile_height);
@@ -716,10 +727,11 @@ static void r300_setup_flags(struct r300_texture* tex)
 static void r300_setup_tiling(struct pipe_screen *screen,
                               struct r300_texture *tex)
 {
+    struct r300_winsys_screen *rws = (struct r300_winsys_screen *)screen->winsys;
     enum pipe_format format = tex->b.b.format;
     boolean rv350_mode = r300_screen(screen)->caps->family >= CHIP_FAMILY_RV350;
 
-    if (util_format_is_compressed(format)) {
+    if (!r300_format_is_plain(format)) {
         return;
     }
 
@@ -735,12 +747,12 @@ static void r300_setup_tiling(struct pipe_screen *screen,
             tex->microtile = R300_BUFFER_TILED;
             break;
 
-        /* XXX Square-tiling doesn't work with kernel older than 2.6.34,
-         * XXX need to check the DRM version */
-        /*case 2:
+        case 2:
         case 8:
-            tex->microtile = R300_BUFFER_SQUARETILED;
-            break;*/
+            if (rws->get_value(rws, R300_VID_SQUARE_TILING_SUPPORT)) {
+                tex->microtile = R300_BUFFER_SQUARETILED;
+            }
+            break;
     }
 
     /* Set macrotiling. */
@@ -843,8 +855,8 @@ struct pipe_resource* r300_texture_create(struct pipe_screen* screen,
 				     tex->size);
     rws->buffer_set_tiling(rws, tex->buffer,
 			   tex->pitch[0],
-			   tex->microtile != R300_BUFFER_LINEAR,
-			   tex->macrotile != R300_BUFFER_LINEAR);
+			   tex->microtile,
+			   tex->macrotile);
 
     if (!tex->buffer) {
         FREE(tex);
diff --git a/src/gallium/drivers/r300/r300_winsys.h b/src/gallium/drivers/r300/r300_winsys.h
index e478802c2f..61ec49c619 100644
--- a/src/gallium/drivers/r300/r300_winsys.h
+++ b/src/gallium/drivers/r300/r300_winsys.h
@@ -30,6 +30,8 @@
 #include "pipe/p_defines.h"
 #include "pipe/p_state.h"
 
+#include "r300_defines.h"
+
 struct r300_winsys_screen;
 
 /* Creates a new r300 screen. */
@@ -42,10 +44,9 @@ enum r300_value_id {
     R300_VID_PCI_ID,
     R300_VID_GB_PIPES,
     R300_VID_Z_PIPES,
+    R300_VID_SQUARE_TILING_SUPPORT
 };
 
-#define R300_USAGE_FLAG_DONT_SYNC (1 << 17)
-
 struct r300_winsys_screen {
     void (*destroy)(struct r300_winsys_screen *ws);
     
@@ -143,8 +144,8 @@ struct r300_winsys_screen {
     void (*buffer_set_tiling)(struct r300_winsys_screen *winsys,
                               struct r300_winsys_buffer *buffer,
                               uint32_t pitch,
-                              boolean microtiled,
-                              boolean macrotiled);
+                              enum r300_buffer_tiling microtiled,
+                              enum r300_buffer_tiling macrotiled);
 
     uint32_t (*get_value)(struct r300_winsys_screen *winsys,
 			  enum r300_value_id vid);
diff --git a/src/gallium/include/state_tracker/dri1_api.h b/src/gallium/include/state_tracker/dri1_api.h
index 0dc91ca58d..777bb9590e 100644
--- a/src/gallium/include/state_tracker/dri1_api.h
+++ b/src/gallium/include/state_tracker/dri1_api.h
@@ -7,7 +7,7 @@
 
 #include "state_tracker/drm_api.h"
 
-#include <drm.h>
+struct drm_clip_rect;
 
 struct pipe_screen;
 struct pipe_winsys;
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.c b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
index 72d571fc22..acfb36f7e0 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
@@ -92,6 +92,10 @@ static void do_ioctls(int fd, struct radeon_libdrm_winsys* winsys)
         exit(1);
     }
 
