Merge remote branch 'origin/master' into lp-binning

Conflicts:
	src/gallium/auxiliary/util/u_surface.c
	src/gallium/drivers/llvmpipe/Makefile
	src/gallium/drivers/llvmpipe/SConscript
	src/gallium/drivers/llvmpipe/lp_bld_arit.c
	src/gallium/drivers/llvmpipe/lp_bld_flow.c
	src/gallium/drivers/llvmpipe/lp_bld_interp.c
	src/gallium/drivers/llvmpipe/lp_clear.c
	src/gallium/drivers/llvmpipe/lp_context.c
	src/gallium/drivers/llvmpipe/lp_context.h
	src/gallium/drivers/llvmpipe/lp_draw_arrays.c
	src/gallium/drivers/llvmpipe/lp_jit.c
	src/gallium/drivers/llvmpipe/lp_jit.h
	src/gallium/drivers/llvmpipe/lp_prim_vbuf.c
	src/gallium/drivers/llvmpipe/lp_setup.c
	src/gallium/drivers/llvmpipe/lp_setup_point.c
	src/gallium/drivers/llvmpipe/lp_state.h
	src/gallium/drivers/llvmpipe/lp_state_blend.c
	src/gallium/drivers/llvmpipe/lp_state_derived.c
	src/gallium/drivers/llvmpipe/lp_state_fs.c
	src/gallium/drivers/llvmpipe/lp_state_sampler.c
	src/gallium/drivers/llvmpipe/lp_state_surface.c
	src/gallium/drivers/llvmpipe/lp_tex_cache.c
	src/gallium/drivers/llvmpipe/lp_tex_cache.h
	src/gallium/drivers/llvmpipe/lp_tex_sample.h
	src/gallium/drivers/llvmpipe/lp_tile_cache.c

12 files changed, 548 insertions, 501 deletions

diff --git a/src/gallium/drivers/nv20/nv20_context.c b/src/gallium/drivers/nv20/nv20_context.c
index b32d0d83ba0..5b80af2d22a 100644
--- a/src/gallium/drivers/nv20/nv20_context.c
+++ b/src/gallium/drivers/nv20/nv20_context.c
@@ -10,10 +10,14 @@ nv20_flush(struct pipe_context *pipe, unsigned flags,
 	   struct pipe_fence_handle **fence)
 {
 	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
 
 	draw_flush(nv20->draw);
 
-	FIRE_RING(fence);
+	FIRE_RING(chan);
+	if (fence)
+		*fence = NULL;
 }
 
 static void
@@ -31,377 +35,352 @@ static void nv20_init_hwctx(struct nv20_context *nv20)
 {
 	struct nv20_screen *screen = nv20->screen;
 	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	int i;
 	float projectionmatrix[16];
-	const boolean is_nv25tcl = (nv20->screen->kelvin->grclass == NV25TCL);
+	const boolean is_nv25tcl = (kelvin->grclass == NV25TCL);
 
-	BEGIN_RING(kelvin, NV20TCL_DMA_NOTIFY, 1);
-	OUT_RING  (screen->sync->handle);
-	BEGIN_RING(kelvin, NV20TCL_DMA_TEXTURE0, 2);
-	OUT_RING  (chan->vram->handle);
-	OUT_RING  (chan->gart->handle); /* TEXTURE1 */
-	BEGIN_RING(kelvin, NV20TCL_DMA_COLOR, 2);
-	OUT_RING  (chan->vram->handle);
-	OUT_RING  (chan->vram->handle); /* ZETA */
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_NOTIFY, 1);
+	OUT_RING  (chan, screen->sync->handle);
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_TEXTURE0, 2);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->gart->handle); /* TEXTURE1 */
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_COLOR, 2);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle); /* ZETA */
 
-	BEGIN_RING(kelvin, NV20TCL_DMA_QUERY, 1);
-	OUT_RING  (0); /* renouveau: beef0351, unique */
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_QUERY, 1);
+	OUT_RING  (chan, 0); /* renouveau: beef0351, unique */
 
-	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 2);
-	OUT_RING  (0);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RT_HORIZ, 2);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
 
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 1);
-	OUT_RING  ((0xfff << 16) | 0x0);
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_VERT(0), 1);
-	OUT_RING  ((0xfff << 16) | 0x0);
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 1);
+	OUT_RING  (chan, (0xfff << 16) | 0x0);
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_VERT(0), 1);
+	OUT_RING  (chan, (0xfff << 16) | 0x0);
 
 	for (i = 1; i < NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE; i++) {
-		BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(i), 1);
-		OUT_RING  (0);
-		BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_VERT(i), 1);
-		OUT_RING  (0);
+		BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(i), 1);
+		OUT_RING  (chan, 0);
+		BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_VERT(i), 1);
+		OUT_RING  (chan, 0);
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_MODE, 1);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_MODE, 1);
+	OUT_RING  (chan, 0);
 
-	BEGIN_RING(kelvin, 0x17e0, 3);
-	OUT_RINGf (0.0);
-	OUT_RINGf (0.0);
-	OUT_RINGf (1.0);
+	BEGIN_RING(chan, kelvin, 0x17e0, 3);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 1.0);
 
 	if (is_nv25tcl) {
-		BEGIN_RING(kelvin, NV20TCL_TX_RCOMP, 1);
-		OUT_RING  (NV20TCL_TX_RCOMP_LEQUAL | 0xdb0);
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_RCOMP, 1);
+		OUT_RING  (chan, NV20TCL_TX_RCOMP_LEQUAL | 0xdb0);
 	} else {
-		BEGIN_RING(kelvin, 0x1e68, 1);
-		OUT_RING  (0x4b800000); /* 16777216.000000 */
-		BEGIN_RING(kelvin, NV20TCL_TX_RCOMP, 1);
-		OUT_RING  (NV20TCL_TX_RCOMP_LEQUAL);
+		BEGIN_RING(chan, kelvin, 0x1e68, 1);
+		OUT_RING  (chan, 0x4b800000); /* 16777216.000000 */
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_RCOMP, 1);
+		OUT_RING  (chan, NV20TCL_TX_RCOMP_LEQUAL);
 	}
 
-	BEGIN_RING(kelvin, 0x290, 1);
-	OUT_RING  ((0x10 << 16) | 1);
-	BEGIN_RING(kelvin, 0x9fc, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, 0x1d80, 1);
-	OUT_RING  (1);
-	BEGIN_RING(kelvin, 0x9f8, 1);
-	OUT_RING  (4);
-	BEGIN_RING(kelvin, 0x17ec, 3);
-	OUT_RINGf (0.0);
-	OUT_RINGf (1.0);
-	OUT_RINGf (0.0);
+	BEGIN_RING(chan, kelvin, 0x290, 1);
+	OUT_RING  (chan, (0x10 << 16) | 1);
+	BEGIN_RING(chan, kelvin, 0x9fc, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, 0x1d80, 1);
+	OUT_RING  (chan, 1);
+	BEGIN_RING(chan, kelvin, 0x9f8, 1);
+	OUT_RING  (chan, 4);
+	BEGIN_RING(chan, kelvin, 0x17ec, 3);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 1.0);
+	OUT_RINGf (chan, 0.0);
 
 	if (is_nv25tcl) {
-		BEGIN_RING(kelvin, 0x1d88, 1);
-		OUT_RING  (3);
+		BEGIN_RING(chan, kelvin, 0x1d88, 1);
+		OUT_RING  (chan, 3);
 
-		BEGIN_RING(kelvin, NV25TCL_DMA_IN_MEMORY9, 1);
-		OUT_RING  (chan->vram->handle);
-		BEGIN_RING(kelvin, NV25TCL_DMA_IN_MEMORY8, 1);
-		OUT_RING  (chan->vram->handle);
+		BEGIN_RING(chan, kelvin, NV25TCL_DMA_IN_MEMORY9, 1);
+		OUT_RING  (chan, chan->vram->handle);
+		BEGIN_RING(chan, kelvin, NV25TCL_DMA_IN_MEMORY8, 1);
+		OUT_RING  (chan, chan->vram->handle);
 	}
-	BEGIN_RING(kelvin, NV20TCL_DMA_FENCE, 1);
-	OUT_RING  (0);	/* renouveau: beef1e10 */
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_FENCE, 1);
+	OUT_RING  (chan, 0);	/* renouveau: beef1e10 */
 
-	BEGIN_RING(kelvin, 0x1e98, 1);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, 0x1e98, 1);
+	OUT_RING  (chan, 0);
 #if 0
 	if (is_nv25tcl) {
-		BEGIN_RING(NvSub3D, NV25TCL_DMA_IN_MEMORY4, 2);
-		OUT_RING  (NvDmaTT);	/* renouveau: beef0202 */
-		OUT_RING  (NvDmaFB);	/* renouveau: beef0201 */
+		BEGIN_RING(chan, NvSub3D, NV25TCL_DMA_IN_MEMORY4, 2);
+		OUT_RING  (chan, NvDmaTT);	/* renouveau: beef0202 */
+		OUT_RING  (chan, NvDmaFB);	/* renouveau: beef0201 */
 
-		BEGIN_RING(NvSub3D, NV20TCL_DMA_TEXTURE1, 1);
-		OUT_RING  (NvDmaTT);	/* renouveau: beef0202 */
+		BEGIN_RING(chan, NvSub3D, NV20TCL_DMA_TEXTURE1, 1);
+		OUT_RING  (chan, NvDmaTT);	/* renouveau: beef0202 */
 	}
 #endif
-	BEGIN_RING(kelvin, NV20TCL_NOTIFY, 1);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, NV20TCL_NOTIFY, 1);
+	OUT_RING  (chan, 0);
 
-	BEGIN_RING(kelvin, 0x120, 3);
-	OUT_RING  (0);
-	OUT_RING  (1);
-	OUT_RING  (2);
+	BEGIN_RING(chan, kelvin, 0x120, 3);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 1);
+	OUT_RING  (chan, 2);
 
