Merge branch 'gallium-nopointsizeminmax'

Conflicts:
	src/gallium/drivers/nv10/nv10_state.c
	src/gallium/drivers/nv20/nv20_state.c
	src/gallium/drivers/nv50/nv50_program.c

28 files changed, 1381 insertions, 104 deletions

diff --git a/src/gallium/auxiliary/draw/draw_pipe_validate.c b/src/gallium/auxiliary/draw/draw_pipe_validate.c
index a69e2633be..153097e543 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_validate.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_validate.c
@@ -105,7 +105,7 @@ draw_need_pipeline(const struct draw_context *draw,
          return TRUE;
 
       /* point sprites */
-      if (rasterizer->point_sprite && draw->pipeline.point_sprite)
+      if (rasterizer->sprite_coord_enable && draw->pipeline.point_sprite)
          return TRUE;
    }
 
@@ -165,7 +165,7 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage )
                  && !draw->rasterizer->line_smooth);
 
    /* drawing large points? */
-   if (draw->rasterizer->point_sprite && draw->pipeline.point_sprite)
+   if (draw->rasterizer->sprite_coord_enable && draw->pipeline.point_sprite)
       wide_points = TRUE;
    else if (draw->rasterizer->point_smooth && draw->pipeline.aapoint)
       wide_points = FALSE;
@@ -197,7 +197,7 @@ static struct draw_stage *validate_pipeline( struct draw_stage *stage )
       precalc_flat = TRUE;
    }
 
-   if (wide_points || draw->rasterizer->point_sprite) {
+   if (wide_points || draw->rasterizer->sprite_coord_enable) {
       draw->pipeline.wide_point->next = next;
       next = draw->pipeline.wide_point;
    }
diff --git a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
index 8dc50c0ab4..fdabce7d44 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
@@ -64,15 +64,14 @@ struct widepoint_stage {
    struct draw_stage stage;
 
    float half_point_size;
-   float point_size_min;
-   float point_size_max;
 
    float xbias;
    float ybias;
 
    uint texcoord_slot[PIPE_MAX_SHADER_OUTPUTS];
-   uint texcoord_mode[PIPE_MAX_SHADER_OUTPUTS];
+   uint texcoord_enable[PIPE_MAX_SHADER_OUTPUTS];
    uint num_texcoords;
+   uint texcoord_mode;
 
    int psize_slot;
 
@@ -98,10 +97,10 @@ static void set_texcoords(const struct widepoint_stage *wide,
 {
    uint i;
    for (i = 0; i < wide->num_texcoords; i++) {
-      if (wide->texcoord_mode[i] != PIPE_SPRITE_COORD_NONE) {
+      if (wide->texcoord_enable[i]) {
          uint j = wide->texcoord_slot[i];
          v->data[j][0] = tc[0];
-         if (wide->texcoord_mode[i] == PIPE_SPRITE_COORD_LOWER_LEFT)
+         if (wide->texcoord_mode == PIPE_SPRITE_COORD_LOWER_LEFT)
             v->data[j][1] = 1.0f - tc[1];
          else
             v->data[j][1] = tc[1];
@@ -129,9 +128,10 @@ static void set_texcoords(const struct widepoint_stage *wide,
 static void widepoint_point( struct draw_stage *stage,
                              struct prim_header *header )
 {
+   /* XXX should take point_quad_rasterization into account? */
    const struct widepoint_stage *wide = widepoint_stage(stage);
    const unsigned pos = draw_current_shader_position_output(stage->draw);
-   const boolean sprite = (boolean) stage->draw->rasterizer->point_sprite;
+   const boolean sprite = (boolean) stage->draw->rasterizer->sprite_coord_enable;
    float half_size;
    float left_adj, right_adj, bot_adj, top_adj;
 
@@ -151,13 +151,6 @@ static void widepoint_point( struct draw_stage *stage,
    /* point size is either per-vertex or fixed size */
    if (wide->psize_slot >= 0) {
       half_size = header->v[0]->data[wide->psize_slot][0];
-
-      /* XXX: temporary -- do this in the vertex shader??
-       */
-      half_size = CLAMP(half_size,
-                        wide->point_size_min,
-                        wide->point_size_max);
-      
       half_size *= 0.5f; 
    }
    else {
@@ -222,8 +215,6 @@ static void widepoint_first_point( struct draw_stage *stage,
    struct draw_context *draw = stage->draw;
 
    wide->half_point_size = 0.5f * draw->rasterizer->point_size;
-   wide->point_size_min = draw->rasterizer->point_size_min;
-   wide->point_size_max = draw->rasterizer->point_size_max;
    wide->xbias = 0.0;
    wide->ybias = 0.0;
 
@@ -233,21 +224,22 @@ static void widepoint_first_point( struct draw_stage *stage,
 
    /* XXX we won't know the real size if it's computed by the vertex shader! */
    if ((draw->rasterizer->point_size > draw->pipeline.wide_point_threshold) ||
-       (draw->rasterizer->point_sprite && draw->pipeline.point_sprite)) {
+       (draw->rasterizer->sprite_coord_enable && draw->pipeline.point_sprite)) {
       stage->point = widepoint_point;
    }
    else {
       stage->point = draw_pipe_passthrough_point;
    }
 
-   if (draw->rasterizer->point_sprite) {
+   if (draw->rasterizer->sprite_coord_enable) {
       /* find vertex shader texcoord outputs */
       const struct draw_vertex_shader *vs = draw->vs.vertex_shader;
       uint i, j = 0;
+      wide->texcoord_mode = draw->rasterizer->sprite_coord_mode;
       for (i = 0; i < vs->info.num_outputs; i++) {
          if (vs->info.output_semantic_name[i] == TGSI_SEMANTIC_GENERIC) {
             wide->texcoord_slot[j] = i;
-            wide->texcoord_mode[j] = draw->rasterizer->sprite_coord_mode[j];
+            wide->texcoord_enable[j] = (draw->rasterizer->sprite_coord_enable >> j) & 1;
             j++;
          }
       }
diff --git a/src/gallium/docs/source/cso/rasterizer.rst b/src/gallium/docs/source/cso/rasterizer.rst
index bfa4a1170a..24cc78c68d 100644
--- a/src/gallium/docs/source/cso/rasterizer.rst
+++ b/src/gallium/docs/source/cso/rasterizer.rst
@@ -119,18 +119,20 @@ point_size_per_vertex
     Whether vertices have a point size element.
 point_size
     The size of points, if not specified per-vertex.
-point_size_min
-    The minimum size of points.
-point_size_max
-    The maximum size of points.
-point_sprite
-    Whether points are drawn as sprites (textured quads)
+sprite_coord_enable
+    Specifies if a coord has its texture coordinates replaced or not. This
+    is a packed bitfield containing the enable for all coords - if all are 0
+    point sprites are effectively disabled, though points may still be
+    rendered slightly different according to point_quad_rasterization.
+    If any coord is non-zero, point_smooth should be disabled, and
+    point_quad_rasterization enabled.
+    If enabled, the four vertices of the resulting quad will be assigned
+    texture coordinates, according to sprite_coord_mode.
 sprite_coord_mode
     Specifies how the value for each shader output should be computed when
-    drawing sprites.  If PIPE_SPRITE_COORD_NONE, don't change the vertex
-    shader output.  Otherwise, the four vertices of the resulting quad will
-    be assigned texture coordinates.  For PIPE_SPRITE_COORD_LOWER_LEFT, the
-    lower left vertex will have coordinate (0,0,0,1).
+    drawing sprites, for each coord which has sprite_coord_enable set.
+    For PIPE_SPRITE_COORD_LOWER_LEFT, the lower left vertex will have
+    coordinate (0,0,0,1).
     For PIPE_SPRITE_COORD_UPPER_LEFT, the upper-left vertex will have
     coordinate (0,0,0,1).
     This state is needed by :ref:`Draw` because that's where each
@@ -139,7 +141,15 @@ sprite_coord_mode
     sprite rendering.
     Note that when geometry shaders are available, this state could be
     removed.  A special geometry shader defined by the state tracker could
-    converts the incoming points into quads with the proper texture coords.
+    convert the incoming points into quads with the proper texture coords.
+point_quad_rasterization
+    This determines if points should be rasterized as quads or points.
+    d3d always uses quad rasterization for points, regardless if point sprites
+    are enabled or not, but OGL has different rules. If point_quad_rasterization
+    is set, point_smooth should be disabled, and points will be rendered as
+    squares even if multisample is enabled.
+    sprite_coord_enable should be zero if point_quad_rasterization is not
+    enabled.
 