+    // Supported since 2.1.0.
+    winsys->squaretiling = version->version_major > 2 ||
+                           version->version_minor >= 1;
+
     info.request = RADEON_INFO_DEVICE_ID;
     retval = drmCommandWriteRead(fd, DRM_RADEON_INFO, &info, sizeof(info));
     if (retval) {
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c
index bdaf95d793..d9870e8519 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c
@@ -71,17 +71,22 @@ radeon_drm_buffer_map(struct pb_buffer *_buf,
     struct radeon_drm_buffer *buf = radeon_drm_buffer(_buf);
     int write;
 
+    if (flags & PIPE_BUFFER_USAGE_DONTBLOCK) {
+	if ((_buf->base.usage & PIPE_BUFFER_USAGE_VERTEX) ||
+	    (_buf->base.usage & PIPE_BUFFER_USAGE_INDEX))
+	    if (radeon_bo_is_referenced_by_cs(buf->bo, buf->mgr->rws->cs))
+		return NULL;
+    }
+
     if (buf->bo->ptr != NULL)
 	return buf->bo->ptr;
-    
+
     if (flags & PIPE_BUFFER_USAGE_DONTBLOCK) {
         uint32_t domain;
-
         if (radeon_bo_is_busy(buf->bo, &domain))
             return NULL;
     }
 
-
     if (radeon_bo_is_referenced_by_cs(buf->bo, buf->mgr->rws->cs)) {
         buf->mgr->rws->flush_cb(buf->mgr->rws->flush_data);
     }
@@ -300,14 +305,19 @@ boolean radeon_drm_bufmgr_get_handle(struct pb_buffer *_buf,
 }
 					   
 
-void radeon_drm_bufmgr_set_tiling(struct pb_buffer *_buf, boolean microtiled, boolean macrotiled, uint32_t pitch)
+void radeon_drm_bufmgr_set_tiling(struct pb_buffer *_buf,
+                                  enum r300_buffer_tiling microtiled,
+                                  enum r300_buffer_tiling macrotiled,
+                                  uint32_t pitch)
 {
     struct radeon_drm_buffer *buf = get_drm_buffer(_buf);
     uint32_t flags = 0, old_flags, old_pitch;
-    if (microtiled)
-	flags |= RADEON_BO_FLAGS_MICRO_TILE;
-    if (macrotiled)
-	flags |= RADEON_BO_FLAGS_MACRO_TILE;
+    if (microtiled == R300_BUFFER_TILED)
+        flags |= RADEON_BO_FLAGS_MICRO_TILE;
+    else if (microtiled == R300_BUFFER_SQUARETILED)
+        flags |= RADEON_BO_FLAGS_MICRO_TILE_SQUARE;
+    if (macrotiled == R300_BUFFER_TILED)
+        flags |= RADEON_BO_FLAGS_MACRO_TILE;
 
     radeon_bo_get_tiling(buf->bo, &old_flags, &old_pitch);
 
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_r300.c b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
index e0895c1113..2f89484f7d 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_r300.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
@@ -43,6 +43,9 @@ radeon_r300_winsys_buffer_create(struct r300_winsys_screen *rws,
 
     if (usage & PIPE_BUFFER_USAGE_CONSTANT)
         provider = ws->mman;
+    else if ((usage & PIPE_BUFFER_USAGE_VERTEX) ||
+	     (usage & PIPE_BUFFER_USAGE_INDEX))
+	provider = ws->cman;
     else
         provider = ws->kman;
     buffer = provider->create_buffer(provider, size, &desc);
@@ -61,8 +64,8 @@ static void radeon_r300_winsys_buffer_destroy(struct r300_winsys_buffer *buf)
 static void radeon_r300_winsys_buffer_set_tiling(struct r300_winsys_screen *rws,
 						  struct r300_winsys_buffer *buf,
 						  uint32_t pitch,
-						  boolean microtiled,
-						  boolean macrotiled)
+						  enum r300_buffer_tiling microtiled,
+						  enum r300_buffer_tiling macrotiled)
 {
     struct pb_buffer *_buf = radeon_pb_buffer(buf);
     radeon_drm_bufmgr_set_tiling(_buf, microtiled, macrotiled, pitch);
@@ -249,6 +252,8 @@ static uint32_t radeon_get_value(struct r300_winsys_screen *rws,
 	return ws->gb_pipes;
     case R300_VID_Z_PIPES:
 	return ws->z_pipes;
+    case R300_VID_SQUARE_TILING_SUPPORT:
+        return ws->squaretiling;
     }
     return 0;
 }
@@ -259,6 +264,7 @@ radeon_winsys_destroy(struct r300_winsys_screen *rws)
     struct radeon_libdrm_winsys *ws = (struct radeon_libdrm_winsys *)rws;
     radeon_cs_destroy(ws->cs);
 