 /* error: ILLEGAL_MTHD, PROTECTION_FAULT
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_TRANSLATE_X, 4);
-	OUT_RINGf (0.0);
-	OUT_RINGf (512.0);
-	OUT_RINGf (0.0);
-	OUT_RINGf (0.0);
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_TRANSLATE_X, 4);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 512.0);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 0.0);
 */
 
 	if (is_nv25tcl) {
-		BEGIN_RING(kelvin, 0x022c, 2);
-		OUT_RING  (0x280);
-		OUT_RING  (0x07d28000);
+		BEGIN_RING(chan, kelvin, 0x022c, 2);
+		OUT_RING  (chan, 0x280);
+		OUT_RING  (chan, 0x07d28000);
 	}
 
 /* * illegal method, protection fault
-	BEGIN_RING(NvSub3D, 0x1c2c, 1);
-	OUT_RING  (0); */
+	BEGIN_RING(chan, NvSub3D, 0x1c2c, 1);
+	OUT_RING  (chan, 0); */
 
 	if (is_nv25tcl) {
-		BEGIN_RING(kelvin, 0x1da4, 1);
-		OUT_RING  (0);
+		BEGIN_RING(chan, kelvin, 0x1da4, 1);
+		OUT_RING  (chan, 0);
 	}
 
 /* * crashes with illegal method, protection fault
-	BEGIN_RING(NvSub3D, 0x1c18, 1);
-	OUT_RING  (0x200); */
+	BEGIN_RING(chan, NvSub3D, 0x1c18, 1);
+	OUT_RING  (chan, 0x200); */
 
-	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 2);
-	OUT_RING  ((0 << 16) | 0);
-	OUT_RING  ((0 << 16) | 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RT_HORIZ, 2);
+	OUT_RING  (chan, (0 << 16) | 0);
+	OUT_RING  (chan, (0 << 16) | 0);
 
 	/* *** Set state *** */
 
-	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_FUNC, 2);
-	OUT_RING  (NV20TCL_ALPHA_FUNC_FUNC_ALWAYS);
-	OUT_RING  (0);			/* NV20TCL_ALPHA_FUNC_REF */
+	BEGIN_RING(chan, kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_ALPHA_FUNC_FUNC, 2);
+	OUT_RING  (chan, NV20TCL_ALPHA_FUNC_FUNC_ALWAYS);
+	OUT_RING  (chan, 0);			/* NV20TCL_ALPHA_FUNC_REF */
 
 	for (i = 0; i < NV20TCL_TX_ENABLE__SIZE; ++i) {
-		BEGIN_RING(kelvin, NV20TCL_TX_ENABLE(i), 1);
-		OUT_RING  (0);
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_ENABLE(i), 1);
+		OUT_RING  (chan, 0);
 	}
-	BEGIN_RING(kelvin, NV20TCL_TX_SHADER_OP, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_TX_SHADER_CULL_MODE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_IN_ALPHA(0), 4);
-	OUT_RING  (0x30d410d0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_OUT_RGB(0), 4);
-	OUT_RING  (0x00000c00);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_ENABLE, 1);
-	OUT_RING  (0x00011101);
-	BEGIN_RING(kelvin, NV20TCL_RC_FINAL0, 2);
-	OUT_RING  (0x130e0300);
-	OUT_RING  (0x0c091c80);
-	BEGIN_RING(kelvin, NV20TCL_RC_OUT_ALPHA(0), 4);
-	OUT_RING  (0x00000c00);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_IN_RGB(0), 4);
-	OUT_RING  (0x20c400c0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_COLOR0, 2);
-	OUT_RING  (0);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_RC_CONSTANT_COLOR0(0), 4);
-	OUT_RING  (0x035125a0);
-	OUT_RING  (0);
-	OUT_RING  (0x40002000);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_MULTISAMPLE_CONTROL, 1);
-	OUT_RING  (0xffff0000);
-
-	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_DITHER_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_STENCIL_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_SRC, 4);
-	OUT_RING  (NV20TCL_BLEND_FUNC_SRC_ONE);
-	OUT_RING  (NV20TCL_BLEND_FUNC_DST_ZERO);
-	OUT_RING  (0);			/* NV20TCL_BLEND_COLOR */
-	OUT_RING  (NV20TCL_BLEND_EQUATION_FUNC_ADD);
-	BEGIN_RING(kelvin, NV20TCL_STENCIL_MASK, 7);
-	OUT_RING  (0xff);
-	OUT_RING  (NV20TCL_STENCIL_FUNC_FUNC_ALWAYS);
-	OUT_RING  (0);			/* NV20TCL_STENCIL_FUNC_REF */
-	OUT_RING  (0xff);		/* NV20TCL_STENCIL_FUNC_MASK */
-	OUT_RING  (NV20TCL_STENCIL_OP_FAIL_KEEP);
-	OUT_RING  (NV20TCL_STENCIL_OP_ZFAIL_KEEP);
-	OUT_RING  (NV20TCL_STENCIL_OP_ZPASS_KEEP);
-
-	BEGIN_RING(kelvin, NV20TCL_COLOR_LOGIC_OP_ENABLE, 2);
-	OUT_RING  (0);
-	OUT_RING  (NV20TCL_COLOR_LOGIC_OP_OP_COPY);
-	BEGIN_RING(kelvin, 0x17cc, 1);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, NV20TCL_TX_SHADER_OP, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_TX_SHADER_CULL_MODE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_IN_ALPHA(0), 4);
+	OUT_RING  (chan, 0x30d410d0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_OUT_RGB(0), 4);
+	OUT_RING  (chan, 0x00000c00);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_ENABLE, 1);
+	OUT_RING  (chan, 0x00011101);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_FINAL0, 2);
+	OUT_RING  (chan, 0x130e0300);
+	OUT_RING  (chan, 0x0c091c80);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_OUT_ALPHA(0), 4);
+	OUT_RING  (chan, 0x00000c00);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_IN_RGB(0), 4);
+	OUT_RING  (chan, 0x20c400c0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_COLOR0, 2);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RC_CONSTANT_COLOR0(0), 4);
+	OUT_RING  (chan, 0x035125a0);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0x40002000);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_MULTISAMPLE_CONTROL, 1);
+	OUT_RING  (chan, 0xffff0000);
+
+	BEGIN_RING(chan, kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_DITHER_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_STENCIL_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_BLEND_FUNC_SRC, 4);
+	OUT_RING  (chan, NV20TCL_BLEND_FUNC_SRC_ONE);
+	OUT_RING  (chan, NV20TCL_BLEND_FUNC_DST_ZERO);
+	OUT_RING  (chan, 0);			/* NV20TCL_BLEND_COLOR */
+	OUT_RING  (chan, NV20TCL_BLEND_EQUATION_FUNC_ADD);
+	BEGIN_RING(chan, kelvin, NV20TCL_STENCIL_MASK, 7);
+	OUT_RING  (chan, 0xff);
+	OUT_RING  (chan, NV20TCL_STENCIL_FUNC_FUNC_ALWAYS);
+	OUT_RING  (chan, 0);			/* NV20TCL_STENCIL_FUNC_REF */
+	OUT_RING  (chan, 0xff);		/* NV20TCL_STENCIL_FUNC_MASK */
+	OUT_RING  (chan, NV20TCL_STENCIL_OP_FAIL_KEEP);
+	OUT_RING  (chan, NV20TCL_STENCIL_OP_ZFAIL_KEEP);
+	OUT_RING  (chan, NV20TCL_STENCIL_OP_ZPASS_KEEP);
+
+	BEGIN_RING(chan, kelvin, NV20TCL_COLOR_LOGIC_OP_ENABLE, 2);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, NV20TCL_COLOR_LOGIC_OP_OP_COPY);
+	BEGIN_RING(chan, kelvin, 0x17cc, 1);
+	OUT_RING  (chan, 0);
 	if (is_nv25tcl) {
-		BEGIN_RING(kelvin, 0x1d84, 1);
-		OUT_RING  (1);
+		BEGIN_RING(chan, kelvin, 0x1d84, 1);
+		OUT_RING  (chan, 1);
 	}
-	BEGIN_RING(kelvin, NV20TCL_LIGHTING_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_LIGHT_CONTROL, 1);
-	OUT_RING  (0x00020000);
-	BEGIN_RING(kelvin, NV20TCL_SEPARATE_SPECULAR_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_ENABLED_LIGHTS, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_NORMALIZE_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_STIPPLE_PATTERN(0),
+	BEGIN_RING(chan, kelvin, NV20TCL_LIGHTING_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_CONTROL, 1);
+	OUT_RING  (chan, 0x00020000);
+	BEGIN_RING(chan, kelvin, NV20TCL_SEPARATE_SPECULAR_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_ENABLED_LIGHTS, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_NORMALIZE_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_STIPPLE_PATTERN(0),
 					NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE);
 	for (i = 0; i < NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE; ++i) {
-		OUT_RING(0xffffffff);
+		OUT_RING(chan, 0xffffffff);
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
-	OUT_RING  (0);
-	OUT_RING  (0);		/* NV20TCL.POLYGON_OFFSET_LINE_ENABLE */
-	OUT_RING  (0);		/* NV20TCL.POLYGON_OFFSET_FILL_ENABLE */
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_FUNC, 1);
-	OUT_RING  (NV20TCL_DEPTH_FUNC_LESS);
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_OFFSET_FACTOR, 2);
-	OUT_RINGf (0.0);
-	OUT_RINGf (0.0);	/* NV20TCL.POLYGON_OFFSET_UNITS */
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_UNK17D8, 1);
-	OUT_RING  (1);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);		/* NV20TCL.POLYGON_OFFSET_LINE_ENABLE */
+	OUT_RING  (chan, 0);		/* NV20TCL.POLYGON_OFFSET_FILL_ENABLE */
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_FUNC, 1);
+	OUT_RING  (chan, NV20TCL_DEPTH_FUNC_LESS);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_OFFSET_FACTOR, 2);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 0.0);	/* NV20TCL.POLYGON_OFFSET_UNITS */
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_UNK17D8, 1);
+	OUT_RING  (chan, 1);
 	if (!is_nv25tcl) {
-		BEGIN_RING(kelvin, 0x1d84, 1);
-		OUT_RING  (3);
+		BEGIN_RING(chan, kelvin, 0x1d84, 1);
+		OUT_RING  (chan, 3);
 	}
-	BEGIN_RING(kelvin, NV20TCL_POINT_SIZE, 1);
+	BEGIN_RING(chan, kelvin, NV20TCL_POINT_SIZE, 1);
 	if (!is_nv25tcl) {
-		OUT_RING  (8);
+		OUT_RING  (chan, 8);
 	} else {
-		OUT_RINGf (1.0);
+		OUT_RINGf (chan, 1.0);
 	}
 	if (!is_nv25tcl) {
-		BEGIN_RING(kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 2);
-		OUT_RING  (0);
-		OUT_RING  (0);		/* NV20TCL.POINT_SMOOTH_ENABLE */
+		BEGIN_RING(chan, kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 2);
+		OUT_RING  (chan, 0);
+		OUT_RING  (chan, 0);		/* NV20TCL.POINT_SMOOTH_ENABLE */
 	} else {
-		BEGIN_RING(kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 1);
-		OUT_RING  (0);
-		BEGIN_RING(kelvin, 0x0a1c, 1);
-		OUT_RING  (0x800);
+		BEGIN_RING(chan, kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 1);
+		OUT_RING  (chan, 0);
+		BEGIN_RING(chan, kelvin, 0x0a1c, 1);
+		OUT_RING  (chan, 0x800);
 	}
-	BEGIN_RING(kelvin, NV20TCL_LINE_WIDTH, 1);
-	OUT_RING  (8);
-	BEGIN_RING(kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
-	OUT_RING  (NV20TCL_POLYGON_MODE_FRONT_FILL);
-	OUT_RING  (NV20TCL_POLYGON_MODE_BACK_FILL);
-	BEGIN_RING(kelvin, NV20TCL_CULL_FACE, 2);
-	OUT_RING  (NV20TCL_CULL_FACE_BACK);
-	OUT_RING  (NV20TCL_FRONT_FACE_CCW);
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_SMOOTH_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_SHADE_MODEL, 1);
-	OUT_RING  (NV20TCL_SHADE_MODEL_SMOOTH);
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_STIPPLE_ENABLE, 1);
-	OUT_RING  (0);
-	BEGIN_RING(kelvin, NV20TCL_TX_GEN_S(0), 4 * NV20TCL_TX_GEN_S__SIZE);
+	BEGIN_RING(chan, kelvin, NV20TCL_LINE_WIDTH, 1);
+	OUT_RING  (chan, 8);
+	BEGIN_RING(chan, kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (chan, NV20TCL_POLYGON_MODE_FRONT_FILL);
+	OUT_RING  (chan, NV20TCL_POLYGON_MODE_BACK_FILL);
+	BEGIN_RING(chan, kelvin, NV20TCL_CULL_FACE, 2);
+	OUT_RING  (chan, NV20TCL_CULL_FACE_BACK);
+	OUT_RING  (chan, NV20TCL_FRONT_FACE_CCW);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_SMOOTH_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_SHADE_MODEL, 1);
+	OUT_RING  (chan, NV20TCL_SHADE_MODEL_SMOOTH);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_STIPPLE_ENABLE, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_TX_GEN_S(0), 4 * NV20TCL_TX_GEN_S__SIZE);
 	for (i=0; i < 4 * NV20TCL_TX_GEN_S__SIZE; ++i) {
-		OUT_RING(0);
+		OUT_RING(chan, 0);
 	}
-	BEGIN_RING(kelvin, NV20TCL_FOG_EQUATION_CONSTANT, 3);
-	OUT_RINGf (1.5);
-	OUT_RINGf (-0.090168);		/* NV20TCL.FOG_EQUATION_LINEAR */
-	OUT_RINGf (0.0);		/* NV20TCL.FOG_EQUATION_QUADRATIC */
-	BEGIN_RING(kelvin, NV20TCL_FOG_MODE, 2);
-	OUT_RING  (NV20TCL_FOG_MODE_EXP_2);
-	OUT_RING  (NV20TCL_FOG_COORD_DIST_COORD_FOG);
-	BEGIN_RING(kelvin, NV20TCL_FOG_ENABLE, 2);
-	OUT_RING  (0);
-	OUT_RING  (0);			/* NV20TCL.FOG_COLOR */
-	BEGIN_RING(kelvin, NV20TCL_ENGINE, 1);
-	OUT_RING  (NV20TCL_ENGINE_FIXED);
+	BEGIN_RING(chan, kelvin, NV20TCL_FOG_EQUATION_CONSTANT, 3);
+	OUT_RINGf (chan, 1.5);
+	OUT_RINGf (chan, -0.090168);		/* NV20TCL.FOG_EQUATION_LINEAR */
+	OUT_RINGf (chan, 0.0);		/* NV20TCL.FOG_EQUATION_QUADRATIC */
+	BEGIN_RING(chan, kelvin, NV20TCL_FOG_MODE, 2);
+	OUT_RING  (chan, NV20TCL_FOG_MODE_EXP_SIGNED);
+	OUT_RING  (chan, NV20TCL_FOG_COORD_FOG);
+	BEGIN_RING(chan, kelvin, NV20TCL_FOG_ENABLE, 2);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 0);			/* NV20TCL.FOG_COLOR */
+	BEGIN_RING(chan, kelvin, NV20TCL_ENGINE, 1);
+	OUT_RING  (chan, NV20TCL_ENGINE_FIXED);
 