 scissor
     Whether the scissor test is enabled.
diff --git a/src/gallium/docs/source/tgsi.rst b/src/gallium/docs/source/tgsi.rst
index fc5482d384..c292cd37d5 100644
--- a/src/gallium/docs/source/tgsi.rst
+++ b/src/gallium/docs/source/tgsi.rst
@@ -1380,10 +1380,8 @@ TGSI_SEMANTIC_PSIZE
 """""""""""""""""""
 
 PSIZE, or point size, is used to specify point sizes per-vertex. It should
-be in ``(p, n, x, f)`` format, where ``p`` is the point size, ``n`` is the minimum
-size, ``x`` is the maximum size, and ``f`` is the fade threshold.
-
-XXX this is arb_vp. is this what we actually do? should double-check...
+be in ``(s, 0, 0, 1)`` format, where ``s`` is the (possibly clamped) point size.
+Only the first component matters when writing from the vertex shader.
 
 When using this semantic, be sure to set the appropriate state in the
 :ref:`rasterizer` first.
diff --git a/src/gallium/drivers/i965/brw_sf.c b/src/gallium/drivers/i965/brw_sf.c
index 6e99eaa09d..9cceb4dbe5 100644
--- a/src/gallium/drivers/i965/brw_sf.c
+++ b/src/gallium/drivers/i965/brw_sf.c
@@ -127,6 +127,7 @@ static enum pipe_error compile_sf_prog( struct brw_context *brw,
 static enum pipe_error upload_sf_prog(struct brw_context *brw)
 {
    const struct brw_fs_signature *sig = &brw->curr.fragment_shader->signature;
+   const struct pipe_rasterizer_state *rast = &brw->curr.rast->templ;
    struct brw_sf_prog_key key;
    enum pipe_error ret;
    unsigned i;
@@ -165,8 +166,8 @@ static enum pipe_error upload_sf_prog(struct brw_context *brw)
    case PIPE_PRIM_TRIANGLES: 
       /* PIPE_NEW_RAST
        */
-      if (brw->curr.rast->templ.fill_cw != PIPE_POLYGON_MODE_FILL ||
-	  brw->curr.rast->templ.fill_ccw != PIPE_POLYGON_MODE_FILL)
+      if (rast->fill_cw != PIPE_POLYGON_MODE_FILL ||
+	  rast->fill_ccw != PIPE_POLYGON_MODE_FILL)
 	 key.primitive = SF_UNFILLED_TRIS;
       else
 	 key.primitive = SF_TRIANGLES;
@@ -179,14 +180,14 @@ static enum pipe_error upload_sf_prog(struct brw_context *brw)
       break;
    }
 
-   key.do_point_sprite = brw->curr.rast->templ.point_sprite;
-   key.sprite_origin_lower_left = 0; /* XXX: ctx->Point.SpriteOrigin - fix rast state */
-   key.do_flat_shading = brw->curr.rast->templ.flatshade;
-   key.do_twoside_color = brw->curr.rast->templ.light_twoside;
+   key.do_point_sprite = rast->sprite_coord_enable ? 1 : 0;
+   key.sprite_origin_lower_left = (rast->sprite_coord_mode == PIPE_SPRITE_COORD_LOWER_LEFT);
+   key.point_coord_replace_attrs = rast->sprite_coord_enable;
+   key.do_flat_shading = rast->flatshade;
+   key.do_twoside_color = rast->light_twoside;
 
    if (key.do_twoside_color) {
-      key.frontface_ccw = (brw->curr.rast->templ.front_winding == 
-			   PIPE_WINDING_CCW);
+      key.frontface_ccw = (rast->front_winding == PIPE_WINDING_CCW);
    }
 
    if (brw_search_cache(&brw->cache, BRW_SF_PROG,
diff --git a/src/gallium/drivers/i965/brw_sf_state.c b/src/gallium/drivers/i965/brw_sf_state.c
index 25dc2b52e0..0ad91e0307 100644
--- a/src/gallium/drivers/i965/brw_sf_state.c
+++ b/src/gallium/drivers/i965/brw_sf_state.c
@@ -123,12 +123,10 @@ sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key)
    key->line_last_pixel_enable = rast->line_last_pixel;
    key->gl_rasterization_rules = rast->gl_rasterization_rules;
 
-   key->point_sprite = rast->point_sprite;
+   key->point_sprite = rast->sprite_coord_enable ? 1 : 0;
    key->point_attenuated = rast->point_size_per_vertex;
 
-   key->point_size = CLAMP(rast->point_size, 
-			   rast->point_size_min, 
-			   rast->point_size_max);
+   key->point_size = rast->point_size;
 }
 