+    ws->cman->destroy(ws->cman);
     ws->kman->destroy(ws->kman);
     ws->mman->destroy(ws->mman);
 
@@ -280,6 +286,10 @@ radeon_setup_winsys(int fd, struct radeon_libdrm_winsys* ws)
     if (!ws->kman)
 	goto fail;
 
+    ws->cman = pb_cache_manager_create(ws->kman, 100000);
+    if (!ws->cman)
+	goto fail;
+
     ws->mman = pb_malloc_bufmgr_create();
     if (!ws->mman)
 	goto fail;
@@ -324,7 +334,8 @@ fail:
     if (ws->bom)
 	radeon_bo_manager_gem_dtor(ws->bom);
 
-
+    if (ws->cman)
+	ws->cman->destroy(ws->cman);
     if (ws->kman)
 	ws->kman->destroy(ws->kman);
     if (ws->mman)
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
index 16cc701ad6..4260dbaad7 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
@@ -38,6 +38,8 @@ struct radeon_libdrm_winsys {
 
     struct pb_manager *kman;
 
+    struct pb_manager *cman;
+
     struct pb_manager *mman;
 
     /* PCI ID */
@@ -55,6 +57,9 @@ struct radeon_libdrm_winsys {
     /* VRAM size. */
     uint32_t vram_size;
 
+    /* Square tiling support. */
+    boolean squaretiling;
+
     /* DRM FD */
     int fd;
 
diff --git a/src/glx/drisw_glx.c b/src/glx/drisw_glx.c
index 1b94a56fd1..3db2d63f1f 100644
--- a/src/glx/drisw_glx.c
+++ b/src/glx/drisw_glx.c
@@ -85,11 +85,14 @@ XCreateDrawable(__GLXDRIdrawablePrivate * pdp,
    visMask = (VisualScreenMask | VisualIDMask);
    pdp->visinfo = XGetVisualInfo(dpy, visMask, &visTemp, &num_visuals);
 
-   pdp->ximage = XCreateImage(dpy, pdp->visinfo->visual, pdp->visinfo->depth, ZPixmap, 0,       /* format, offset */
-                              NULL,     /* data */
-                              0, 0,     /* size */
-                              32,       /* bitmap_pad */
-                              0);       /* bytes_per_line */
+   pdp->ximage = XCreateImage(dpy,
+                              pdp->visinfo->visual,
+                              pdp->visinfo->depth,
+                              ZPixmap, 0,             /* format, offset */
+                              NULL,                   /* data */
+                              0, 0,                   /* size */
+                              32,                     /* bitmap_pad */
+                              0);                     /* bytes_per_line */
 
    /* get the true number of bits per pixel */
    pdp->bpp = pdp->ximage->bits_per_pixel;
@@ -334,10 +337,17 @@ driCreateDrawable(__GLXscreenConfigs * psc,
    return pdraw;
 }
 
-static void
-driSwapBuffers(__GLXDRIdrawable * pdraw)
+static int64_t
+driSwapBuffers(__GLXDRIdrawable * pdraw,
+               int64_t target_msc, int64_t divisor, int64_t remainder)
 {
+   (void) target_msc;
+   (void) divisor;
+   (void) remainder;
+
    (*pdraw->psc->core->swapBuffers) (pdraw->driDrawable);
+
+   return 0;
 }
 
 static void
diff --git a/src/mesa/drivers/dri/common/dri_sw.c b/src/mesa/drivers/dri/common/drisw_util.c
index b7f9036f47..141ca302d0 100644
--- a/src/mesa/drivers/dri/common/dri_sw.c
+++ b/src/mesa/drivers/dri/common/drisw_util.c
@@ -22,12 +22,12 @@
  */
 