 	for (i = 0; i < NV20TCL_TX_MATRIX_ENABLE__SIZE; ++i) {
-		BEGIN_RING(kelvin, NV20TCL_TX_MATRIX_ENABLE(i), 1);
-		OUT_RING  (0);
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_MATRIX_ENABLE(i), 1);
+		OUT_RING  (chan, 0);
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_VTX_ATTR_4F_X(1), 4 * 15);
-	OUT_RINGf(1.0); OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(1.0);
-	OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(1.0); OUT_RINGf(1.0);
-	OUT_RINGf(1.0); OUT_RINGf(1.0); OUT_RINGf(1.0); OUT_RINGf(1.0);
+	BEGIN_RING(chan, kelvin, NV20TCL_VTX_ATTR_4F_X(1), 4 * 15);
+	OUT_RINGf(chan, 1.0); OUT_RINGf(chan, 0.0); OUT_RINGf(chan, 0.0); OUT_RINGf(chan, 1.0);
+	OUT_RINGf(chan, 0.0); OUT_RINGf(chan, 0.0); OUT_RINGf(chan, 1.0); OUT_RINGf(chan, 1.0);
+	OUT_RINGf(chan, 1.0); OUT_RINGf(chan, 1.0); OUT_RINGf(chan, 1.0); OUT_RINGf(chan, 1.0);
 	for (i = 4; i < 16; ++i) {
-		OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(0.0);	OUT_RINGf(1.0);
+		OUT_RINGf(chan, 0.0);
+		OUT_RINGf(chan, 0.0);
+		OUT_RINGf(chan, 0.0);
+		OUT_RINGf(chan, 1.0);
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_EDGEFLAG_ENABLE, 1);
-	OUT_RING  (1);
-	BEGIN_RING(kelvin, NV20TCL_COLOR_MASK, 1);
-	OUT_RING (0x00010101);
-	BEGIN_RING(kelvin, NV20TCL_CLEAR_VALUE, 1);
-	OUT_RING (0);
+	BEGIN_RING(chan, kelvin, NV20TCL_EDGEFLAG_ENABLE, 1);
+	OUT_RING  (chan, 1);
+	BEGIN_RING(chan, kelvin, NV20TCL_COLOR_MASK, 1);
+	OUT_RING (chan, 0x00010101);
+	BEGIN_RING(chan, kelvin, NV20TCL_CLEAR_VALUE, 1);
+	OUT_RING (chan, 0);
 
 	memset(projectionmatrix, 0, sizeof(projectionmatrix));
 	projectionmatrix[0*4+0] = 1.0;
 	projectionmatrix[1*4+1] = 1.0;
 	projectionmatrix[2*4+2] = 16777215.0;
 	projectionmatrix[3*4+3] = 1.0;
-	BEGIN_RING(kelvin, NV20TCL_PROJECTION_MATRIX(0), 16);
+	BEGIN_RING(chan, kelvin, NV20TCL_PROJECTION_MATRIX(0), 16);
 	for (i = 0; i < 16; i++) {
-		OUT_RINGf  (projectionmatrix[i]);
+		OUT_RINGf  (chan, projectionmatrix[i]);
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_RANGE_NEAR, 2);
-	OUT_RINGf (0.0);
-	OUT_RINGf (16777216.0); /* [0, 1] scaled approx to [0, 2^24] */
-
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_SCALE0_X, 4);
-	OUT_RINGf (0.0); /* x-offset, w/2 + 1.031250 */
-	OUT_RINGf (0.0); /* y-offset, h/2 + 0.030762 */
-	OUT_RINGf (0.0);
-	OUT_RINGf (16777215.0);
-
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_SCALE1_X, 4);
-	OUT_RINGf (0.0); /* no effect?, w/2 */
-	OUT_RINGf (0.0); /* no effect?, h/2 */
-	OUT_RINGf (16777215.0 * 0.5);
-	OUT_RINGf (65535.0);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_RANGE_NEAR, 2);
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 16777216.0); /* [0, 1] scaled approx to [0, 2^24] */
 
-	FIRE_RING (NULL);
-}
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_TRANSLATE_X, 4);
+	OUT_RINGf (chan, 0.0); /* x-offset, w/2 + 1.031250 */
+	OUT_RINGf (chan, 0.0); /* y-offset, h/2 + 0.030762 */
+	OUT_RINGf (chan, 0.0);
+	OUT_RINGf (chan, 16777215.0);
 
-static void
-nv20_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
-{
-}
-
-
-static unsigned int
-nv20_is_texture_referenced( struct pipe_context *pipe,
-			    struct pipe_texture *texture,
-			    unsigned face, unsigned level)
-{
-   /**
-    * FIXME: Optimize.
-    */
-
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
-}
-
-static unsigned int
-nv20_is_buffer_referenced( struct pipe_context *pipe,
-			   struct pipe_buffer *buf)
-{
-   /**
-    * FIXME: Optimize.
-    */
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_SCALE_X, 4);
+	OUT_RINGf (chan, 0.0); /* no effect?, w/2 */
+	OUT_RINGf (chan, 0.0); /* no effect?, h/2 */
+	OUT_RINGf (chan, 16777215.0 * 0.5);
+	OUT_RINGf (chan, 65535.0);
 