 static enum pipe_error
diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c
new file mode 100644
index 0000000000..6f674d6e13
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_state.c
@@ -0,0 +1,590 @@
+#include "draw/draw_context.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+#include "pipe/p_inlines.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv10_context.h"
+#include "nv10_state.h"
+
+static void *
+nv10_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv10_blend_state *cb;
+
+	cb = MALLOC(sizeof(struct nv10_blend_state));
+
+	cb->b_enable = cso->blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_dst_factor)));
+
+	cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
+
+	cb->d_enable = cso->dither ? 1 : 0;
+
+	return (void *)cb;
+}
+
+static void
+nv10_blend_state_bind(struct pipe_context *pipe, void *blend)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->blend = (struct nv10_blend_state*)blend;
+
+	nv10->dirty |= NV10_NEW_BLEND;
+}
+
+static void
+nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT;
+		break;
+	}
+
+	return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT;
+}
+
+static void *
+nv10_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv10_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv10_sampler_state));
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT));
+
+	ps->en = 0;
+	if (cso->max_anisotropy > 1.0) {
+		/* no idea, binary driver sets it, works without it.. meh.. */
+		ps->wrap |= (1 << 5);
+
+/*		if (cso->max_anisotropy >= 16.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_16X;
+		} else
+		if (cso->max_anisotropy >= 12.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_12X;
+		} else
+		if (cso->max_anisotropy >= 10.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_10X;
+		} else
+		if (cso->max_anisotropy >= 8.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_8X;
+		} else
+		if (cso->max_anisotropy >= 6.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_6X;
+		} else
+		if (cso->max_anisotropy >= 4.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_4X;
+		} else {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_2X;
+		}*/
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+/*	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}*/
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv10->tex_sampler[unit] = sampler[unit];
+		nv10->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void
+nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit];
+		nv10->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void *
+nv10_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv10_rasterizer_state *rs;
+	int i;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	offset_cw/ccw -nohw
+	 * 	scissor
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 * 	offset_units / offset_scale
+	 */
+	rs = MALLOC(sizeof(struct nv10_rasterizer_state));
+
+	rs->templ = cso;
+	
+	rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01;
+
+	rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff;
+	rs->line_smooth_en = cso->line_smooth ? 1 : 0;
+
+	rs->point_size = *(uint32_t*)&cso->point_size;
+
+	rs->poly_smooth_en = cso->poly_smooth ? 1 : 0;
+
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		rs->front_face = NV10TCL_FRONT_FACE_CCW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_cw);
+	} else {
+		rs->front_face = NV10TCL_FRONT_FACE_CW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_ccw);
+	}
+
+	switch (cso->cull_mode) {
+	case PIPE_WINDING_CCW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CCW)
+			rs->cull_face    = NV10TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV10TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_CW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CW)
+			rs->cull_face    = NV10TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV10TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_BOTH:
+		rs->cull_face_en = 1;
+		rs->cull_face    = NV10TCL_CULL_FACE_FRONT_AND_BACK;
+		break;
+	case PIPE_WINDING_NONE:
+	default:
+		rs->cull_face_en = 0;
+		rs->cull_face    = 0;
+		break;
+	}
+
+	if (cso->sprite_coord_enable) {
+		rs->point_sprite = (1 << 0);
+		for (i = 0; i < 8; i++) {
+			if ((cso->sprite_coord_enable >> i) & 1)
+				rs->point_sprite |= (1 << (8 + i));
+		}
+	} else {
+		rs->point_sprite = 0;
+	}
+
+	return (void *)rs;
+}
+
+static void
+nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->rast = (struct nv10_rasterizer_state*)rast;
+
+	draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL));
+
+	nv10->dirty |= NV10_NEW_RAST;
+}
+
+static void
+nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv10_depth_stencil_alpha_state *hw;
+
+	hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state));
+
+	hw->depth.func		= nvgl_comparison_op(cso->depth.func);
+	hw->depth.write_enable	= cso->depth.writemask ? 1 : 0;
+	hw->depth.test_enable	= cso->depth.enabled ? 1 : 0;
+
+	hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0;
+	hw->stencil.wmask = cso->stencil[0].writemask;
+	hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func);
+	hw->stencil.ref	= cso->stencil[0].ref_value;
+	hw->stencil.vmask = cso->stencil[0].valuemask;
+	hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op);
+	hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op);
+	hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op);
+
+	hw->alpha.enabled = cso->alpha.enabled ? 1 : 0;
+	hw->alpha.func = nvgl_comparison_op(cso->alpha.func);
+	hw->alpha.ref  = float_to_ubyte(cso->alpha.ref_value);
+
+	return (void *)hw;
+}
+
+static void
+nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa;
+
+	nv10->dirty |= NV10_NEW_DSA;
+}
+
+static void
+nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv10_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *templ)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	return draw_create_vertex_shader(nv10->draw, templ);
+}
+
+static void
+nv10_vp_state_bind(struct pipe_context *pipe, void *shader)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader);
+
+	nv10->dirty |= NV10_NEW_VERTPROG;
+}
+
+static void
+nv10_vp_state_delete(struct pipe_context *pipe, void *shader)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader);
+}
+
+static void *
+nv10_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv10_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv10_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	
+	tgsi_scan_shader(cso->tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nv10_fragment_program *fp = hwcso;
+
+	nv10->fragprog.current = fp;
+	nv10->dirty |= NV10_NEW_FRAGPROG;
+}
+
+static void
+nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nv10_fragment_program *fp = hwcso;
+
+	nv10_fragprog_destroy(nv10, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv10_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->blend_color = (struct pipe_blend_color*)bcol;
+
+	nv10->dirty |= NV10_NEW_BLENDCOL;
+}
+
+static void
+nv10_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_set_clip_state(nv10->draw, clip);
+}
+
+static void
+nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct pipe_screen *pscreen = pipe->screen;
+
+	assert(shader < PIPE_SHADER_TYPES);
+	assert(index == 0);
+
+	if (buf) {
+		void *mapped;
+		if (buf->buffer && buf->buffer->size &&
+                    (mapped = pipe_buffer_map(pscreen, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ)))
+		{
+			memcpy(nv10->constbuf[shader], mapped, buf->buffer->size);
+			nv10->constbuf_nr[shader] =
+				buf->buffer->size / (4 * sizeof(float));
+			pipe_buffer_unmap(pscreen, buf->buffer);
+		}
+	}
+}
+
+static void
+nv10_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->framebuffer = (struct pipe_framebuffer_state*)fb;
+
+	nv10->dirty |= NV10_NEW_FRAMEBUFFER;
+}
+
+static void
+nv10_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	NOUVEAU_ERR("line stipple hahaha\n");
+}
+
+static void
+nv10_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->scissor = (struct pipe_scissor_state*)s;
+
+	nv10->dirty |= NV10_NEW_SCISSOR;
+}
+
+static void
+nv10_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->viewport = (struct pipe_viewport_state*)vpt;
+
+	draw_set_viewport_state(nv10->draw, nv10->viewport);
+
+	nv10->dirty |= NV10_NEW_VIEWPORT;
+}
+
+static void
+nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count);
+	nv10->dirty |= NV10_NEW_VTXARRAYS;
+
+	draw_set_vertex_buffers(nv10->draw, count, vb);
+}
+
+static void
+nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	memcpy(nv10->vtxelt, ve, sizeof(*ve) * count);
+	nv10->dirty |= NV10_NEW_VTXARRAYS;
+
+	draw_set_vertex_elements(nv10->draw, count, ve);
+}
+
+void
+nv10_init_state_functions(struct nv10_context *nv10)
+{
+	nv10->pipe.create_blend_state = nv10_blend_state_create;
+	nv10->pipe.bind_blend_state = nv10_blend_state_bind;
+	nv10->pipe.delete_blend_state = nv10_blend_state_delete;
+
+	nv10->pipe.create_sampler_state = nv10_sampler_state_create;
+	nv10->pipe.bind_fragment_sampler_states = nv10_sampler_state_bind;
+	nv10->pipe.delete_sampler_state = nv10_sampler_state_delete;
+	nv10->pipe.set_fragment_sampler_textures = nv10_set_sampler_texture;
+
+	nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create;
+	nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind;
+	nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete;
+
+	nv10->pipe.create_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_create;
+	nv10->pipe.bind_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_bind;
+	nv10->pipe.delete_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_delete;
+
+	nv10->pipe.create_vs_state = nv10_vp_state_create;
+	nv10->pipe.bind_vs_state = nv10_vp_state_bind;
+	nv10->pipe.delete_vs_state = nv10_vp_state_delete;
+
+	nv10->pipe.create_fs_state = nv10_fp_state_create;
+	nv10->pipe.bind_fs_state = nv10_fp_state_bind;
+	nv10->pipe.delete_fs_state = nv10_fp_state_delete;
+
+	nv10->pipe.set_blend_color = nv10_set_blend_color;
+	nv10->pipe.set_clip_state = nv10_set_clip_state;
+	nv10->pipe.set_constant_buffer = nv10_set_constant_buffer;
+	nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state;
+	nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple;
+	nv10->pipe.set_scissor_state = nv10_set_scissor_state;
+	nv10->pipe.set_viewport_state = nv10_set_viewport_state;
+
+	nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers;
+	nv10->pipe.set_vertex_elements = nv10_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c
new file mode 100644
index 0000000000..3d9a276fa1
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_state.c
@@ -0,0 +1,583 @@
+#include "draw/draw_context.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+#include "pipe/p_inlines.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv20_context.h"
+#include "nv20_state.h"
+
+static void *
+nv20_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv20_blend_state *cb;
+
+	cb = MALLOC(sizeof(struct nv20_blend_state));