 /**
- * \file dri_sw.c
+ * \file drisw_util.c
  *
  * DRISW utility functions, i.e. dri_util.c stripped from drm-specific bits.
  */
 
-#include "dri_sw.h"
+#include "drisw_util.h"
 #include "utils.h"
 
 
diff --git a/src/mesa/drivers/dri/common/dri_sw.h b/src/mesa/drivers/dri/common/drisw_util.h
index e353e26b34..e353e26b34 100644
--- a/src/mesa/drivers/dri/common/dri_sw.h
+++ b/src/mesa/drivers/dri/common/drisw_util.h
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_render.h b/src/mesa/drivers/dri/nouveau/nouveau_render.h
index bff0ccfd76..923b79b2cf 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_render.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_render.h
@@ -32,8 +32,8 @@
 struct nouveau_array_state;
 
 typedef void (*dispatch_t)(GLcontext *, unsigned int, int, unsigned int);
-typedef unsigned (*extract_u_t)(struct nouveau_array_state *a, int i, int j);
-typedef float (*extract_f_t)(struct nouveau_array_state *a, int i, int j);
+typedef unsigned (*extract_u_t)(struct nouveau_array_state *, int, int);
+typedef float (*extract_f_t)(struct nouveau_array_state *, int, int);
 
 struct nouveau_attr_info {
 	int vbo_index;
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c
index ef2cc787de..7697090f6a 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c
@@ -189,6 +189,7 @@ nouveau_enable(GLcontext *ctx, GLenum cap, GLboolean state)
 	case GL_LIGHTING:
 		context_dirty(ctx, FRAG);
 		context_dirty(ctx, MODELVIEW);
+		context_dirty(ctx, LIGHT_MODEL);
 		context_dirty(ctx, LIGHT_ENABLE);
 
 		for (i = 0; i < MAX_LIGHTS; i++) {
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_texture.h b/src/mesa/drivers/dri/nouveau/nouveau_texture.h
index b91facbdeb..251f537bba 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_texture.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_texture.h
@@ -41,7 +41,7 @@ struct nouveau_texture {
 #define to_nouveau_texture(x) ((struct nouveau_texture *)(x))
 
 #define texture_dirty(t) \
-	to_nouveau_texture(t)->dirty = GL_TRUE;
+	to_nouveau_texture(t)->dirty = GL_TRUE
 
 void
 nouveau_set_texbuffer(__DRIcontext *dri_ctx,
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c b/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c
index 0c29eec8ee..e5858f8268 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c
@@ -85,6 +85,18 @@ vbo_deinit_array(struct nouveau_array_state *a)
 	a->fields = 0;
 }
 
+static int
+get_array_stride(GLcontext *ctx, const struct gl_client_array *a)
+{
+	struct nouveau_render_state *render = to_render_state(ctx);
+
+	if (render->mode == VBO && !_mesa_is_bufferobj(a->BufferObj))
+		/* Pack client buffers. */
+		return align(_mesa_sizeof_type(a->Type) * a->Size, 4);
+	else
+		return a->StrideB;
+}
+
 static void
 vbo_init_arrays(GLcontext *ctx, const struct _mesa_index_buffer *ib,
 		const struct gl_client_array **arrays)
@@ -101,18 +113,10 @@ vbo_init_arrays(GLcontext *ctx, const struct _mesa_index_buffer *ib,
 
 		if (attr >= 0) {
 			const struct gl_client_array *array = arrays[attr];
-			int stride;
-
-			if (render->mode == VBO &&
-			    !_mesa_is_bufferobj(array->BufferObj))
-				/* Pack client buffers. */
-				stride = align(_mesa_sizeof_type(array->Type)
-					       * array->Size, 4);
-			else
-				stride = array->StrideB;
 
 			vbo_init_array(&render->attrs[attr], attr,
-				       stride, array->Size, array->Type,
+				       get_array_stride(ctx, array),
+				       array->Size, array->Type,
 				       array->BufferObj, array->Ptr,
 				       render->mode == IMM);
 		}
@@ -245,7 +249,7 @@ vbo_choose_attrs(GLcontext *ctx, const struct gl_client_array **arrays)
 	vbo_emit_attr(ctx, arrays, VERT_ATTRIB_POS);
 }
 