-   return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
+	FIRE_RING (chan);
 }
 
 struct pipe_context *
@@ -423,14 +402,13 @@ nv20_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv20->pipe.winsys = ws;
 	nv20->pipe.screen = pscreen;
 	nv20->pipe.destroy = nv20_destroy;
-	nv20->pipe.set_edgeflags = nv20_set_edgeflags;
 	nv20->pipe.draw_arrays = nv20_draw_arrays;
 	nv20->pipe.draw_elements = nv20_draw_elements;
 	nv20->pipe.clear = nv20_clear;
 	nv20->pipe.flush = nv20_flush;
 
-	nv20->pipe.is_texture_referenced = nv20_is_texture_referenced;
-	nv20->pipe.is_buffer_referenced = nv20_is_buffer_referenced;
+	nv20->pipe.is_texture_referenced = nouveau_is_texture_referenced;
+	nv20->pipe.is_buffer_referenced = nouveau_is_buffer_referenced;
 
 	nv20_init_surface_functions(nv20);
 	nv20_init_state_functions(nv20);
diff --git a/src/gallium/drivers/nv20/nv20_context.h b/src/gallium/drivers/nv20/nv20_context.h
index fc932f1f90e..c7dfadaa311 100644
--- a/src/gallium/drivers/nv20/nv20_context.h
+++ b/src/gallium/drivers/nv20/nv20_context.h
@@ -13,10 +13,7 @@
 
 #include "nouveau/nouveau_winsys.h"
 #include "nouveau/nouveau_gldefs.h"
-
-#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
-	struct nv20_screen *ctx = nv20->screen
-#include "nouveau/nouveau_push.h"
+#include "nouveau/nouveau_context.h"
 
 #include "nv20_state.h"
 
@@ -142,9 +139,9 @@ extern void nv20_emit_hw_state(struct nv20_context *nv20);
 extern void nv20_state_tex_update(struct nv20_context *nv20);
 
 /* nv20_vbo.c */
-extern boolean nv20_draw_arrays(struct pipe_context *, unsigned mode,
+extern void nv20_draw_arrays(struct pipe_context *, unsigned mode,
 				unsigned start, unsigned count);
-extern boolean nv20_draw_elements( struct pipe_context *pipe,
+extern void nv20_draw_elements( struct pipe_context *pipe,
                     struct pipe_buffer *indexBuffer,
                     unsigned indexSize,
                     unsigned prim, unsigned start, unsigned count);
diff --git a/src/gallium/drivers/nv20/nv20_fragtex.c b/src/gallium/drivers/nv20/nv20_fragtex.c
index 495a7be9127..dedbec73f39 100644
--- a/src/gallium/drivers/nv20/nv20_fragtex.c
+++ b/src/gallium/drivers/nv20/nv20_fragtex.c
@@ -52,6 +52,9 @@ nv20_fragtex_build(struct nv20_context *nv20, int unit)
 	struct nv20_miptree *nv20mt = nv20->tex_miptree[unit];
 	struct pipe_texture *pt = &nv20mt->base;
 	struct nv20_texture_format *tf;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	uint32_t txf, txs, txp;
 
 	tf = nv20_fragtex_format(pt->format);
@@ -62,9 +65,9 @@ nv20_fragtex_build(struct nv20_context *nv20, int unit)
 
 	txf  = tf->format << 8;
 	txf |= (pt->last_level + 1) << 16;
-	txf |= log2i(pt->width[0]) << 20;
-	txf |= log2i(pt->height[0]) << 24;
-	txf |= log2i(pt->depth[0]) << 28;
+	txf |= log2i(pt->width0) << 20;
+	txf |= log2i(pt->height0) << 24;
+	txf |= log2i(pt->depth0) << 28;
 	txf |= 8;
 
 	switch (pt->target) {
@@ -82,15 +85,15 @@ nv20_fragtex_build(struct nv20_context *nv20, int unit)
 		return;
 	}
 
-	BEGIN_RING(kelvin, NV10TCL_TX_OFFSET(unit), 8);
-	OUT_RELOCl(nv20mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
-	OUT_RELOCd(nv20mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
-	OUT_RING  (ps->wrap);
-	OUT_RING  (0x40000000); /* enable */
-	OUT_RING  (txs);
-	OUT_RING  (ps->filt | 0x2000 /* magic */);
-	OUT_RING  ((pt->width[0] << 16) | pt->height[0]);
-	OUT_RING  (ps->bcol);
+	BEGIN_RING(chan, kelvin, NV10TCL_TX_OFFSET(unit), 8);
+	OUT_RELOCl(chan, nouveau_bo(nv20mt->buffer), 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RELOCd(chan, nouveau_bo(nv20mt->buffer),txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
+	OUT_RING  (chan, ps->wrap);
+	OUT_RING  (chan, 0x40000000); /* enable */
+	OUT_RING  (chan, txs);
+	OUT_RING  (chan, ps->filt | 0x2000 /* magic */);
+	OUT_RING  (chan, (pt->width0 << 16) | pt->height0);
+	OUT_RING  (chan, ps->bcol);
 #endif
 }
 
@@ -99,6 +102,9 @@ nv20_fragtex_bind(struct nv20_context *nv20)
 {
 #if 0
 	struct nv20_fragment_program *fp = nv20->fragprog.active;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	unsigned samplers, unit;
 
 	samplers = nv20->fp_samplers & ~fp->samplers;
@@ -106,8 +112,8 @@ nv20_fragtex_bind(struct nv20_context *nv20)
 		unit = ffs(samplers) - 1;
 		samplers &= ~(1 << unit);
 
-		BEGIN_RING(kelvin, NV10TCL_TX_ENABLE(unit), 1);
-		OUT_RING  (0);
+		BEGIN_RING(chan, kelvin, NV10TCL_TX_ENABLE(unit), 1);
+		OUT_RING  (chan, 0);
 	}
 
 	samplers = nv20->dirty_samplers & fp->samplers;
diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c
index 185fbf53e0f..8f7538e7f57 100644
--- a/src/gallium/drivers/nv20/nv20_miptree.c
+++ b/src/gallium/drivers/nv20/nv20_miptree.c
@@ -1,15 +1,18 @@
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
+#include "util/u_format.h"
+#include "util/u_math.h"
 
 #include "nv20_context.h"
 #include "nv20_screen.h"
+#include "../nv04/nv04_surface_2d.h"
 
 static void
 nv20_miptree_layout(struct nv20_miptree *nv20mt)
 {
 	struct pipe_texture *pt = &nv20mt->base;
-	uint width = pt->width[0], height = pt->height[0];
+	uint width = pt->width0;
 	uint offset = 0;
 	int nr_faces, l, f;
 	uint wide_pitch = pt->tex_usage & (PIPE_TEXTURE_USAGE_SAMPLER |
@@ -25,21 +28,15 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt)
 	}
 	
 	for (l = 0; l <= pt->last_level; l++) {
-		pt->width[l] = width;
-		pt->height[l] = height;
-		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
-		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
-
 		if (wide_pitch && (pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
-			nv20mt->level[l].pitch = align(pt->width[0] * pt->block.size, 64);
+			nv20mt->level[l].pitch = align(util_format_get_stride(pt->format, pt->width0), 64);
 		else
-			nv20mt->level[l].pitch = pt->width[l] * pt->block.size;
+			nv20mt->level[l].pitch = util_format_get_stride(pt->format, width);
 
 		nv20mt->level[l].image_offset =
 			CALLOC(nr_faces, sizeof(unsigned));
 
-		width  = MAX2(1, width  >> 1);
-		height = MAX2(1, height >> 1);
+		width  = u_minify(width, 1);
 	}
 
 	for (f = 0; f < nr_faces; f++) {
@@ -47,14 +44,14 @@ nv20_miptree_layout(struct nv20_miptree *nv20mt)
 			nv20mt->level[l].image_offset[f] = offset;
 
 			if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR) &&
-			    pt->width[l + 1] > 1 && pt->height[l + 1] > 1)
-				offset += align(nv20mt->level[l].pitch * pt->height[l], 64);
+			    u_minify(pt->width0, l + 1) > 1 && u_minify(pt->height0, l + 1) > 1)
+				offset += align(nv20mt->level[l].pitch * u_minify(pt->height0, l), 64);
 			else
-				offset += nv20mt->level[l].pitch * pt->height[l];
+				offset += nv20mt->level[l].pitch * u_minify(pt->height0, l);
 		}
 
 		nv20mt->level[l].image_offset[f] = offset;
-		offset += nv20mt->level[l].pitch * pt->height[l];
+		offset += nv20mt->level[l].pitch * u_minify(pt->height0, l);
 	}
 
 	nv20mt->total_size = offset;
@@ -68,7 +65,7 @@ nv20_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 
 	/* Only supports 2D, non-mipmapped textures for the moment */
 	if (pt->target != PIPE_TEXTURE_2D || pt->last_level != 0 ||
-	    pt->depth[0] != 1)
+	    pt->depth0 != 1)
 		return NULL;
 
 	mt = CALLOC_STRUCT(nv20_miptree);
@@ -82,6 +79,7 @@ nv20_miptree_blanket(struct pipe_screen *pscreen, const struct pipe_texture *pt,
 	mt->level[0].image_offset = CALLOC(1, sizeof(unsigned));
 
 	pipe_buffer_reference(&mt->buffer, pb);
+	mt->bo = nouveau_bo(mt->buffer);
 	return &mt->base;
 }
 
@@ -100,8 +98,8 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 	mt->base.screen = screen;
 
 	/* Swizzled textures must be POT */
-	if (pt->width[0] & (pt->width[0] - 1) ||
-	    pt->height[0] & (pt->height[0] - 1))
+	if (pt->width0 & (pt->width0 - 1) ||
+	    pt->height0 & (pt->height0 - 1))
 		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
 	else
 	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
@@ -130,6 +128,12 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 	if (pt->tex_usage & PIPE_TEXTURE_USAGE_DYNAMIC)
 		buf_usage |= PIPE_BUFFER_USAGE_CPU_READ_WRITE;
 