+
+	cb->b_enable = cso->blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_dst_factor)));
+
+	cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
+
+	cb->d_enable = cso->dither ? 1 : 0;
+
+	return (void *)cb;
+}
+
+static void
+nv20_blend_state_bind(struct pipe_context *pipe, void *blend)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->blend = (struct nv20_blend_state*)blend;
+
+	nv20->dirty |= NV20_NEW_BLEND;
+}
+
+static void
+nv20_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV20TCL_TX_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV20TCL_TX_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV20TCL_TX_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV20TCL_TX_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV20TCL_TX_WRAP_S_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV20TCL_TX_WRAP_S_REPEAT;
+		break;
+	}
+
+	return (ret >> NV20TCL_TX_WRAP_S_SHIFT);
+}
+
+static void *
+nv20_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv20_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv20_sampler_state));
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV20TCL_TX_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV20TCL_TX_WRAP_T_SHIFT));
+
+	ps->en = 0;
+	if (cso->max_anisotropy > 1.0) {
+		/* no idea, binary driver sets it, works without it.. meh.. */
+		ps->wrap |= (1 << 5);
+
+/*		if (cso->max_anisotropy >= 8.0) {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_8X;
+		} else
+		if (cso->max_anisotropy >= 4.0) {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_4X;
+		} else {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_2X;
+		}*/
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV20TCL_TX_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV20TCL_TX_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV20TCL_TX_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV20TCL_TX_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+/*	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}*/
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv20_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv20->tex_sampler[unit] = sampler[unit];
+		nv20->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void
+nv20_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv20_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv20->tex_miptree[unit] = (struct nv20_miptree *)miptree[unit];
+		nv20->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void *
+nv20_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv20_rasterizer_state *rs;
+	int i;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	offset_cw/ccw -nohw
+	 * 	scissor
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 * 	offset_units / offset_scale
+	 */
+	rs = MALLOC(sizeof(struct nv20_rasterizer_state));
+
+	rs->templ = cso;
+	
+	rs->shade_model = cso->flatshade ? NV20TCL_SHADE_MODEL_FLAT :
+						NV20TCL_SHADE_MODEL_SMOOTH;
+
+	rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff;
+	rs->line_smooth_en = cso->line_smooth ? 1 : 0;
+
+	/* XXX: nv20 and nv25 different! */
+	rs->point_size = *(uint32_t*)&cso->point_size;
+
+	rs->poly_smooth_en = cso->poly_smooth ? 1 : 0;
+
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		rs->front_face = NV20TCL_FRONT_FACE_CCW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_cw);
+	} else {
+		rs->front_face = NV20TCL_FRONT_FACE_CW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_ccw);
+	}
+
+	switch (cso->cull_mode) {
+	case PIPE_WINDING_CCW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CCW)
+			rs->cull_face    = NV20TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV20TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_CW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CW)
+			rs->cull_face    = NV20TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV20TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_BOTH:
+		rs->cull_face_en = 1;
+		rs->cull_face    = NV20TCL_CULL_FACE_FRONT_AND_BACK;
+		break;
+	case PIPE_WINDING_NONE:
+	default:
+		rs->cull_face_en = 0;
+		rs->cull_face    = 0;
+		break;
+	}
+
+	if (cso->sprite_coord_enable) {
+		rs->point_sprite = (1 << 0);
+		for (i = 0; i < 8; i++) {
+			if ((cso->sprite_coord_enable >> i) & 1)
+				rs->point_sprite |= (1 << (8 + i));
+		}
+	} else {
+		rs->point_sprite = 0;
+	}
+
+	return (void *)rs;
+}
+
+static void
+nv20_rasterizer_state_bind(struct pipe_context *pipe, void *rast)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->rast = (struct nv20_rasterizer_state*)rast;
+
+	draw_set_rasterizer_state(nv20->draw, (nv20->rast ? nv20->rast->templ : NULL));
+
+	nv20->dirty |= NV20_NEW_RAST;
+}
+
+static void
+nv20_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv20_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv20_depth_stencil_alpha_state *hw;
+
+	hw = MALLOC(sizeof(struct nv20_depth_stencil_alpha_state));
+
+	hw->depth.func		= nvgl_comparison_op(cso->depth.func);
+	hw->depth.write_enable	= cso->depth.writemask ? 1 : 0;
+	hw->depth.test_enable	= cso->depth.enabled ? 1 : 0;
+
+	hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0;
+	hw->stencil.wmask = cso->stencil[0].writemask;
+	hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func);
+	hw->stencil.ref	= cso->stencil[0].ref_value;
+	hw->stencil.vmask = cso->stencil[0].valuemask;
+	hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op);
+	hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op);
+	hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op);
+
+	hw->alpha.enabled = cso->alpha.enabled ? 1 : 0;
+	hw->alpha.func = nvgl_comparison_op(cso->alpha.func);
+	hw->alpha.ref  = float_to_ubyte(cso->alpha.ref_value);
+
+	return (void *)hw;
+}
+
+static void
+nv20_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->dsa = (struct nv20_depth_stencil_alpha_state*)dsa;
+
+	nv20->dirty |= NV20_NEW_DSA;
+}
+
+static void
+nv20_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv20_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *templ)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	return draw_create_vertex_shader(nv20->draw, templ);
+}
+
+static void
+nv20_vp_state_bind(struct pipe_context *pipe, void *shader)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_bind_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader);
+
+	nv20->dirty |= NV20_NEW_VERTPROG;
+}
+
+static void
+nv20_vp_state_delete(struct pipe_context *pipe, void *shader)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_delete_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader);
+}
+
+static void *
+nv20_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv20_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv20_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	
+	tgsi_scan_shader(cso->tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv20_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nv20_fragment_program *fp = hwcso;
+
+	nv20->fragprog.current = fp;
+	nv20->dirty |= NV20_NEW_FRAGPROG;
+}
+
+static void
+nv20_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nv20_fragment_program *fp = hwcso;
+
+	nv20_fragprog_destroy(nv20, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv20_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->blend_color = (struct pipe_blend_color*)bcol;
+
+	nv20->dirty |= NV20_NEW_BLENDCOL;
+}
+
+static void
+nv20_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_set_clip_state(nv20->draw, clip);
+}
+
+static void
+nv20_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct pipe_screen *pscreen = pipe->screen;
+
+	assert(shader < PIPE_SHADER_TYPES);
+	assert(index == 0);
+
+	if (buf) {
+		void *mapped;
+		if (buf->buffer && buf->buffer->size &&
+                    (mapped = pipe_buffer_map(pscreen, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ)))
+		{
+			memcpy(nv20->constbuf[shader], mapped, buf->buffer->size);
+			nv20->constbuf_nr[shader] =
+				buf->buffer->size / (4 * sizeof(float));
+			pipe_buffer_unmap(pscreen, buf->buffer);
+		}
+	}
+}
+
+static void
+nv20_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->framebuffer = (struct pipe_framebuffer_state*)fb;
+
+	nv20->dirty |= NV20_NEW_FRAMEBUFFER;
+}
+
+static void
+nv20_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	NOUVEAU_ERR("line stipple hahaha\n");
+}
+
+static void
+nv20_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->scissor = (struct pipe_scissor_state*)s;
+
+	nv20->dirty |= NV20_NEW_SCISSOR;
+}
+
+static void
+nv20_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->viewport = (struct pipe_viewport_state*)vpt;
+
+	draw_set_viewport_state(nv20->draw, nv20->viewport);
+
+	nv20->dirty |= NV20_NEW_VIEWPORT;
+}
+
+static void
+nv20_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	memcpy(nv20->vtxbuf, vb, sizeof(*vb) * count);
+	nv20->dirty |= NV20_NEW_VTXARRAYS;
+
+	draw_set_vertex_buffers(nv20->draw, count, vb);
+}
+
+static void
+nv20_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	memcpy(nv20->vtxelt, ve, sizeof(*ve) * count);
+	nv20->dirty |= NV20_NEW_VTXARRAYS;
+
+	draw_set_vertex_elements(nv20->draw, count, ve);
+}
+
+void
+nv20_init_state_functions(struct nv20_context *nv20)
+{
+	nv20->pipe.create_blend_state = nv20_blend_state_create;
+	nv20->pipe.bind_blend_state = nv20_blend_state_bind;
+	nv20->pipe.delete_blend_state = nv20_blend_state_delete;
+
+	nv20->pipe.create_sampler_state = nv20_sampler_state_create;
+	nv20->pipe.bind_fragment_sampler_states = nv20_sampler_state_bind;
+	nv20->pipe.delete_sampler_state = nv20_sampler_state_delete;
+	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;
+	nv20->pipe.delete_rasterizer_state = nv20_rasterizer_state_delete;
+
+	nv20->pipe.create_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_create;
+	nv20->pipe.bind_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_bind;
+	nv20->pipe.delete_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_delete;
+
+	nv20->pipe.create_vs_state = nv20_vp_state_create;
+	nv20->pipe.bind_vs_state = nv20_vp_state_bind;
+	nv20->pipe.delete_vs_state = nv20_vp_state_delete;
+
+	nv20->pipe.create_fs_state = nv20_fp_state_create;
+	nv20->pipe.bind_fs_state = nv20_fp_state_bind;
+	nv20->pipe.delete_fs_state = nv20_fp_state_delete;
+
+	nv20->pipe.set_blend_color = nv20_set_blend_color;
+	nv20->pipe.set_clip_state = nv20_set_clip_state;
+	nv20->pipe.set_constant_buffer = nv20_set_constant_buffer;
+	nv20->pipe.set_framebuffer_state = nv20_set_framebuffer_state;
+	nv20->pipe.set_polygon_stipple = nv20_set_polygon_stipple;
+	nv20->pipe.set_scissor_state = nv20_set_scissor_state;
+	nv20->pipe.set_viewport_state = nv20_set_viewport_state;
+
+	nv20->pipe.set_vertex_buffers = nv20_set_vertex_buffers;
+	nv20->pipe.set_vertex_elements = nv20_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c
index 25532e3a54..f775938ba7 100644
--- a/src/gallium/drivers/nv30/nv30_state.c
+++ b/src/gallium/drivers/nv30/nv30_state.c
@@ -391,11 +391,11 @@ nv30_rasterizer_state_create(struct pipe_context *pipe,
 	}
 