-static unsigned
+static int
 get_max_client_stride(GLcontext *ctx, const struct gl_client_array **arrays)
 {
 	struct nouveau_render_state *render = to_render_state(ctx);
@@ -258,7 +262,7 @@ get_max_client_stride(GLcontext *ctx, const struct gl_client_array **arrays)
 			const struct gl_client_array *a = arrays[attr];
 
 			if (!_mesa_is_bufferobj(a->BufferObj))
-				s = MAX2(a->StrideB, s);
+				s = MAX2(s, get_array_stride(ctx, a));
 		}
 	}
 
@@ -327,6 +331,7 @@ vbo_bind_vertices(GLcontext *ctx, const struct gl_client_array **arrays,
 				* array->StrideB;
 
 			if (a->bo) {
+				/* Array in a buffer obj. */
 				a->offset = (intptr_t)array->Ptr + delta;
 			} else {
 				int j, n = max_index - min_index + 1;
@@ -334,6 +339,8 @@ vbo_bind_vertices(GLcontext *ctx, const struct gl_client_array **arrays,
 				char *dp = get_scratch_vbo(ctx, n * a->stride,
 							   &a->bo, &a->offset);
 
+				/* Array in client memory, move it to
+				 * a scratch buffer obj. */
 				for (j = 0; j < n; j++)
 					memcpy(dp + j * a->stride,
 					       sp + j * array->StrideB,
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_fb.c b/src/mesa/drivers/dri/nouveau/nv10_state_fb.c
index f7c3d36e1c..a2fcb6b695 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state_fb.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state_fb.c
@@ -172,15 +172,13 @@ nv10_emit_viewport(GLcontext *ctx, int emit)
 	struct nouveau_grobj *celsius = context_eng3d(ctx);
 	struct gl_framebuffer *fb = ctx->DrawBuffer;
 	float a[4] = {};
-	int i;
 
 	get_viewport_translate(ctx, a);
 	a[0] -= 2048;
 	a[1] -= 2048;
 
 	BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_TRANSLATE_X, 4);
-	for (i = 0; i < 4; i++)
-		OUT_RINGf(chan, a[i]);
+	OUT_RINGp(chan, a, 4);
 
 	BEGIN_RING(chan, celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
 	OUT_RING(chan, (fb->Width - 1) << 16 | 0x08000800);
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_tex.c b/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
index 35f41d7295..6dedb18c72 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
@@ -54,10 +54,7 @@ nv10_emit_tex_gen(GLcontext *ctx, int emit)
 			if (k) {
 				BEGIN_RING(chan, celsius,
 					   TX_GEN_COEFF(i, j), 4);
-				OUT_RINGf(chan, k[0]);
-				OUT_RINGf(chan, k[1]);
-				OUT_RINGf(chan, k[2]);
-				OUT_RINGf(chan, k[3]);
+				OUT_RINGp(chan, k, 4);
 			}
 
 			BEGIN_RING(chan, celsius, TX_GEN_MODE(i, j), 1);
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c b/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c
index 2624c9bf30..0e592a1629 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c
@@ -140,9 +140,7 @@ nv10_emit_fog(GLcontext *ctx, int emit)
 	OUT_RING(chan, pack_rgba_f(MESA_FORMAT_RGBA8888_REV, f->Color));
 
 	BEGIN_RING(chan, celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3);
-	OUT_RINGf(chan, k[0]);
-	OUT_RINGf(chan, k[1]);
-	OUT_RINGf(chan, k[2]);
+	OUT_RINGp(chan, k, 3);
 
 	context_dirty(ctx, FRAG);
 }
@@ -284,9 +282,7 @@ nv10_emit_light_source(GLcontext *ctx, int emit)
 
 	if (l->_Flags & LIGHT_POSITIONAL) {
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_POSITION_X(i), 3);
-		OUT_RINGf(chan, l->_Position[0]);
-		OUT_RINGf(chan, l->_Position[1]);
-		OUT_RINGf(chan, l->_Position[2]);
+		OUT_RINGp(chan, l->_Position, 3);
 