+	/* apparently we can't render to swizzled surfaces smaller than 64 bytes, so make them linear.
+	 * If the user did not ask for a render target, they can still render to it, but it will cost them an extra copy.
+	 * This also happens for small mipmaps of large textures. */
+	if (pt->tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET && util_format_get_stride(pt->format, pt->width0) < 64)
+		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+
 	nv20_miptree_layout(mt);
 
 	mt->buffer = screen->buffer_create(screen, 256, buf_usage, mt->total_size);
@@ -137,6 +141,7 @@ nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
 		FREE(mt);
 		return NULL;
 	}
+	mt->bo = nouveau_bo(mt->buffer);
 	
 	return &mt->base;
 }
@@ -167,8 +172,8 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 		return NULL;
 	pipe_texture_reference(&ns->base.texture, pt);
 	ns->base.format = pt->format;
-	ns->base.width = pt->width[level];
-	ns->base.height = pt->height[level];
+	ns->base.width = u_minify(pt->width0, level);
+	ns->base.height = u_minify(pt->height0, level);
 	ns->base.usage = flags;
 	pipe_reference_init(&ns->base.reference, 1);
 	ns->base.face = face;
@@ -185,12 +190,27 @@ nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
 		ns->base.offset = nv20mt->level[level].image_offset[0];
 	}
 
+	/* create a linear temporary that we can render into if necessary.
+	 * Note that ns->pitch is always a multiple of 64 for linear surfaces and swizzled surfaces are POT, so
+	 * ns->pitch & 63 is equivalent to (ns->pitch < 64 && swizzled)*/
+	if((ns->pitch & 63) && (ns->base.usage & (PIPE_BUFFER_USAGE_GPU_WRITE | NOUVEAU_BUFFER_USAGE_NO_RENDER)) == PIPE_BUFFER_USAGE_GPU_WRITE)
+		return &nv04_surface_wrap_for_render(screen, ((struct nv20_screen*)screen)->eng2d, ns)->base;
+
 	return &ns->base;
 }
 
 static void
 nv20_miptree_surface_destroy(struct pipe_surface *ps)
 {
+	struct nv04_surface* ns = (struct nv04_surface*)ps;
+	if(ns->backing)
+	{
+		struct nv20_screen* screen = (struct nv20_screen*)ps->texture->screen;
+		if(ns->backing->base.usage & PIPE_BUFFER_USAGE_GPU_WRITE)
+			screen->eng2d->copy(screen->eng2d, &ns->backing->base, 0, 0, ps, 0, 0, ns->base.width, ns->base.height);
+		nv20_miptree_surface_destroy(&ns->backing->base);
+	}
+	
 	pipe_texture_reference(&ps->texture, NULL);
 	FREE(ps);
 }
diff --git a/src/gallium/drivers/nv20/nv20_prim_vbuf.c b/src/gallium/drivers/nv20/nv20_prim_vbuf.c
index ddfcdb8057a..2e145672da1 100644
--- a/src/gallium/drivers/nv20/nv20_prim_vbuf.c
+++ b/src/gallium/drivers/nv20/nv20_prim_vbuf.c
@@ -81,12 +81,15 @@ nv20_vbuf_render(struct vbuf_render *render)
 void nv20_vtxbuf_bind( struct nv20_context* nv20 )
 {
 #if 0
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	int i;
 	for(i = 0; i < NV20TCL_VTXBUF_ADDRESS__SIZE; i++) {
-		BEGIN_RING(kelvin, NV20TCL_VTXBUF_ADDRESS(i), 1);
-		OUT_RING(0/*nv20->vtxbuf*/);
-		BEGIN_RING(kelvin, NV20TCL_VTXFMT(i) ,1);
-		OUT_RING(0/*XXX*/);
+		BEGIN_RING(chan, kelvin, NV20TCL_VTXBUF_ADDRESS(i), 1);
+		OUT_RING(chan, 0/*nv20->vtxbuf*/);
+		BEGIN_RING(chan, kelvin, NV20TCL_VTXFMT(i) ,1);
+		OUT_RING(chan, 0/*XXX*/);
 	}
 #endif
 }
@@ -202,6 +205,9 @@ nv20__vtxhwformat(unsigned stride, unsigned fields, unsigned type)
 static unsigned
 nv20__emit_format(struct nv20_context *nv20, enum attrib_emit type, int hwattr)
 {
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	uint32_t hwfmt = 0;
 	unsigned fields;
 
@@ -231,8 +237,8 @@ nv20__emit_format(struct nv20_context *nv20, enum attrib_emit type, int hwattr)
 		return 0;
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_VTXFMT(hwattr), 1);
-	OUT_RING(hwfmt);
+	BEGIN_RING(chan, kelvin, NV20TCL_VTXFMT(hwattr), 1);
+	OUT_RING(chan, hwfmt);
 	return fields;
 }
 
@@ -262,6 +268,9 @@ nv20__draw_mbuffer(struct nv20_vbuf_render *nv20_render,
 		uint nr_indices)
 {
 	struct nv20_context *nv20 = nv20_render->nv20;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	struct vertex_info *vinfo = &nv20->vertex_info;
 	unsigned nr_fields;
 	int max_push;
@@ -270,29 +279,29 @@ nv20__draw_mbuffer(struct nv20_vbuf_render *nv20_render,
 
 	nr_fields = nv20__emit_vertex_array_format(nv20);
 
-	BEGIN_RING(kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
-	OUT_RING(nv20_render->hwprim);
+	BEGIN_RING(chan, kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
+	OUT_RING(chan, nv20_render->hwprim);
 
 	max_push = 1200 / nr_fields;
 	while (nr_indices) {
 		int i;
 		int push = MIN2(nr_indices, max_push);
 
-		BEGIN_RING_NI(kelvin, NV20TCL_VERTEX_DATA, push * nr_fields);
+		BEGIN_RING_NI(chan, kelvin, NV20TCL_VERTEX_DATA, push * nr_fields);
 		for (i = 0; i < push; i++) {
 			/* XXX: fixme to handle other than floats? */
 			int f = nr_fields;
 			float *attrv = (float*)&data[indices[i] * vsz];
 			while (f-- > 0)
-				OUT_RINGf(*attrv++);
+				OUT_RINGf(chan, *attrv++);
 		}
 
 		nr_indices -= push;
 		indices += push;
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
-	OUT_RING(NV20TCL_VERTEX_BEGIN_END_STOP);
+	BEGIN_RING(chan, kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
+	OUT_RING(chan, NV20TCL_VERTEX_BEGIN_END_STOP);
 }
 
 static void
@@ -301,20 +310,23 @@ nv20__draw_pbuffer(struct nv20_vbuf_render *nv20_render,
 		uint nr_indices)
 {
 	struct nv20_context *nv20 = nv20_render->nv20;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 	int push, i;
 
 	NOUVEAU_ERR("nv20__draw_pbuffer: this path is broken.\n");
 
-	BEGIN_RING(kelvin, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1);
-	OUT_RELOCl(nv20_render->pbuffer, 0,
+	BEGIN_RING(chan, kelvin, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1);
+	OUT_RELOCl(chan, nouveau_bo(nv20_render->pbuffer), 0,
 			NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
 
-	BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
-	OUT_RING(nv20_render->hwprim);
+	BEGIN_RING(chan, kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING(chan, nv20_render->hwprim);
 
 	if (nr_indices & 1) {
-		BEGIN_RING(kelvin, NV10TCL_VB_ELEMENT_U32, 1);
-		OUT_RING  (indices[0]);
+		BEGIN_RING(chan, kelvin, NV10TCL_VB_ELEMENT_U32, 1);
+		OUT_RING  (chan, indices[0]);
 		indices++; nr_indices--;
 	}
 
@@ -322,16 +334,16 @@ nv20__draw_pbuffer(struct nv20_vbuf_render *nv20_render,
 		// XXX too big/small ? check the size
 		push = MIN2(nr_indices, 1200 * 2);
 
-		BEGIN_RING_NI(kelvin, NV10TCL_VB_ELEMENT_U16, push >> 1);
+		BEGIN_RING_NI(chan, kelvin, NV10TCL_VB_ELEMENT_U16, push >> 1);
 		for (i = 0; i < push; i+=2)
-			OUT_RING((indices[i+1] << 16) | indices[i]);
+			OUT_RING(chan, (indices[i+1] << 16) | indices[i]);
 
 		nr_indices -= push;
 		indices  += push;
 	}
 
-	BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
-	OUT_RING  (0);
+	BEGIN_RING(chan, kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING  (chan, 0);
 }
 
 static void
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
index 4eeacd1afd5..d091335063b 100644
--- a/src/gallium/drivers/nv20/nv20_screen.c
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -115,6 +115,9 @@ nv20_screen_destroy(struct pipe_screen *pscreen)
 
 	nouveau_notifier_free(&screen->sync);
 	nouveau_grobj_free(&screen->kelvin);
+	nv04_surface_2d_takedown(&screen->eng2d);
+
+	nouveau_screen_fini(&screen->base);
 
 	FREE(pscreen);
 }
@@ -173,7 +176,6 @@ nv20_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
 		return FALSE;
 	}
-	BIND_RING(chan, screen->kelvin, 7);
 