 	so_method(so, rankine, NV34TCL_POINT_SPRITE, 1);
-	if (cso->point_sprite) {
+	if (cso->point_quad_rasterization) {
 		unsigned psctl = (1 << 0), i;
 
 		for (i = 0; i < 8; i++) {
-			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+			if ((cso->sprite_coord_enable >> i) & 1)
 				psctl |= (1 << (8 + i));
 		}
 
diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c
index b09c450dc2..51b40e51e4 100644
--- a/src/gallium/drivers/nv40/nv40_state.c
+++ b/src/gallium/drivers/nv40/nv40_state.c
@@ -401,11 +401,11 @@ nv40_rasterizer_state_create(struct pipe_context *pipe,
 	}
 
 	so_method(so, curie, NV40TCL_POINT_SPRITE, 1);
-	if (cso->point_sprite) {
+	if (cso->point_quad_rasterization) {
 		unsigned psctl = (1 << 0), i;
 
 		for (i = 0; i < 8; i++) {
-			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+			if ((cso->sprite_coord_enable >> i) & 1)
 				psctl |= (1 << (8 + i));
 		}
 
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index 3653e606ef..2372cbbef6 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -4406,16 +4406,14 @@ nv50_pntc_replace(struct nv50_context *nv50, uint32_t pntc[8], unsigned base)
 			    vp->cfg.out[j].si == fp->cfg.in[i].si)
 				break;
 
-		if (j < vp->cfg.out_nr) {
-			ubyte mode = nv50->rasterizer->pipe.sprite_coord_mode[
-				vp->cfg.out[j].si];
+		if (j < vp->info.num_outputs) {
+			ubyte enable =
+				 (nv50->rasterizer->pipe.sprite_coord_enable >> vp->cfg.out[j].si) & 1;
 
-			if (mode == PIPE_SPRITE_COORD_NONE) {
+			if (enable == 0) {
 				m += n;
 				continue;
-			} else
-			if (mode == PIPE_SPRITE_COORD_LOWER_LEFT)
-				origin = 0;
+			}
 		}
 
 		/* this is either PointCoord or replaced by sprite coords */
@@ -4426,7 +4424,7 @@ nv50_pntc_replace(struct nv50_context *nv50, uint32_t pntc[8], unsigned base)
 			++m;
 		}
 	}
-	return origin;
+	return (nv50->rasterizer->pipe.sprite_coord_mode == PIPE_SPRITE_COORD_LOWER_LEFT ? 0 : origin);
 }
 
 static int
@@ -4570,7 +4568,7 @@ nv50_fp_linkage_validate(struct nv50_context *nv50)
 	so_method(so, tesla, NV50TCL_NOPERSPECTIVE_BITMAP(0), 4);
 	so_datap (so, lin, 4);
 
-	if (nv50->rasterizer->pipe.point_sprite) {
+	if (nv50->rasterizer->pipe.sprite_coord_enable) {
 		so_method(so, tesla, NV50TCL_POINT_SPRITE_CTRL, 1);
 		so_data  (so,
 			  nv50_pntc_replace(nv50, pcrd, (reg[4] >> 8) & 0xff));
diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c
index 7eb326adf5..7c531b50a5 100644
--- a/src/gallium/drivers/nv50/nv50_state.c
+++ b/src/gallium/drivers/nv50/nv50_state.c
@@ -342,7 +342,7 @@ nv50_rasterizer_state_create(struct pipe_context *pipe,
 	so_data  (so, fui(cso->point_size));
 
 	so_method(so, tesla, NV50TCL_POINT_SPRITE_ENABLE, 1);
-	so_data  (so, cso->point_sprite);
+	so_data  (so, cso->point_quad_rasterization ? 1 : 0);
 
 	so_method(so, tesla, NV50TCL_POLYGON_MODE_FRONT, 3);
 	if (cso->front_winding == PIPE_WINDING_CCW) {
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 129eae1eae..d07e90860c 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -628,10 +628,12 @@ static void* r300_create_rs_state(struct pipe_context* pipe,
     rs->point_size = pack_float_16_6x(state->point_size) |
         (pack_float_16_6x(state->point_size) << R300_POINTSIZE_X_SHIFT);
 