 		BEGIN_RING(chan, celsius,
 			   NV10TCL_LIGHT_ATTENUATION_CONSTANT(i), 3);
@@ -296,14 +292,10 @@ nv10_emit_light_source(GLcontext *ctx, int emit)
 
 	} else {
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_DIRECTION_X(i), 3);
-		OUT_RINGf(chan, l->_VP_inf_norm[0]);
-		OUT_RINGf(chan, l->_VP_inf_norm[1]);
-		OUT_RINGf(chan, l->_VP_inf_norm[2]);
+		OUT_RINGp(chan, l->_VP_inf_norm, 3);
 
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_HALF_VECTOR_X(i), 3);
-		OUT_RINGf(chan, l->_h_inf_norm[0]);
-		OUT_RINGf(chan, l->_h_inf_norm[1]);
-		OUT_RINGf(chan, l->_h_inf_norm[2]);
+		OUT_RINGp(chan, l->_h_inf_norm, 3);
 	}
 
 	if (l->_Flags & LIGHT_SPOT) {
@@ -312,13 +304,7 @@ nv10_emit_light_source(GLcontext *ctx, int emit)
 		nv10_get_spot_coeff(l, k);
 
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_SPOT_CUTOFF_A(i), 7);
-		OUT_RINGf(chan, k[0]);
-		OUT_RINGf(chan, k[1]);
-		OUT_RINGf(chan, k[2]);
-		OUT_RINGf(chan, k[3]);
-		OUT_RINGf(chan, k[4]);
-		OUT_RINGf(chan, k[5]);
-		OUT_RINGf(chan, k[6]);
+		OUT_RINGp(chan, k, 7);
 	}
 }
 
@@ -350,15 +336,11 @@ nv10_emit_material_ambient(GLcontext *ctx, int emit)
 	}
 
 	BEGIN_RING(chan, celsius, NV10TCL_LIGHT_MODEL_AMBIENT_R, 3);
-	OUT_RINGf(chan, c_scene[0]);
-	OUT_RINGf(chan, c_scene[1]);
-	OUT_RINGf(chan, c_scene[2]);
+	OUT_RINGp(chan, c_scene, 3);
 
 	if (ctx->Light.ColorMaterialEnabled) {
 		BEGIN_RING(chan, celsius, NV10TCL_MATERIAL_FACTOR_R, 3);
-		OUT_RINGf(chan, c_factor[0]);
-		OUT_RINGf(chan, c_factor[1]);
-		OUT_RINGf(chan, c_factor[2]);
+		OUT_RINGp(chan, c_factor, 3);
 	}
 
 	foreach(l, &ctx->Light.EnabledList) {
@@ -368,9 +350,7 @@ nv10_emit_material_ambient(GLcontext *ctx, int emit)
 				  l->_MatAmbient[0]);
 
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_AMBIENT_R(i), 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -392,9 +372,7 @@ nv10_emit_material_diffuse(GLcontext *ctx, int emit)
 				  l->_MatDiffuse[0]);
 
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_DIFFUSE_R(i), 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -412,9 +390,7 @@ nv10_emit_material_specular(GLcontext *ctx, int emit)
 				  l->_MatSpecular[0]);
 
 		BEGIN_RING(chan, celsius, NV10TCL_LIGHT_SPECULAR_R(i), 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -455,12 +431,7 @@ nv10_emit_material_shininess(GLcontext *ctx, int emit)
 		k);
 