 	/* 2D engine setup */
 	screen->eng2d = nv04_surface_2d_init(&screen->base);
diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c
index ed4084980f2..3a82e63423d 100644
--- a/src/gallium/drivers/nv20/nv20_state.c
+++ b/src/gallium/drivers/nv20/nv20_state.c
@@ -546,9 +546,9 @@ nv20_init_state_functions(struct nv20_context *nv20)
 	nv20->pipe.delete_blend_state = nv20_blend_state_delete;
 
 	nv20->pipe.create_sampler_state = nv20_sampler_state_create;
-	nv20->pipe.bind_sampler_states = nv20_sampler_state_bind;
+	nv20->pipe.bind_fragment_sampler_states = nv20_sampler_state_bind;
 	nv20->pipe.delete_sampler_state = nv20_sampler_state_delete;
-	nv20->pipe.set_sampler_textures = nv20_set_sampler_texture;
+	nv20->pipe.set_fragment_sampler_textures = nv20_set_sampler_texture;
 
 	nv20->pipe.create_rasterizer_state = nv20_rasterizer_state_create;
 	nv20->pipe.bind_rasterizer_state = nv20_rasterizer_state_bind;
diff --git a/src/gallium/drivers/nv20/nv20_state.h b/src/gallium/drivers/nv20/nv20_state.h
index 34f402fdcbf..dde41065685 100644
--- a/src/gallium/drivers/nv20/nv20_state.h
+++ b/src/gallium/drivers/nv20/nv20_state.h
@@ -126,6 +126,7 @@ struct nv20_depth_stencil_alpha_state {
 
 struct nv20_miptree {
 	struct pipe_texture base;
+	struct nouveau_bo *bo;
 
 	struct pipe_buffer *buffer;
 	uint total_size;
diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c
index 4042f46d053..6bbd1fdae9d 100644
--- a/src/gallium/drivers/nv20/nv20_state_emit.c
+++ b/src/gallium/drivers/nv20/nv20_state_emit.c
@@ -5,27 +5,34 @@
 static void nv20_state_emit_blend(struct nv20_context* nv20)
 {
 	struct nv20_blend_state *b = nv20->blend;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 
-	BEGIN_RING(kelvin, NV20TCL_DITHER_ENABLE, 1);
-	OUT_RING  (b->d_enable);
+	BEGIN_RING(chan, kelvin, NV20TCL_DITHER_ENABLE, 1);
+	OUT_RING  (chan, b->d_enable);
 
-	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
-	OUT_RING  (b->b_enable);
+	BEGIN_RING(chan, kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
+	OUT_RING  (chan, b->b_enable);
 
-	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_SRC, 2);
-	OUT_RING  (b->b_srcfunc);
-	OUT_RING  (b->b_dstfunc);
+	BEGIN_RING(chan, kelvin, NV20TCL_BLEND_FUNC_SRC, 2);
+	OUT_RING  (chan, b->b_srcfunc);
+	OUT_RING  (chan, b->b_dstfunc);
 
-	BEGIN_RING(kelvin, NV20TCL_COLOR_MASK, 1);
-	OUT_RING  (b->c_mask);
+	BEGIN_RING(chan, kelvin, NV20TCL_COLOR_MASK, 1);
+	OUT_RING  (chan, b->c_mask);
 }
 
 static void nv20_state_emit_blend_color(struct nv20_context* nv20)
 {
 	struct pipe_blend_color *c = nv20->blend_color;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 
-	BEGIN_RING(kelvin, NV20TCL_BLEND_COLOR, 1);
-	OUT_RING  ((float_to_ubyte(c->color[3]) << 24)|
+	BEGIN_RING(chan, kelvin, NV20TCL_BLEND_COLOR, 1);
+	OUT_RING  (chan,
+		   (float_to_ubyte(c->color[3]) << 24)|
 		   (float_to_ubyte(c->color[0]) << 16)|
 		   (float_to_ubyte(c->color[1]) << 8) |
 		   (float_to_ubyte(c->color[2]) << 0));
@@ -34,63 +41,69 @@ static void nv20_state_emit_blend_color(struct nv20_context* nv20)
 static void nv20_state_emit_rast(struct nv20_context* nv20)
 {
 	struct nv20_rasterizer_state *r = nv20->rast;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 
-	BEGIN_RING(kelvin, NV20TCL_SHADE_MODEL, 2);
-	OUT_RING  (r->shade_model);
-	OUT_RING  (r->line_width);
+	BEGIN_RING(chan, kelvin, NV20TCL_SHADE_MODEL, 2);
+	OUT_RING  (chan, r->shade_model);
+	OUT_RING  (chan, r->line_width);
 
 
-	BEGIN_RING(kelvin, NV20TCL_POINT_SIZE, 1);
-	OUT_RING  (r->point_size);
+	BEGIN_RING(chan, kelvin, NV20TCL_POINT_SIZE, 1);
+	OUT_RING  (chan, r->point_size);
 
-	BEGIN_RING(kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
-	OUT_RING  (r->poly_mode_front);
-	OUT_RING  (r->poly_mode_back);
+	BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (chan, r->poly_mode_front);
+	OUT_RING  (chan, r->poly_mode_back);
 
 
-	BEGIN_RING(kelvin, NV20TCL_CULL_FACE, 2);
-	OUT_RING  (r->cull_face);
-	OUT_RING  (r->front_face);
+	BEGIN_RING(chan, kelvin, NV20TCL_CULL_FACE, 2);
+	OUT_RING  (chan, r->cull_face);
+	OUT_RING  (chan, r->front_face);
 
-	BEGIN_RING(kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 2);
-	OUT_RING  (r->line_smooth_en);
-	OUT_RING  (r->poly_smooth_en);
+	BEGIN_RING(chan, kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 2);
+	OUT_RING  (chan, r->line_smooth_en);
+	OUT_RING  (chan, r->poly_smooth_en);
 
-	BEGIN_RING(kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
-	OUT_RING  (r->cull_face_en);
+	BEGIN_RING(chan, kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (chan, r->cull_face_en);
 }
 
 static void nv20_state_emit_dsa(struct nv20_context* nv20)
 {
 	struct nv20_depth_stencil_alpha_state *d = nv20->dsa;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_FUNC, 1);
-	OUT_RING (d->depth.func);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_FUNC, 1);
+	OUT_RING (chan, d->depth.func);
 
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING (d->depth.write_enable);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING (chan, d->depth.write_enable);
 
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
-	OUT_RING (d->depth.test_enable);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING (chan, d->depth.test_enable);
 
-	BEGIN_RING(kelvin, NV20TCL_DEPTH_UNK17D8, 1);
-	OUT_RING (1);
+	BEGIN_RING(chan, kelvin, NV20TCL_DEPTH_UNK17D8, 1);
+	OUT_RING (chan, 1);
 
 #if 0
-	BEGIN_RING(kelvin, NV20TCL_STENCIL_ENABLE, 1);
-	OUT_RING (d->stencil.enable);
-	BEGIN_RING(kelvin, NV20TCL_STENCIL_MASK, 7);
-	OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7);
+	BEGIN_RING(chan, kelvin, NV20TCL_STENCIL_ENABLE, 1);
+	OUT_RING (chan, d->stencil.enable);
+	BEGIN_RING(chan, kelvin, NV20TCL_STENCIL_MASK, 7);
+	OUT_RINGp (chan, (uint32_t *)&(d->stencil.wmask), 7);
 #endif
 
-	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
-	OUT_RING (d->alpha.enabled);
+	BEGIN_RING(chan, kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING (chan, d->alpha.enabled);
 
-	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_FUNC, 1);
-	OUT_RING (d->alpha.func);
+	BEGIN_RING(chan, kelvin, NV20TCL_ALPHA_FUNC_FUNC, 1);
+	OUT_RING (chan, d->alpha.func);
 
-	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_REF, 1);
-	OUT_RING (d->alpha.ref);
+	BEGIN_RING(chan, kelvin, NV20TCL_ALPHA_FUNC_REF, 1);
+	OUT_RING (chan, d->alpha.ref);
 }
 
 static void nv20_state_emit_viewport(struct nv20_context* nv20)
@@ -101,9 +114,13 @@ static void nv20_state_emit_scissor(struct nv20_context* nv20)
 {
 	/* NV20TCL_SCISSOR_* is probably a software method */
 /*	struct pipe_scissor_state *s = nv20->scissor;
-	BEGIN_RING(kelvin, NV20TCL_SCISSOR_HORIZ, 2);
-	OUT_RING  (((s->maxx - s->minx) << 16) | s->minx);
-	OUT_RING  (((s->maxy - s->miny) << 16) | s->miny);*/
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
+
+	BEGIN_RING(chan, kelvin, NV20TCL_SCISSOR_HORIZ, 2);
+	OUT_RING  (chan, ((s->maxx - s->minx) << 16) | s->minx);
+	OUT_RING  (chan, ((s->maxy - s->miny) << 16) | s->miny);*/
 }
 
 static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
@@ -113,6 +130,9 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 	uint32_t rt_format, w, h;
 	int colour_format = 0, zeta_format = 0;
 	struct nv20_miptree *nv20mt = 0;
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
 
 	w = fb->cbufs[0]->width;
 	h = fb->cbufs[0]->height;
@@ -135,6 +155,9 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 	rt_format = NV20TCL_RT_FORMAT_TYPE_LINEAR | 0x20;
 
 	switch (colour_format) {
+	case PIPE_FORMAT_X8R8G8B8_UNORM:
+		rt_format |= NV20TCL_RT_FORMAT_COLOR_X8R8G8B8;
+		break;
 	case PIPE_FORMAT_A8R8G8B8_UNORM:
 	case 0:
 		rt_format |= NV20TCL_RT_FORMAT_COLOR_A8R8G8B8;
@@ -147,11 +170,11 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 	}
 
 	if (zeta) {
-		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
-		OUT_RING  (rt->pitch | (zeta->pitch << 16));
+		BEGIN_RING(chan, kelvin, NV20TCL_RT_PITCH, 1);
+		OUT_RING  (chan, rt->pitch | (zeta->pitch << 16));
 	} else {
-		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
-		OUT_RING  (rt->pitch | (rt->pitch << 16));
+		BEGIN_RING(chan, kelvin, NV20TCL_RT_PITCH, 1);
+		OUT_RING  (chan, rt->pitch | (rt->pitch << 16));
 	}
 
 	nv20mt = (struct nv20_miptree *)rt->base.texture;
@@ -163,13 +186,13 @@ static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
 		nv20->zeta = nv20mt->buffer;
 	}
 
-	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 3);
-	OUT_RING  ((w << 16) | 0);
-	OUT_RING  ((h << 16) | 0); /*NV20TCL_RT_VERT */
-	OUT_RING  (rt_format); /* NV20TCL_RT_FORMAT */
-	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 2);
-	OUT_RING  (((w - 1) << 16) | 0);
-	OUT_RING  (((h - 1) << 16) | 0);
+	BEGIN_RING(chan, kelvin, NV20TCL_RT_HORIZ, 3);
+	OUT_RING  (chan, (w << 16) | 0);
+	OUT_RING  (chan, (h << 16) | 0); /*NV20TCL_RT_VERT */
+	OUT_RING  (chan, rt_format); /* NV20TCL_RT_FORMAT */
+	BEGIN_RING(chan, kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 2);
+	OUT_RING  (chan, ((w - 1) << 16) | 0);
+	OUT_RING  (chan, ((h - 1) << 16) | 0);
 }
 
 static void nv20_vertex_layout(struct nv20_context *nv20)
@@ -225,7 +248,7 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 	}
 