-    rs->point_minmax =
-        ((int)(state->point_size_min * 6.0) <<
+        /* set hw limits - clamping done by state tracker in vs or point_size
+           XXX always need to emit this? */
+        rs->point_minmax =
+        ((int)(0.0 * 6.0) <<
          R300_GA_POINT_MINMAX_MIN_SHIFT) |
-        ((int)(state->point_size_max * 6.0) <<
+        ((int)(4096.0 * 6.0) <<
          R300_GA_POINT_MINMAX_MAX_SHIFT);
 
     rs->line_control = pack_float_16_6x(state->line_width) |
diff --git a/src/gallium/drivers/softpipe/sp_video_context.c b/src/gallium/drivers/softpipe/sp_video_context.c
index 47767f985d..d8b5b31e95 100644
--- a/src/gallium/drivers/softpipe/sp_video_context.c
+++ b/src/gallium/drivers/softpipe/sp_video_context.c
@@ -166,7 +166,7 @@ init_pipe_state(struct sp_mpeg12_context *ctx)
    rast.scissor = 0;
    rast.poly_smooth = 0;
    rast.poly_stipple_enable = 0;
-   rast.point_sprite = 0;
+   rast.sprite_coord_enable = 0;
    rast.point_size_per_vertex = 0;
    rast.multisample = 0;
    rast.line_smooth = 0;
@@ -177,10 +177,10 @@ init_pipe_state(struct sp_mpeg12_context *ctx)
    rast.bypass_vs_clip_and_viewport = 0;
    rast.line_width = 1;
    rast.point_smooth = 0;
+   rast.point_quad_rasterization = 0;
    rast.point_size = 1;
    rast.offset_units = 1;
    rast.offset_scale = 1;
-   /*rast.sprite_coord_mode[i] = ;*/
    ctx->rast = ctx->pipe->create_rasterizer_state(ctx->pipe, &rast);
    ctx->pipe->bind_rasterizer_state(ctx->pipe, ctx->rast);
 
diff --git a/src/gallium/drivers/svga/svga_context.h b/src/gallium/drivers/svga/svga_context.h
index e93acc5f9f..f9a641c6df 100644
--- a/src/gallium/drivers/svga/svga_context.h
+++ b/src/gallium/drivers/svga/svga_context.h
@@ -37,6 +37,7 @@
 
 
 #define SVGA_TEX_UNITS 8
+#define SVGA_MAX_POINTSIZE 80.0
 
 struct draw_vertex_shader;
 struct svga_shader_result;
@@ -145,8 +146,6 @@ struct svga_rasterizer_state {
    float slopescaledepthbias;
    float depthbias;
    float pointsize;
-   float pointsize_min;
-   float pointsize_max;
    
    unsigned hw_unfilled:16;         /* PIPE_POLYGON_MODE_x */
    unsigned need_pipeline:16;    /* which prims do we need help for? */
diff --git a/src/gallium/drivers/svga/svga_pipe_rasterizer.c b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
index 2f0b54092b..3571778867 100644
--- a/src/gallium/drivers/svga/svga_pipe_rasterizer.c
+++ b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
@@ -70,7 +70,8 @@ svga_create_rasterizer_state(struct pipe_context *pipe,
    /* light_twoside          - XXX: need fragment shader varient */
    /* poly_smooth            - XXX: no fallback available */
    /* poly_stipple_enable    - draw module */
-   /* point_sprite           - ? */
+   /* sprite_coord_enable    - ? */
+   /* point_quad_rasterization - ? */
    /* point_size_per_vertex  - ? */
    /* sprite_coord_mode      - ??? */
    /* bypass_vs_viewport_and_clip        - handled by viewport setup */
@@ -87,8 +88,6 @@ svga_create_rasterizer_state(struct pipe_context *pipe,
    rast->antialiasedlineenable = templ->line_smooth;
    rast->lastpixel = templ->line_last_pixel;
    rast->pointsize = templ->point_size;
-   rast->pointsize_min = templ->point_size_min;
-   rast->pointsize_max = templ->point_size_max;
    rast->hw_unfilled = PIPE_POLYGON_MODE_FILL;
 
    /* Use swtnl + decomposition implement these:
diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index 5e11163d18..8143be5024 100644
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -101,7 +101,7 @@ svga_get_paramf(struct pipe_screen *screen, int param)
       /* Keep this to a reasonable size to avoid failures in
        * conform/pntaa.c:
        */
-      return 80.0;
+      return SVGA_MAX_POINTSIZE;
 
    case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
       return 4.0;
diff --git a/src/gallium/drivers/svga/svga_state_rss.c b/src/gallium/drivers/svga/svga_state_rss.c
index ff656a7e05..5ce9b4ef4f 100644
--- a/src/gallium/drivers/svga/svga_state_rss.c
+++ b/src/gallium/drivers/svga/svga_state_rss.c
@@ -193,8 +193,9 @@ static int emit_rss( struct svga_context *svga,
       EMIT_RS( svga, curr->lastpixel, LASTPIXEL, fail );
       EMIT_RS( svga, curr->linepattern, LINEPATTERN, fail );
       EMIT_RS_FLOAT( svga, curr->pointsize, POINTSIZE, fail );
-      EMIT_RS_FLOAT( svga, curr->pointsize_min, POINTSIZEMIN, fail );
-      EMIT_RS_FLOAT( svga, curr->pointsize_max, POINTSIZEMAX, fail );
+      /* XXX still need to set this? */
+      EMIT_RS_FLOAT( svga, 0.0, POINTSIZEMIN, fail );
+      EMIT_RS_FLOAT( svga, SVGA_MAX_POINTSIZE, POINTSIZEMAX, fail );
    }
 
    if (dirty & (SVGA_NEW_RAST | SVGA_NEW_FRAME_BUFFER | SVGA_NEW_NEED_PIPELINE))
diff --git a/src/gallium/drivers/trace/tr_dump_state.c b/src/gallium/drivers/trace/tr_dump_state.c
index 16b24f1c8f..6648539a0f 100644
--- a/src/gallium/drivers/trace/tr_dump_state.c
+++ b/src/gallium/drivers/trace/tr_dump_state.c
@@ -112,7 +112,9 @@ void trace_dump_rasterizer_state(const struct pipe_rasterizer_state *state)
    trace_dump_member(bool, state, poly_smooth);
    trace_dump_member(bool, state, poly_stipple_enable);
    trace_dump_member(bool, state, point_smooth);
-   trace_dump_member(bool, state, point_sprite);
+   trace_dump_member(uint, state, sprite_coord_enable);
+   trace_dump_member(bool, state, sprite_coord_mode);
+   trace_dump_member(bool, state, point_quad_rasterization);
    trace_dump_member(bool, state, point_size_per_vertex);
    trace_dump_member(bool, state, multisample);
    trace_dump_member(bool, state, line_smooth);
@@ -126,13 +128,9 @@ void trace_dump_rasterizer_state(const struct pipe_rasterizer_state *state)
 
    trace_dump_member(float, state, line_width);
    trace_dump_member(float, state, point_size);
-   trace_dump_member(float, state, point_size_min);
-   trace_dump_member(float, state, point_size_max);
    trace_dump_member(float, state, offset_units);
    trace_dump_member(float, state, offset_scale);
 
-   trace_dump_member_array(uint, state, sprite_coord_mode);
-
    trace_dump_struct_end();
 }
 
diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h
index 63ba3119b8..5cebd43ace 100644
--- a/src/gallium/include/pipe/p_defines.h
+++ b/src/gallium/include/pipe/p_defines.h
@@ -364,9 +364,8 @@ enum pipe_transfer_usage {
 /**
  * Point sprite coord modes
  */
-#define PIPE_SPRITE_COORD_NONE       0
-#define PIPE_SPRITE_COORD_UPPER_LEFT 1
-#define PIPE_SPRITE_COORD_LOWER_LEFT 2
+#define PIPE_SPRITE_COORD_UPPER_LEFT 0
+#define PIPE_SPRITE_COORD_LOWER_LEFT 1
 