 	BEGIN_RING(chan, celsius, NV10TCL_MATERIAL_SHININESS(0), 6);
-	OUT_RINGf(chan, k[0]);
-	OUT_RINGf(chan, k[1]);
-	OUT_RINGf(chan, k[2]);
-	OUT_RINGf(chan, k[3]);
-	OUT_RINGf(chan, k[4]);
-	OUT_RINGf(chan, k[5]);
+	OUT_RINGp(chan, k, 6);
 }
 
 void
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_fb.c b/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
index d638541df9..21da4f7af1 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
@@ -106,13 +106,11 @@ nv20_emit_viewport(GLcontext *ctx, int emit)
 	struct nouveau_grobj *kelvin = context_eng3d(ctx);
 	struct gl_framebuffer *fb = ctx->DrawBuffer;
 	float a[4] = {};
-	int i;
 
 	get_viewport_translate(ctx, a);
 
 	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_TRANSLATE_X, 4);
-	for (i = 0; i < 4; i++)
-		OUT_RINGf(chan, a[i]);
+	OUT_RINGp(chan, a, 4);
 
 	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 1);
 	OUT_RING(chan, (fb->Width - 1) << 16);
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
index bb8a79c2c9..e46118e4fc 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
@@ -53,10 +53,7 @@ nv20_emit_tex_gen(GLcontext *ctx, int emit)
 
 			if (k) {
 				BEGIN_RING(chan, kelvin, TX_GEN_COEFF(i, j), 4);
-				OUT_RINGf(chan, k[0]);
-				OUT_RINGf(chan, k[1]);
-				OUT_RINGf(chan, k[2]);
-				OUT_RINGf(chan, k[3]);
+				OUT_RINGp(chan, k, 4);
 			}
 
 			BEGIN_RING(chan, kelvin, TX_GEN_MODE(i, j), 1);
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c b/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c
index df22adf272..62efe80fe4 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c
@@ -139,9 +139,7 @@ nv20_emit_fog(GLcontext *ctx, int emit)
 	OUT_RING(chan, pack_rgba_f(MESA_FORMAT_RGBA8888_REV, f->Color));
 
 	BEGIN_RING(chan, kelvin, NV20TCL_FOG_EQUATION_CONSTANT, 3);
-	OUT_RINGf(chan, k[0]);
-	OUT_RINGf(chan, k[1]);
-	OUT_RINGf(chan, k[2]);
+	OUT_RINGp(chan, k, 3);
 }
 
 void
@@ -176,9 +174,7 @@ nv20_emit_light_source(GLcontext *ctx, int emit)
 
 	if (l->_Flags & LIGHT_POSITIONAL) {
 		BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_POSITION_X(i), 3);
-		OUT_RINGf(chan, l->_Position[0]);
-		OUT_RINGf(chan, l->_Position[1]);
-		OUT_RINGf(chan, l->_Position[2]);
+		OUT_RINGp(chan, l->_Position, 3);
 
 		BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_ATTENUATION_CONSTANT(i), 3);
 		OUT_RINGf(chan, l->ConstantAttenuation);
@@ -187,14 +183,10 @@ nv20_emit_light_source(GLcontext *ctx, int emit)
 
 	} else {
 		BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_DIRECTION_X(i), 3);
-		OUT_RINGf(chan, l->_VP_inf_norm[0]);
-		OUT_RINGf(chan, l->_VP_inf_norm[1]);
-		OUT_RINGf(chan, l->_VP_inf_norm[2]);
+		OUT_RINGp(chan, l->_VP_inf_norm, 3);
 
 		BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_HALF_VECTOR_X(i), 3);
-		OUT_RINGf(chan, l->_h_inf_norm[0]);
-		OUT_RINGf(chan, l->_h_inf_norm[1]);
-		OUT_RINGf(chan, l->_h_inf_norm[2]);
+		OUT_RINGp(chan, l->_h_inf_norm, 3);
 	}
 
 	if (l->_Flags & LIGHT_SPOT) {
@@ -203,13 +195,7 @@ nv20_emit_light_source(GLcontext *ctx, int emit)
 		nv10_get_spot_coeff(l, k);
 
 		BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_SPOT_CUTOFF_A(i), 7);
-		OUT_RINGf(chan, k[0]);
-		OUT_RINGf(chan, k[1]);
-		OUT_RINGf(chan, k[2]);
-		OUT_RINGf(chan, k[3]);
-		OUT_RINGf(chan, k[4]);
-		OUT_RINGf(chan, k[5]);
-		OUT_RINGf(chan, k[6]);
+		OUT_RINGp(chan, k, 7);
 	}
 }
 