 	/* always do position */ {
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_POSITION, 0);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_POSITION, 0);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_LINEAR, src);
 		vinfo->hwfmt[0] |= (1 << 0);
 	}
@@ -234,19 +257,19 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 	for (i = 4; i < 6; i++) {
 		if (!generics[i])
 			continue;
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_GENERIC, i);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_GENERIC, i);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
 		vinfo->hwfmt[0] |= (1 << (i - 3));
 	}
 
 	if (colors[0]) {
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_COLOR, 0);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_COLOR, 0);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
 		vinfo->hwfmt[0] |= (1 << 3);
 	}
 
 	if (colors[1]) {
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_COLOR, 1);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_COLOR, 1);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
 		vinfo->hwfmt[0] |= (1 << 4);
 	}
@@ -255,7 +278,7 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 	for (i = 6; i < 10; i++) {
 		if (!generics[i])
 			continue;
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_GENERIC, i);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_GENERIC, i);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
 		vinfo->hwfmt[0] |= (1 << (i - 1));
 	}
@@ -264,7 +287,7 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 	for (i = 0; i < 4; i++) {
 		if (!generics[i])
 			continue;
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_GENERIC, i);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_GENERIC, i);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
 		vinfo->hwfmt[0] |= (1 << (i + 9));
 	}
@@ -273,13 +296,13 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 	for (i = 10; i < 12; i++) {
 		if (!generics[i])
 			continue;
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_GENERIC, i);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_GENERIC, i);
 		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
 		vinfo->hwfmt[0] |= (1 << (i + 3));
 	}
 
 	if (fog) {
-		src = draw_find_vs_output(dc, TGSI_SEMANTIC_FOG, 0);
+		src = draw_find_shader_output(dc, TGSI_SEMANTIC_FOG, 0);
 		draw_emit_vertex_attr(vinfo, EMIT_1F, INTERP_PERSPECTIVE, src);
 		vinfo->hwfmt[0] |= (1 << 15);
 	}
@@ -290,6 +313,10 @@ static void nv20_vertex_layout(struct nv20_context *nv20)
 void
 nv20_emit_hw_state(struct nv20_context *nv20)
 {
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_channel *chan = screen->base.channel;
+	struct nouveau_grobj *kelvin = screen->kelvin;
+	struct nouveau_bo *rt_bo;
 	int i;
 
 	if (nv20->dirty & NV20_NEW_VERTPROG) {
@@ -358,36 +385,39 @@ nv20_emit_hw_state(struct nv20_context *nv20)
 	 */
 
 	/* Render target */
-	BEGIN_RING(kelvin, NV20TCL_DMA_COLOR, 1);
-	OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
-	BEGIN_RING(kelvin, NV20TCL_COLOR_OFFSET, 1);
-	OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	rt_bo = nouveau_bo(nv20->rt[0]);
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_COLOR, 1);
+	OUT_RELOCo(chan, rt_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(chan, kelvin, NV20TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(chan, rt_bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
 	if (nv20->zeta) {
-		BEGIN_RING(kelvin, NV20TCL_DMA_ZETA, 1);
-		OUT_RELOCo(nv20->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
-		BEGIN_RING(kelvin, NV20TCL_ZETA_OFFSET, 1);
-		OUT_RELOCl(nv20->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+		struct nouveau_bo *zeta_bo = nouveau_bo(nv20->zeta);
+		BEGIN_RING(chan, kelvin, NV20TCL_DMA_ZETA, 1);
+		OUT_RELOCo(chan, zeta_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+		BEGIN_RING(chan, kelvin, NV20TCL_ZETA_OFFSET, 1);
+		OUT_RELOCl(chan, zeta_bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 		/* XXX for when we allocate LMA on nv17 */
-/*		BEGIN_RING(kelvin, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1);
-		OUT_RELOCl(nv20->zeta + lma_offset);*/
+/*		BEGIN_RING(chan, kelvin, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1);
+		OUT_RELOCl(chan, nouveau_bo(nv20->zeta + lma_offset));*/
 	}
 
 	/* Vertex buffer */
-	BEGIN_RING(kelvin, NV20TCL_DMA_VTXBUF0, 1);
-	OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
-	BEGIN_RING(kelvin, NV20TCL_COLOR_OFFSET, 1);
-	OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(chan, kelvin, NV20TCL_DMA_VTXBUF0, 1);
+	OUT_RELOCo(chan, rt_bo, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(chan, kelvin, NV20TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(chan, rt_bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
 
 	/* Texture images */
 	for (i = 0; i < 2; i++) {
 		if (!(nv20->fp_samplers & (1 << i)))
 			continue;
-		BEGIN_RING(kelvin, NV20TCL_TX_OFFSET(i), 1);
-		OUT_RELOCl(nv20->tex[i].buffer, 0, NOUVEAU_BO_VRAM |
+		struct nouveau_bo *bo = nouveau_bo(nv20->tex[i].buffer);
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_OFFSET(i), 1);
+		OUT_RELOCl(chan, bo, 0, NOUVEAU_BO_VRAM |
 			   NOUVEAU_BO_GART | NOUVEAU_BO_RD);
-		BEGIN_RING(kelvin, NV20TCL_TX_FORMAT(i), 1);
-		OUT_RELOCd(nv20->tex[i].buffer, nv20->tex[i].format,
+		BEGIN_RING(chan, kelvin, NV20TCL_TX_FORMAT(i), 1);
+		OUT_RELOCd(chan, bo, nv20->tex[i].format,
 			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD |
 			   NOUVEAU_BO_OR, NV20TCL_TX_FORMAT_DMA0,
 			   NV20TCL_TX_FORMAT_DMA1);
diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c
index 81b4f1a9177..699773e8e6f 100644
--- a/src/gallium/drivers/nv20/nv20_transfer.c
+++ b/src/gallium/drivers/nv20/nv20_transfer.c
@@ -1,7 +1,9 @@
 #include <pipe/p_state.h>
 #include <pipe/p_defines.h>
 #include <pipe/p_inlines.h>
+#include <util/u_format.h>
 #include <util/u_memory.h>
+#include <util/u_math.h>
 #include <nouveau/nouveau_winsys.h>
 #include "nv20_context.h"
 #include "nv20_screen.h"
@@ -10,22 +12,19 @@
 struct nv20_transfer {
 	struct pipe_transfer base;
 	struct pipe_surface *surface;
-	bool direct;
+	boolean direct;
 };
 
 static void
-nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level,
+nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned width, unsigned height,
                              struct pipe_texture *template)
 {
 	memset(template, 0, sizeof(struct pipe_texture));
 	template->target = pt->target;
 	template->format = pt->format;
-	template->width[0] = pt->width[level];
-	template->height[0] = pt->height[level];
-	template->depth[0] = 1;
-	template->block = pt->block;
-	template->nblocksx[0] = pt->nblocksx[level];
-	template->nblocksy[0] = pt->nblocksx[level];
+	template->width0 = width;
+	template->height0 = height;
+	template->depth0 = 1;
 	template->last_level = 0;
 	template->nr_samples = pt->nr_samples;
 
@@ -48,14 +47,10 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 
 	pipe_texture_reference(&tx->base.texture, pt);
-	tx->base.format = pt->format;
 	tx->base.x = x;
 	tx->base.y = y;
 	tx->base.width = w;
 	tx->base.height = h;
-	tx->base.block = pt->block;
-	tx->base.nblocksx = pt->nblocksx[level];
-	tx->base.nblocksy = pt->nblocksy[level];
 	tx->base.stride = mt->level[level].pitch;
 	tx->base.usage = usage;
 	tx->base.face = face;
@@ -76,7 +71,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 
 	tx->direct = false;
 
-	nv20_compatible_transfer_tex(pt, level, &tx_tex_template);
+	nv20_compatible_transfer_tex(pt, w, h, &tx_tex_template);
 
 	tx_tex = pscreen->texture_create(pscreen, &tx_tex_template);
 	if (!tx_tex)
@@ -85,6 +80,8 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		return NULL;
 	}
 
+	tx->base.stride = ((struct nv20_miptree*)tx_tex)->level[0].pitch;
+
 	tx->surface = pscreen->get_tex_surface(pscreen, tx_tex,
 	                                       face, level, zslice,
 	                                       pipe_transfer_buffer_flags(&tx->base));
@@ -110,8 +107,8 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 		/* TODO: Check if SIFM can un-swizzle */
 		nvscreen->eng2d->copy(nvscreen->eng2d,
 		                      tx->surface, 0, 0,
-		                      src, 0, 0,
-		                      src->width, src->height);
+		                      src, x, y,
+		                      w, h);
 
 		pipe_surface_reference(&src, NULL);
 	}
@@ -131,13 +128,13 @@ nv20_transfer_del(struct pipe_transfer *ptx)
 
 		dst = pscreen->get_tex_surface(pscreen, ptx->texture,
 	                                       ptx->face, ptx->level, ptx->zslice,
-	                                       PIPE_BUFFER_USAGE_GPU_WRITE);
+	                                       PIPE_BUFFER_USAGE_GPU_WRITE | NOUVEAU_BUFFER_USAGE_NO_RENDER);
 
 		/* TODO: Check if SIFM can deal with x,y,w,h when swizzling */
 		nvscreen->eng2d->copy(nvscreen->eng2d,
-		                      dst, 0, 0,
+		                      dst, tx->base.x, tx->base.y,
 		                      tx->surface, 0, 0,
-		                      dst->width, dst->height);
+		                      tx->base.width, tx->base.height);
 
 		pipe_surface_reference(&dst, NULL);
 	}
@@ -156,8 +153,10 @@ nv20_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx)
 	void *map = pipe_buffer_map(pscreen, mt->buffer,
 	                            pipe_transfer_buffer_flags(ptx));
 