 
 /**
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index 9715862567..68369570b9 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -102,7 +102,9 @@ struct pipe_rasterizer_state
    unsigned poly_smooth:1;
    unsigned poly_stipple_enable:1;
    unsigned point_smooth:1;
-   unsigned point_sprite:1;
+   unsigned sprite_coord_enable:PIPE_MAX_SHADER_OUTPUTS;
+   unsigned sprite_coord_mode:1;     /**< PIPE_SPRITE_COORD_ */
+   unsigned point_quad_rasterization:1; /** points rasterized as quads or points */
    unsigned point_size_per_vertex:1; /**< size computed in vertex shader */
    unsigned multisample:1;         /* XXX maybe more ms state in future */
    unsigned line_smooth:1;
@@ -142,11 +144,8 @@ struct pipe_rasterizer_state
 
    float line_width;
    float point_size;           /**< used when no per-vertex size */
-   float point_size_min;        /* XXX - temporary, will go away */
-   float point_size_max;        /* XXX - temporary, will go away */
    float offset_units;
    float offset_scale;
-   ubyte sprite_coord_mode[PIPE_MAX_SHADER_OUTPUTS]; /**< PIPE_SPRITE_COORD_ */
 };
 
 
diff --git a/src/gallium/state_trackers/python/retrace/interpreter.py b/src/gallium/state_trackers/python/retrace/interpreter.py
index f9b330e5c7..190db43b08 100755
--- a/src/gallium/state_trackers/python/retrace/interpreter.py
+++ b/src/gallium/state_trackers/python/retrace/interpreter.py
@@ -113,7 +113,7 @@ struct_factories = {
 
 
 member_array_factories = {
-    "pipe_rasterizer_state": {"sprite_coord_mode": gallium.ByteArray},                          
+    #"pipe_rasterizer_state": {"sprite_coord_mode": gallium.ByteArray},                          
     "pipe_poly_stipple": {"stipple": gallium.UnsignedArray},                          
     "pipe_viewport_state": {"scale": gallium.FloatArray, "translate": gallium.FloatArray},                          
     #"pipe_clip_state": {"ucp": gallium.FloatArray},
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 2d1db29cbf..867a55242c 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1496,7 +1496,7 @@ static void build_texture_transform( struct tnl_program *p )
 static void build_atten_pointsize( struct tnl_program *p )
 {
    struct ureg eye = get_eye_position_z(p);
-   struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
+   struct ureg state_size = register_param2(p, STATE_INTERNAL, STATE_POINT_SIZE_CLAMPED);
    struct ureg state_attenuation = register_param1(p, STATE_POINT_ATTENUATION);
    struct ureg out = register_output(p, VERT_RESULT_PSIZ);
    struct ureg ut = get_temp(p);
diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c
index 20321dd01f..a0be1acfca 100644
--- a/src/mesa/shader/prog_statevars.c
+++ b/src/mesa/shader/prog_statevars.c
@@ -445,6 +445,61 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
          value[3] = (GLfloat)(ctx->Fog.Density * ONE_DIV_SQRT_LN2);
          return;
 
+      case STATE_POINT_SIZE_CLAMPED:
+         {
+           /* this includes implementation dependent limits, to avoid
+            * another potentially necessary clamp.
+            * Note: for sprites, point smooth (point AA) is ignored
+            * and we'll clamp to MinPointSizeAA and MaxPointSize, because we
+            * expect drivers will want to say their minimum for AA size is 0.0
+            * but for non-AA it's 1.0 (because normal points with size below 1.0
+            * need to get rounded up to 1.0, hence never disappear). GL does
+            * not specify max clamp size for sprites, other than it needs to be
+            * at least as large as max AA size, hence use non-AA size there.
+            */
+            GLfloat minImplSize;
+            GLfloat maxImplSize;
+            if (ctx->Point.PointSprite) {
+               minImplSize = ctx->Const.MinPointSizeAA;
+               maxImplSize = ctx->Const.MaxPointSize;
+            }
+            else if (ctx->Point.SmoothFlag || ctx->Multisample._Enabled) {
+               minImplSize = ctx->Const.MinPointSizeAA;
+               maxImplSize = ctx->Const.MaxPointSizeAA;
+            }
+            else {
+               minImplSize = ctx->Const.MinPointSize;
+               maxImplSize = ctx->Const.MaxPointSize;
+            }
+            value[0] = ctx->Point.Size;
+            value[1] = ctx->Point.MinSize >= minImplSize ? ctx->Point.MinSize : minImplSize;
+            value[2] = ctx->Point.MaxSize <= maxImplSize ? ctx->Point.MaxSize : maxImplSize;
+            value[3] = ctx->Point.Threshold;
+         }
+         return;
+      case STATE_POINT_SIZE_IMPL_CLAMP:
+         {
+           /* for implementation clamp only in vs */
+            GLfloat minImplSize;
+            GLfloat maxImplSize;
+            if (ctx->Point.PointSprite) {
+               minImplSize = ctx->Const.MinPointSizeAA;
+               maxImplSize = ctx->Const.MaxPointSize;
+            }
+            else if (ctx->Point.SmoothFlag || ctx->Multisample._Enabled) {
+               minImplSize = ctx->Const.MinPointSizeAA;
+               maxImplSize = ctx->Const.MaxPointSizeAA;
+            }
+            else {
+               minImplSize = ctx->Const.MinPointSize;
+               maxImplSize = ctx->Const.MaxPointSize;
+            }
+            value[0] = ctx->Point.Size;
+            value[1] = minImplSize;
+            value[2] = maxImplSize;
+            value[3] = ctx->Point.Threshold;
+         }
+         return;
       case STATE_LIGHT_SPOT_DIR_NORMALIZED:
          {
             /* here, state[2] is the light number */
@@ -640,6 +695,9 @@ _mesa_program_state_flags(const gl_state_index state[STATE_LENGTH])
 	 return _NEW_TEXTURE;
       case STATE_FOG_PARAMS_OPTIMIZED:
 	 return _NEW_FOG;
+      case STATE_POINT_SIZE_CLAMPED:
+      case STATE_POINT_SIZE_IMPL_CLAMP:
+         return _NEW_POINT | _NEW_MULTISAMPLE;
       case STATE_LIGHT_SPOT_DIR_NORMALIZED:
       case STATE_LIGHT_POSITION:
       case STATE_LIGHT_POSITION_NORMALIZED:
@@ -831,6 +889,12 @@ append_token(char *dst, gl_state_index k)
    case STATE_FOG_PARAMS_OPTIMIZED:
       append(dst, "fogParamsOptimized");
       break;
+   case STATE_POINT_SIZE_CLAMPED:
+      append(dst, "pointSizeClamped");
+      break;
+   case STATE_POINT_SIZE_IMPL_CLAMP:
+      append(dst, "pointSizeImplClamp");
+      break;
    case STATE_LIGHT_SPOT_DIR_NORMALIZED:
       append(dst, "lightSpotDirNormalized");
       break;
diff --git a/src/mesa/shader/prog_statevars.h b/src/mesa/shader/prog_statevars.h
index 1180d9eaa4..1753471ffb 100644
--- a/src/mesa/shader/prog_statevars.h
+++ b/src/mesa/shader/prog_statevars.h
@@ -108,6 +108,8 @@ typedef enum gl_state_index_ {
    STATE_NORMAL_SCALE,
    STATE_TEXRECT_SCALE,
    STATE_FOG_PARAMS_OPTIMIZED,  /* for faster fog calc */
+   STATE_POINT_SIZE_CLAMPED,    /* includes implementation dependent size clamp */
+   STATE_POINT_SIZE_IMPL_CLAMP, /* for implementation clamp only in vs */
    STATE_LIGHT_SPOT_DIR_NORMALIZED,   /* pre-normalized spot dir */
    STATE_LIGHT_POSITION,              /* object vs eye space */
    STATE_LIGHT_POSITION_NORMALIZED,   /* object vs eye space */
diff --git a/src/mesa/state_tracker/st_atom_rasterizer.c b/src/mesa/state_tracker/st_atom_rasterizer.c
index 36b28cb4df..9c9a99bcfc 100644
--- a/src/mesa/state_tracker/st_atom_rasterizer.c
+++ b/src/mesa/state_tracker/st_atom_rasterizer.c
@@ -183,27 +183,26 @@ static void update_raster_state( struct st_context *st )
 