@@ -246,15 +232,11 @@ nv20_emit_material_ambient(GLcontext *ctx, int emit)
 	}
 
 	BEGIN_RING(chan, kelvin, m_scene[side], 3);
-	OUT_RINGf(chan, c_scene[0]);
-	OUT_RINGf(chan, c_scene[1]);
-	OUT_RINGf(chan, c_scene[2]);
+	OUT_RINGp(chan, c_scene, 3);
 
 	if (ctx->Light.ColorMaterialEnabled) {
 		BEGIN_RING(chan, kelvin, m_factor[side], 3);
-		OUT_RINGf(chan, c_factor[0]);
-		OUT_RINGf(chan, c_factor[1]);
-		OUT_RINGf(chan, c_factor[2]);
+		OUT_RINGp(chan, c_factor, 3);
 	}
 
 	foreach(l, &ctx->Light.EnabledList) {
@@ -266,9 +248,7 @@ nv20_emit_material_ambient(GLcontext *ctx, int emit)
 				  l->_MatAmbient[side]);
 
 		BEGIN_RING(chan, kelvin, m_light[side], 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -295,9 +275,7 @@ nv20_emit_material_diffuse(GLcontext *ctx, int emit)
 				  l->_MatDiffuse[side]);
 
 		BEGIN_RING(chan, kelvin, m_light[side], 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -318,9 +296,7 @@ nv20_emit_material_specular(GLcontext *ctx, int emit)
 				  l->_MatSpecular[side]);
 
 		BEGIN_RING(chan, kelvin, m_light[side], 3);
-		OUT_RINGf(chan, c_light[0]);
-		OUT_RINGf(chan, c_light[1]);
-		OUT_RINGf(chan, c_light[2]);
+		OUT_RINGp(chan, c_light, 3);
 	}
 }
 
@@ -340,12 +316,7 @@ nv20_emit_material_shininess(GLcontext *ctx, int emit)
 		k);
 
 	BEGIN_RING(chan, kelvin, mthd[side], 6);
-	OUT_RINGf(chan, k[0]);
-	OUT_RINGf(chan, k[1]);
-	OUT_RINGf(chan, k[2]);
-	OUT_RINGf(chan, k[3]);
-	OUT_RINGf(chan, k[4]);
-	OUT_RINGf(chan, k[5]);
+	OUT_RINGp(chan, k, 6);
 }
 
 void
diff --git a/src/mesa/drivers/dri/swrast/Makefile b/src/mesa/drivers/dri/swrast/Makefile
index cc59eefdb2..d2cf6dbc55 100644
--- a/src/mesa/drivers/dri/swrast/Makefile
+++ b/src/mesa/drivers/dri/swrast/Makefile
@@ -5,6 +5,8 @@ include $(TOP)/configs/current
 
 LIBNAME = swrast_dri.so
 
+DRIVER_DEFINES = -D__NOT_HAVE_DRM_H
+
 DRIVER_SOURCES = \
 	swrast.c \
 	swrast_span.c
@@ -18,7 +20,7 @@ ASM_SOURCES =
 SWRAST_COMMON_SOURCES = \
 	../../common/driverfuncs.c \
 	../common/utils.c \
-	../common/dri_sw.c
+	../common/drisw_util.c
 
 include ../Makefile.template
 
diff --git a/src/mesa/drivers/dri/swrast/swrast_priv.h b/src/mesa/drivers/dri/swrast/swrast_priv.h
index 77670d89a5..007642be95 100644
--- a/src/mesa/drivers/dri/swrast/swrast_priv.h
+++ b/src/mesa/drivers/dri/swrast/swrast_priv.h
@@ -30,7 +30,7 @@
 #include <GL/gl.h>
 #include <GL/internal/dri_interface.h>
 #include "main/mtypes.h"
-#include "dri_sw.h"
+#include "drisw_util.h"
 
 
 /**