-	return map + ns->base.offset +
-	       ptx->y * ns->pitch + ptx->x * ptx->block.size;
+	if(!tx->direct)
+		return map + ns->base.offset;
+	else
+		return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * util_format_get_blocksize(ptx->texture->format);
 }
 
 static void
diff --git a/src/gallium/drivers/nv20/nv20_vbo.c b/src/gallium/drivers/nv20/nv20_vbo.c
index 84d7db6c5e2..52991a0d856 100644
--- a/src/gallium/drivers/nv20/nv20_vbo.c
+++ b/src/gallium/drivers/nv20/nv20_vbo.c
@@ -9,7 +9,7 @@
 #include "nouveau/nouveau_channel.h"
 #include "nouveau/nouveau_pushbuf.h"
 
-boolean nv20_draw_elements( struct pipe_context *pipe,
+void nv20_draw_elements( struct pipe_context *pipe,
                     struct pipe_buffer *indexBuffer,
                     unsigned indexSize,
                     unsigned prim, unsigned start, unsigned count)
@@ -45,7 +45,7 @@ boolean nv20_draw_elements( struct pipe_context *pipe,
 		draw_set_mapped_element_buffer(draw, 0, NULL);
 	}
 
-	draw_set_mapped_constant_buffer(draw,
+	draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX,
 					nv20->constbuf[PIPE_SHADER_VERTEX],
 					nv20->constbuf_nr[PIPE_SHADER_VERTEX]);
 
@@ -67,13 +67,12 @@ boolean nv20_draw_elements( struct pipe_context *pipe,
 	}
 
 	draw_flush(nv20->draw);
-	return TRUE;
 }
 
-boolean nv20_draw_arrays( struct pipe_context *pipe,
+void nv20_draw_arrays( struct pipe_context *pipe,
 				 unsigned prim, unsigned start, unsigned count)
 {
-	return nv20_draw_elements(pipe, NULL, 0, prim, start, count);
+	nv20_draw_elements(pipe, NULL, 0, prim, start, count);
 }
 
 
diff --git a/src/gallium/drivers/nv20/nv20_vertprog.c b/src/gallium/drivers/nv20/nv20_vertprog.c
index 388245ecb04..7886c2af7e6 100644
--- a/src/gallium/drivers/nv20/nv20_vertprog.c
+++ b/src/gallium/drivers/nv20/nv20_vertprog.c
@@ -253,32 +253,32 @@ static INLINE struct nv20_sreg
 tgsi_src(struct nv20_vpc *vpc, const struct tgsi_full_src_register *fsrc) {
 	struct nv20_sreg src;
 
-	switch (fsrc->SrcRegister.File) {
+	switch (fsrc->Register.File) {
 	case TGSI_FILE_INPUT:
-		src = nv20_sr(NV30SR_INPUT, fsrc->SrcRegister.Index);
+		src = nv20_sr(NV30SR_INPUT, fsrc->Register.Index);
 		break;
 	case TGSI_FILE_CONSTANT:
-		src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0);
+		src = constant(vpc, fsrc->Register.Index, 0, 0, 0, 0);
 		break;
 	case TGSI_FILE_IMMEDIATE:
-		src = vpc->imm[fsrc->SrcRegister.Index];
+		src = vpc->imm[fsrc->Register.Index];
 		break;
 	case TGSI_FILE_TEMPORARY:
-		if (vpc->high_temp < fsrc->SrcRegister.Index)
-			vpc->high_temp = fsrc->SrcRegister.Index;
-		src = nv20_sr(NV30SR_TEMP, fsrc->SrcRegister.Index);
+		if (vpc->high_temp < fsrc->Register.Index)
+			vpc->high_temp = fsrc->Register.Index;
+		src = nv20_sr(NV30SR_TEMP, fsrc->Register.Index);
 		break;
 	default:
 		NOUVEAU_ERR("bad src file\n");
 		break;
 	}
 
-	src.abs = fsrc->SrcRegisterExtMod.Absolute;
-	src.negate = fsrc->SrcRegister.Negate;
-	src.swz[0] = fsrc->SrcRegister.SwizzleX;
-	src.swz[1] = fsrc->SrcRegister.SwizzleY;
-	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
-	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	src.abs = fsrc->Register.Absolute;
+	src.negate = fsrc->Register.Negate;
+	src.swz[0] = fsrc->Register.SwizzleX;
+	src.swz[1] = fsrc->Register.SwizzleY;
+	src.swz[2] = fsrc->Register.SwizzleZ;
+	src.swz[3] = fsrc->Register.SwizzleW;
 	return src;
 }
 
@@ -286,14 +286,14 @@ static INLINE struct nv20_sreg
 tgsi_dst(struct nv20_vpc *vpc, const struct tgsi_full_dst_register *fdst) {
 	struct nv20_sreg dst;
 
-	switch (fdst->DstRegister.File) {
+	switch (fdst->Register.File) {
 	case TGSI_FILE_OUTPUT:
 		dst = nv20_sr(NV30SR_OUTPUT,
-			      vpc->output_map[fdst->DstRegister.Index]);
+			      vpc->output_map[fdst->Register.Index]);
 
 		break;
 	case TGSI_FILE_TEMPORARY:
-		dst = nv20_sr(NV30SR_TEMP, fdst->DstRegister.Index);
+		dst = nv20_sr(NV30SR_TEMP, fdst->Register.Index);
 		if (vpc->high_temp < dst.index)
 			vpc->high_temp = dst.index;
 		break;
@@ -334,8 +334,8 @@ nv20_vertprog_parse_instruction(struct nv20_vpc *vpc,
 	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
 		const struct tgsi_full_src_register *fsrc;
 
-		fsrc = &finst->FullSrcRegisters[i];
-		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+		fsrc = &finst->Src[i];
+		if (fsrc->Register.File == TGSI_FILE_TEMPORARY) {
 			src[i] = tgsi_src(vpc, fsrc);
 		}
 	}
@@ -343,11 +343,11 @@ nv20_vertprog_parse_instruction(struct nv20_vpc *vpc,
 	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
 		const struct tgsi_full_src_register *fsrc;
 
-		fsrc = &finst->FullSrcRegisters[i];
-		switch (fsrc->SrcRegister.File) {
+		fsrc = &finst->Src[i];
+		switch (fsrc->Register.File) {
 		case TGSI_FILE_INPUT:
-			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
-				ai = fsrc->SrcRegister.Index;
+			if (ai == -1 || ai == fsrc->Register.Index) {
+				ai = fsrc->Register.Index;
 				src[i] = tgsi_src(vpc, fsrc);
 			} else {
 				src[i] = temp(vpc);
@@ -360,8 +360,8 @@ nv20_vertprog_parse_instruction(struct nv20_vpc *vpc,
 		 */
 		case TGSI_FILE_CONSTANT:
 		case TGSI_FILE_IMMEDIATE:
-			if (ci == -1 || ci == fsrc->SrcRegister.Index) {
-				ci = fsrc->SrcRegister.Index;
+			if (ci == -1 || ci == fsrc->Register.Index) {
+				ci = fsrc->Register.Index;
 				src[i] = tgsi_src(vpc, fsrc);
 			} else {
 				src[i] = temp(vpc);
@@ -378,8 +378,8 @@ nv20_vertprog_parse_instruction(struct nv20_vpc *vpc,
 		}
 	}
 
-	dst  = tgsi_dst(vpc, &finst->FullDstRegisters[0]);
-	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+	dst  = tgsi_dst(vpc, &finst->Dst[0]);
+	mask = tgsi_mask(finst->Dst[0].Register.WriteMask);
 
 	switch (finst->Instruction.Opcode) {
 	case TGSI_OPCODE_ABS:
@@ -490,15 +490,15 @@ nv20_vertprog_parse_decl_output(struct nv20_vpc *vpc,
 {
 	int hw;
 
-	switch (fdec->Semantic.SemanticName) {
+	switch (fdec->Semantic.Name) {
 	case TGSI_SEMANTIC_POSITION:
 		hw = NV30_VP_INST_DEST_POS;
 		break;
 	case TGSI_SEMANTIC_COLOR:
-		if (fdec->Semantic.SemanticIndex == 0) {
+		if (fdec->Semantic.Index == 0) {
 			hw = NV30_VP_INST_DEST_COL0;
 		} else
-		if (fdec->Semantic.SemanticIndex == 1) {
+		if (fdec->Semantic.Index == 1) {
 			hw = NV30_VP_INST_DEST_COL1;
 		} else {
 			NOUVEAU_ERR("bad colour semantic index\n");
@@ -506,10 +506,10 @@ nv20_vertprog_parse_decl_output(struct nv20_vpc *vpc,
 		}
 		break;
 	case TGSI_SEMANTIC_BCOLOR:
-		if (fdec->Semantic.SemanticIndex == 0) {
+		if (fdec->Semantic.Index == 0) {
 			hw = NV30_VP_INST_DEST_BFC0;
 		} else
-		if (fdec->Semantic.SemanticIndex == 1) {
+		if (fdec->Semantic.Index == 1) {
 			hw = NV30_VP_INST_DEST_BFC1;
 		} else {
 			NOUVEAU_ERR("bad bcolour semantic index\n");
@@ -523,19 +523,22 @@ nv20_vertprog_parse_decl_output(struct nv20_vpc *vpc,
 		hw = NV30_VP_INST_DEST_PSZ;
 		break;
 	case TGSI_SEMANTIC_GENERIC:
-		if (fdec->Semantic.SemanticIndex <= 7) {
-			hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex);
+		if (fdec->Semantic.Index <= 7) {
+			hw = NV30_VP_INST_DEST_TC(fdec->Semantic.Index);
 		} else {
 			NOUVEAU_ERR("bad generic semantic index\n");
 			return FALSE;
 		}
 		break;
+	case TGSI_SEMANTIC_EDGEFLAG:
+		NOUVEAU_ERR("cannot handle edgeflag output\n");
+		return FALSE;
 	default:
 		NOUVEAU_ERR("bad output semantic\n");
 		return FALSE;
 	}
 
-	vpc->output_map[fdec->DeclarationRange.First] = hw;
+	vpc->output_map[fdec->Range.First] = hw;
 	return TRUE;
 }