    if (ctx->Polygon.StippleFlag)
       raster->poly_stipple_enable = 1;
-     
+
    /* _NEW_POINT
     */
    raster->point_size = ctx->Point.Size;
 
-   raster->point_size_min = 0;    /* temporary, will go away */
-   raster->point_size_max = 1000; /* temporary, will go away */
+   if (!ctx->Point.PointSprite && ctx->Point.SmoothFlag)
+      raster->point_smooth = 1;
 
-   raster->point_smooth = ctx->Point.SmoothFlag;
-   raster->point_sprite = ctx->Point.PointSprite;
-   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
-      if (ctx->Point.CoordReplace[i]) {
-         if ((ctx->Point.SpriteOrigin == GL_UPPER_LEFT) ^
-             (st_fb_orientation(ctx->DrawBuffer) == Y_0_BOTTOM))
-            raster->sprite_coord_mode[i] = PIPE_SPRITE_COORD_UPPER_LEFT;
-         else 
-            raster->sprite_coord_mode[i] = PIPE_SPRITE_COORD_LOWER_LEFT;
-      }
-      else {
-         raster->sprite_coord_mode[i] = PIPE_SPRITE_COORD_NONE;
+   if (ctx->Point.PointSprite) {
+      if ((ctx->Point.SpriteOrigin == GL_UPPER_LEFT) ^
+          (st_fb_orientation(ctx->DrawBuffer) == Y_0_BOTTOM))
+         raster->sprite_coord_mode = PIPE_SPRITE_COORD_UPPER_LEFT;
+      else 
+         raster->sprite_coord_mode = PIPE_SPRITE_COORD_LOWER_LEFT;
+      for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
+         if (ctx->Point.CoordReplace[i]) {
+            raster->sprite_coord_enable |= 1 << i;
+         }
       }
+      raster->point_quad_rasterization = 1;
    }
 
    /* ST_NEW_VERTEX_PROGRAM
diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c
index f2a62f9b69..d5f5854661 100644
--- a/src/mesa/state_tracker/st_extensions.c
+++ b/src/mesa/state_tracker/st_extensions.c
@@ -113,6 +113,15 @@ void st_init_limits(struct st_context *st)
       = _maxf(1.0f, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH));
    c->MaxPointSizeAA
       = _maxf(1.0f, screen->get_paramf(screen, PIPE_CAP_MAX_POINT_WIDTH_AA));
+   /* called after _mesa_create_context/_mesa_init_point, fix default user
+    * settable max point size up
+    */
+   st->ctx->Point.MaxSize = MAX2(c->MaxPointSize, c->MaxPointSizeAA);
+   /* these are not queryable. Note that GL basically mandates a 1.0 minimum
+    * for non-aa sizes, but we can go down to 0.0 for aa points.
+    */
+   c->MinPointSize = 1.0f;
+   c->MinPointSizeAA = 0.0f;
 
    c->MaxTextureMaxAnisotropy
       = _maxf(2.0f, screen->get_paramf(screen, PIPE_CAP_MAX_TEXTURE_ANISOTROPY));
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 06cf6d21c6..537a6a8648 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -62,6 +62,10 @@ struct st_translate {
    struct ureg_src inputs[PIPE_MAX_SHADER_INPUTS];
    struct ureg_dst address[1];
    struct ureg_src samplers[PIPE_MAX_SAMPLERS];
+   struct ureg_dst psizregreal;
+   struct ureg_src pointSizeConst;
+   GLint psizoutindex;
+   GLboolean prevInstWrotePsiz;
 
    const GLuint *inputMapping;
    const GLuint *outputMapping;
@@ -150,6 +154,8 @@ dst_register( struct st_translate *t,
       return t->temps[index];
 
    case PROGRAM_OUTPUT:
+      if (index == t->psizoutindex)
+         t->prevInstWrotePsiz = GL_TRUE;
       return t->outputs[t->outputMapping[index]];
 
    case PROGRAM_ADDRESS:
@@ -816,6 +822,8 @@ st_translate_mesa_program(
    t->inputMapping = inputMapping;
    t->outputMapping = outputMapping;
    t->ureg = ureg;
+   t->psizoutindex = -1;
+   t->prevInstWrotePsiz = GL_FALSE;
 
    /*_mesa_print_program(program);*/
 
@@ -928,6 +936,21 @@ st_translate_mesa_program(
          t->outputs[i] = ureg_DECL_output( ureg,
                                            outputSemanticName[i],
                                            outputSemanticIndex[i] );
+         if ((outputSemanticName[i] == TGSI_SEMANTIC_PSIZE) && program->Id) {
+            static const gl_state_index pointSizeClampState[STATE_LENGTH]
+               = { STATE_INTERNAL, STATE_POINT_SIZE_IMPL_CLAMP, 0, 0, 0 };
+               /* XXX: note we are modifying the incoming shader here!  Need to
+               * do this before emitting the constant decls below, or this
+               * will be missed:
+               */
+            unsigned pointSizeClampConst = _mesa_add_state_reference(program->Parameters,
+                                                                     pointSizeClampState);
+            struct ureg_dst psizregtemp = ureg_DECL_temporary( ureg );
+            t->pointSizeConst = ureg_DECL_constant( ureg, pointSizeClampConst );
+            t->psizregreal = t->outputs[i];
+            t->psizoutindex = i;
+            t->outputs[i] = psizregtemp;
+         }
       }
       if (passthrough_edgeflags)
          emit_edgeflags( t, program );
@@ -995,6 +1018,19 @@ st_translate_mesa_program(
    for (i = 0; i < program->NumInstructions; i++) {
       set_insn_start( t, ureg_get_instruction_number( ureg ));
       compile_instruction( t, &program->Instructions[i] );
+
+      /* note can't do that easily at the end of prog due to
+         possible early return */
+      if (t->prevInstWrotePsiz && program->Id) {
+         set_insn_start( t, ureg_get_instruction_number( ureg ));
+         ureg_MAX( t->ureg, ureg_writemask(t->outputs[t->psizoutindex], WRITEMASK_X),
+                   ureg_src(t->outputs[t->psizoutindex]),
+                   ureg_swizzle(t->pointSizeConst, 1,1,1,1));
+         ureg_MIN( t->ureg, ureg_writemask(t->psizregreal, WRITEMASK_X),
+                   ureg_src(t->outputs[t->psizoutindex]),
+                   ureg_swizzle(t->pointSizeConst, 2,2,2,2));
+      }
+      t->prevInstWrotePsiz = GL_FALSE;
    }
 
    /* Fix up all emitted labels: