From eedebfa5793844728f1d41a6b0a899825eba176a Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 16 Jan 2010 17:45:34 -0800
Subject: r300g: Respect fill modes in indices fallback.

---
 src/gallium/drivers/r300/r300_render.c | 17 +++++++++++++++--
 src/gallium/drivers/r300/r300_state.c  |  3 ---
 2 files changed, 15 insertions(+), 5 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 710d850163..64083f2dd5 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -257,10 +257,23 @@ static struct pipe_buffer* r300_translate_elts(struct r300_context* r300,
     struct pipe_buffer* new_elts;
     void *in_map, *out_map;
     unsigned out_prim, out_index_size, out_nr;
+    struct pipe_rasterizer_state* rs;
     u_translate_func out_translate;
 
-    (void)u_index_translator(~0, *mode, *size, *count, PV_LAST, PV_LAST,
-        &out_prim, &out_index_size, &out_nr, &out_translate);
+    rs = &((struct r300_rs_state*)r300->rs_state.state)->rs;
+
+    if (rs->fill_cw == rs->fill_ccw &&
+        rs->fill_cw != PIPE_POLYGON_MODE_FILL) {
+        (void)u_unfilled_translator(*mode, *size, *count, rs->fill_cw,
+            &out_prim, &out_index_size, &out_nr, &out_translate);
+    } else {
+        (void)u_index_translator(~0, *mode, *size, *count, PV_LAST, PV_LAST,
+            &out_prim, &out_index_size, &out_nr, &out_translate);
+    }
+
+    debug_printf("r300: old mode %d, new mode %d\n", *mode, out_prim);
+    debug_printf("r300: old count %d, new count %d\n", *count, out_nr);
+    debug_printf("r300: old size %d, new size %d\n", *size, out_index_size);
 
     new_elts = screen->buffer_create(screen, 32,
                                      PIPE_BUFFER_USAGE_INDEX |
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 60ad763cf4..609b910e12 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -602,9 +602,6 @@ static void* r300_create_rs_state(struct pipe_context* pipe,
     rs->line_control = pack_float_16_6x(state->line_width) |
         R300_GA_LINE_CNTL_END_TYPE_COMP;
 
-    /* XXX I think there is something wrong with the polygon mode,
-     * XXX re-test when r300g is in a better shape */
-
     /* Enable polygon mode */
     if (state->fill_cw != PIPE_POLYGON_MODE_FILL ||
         state->fill_ccw != PIPE_POLYGON_MODE_FILL) {
-- 
cgit v1.2.3


From c69bd0ead755b695ee5c64dc9efd45c53b299f21 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 16 Jan 2010 18:16:26 -0800
Subject: r300g: Stop using indices.

Sorry, it was a good game, but it just doesn't work.
---
 src/gallium/drivers/r300/r300_render.c | 54 +++++++++++-----------------------
 1 file changed, 17 insertions(+), 37 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 64083f2dd5..90de062bcd 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -26,8 +26,6 @@
 #include "draw/draw_context.h"
 #include "draw/draw_vbuf.h"
 
-#include "indices/u_indices.h"
-
 #include "pipe/p_inlines.h"
 
 #include "util/u_memory.h"
@@ -247,53 +245,35 @@ validate:
     return TRUE;
 }
 
-static struct pipe_buffer* r300_translate_elts(struct r300_context* r300,
-                                               struct pipe_buffer* elts,
-                                               unsigned* size,
-                                               unsigned* mode,
-                                               unsigned* count)
+static void r300_shorten_ubyte_elts(struct r300_context* r300,
+                                    struct pipe_buffer** elts,
+                                    unsigned count)
 {
     struct pipe_screen* screen = r300->context.screen;
     struct pipe_buffer* new_elts;
-    void *in_map, *out_map;
-    unsigned out_prim, out_index_size, out_nr;
-    struct pipe_rasterizer_state* rs;
-    u_translate_func out_translate;
-
-    rs = &((struct r300_rs_state*)r300->rs_state.state)->rs;
-
-    if (rs->fill_cw == rs->fill_ccw &&
-        rs->fill_cw != PIPE_POLYGON_MODE_FILL) {
-        (void)u_unfilled_translator(*mode, *size, *count, rs->fill_cw,
-            &out_prim, &out_index_size, &out_nr, &out_translate);
-    } else {
-        (void)u_index_translator(~0, *mode, *size, *count, PV_LAST, PV_LAST,
-            &out_prim, &out_index_size, &out_nr, &out_translate);
-    }
-
-    debug_printf("r300: old mode %d, new mode %d\n", *mode, out_prim);
-    debug_printf("r300: old count %d, new count %d\n", *count, out_nr);
-    debug_printf("r300: old size %d, new size %d\n", *size, out_index_size);
+    unsigned char *in_map;
+    unsigned short *out_map;
+    unsigned i;
 
     new_elts = screen->buffer_create(screen, 32,
                                      PIPE_BUFFER_USAGE_INDEX |
                                      PIPE_BUFFER_USAGE_CPU_WRITE |
                                      PIPE_BUFFER_USAGE_GPU_READ,
-                                     out_index_size * out_nr);
+                                     2 * count);
 
-    in_map = pipe_buffer_map(screen, elts, PIPE_BUFFER_USAGE_CPU_READ);
+    in_map = pipe_buffer_map(screen, *elts, PIPE_BUFFER_USAGE_CPU_READ);
     out_map = pipe_buffer_map(screen, new_elts, PIPE_BUFFER_USAGE_CPU_WRITE);
 
-    out_translate(in_map, *count, out_map);
+    for (i = 0; i < count; i++) {
+        *out_map = (unsigned short)*in_map;
+        in_map++;
+        out_map++;
+    }
 
-    pipe_buffer_unmap(screen, elts);
+    pipe_buffer_unmap(screen, *elts);
     pipe_buffer_unmap(screen, new_elts);
 
-    *size = out_index_size;
-    *mode = out_prim;
-    *count = out_nr;
-
-    return new_elts;
+    *elts = new_elts;
 }
 
 /* This is the fast-path drawing & emission for HW TCL. */
@@ -327,8 +307,8 @@ void r300_draw_range_elements(struct pipe_context* pipe,
     }
 
     if (indexSize == 1) {
-        indexBuffer = r300_translate_elts(r300, indexBuffer,
-            &indexSize, &mode, &count);
+        r300_shorten_ubyte_elts(r300, &indexBuffer, count);
+        indexSize = 2;
     }
 
     if (!r300->winsys->add_buffer(r300->winsys, indexBuffer,
-- 
cgit v1.2.3


From 37897a65e27842bbbfe32c37d2fcf775a5c127eb Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sat, 16 Jan 2010 00:56:33 +0100
Subject: r300g: fix the CS size when emitting invariant state

---
 src/gallium/drivers/r300/r300_state_invariant.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
index b0f309695c..47d7e60a40 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.c
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -79,7 +79,8 @@ void r300_emit_invariant_state(struct r300_context* r300)
     END_CS;
 
     /* XXX unsorted stuff from surface_fill */
-    BEGIN_CS(44 + (caps->has_tcl ? 7 : 0) + (caps->is_r500 ? 4 : 0));
+    BEGIN_CS(44 + (caps->has_tcl ? 7 : 0) +
+             (caps->family >= CHIP_FAMILY_RV350 ? 4 : 0));
 
     if (caps->has_tcl) {
         /*Flushing PVS is required before the VAP_GB registers can be changed*/
-- 
cgit v1.2.3


From 21f3bbf1495ce44ecde4f353fed9f85e6e0fdd2d Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sat, 16 Jan 2010 00:48:53 +0100
Subject: r300g: disable the MIP filter for NPOT textures

Safety first.
---
 src/gallium/drivers/r300/r300_emit.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 9f93327e59..11bd3346c7 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -705,12 +705,18 @@ void r300_emit_texture(struct r300_context* r300,
         filter0 |= R300_TX_WRAP_T(R300_TX_CLAMP_TO_EDGE);
     }
 
-    /* determine min/max levels */
-    /* the MAX_MIP level is the largest (finest) one */
-    max_level = MIN2(sampler->max_lod, tex->tex.last_level);
-    min_level = MIN2(sampler->min_lod, max_level);
-    format0 |= R300_TX_NUM_LEVELS(max_level);
-    filter0 |= R300_TX_MAX_MIP_LEVEL(min_level);
+    if (tex->is_npot) {
+        /* NPOT textures don't support mip filter, unfortunately.
+         * This prevents incorrect rendering. */
+        filter0 &= ~R300_TX_MIN_FILTER_MIP_MASK;
+    } else {
+        /* determine min/max levels */
+        /* the MAX_MIP level is the largest (finest) one */
+        max_level = MIN2(sampler->max_lod, tex->tex.last_level);
+        min_level = MIN2(sampler->min_lod, max_level);
+        format0 |= R300_TX_NUM_LEVELS(max_level);
+        filter0 |= R300_TX_MAX_MIP_LEVEL(min_level);
+    }
 
     BEGIN_CS(16);
     OUT_CS_REG(R300_TX_FILTER0_0 + (offset * 4), filter0 |
-- 
cgit v1.2.3


From 08edbe9f14712f719f838769c61123f83768af76 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sat, 9 Jan 2010 18:38:54 +0100
Subject: r300g: move debug flags into r300_screen

---
 src/gallium/drivers/r300/r300_context.c |  2 --
 src/gallium/drivers/r300/r300_context.h | 39 ++++++++----------------------
 src/gallium/drivers/r300/r300_debug.c   |  8 +++---
 src/gallium/drivers/r300/r300_screen.c  |  1 +
 src/gallium/drivers/r300/r300_screen.h  | 43 +++++++++++++++++++++++++++++++++
 5 files changed, 58 insertions(+), 35 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 5e4f6552c3..6bd2b0278f 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -153,8 +153,6 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     r300->context.winsys = (struct pipe_winsys*)radeon_winsys;
     r300->context.screen = screen;
 
-    r300_init_debug(r300);
-
     r300->context.destroy = r300_destroy_context;
 
     r300->context.clear = r300_clear;
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index 682b9179c8..c65d23823c 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -30,6 +30,8 @@
 #include "pipe/p_context.h"
 #include "pipe/p_inlines.h"
 
+#include "r300_screen.h"
+
 struct r300_context;
 
 struct r300_fragment_shader;
@@ -324,9 +326,6 @@ struct r300_context {
     uint32_t dirty_hw;
     /* Whether the TCL engine should be in bypass mode. */
     boolean tcl_bypass;
-
-    /** Combination of DBG_xxx flags */
-    unsigned debug;
 };
 
 /* Convenience cast wrapper. */
@@ -340,35 +339,15 @@ struct draw_stage* r300_draw_stage(struct r300_context* r300);
 void r300_init_state_functions(struct r300_context* r300);
 void r300_init_surface_functions(struct r300_context* r300);
 
-/* Debug functionality. */
-
-/**
- * Debug flags to disable/enable certain groups of debugging outputs.
- *
- * \note These may be rather coarse, and the grouping may be impractical.
- * If you find, while debugging the driver, that a different grouping
- * of these flags would be beneficial, just feel free to change them
- * but make sure to update the documentation in r300_debug.c to reflect
- * those changes.
- */
-/*@{*/
-#define DBG_HELP    0x0000001
-#define DBG_FP      0x0000002
-#define DBG_VP      0x0000004
-#define DBG_CS      0x0000008
-#define DBG_DRAW    0x0000010
-#define DBG_TEX     0x0000020
-#define DBG_FALL    0x0000040
-/*@}*/
-
-static INLINE boolean DBG_ON(struct r300_context * ctx, unsigned flags)
+static INLINE boolean CTX_DBG_ON(struct r300_context * ctx, unsigned flags)
 {
-    return (ctx->debug & flags) ? TRUE : FALSE;
+    return SCREEN_DBG_ON(r300_screen(ctx->context.screen), flags);
 }
 
-static INLINE void DBG(struct r300_context * ctx, unsigned flags, const char * fmt, ...)
+static INLINE void CTX_DBG(struct r300_context * ctx, unsigned flags,
+                       const char * fmt, ...)
 {
-    if (DBG_ON(ctx, flags)) {
+    if (CTX_DBG_ON(ctx, flags)) {
         va_list va;
         va_start(va, fmt);
         debug_vprintf(fmt, va);
@@ -376,6 +355,8 @@ static INLINE void DBG(struct r300_context * ctx, unsigned flags, const char * f
     }
 }
 
-void r300_init_debug(struct r300_context * ctx);
+#define DBG_ON  CTX_DBG_ON
+#define DBG     CTX_DBG
 
 #endif /* R300_CONTEXT_H */
+
diff --git a/src/gallium/drivers/r300/r300_debug.c b/src/gallium/drivers/r300/r300_debug.c
index 2a6ed54ac9..00d4f31c2b 100644
--- a/src/gallium/drivers/r300/r300_debug.c
+++ b/src/gallium/drivers/r300/r300_debug.c
@@ -46,7 +46,7 @@ static struct debug_option debug_options[] = {
     { 0, 0, 0 }
 };
 
-void r300_init_debug(struct r300_context * ctx)
+void r300_init_debug(struct r300_screen * screen)
 {
     const char * options = debug_get_option("RADEON_DEBUG", 0);
     boolean printhint = FALSE;
@@ -64,7 +64,7 @@ void r300_init_debug(struct r300_context * ctx)
 
             for(opt = debug_options; opt->name; ++opt) {
                 if (!strncmp(options, opt->name, length)) {
-                    ctx->debug |= opt->flag;
+                    screen->debug |= opt->flag;
                     break;
                 }
             }
@@ -77,11 +77,11 @@ void r300_init_debug(struct r300_context * ctx)
             options += length;
         }
 
-        if (!ctx->debug)
+        if (!screen->debug)
             printhint = TRUE;
     }
 
-    if (printhint || ctx->debug & DBG_HELP) {
+    if (printhint || screen->debug & DBG_HELP) {
         debug_printf("You can enable debug output by setting the RADEON_DEBUG environment variable\n"
                      "to a comma-separated list of debug options. Available options are:\n");
         for(opt = debug_options; opt->name; ++opt) {
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 287664b1d2..e16d659ec8 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -396,6 +396,7 @@ struct pipe_screen* r300_create_screen(struct radeon_winsys* radeon_winsys)
     caps->num_frag_pipes = radeon_winsys->gb_pipes;
     caps->num_z_pipes = radeon_winsys->z_pipes;
 
+    r300_init_debug(r300screen);
     r300_parse_chipset(caps);
 
     r300screen->caps = caps;
diff --git a/src/gallium/drivers/r300/r300_screen.h b/src/gallium/drivers/r300/r300_screen.h
index 2217988add..580fda3984 100644
--- a/src/gallium/drivers/r300/r300_screen.h
+++ b/src/gallium/drivers/r300/r300_screen.h
@@ -35,6 +35,9 @@ struct r300_screen {
 
     /* Chipset capabilities */
     struct r300_capabilities* caps;
+
+    /** Combination of DBG_xxx flags */
+    unsigned debug;
 };
 
 struct r300_transfer {
@@ -60,4 +63,44 @@ r300_transfer(struct pipe_transfer* transfer)
 /* Creates a new r300 screen. */
 struct pipe_screen* r300_create_screen(struct radeon_winsys* radeon_winsys);
 
+/* Debug functionality. */
+
+/**
+ * Debug flags to disable/enable certain groups of debugging outputs.
+ *
+ * \note These may be rather coarse, and the grouping may be impractical.
+ * If you find, while debugging the driver, that a different grouping
+ * of these flags would be beneficial, just feel free to change them
+ * but make sure to update the documentation in r300_debug.c to reflect
+ * those changes.
+ */
+/*@{*/
+#define DBG_HELP    0x0000001
+#define DBG_FP      0x0000002
+#define DBG_VP      0x0000004
+#define DBG_CS      0x0000008
+#define DBG_DRAW    0x0000010
+#define DBG_TEX     0x0000020
+#define DBG_FALL    0x0000040
+/*@}*/
+
+static INLINE boolean SCREEN_DBG_ON(struct r300_screen * screen, unsigned flags)
+{
+    return (screen->debug & flags) ? TRUE : FALSE;
+}
+
+static INLINE void SCREEN_DBG(struct r300_screen * screen, unsigned flags,
+                              const char * fmt, ...)
+{
+    if (SCREEN_DBG_ON(screen, flags)) {
+        va_list va;
+        va_start(va, fmt);
+        debug_vprintf(fmt, va);
+        va_end(va);
+    }
+}
+
+void r300_init_debug(struct r300_screen* ctx);
+
 #endif /* R300_SCREEN_H */
+
-- 
cgit v1.2.3


From 8959608051ef8e4a97214323009d09c5e9e633fb Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sat, 9 Jan 2010 19:19:44 +0100
Subject: r300g: output texture debug messages if only RADEON_DEBUG=tex is set

---
 src/gallium/drivers/r300/r300_context.c         |  2 +-
 src/gallium/drivers/r300/r300_screen.c          |  6 ++--
 src/gallium/drivers/r300/r300_texture.c         | 37 +++++++++++++++----------
 src/gallium/drivers/r300/r300_texture.h         |  6 ++--
 src/gallium/drivers/r300/r300_winsys.h          |  3 +-
 src/gallium/winsys/drm/radeon/core/radeon_drm.c |  7 +++--
 src/gallium/winsys/drm/radeon/core/radeon_drm.h |  1 +
 7 files changed, 36 insertions(+), 26 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 6bd2b0278f..94a9ab3ef3 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -87,7 +87,7 @@ r300_is_texture_referenced(struct pipe_context *pipe,
 {
     struct pipe_buffer* buf = 0;
 
-    r300_get_texture_buffer(texture, &buf, NULL);
+    r300_get_texture_buffer(pipe->screen, texture, &buf, NULL);
 
     return pipe->is_buffer_referenced(pipe, buf);
 }
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index e16d659ec8..d0f2c13df1 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -319,6 +319,7 @@ r300_get_tex_transfer(struct pipe_screen *screen,
 {
     struct r300_texture *tex = (struct r300_texture *)texture;
     struct r300_transfer *trans;
+    struct r300_screen *rscreen = r300_screen(screen);
     unsigned offset;
 
     offset = r300_texture_get_offset(tex, level, zslice, face);  /* in bytes */
@@ -330,11 +331,8 @@ r300_get_tex_transfer(struct pipe_screen *screen,
         trans->transfer.y = y;
         trans->transfer.width = w;
         trans->transfer.height = h;
-        trans->transfer.stride = r300_texture_get_stride(tex, level);
+        trans->transfer.stride = r300_texture_get_stride(rscreen, tex, level);
         trans->transfer.usage = usage;
-
-        /* XXX not sure whether it's required to set these two,
-               the driver doesn't use them */
         trans->transfer.zslice = zslice;
         trans->transfer.face = face;
 
diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index a9bbdd56d8..c6dd27272b 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -42,10 +42,11 @@ static const unsigned microblock_table[5][3][2] = {
     {{ 2, 1}, {0, 0}, {0, 0}}  /* 128 bits per pixel */
 };
 
-static void r300_setup_texture_state(struct r300_texture* tex, boolean is_r500)
+static void r300_setup_texture_state(struct r300_screen* screen, struct r300_texture* tex)
 {
     struct r300_texture_state* state = &tex->state;
     struct pipe_texture *pt = &tex->tex;
+    boolean is_r500 = screen->caps->is_r500;
 
     state->format0 = R300_TX_WIDTH((pt->width0 - 1) & 0x7ff) |
                      R300_TX_HEIGHT((pt->height0 - 1) & 0x7ff);
@@ -79,8 +80,8 @@ static void r300_setup_texture_state(struct r300_texture* tex, boolean is_r500)
     }
     assert(is_r500 || (pt->width0 <= 2048 && pt->height0 <= 2048));
 
-    debug_printf("r300: Set texture state (%dx%d, %d levels)\n",
-		 pt->width0, pt->height0, pt->last_level);
+    SCREEN_DBG(screen, DBG_TEX, "r300: Set texture state (%dx%d, %d levels)\n",
+               pt->width0, pt->height0, pt->last_level);
 }
 
 unsigned r300_texture_get_offset(struct r300_texture* tex, unsigned level,
@@ -123,7 +124,8 @@ static unsigned r300_texture_get_tile_size(struct r300_texture* tex, int dim)
  * Return the stride, in bytes, of the texture images of the given texture
  * at the given level.
  */
-unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level)
+unsigned r300_texture_get_stride(struct r300_screen* screen,
+                                 struct r300_texture* tex, unsigned level)
 {
     unsigned tile_width, width;
 
@@ -132,8 +134,8 @@ unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level)
 
     /* Check the level. */
     if (level > tex->tex.last_level) {
-        debug_printf("%s: level (%u) > last_level (%u)\n", __FUNCTION__,
-            level, tex->tex.last_level);
+        SCREEN_DBG(screen, DBG_TEX, "%s: level (%u) > last_level (%u)\n",
+                   __FUNCTION__, level, tex->tex.last_level);
         return 0;
     }
 
@@ -155,15 +157,17 @@ static unsigned r300_texture_get_nblocksy(struct r300_texture* tex,
     return util_format_get_nblocksy(tex->tex.format, height);
 }
 
-static void r300_setup_miptree(struct r300_texture* tex)
+static void r300_setup_miptree(struct r300_screen* screen,
+                               struct r300_texture* tex)
 {
     struct pipe_texture* base = &tex->tex;
     unsigned stride, size, layer_size, nblocksy, i;
 
-    debug_printf("r300: Making miptree for texture, format %s\n", pf_name(base->format));
+    SCREEN_DBG(screen, DBG_TEX, "r300: Making miptree for texture, format %s\n",
+               pf_name(base->format));
 
     for (i = 0; i <= base->last_level; i++) {
-        stride = r300_texture_get_stride(tex, i);
+        stride = r300_texture_get_stride(screen, tex, i);
         nblocksy = r300_texture_get_nblocksy(tex, i);
         layer_size = stride * nblocksy;
 
@@ -177,7 +181,7 @@ static void r300_setup_miptree(struct r300_texture* tex)
         tex->layer_size[i] = layer_size;
         tex->pitch[i] = stride / util_format_get_blocksize(base->format);
 
-        debug_printf("r300: Texture miptree: Level %d "
+        SCREEN_DBG(screen, DBG_TEX, "r300: Texture miptree: Level %d "
                 "(%dx%dx%d px, pitch %d bytes) %d bytes total\n",
                 i, u_minify(base->width0, i), u_minify(base->height0, i),
                 u_minify(base->depth0, i), stride, tex->size);
@@ -196,6 +200,7 @@ static struct pipe_texture*
                         const struct pipe_texture* template)
 {
     struct r300_texture* tex = CALLOC_STRUCT(r300_texture);
+    struct r300_screen* rscreen = r300_screen(screen);
 
     if (!tex) {
         return NULL;
@@ -206,8 +211,8 @@ static struct pipe_texture*
     tex->tex.screen = screen;
 
     r300_setup_flags(tex);
-    r300_setup_miptree(tex);
-    r300_setup_texture_state(tex, r300_screen(screen)->caps->is_r500);
+    r300_setup_miptree(rscreen, tex);
+    r300_setup_texture_state(rscreen, tex);
 
     tex->buffer = screen->buffer_create(screen, 2048,
                                         PIPE_BUFFER_USAGE_PIXEL,
@@ -273,6 +278,7 @@ static struct pipe_texture*
                          struct pipe_buffer* buffer)
 {
     struct r300_texture* tex;
+    struct r300_screen* rscreen = r300_screen(screen);
 
     /* Support only 2D textures without mipmaps */
     if (base->target != PIPE_TEXTURE_2D ||
@@ -294,7 +300,7 @@ static struct pipe_texture*
     tex->pitch[0] = *stride / util_format_get_blocksize(base->format);
 
     r300_setup_flags(tex);
-    r300_setup_texture_state(tex, r300_screen(screen)->caps->is_r500);
+    r300_setup_texture_state(rscreen, tex);
 
     pipe_buffer_reference(&tex->buffer, buffer);
 
@@ -361,7 +367,8 @@ void r300_init_screen_texture_functions(struct pipe_screen* screen)
     screen->video_surface_destroy= r300_video_surface_destroy;
 }
 
-boolean r300_get_texture_buffer(struct pipe_texture* texture,
+boolean r300_get_texture_buffer(struct pipe_screen* screen,
+                                struct pipe_texture* texture,
                                 struct pipe_buffer** buffer,
                                 unsigned* stride)
 {
@@ -373,7 +380,7 @@ boolean r300_get_texture_buffer(struct pipe_texture* texture,
     pipe_buffer_reference(buffer, tex->buffer);
 
     if (stride) {
-        *stride = r300_texture_get_stride(tex, 0);
+        *stride = r300_texture_get_stride(r300_screen(screen), tex, 0);
     }
 
     return TRUE;
diff --git a/src/gallium/drivers/r300/r300_texture.h b/src/gallium/drivers/r300/r300_texture.h
index 55ceb1a513..1be1e6843c 100644
--- a/src/gallium/drivers/r300/r300_texture.h
+++ b/src/gallium/drivers/r300/r300_texture.h
@@ -31,7 +31,8 @@ struct r300_texture;
 
 void r300_init_screen_texture_functions(struct pipe_screen* screen);
 
-unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level);
+unsigned r300_texture_get_stride(struct r300_screen* screen,
+                                 struct r300_texture* tex, unsigned level);
 
 unsigned r300_texture_get_offset(struct r300_texture* tex, unsigned level,
                                  unsigned zslice, unsigned face);
@@ -115,7 +116,8 @@ r300_video_surface(struct pipe_video_surface *pvs)
 
 #ifndef R300_WINSYS_H
 
-boolean r300_get_texture_buffer(struct pipe_texture* texture,
+boolean r300_get_texture_buffer(struct pipe_screen* screen,
+                                struct pipe_texture* texture,
                                 struct pipe_buffer** buffer,
                                 unsigned* stride);
 
diff --git a/src/gallium/drivers/r300/r300_winsys.h b/src/gallium/drivers/r300/r300_winsys.h
index 1ae6de70fe..bdb8b54bab 100644
--- a/src/gallium/drivers/r300/r300_winsys.h
+++ b/src/gallium/drivers/r300/r300_winsys.h
@@ -40,7 +40,8 @@ extern "C" {
 struct pipe_context* r300_create_context(struct pipe_screen* screen,
                                          struct radeon_winsys* radeon_winsys);
 
-boolean r300_get_texture_buffer(struct pipe_texture* texture,
+boolean r300_get_texture_buffer(struct pipe_screen* screen,
+                                struct pipe_texture* texture,
                                 struct pipe_buffer** buffer,
                                 unsigned* stride);
 
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.c b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
index 851c223697..dd73b33ee0 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
@@ -136,12 +136,13 @@ struct pipe_context* radeon_create_context(struct drm_api* api,
 }
 
 boolean radeon_buffer_from_texture(struct drm_api* api,
+                                   struct pipe_screen* screen,
                                    struct pipe_texture* texture,
                                    struct pipe_buffer** buffer,
                                    unsigned* stride)
 {
     /* XXX fix this */
-    return r300_get_texture_buffer(texture, buffer, stride);
+    return r300_get_texture_buffer(screen, texture, buffer, stride);
 }
 
 /* Create a buffer from a handle. */
@@ -208,7 +209,7 @@ static boolean radeon_shared_handle_from_texture(struct drm_api *api,
     struct radeon_pipe_buffer* radeon_buffer;
     struct pipe_buffer *buffer = NULL;
 
-    if (!radeon_buffer_from_texture(api, texture, &buffer, stride)) {
+    if (!radeon_buffer_from_texture(api, screen, texture, &buffer, stride)) {
         return FALSE;
     }
 
@@ -240,7 +241,7 @@ static boolean radeon_local_handle_from_texture(struct drm_api *api,
                                                 unsigned *handle)
 {
     struct pipe_buffer *buffer = NULL;
-    if (!radeon_buffer_from_texture(api, texture, &buffer, stride)) {
+    if (!radeon_buffer_from_texture(api, screen, texture, &buffer, stride)) {
         return FALSE;
     }
 
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.h b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
index bf0e78138d..28f2ba818e 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
@@ -56,6 +56,7 @@ struct pipe_context* radeon_create_context(struct drm_api* api,
                                            struct pipe_screen* screen);
 
 boolean radeon_buffer_from_texture(struct drm_api* api,
+                                   struct pipe_screen* screen,
                                    struct pipe_texture* texture,
                                    struct pipe_buffer** buffer,
                                    unsigned* stride);
-- 
cgit v1.2.3


From c4a2f13a5c59f3527e4e04cb2d8f55e639e1fd46 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 02:25:50 +0100
Subject: r300g: simplify is_format_supported

---
 src/gallium/drivers/r300/r300_screen.c | 57 +++++++---------------------------
 1 file changed, 11 insertions(+), 46 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index d0f2c13df1..e9727e99df 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -183,10 +183,14 @@ static float r300_get_paramf(struct pipe_screen* pscreen, int param)
     }
 }
 
-static boolean check_tex_format(enum pipe_format format, uint32_t usage,
-                                boolean is_r500)
+static boolean r300_is_format_supported(struct pipe_screen* screen,
+                                        enum pipe_format format,
+                                        enum pipe_texture_target target,
+                                        unsigned usage,
+                                        unsigned geom_flags)
 {
     uint32_t retval = 0;
+    boolean is_r500 = r300_screen(screen)->caps->is_r500;
 
     switch (format) {
         /* Supported formats. */
@@ -247,28 +251,13 @@ static boolean check_tex_format(enum pipe_format format, uint32_t usage,
         case PIPE_FORMAT_Z32_UNORM:
         case PIPE_FORMAT_S8Z24_UNORM:
         case PIPE_FORMAT_X8Z24_UNORM:
-            debug_printf("r300: Note: Got unsupported format: %s in %s\n",
-                pf_name(format), __FUNCTION__);
+            SCREEN_DBG(r300_screen(screen), DBG_TEX,
+                       "r300: Note: Got unsupported format: %s in %s\n",
+                       pf_name(format), __FUNCTION__);
             return FALSE;
 
-        /* XXX These don't even exist
-        case PIPE_FORMAT_A32R32G32B32:
-        case PIPE_FORMAT_A16R16G16B16: */
-        /* XXX What the deuce is UV88? (r3xx accel page 14)
-            debug_printf("r300: Warning: Got unimplemented format: %s in %s\n",
-                pf_name(format), __FUNCTION__);
-            return FALSE; */
-
-        /* XXX Supported yet unimplemented r5xx formats: */
-        /* XXX Again, what is UV1010 this time? (r5xx accel page 148) */
-        /* XXX Even more that don't exist
-        case PIPE_FORMAT_A10R10G10B10_UNORM:
-        case PIPE_FORMAT_A2R10G10B10_UNORM:
-        case PIPE_FORMAT_I10_UNORM:
-            debug_printf(
-                "r300: Warning: Got unimplemented r500 format: %s in %s\n",
-                pf_name(format), __FUNCTION__);
-            return FALSE; */
+        /* XXX Add all remaining gallium-supported formats,
+         * see util/u_format.csv. */
 
         default:
             /* Unknown format... */
@@ -286,30 +275,6 @@ static boolean check_tex_format(enum pipe_format format, uint32_t usage,
     return (retval >= usage);
 }
 
-static boolean r300_is_format_supported(struct pipe_screen* pscreen,
-                                        enum pipe_format format,
-                                        enum pipe_texture_target target,
-                                        unsigned tex_usage,
-                                        unsigned geom_flags)
-{
-    switch (target) {
-        case PIPE_TEXTURE_1D:   /* handle 1D textures as 2D ones */
-        case PIPE_TEXTURE_2D:
-        case PIPE_TEXTURE_3D:
-        case PIPE_TEXTURE_CUBE:
-            return check_tex_format(format, tex_usage,
-                r300_screen(pscreen)->caps->is_r500);
-
-        default:
-            debug_printf("r300: Fatal: This is not a format target: %d\n",
-                target);
-            assert(0);
-            break;
-    }
-
-    return FALSE;
-}
-
 static struct pipe_transfer*
 r300_get_tex_transfer(struct pipe_screen *screen,
                       struct pipe_texture *texture,
-- 
cgit v1.2.3


From 7d3090f0fe41757582729bb5a580ad5b9a03f765 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 04:41:51 +0100
Subject: r300g: fix vertex colors with 8 bits per channel

The piglit BGRA tests pass now.
---
 src/gallium/drivers/r300/r300_state_inlines.h | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index 35be00e1b0..a9ee0a26f1 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -537,6 +537,7 @@ r300_translate_vertex_data_type(enum pipe_format format) {
 static INLINE uint16_t
 r300_translate_vertex_data_swizzle(enum pipe_format format) {
     const struct util_format_description *desc = util_format_description(format);
+    unsigned swizzle[4], i;
 
     assert(format);
 
@@ -547,11 +548,22 @@ r300_translate_vertex_data_swizzle(enum pipe_format format) {
         return 0;
     }
 
-    return ((desc->swizzle[0] << R300_SWIZZLE_SELECT_X_SHIFT) |
-        (desc->swizzle[1] << R300_SWIZZLE_SELECT_Y_SHIFT) |
-        (desc->swizzle[2] << R300_SWIZZLE_SELECT_Z_SHIFT) |
-        (desc->swizzle[3] << R300_SWIZZLE_SELECT_W_SHIFT) |
-        (0xf << R300_WRITE_ENA_SHIFT));
+    /* Swizzles for 8bits formats are in the reversed order, not sure why. */
+    if (desc->channel[0].size == 8) {
+        for (i = 0; i < 4; i++) {
+            swizzle[i] = 3 - desc->swizzle[i];
+        }
+    } else {
+        for (i = 0; i < 4; i++) {
+            swizzle[i] = desc->swizzle[i];
+        }
+    }
+
+    return ((swizzle[0] << R300_SWIZZLE_SELECT_X_SHIFT) |
+            (swizzle[1] << R300_SWIZZLE_SELECT_Y_SHIFT) |
+            (swizzle[2] << R300_SWIZZLE_SELECT_Z_SHIFT) |
+            (swizzle[3] << R300_SWIZZLE_SELECT_W_SHIFT) |
+            (0xf << R300_WRITE_ENA_SHIFT));
 }
 
 #endif /* R300_STATE_INLINES_H */
-- 
cgit v1.2.3


From b62d339301f32d29c0a0fdaba2a0e192bff42c23 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 04:49:07 +0100
Subject: r300g: fix the rasterization of secondary color

When the secondary color is used, the primary color must always be rasterized
regardless of whether it is used or not.
---
 src/gallium/drivers/r300/r300_state_derived.c | 3 ++-
 src/gallium/drivers/r300/r300_vs.c            | 9 ++++++---
 2 files changed, 8 insertions(+), 4 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 192846411b..99c2720897 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -350,7 +350,8 @@ static void r300_update_rs_block(struct r300_context* r300,
 
     /* Rasterize colors. */
     for (i = 0; i < ATTR_COLOR_COUNT; i++) {
-        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used) {
+        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used ||
+            vs_outputs->color[1] != ATTR_UNUSED) {
             /* Always rasterize if it's written by the VS,
              * otherwise it locks up. */
             rX00_rs_col(rs, col_count, i, FALSE);
diff --git a/src/gallium/drivers/r300/r300_vs.c b/src/gallium/drivers/r300/r300_vs.c
index 68aef70872..9fbb830047 100644
--- a/src/gallium/drivers/r300/r300_vs.c
+++ b/src/gallium/drivers/r300/r300_vs.c
@@ -124,7 +124,8 @@ static void r300_shader_vap_output_fmt(struct r300_vertex_shader* vs)
 
     /* Colors. */
     for (i = 0; i < ATTR_COLOR_COUNT; i++) {
-        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used) {
+        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used ||
+            vs_outputs->color[1] != ATTR_UNUSED) {
             hwfmt[1] |= R300_INPUT_CNTL_COLOR;
             hwfmt[2] |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT << i;
         }
@@ -182,7 +183,8 @@ static void r300_stream_locations_notcl(
 
     /* Colors. */
     for (i = 0; i < ATTR_COLOR_COUNT; i++) {
-        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used) {
+        if (vs_outputs->color[i] != ATTR_UNUSED || any_bcolor_used ||
+            vs_outputs->color[1] != ATTR_UNUSED) {
             stream_loc[tabi++] = 2 + i;
         }
     }
@@ -259,7 +261,8 @@ static void set_vertex_inputs_outputs(struct r300_vertex_program_compiler * c)
     for (i = 0; i < ATTR_COLOR_COUNT; i++) {
         if (outputs->color[i] != ATTR_UNUSED) {
             c->code->outputs[outputs->color[i]] = reg++;
-        } else if (any_bcolor_used) {
+        } else if (any_bcolor_used ||
+                   outputs->color[1] != ATTR_UNUSED) {
             reg++;
         }
     }
-- 
cgit v1.2.3


From 945cea01d9620ddd6a459e30517080e2bceb6bc5 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sun, 17 Jan 2010 01:13:45 -0800
Subject: r300g: Check for bogus texture targets.

---
 src/gallium/drivers/r300/r300_screen.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index e9727e99df..67325c6b80 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -192,6 +192,12 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
     uint32_t retval = 0;
     boolean is_r500 = r300_screen(screen)->caps->is_r500;
 
+    if (target >= PIPE_MAX_TEXTURE_TYPES) {
+        debug_printf("r300: Implementation error: Received bogus texture "
+            "target %d in %s\n", target, __FUNCTION__);
+        return FALSE;
+    }
+
     switch (format) {
         /* Supported formats. */
         /* Colorbuffer */
-- 
cgit v1.2.3


From a4bbabf494a127dcffc51f9c4734f2ecf75b3961 Mon Sep 17 00:00:00 2001
From: Jakob Bornecrantz <jakob@vmware.com>
Date: Sun, 17 Jan 2010 14:18:26 +0000
Subject: softpipe: Move st/python's malloc winsys to softpipe

Users of softpipe can still specify thier own winsys
but now there is a default one also that uses malloc.
This allows the malloc winsys to be shared with other
users of softpipe.
---
 src/gallium/drivers/softpipe/Makefile              |   3 +-
 src/gallium/drivers/softpipe/SConscript            |   1 +
 src/gallium/drivers/softpipe/sp_winsys.c           | 245 +++++++++++++++++++++
 src/gallium/drivers/softpipe/sp_winsys.h           |  13 +-
 .../state_trackers/python/st_softpipe_winsys.c     | 219 +-----------------
 5 files changed, 260 insertions(+), 221 deletions(-)
 create mode 100644 src/gallium/drivers/softpipe/sp_winsys.c

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile
index bcb887a0b2..e4ac49fa85 100644
--- a/src/gallium/drivers/softpipe/Makefile
+++ b/src/gallium/drivers/softpipe/Makefile
@@ -32,6 +32,7 @@ C_SOURCES = \
 	sp_tex_tile_cache.c \
 	sp_tile_cache.c \
 	sp_surface.c \
-	sp_video_context.c
+	sp_video_context.c \
+	sp_winsys.c
 
 include ../../Makefile.template
diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript
index aac9edf44e..3042e556c6 100644
--- a/src/gallium/drivers/softpipe/SConscript
+++ b/src/gallium/drivers/softpipe/SConscript
@@ -34,6 +34,7 @@ softpipe = env.ConvenienceLibrary(
 		'sp_texture.c',
 		'sp_tile_cache.c',
 		'sp_video_context.c',
+		'sp_winsys.c'
 	])
 
 Export('softpipe')
diff --git a/src/gallium/drivers/softpipe/sp_winsys.c b/src/gallium/drivers/softpipe/sp_winsys.c
new file mode 100644
index 0000000000..8169071dc9
--- /dev/null
+++ b/src/gallium/drivers/softpipe/sp_winsys.c
@@ -0,0 +1,245 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Bismarck, ND., USA
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * 
+ **************************************************************************/
+
+/**
+ * @file
+ * Malloc softpipe winsys. Uses malloc for all memory allocations.
+ * 
+ * @author Keith Whitwell
+ * @author Brian Paul
+ * @author Jose Fonseca
+ */
+
+
+#include "pipe/internal/p_winsys_screen.h"/* port to just p_screen */
+#include "pipe/p_format.h"
+#include "pipe/p_context.h"
+#include "pipe/p_inlines.h"
+#include "util/u_format.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "softpipe/sp_winsys.h"
+
+
+struct st_softpipe_buffer
+{
+   struct pipe_buffer base;
+   boolean userBuffer;  /** Is this a user-space buffer? */
+   void *data;
+   void *mapped;
+};
+
+
+/** Cast wrapper */
+static INLINE struct st_softpipe_buffer *
+st_softpipe_buffer( struct pipe_buffer *buf )
+{
+   return (struct st_softpipe_buffer *)buf;
+}
+
+
+static void *
+st_softpipe_buffer_map(struct pipe_winsys *winsys, 
+                       struct pipe_buffer *buf,
+                       unsigned flags)
+{
+   struct st_softpipe_buffer *st_softpipe_buf = st_softpipe_buffer(buf);
+   st_softpipe_buf->mapped = st_softpipe_buf->data;
+   return st_softpipe_buf->mapped;
+}
+
+
+static void
+st_softpipe_buffer_unmap(struct pipe_winsys *winsys, 
+                         struct pipe_buffer *buf)
+{
+   struct st_softpipe_buffer *st_softpipe_buf = st_softpipe_buffer(buf);
+   st_softpipe_buf->mapped = NULL;
+}
+
+
+static void
+st_softpipe_buffer_destroy(struct pipe_buffer *buf)
+{
+   struct st_softpipe_buffer *oldBuf = st_softpipe_buffer(buf);
+
+   if (oldBuf->data) {
+      if (!oldBuf->userBuffer)
+         align_free(oldBuf->data);
+
+      oldBuf->data = NULL;
+   }
+
+   FREE(oldBuf);
+}
+
+
+static void
+st_softpipe_flush_frontbuffer(struct pipe_winsys *winsys,
+                              struct pipe_surface *surf,
+                              void *context_private)
+{
+}
+
+
+
+static const char *
+st_softpipe_get_name(struct pipe_winsys *winsys)
+{
+   return "softpipe";
+}
+
+
+static struct pipe_buffer *
+st_softpipe_buffer_create(struct pipe_winsys *winsys, 
+                          unsigned alignment, 
+                          unsigned usage,
+                          unsigned size)
+{
+   struct st_softpipe_buffer *buffer = CALLOC_STRUCT(st_softpipe_buffer);
+
+   pipe_reference_init(&buffer->base.reference, 1);
+   buffer->base.alignment = alignment;
+   buffer->base.usage = usage;
+   buffer->base.size = size;
+
+   buffer->data = align_malloc(size, alignment);
+
+   return &buffer->base;
+}
+
+
+/**
+ * Create buffer which wraps user-space data.
+ */
+static struct pipe_buffer *
+st_softpipe_user_buffer_create(struct pipe_winsys *winsys, 
+                               void *ptr, 
+                               unsigned bytes)
+{
+   struct st_softpipe_buffer *buffer;
+   
+   buffer = CALLOC_STRUCT(st_softpipe_buffer);
+   if(!buffer)
+      return NULL;
+   
+   pipe_reference_init(&buffer->base.reference, 1);
+   buffer->base.size = bytes;
+   buffer->userBuffer = TRUE;
+   buffer->data = ptr;
+
+   return &buffer->base;
+}
+
+
+static struct pipe_buffer *
+st_softpipe_surface_buffer_create(struct pipe_winsys *winsys,
+                                  unsigned width, unsigned height,
+                                  enum pipe_format format,
+                                  unsigned usage,
+                                  unsigned tex_usage,
+                                  unsigned *stride)
+{
+   const unsigned alignment = 64;
+   unsigned nblocksy;
+
+   nblocksy = util_format_get_nblocksy(format, height);
+   *stride = align(util_format_get_stride(format, width), alignment);
+
+   return winsys->buffer_create(winsys, alignment,
+                                usage,
+                                *stride * nblocksy);
+}
+
+
+static void
+st_softpipe_fence_reference(struct pipe_winsys *winsys, 
+                            struct pipe_fence_handle **ptr,
+                            struct pipe_fence_handle *fence)
+{
+}
+
+
+static int
+st_softpipe_fence_signalled(struct pipe_winsys *winsys, 
+                            struct pipe_fence_handle *fence,
+                            unsigned flag)
+{
+   return 0;
+}
+
+
+static int
+st_softpipe_fence_finish(struct pipe_winsys *winsys, 
+                         struct pipe_fence_handle *fence,
+                         unsigned flag)
+{
+   return 0;
+}
+
+
+static void
+st_softpipe_destroy(struct pipe_winsys *winsys)
+{
+   FREE(winsys);
+}
+
+
+struct pipe_screen *
+softpipe_create_screen_malloc(void)
+{
+   static struct pipe_winsys *winsys;
+   struct pipe_screen *screen;
+
+   winsys = CALLOC_STRUCT(pipe_winsys);
+   if(!winsys)
+      return NULL;
+
+   winsys->destroy = st_softpipe_destroy;
+   
+   winsys->buffer_create = st_softpipe_buffer_create;
+   winsys->user_buffer_create = st_softpipe_user_buffer_create;
+   winsys->buffer_map = st_softpipe_buffer_map;
+   winsys->buffer_unmap = st_softpipe_buffer_unmap;
+   winsys->buffer_destroy = st_softpipe_buffer_destroy;
+
+   winsys->surface_buffer_create = st_softpipe_surface_buffer_create;
+
+   winsys->fence_reference = st_softpipe_fence_reference;
+   winsys->fence_signalled = st_softpipe_fence_signalled;
+   winsys->fence_finish = st_softpipe_fence_finish;
+
+   winsys->flush_frontbuffer = st_softpipe_flush_frontbuffer;
+   winsys->get_name = st_softpipe_get_name;
+
+   screen = softpipe_create_screen(winsys);
+   if(!screen)
+      st_softpipe_destroy(winsys);
+
+   return screen;
+}
diff --git a/src/gallium/drivers/softpipe/sp_winsys.h b/src/gallium/drivers/softpipe/sp_winsys.h
index f203ded29e..3042e01a05 100644
--- a/src/gallium/drivers/softpipe/sp_winsys.h
+++ b/src/gallium/drivers/softpipe/sp_winsys.h
@@ -49,10 +49,17 @@ struct pipe_buffer;
 
 struct pipe_context *softpipe_create( struct pipe_screen * );
 
+/**
+ * Create a softpipe screen that uses the
+ * given winsys for allocating buffers.
+ */
+struct pipe_screen *softpipe_create_screen( struct pipe_winsys * );
 
-struct pipe_screen *
-softpipe_create_screen(struct pipe_winsys *);
-
+/**
+ * Create a softpipe screen that uses
+ * regular malloc to create all its buffers.
+ */
+struct pipe_screen *softpipe_create_screen_malloc(void);
 
 boolean
 softpipe_get_texture_buffer( struct pipe_texture *texture,
diff --git a/src/gallium/state_trackers/python/st_softpipe_winsys.c b/src/gallium/state_trackers/python/st_softpipe_winsys.c
index a3294e877a..dfe3e465f7 100644
--- a/src/gallium/state_trackers/python/st_softpipe_winsys.c
+++ b/src/gallium/state_trackers/python/st_softpipe_winsys.c
@@ -35,225 +35,10 @@
  * @author Jose Fonseca
  */
 
-
-#include "pipe/internal/p_winsys_screen.h"/* port to just p_screen */
-#include "pipe/p_format.h"
-#include "pipe/p_context.h"
-#include "pipe/p_inlines.h"
-#include "util/u_format.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
 #include "softpipe/sp_winsys.h"
 #include "st_winsys.h"
 
-
-struct st_softpipe_buffer
-{
-   struct pipe_buffer base;
-   boolean userBuffer;  /** Is this a user-space buffer? */
-   void *data;
-   void *mapped;
-};
-
-
-/** Cast wrapper */
-static INLINE struct st_softpipe_buffer *
-st_softpipe_buffer( struct pipe_buffer *buf )
-{
-   return (struct st_softpipe_buffer *)buf;
-}
-
-
-static void *
-st_softpipe_buffer_map(struct pipe_winsys *winsys, 
-                       struct pipe_buffer *buf,
-                       unsigned flags)
-{
-   struct st_softpipe_buffer *st_softpipe_buf = st_softpipe_buffer(buf);
-   st_softpipe_buf->mapped = st_softpipe_buf->data;
-   return st_softpipe_buf->mapped;
-}
-
-
-static void
-st_softpipe_buffer_unmap(struct pipe_winsys *winsys, 
-                         struct pipe_buffer *buf)
-{
-   struct st_softpipe_buffer *st_softpipe_buf = st_softpipe_buffer(buf);
-   st_softpipe_buf->mapped = NULL;
-}
-
-
-static void
-st_softpipe_buffer_destroy(struct pipe_buffer *buf)
-{
-   struct st_softpipe_buffer *oldBuf = st_softpipe_buffer(buf);
-
-   if (oldBuf->data) {
-      if (!oldBuf->userBuffer)
-         align_free(oldBuf->data);
-
-      oldBuf->data = NULL;
-   }
-
-   FREE(oldBuf);
-}
-
-
-static void
-st_softpipe_flush_frontbuffer(struct pipe_winsys *winsys,
-                              struct pipe_surface *surf,
-                              void *context_private)
-{
-}
-
-
-
-static const char *
-st_softpipe_get_name(struct pipe_winsys *winsys)
-{
-   return "softpipe";
-}
-
-
-static struct pipe_buffer *
-st_softpipe_buffer_create(struct pipe_winsys *winsys, 
-                          unsigned alignment, 
-                          unsigned usage,
-                          unsigned size)
-{
-   struct st_softpipe_buffer *buffer = CALLOC_STRUCT(st_softpipe_buffer);
-
-   pipe_reference_init(&buffer->base.reference, 1);
-   buffer->base.alignment = alignment;
-   buffer->base.usage = usage;
-   buffer->base.size = size;
-
-   buffer->data = align_malloc(size, alignment);
-
-   return &buffer->base;
-}
-
-
-/**
- * Create buffer which wraps user-space data.
- */
-static struct pipe_buffer *
-st_softpipe_user_buffer_create(struct pipe_winsys *winsys, 
-                               void *ptr, 
-                               unsigned bytes)
-{
-   struct st_softpipe_buffer *buffer;
-   
-   buffer = CALLOC_STRUCT(st_softpipe_buffer);
-   if(!buffer)
-      return NULL;
-   
-   pipe_reference_init(&buffer->base.reference, 1);
-   buffer->base.size = bytes;
-   buffer->userBuffer = TRUE;
-   buffer->data = ptr;
-
-   return &buffer->base;
-}
-
-
-static struct pipe_buffer *
-st_softpipe_surface_buffer_create(struct pipe_winsys *winsys,
-                                  unsigned width, unsigned height,
-                                  enum pipe_format format,
-                                  unsigned usage,
-                                  unsigned tex_usage,
-                                  unsigned *stride)
-{
-   const unsigned alignment = 64;
-   unsigned nblocksy;
-
-   nblocksy = util_format_get_nblocksy(format, height);
-   *stride = align(util_format_get_stride(format, width), alignment);
-
-   return winsys->buffer_create(winsys, alignment,
-                                usage,
-                                *stride * nblocksy);
-}
-
-
-static void
-st_softpipe_fence_reference(struct pipe_winsys *winsys, 
-                            struct pipe_fence_handle **ptr,
-                            struct pipe_fence_handle *fence)
-{
-}
-
-
-static int
-st_softpipe_fence_signalled(struct pipe_winsys *winsys, 
-                            struct pipe_fence_handle *fence,
-                            unsigned flag)
-{
-   return 0;
-}
-
-
-static int
-st_softpipe_fence_finish(struct pipe_winsys *winsys, 
-                         struct pipe_fence_handle *fence,
-                         unsigned flag)
-{
-   return 0;
-}
-
-
-static void
-st_softpipe_destroy(struct pipe_winsys *winsys)
-{
-   FREE(winsys);
-}
-
-
-static struct pipe_screen *
-st_softpipe_screen_create(void)
-{
-   static struct pipe_winsys *winsys;
-   struct pipe_screen *screen;
-
-   winsys = CALLOC_STRUCT(pipe_winsys);
-   if(!winsys)
-      return NULL;
-
-   winsys->destroy = st_softpipe_destroy;
-   
-   winsys->buffer_create = st_softpipe_buffer_create;
-   winsys->user_buffer_create = st_softpipe_user_buffer_create;
-   winsys->buffer_map = st_softpipe_buffer_map;
-   winsys->buffer_unmap = st_softpipe_buffer_unmap;
-   winsys->buffer_destroy = st_softpipe_buffer_destroy;
-
-   winsys->surface_buffer_create = st_softpipe_surface_buffer_create;
-
-   winsys->fence_reference = st_softpipe_fence_reference;
-   winsys->fence_signalled = st_softpipe_fence_signalled;
-   winsys->fence_finish = st_softpipe_fence_finish;
-
-   winsys->flush_frontbuffer = st_softpipe_flush_frontbuffer;
-   winsys->get_name = st_softpipe_get_name;
-
-   screen = softpipe_create_screen(winsys);
-   if(!screen)
-      st_softpipe_destroy(winsys);
-
-   return screen;
-}
-
-
-static struct pipe_context *
-st_softpipe_context_create(struct pipe_screen *screen)
-{
-   return softpipe_create(screen);
-}
-
-
 const struct st_winsys st_softpipe_winsys = {
-   &st_softpipe_screen_create,
-   &st_softpipe_context_create,
+   &softpipe_create_screen_malloc,
+   &softpipe_create,
 };
-- 
cgit v1.2.3


From 1448d2f252efc1ab8d3175c7fd9701dfe3079577 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Sun, 17 Jan 2010 17:37:55 +0100
Subject: nv50: implement instanced drawing

Too bad we don't have hw array divisors or a method for
setting startInstance.
---
 src/gallium/drivers/nv50/nv50_context.c        |   2 +
 src/gallium/drivers/nv50/nv50_context.h        |  16 +-
 src/gallium/drivers/nv50/nv50_program.c        |  22 +-
 src/gallium/drivers/nv50/nv50_screen.c         |  16 +-
 src/gallium/drivers/nv50/nv50_screen.h         |   2 +
 src/gallium/drivers/nv50/nv50_state_validate.c |   3 +
 src/gallium/drivers/nv50/nv50_vbo.c            | 292 ++++++++++++++++++++++---
 7 files changed, 314 insertions(+), 39 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c
index 22a8498c69..1e69746322 100644
--- a/src/gallium/drivers/nv50/nv50_context.c
+++ b/src/gallium/drivers/nv50/nv50_context.c
@@ -104,7 +104,9 @@ nv50_create(struct pipe_screen *pscreen, unsigned pctx_id)
 	nv50->pipe.destroy = nv50_destroy;
 
 	nv50->pipe.draw_arrays = nv50_draw_arrays;
+	nv50->pipe.draw_arrays_instanced = nv50_draw_arrays_instanced;
 	nv50->pipe.draw_elements = nv50_draw_elements;
+	nv50->pipe.draw_elements_instanced = nv50_draw_elements_instanced;
 	nv50->pipe.clear = nv50_clear;
 
 	nv50->pipe.flush = nv50_flush;
diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h
index d024be6ea9..bebcd95054 100644
--- a/src/gallium/drivers/nv50/nv50_context.h
+++ b/src/gallium/drivers/nv50/nv50_context.h
@@ -29,9 +29,7 @@
 #define NV50_CB_PVP		1
 #define NV50_CB_PFP		2
 #define NV50_CB_PGP		3
-#define NV50_CB_TIC		4
-#define NV50_CB_TSC		5
-#define NV50_CB_PUPLOAD         6
+#define NV50_CB_AUX		4
 
 #define NV50_NEW_BLEND		(1 << 0)
 #define NV50_NEW_ZSA		(1 << 1)
@@ -137,6 +135,7 @@ struct nv50_state {
 	struct nouveau_stateobj *vtxfmt;
 	struct nouveau_stateobj *vtxbuf;
 	struct nouveau_stateobj *vtxattr;
+	struct nouveau_stateobj *instbuf;
 	unsigned vtxelt_nr;
 };
 
@@ -198,11 +197,22 @@ extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50);
 /* nv50_vbo.c */
 extern void nv50_draw_arrays(struct pipe_context *, unsigned mode,
 				unsigned start, unsigned count);
+extern void nv50_draw_arrays_instanced(struct pipe_context *, unsigned mode,
+					unsigned start, unsigned count,
+					unsigned startInstance,
+					unsigned instanceCount);
 extern void nv50_draw_elements(struct pipe_context *pipe,
 				  struct pipe_buffer *indexBuffer,
 				  unsigned indexSize,
 				  unsigned mode, unsigned start,
 				  unsigned count);
+extern void nv50_draw_elements_instanced(struct pipe_context *pipe,
+					 struct pipe_buffer *indexBuffer,
+					 unsigned indexSize,
+					 unsigned mode, unsigned start,
+					 unsigned count,
+					 unsigned startInstance,
+					 unsigned instanceCount);
 extern void nv50_vbo_validate(struct nv50_context *nv50);
 
 /* nv50_clear.c */
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index acd695b39b..4111d212e6 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -95,6 +95,8 @@ struct nv50_reg {
 
 	int vtx; /* vertex index, for GP inputs (TGSI Dimension.Index) */
 	int indirect[2]; /* index into pc->addr, or -1 */
+
+	ubyte buf_index; /* c{0 .. 15}[] or g{0 .. 15}[] */
 };
 
 #define NV50_MOD_NEG 1
@@ -188,6 +190,7 @@ ctor_reg(struct nv50_reg *reg, unsigned type, int index, int hw)
 	reg->vtx = -1;
 	reg->acc = 0;
 	reg->indirect[0] = reg->indirect[1] = -1;
+	reg->buf_index = (type == P_CONST) ? 1 : 0;
 }
 
 static INLINE unsigned
@@ -631,7 +634,7 @@ set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s,
 		set_addr(e, pc->addr[src->indirect[0]]);
 	}
 
-	e->inst[1] |= (((src->type == P_IMMD) ? 0 : 1) << 22);
+	e->inst[1] |= (src->buf_index << 22);
 }
 
 /* Never apply nv50_reg::mod in emit_mov, or carefully check the code !!! */
@@ -3482,6 +3485,19 @@ load_frontfacing(struct nv50_pc *pc, struct nv50_reg *sv)
 	free_temp(pc, temp);
 }
 
+static void
+load_instance_id(struct nv50_pc *pc, unsigned index)
+{
+	struct nv50_reg reg, mem;
+
+	ctor_reg(&reg, P_TEMP, -1, -1);
+	ctor_reg(&mem, P_CONST, -1, 24); /* startInstance */
+	mem.buf_index = 2;
+
+	emit_add_b32(pc, &reg, &pc->sysval[index], &mem);
+	pc->sysval[index] = reg;
+}
+
 static void
 copy_semantic_info(struct nv50_program *p)
 {
@@ -3668,8 +3684,10 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 			}
 			if (p->cfg.regs[0] & (1 << 0))
 				pc->sysval[vertex_id].hw = rid++;
-			if (p->cfg.regs[0] & (1 << 4))
+			if (p->cfg.regs[0] & (1 << 4)) {
 				pc->sysval[instance_id].hw = rid++;
+				load_instance_id(pc, instance_id);
+			}
 		}
 
 		for (i = 0, rid = 0; i < pc->result_nr; ++i) {
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index 73205652cc..9d58f3c965 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -329,7 +329,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	so_ref(NULL, &so);
 
 	/* Static tesla init */
-	so = so_new(44, 90, 22);
+	so = so_new(47, 95, 24);
 
 	so_method(so, screen->tesla, NV50TCL_COND_MODE, 1);
 	so_data  (so, NV50TCL_COND_MODE_ALWAYS);
@@ -372,7 +372,7 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	}
 
 	for (i = 0; i < 3; i++) {
-		ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (128 * 4) * 4,
+		ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM, 0, (256 * 4) * 4,
 				     &screen->constbuf_parm[i]);
 		if (ret) {
 			nv50_screen_destroy(pscreen);
@@ -411,6 +411,18 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev)
 	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
 	so_data  (so, 0x00000031 | (NV50_CB_PMISC << 12));
 
+	/* bind auxiliary constbuf to immediate data bo */
+	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
+	so_reloc (so, screen->constbuf_misc[0], (128 * 4) * 4,
+		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->constbuf_misc[0], (128 * 4) * 4,
+		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, (NV50_CB_AUX << 16) | 0x00000200);
+	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
+	so_data  (so, 0x00000201 | (NV50_CB_AUX << 12));
+	so_method(so, screen->tesla, NV50TCL_SET_PROGRAM_CB, 1);
+	so_data  (so, 0x00000221 | (NV50_CB_AUX << 12));
+
 	so_method(so, screen->tesla, NV50TCL_CB_DEF_ADDRESS_HIGH, 3);
 	so_reloc (so, screen->constbuf_parm[PIPE_SHADER_VERTEX], 0,
 		  NOUVEAU_BO_VRAM | NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h
index 7a155ca9c4..0d786b0f2e 100644
--- a/src/gallium/drivers/nv50/nv50_screen.h
+++ b/src/gallium/drivers/nv50/nv50_screen.h
@@ -23,6 +23,8 @@ struct nv50_screen {
 	struct nouveau_resource *immd_heap[1];
 	struct nouveau_resource *parm_heap[PIPE_SHADER_TYPES];
 
+	struct pipe_buffer *strm_vbuf[16];
+
 	struct nouveau_bo *tic;
 	struct nouveau_bo *tsc;
 
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
index fcd07b59cd..956da9b304 100644
--- a/src/gallium/drivers/nv50/nv50_state_validate.c
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -274,6 +274,9 @@ nv50_state_flush_notify(struct nouveau_channel *chan)
 	so_emit_reloc_markers(chan, nv50->state.fragprog);
 	so_emit_reloc_markers(chan, nv50->state.vtxbuf);
 	so_emit_reloc_markers(chan, nv50->screen->static_init);
+
+	if (nv50->state.instbuf)
+		so_emit_reloc_markers(chan, nv50->state.instbuf);
 }
 
 boolean
diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c
index 89a94d2fe8..3238629461 100644
--- a/src/gallium/drivers/nv50/nv50_vbo.c
+++ b/src/gallium/drivers/nv50/nv50_vbo.c
@@ -160,6 +160,188 @@ nv50_vbo_vtxelt_to_hw(struct pipe_vertex_element *ve)
 	return (hw_type | hw_size);
 }
 
+/* For instanced drawing from user buffers, hitting the FIFO repeatedly
+ * with the same vertex data is probably worse than uploading all data.
+ */
+static boolean
+nv50_upload_vtxbuf(struct nv50_context *nv50, unsigned i)
+{
+	struct nv50_screen *nscreen = nv50->screen;
+	struct pipe_screen *pscreen = &nscreen->base.base;
+	struct pipe_buffer *buf = nscreen->strm_vbuf[i];
+	struct pipe_vertex_buffer *vb = &nv50->vtxbuf[i];
+	uint8_t *src;
+	unsigned size = MAX2(vb->buffer->size, 4096);
+
+	if (buf && buf->size < size)
+		pipe_buffer_reference(&nscreen->strm_vbuf[i], NULL);
+
+	if (!nscreen->strm_vbuf[i]) {
+		nscreen->strm_vbuf[i] = pipe_buffer_create(
+			pscreen, 0, PIPE_BUFFER_USAGE_VERTEX, size);
+		buf = nscreen->strm_vbuf[i];
+	}
+
+	src = pipe_buffer_map(pscreen, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ);
+	if (!src)
+		return FALSE;
+	src += vb->buffer_offset;
+
+	size = (vb->max_index + 1) * vb->stride + 16; /* + 16 is for stride 0 */
+	if (vb->buffer_offset + size > vb->buffer->size)
+		size = vb->buffer->size - vb->buffer_offset;
+
+	pipe_buffer_write(pscreen, buf, vb->buffer_offset, size, src);
+	pipe_buffer_unmap(pscreen, vb->buffer);
+
+	vb->buffer = buf; /* don't pipe_reference, this is a private copy */
+	return TRUE;
+}
+
+static void
+nv50_upload_user_vbufs(struct nv50_context *nv50)
+{
+	unsigned i;
+
+	if (nv50->vbo_fifo)
+		nv50->dirty |= NV50_NEW_ARRAYS;
+	if (!(nv50->dirty & NV50_NEW_ARRAYS))
+		return;
+
+	for (i = 0; i < nv50->vtxbuf_nr; ++i) {
+		if (nv50->vtxbuf[i].buffer->usage & PIPE_BUFFER_USAGE_VERTEX)
+			continue;
+		nv50_upload_vtxbuf(nv50, i);
+	}
+}
+
+static unsigned
+init_per_instance_arrays(struct nv50_context *nv50,
+			 unsigned startInstance,
+			 unsigned pos[16], unsigned step[16])
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_channel *chan = tesla->channel;
+	struct nouveau_bo *bo;
+	struct nouveau_stateobj *so;
+	unsigned i, b, count = 0;
+	const uint32_t rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+
+	so = so_new(nv50->vtxelt_nr, nv50->vtxelt_nr * 2, nv50->vtxelt_nr * 2);
+
+	for (i = 0; i < nv50->vtxelt_nr; ++i) {
+		if (!nv50->vtxelt[i].instance_divisor)
+			continue;
+		++count;
+		b = nv50->vtxelt[i].vertex_buffer_index;
+
+		pos[i] = nv50->vtxelt[i].src_offset +
+			nv50->vtxbuf[b].buffer_offset +
+			startInstance * nv50->vtxbuf[b].stride;
+
+		if (!startInstance) {
+			step[i] = 0;
+			continue;
+		}
+		step[i] = startInstance % nv50->vtxelt[i].instance_divisor;
+
+		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
+
+		so_method(so, tesla, NV50TCL_VERTEX_ARRAY_START_HIGH(i), 2);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_HIGH, 0, 0);
+	}
+
+	if (count) {
+		so_ref (so, &nv50->state.instbuf); /* for flush notify */
+		so_emit(chan, nv50->state.instbuf);
+	}
+	so_ref (NULL, &so);
+
+	return count;
+}
+
+static void
+step_per_instance_arrays(struct nv50_context *nv50,
+			 unsigned pos[16], unsigned step[16])
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_channel *chan = tesla->channel;
+	struct nouveau_bo *bo;
+	struct nouveau_stateobj *so;
+	unsigned i, b;
+	const uint32_t rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+
+	so = so_new(nv50->vtxelt_nr, nv50->vtxelt_nr * 2, nv50->vtxelt_nr * 2);
+
+	for (i = 0; i < nv50->vtxelt_nr; ++i) {
+		if (!nv50->vtxelt[i].instance_divisor)
+			continue;
+		b = nv50->vtxelt[i].vertex_buffer_index;
+
+		if (++step[i] == nv50->vtxelt[i].instance_divisor) {
+			step[i] = 0;
+			pos[i] += nv50->vtxbuf[b].stride;
+		}
+
+		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
+
+		so_method(so, tesla, NV50TCL_VERTEX_ARRAY_START_HIGH(i), 2);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_HIGH, 0, 0);
+	}
+
+	so_ref (so, &nv50->state.instbuf); /* for flush notify */
+	so_ref (NULL, &so);
+
+	so_emit(chan, nv50->state.instbuf);
+}
+
+void
+nv50_draw_arrays_instanced(struct pipe_context *pipe,
+			   unsigned mode, unsigned start, unsigned count,
+			   unsigned startInstance, unsigned instanceCount)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct nouveau_channel *chan = nv50->screen->tesla->channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	unsigned i, nz_divisors;
+	unsigned step[16], pos[16];
+
+	nv50_upload_user_vbufs(nv50);
+
+	nv50_state_validate(nv50);
+
+	nz_divisors = init_per_instance_arrays(nv50, startInstance, pos, step);
+
+	BEGIN_RING(chan, tesla, NV50TCL_CB_ADDR, 2);
+	OUT_RING  (chan, NV50_CB_AUX | (24 << 8));
+	OUT_RING  (chan, startInstance);
+
+	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
+	OUT_RING  (chan, nv50_prim(mode));
+	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+	OUT_RING  (chan, start);
+	OUT_RING  (chan, count);
+	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
+	OUT_RING  (chan, 0);
+
+	for (i = 1; i < instanceCount; i++) {
+		if (nz_divisors) /* any non-zero array divisors ? */
+			step_per_instance_arrays(nv50, pos, step);
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
+		OUT_RING  (chan, nv50_prim(mode) | (1 << 28));
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+		OUT_RING  (chan, start);
+		OUT_RING  (chan, count);
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
+		OUT_RING  (chan, 0);
+	}
+
+	so_ref(NULL, &nv50->state.instbuf);
+}
+
 void
 nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start,
 		 unsigned count)
@@ -285,6 +467,75 @@ nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint32_t *map,
 	return TRUE;
 }
 
+static INLINE void
+nv50_draw_elements_inline(struct nv50_context *nv50,
+			  void *map, unsigned indexSize,
+			  unsigned start, unsigned count)
+{
+	switch (indexSize) {
+	case 1:
+		nv50_draw_elements_inline_u08(nv50, map, start, count);
+		break;
+	case 2:
+		nv50_draw_elements_inline_u16(nv50, map, start, count);
+		break;
+	case 4:
+		nv50_draw_elements_inline_u32(nv50, map, start, count);
+		break;
+	}
+}
+
+void
+nv50_draw_elements_instanced(struct pipe_context *pipe,
+			     struct pipe_buffer *indexBuffer,
+			     unsigned indexSize,
+			     unsigned mode, unsigned start, unsigned count,
+			     unsigned startInstance, unsigned instanceCount)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_channel *chan = tesla->channel;
+	struct pipe_screen *pscreen = pipe->screen;
+	void *map;
+	unsigned i, nz_divisors;
+	unsigned step[16], pos[16];
+
+	map = pipe_buffer_map(pscreen, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ);
+
+	nv50_upload_user_vbufs(nv50);
+
+	nv50_state_validate(nv50);
+
+	nz_divisors = init_per_instance_arrays(nv50, startInstance, pos, step);
+
+	BEGIN_RING(chan, tesla, NV50TCL_CB_ADDR, 2);
+	OUT_RING  (chan, NV50_CB_AUX | (24 << 8));
+	OUT_RING  (chan, startInstance);
+
+	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
+	OUT_RING  (chan, nv50_prim(mode) | (1 << 28));
+
+	nv50_draw_elements_inline(nv50, map, indexSize, start, count);
+
+	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
+	OUT_RING  (chan, 0);
+
+	for (i = 1; i < instanceCount; ++i) {
+		if (nz_divisors) /* any non-zero array divisors ? */
+			step_per_instance_arrays(nv50, pos, step);
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
+		OUT_RING  (chan, nv50_prim(mode) | (1 << 28));
+
+		nv50_draw_elements_inline(nv50, map, indexSize, start, count);
+
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
+		OUT_RING  (chan, 0);
+	}
+
+	so_ref(NULL, &nv50->state.instbuf);
+}
+
 void
 nv50_draw_elements(struct pipe_context *pipe,
 		   struct pipe_buffer *indexBuffer, unsigned indexSize,
@@ -295,7 +546,6 @@ nv50_draw_elements(struct pipe_context *pipe,
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct pipe_screen *pscreen = pipe->screen;
 	void *map;
-	boolean ret;
 	
 	map = pipe_buffer_map(pscreen, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ);
 
@@ -308,29 +558,13 @@ nv50_draw_elements(struct pipe_context *pipe,
 
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
 	OUT_RING  (chan, nv50_prim(mode));
-	switch (indexSize) {
-	case 1:
-		ret = nv50_draw_elements_inline_u08(nv50, map, start, count);
-		break;
-	case 2:
-		ret = nv50_draw_elements_inline_u16(nv50, map, start, count);
-		break;
-	case 4:
-		ret = nv50_draw_elements_inline_u32(nv50, map, start, count);
-		break;
-	default:
-		assert(0);
-		ret = FALSE;
-		break;
-	}
+
+	nv50_draw_elements_inline(nv50, map, indexSize, start, count);
+
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
 	pipe_buffer_unmap(pscreen, indexBuffer);
-        
-        /* XXX: what to do if ret != TRUE?  Flush and retry?
-         */
-	assert(ret);
 }
 
 static INLINE boolean
@@ -343,23 +577,16 @@ nv50_vbo_static_attrib(struct nv50_context *nv50, unsigned attrib,
 	struct nouveau_stateobj *so;
 	struct nouveau_grobj *tesla = nv50->screen->tesla;
 	struct nouveau_bo *bo = nouveau_bo(vb->buffer);
-	float *v;
+	float v[4];
 	int ret;
-	enum pipe_format pf = ve->src_format;
-	const struct util_format_description *desc;
-
-	desc = util_format_description(pf);
-	assert(desc);
-
-	if ((desc->channel[0].type != UTIL_FORMAT_TYPE_FLOAT) ||
-	    util_format_get_component_bits(pf, UTIL_FORMAT_COLORSPACE_RGB, 0) != 32)
-		return FALSE;
 
 	ret = nouveau_bo_map(bo, NOUVEAU_BO_RD);
 	if (ret)
 		return FALSE;
-	v = (float *)(bo->map + (vb->buffer_offset + ve->src_offset));
 
+	util_format_read_4f(ve->src_format, v, 0, (uint8_t *)bo->map +
+			    (vb->buffer_offset + ve->src_offset), 0,
+			    0, 0, 1, 1);
 	so = *pso;
 	if (!so)
 		*pso = so = so_new(nv50->vtxelt_nr, nv50->vtxelt_nr * 4, 0);
@@ -455,7 +682,8 @@ nv50_vbo_validate(struct nv50_context *nv50)
 		}
 
 		so_method(vtxbuf, tesla, NV50TCL_VERTEX_ARRAY_FORMAT(i), 3);
-		so_data  (vtxbuf, 0x20000000 | vb->stride);
+		so_data  (vtxbuf, 0x20000000 |
+			  (ve->instance_divisor ? 0 : vb->stride));
 		so_reloc (vtxbuf, bo, vb->buffer_offset +
 			  ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART |
 			  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
-- 
cgit v1.2.3


From 2a60302e51987ae65455b083fa3d9e9bb0e2087b Mon Sep 17 00:00:00 2001
From: Jakob Bornecrantz <jakob@vmware.com>
Date: Mon, 18 Jan 2010 00:04:48 +0000
Subject: softpipe: Fix leak in vbuf draw module

---
 src/gallium/drivers/softpipe/sp_prim_vbuf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
index 7f573aef3c..5812d1eefe 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
@@ -526,6 +526,7 @@ static void
 sp_vbuf_destroy(struct vbuf_render *vbr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
+   align_free(cvbr->vertex_buffer);
    sp_setup_destroy_context(cvbr->setup);
    FREE(cvbr);
 }
@@ -541,7 +542,6 @@ sp_create_vbuf_backend(struct softpipe_context *sp)
 
    assert(sp->draw);
 
-
    cvbr->base.max_indices = SP_MAX_VBUF_INDEXES;
    cvbr->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE;
 
-- 
cgit v1.2.3


From cd553ec8be14eea7db2d21b9f26ea6e72474a471 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sat, 16 Jan 2010 01:45:10 +0100
Subject: r300g: do not align compressed textures to a tile size

---
 src/gallium/drivers/r300/r300_texture.c | 20 ++++++++++++++------
 1 file changed, 14 insertions(+), 6 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index c6dd27272b..1f73f74c26 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -139,11 +139,15 @@ unsigned r300_texture_get_stride(struct r300_screen* screen,
         return 0;
     }
 
-    tile_width = r300_texture_get_tile_size(tex, TILE_WIDTH);
-    width = align(u_minify(tex->tex.width0, level), tile_width);
+    width = u_minify(tex->tex.width0, level);
 
-    /* Should already be aligned except for S3TC. */
-    return align(util_format_get_stride(tex->tex.format, width), 32);
+    if (!util_format_is_compressed(tex->tex.format)) {
+        tile_width = r300_texture_get_tile_size(tex, TILE_WIDTH);
+        width = align(width, tile_width);
+        return util_format_get_stride(tex->tex.format, width);
+    } else {
+        return align(util_format_get_stride(tex->tex.format, width), 32);
+    }
 }
 
 static unsigned r300_texture_get_nblocksy(struct r300_texture* tex,
@@ -151,8 +155,12 @@ static unsigned r300_texture_get_nblocksy(struct r300_texture* tex,
 {
     unsigned height, tile_height;
 
-    tile_height = r300_texture_get_tile_size(tex, TILE_HEIGHT);
-    height = align(u_minify(tex->tex.height0, level), tile_height);
+    height = u_minify(tex->tex.height0, level);
+
+    if (!util_format_is_compressed(tex->tex.format)) {
+        tile_height = r300_texture_get_tile_size(tex, TILE_HEIGHT);
+        height = align(height, tile_height);
+    }
 
     return util_format_get_nblocksy(tex->tex.format, height);
 }
-- 
cgit v1.2.3


From d3d41cbc3504536c2d2dfa2faaae86b2ae5eb670 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 14:54:35 +0100
Subject: r300g: fix vertex data swizzle

---
 src/gallium/drivers/r300/r300_state_inlines.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index a9ee0a26f1..e2180b33b7 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -551,7 +551,11 @@ r300_translate_vertex_data_swizzle(enum pipe_format format) {
     /* Swizzles for 8bits formats are in the reversed order, not sure why. */
     if (desc->channel[0].size == 8) {
         for (i = 0; i < 4; i++) {
-            swizzle[i] = 3 - desc->swizzle[i];
+            if (desc->swizzle[i] <= 3) {
+                swizzle[i] = 3 - desc->swizzle[i];
+            } else {
+                swizzle[i] = desc->swizzle[i];
+            }
         }
     } else {
         for (i = 0; i < 4; i++) {
-- 
cgit v1.2.3


From 673ae6266ebf66145bf2bd7d37965f3fe60371d5 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 19:21:03 +0100
Subject: r300g: fix polygon offset

As per classic r300.

Not sure why glean/polygonOffset still fails, but we do emit the same register
values as in classic.
---
 src/gallium/drivers/r300/r300_context.h | 12 ++++++++----
 src/gallium/drivers/r300/r300_emit.c    | 30 ++++++++++++++++++++++++------
 src/gallium/drivers/r300/r300_state.c   | 26 ++++++++++++++++++++++----
 3 files changed, 54 insertions(+), 14 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index c65d23823c..5e33dc042a 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -87,10 +87,10 @@ struct r300_rs_state {
     uint32_t point_size;            /* R300_GA_POINT_SIZE: 0x421c */
     uint32_t point_minmax;          /* R300_GA_POINT_MINMAX: 0x4230 */
     uint32_t line_control;          /* R300_GA_LINE_CNTL: 0x4234 */
-    uint32_t depth_scale_front;  /* R300_SU_POLY_OFFSET_FRONT_SCALE: 0x42a4 */
-    uint32_t depth_offset_front;/* R300_SU_POLY_OFFSET_FRONT_OFFSET: 0x42a8 */
-    uint32_t depth_scale_back;    /* R300_SU_POLY_OFFSET_BACK_SCALE: 0x42ac */
-    uint32_t depth_offset_back;  /* R300_SU_POLY_OFFSET_BACK_OFFSET: 0x42b0 */
+    float depth_scale;            /* R300_SU_POLY_OFFSET_FRONT_SCALE: 0x42a4 */
+                                  /* R300_SU_POLY_OFFSET_BACK_SCALE: 0x42ac */
+    float depth_offset;           /* R300_SU_POLY_OFFSET_FRONT_OFFSET: 0x42a8 */
+                                  /* R300_SU_POLY_OFFSET_BACK_OFFSET: 0x42b0 */
     uint32_t polygon_offset_enable; /* R300_SU_POLY_OFFSET_ENABLE: 0x42b4 */
     uint32_t cull_mode;             /* R300_SU_CULL_MODE: 0x42b8 */
     uint32_t line_stipple_config;   /* R300_GA_LINE_STIPPLE_CONFIG: 0x4328 */
@@ -326,6 +326,10 @@ struct r300_context {
     uint32_t dirty_hw;
     /* Whether the TCL engine should be in bypass mode. */
     boolean tcl_bypass;
+    /* Whether polygon offset is enabled. */
+    boolean polygon_offset_enabled;
+    /* Z buffer bit depth. */
+    uint32_t zbuffer_bpp;
 };
 
 /* Convenience cast wrapper. */
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 11bd3346c7..2ea9fab015 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -584,19 +584,37 @@ void r300_emit_query_end(struct r300_context* r300)
 void r300_emit_rs_state(struct r300_context* r300, void* state)
 {
     struct r300_rs_state* rs = (struct r300_rs_state*)state;
+    float scale, offset;
     CS_LOCALS(r300);
 
-    BEGIN_CS(22);
+    BEGIN_CS(18 + (rs->polygon_offset_enable ? 5 : 0));
     OUT_CS_REG(R300_VAP_CNTL_STATUS, rs->vap_control_status);
     OUT_CS_REG(R300_GA_POINT_SIZE, rs->point_size);
     OUT_CS_REG_SEQ(R300_GA_POINT_MINMAX, 2);
     OUT_CS(rs->point_minmax);
     OUT_CS(rs->line_control);
-    OUT_CS_REG_SEQ(R300_SU_POLY_OFFSET_FRONT_SCALE, 6);
-    OUT_CS(rs->depth_scale_front);
-    OUT_CS(rs->depth_offset_front);
-    OUT_CS(rs->depth_scale_back);
-    OUT_CS(rs->depth_offset_back);
+
+    if (rs->polygon_offset_enable) {
+        scale = rs->depth_scale * 12;
+        offset = rs->depth_offset;
+
+        switch (r300->zbuffer_bpp) {
+            case 16:
+                offset *= 4;
+                break;
+            case 24:
+                offset *= 2;
+                break;
+        }
+
+        OUT_CS_REG_SEQ(R300_SU_POLY_OFFSET_FRONT_SCALE, 4);
+        OUT_CS_32F(scale);
+        OUT_CS_32F(offset);
+        OUT_CS_32F(scale);
+        OUT_CS_32F(offset);
+    }
+
+    OUT_CS_REG_SEQ(R300_SU_POLY_OFFSET_ENABLE, 2);
     OUT_CS(rs->polygon_offset_enable);
     OUT_CS(rs->cull_mode);
     OUT_CS_REG(R300_GA_LINE_STIPPLE_CONFIG, rs->line_stipple_config);
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 609b910e12..e2ec0bc5bd 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -486,6 +486,7 @@ static void
                                const struct pipe_framebuffer_state* state)
 {
     struct r300_context* r300 = r300_context(pipe);
+    uint32_t zbuffer_bpp = 0;
 
     if (r300->draw) {
         draw_flush(r300->draw);
@@ -499,6 +500,23 @@ static void
     r300->blend_state.dirty = TRUE;
     r300->dsa_state.dirty = TRUE;
     r300->scissor_state.dirty = TRUE;
+
+    /* Polyfon offset depends on the zbuffer bit depth. */
+    if (state->zsbuf && r300->polygon_offset_enabled) {
+        switch (util_format_get_blocksize(state->zsbuf->texture->format)) {
+            case 2:
+                zbuffer_bpp = 16;
+                break;
+            case 4:
+                zbuffer_bpp = 24;
+                break;
+        }
+
+        if (r300->zbuffer_bpp != zbuffer_bpp) {
+            r300->zbuffer_bpp = zbuffer_bpp;
+            r300->rs_state.dirty = TRUE;
+        }
+    }
 }
 
 /* Create fragment shader state. */
@@ -654,10 +672,8 @@ static void* r300_create_rs_state(struct pipe_context* pipe,
     }
 
     if (rs->polygon_offset_enable) {
-        rs->depth_offset_front = rs->depth_offset_back =
-            fui(state->offset_units);
-        rs->depth_scale_front = rs->depth_scale_back =
-            fui(state->offset_scale);
+        rs->depth_offset = state->offset_units;
+        rs->depth_scale = state->offset_scale;
     }
 
     if (state->line_stipple_enable) {
@@ -691,8 +707,10 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state)
 
     if (rs) {
         r300->tcl_bypass = rs->rs.bypass_vs_clip_and_viewport;
+        r300->polygon_offset_enabled = rs->rs.offset_cw || rs->rs.offset_ccw;
     } else {
         r300->tcl_bypass = FALSE;
+        r300->polygon_offset_enabled = FALSE;
     }
 
     r300->rs_state.state = rs;
-- 
cgit v1.2.3


From 9fb9c9f9ca27fbdb90fc0a7a01bdff64d32c05bb Mon Sep 17 00:00:00 2001
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Date: Mon, 18 Jan 2010 16:12:38 +0100
Subject: nv50: fix nv50_program->immd memory leak

---
 src/gallium/drivers/nv50/nv50_program.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index 4111d212e6..22fa6e6ae4 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -4659,6 +4659,7 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
 
 	nouveau_bo_ref(NULL, &p->bo);
 
+	FREE(p->immd);
 	nouveau_resource_free(&p->data[0]);
 
 	p->translated = 0;
-- 
cgit v1.2.3


From e74db4f20fdc4bd71259480f6f93436c98275cb9 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Mon, 18 Jan 2010 16:22:59 +0100
Subject: nv50: cannot exit shaders on a control flow instruction

Fixes lockup triggered by this ingenious shader:
   1: CALL :3
   2: END
   3: BGNSUB ...
---
 src/gallium/drivers/nv50/nv50_program.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index 22fa6e6ae4..20db51070f 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -477,6 +477,12 @@ is_join(struct nv50_program_exec *e)
 	return FALSE;
 }
 
+static INLINE boolean
+is_control_flow(struct nv50_program_exec *e)
+{
+	return (e->inst[0] & 2);
+}
+
 static INLINE void
 set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx,
 	 struct nv50_program_exec *e)
@@ -3166,7 +3172,9 @@ nv50_program_tx_insn(struct nv50_pc *pc,
 		if (!is_long(pc->p->exec_tail))
 			convert_to_long(pc, pc->p->exec_tail);
 		else
-		if (is_immd(pc->p->exec_tail) || is_join(pc->p->exec_tail))
+		if (is_immd(pc->p->exec_tail) ||
+		    is_join(pc->p->exec_tail) ||
+		    is_control_flow(pc->p->exec_tail))
 			emit_nop(pc);
 
 		pc->p->exec_tail->inst[1] |= 1; /* set exit bit */
-- 
cgit v1.2.3


From 0d6e3dd84d21a9c0a60cd64304d39d8928f37238 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Mon, 18 Jan 2010 19:52:34 +0100
Subject: nv50: make instanced drawing work with edge flags

And fix some obvious mistakes introduced in the
previous instancing commit.
---
 src/gallium/drivers/nv50/nv50_vbo.c | 207 ++++++++++++++++++++++++++++--------
 1 file changed, 162 insertions(+), 45 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c
index 3238629461..b9cf0754b7 100644
--- a/src/gallium/drivers/nv50/nv50_vbo.c
+++ b/src/gallium/drivers/nv50/nv50_vbo.c
@@ -40,6 +40,8 @@ nv50_push_elements_u32(struct nv50_context *, uint32_t *, unsigned);
 static boolean
 nv50_push_arrays(struct nv50_context *, unsigned, unsigned);
 
+#define NV50_USING_LOATHED_EDGEFLAG(ctx) ((ctx)->vertprog->cfg.edgeflag_in < 16)
+
 static INLINE unsigned
 nv50_prim(unsigned mode)
 {
@@ -171,7 +173,7 @@ nv50_upload_vtxbuf(struct nv50_context *nv50, unsigned i)
 	struct pipe_buffer *buf = nscreen->strm_vbuf[i];
 	struct pipe_vertex_buffer *vb = &nv50->vtxbuf[i];
 	uint8_t *src;
-	unsigned size = MAX2(vb->buffer->size, 4096);
+	unsigned size = align(vb->buffer->size, 4096);
 
 	if (buf && buf->size < size)
 		pipe_buffer_reference(&nscreen->strm_vbuf[i], NULL);
@@ -215,6 +217,74 @@ nv50_upload_user_vbufs(struct nv50_context *nv50)
 	}
 }
 
+static void
+nv50_set_static_vtxattr(struct nv50_context *nv50, unsigned i, void *data)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_channel *chan = tesla->channel;
+	float v[4];
+
+	util_format_read_4f(nv50->vtxelt[i].src_format,
+			    v, 0, data, 0, 0, 0, 1, 1);
+
+	switch (nv50->vtxelt[i].nr_components) {
+	case 4:
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_4F_X(i), 4);
+		OUT_RING  (chan, v[0]);
+		OUT_RING  (chan, v[1]);
+		OUT_RING  (chan, v[2]);
+		OUT_RING  (chan, v[3]);
+		break;
+	case 3:
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_3F_X(i), 4);
+		OUT_RING  (chan, v[0]);
+		OUT_RING  (chan, v[1]);
+		OUT_RING  (chan, v[2]);
+		break;
+	case 2:
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_2F_X(i), 4);
+		OUT_RING  (chan, v[0]);
+		OUT_RING  (chan, v[1]);
+		break;
+	case 1:
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_1F(i), 4);
+		OUT_RING  (chan, v[0]);
+		break;
+	default:
+		assert(0);
+		break;
+	}
+}
+
+static unsigned
+init_per_instance_arrays_immd(struct nv50_context *nv50,
+			      unsigned startInstance,
+			      unsigned pos[16], unsigned step[16])
+{
+	struct nouveau_bo *bo;
+	unsigned i, b, count = 0;
+
+	for (i = 0; i < nv50->vtxelt_nr; ++i) {
+		if (!nv50->vtxelt[i].instance_divisor)
+			continue;
+		++count;
+		b = nv50->vtxelt[i].vertex_buffer_index;
+
+		pos[i] = nv50->vtxelt[i].src_offset +
+			nv50->vtxbuf[b].buffer_offset +
+			startInstance * nv50->vtxbuf[b].stride;
+		step[i] = startInstance % nv50->vtxelt[i].instance_divisor;
+
+		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
+		if (!bo->map)
+			nouveau_bo_map(bo, NOUVEAU_BO_RD);
+
+		nv50_set_static_vtxattr(nv50, i, (uint8_t *)bo->map + pos[i]);
+	}
+
+	return count;
+}
+
 static unsigned
 init_per_instance_arrays(struct nv50_context *nv50,
 			 unsigned startInstance,
@@ -227,6 +297,10 @@ init_per_instance_arrays(struct nv50_context *nv50,
 	unsigned i, b, count = 0;
 	const uint32_t rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
 
+	if (nv50->vbo_fifo)
+		return init_per_instance_arrays_immd(nv50, startInstance,
+						     pos, step);
+
 	so = so_new(nv50->vtxelt_nr, nv50->vtxelt_nr * 2, nv50->vtxelt_nr * 2);
 
 	for (i = 0; i < nv50->vtxelt_nr; ++i) {
@@ -248,11 +322,11 @@ init_per_instance_arrays(struct nv50_context *nv50,
 		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
 
 		so_method(so, tesla, NV50TCL_VERTEX_ARRAY_START_HIGH(i), 2);
-		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
 		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_HIGH, 0, 0);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
 	}
 
-	if (count) {
+	if (count && startInstance) {
 		so_ref (so, &nv50->state.instbuf); /* for flush notify */
 		so_emit(chan, nv50->state.instbuf);
 	}
@@ -261,6 +335,28 @@ init_per_instance_arrays(struct nv50_context *nv50,
 	return count;
 }
 
+static void
+step_per_instance_arrays_immd(struct nv50_context *nv50,
+			      unsigned pos[16], unsigned step[16])
+{
+	struct nouveau_bo *bo;
+	unsigned i, b;
+
+	for (i = 0; i < nv50->vtxelt_nr; ++i) {
+		if (!nv50->vtxelt[i].instance_divisor)
+			continue;
+		if (++step[i] != nv50->vtxelt[i].instance_divisor)
+			continue;
+		b = nv50->vtxelt[i].vertex_buffer_index;
+		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
+
+		step[i] = 0;
+		pos[i] += nv50->vtxbuf[b].stride;
+
+		nv50_set_static_vtxattr(nv50, i, (uint8_t *)bo->map + pos[i]);
+	}
+}
+
 static void
 step_per_instance_arrays(struct nv50_context *nv50,
 			 unsigned pos[16], unsigned step[16])
@@ -272,6 +368,11 @@ step_per_instance_arrays(struct nv50_context *nv50,
 	unsigned i, b;
 	const uint32_t rl = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
 
+	if (nv50->vbo_fifo) {
+		step_per_instance_arrays_immd(nv50, pos, step);
+		return;
+	}
+
 	so = so_new(nv50->vtxelt_nr, nv50->vtxelt_nr * 2, nv50->vtxelt_nr * 2);
 
 	for (i = 0; i < nv50->vtxelt_nr; ++i) {
@@ -287,8 +388,8 @@ step_per_instance_arrays(struct nv50_context *nv50,
 		bo = nouveau_bo(nv50->vtxbuf[b].buffer);
 
 		so_method(so, tesla, NV50TCL_VERTEX_ARRAY_START_HIGH(i), 2);
-		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
 		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_HIGH, 0, 0);
+		so_reloc (so, bo, pos[i], rl | NOUVEAU_BO_LOW, 0, 0);
 	}
 
 	so_ref (so, &nv50->state.instbuf); /* for flush notify */
@@ -297,6 +398,16 @@ step_per_instance_arrays(struct nv50_context *nv50,
 	so_emit(chan, nv50->state.instbuf);
 }
 
+static INLINE void
+nv50_unmap_vbufs(struct nv50_context *nv50)
+{
+        unsigned i;
+
+        for (i = 0; i < nv50->vtxbuf_nr; ++i)
+                if (nouveau_bo(nv50->vtxbuf[i].buffer)->map)
+                        nouveau_bo_unmap(nouveau_bo(nv50->vtxbuf[i].buffer));
+}
+
 void
 nv50_draw_arrays_instanced(struct pipe_context *pipe,
 			   unsigned mode, unsigned start, unsigned count,
@@ -308,7 +419,8 @@ nv50_draw_arrays_instanced(struct pipe_context *pipe,
 	unsigned i, nz_divisors;
 	unsigned step[16], pos[16];
 
-	nv50_upload_user_vbufs(nv50);
+	if (!NV50_USING_LOATHED_EDGEFLAG(nv50))
+		nv50_upload_user_vbufs(nv50);
 
 	nv50_state_validate(nv50);
 
@@ -320,9 +432,14 @@ nv50_draw_arrays_instanced(struct pipe_context *pipe,
 
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
 	OUT_RING  (chan, nv50_prim(mode));
-	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
-	OUT_RING  (chan, start);
-	OUT_RING  (chan, count);
+
+	if (nv50->vbo_fifo)
+		nv50_push_arrays(nv50, start, count);
+	else {
+		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+		OUT_RING  (chan, start);
+		OUT_RING  (chan, count);
+	}
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
@@ -332,12 +449,18 @@ nv50_draw_arrays_instanced(struct pipe_context *pipe,
 
 		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
 		OUT_RING  (chan, nv50_prim(mode) | (1 << 28));
-		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
-		OUT_RING  (chan, start);
-		OUT_RING  (chan, count);
+
+		if (nv50->vbo_fifo)
+			nv50_push_arrays(nv50, start, count);
+		else {
+			BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+			OUT_RING  (chan, start);
+			OUT_RING  (chan, count);
+		}
 		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 		OUT_RING  (chan, 0);
 	}
+	nv50_unmap_vbufs(nv50);
 
 	so_ref(NULL, &nv50->state.instbuf);
 }
@@ -372,6 +495,8 @@ nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start,
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
+	nv50_unmap_vbufs(nv50);
+
         /* XXX: not sure what to do if ret != TRUE: flush and retry?
          */
         assert(ret);
@@ -400,7 +525,7 @@ nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map,
 		unsigned nr = count > 2046 ? 2046 : count;
 		int i;
 
-		BEGIN_RING(chan, tesla, NV50TCL_VB_ELEMENT_U16 | 0x40000000, nr >> 1);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VB_ELEMENT_U16, nr >> 1);
 		for (i = 0; i < nr; i += 2)
 			OUT_RING  (chan, (map[i + 1] << 16) | map[i]);
 
@@ -433,7 +558,7 @@ nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map,
 		unsigned nr = count > 2046 ? 2046 : count;
 		int i;
 
-		BEGIN_RING(chan, tesla, NV50TCL_VB_ELEMENT_U16 | 0x40000000, nr >> 1);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VB_ELEMENT_U16, nr >> 1);
 		for (i = 0; i < nr; i += 2)
 			OUT_RING  (chan, (map[i + 1] << 16) | map[i]);
 
@@ -458,7 +583,7 @@ nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint32_t *map,
 	while (count) {
 		unsigned nr = count > 2047 ? 2047 : count;
 
-		BEGIN_RING(chan, tesla, NV50TCL_VB_ELEMENT_U32 | 0x40000000, nr);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VB_ELEMENT_U32, nr);
 		OUT_RINGp (chan, map, nr);
 
 		count -= nr;
@@ -502,7 +627,8 @@ nv50_draw_elements_instanced(struct pipe_context *pipe,
 
 	map = pipe_buffer_map(pscreen, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ);
 
-	nv50_upload_user_vbufs(nv50);
+	if (!NV50_USING_LOATHED_EDGEFLAG(nv50))
+		nv50_upload_user_vbufs(nv50);
 
 	nv50_state_validate(nv50);
 
@@ -513,7 +639,7 @@ nv50_draw_elements_instanced(struct pipe_context *pipe,
 	OUT_RING  (chan, startInstance);
 
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_BEGIN, 1);
-	OUT_RING  (chan, nv50_prim(mode) | (1 << 28));
+	OUT_RING  (chan, nv50_prim(mode));
 
 	nv50_draw_elements_inline(nv50, map, indexSize, start, count);
 
@@ -532,6 +658,7 @@ nv50_draw_elements_instanced(struct pipe_context *pipe,
 		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 		OUT_RING  (chan, 0);
 	}
+	nv50_unmap_vbufs(nv50);
 
 	so_ref(NULL, &nv50->state.instbuf);
 }
@@ -564,6 +691,8 @@ nv50_draw_elements(struct pipe_context *pipe,
 	BEGIN_RING(chan, tesla, NV50TCL_VERTEX_END, 1);
 	OUT_RING  (chan, 0);
 
+	nv50_unmap_vbufs(nv50);
+
 	pipe_buffer_unmap(pscreen, indexBuffer);
 }
 
@@ -644,7 +773,7 @@ nv50_vbo_validate(struct nv50_context *nv50)
 		    !(nv50->vtxbuf[i].buffer->usage & PIPE_BUFFER_USAGE_VERTEX))
 			nv50->vbo_fifo = 0xffff;
 
-	if (nv50->vertprog->cfg.edgeflag_in < 16)
+	if (NV50_USING_LOATHED_EDGEFLAG(nv50))
 		nv50->vbo_fifo = 0xffff; /* vertprog can't set edgeflag */
 
 	n_ve = MAX2(nv50->vtxelt_nr, nv50->state.vtxelt_nr);
@@ -672,14 +801,16 @@ nv50_vbo_validate(struct nv50_context *nv50)
 			nv50->vbo_fifo &= ~(1 << i);
 			continue;
 		}
-		so_data(vtxfmt, hw | i);
 
 		if (nv50->vbo_fifo) {
+			so_data  (vtxfmt, hw |
+				  (ve->instance_divisor ? (1 << 4) : i));
 			so_method(vtxbuf, tesla,
 				  NV50TCL_VERTEX_ARRAY_FORMAT(i), 1);
 			so_data  (vtxbuf, 0);
 			continue;
 		}
+		so_data(vtxfmt, hw | i);
 
 		so_method(vtxbuf, tesla, NV50TCL_VERTEX_ARRAY_FORMAT(i), 3);
 		so_data  (vtxbuf, 0x20000000 |
@@ -721,7 +852,7 @@ typedef void (*pfn_push)(struct nouveau_channel *, void *);
 struct nv50_vbo_emitctx
 {
 	pfn_push push[16];
-	void *map[16];
+	uint8_t *map[16];
 	unsigned stride[16];
 	unsigned nr_ve;
 	unsigned vtx_dwords;
@@ -759,19 +890,18 @@ nv50_map_vbufs(struct nv50_context *nv50)
 
 	for (i = 0; i < nv50->vtxbuf_nr; ++i) {
 		struct pipe_vertex_buffer *vb = &nv50->vtxbuf[i];
-		unsigned size, delta;
+		unsigned size = vb->stride * (vb->max_index + 1) + 16;
 
 		if (nouveau_bo(vb->buffer)->map)
 			continue;
 
-		size = vb->stride * (vb->max_index + 1);
-		delta = vb->buffer_offset;
-
+		size = vb->stride * (vb->max_index + 1) + 16;
+		size = MIN2(size, vb->buffer->size);
 		if (!size)
-			size = vb->buffer->size - vb->buffer_offset;
+			size = vb->buffer->size;
 
 		if (nouveau_bo_map_range(nouveau_bo(vb->buffer),
-					 delta, size, NOUVEAU_BO_RD))
+					 0, size, NOUVEAU_BO_RD))
 			break;
 	}
 
@@ -782,16 +912,6 @@ nv50_map_vbufs(struct nv50_context *nv50)
 	return FALSE;
 }
 
-static INLINE void
-nv50_unmap_vbufs(struct nv50_context *nv50)
-{
-        unsigned i;
-
-        for (i = 0; i < nv50->vtxbuf_nr; ++i)
-                if (nouveau_bo(nv50->vtxbuf[i].buffer)->map)
-                        nouveau_bo_unmap(nouveau_bo(nv50->vtxbuf[i].buffer));
-}
-
 static void
 emit_b32_1(struct nouveau_channel *chan, void *data)
 {
@@ -886,12 +1006,13 @@ emit_prepare(struct nv50_context *nv50, struct nv50_vbo_emitctx *emit,
 
 		ve = &nv50->vtxelt[i];
 		vb = &nv50->vtxbuf[ve->vertex_buffer_index];
-		if (!(nv50->vbo_fifo & (1 << i)))
+		if (!(nv50->vbo_fifo & (1 << i)) || ve->instance_divisor)
 			continue;
 		n = emit->nr_ve++;
 
 		emit->stride[n] = vb->stride;
-		emit->map[n] = nouveau_bo(vb->buffer)->map +
+		emit->map[n] = (uint8_t *)nouveau_bo(vb->buffer)->map +
+			vb->buffer_offset +
 			(start * vb->stride + ve->src_offset);
 
 		desc = util_format_description(ve->src_format);
@@ -981,13 +1102,12 @@ nv50_push_arrays(struct nv50_context *nv50, unsigned start, unsigned count)
 
 		set_edgeflag(chan, tesla, &emit, 0); /* nr will be 1 */
 
-		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VERTEX_DATA, dw);
 		for (i = 0; i < nr; ++i)
 			emit_vtx_next(chan, &emit);
 
 		count -= nr;
 	}
-	nv50_unmap_vbufs(nv50);
 
 	return TRUE;
 }
@@ -1008,13 +1128,12 @@ nv50_push_elements_u32(struct nv50_context *nv50, uint32_t *map, unsigned count)
 
 		set_edgeflag(chan, tesla, &emit, *map);
 
-		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VERTEX_DATA, dw);
 		for (i = 0; i < nr; ++i)
 			emit_vtx(chan, &emit, *map++);
 
 		count -= nr;
 	}
-	nv50_unmap_vbufs(nv50);
 
 	return TRUE;
 }
@@ -1035,13 +1154,12 @@ nv50_push_elements_u16(struct nv50_context *nv50, uint16_t *map, unsigned count)
 
 		set_edgeflag(chan, tesla, &emit, *map);
 
-		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VERTEX_DATA, dw);
 		for (i = 0; i < nr; ++i)
 			emit_vtx(chan, &emit, *map++);
 
 		count -= nr;
 	}
-	nv50_unmap_vbufs(nv50);
 
 	return TRUE;
 }
@@ -1062,13 +1180,12 @@ nv50_push_elements_u08(struct nv50_context *nv50, uint8_t *map, unsigned count)
 
 		set_edgeflag(chan, tesla, &emit, *map);
 
-		BEGIN_RING(chan, tesla, NV50TCL_VERTEX_DATA | 0x40000000, dw);
+		BEGIN_RING_NI(chan, tesla, NV50TCL_VERTEX_DATA, dw);
 		for (i = 0; i < nr; ++i)
 			emit_vtx(chan, &emit, *map++);
 
 		count -= nr;
 	}
-	nv50_unmap_vbufs(nv50);
 
 	return TRUE;
 }
-- 
cgit v1.2.3


From c6559b7e01bbbf7f8d23d6ca58b5fe3d8514ad97 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Mon, 18 Jan 2010 21:37:33 +0100
Subject: nv50: fix constant vtxattr methods

This function was untested, sorry.
---
 src/gallium/drivers/nv50/nv50_vbo.c | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c
index b9cf0754b7..bfb1b34d27 100644
--- a/src/gallium/drivers/nv50/nv50_vbo.c
+++ b/src/gallium/drivers/nv50/nv50_vbo.c
@@ -230,25 +230,25 @@ nv50_set_static_vtxattr(struct nv50_context *nv50, unsigned i, void *data)
 	switch (nv50->vtxelt[i].nr_components) {
 	case 4:
 		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_4F_X(i), 4);
-		OUT_RING  (chan, v[0]);
-		OUT_RING  (chan, v[1]);
-		OUT_RING  (chan, v[2]);
-		OUT_RING  (chan, v[3]);
+		OUT_RINGf (chan, v[0]);
+		OUT_RINGf (chan, v[1]);
+		OUT_RINGf (chan, v[2]);
+		OUT_RINGf (chan, v[3]);
 		break;
 	case 3:
-		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_3F_X(i), 4);
-		OUT_RING  (chan, v[0]);
-		OUT_RING  (chan, v[1]);
-		OUT_RING  (chan, v[2]);
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_3F_X(i), 3);
+		OUT_RINGf (chan, v[0]);
+		OUT_RINGf (chan, v[1]);
+		OUT_RINGf (chan, v[2]);
 		break;
 	case 2:
-		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_2F_X(i), 4);
-		OUT_RING  (chan, v[0]);
-		OUT_RING  (chan, v[1]);
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_2F_X(i), 2);
+		OUT_RINGf (chan, v[0]);
+		OUT_RINGf (chan, v[1]);
 		break;
 	case 1:
-		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_1F(i), 4);
-		OUT_RING  (chan, v[0]);
+		BEGIN_RING(chan, tesla, NV50TCL_VTX_ATTR_1F(i), 1);
+		OUT_RINGf (chan, v[0]);
 		break;
 	default:
 		assert(0);
-- 
cgit v1.2.3


From dc298559c6c838d8ad744e07c134a1c8846c9b33 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Thu, 14 Jan 2010 15:06:48 +0000
Subject: svga: Ensure buffer size/alignment is not zero.

---
 src/gallium/drivers/svga/svga_screen_buffer.c | 3 +++
 src/gallium/drivers/svga/svga_swtnl_backend.c | 6 +++---
 2 files changed, 6 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index 1f8a889672..c014f2ee20 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -594,6 +594,9 @@ svga_buffer_create(struct pipe_screen *screen,
    struct svga_screen *ss = svga_screen(screen);
    struct svga_buffer *sbuf;
    
+   assert(size);
+   assert(alignment);
+
    sbuf = CALLOC_STRUCT(svga_buffer);
    if(!sbuf)
       goto error1;
diff --git a/src/gallium/drivers/svga/svga_swtnl_backend.c b/src/gallium/drivers/svga/svga_swtnl_backend.c
index b4f757a47a..3d82e4f880 100644
--- a/src/gallium/drivers/svga/svga_swtnl_backend.c
+++ b/src/gallium/drivers/svga/svga_swtnl_backend.c
@@ -87,13 +87,13 @@ svga_vbuf_render_allocate_vertices( struct vbuf_render *render,
    if (!svga_render->vbuf) {
       svga_render->vbuf_size = MAX2(size, svga_render->vbuf_alloc_size);
       svga_render->vbuf = pipe_buffer_create(screen,
-                                             0,
+                                             16,
                                              PIPE_BUFFER_USAGE_VERTEX,
                                              svga_render->vbuf_size);
       if(!svga_render->vbuf) {
          svga_context_flush(svga, NULL);
          svga_render->vbuf = pipe_buffer_create(screen,
-                                                0,
+                                                16,
                                                 PIPE_BUFFER_USAGE_VERTEX,
                                                 svga_render->vbuf_size);
          assert(svga_render->vbuf);
@@ -259,7 +259,7 @@ svga_vbuf_render_draw( struct vbuf_render *render,
    if (!svga_render->ibuf) {
       svga_render->ibuf_size = MAX2(size, svga_render->ibuf_alloc_size);
       svga_render->ibuf = pipe_buffer_create(screen,
-                                             0,
+                                             2,
                                              PIPE_BUFFER_USAGE_VERTEX,
                                              svga_render->ibuf_size);
       svga_render->ibuf_offset = 0;
-- 
cgit v1.2.3


From 65fa11959f118a75a82d1e432df0b635d316889f Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Mon, 18 Jan 2010 19:33:13 -0800
Subject: softpipe: Skip depth testing for PIPE_S8_UNORM.

---
 src/gallium/drivers/softpipe/sp_quad_depth_test.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
index 0ca86c4e1c..03f569d466 100644
--- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c
+++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
@@ -667,6 +667,7 @@ depth_test_quads_fallback(struct quad_stage *qs,
    }
 
    if (qs->softpipe->framebuffer.zsbuf && 
+       pf_get_component_bits(qs->softpipe->framebuffer.zsbuf->format, PIPE_FORMAT_COMP_Z) &&
        (qs->softpipe->depth_stencil->depth.enabled ||
         qs->softpipe->depth_stencil->stencil[0].enabled)) {
 
@@ -885,6 +886,7 @@ choose_depth_test(struct quad_stage *qs,
    boolean alpha = qs->softpipe->depth_stencil->alpha.enabled;
 
    boolean depth = (qs->softpipe->framebuffer.zsbuf && 
+                    pf_get_component_bits(qs->softpipe->framebuffer.zsbuf->format, PIPE_FORMAT_COMP_Z) &&
                     qs->softpipe->depth_stencil->depth.enabled);
 
    unsigned depthfunc = qs->softpipe->depth_stencil->depth.func;
@@ -895,7 +897,6 @@ choose_depth_test(struct quad_stage *qs,
 
    boolean occlusion = qs->softpipe->active_query_count;
 
-
    if (!alpha &&
        !depth &&
        !stencil) {
-- 
cgit v1.2.3


From e5da7b8548842500e751b133b96b28113fbe0d76 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Tue, 19 Jan 2010 05:39:18 -0800
Subject: softpipe: Fix vertex buffer memory leak.

---
 src/gallium/drivers/softpipe/sp_prim_vbuf.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
index 7f573aef3c..43a0bd4415 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
@@ -526,6 +526,8 @@ static void
 sp_vbuf_destroy(struct vbuf_render *vbr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
+   if(cvbr->vertex_buffer)
+      align_free(cvbr->vertex_buffer);
    sp_setup_destroy_context(cvbr->setup);
    FREE(cvbr);
 }
-- 
cgit v1.2.3


From a5fb5d6a54195d19942482873d0738eaade9e986 Mon Sep 17 00:00:00 2001
From: Jakob Bornecrantz <jakob@vmware.com>
Date: Tue, 19 Jan 2010 16:46:28 +0000
Subject: st/drm: Expose kernel driver name

Based on patch by Chia-I Wu <olvaffe@gmail.com>

Expose the name of the kernel driver as accepted by drmOpenByName.
---
 src/gallium/drivers/trace/tr_drm.c                   | 1 +
 src/gallium/include/state_tracker/drm_api.h          | 5 +++++
 src/gallium/winsys/drm/intel/gem/intel_drm_api.c     | 1 +
 src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c | 1 +
 src/gallium/winsys/drm/radeon/core/radeon_drm.c      | 1 +
 src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c  | 1 +
 6 files changed, 10 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/trace/tr_drm.c b/src/gallium/drivers/trace/tr_drm.c
index 48d1c4051c..e7ca3a86ea 100644
--- a/src/gallium/drivers/trace/tr_drm.c
+++ b/src/gallium/drivers/trace/tr_drm.c
@@ -173,6 +173,7 @@ trace_drm_create(struct drm_api *api)
    if (!tr_api)
       goto error;
 
+   tr_api->base.driver_name = api->driver_name;
    tr_api->base.create_screen = trace_drm_create_screen;
    tr_api->base.create_context = trace_drm_create_context;
    tr_api->base.texture_from_shared_handle = trace_drm_texture_from_shared_handle;
diff --git a/src/gallium/include/state_tracker/drm_api.h b/src/gallium/include/state_tracker/drm_api.h
index 4d1259e1ee..78882f3aa7 100644
--- a/src/gallium/include/state_tracker/drm_api.h
+++ b/src/gallium/include/state_tracker/drm_api.h
@@ -28,6 +28,11 @@ struct drm_create_screen_arg {
 
 struct drm_api
 {
+	/**
+	 * Kernel driver name, as accepted by drmOpenByName.
+	 */
+	const char *driver_name;
+
 	/**
 	 * Special buffer functions
 	 */
diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
index 5ed2a10af1..89b72fd613 100644
--- a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
+++ b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c
@@ -195,6 +195,7 @@ destroy(struct drm_api *api)
 
 struct drm_api intel_drm_api =
 {
+   .driver_name = "i915",
    .create_context = intel_drm_create_context,
    .create_screen = intel_drm_create_screen,
    .texture_from_shared_handle = intel_drm_texture_from_shared_handle,
diff --git a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
index 317dc44d22..6fccc358e9 100644
--- a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
+++ b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
@@ -252,6 +252,7 @@ nouveau_drm_handle_from_pt(struct drm_api *api, struct pipe_screen *pscreen,
 }
 
 struct drm_api drm_api_hooks = {
+	.driver_name = "nouveau",
 	.create_screen = nouveau_drm_create_screen,
 	.create_context = nouveau_drm_create_context,
 	.texture_from_shared_handle = nouveau_drm_pt_from_name,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.c b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
index 69f14e54f2..4d962c3abc 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.c
@@ -176,6 +176,7 @@ static boolean radeon_local_handle_from_texture(struct drm_api *api,
 }
 
 struct drm_api drm_api_hooks = {
+    .driver_name = "radeon",
     .create_screen = radeon_create_screen,
     .create_context = radeon_create_context,
     .texture_from_shared_handle = radeon_texture_from_shared_handle,
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
index 7c47d434f0..dbbe1701f1 100644
--- a/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
@@ -375,6 +375,7 @@ static struct dri1_api dri1_api_hooks = {
 };
 
 static struct drm_api vmw_drm_api_hooks = {
+   .driver_name = "vmwgfx",
    .create_screen = vmw_drm_create_screen,
    .create_context = vmw_drm_create_context,
    .texture_from_shared_handle = vmw_drm_texture_from_handle,
-- 
cgit v1.2.3


From aec3fe82e490efa36c432b2f43cbeebefa91e88f Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Tue, 19 Jan 2010 11:14:00 -0700
Subject: softpipe: remove redundant assignments

We initialized width/height/depth from pt->width0/height0/depth0 above.
---
 src/gallium/drivers/softpipe/sp_texture.c | 5 -----
 1 file changed, 5 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index a9436a3394..fae72c81aa 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -57,13 +57,8 @@ softpipe_texture_layout(struct pipe_screen *screen,
    unsigned width = pt->width0;
    unsigned height = pt->height0;
    unsigned depth = pt->depth0;
-
    unsigned buffer_size = 0;
 
-   pt->width0 = width;
-   pt->height0 = height;
-   pt->depth0 = depth;
-
    for (level = 0; level <= pt->last_level; level++) {
       spt->stride[level] = util_format_get_stride(pt->format, width);
 
-- 
cgit v1.2.3


From bea9ed4dc6abbcf2b95f5f55554800d4f41da47a Mon Sep 17 00:00:00 2001
From: Jakob Bornecrantz <jakob@vmware.com>
Date: Tue, 19 Jan 2010 22:26:01 +0100
Subject: svga: Do a more propper creation of textures from handles

---
 src/gallium/drivers/svga/svga_screen_texture.c     | 68 +++++++++++++++++++++-
 src/gallium/drivers/svga/svga_winsys.h             |  6 ++
 .../winsys/drm/vmware/core/vmw_screen_dri.c        | 68 ++++++----------------
 3 files changed, 90 insertions(+), 52 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_texture.c b/src/gallium/drivers/svga/svga_screen_texture.c
index 1eb03db280..9ad4edafef 100644
--- a/src/gallium/drivers/svga/svga_screen_texture.c
+++ b/src/gallium/drivers/svga/svga_screen_texture.c
@@ -308,11 +308,19 @@ svga_texture_create(struct pipe_screen *screen,
       tex->key.numFaces = 1;
    }
 
+   tex->key.cachable = 1;
+
    if(templat->tex_usage & PIPE_TEXTURE_USAGE_SAMPLER)
       tex->key.flags |= SVGA3D_SURFACE_HINT_TEXTURE;
 
-   if(templat->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY)
+   if(templat->tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) {
+      tex->key.cachable = 0;
+   }
+
+   if(templat->tex_usage & PIPE_TEXTURE_USAGE_PRIMARY) {
       tex->key.flags |= SVGA3D_SURFACE_HINT_SCANOUT;
+      tex->key.cachable = 0;
+   }
    
    /* 
     * XXX: Never pass the SVGA3D_SURFACE_HINT_RENDERTARGET hint. Mesa cannot
@@ -335,8 +343,6 @@ svga_texture_create(struct pipe_screen *screen,
    if(tex->key.format == SVGA3D_FORMAT_INVALID)
       goto error2;
 
-   tex->key.cachable = 1;
-   
    SVGA_DBG(DEBUG_DMA, "surface_create for texture\n", tex->handle);
    tex->handle = svga_screen_surface_create(svgascreen, &tex->key);
    if (tex->handle)
@@ -418,6 +424,62 @@ svga_texture_blanket(struct pipe_screen * screen,
 }
 
 
+struct pipe_texture *
+svga_screen_texture_wrap_surface(struct pipe_screen *screen,
+				 struct pipe_texture *base,
+				 enum SVGA3dSurfaceFormat format,
+				 struct svga_winsys_surface *srf)
+{
+   struct svga_texture *tex;
+   assert(screen);
+
+   /* Only supports one type */
+   if (base->target != PIPE_TEXTURE_2D ||
+       base->last_level != 0 ||
+       base->depth[0] != 1) {
+      return NULL;
+   }
+
+   if (!srf)
+      return NULL;
+
+   if (svga_translate_format(base->format) != format) {
+      unsigned f1 = svga_translate_format(base->format);
+      unsigned f2 = format;
+
+      /* It's okay for XRGB and ARGB or depth with/out stencil to get mixed up */
+      if ( !( (f1 == SVGA3D_X8R8G8B8 && f2 == SVGA3D_A8R8G8B8) ||
+              (f1 == SVGA3D_A8R8G8B8 && f2 == SVGA3D_X8R8G8B8) ||
+              (f1 == SVGA3D_Z_D24X8 && f2 == SVGA3D_Z_D24S8) ) ) {
+         debug_printf("%s wrong format %u != %u\n", __FUNCTION__, f1, f2);
+         return NULL;
+      }
+   }
+
+   tex = CALLOC_STRUCT(svga_texture);
+   if (!tex)
+      return NULL;
+
+   tex->base = *base;
+   
+
+   if (format == 1)
+      tex->base.format = PIPE_FORMAT_X8R8G8B8_UNORM;
+   else if (format == 2)
+      tex->base.format = PIPE_FORMAT_A8R8G8B8_UNORM;
+
+   pipe_reference_init(&tex->base.reference, 1);
+   tex->base.screen = screen;
+
+   SVGA_DBG(DEBUG_DMA, "wrap surface sid %p\n", srf);
+
+   tex->key.cachable = 0;
+   tex->handle = srf;
+
+   return &tex->base;
+}
+
+
 static void
 svga_texture_destroy(struct pipe_texture *pt)
 {
diff --git a/src/gallium/drivers/svga/svga_winsys.h b/src/gallium/drivers/svga/svga_winsys.h
index 59f299c185..27b99fe4c1 100644
--- a/src/gallium/drivers/svga/svga_winsys.h
+++ b/src/gallium/drivers/svga/svga_winsys.h
@@ -296,4 +296,10 @@ svga_screen_buffer_from_texture(struct pipe_texture *texture,
 				struct pipe_buffer **buffer,
 				unsigned *stride);
 
+struct pipe_texture *
+svga_screen_texture_wrap_surface(struct pipe_screen *screen,
+				 struct pipe_texture *base,
+				 enum SVGA3dSurfaceFormat format,
+				 struct svga_winsys_surface *srf);
+
 #endif /* SVGA_WINSYS_H_ */
diff --git a/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
index dbbe1701f1..60491cb519 100644
--- a/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
+++ b/src/gallium/winsys/drm/vmware/core/vmw_screen_dri.c
@@ -237,22 +237,19 @@ vmw_dri1_present_locked(struct pipe_context *locked_pipe,
    vmw_svga_winsys_surface_reference(&vsrf, NULL);
 }
 
-/**
- * FIXME: We'd probably want to cache these buffers in the
- * screen, based on handle.
- */
-
-static struct pipe_buffer *
-vmw_drm_buffer_from_handle(struct drm_api *drm_api,
-                           struct pipe_screen *screen,
-			   const char *name,
-			   unsigned handle)
+static struct pipe_texture *
+vmw_drm_texture_from_handle(struct drm_api *drm_api,
+			    struct pipe_screen *screen,
+			    struct pipe_texture *templat,
+			    const char *name,
+			    unsigned stride,
+			    unsigned handle)
 {
     struct vmw_svga_winsys_surface *vsrf;
     struct svga_winsys_surface *ssrf;
     struct vmw_winsys_screen *vws =
 	vmw_winsys_screen(svga_winsys_screen(screen));
-    struct pipe_buffer *buf;
+    struct pipe_texture *tex;
     union drm_vmw_surface_reference_arg arg;
     struct drm_vmw_surface_arg *req = &arg.req;
     struct drm_vmw_surface_create_req *rep = &arg.rep;
@@ -299,43 +296,28 @@ vmw_drm_buffer_from_handle(struct drm_api *drm_api,
 
     pipe_reference_init(&vsrf->refcnt, 1);
     p_atomic_set(&vsrf->validated, 0);
+    vsrf->screen = vws;
     vsrf->sid = handle;
     ssrf = svga_winsys_surface(vsrf);
-    buf = svga_screen_buffer_wrap_surface(screen, rep->format, ssrf);
-    if (!buf)
+    tex = svga_screen_texture_wrap_surface(screen, templat, rep->format, ssrf);
+    if (!tex)
 	vmw_svga_winsys_surface_reference(&vsrf, NULL);
 
-    return buf;
+    return tex;
   out_mip:
     vmw_ioctl_surface_destroy(vws, handle);
     return NULL;
 }
 
-static struct pipe_texture *
-vmw_drm_texture_from_handle(struct drm_api *drm_api,
-			    struct pipe_screen *screen,
-			    struct pipe_texture *templat,
-			    const char *name,
-			    unsigned stride,
-			    unsigned handle)
-{
-    struct pipe_buffer *buffer;
-    buffer = vmw_drm_buffer_from_handle(drm_api, screen, name, handle);
-
-    if (!buffer)
-	return NULL;
-
-    return screen->texture_blanket(screen, templat, &stride, buffer);
-}
-
 static boolean
-vmw_drm_handle_from_buffer(struct drm_api *drm_api,
+vmw_drm_handle_from_texture(struct drm_api *drm_api,
                            struct pipe_screen *screen,
-			   struct pipe_buffer *buffer,
+			   struct pipe_texture *texture,
+			   unsigned *stride,
 			   unsigned *handle)
 {
     struct svga_winsys_surface *surface =
-	svga_screen_buffer_get_winsys_surface(buffer);
+	svga_screen_texture_get_winsys_surface(texture);
     struct vmw_svga_winsys_surface *vsrf;
 
     if (!surface)
@@ -343,25 +325,13 @@ vmw_drm_handle_from_buffer(struct drm_api *drm_api,
 
     vsrf = vmw_svga_winsys_surface(surface);
     *handle = vsrf->sid;
+    *stride = pf_get_nblocksx(&texture->block, texture->width[0]) *
+	texture->block.size;
+
     vmw_svga_winsys_surface_reference(&vsrf, NULL);
     return TRUE;
 }
 
-static boolean
-vmw_drm_handle_from_texture(struct drm_api *drm_api,
-			    struct pipe_screen *screen,
-			    struct pipe_texture *texture,
-			    unsigned *stride,
-			    unsigned *handle)
-{
-    struct pipe_buffer *buffer;
-
-    if (!svga_screen_buffer_from_texture(texture, &buffer, stride))
-	return FALSE;
-
-    return vmw_drm_handle_from_buffer(drm_api, screen, buffer, handle);
-}
-
 static struct pipe_context*
 vmw_drm_create_context(struct drm_api *drm_api,
                        struct pipe_screen *screen)
-- 
cgit v1.2.3


From fcbd285e421903ee0a65f19f5d633b25b5923c24 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 19 Jan 2010 17:56:12 -0800
Subject: r300g: Make invariant state into an atom.

---
 src/gallium/drivers/r300/r300_context.c         | 3 ++-
 src/gallium/drivers/r300/r300_context.h         | 3 +++
 src/gallium/drivers/r300/r300_flush.c           | 1 -
 src/gallium/drivers/r300/r300_state_invariant.c | 2 +-
 src/gallium/drivers/r300/r300_state_invariant.h | 2 +-
 5 files changed, 7 insertions(+), 4 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 94a9ab3ef3..67ddec140a 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -129,6 +129,7 @@ static void r300_setup_atoms(struct r300_context* r300)
      * an upper bound on each atom, to keep the emission machinery from
      * underallocating space. */
     make_empty_list(&r300->atom_list);
+    R300_INIT_ATOM(invariant, 73);
     R300_INIT_ATOM(ztop, 2);
     R300_INIT_ATOM(blend, 8);
     R300_INIT_ATOM(blend_color, 3);
@@ -208,7 +209,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     r300_init_state_functions(r300);
 
-    r300_emit_invariant_state(r300);
+    r300->invariant_state.dirty = TRUE;
 
     r300->winsys->set_flush_cb(r300->winsys, r300_flush_cb, r300);
     r300->dirty_state = R300_NEW_KITCHEN_SINK;
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index 5e33dc042a..ec8940f420 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -313,6 +313,9 @@ struct r300_context {
     /* ZTOP state. */
     struct r300_atom ztop_state;
 
+    /* Invariant state. This must be emitted to get the engine started. */
+    struct r300_atom invariant_state;
+
     /* Vertex buffers for Gallium. */
     struct pipe_vertex_buffer vertex_buffer[PIPE_MAX_ATTRIBS];
     int vertex_buffer_count;
diff --git a/src/gallium/drivers/r300/r300_flush.c b/src/gallium/drivers/r300/r300_flush.c
index 59819cb106..15e612d8a6 100644
--- a/src/gallium/drivers/r300/r300_flush.c
+++ b/src/gallium/drivers/r300/r300_flush.c
@@ -53,7 +53,6 @@ static void r300_flush(struct pipe_context* pipe,
 
     if (r300->dirty_hw) {
         FLUSH_CS;
-        r300_emit_invariant_state(r300);
         r300->dirty_state = R300_NEW_KITCHEN_SINK;
         r300->dirty_hw = 0;
 
diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
index 47d7e60a40..f31b2e30df 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.c
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -38,7 +38,7 @@ struct pipe_viewport_state r300_viewport_identity = {
  *
  * Note that eventually this should be empty, but it's useful for development
  * and general unduplication of code. */
-void r300_emit_invariant_state(struct r300_context* r300)
+void r300_emit_invariant_state(struct r300_context* r300, void* state)
 {
     struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps;
     CS_LOCALS(r300);
diff --git a/src/gallium/drivers/r300/r300_state_invariant.h b/src/gallium/drivers/r300/r300_state_invariant.h
index 05cff0d6df..5d1a963654 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.h
+++ b/src/gallium/drivers/r300/r300_state_invariant.h
@@ -25,6 +25,6 @@
 
 struct r300_context;
 
-void r300_emit_invariant_state(struct r300_context* r300);
+void r300_emit_invariant_state(struct r300_context* r300, void* state);
 
 #endif /* R300_STATE_INVARIANT_H */
-- 
cgit v1.2.3


From d4a3f7d3ca377b77aa8c268037740a5372a2bdbd Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 19 Jan 2010 18:09:28 -0800
Subject: r300g: Move GB_AA_CONFIG to its eventual new home.

---
 src/gallium/drivers/r300/r300_context.c         | 4 ++--
 src/gallium/drivers/r300/r300_context.h         | 1 +
 src/gallium/drivers/r300/r300_emit.c            | 5 ++++-
 src/gallium/drivers/r300/r300_state_invariant.c | 4 +---
 4 files changed, 8 insertions(+), 6 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 67ddec140a..3bf9c8b18d 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -129,13 +129,13 @@ static void r300_setup_atoms(struct r300_context* r300)
      * an upper bound on each atom, to keep the emission machinery from
      * underallocating space. */
     make_empty_list(&r300->atom_list);
-    R300_INIT_ATOM(invariant, 73);
+    R300_INIT_ATOM(invariant, 71);
     R300_INIT_ATOM(ztop, 2);
     R300_INIT_ATOM(blend, 8);
     R300_INIT_ATOM(blend_color, 3);
     R300_INIT_ATOM(clip, 29);
     R300_INIT_ATOM(dsa, 8);
-    R300_INIT_ATOM(rs, 22);
+    R300_INIT_ATOM(rs, 25);
     R300_INIT_ATOM(scissor, 3);
     R300_INIT_ATOM(viewport, 9);
 }
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index ec8940f420..48c86fdad7 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -84,6 +84,7 @@ struct r300_rs_state {
     struct pipe_rasterizer_state rs;
 
     uint32_t vap_control_status;    /* R300_VAP_CNTL_STATUS: 0x2140 */
+    uint32_t antialiasing_config;   /* R300_GB_AA_CONFIG: 0x4020 */
     uint32_t point_size;            /* R300_GA_POINT_SIZE: 0x421c */
     uint32_t point_minmax;          /* R300_GA_POINT_MINMAX: 0x4230 */
     uint32_t line_control;          /* R300_GA_LINE_CNTL: 0x4234 */
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 2ea9fab015..921170aef1 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -587,8 +587,11 @@ void r300_emit_rs_state(struct r300_context* r300, void* state)
     float scale, offset;
     CS_LOCALS(r300);
 
-    BEGIN_CS(18 + (rs->polygon_offset_enable ? 5 : 0));
+    BEGIN_CS(20 + (rs->polygon_offset_enable ? 5 : 0));
     OUT_CS_REG(R300_VAP_CNTL_STATUS, rs->vap_control_status);
+
+    OUT_CS_REG(R300_GB_AA_CONFIG, rs->antialiasing_config);
+
     OUT_CS_REG(R300_GA_POINT_SIZE, rs->point_size);
     OUT_CS_REG_SEQ(R300_GA_POINT_MINMAX, 2);
     OUT_CS(rs->point_minmax);
diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c
index f31b2e30df..97927acf1b 100644
--- a/src/gallium/drivers/r300/r300_state_invariant.c
+++ b/src/gallium/drivers/r300/r300_state_invariant.c
@@ -43,7 +43,7 @@ void r300_emit_invariant_state(struct r300_context* r300, void* state)
     struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps;
     CS_LOCALS(r300);
 
-    BEGIN_CS(16 + (caps->has_tcl ? 2: 0));
+    BEGIN_CS(14 + (caps->has_tcl ? 2: 0));
 
     /*** Graphics Backend (GB) ***/
     /* Various GB enables */
@@ -58,8 +58,6 @@ void r300_emit_invariant_state(struct r300_context* r300, void* state)
      */
     /* Source of fog depth */
     OUT_CS_REG(R300_GB_SELECT, R300_GB_FOG_SELECT_1_1_W);
-    /* AA enable */
-    OUT_CS_REG(R300_GB_AA_CONFIG, 0x0);
 
     /*** Fog (FG) ***/
     OUT_CS_REG(R300_FG_FOG_BLEND, 0x0);
-- 
cgit v1.2.3


From ac2a665fd75249e02838ec63ef4a5b3db093ceb1 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Tue, 19 Jan 2010 22:47:03 -0800
Subject: i915g: Remove unnecessary headers.

---
 src/gallium/drivers/i915/i915_clear.c         | 1 -
 src/gallium/drivers/i915/i915_context.c       | 3 ---
 src/gallium/drivers/i915/i915_debug.c         | 1 -
 src/gallium/drivers/i915/i915_debug_fp.c      | 1 -
 src/gallium/drivers/i915/i915_state.c         | 2 --
 src/gallium/drivers/i915/i915_state_derived.c | 1 -
 src/gallium/drivers/i915/i915_state_sampler.c | 1 -
 src/gallium/drivers/i915/i915_surface.c       | 5 -----
 src/gallium/drivers/i915/i915_texture.c       | 2 --
 9 files changed, 17 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i915/i915_clear.c b/src/gallium/drivers/i915/i915_clear.c
index 90530f2826..0d0859f8f3 100644
--- a/src/gallium/drivers/i915/i915_clear.c
+++ b/src/gallium/drivers/i915/i915_clear.c
@@ -32,7 +32,6 @@
 
 #include "util/u_clear.h"
 #include "i915_context.h"
-#include "i915_state.h"
 
 
 /**
diff --git a/src/gallium/drivers/i915/i915_context.c b/src/gallium/drivers/i915/i915_context.c
index 94c8aee30f..16ca638119 100644
--- a/src/gallium/drivers/i915/i915_context.c
+++ b/src/gallium/drivers/i915/i915_context.c
@@ -29,12 +29,9 @@
 #include "i915_state.h"
 #include "i915_screen.h"
 #include "i915_batch.h"
-#include "i915_texture.h"
-#include "i915_reg.h"
 
 #include "draw/draw_context.h"
 #include "pipe/p_defines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "pipe/p_inlines.h"
 #include "util/u_memory.h"
 #include "pipe/p_screen.h"
diff --git a/src/gallium/drivers/i915/i915_debug.c b/src/gallium/drivers/i915/i915_debug.c
index c6e6d6fd31..237654d26b 100644
--- a/src/gallium/drivers/i915/i915_debug.c
+++ b/src/gallium/drivers/i915/i915_debug.c
@@ -29,7 +29,6 @@
 #include "i915_context.h"
 #include "i915_debug.h"
 #include "i915_batch.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_debug.h"
 
 
diff --git a/src/gallium/drivers/i915/i915_debug_fp.c b/src/gallium/drivers/i915/i915_debug_fp.c
index 9c5b117b6d..f9c40d8a11 100644
--- a/src/gallium/drivers/i915/i915_debug_fp.c
+++ b/src/gallium/drivers/i915/i915_debug_fp.c
@@ -29,7 +29,6 @@
 #include "i915_reg.h"
 #include "i915_debug.h"
 #include "pipe/internal/p_winsys_screen.h"
-#include "util/u_memory.h"
 
 
 static void
diff --git a/src/gallium/drivers/i915/i915_state.c b/src/gallium/drivers/i915/i915_state.c
index a04668d32f..1c8cb91525 100644
--- a/src/gallium/drivers/i915/i915_state.c
+++ b/src/gallium/drivers/i915/i915_state.c
@@ -30,7 +30,6 @@
 
 
 #include "draw/draw_context.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "pipe/p_inlines.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
@@ -38,7 +37,6 @@
 
 #include "i915_context.h"
 #include "i915_reg.h"
-#include "i915_state.h"
 #include "i915_state_inlines.h"
 #include "i915_fpc.h"
 
diff --git a/src/gallium/drivers/i915/i915_state_derived.c b/src/gallium/drivers/i915/i915_state_derived.c
index 178d4e8781..1373f9e097 100644
--- a/src/gallium/drivers/i915/i915_state_derived.c
+++ b/src/gallium/drivers/i915/i915_state_derived.c
@@ -33,7 +33,6 @@
 #include "i915_context.h"
 #include "i915_state.h"
 #include "i915_reg.h"
-#include "i915_fpc.h"
 
 
 
diff --git a/src/gallium/drivers/i915/i915_state_sampler.c b/src/gallium/drivers/i915/i915_state_sampler.c
index c5e9084d12..86a019d250 100644
--- a/src/gallium/drivers/i915/i915_state_sampler.c
+++ b/src/gallium/drivers/i915/i915_state_sampler.c
@@ -27,7 +27,6 @@
 
 #include "pipe/p_context.h"
 #include "pipe/p_state.h"
-#include "util/u_memory.h"
 
 #include "i915_state_inlines.h"
 #include "i915_context.h"
diff --git a/src/gallium/drivers/i915/i915_surface.c b/src/gallium/drivers/i915/i915_surface.c
index ab8331f3e6..9b4366daff 100644
--- a/src/gallium/drivers/i915/i915_surface.c
+++ b/src/gallium/drivers/i915/i915_surface.c
@@ -27,13 +27,8 @@
 
 #include "i915_context.h"
 #include "i915_blit.h"
-#include "i915_state.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "util/u_tile.h"
-#include "util/u_rect.h"
 
 
 /* Assumes all values are within bounds -- no checking at this level -
diff --git a/src/gallium/drivers/i915/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c
index 286c9ace8e..9c0de8fa5f 100644
--- a/src/gallium/drivers/i915/i915_texture.c
+++ b/src/gallium/drivers/i915/i915_texture.c
@@ -34,13 +34,11 @@
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
 #include "i915_context.h"
 #include "i915_texture.h"
-#include "i915_debug.h"
 #include "i915_screen.h"
 #include "intel_winsys.h"
 
-- 
cgit v1.2.3


From e3cb614f932ac417d446c84bd54f86d01b1cdcd9 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Wed, 20 Jan 2010 00:08:54 -0800
Subject: softpipe: Remove unnecessary headers.

---
 src/gallium/drivers/softpipe/sp_context.c         | 1 -
 src/gallium/drivers/softpipe/sp_flush.c           | 2 --
 src/gallium/drivers/softpipe/sp_quad_blend.c      | 1 -
 src/gallium/drivers/softpipe/sp_quad_depth_test.c | 1 -
 src/gallium/drivers/softpipe/sp_quad_fs.c         | 2 --
 src/gallium/drivers/softpipe/sp_setup.c           | 1 -
 src/gallium/drivers/softpipe/sp_state_surface.c   | 1 -
 src/gallium/drivers/softpipe/sp_state_vertex.c    | 1 -
 src/gallium/drivers/softpipe/sp_tex_tile_cache.c  | 1 -
 src/gallium/drivers/softpipe/sp_texture.c         | 1 -
 10 files changed, 12 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index 5f60139968..beeb6f6b45 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -43,7 +43,6 @@
 #include "sp_surface.h"
 #include "sp_tile_cache.h"
 #include "sp_tex_tile_cache.h"
-#include "sp_texture.h"
 #include "sp_winsys.h"
 #include "sp_query.h"
 
diff --git a/src/gallium/drivers/softpipe/sp_flush.c b/src/gallium/drivers/softpipe/sp_flush.c
index e38b767cf2..db293a51a9 100644
--- a/src/gallium/drivers/softpipe/sp_flush.c
+++ b/src/gallium/drivers/softpipe/sp_flush.c
@@ -34,11 +34,9 @@
 #include "draw/draw_context.h"
 #include "sp_flush.h"
 #include "sp_context.h"
-#include "sp_surface.h"
 #include "sp_state.h"
 #include "sp_tile_cache.h"
 #include "sp_tex_tile_cache.h"
-#include "sp_winsys.h"
 
 
 void
diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c
index d9babe81da..3b8c2d5789 100644
--- a/src/gallium/drivers/softpipe/sp_quad_blend.c
+++ b/src/gallium/drivers/softpipe/sp_quad_blend.c
@@ -35,7 +35,6 @@
 #include "util/u_memory.h"
 #include "sp_context.h"
 #include "sp_quad.h"
-#include "sp_surface.h"
 #include "sp_tile_cache.h"
 #include "sp_quad_pipe.h"
 
diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
index 03f569d466..5cb17c5ae7 100644
--- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c
+++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c
@@ -34,7 +34,6 @@
 #include "tgsi/tgsi_scan.h"
 #include "sp_context.h"
 #include "sp_quad.h"
-#include "sp_surface.h"
 #include "sp_quad_pipe.h"
 #include "sp_tile_cache.h"
 #include "sp_state.h"           /* for sp_fragment_shader */
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 1e7533d0f9..e799df136e 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -45,8 +45,6 @@
 #include "sp_state.h"
 #include "sp_quad.h"
 #include "sp_quad_pipe.h"
-#include "sp_texture.h"
-#include "sp_tex_sample.h"
 
 
 struct quad_shade_stage
diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c
index 615581b95f..478d8efd1b 100644
--- a/src/gallium/drivers/softpipe/sp_setup.c
+++ b/src/gallium/drivers/softpipe/sp_setup.c
@@ -41,7 +41,6 @@
 #include "draw/draw_private.h"
 #include "draw/draw_vertex.h"
 #include "pipe/p_shader_tokens.h"
-#include "pipe/p_thread.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
diff --git a/src/gallium/drivers/softpipe/sp_state_surface.c b/src/gallium/drivers/softpipe/sp_state_surface.c
index bc0e201130..794a9225b8 100644
--- a/src/gallium/drivers/softpipe/sp_state_surface.c
+++ b/src/gallium/drivers/softpipe/sp_state_surface.c
@@ -30,7 +30,6 @@
 
 #include "sp_context.h"
 #include "sp_state.h"
-#include "sp_surface.h"
 #include "sp_tile_cache.h"
 
 #include "draw/draw_context.h"
diff --git a/src/gallium/drivers/softpipe/sp_state_vertex.c b/src/gallium/drivers/softpipe/sp_state_vertex.c
index 46b6991195..b491d92ed1 100644
--- a/src/gallium/drivers/softpipe/sp_state_vertex.c
+++ b/src/gallium/drivers/softpipe/sp_state_vertex.c
@@ -31,7 +31,6 @@
 
 #include "sp_context.h"
 #include "sp_state.h"
-#include "sp_surface.h"
 
 #include "draw/draw_context.h"
 
diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
index 407a22a9f4..0312a31460 100644
--- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
+++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c
@@ -36,7 +36,6 @@
 #include "util/u_memory.h"
 #include "util/u_tile.h"
 #include "sp_context.h"
-#include "sp_surface.h"
 #include "sp_texture.h"
 #include "sp_tex_tile_cache.h"
 
diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index 7caf2928b4..64a70bf44b 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -36,7 +36,6 @@
 #include "util/u_memory.h"
 
 #include "sp_context.h"
-#include "sp_state.h"
 #include "sp_texture.h"
 #include "sp_screen.h"
 #include "sp_winsys.h"
-- 
cgit v1.2.3


From 04cb5dfb1e0968300204980b08748a140e5e7918 Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Wed, 20 Jan 2010 18:27:53 +0100
Subject: llvmpipe: adapt to per-rendertarget blend changes

---
 src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c | 19 +++++++++++--------
 src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c | 12 ++++++------
 src/gallium/drivers/llvmpipe/lp_state_fs.c      | 20 ++++++++++----------
 3 files changed, 27 insertions(+), 24 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
index ced7b9c11d..ad1dbbc9b1 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_aos.c
@@ -314,9 +314,10 @@ lp_build_blend_aos(LLVMBuilderRef builder,
    LLVMValueRef dst_term;
 
    /* FIXME */
-   assert(blend->colormask == 0xf);
+   assert(blend->independent_blend_enable == 0);
+   assert(blend->rt[0].colormask == 0xf);
 
-   if(!blend->blend_enable)
+   if(!blend->rt[0].blend_enable)
       return src;
 
    /* It makes no sense to blend unless values are normalized */
@@ -333,14 +334,16 @@ lp_build_blend_aos(LLVMBuilderRef builder,
     * combinations it is possible to reorder the operations and therefore saving
     * some instructions. */
 
-   src_term = lp_build_blend_factor(&bld, src, blend->rgb_src_factor, blend->alpha_src_factor, alpha_swizzle);
-   dst_term = lp_build_blend_factor(&bld, dst, blend->rgb_dst_factor, blend->alpha_dst_factor, alpha_swizzle);
+   src_term = lp_build_blend_factor(&bld, src, blend->rt[0].rgb_src_factor,
+                                    blend->rt[0].alpha_src_factor, alpha_swizzle);
+   dst_term = lp_build_blend_factor(&bld, dst, blend->rt[0].rgb_dst_factor,
+                                    blend->rt[0].alpha_dst_factor, alpha_swizzle);
 
    lp_build_name(src_term, "src_term");
    lp_build_name(dst_term, "dst_term");
 
-   if(blend->rgb_func == blend->alpha_func) {
-      return lp_build_blend_func(&bld.base, blend->rgb_func, src_term, dst_term);
+   if(blend->rt[0].rgb_func == blend->rt[0].alpha_func) {
+      return lp_build_blend_func(&bld.base, blend->rt[0].rgb_func, src_term, dst_term);
    }
    else {
       /* Seperate RGB / A functions */
@@ -348,8 +351,8 @@ lp_build_blend_aos(LLVMBuilderRef builder,
       LLVMValueRef rgb;
       LLVMValueRef alpha;
 
-      rgb   = lp_build_blend_func(&bld.base, blend->rgb_func,   src_term, dst_term);
-      alpha = lp_build_blend_func(&bld.base, blend->alpha_func, src_term, dst_term);
+      rgb   = lp_build_blend_func(&bld.base, blend->rt[0].rgb_func,   src_term, dst_term);
+      alpha = lp_build_blend_func(&bld.base, blend->rt[0].alpha_func, src_term, dst_term);
 
       return lp_build_blend_swizzle(&bld, rgb, alpha, LP_BUILD_BLEND_SWIZZLE_RGBA, alpha_swizzle);
    }
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
index 9511299d55..284977b7c3 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
@@ -218,7 +218,7 @@ lp_build_blend_soa(LLVMBuilderRef builder,
    }
 
    for (i = 0; i < 4; ++i) {
-      if (blend->colormask & (1 << i)) {
+      if (blend->rt[0].colormask & (1 << i)) {
          if (blend->logicop_enable) {
             if(!type.floating) {
                res[i] = lp_build_logicop(builder, blend->logicop_func, src[i], dst[i]);
@@ -226,10 +226,10 @@ lp_build_blend_soa(LLVMBuilderRef builder,
             else
                res[i] = dst[i];
          }
-         else if (blend->blend_enable) {
-            unsigned src_factor = i < 3 ? blend->rgb_src_factor : blend->alpha_src_factor;
-            unsigned dst_factor = i < 3 ? blend->rgb_dst_factor : blend->alpha_dst_factor;
-            unsigned func = i < 3 ? blend->rgb_func : blend->alpha_func;
+         else if (blend->rt[0].blend_enable) {
+            unsigned src_factor = i < 3 ? blend->rt[0].rgb_src_factor : blend->rt[0].alpha_src_factor;
+            unsigned dst_factor = i < 3 ? blend->rt[0].rgb_dst_factor : blend->rt[0].alpha_dst_factor;
+            unsigned func = i < 3 ? blend->rt[0].rgb_func : blend->rt[0].alpha_func;
             boolean func_commutative = lp_build_blend_func_commutative(func);
 
             /* It makes no sense to blend unless values are normalized */
@@ -270,7 +270,7 @@ lp_build_blend_soa(LLVMBuilderRef builder,
 
             /* See if this function has been previously applied */
             for(j = 0; j < i; ++j) {
-               unsigned prev_func = j < 3 ? blend->rgb_func : blend->alpha_func;
+               unsigned prev_func = j < 3 ? blend->rt[0].rgb_func : blend->rt[0].alpha_func;
                unsigned func_reverse = lp_build_blend_func_reverse(func, prev_func);
 
                if((!func_reverse &&
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 9f4bbef73f..c6d97bb3e9 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -354,7 +354,7 @@ generate_blend(const struct pipe_blend_state *blend,
    lp_build_blend_soa(builder, blend, type, src, dst, con, res);
 
    for(chan = 0; chan < 4; ++chan) {
-      if(blend->colormask & (1 << chan)) {
+      if(blend->rt[0].colormask & (1 << chan)) {
          LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), chan, 0);
          lp_build_name(res[chan], "res.%c", "rgba"[chan]);
          res[chan] = lp_build_select(&bld, mask, res[chan], dst[chan]);
@@ -423,15 +423,15 @@ generate_fragment(struct llvmpipe_context *lp,
       if(key->blend.logicop_enable) {
          debug_printf("blend.logicop_func = %u\n", key->blend.logicop_func);
       }
-      else if(key->blend.blend_enable) {
-         debug_printf("blend.rgb_func = %s\n",   debug_dump_blend_func  (key->blend.rgb_func, TRUE));
-         debug_printf("rgb_src_factor = %s\n",   debug_dump_blend_factor(key->blend.rgb_src_factor, TRUE));
-         debug_printf("rgb_dst_factor = %s\n",   debug_dump_blend_factor(key->blend.rgb_dst_factor, TRUE));
-         debug_printf("alpha_func = %s\n",       debug_dump_blend_func  (key->blend.alpha_func, TRUE));
-         debug_printf("alpha_src_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_src_factor, TRUE));
-         debug_printf("alpha_dst_factor = %s\n", debug_dump_blend_factor(key->blend.alpha_dst_factor, TRUE));
+      else if(key->blend.rt[0].blend_enable) {
+         debug_printf("blend.rgb_func = %s\n",   debug_dump_blend_func  (key->blend.rt[0].rgb_func, TRUE));
+         debug_printf("rgb_src_factor = %s\n",   debug_dump_blend_factor(key->blend.rt[0].rgb_src_factor, TRUE));
+         debug_printf("rgb_dst_factor = %s\n",   debug_dump_blend_factor(key->blend.rt[0].rgb_dst_factor, TRUE));
+         debug_printf("alpha_func = %s\n",       debug_dump_blend_func  (key->blend.rt[0].alpha_func, TRUE));
+         debug_printf("alpha_src_factor = %s\n", debug_dump_blend_factor(key->blend.rt[0].alpha_src_factor, TRUE));
+         debug_printf("alpha_dst_factor = %s\n", debug_dump_blend_factor(key->blend.rt[0].alpha_dst_factor, TRUE));
       }
-      debug_printf("blend.colormask = 0x%x\n", key->blend.colormask);
+      debug_printf("blend.colormask = 0x%x\n", key->blend.rt[0].colormask);
       for(i = 0; i < PIPE_MAX_SAMPLERS; ++i) {
          if(key->sampler[i].format) {
             debug_printf("sampler[%u] = \n", i);
@@ -782,7 +782,7 @@ make_variant_key(struct llvmpipe_context *lp,
       for(chan = 0; chan < 4; ++chan) {
          enum util_format_swizzle swizzle = format_desc->swizzle[chan];
          if(swizzle > 4)
-            key->blend.colormask &= ~(1 << chan);
+            key->blend.rt[0].colormask &= ~(1 << chan);
       }
    }
 
-- 
cgit v1.2.3


From 208e815c888d5267d146ac3a89a70d733728826c Mon Sep 17 00:00:00 2001
From: Alan Coopersmith <alan.coopersmith@sun.com>
Date: Tue, 19 Jan 2010 18:59:39 -0800
Subject: gallium/i965: Make brw_batchbuffer prototypes match in source &
 header

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
---
 src/gallium/drivers/i965/brw_batchbuffer.c | 2 +-
 src/gallium/drivers/i965/brw_batchbuffer.h | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i965/brw_batchbuffer.c b/src/gallium/drivers/i965/brw_batchbuffer.c
index 22607dc608..003b1fd5bf 100644
--- a/src/gallium/drivers/i965/brw_batchbuffer.c
+++ b/src/gallium/drivers/i965/brw_batchbuffer.c
@@ -155,7 +155,7 @@ _brw_batchbuffer_flush(struct brw_batchbuffer *batch,
 enum pipe_error
 brw_batchbuffer_emit_reloc(struct brw_batchbuffer *batch,
 			   struct brw_winsys_buffer *buffer,
-			   uint32_t usage,
+			   enum brw_buffer_usage usage,
 			   uint32_t delta)
 {
    int ret;
diff --git a/src/gallium/drivers/i965/brw_batchbuffer.h b/src/gallium/drivers/i965/brw_batchbuffer.h
index 7473f5bea4..6ca9f617f5 100644
--- a/src/gallium/drivers/i965/brw_batchbuffer.h
+++ b/src/gallium/drivers/i965/brw_batchbuffer.h
@@ -64,12 +64,12 @@ brw_batchbuffer_reset(struct brw_batchbuffer *batch);
  * Consider it a convenience function wrapping multple
  * intel_buffer_dword() calls.
  */
-int brw_batchbuffer_data(struct brw_batchbuffer *batch,
+enum pipe_error brw_batchbuffer_data(struct brw_batchbuffer *batch,
                             const void *data, GLuint bytes,
 			    enum cliprect_mode cliprect_mode);
 
 
-int brw_batchbuffer_emit_reloc(struct brw_batchbuffer *batch,
+enum pipe_error brw_batchbuffer_emit_reloc(struct brw_batchbuffer *batch,
 			       struct brw_winsys_buffer *buffer,
 			       enum brw_buffer_usage usage,
 			       uint32_t offset);
-- 
cgit v1.2.3


From 074abd889f7f5d2c7805a2c0fe118f463101a63d Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 20 Jan 2010 17:40:48 -0700
Subject: softpipe: added region assertions in softpipe_get_tex_transfer()

---
 src/gallium/drivers/softpipe/sp_texture.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index fae72c81aa..45e80c5447 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -291,6 +291,10 @@ softpipe_get_tex_transfer(struct pipe_screen *screen,
    assert(texture);
    assert(level <= texture->last_level);
 
+   /* make sure the requested region is in the image bounds */
+   assert(x + w <= u_minify(texture->width0, level));
+   assert(y + h <= u_minify(texture->height0, level));
+
    spt = CALLOC_STRUCT(softpipe_transfer);
    if (spt) {
       struct pipe_transfer *pt = &spt->base;
-- 
cgit v1.2.3


From c50fe2c55cfe9c858b7b7859da79edb0693ee8bd Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 21 Jan 2010 12:12:11 -0800
Subject: svga: Remove unnecessary headers.

---
 src/gallium/drivers/svga/svga_draw_arrays.c       | 1 -
 src/gallium/drivers/svga/svga_draw_elements.c     | 1 -
 src/gallium/drivers/svga/svga_pipe_blend.c        | 1 -
 src/gallium/drivers/svga/svga_pipe_constants.c    | 3 ---
 src/gallium/drivers/svga/svga_pipe_depthstencil.c | 1 -
 src/gallium/drivers/svga/svga_pipe_draw.c         | 1 -
 src/gallium/drivers/svga/svga_pipe_flush.c        | 5 -----
 src/gallium/drivers/svga/svga_pipe_fs.c           | 2 --
 src/gallium/drivers/svga/svga_pipe_misc.c         | 6 ------
 src/gallium/drivers/svga/svga_pipe_query.c        | 1 -
 src/gallium/drivers/svga/svga_pipe_rasterizer.c   | 1 -
 src/gallium/drivers/svga/svga_pipe_sampler.c      | 3 ---
 src/gallium/drivers/svga/svga_pipe_vertex.c       | 4 ----
 src/gallium/drivers/svga/svga_pipe_vs.c           | 1 -
 src/gallium/drivers/svga/svga_screen.c            | 2 --
 src/gallium/drivers/svga/svga_state_framebuffer.c | 2 --
 src/gallium/drivers/svga/svga_state_rss.c         | 3 ---
 src/gallium/drivers/svga/svga_state_tss.c         | 2 --
 src/gallium/drivers/svga/svga_swtnl_backend.c     | 1 -
 src/gallium/drivers/svga/svga_swtnl_draw.c        | 1 -
 src/gallium/drivers/svga/svga_swtnl_state.c       | 1 -
 src/gallium/drivers/svga/svga_tgsi_decl_sm20.c    | 3 ---
 src/gallium/drivers/svga/svga_tgsi_decl_sm30.c    | 1 -
 23 files changed, 47 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_draw_arrays.c b/src/gallium/drivers/svga/svga_draw_arrays.c
index 75492dffca..6b6ebc9b58 100644
--- a/src/gallium/drivers/svga/svga_draw_arrays.c
+++ b/src/gallium/drivers/svga/svga_draw_arrays.c
@@ -26,7 +26,6 @@
 #include "svga_cmd.h"
 
 #include "pipe/p_inlines.h"
-#include "util/u_prim.h"
 #include "indices/u_indices.h"
 
 #include "svga_hw_reg.h"
diff --git a/src/gallium/drivers/svga/svga_draw_elements.c b/src/gallium/drivers/svga/svga_draw_elements.c
index 167d817831..022b444eb9 100644
--- a/src/gallium/drivers/svga/svga_draw_elements.c
+++ b/src/gallium/drivers/svga/svga_draw_elements.c
@@ -24,7 +24,6 @@
  **********************************************************/
 
 #include "pipe/p_inlines.h"
-#include "util/u_prim.h"
 #include "util/u_upload_mgr.h"
 #include "indices/u_indices.h"
 
diff --git a/src/gallium/drivers/svga/svga_pipe_blend.c b/src/gallium/drivers/svga/svga_pipe_blend.c
index 855d228755..3ad3f97816 100644
--- a/src/gallium/drivers/svga/svga_pipe_blend.c
+++ b/src/gallium/drivers/svga/svga_pipe_blend.c
@@ -29,7 +29,6 @@
 #include "util/u_memory.h"
 
 #include "svga_context.h"
-#include "svga_state.h"
 
 #include "svga_hw_reg.h"
 
diff --git a/src/gallium/drivers/svga/svga_pipe_constants.c b/src/gallium/drivers/svga/svga_pipe_constants.c
index 10e7a12189..a6dc344bdd 100644
--- a/src/gallium/drivers/svga/svga_pipe_constants.c
+++ b/src/gallium/drivers/svga/svga_pipe_constants.c
@@ -30,9 +30,6 @@
 #include "tgsi/tgsi_parse.h"
 
 #include "svga_context.h"
-#include "svga_state.h"
-#include "svga_hw_reg.h"
-#include "svga_cmd.h"
 
 /***********************************************************************
  * Constant buffers 
diff --git a/src/gallium/drivers/svga/svga_pipe_depthstencil.c b/src/gallium/drivers/svga/svga_pipe_depthstencil.c
index df636c08a0..34e60cb341 100644
--- a/src/gallium/drivers/svga/svga_pipe_depthstencil.c
+++ b/src/gallium/drivers/svga/svga_pipe_depthstencil.c
@@ -29,7 +29,6 @@
 #include "util/u_memory.h"
 
 #include "svga_context.h"
-#include "svga_state.h"
 #include "svga_hw_reg.h"
 
 
diff --git a/src/gallium/drivers/svga/svga_pipe_draw.c b/src/gallium/drivers/svga/svga_pipe_draw.c
index 71a552862e..719b3419f8 100644
--- a/src/gallium/drivers/svga/svga_pipe_draw.c
+++ b/src/gallium/drivers/svga/svga_pipe_draw.c
@@ -33,7 +33,6 @@
 #include "svga_hw_reg.h"
 #include "svga_context.h"
 #include "svga_screen.h"
-#include "svga_winsys.h"
 #include "svga_draw.h"
 #include "svga_state.h"
 #include "svga_swtnl.h"
diff --git a/src/gallium/drivers/svga/svga_pipe_flush.c b/src/gallium/drivers/svga/svga_pipe_flush.c
index 0becb0765a..3eb10336c4 100644
--- a/src/gallium/drivers/svga/svga_pipe_flush.c
+++ b/src/gallium/drivers/svga/svga_pipe_flush.c
@@ -28,13 +28,8 @@
 #include "svga_screen_texture.h"
 #include "svga_context.h"
 #include "svga_winsys.h"
-#include "svga_draw.h"
 #include "svga_debug.h"
 
-#include "svga_hw_reg.h"
-
-
-
 
 static void svga_flush( struct pipe_context *pipe,
                         unsigned flags,
diff --git a/src/gallium/drivers/svga/svga_pipe_fs.c b/src/gallium/drivers/svga/svga_pipe_fs.c
index 5f1213e46a..32f07fb261 100644
--- a/src/gallium/drivers/svga/svga_pipe_fs.c
+++ b/src/gallium/drivers/svga/svga_pipe_fs.c
@@ -32,11 +32,9 @@
 
 #include "svga_screen.h"
 #include "svga_context.h"
-#include "svga_state.h"
 #include "svga_tgsi.h"
 #include "svga_hw_reg.h"
 #include "svga_cmd.h"
-#include "svga_draw.h"
 #include "svga_debug.h"
 
 
diff --git a/src/gallium/drivers/svga/svga_pipe_misc.c b/src/gallium/drivers/svga/svga_pipe_misc.c
index 58cb1e6e23..8cf1f2e083 100644
--- a/src/gallium/drivers/svga/svga_pipe_misc.c
+++ b/src/gallium/drivers/svga/svga_pipe_misc.c
@@ -27,12 +27,6 @@
 
 #include "svga_context.h"
 #include "svga_screen_texture.h"
-#include "svga_state.h"
-#include "svga_winsys.h"
-
-#include "svga_hw_reg.h"
-
-
 
 
 static void svga_set_scissor_state( struct pipe_context *pipe,
diff --git a/src/gallium/drivers/svga/svga_pipe_query.c b/src/gallium/drivers/svga/svga_pipe_query.c
index 01336b0a2c..08283e3731 100644
--- a/src/gallium/drivers/svga/svga_pipe_query.c
+++ b/src/gallium/drivers/svga/svga_pipe_query.c
@@ -32,7 +32,6 @@
 #include "svga_screen.h"
 #include "svga_screen_buffer.h"
 #include "svga_winsys.h"
-#include "svga_draw.h"
 #include "svga_debug.h"
 
 
diff --git a/src/gallium/drivers/svga/svga_pipe_rasterizer.c b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
index b03f8eb9cf..9ea11aad9a 100644
--- a/src/gallium/drivers/svga/svga_pipe_rasterizer.c
+++ b/src/gallium/drivers/svga/svga_pipe_rasterizer.c
@@ -30,7 +30,6 @@
 #include "util/u_memory.h"
 
 #include "svga_context.h"
-#include "svga_state.h"
 
 #include "svga_hw_reg.h"
 
diff --git a/src/gallium/drivers/svga/svga_pipe_sampler.c b/src/gallium/drivers/svga/svga_pipe_sampler.c
index 3eeca6b784..9cc69c8266 100644
--- a/src/gallium/drivers/svga/svga_pipe_sampler.c
+++ b/src/gallium/drivers/svga/svga_pipe_sampler.c
@@ -32,9 +32,6 @@
 
 #include "svga_context.h"
 #include "svga_screen_texture.h"
-#include "svga_state.h"
-
-#include "svga_hw_reg.h"
 
 #include "svga_debug.h"
 
diff --git a/src/gallium/drivers/svga/svga_pipe_vertex.c b/src/gallium/drivers/svga/svga_pipe_vertex.c
index 28e2787e0d..87ac9231d5 100644
--- a/src/gallium/drivers/svga/svga_pipe_vertex.c
+++ b/src/gallium/drivers/svga/svga_pipe_vertex.c
@@ -32,10 +32,6 @@
 #include "svga_screen.h"
 #include "svga_screen_buffer.h"
 #include "svga_context.h"
-#include "svga_state.h"
-#include "svga_winsys.h"
-
-#include "svga_hw_reg.h"
 
 
 static void svga_set_vertex_buffers(struct pipe_context *pipe,
diff --git a/src/gallium/drivers/svga/svga_pipe_vs.c b/src/gallium/drivers/svga/svga_pipe_vs.c
index fd9864c51a..7f2b2bce79 100644
--- a/src/gallium/drivers/svga/svga_pipe_vs.c
+++ b/src/gallium/drivers/svga/svga_pipe_vs.c
@@ -33,7 +33,6 @@
 
 #include "svga_screen.h"
 #include "svga_context.h"
-#include "svga_state.h"
 #include "svga_tgsi.h"
 #include "svga_hw_reg.h"
 #include "svga_cmd.h"
diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index fc1b3c980e..68c09e80af 100644
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -33,10 +33,8 @@
 #include "svga_screen.h"
 #include "svga_screen_texture.h"
 #include "svga_screen_buffer.h"
-#include "svga_cmd.h"
 #include "svga_debug.h"
 
-#include "svga_hw_reg.h"
 #include "svga3d_shaderdefs.h"
 
 
diff --git a/src/gallium/drivers/svga/svga_state_framebuffer.c b/src/gallium/drivers/svga/svga_state_framebuffer.c
index cfdcae4ee4..eda1aefd67 100644
--- a/src/gallium/drivers/svga/svga_state_framebuffer.c
+++ b/src/gallium/drivers/svga/svga_state_framebuffer.c
@@ -32,8 +32,6 @@
 #include "svga_cmd.h"
 #include "svga_debug.h"
 
-#include "svga_hw_reg.h"
-
 
 /***********************************************************************
  * Hardware state update
diff --git a/src/gallium/drivers/svga/svga_state_rss.c b/src/gallium/drivers/svga/svga_state_rss.c
index 8b6803a285..2f9adaeb56 100644
--- a/src/gallium/drivers/svga/svga_state_rss.c
+++ b/src/gallium/drivers/svga/svga_state_rss.c
@@ -31,9 +31,6 @@
 #include "svga_state.h"
 #include "svga_cmd.h"
 
-#include "svga_hw_reg.h"
-
-
 
 struct rs_queue {
    unsigned rs_count;
diff --git a/src/gallium/drivers/svga/svga_state_tss.c b/src/gallium/drivers/svga/svga_state_tss.c
index b313794520..b3c9687b1a 100644
--- a/src/gallium/drivers/svga/svga_state_tss.c
+++ b/src/gallium/drivers/svga/svga_state_tss.c
@@ -33,8 +33,6 @@
 #include "svga_state.h"
 #include "svga_cmd.h"
 
-#include "svga_hw_reg.h"
-
 
 void svga_cleanup_tss_binding(struct svga_context *svga)
 {
diff --git a/src/gallium/drivers/svga/svga_swtnl_backend.c b/src/gallium/drivers/svga/svga_swtnl_backend.c
index 3d82e4f880..cfbdcf2554 100644
--- a/src/gallium/drivers/svga/svga_swtnl_backend.c
+++ b/src/gallium/drivers/svga/svga_swtnl_backend.c
@@ -31,7 +31,6 @@
 #include "pipe/p_inlines.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
-#include "util/u_simple_shaders.h"
 
 #include "svga_context.h"
 #include "svga_state.h"
diff --git a/src/gallium/drivers/svga/svga_swtnl_draw.c b/src/gallium/drivers/svga/svga_swtnl_draw.c
index 8b14c913f7..6729cc7381 100644
--- a/src/gallium/drivers/svga/svga_swtnl_draw.c
+++ b/src/gallium/drivers/svga/svga_swtnl_draw.c
@@ -27,7 +27,6 @@
 #include "draw/draw_vbuf.h"
 #include "pipe/p_inlines.h"
 #include "pipe/p_state.h"
-#include "util/u_memory.h"
 
 #include "svga_context.h"
 #include "svga_swtnl.h"
diff --git a/src/gallium/drivers/svga/svga_swtnl_state.c b/src/gallium/drivers/svga/svga_swtnl_state.c
index 1616312113..d57ed82866 100644
--- a/src/gallium/drivers/svga/svga_swtnl_state.c
+++ b/src/gallium/drivers/svga/svga_swtnl_state.c
@@ -27,7 +27,6 @@
 #include "draw/draw_vbuf.h"
 #include "pipe/p_inlines.h"
 #include "pipe/p_state.h"
-#include "util/u_memory.h"
 
 #include "svga_context.h"
 #include "svga_swtnl.h"
diff --git a/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c b/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c
index 54457082a0..0c3430127a 100644
--- a/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c
+++ b/src/gallium/drivers/svga/svga_tgsi_decl_sm20.c
@@ -29,9 +29,6 @@
 #include "util/u_memory.h"
 
 #include "svga_tgsi_emit.h"
-#include "svga_context.h"
-
-
 
 
 static boolean ps20_input( struct svga_shader_emitter *emit,
diff --git a/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c b/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c
index 08e7dfb117..2291cf116d 100644
--- a/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c
+++ b/src/gallium/drivers/svga/svga_tgsi_decl_sm30.c
@@ -29,7 +29,6 @@
 #include "util/u_memory.h"
 
 #include "svga_tgsi_emit.h"
-#include "svga_context.h"
 
 static boolean translate_vs_ps_semantic( struct tgsi_declaration_semantic semantic,
                                          unsigned *usage,
-- 
cgit v1.2.3


From 4222bf3d3c7c85112f2e21bd3ea29b1f496ed8bb Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 21 Jan 2010 14:31:15 -0800
Subject: llvmpipe: Remove unnecessary headers.

---
 src/gallium/drivers/llvmpipe/lp_bld_alpha.c       | 1 -
 src/gallium/drivers/llvmpipe/lp_bld_arit.c        | 1 -
 src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c   | 1 -
 src/gallium/drivers/llvmpipe/lp_bld_conv.c        | 2 --
 src/gallium/drivers/llvmpipe/lp_bld_format_aos.c  | 1 -
 src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c    | 2 --
 src/gallium/drivers/llvmpipe/lp_draw_arrays.c     | 2 --
 src/gallium/drivers/llvmpipe/lp_flush.c           | 2 --
 src/gallium/drivers/llvmpipe/lp_jit.c             | 1 -
 src/gallium/drivers/llvmpipe/lp_setup.c           | 1 -
 src/gallium/drivers/llvmpipe/lp_state_fs.c        | 2 --
 src/gallium/drivers/llvmpipe/lp_state_surface.c   | 1 -
 src/gallium/drivers/llvmpipe/lp_state_vertex.c    | 1 -
 src/gallium/drivers/llvmpipe/lp_test_blend.c      | 1 -
 src/gallium/drivers/llvmpipe/lp_tex_cache.c       | 1 -
 src/gallium/drivers/llvmpipe/lp_tex_sample_c.c    | 1 -
 src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c | 1 -
 src/gallium/drivers/llvmpipe/lp_texture.c         | 2 --
 src/gallium/drivers/llvmpipe/lp_tile_cache.c      | 2 --
 19 files changed, 26 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
index 2b4bc5c819..7245730350 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_alpha.c
@@ -35,7 +35,6 @@
 
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
-#include "lp_bld_arit.h"
 #include "lp_bld_logic.h"
 #include "lp_bld_flow.h"
 #include "lp_bld_debug.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.c b/src/gallium/drivers/llvmpipe/lp_bld_arit.c
index 9c59677a74..f5d9db70fb 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_arit.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.c
@@ -56,7 +56,6 @@
 #include "lp_bld_intr.h"
 #include "lp_bld_logic.h"
 #include "lp_bld_pack.h"
-#include "lp_bld_debug.h"
 #include "lp_bld_arit.h"
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
index 9511299d55..5d5ca7a5d4 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_blend_soa.c
@@ -71,7 +71,6 @@
 #include "pipe/p_state.h"
 
 #include "lp_bld_type.h"
-#include "lp_bld_const.h"
 #include "lp_bld_arit.h"
 #include "lp_bld_blend.h"
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.c b/src/gallium/drivers/llvmpipe/lp_bld_conv.c
index 9935209437..ebf554cd04 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_conv.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.c
@@ -63,11 +63,9 @@
 
 #include "util/u_debug.h"
 #include "util/u_math.h"
-#include "util/u_cpu_detect.h"
 
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
-#include "lp_bld_intr.h"
 #include "lp_bld_arit.h"
 #include "lp_bld_pack.h"
 #include "lp_bld_conv.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c b/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
index 10e82f120b..dfa080b853 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_format_aos.c
@@ -38,7 +38,6 @@
 
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
-#include "lp_bld_logic.h"
 #include "lp_bld_swizzle.h"
 #include "lp_bld_format.h"
 
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
index 0dea2cd4c8..e722bca1a6 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_tgsi_soa.c
@@ -47,13 +47,11 @@
 #include "tgsi/tgsi_exec.h"
 #include "lp_bld_type.h"
 #include "lp_bld_const.h"
-#include "lp_bld_intr.h"
 #include "lp_bld_arit.h"
 #include "lp_bld_logic.h"
 #include "lp_bld_swizzle.h"
 #include "lp_bld_flow.h"
 #include "lp_bld_tgsi.h"
-#include "lp_bld_debug.h"
 
 
 #define LP_MAX_TEMPS 256
diff --git a/src/gallium/drivers/llvmpipe/lp_draw_arrays.c b/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
index 0aa13a1fc6..a43e438064 100644
--- a/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
+++ b/src/gallium/drivers/llvmpipe/lp_draw_arrays.c
@@ -33,8 +33,6 @@
 
 #include "pipe/p_defines.h"
 #include "pipe/p_context.h"
-#include "pipe/internal/p_winsys_screen.h"
-#include "pipe/p_inlines.h"
 #include "util/u_prim.h"
 
 #include "lp_buffer.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_flush.c b/src/gallium/drivers/llvmpipe/lp_flush.c
index cd8381fe30..6c81012e84 100644
--- a/src/gallium/drivers/llvmpipe/lp_flush.c
+++ b/src/gallium/drivers/llvmpipe/lp_flush.c
@@ -37,8 +37,6 @@
 #include "lp_surface.h"
 #include "lp_state.h"
 #include "lp_tile_cache.h"
-#include "lp_tex_cache.h"
-#include "lp_winsys.h"
 
 
 void
diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c
index 13535dd638..b5aa7d680f 100644
--- a/src/gallium/drivers/llvmpipe/lp_jit.c
+++ b/src/gallium/drivers/llvmpipe/lp_jit.c
@@ -39,7 +39,6 @@
 #include "util/u_cpu_detect.h"
 #include "lp_screen.h"
 #include "lp_bld_intr.h"
-#include "lp_bld_misc.h"
 #include "lp_jit.h"
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c
index ffcbc9a379..95bc66e624 100644
--- a/src/gallium/drivers/llvmpipe/lp_setup.c
+++ b/src/gallium/drivers/llvmpipe/lp_setup.c
@@ -40,7 +40,6 @@
 #include "draw/draw_private.h"
 #include "draw/draw_vertex.h"
 #include "pipe/p_shader_tokens.h"
-#include "pipe/p_thread.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "lp_bld_debug.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 8e2aae40af..2297cbb76f 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -62,7 +62,6 @@
 #include "util/u_memory.h"
 #include "util/u_format.h"
 #include "util/u_debug_dump.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "pipe/p_shader_tokens.h"
 #include "draw/draw_context.h"
 #include "tgsi/tgsi_dump.h"
@@ -85,7 +84,6 @@
 #include "lp_context.h"
 #include "lp_buffer.h"
 #include "lp_state.h"
-#include "lp_quad.h"
 #include "lp_tex_sample.h"
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_state_surface.c b/src/gallium/drivers/llvmpipe/lp_state_surface.c
index c06ce8b75c..9f74585674 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_surface.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_surface.c
@@ -30,7 +30,6 @@
 
 #include "lp_context.h"
 #include "lp_state.h"
-#include "lp_surface.h"
 #include "lp_tile_cache.h"
 
 #include "draw/draw_context.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_state_vertex.c b/src/gallium/drivers/llvmpipe/lp_state_vertex.c
index 1a17631a4c..57ac25ea0c 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_vertex.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_vertex.c
@@ -31,7 +31,6 @@
 
 #include "lp_context.h"
 #include "lp_state.h"
-#include "lp_surface.h"
 
 #include "draw/draw_context.h"
 
diff --git a/src/gallium/drivers/llvmpipe/lp_test_blend.c b/src/gallium/drivers/llvmpipe/lp_test_blend.c
index 29fff91981..411e99d06c 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_blend.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_blend.c
@@ -38,7 +38,6 @@
 
 
 #include "lp_bld_type.h"
-#include "lp_bld_arit.h"
 #include "lp_bld_blend.h"
 #include "lp_bld_debug.h"
 #include "lp_test.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_cache.c b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
index 773e848242..b28f8dc5da 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_cache.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_cache.c
@@ -37,7 +37,6 @@
 #include "util/u_tile.h"
 #include "util/u_format.h"
 #include "lp_context.h"
-#include "lp_surface.h"
 #include "lp_texture.h"
 #include "lp_tex_cache.h"
 
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
index 699394c0de..d1f5d9505d 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
@@ -35,7 +35,6 @@
 
 #include "lp_context.h"
 #include "lp_quad.h"
-#include "lp_surface.h"
 #include "lp_texture.h"
 #include "lp_tex_sample.h"
 #include "lp_tex_cache.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c
index d2a6ae21f5..5138ccf7c9 100644
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c
+++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c
@@ -44,7 +44,6 @@
 #include "pipe/p_shader_tokens.h"
 #include "lp_bld_debug.h"
 #include "lp_bld_type.h"
-#include "lp_bld_intr.h"
 #include "lp_bld_sample.h"
 #include "lp_bld_tgsi.h"
 #include "lp_state.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index a00f2495df..948e3de34d 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -33,14 +33,12 @@
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
 #include "lp_context.h"
 #include "lp_state.h"
 #include "lp_texture.h"
-#include "lp_tex_cache.h"
 #include "lp_screen.h"
 #include "lp_winsys.h"
 
diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.c b/src/gallium/drivers/llvmpipe/lp_tile_cache.c
index ec3e002d62..a555149730 100644
--- a/src/gallium/drivers/llvmpipe/lp_tile_cache.c
+++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.c
@@ -38,8 +38,6 @@
 #include "util/u_tile.h"
 #include "util/u_rect.h"
 #include "lp_context.h"
-#include "lp_surface.h"
-#include "lp_texture.h"
 #include "lp_tile_soa.h"
 #include "lp_tile_cache.h"
 
-- 
cgit v1.2.3


From 0ae076bf40782c48b1b26ca63ed2c349532dd81e Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Thu, 21 Jan 2010 12:12:33 -0800
Subject: svga: Follow buffer usage semantics properly.

It's necessary to download buffers from the host always, except if the
buffer is undefined, because:
- just PIPE_BUFFER_USAGE_CPU_WRITE doesn't guarantee all data is written
  -- old contents may still pierce through
- PIPE_BUFFER_USAGE_DISCARD refers to a range, not the whole buffer, so
  unless we track which parts have been modified and not we still need
  to download the data.
---
 src/gallium/drivers/svga/svga_screen_buffer.c | 10 ++++++----
 src/gallium/drivers/svga/svga_screen_buffer.h |  5 +++++
 2 files changed, 11 insertions(+), 4 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index c014f2ee20..cc2d3cd9e9 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -355,6 +355,8 @@ svga_buffer_upload_flush(struct svga_context *svga,
    sbuf->hw.svga = NULL;
    sbuf->hw.boxes = NULL;
 
+   sbuf->host_written = TRUE;
+
    /* Decrement reference count */
    pipe_buffer_reference((struct pipe_buffer **)&sbuf, NULL);
 }
@@ -436,17 +438,17 @@ svga_buffer_map_range( struct pipe_screen *screen,
    }
    else {
       if(!sbuf->hw.buf) {
-         struct svga_winsys_surface *handle = sbuf->handle;
-
          if(svga_buffer_create_hw_storage(ss, sbuf) != PIPE_OK)
             return NULL;
          
          /* Populate the hardware storage if the host surface pre-existed */
-         if((usage & PIPE_BUFFER_USAGE_CPU_READ) && handle) {
+         if(sbuf->host_written) {
             SVGA3dSurfaceDMAFlags flags;
             enum pipe_error ret;
             struct pipe_fence_handle *fence = NULL;
             
+            assert(sbuf->handle);
+
             SVGA_DBG(DEBUG_DMA|DEBUG_PERF, "dma from sid %p (buffer), bytes %u - %u\n", 
                      sbuf->handle, 0, sbuf->base.size);
 
@@ -478,7 +480,7 @@ svga_buffer_map_range( struct pipe_screen *screen,
          }
       }
       else {
-         if((usage & PIPE_BUFFER_USAGE_CPU_READ) && !sbuf->needs_flush) {
+         if(!(usage & PIPE_BUFFER_USAGE_DISCARD) && !sbuf->needs_flush) {
             /* We already had the hardware storage but we would have to issue
              * a download if we hadn't, so move the buffer to the begginning
              * of the LRU list.
diff --git a/src/gallium/drivers/svga/svga_screen_buffer.h b/src/gallium/drivers/svga/svga_screen_buffer.h
index 5d7af5a7c5..c9bbe37f32 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.h
+++ b/src/gallium/drivers/svga/svga_screen_buffer.h
@@ -135,6 +135,11 @@ struct svga_buffer
     */
    struct svga_winsys_surface *handle;
    
+   /**
+    * Whether the host has been ever written.
+    */
+   boolean host_written;
+
    struct {
       unsigned count;
       boolean writing;
-- 
cgit v1.2.3


From 0bc885188464446dd94542e0af15f3985dcdf684 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Thu, 21 Jan 2010 12:13:27 -0800
Subject: svga: Avoid synchronization issues when doing SW TNL.

---
 src/gallium/drivers/svga/svga_swtnl_backend.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_swtnl_backend.c b/src/gallium/drivers/svga/svga_swtnl_backend.c
index cfbdcf2554..aafb3e26b0 100644
--- a/src/gallium/drivers/svga/svga_swtnl_backend.c
+++ b/src/gallium/drivers/svga/svga_swtnl_backend.c
@@ -122,7 +122,9 @@ svga_vbuf_render_map_vertices( struct vbuf_render *render )
    char *ptr = (char*)pipe_buffer_map(screen,
                                       svga_render->vbuf,
                                       PIPE_BUFFER_USAGE_CPU_WRITE | 
-                                      PIPE_BUFFER_USAGE_FLUSH_EXPLICIT);
+                                      PIPE_BUFFER_USAGE_FLUSH_EXPLICIT |
+                                      PIPE_BUFFER_USAGE_DISCARD |
+                                      PIPE_BUFFER_USAGE_UNSYNCHRONIZED);
    return ptr + svga_render->vbuf_offset;
 }
 
@@ -264,8 +266,8 @@ svga_vbuf_render_draw( struct vbuf_render *render,
       svga_render->ibuf_offset = 0;
    }
 
-   pipe_buffer_write(screen, svga_render->ibuf,
-                     svga_render->ibuf_offset, 2 * nr_indices, indices);
+   pipe_buffer_write_nooverlap(screen, svga_render->ibuf,
+                                 svga_render->ibuf_offset, 2 * nr_indices, indices);
 
 
    /* off to hardware */
-- 
cgit v1.2.3


From 9782d7cdf39c141d816efc48333d6a70ed3f2ac6 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Wed, 20 Jan 2010 03:45:17 -0800
Subject: svga: Don't swap buffers to system memory -- winsys' responsibility.

---
 src/gallium/drivers/svga/svga_screen_buffer.c | 60 ---------------------------
 src/gallium/drivers/svga/svga_screen_buffer.h |  3 --
 2 files changed, 63 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index cc2d3cd9e9..fa1fd1367e 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -121,60 +121,6 @@ svga_buffer_destroy_hw_storage(struct svga_screen *ss, struct svga_buffer *sbuf)
    }
 }
 
-static INLINE enum pipe_error
-svga_buffer_backup(struct svga_screen *ss, struct svga_buffer *sbuf)
-{
-   if (sbuf->hw.buf && sbuf->hw.num_ranges) {
-      void *src;
-
-      if (!sbuf->swbuf)
-	 sbuf->swbuf = align_malloc(sbuf->base.size, sbuf->base.alignment);
-      if (!sbuf->swbuf)
-	 return PIPE_ERROR_OUT_OF_MEMORY;
-
-      src = ss->sws->buffer_map(ss->sws, sbuf->hw.buf,
-				PIPE_BUFFER_USAGE_CPU_READ);
-      if (!src)
-	 return PIPE_ERROR;
-
-      memcpy(sbuf->swbuf, src, sbuf->base.size);
-      ss->sws->buffer_unmap(ss->sws, sbuf->hw.buf);
-   }
-
-   return PIPE_OK;
-}
-
-/**
- * Try to make GMR space available by freeing the hardware storage of 
- * unmapped
- */
-boolean
-svga_buffer_free_cached_hw_storage(struct svga_screen *ss)
-{
-   struct list_head *curr;
-   struct svga_buffer *sbuf;
-   enum pipe_error ret = PIPE_OK;
-
-   curr = ss->cached_buffers.prev;
-   
-   /* free the least recently used buffer's hw storage which is not mapped */
-   do {
-      if(curr == &ss->cached_buffers)
-         return FALSE;
-
-      sbuf = LIST_ENTRY(struct svga_buffer, curr, head);
-      
-      curr = curr->prev;
-      if (sbuf->map.count == 0)
-	 ret = svga_buffer_backup(ss, sbuf);
-
-   } while(sbuf->map.count != 0 || ret != PIPE_OK);
-   
-   svga_buffer_destroy_hw_storage(ss, sbuf);
-   
-   return TRUE;
-}
-
 struct svga_winsys_buffer *
 svga_winsys_buffer_create( struct svga_screen *ss,
                            unsigned alignment, 
@@ -195,12 +141,6 @@ svga_winsys_buffer_create( struct svga_screen *ss,
       svga_screen_flush(ss, NULL);
       buf = sws->buffer_create(sws, alignment, usage, size);
 
-      SVGA_DBG(DEBUG_DMA|DEBUG_PERF, "evicting buffers to find %d bytes GMR\n", 
-               size);
-
-      /* Try evicing all buffer storage */
-      while(!buf && svga_buffer_free_cached_hw_storage(ss))
-         buf = sws->buffer_create(sws, alignment, usage, size);
    }
    
    return buf;
diff --git a/src/gallium/drivers/svga/svga_screen_buffer.h b/src/gallium/drivers/svga/svga_screen_buffer.h
index c9bbe37f32..448ac107c7 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.h
+++ b/src/gallium/drivers/svga/svga_screen_buffer.h
@@ -183,9 +183,6 @@ svga_buffer_handle(struct svga_context *svga,
 void
 svga_context_flush_buffers(struct svga_context *svga);
 
-boolean
-svga_buffer_free_cached_hw_storage(struct svga_screen *ss);
-
 struct svga_winsys_buffer *
 svga_winsys_buffer_create(struct svga_screen *ss,
                           unsigned alignment, 
-- 
cgit v1.2.3


From 50edefec2dd5ec2990ab2cd699478139173d5fd2 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Thu, 21 Jan 2010 14:24:53 -0800
Subject: svga: Remove unused code.

---
 src/gallium/drivers/svga/svga_screen.c        |  2 --
 src/gallium/drivers/svga/svga_screen.h        |  6 ------
 src/gallium/drivers/svga/svga_screen_buffer.c | 26 ++++----------------------
 3 files changed, 4 insertions(+), 30 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index 68c09e80af..cd1ed7bac8 100644
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -391,8 +391,6 @@ svga_screen_create(struct svga_winsys_screen *sws)
    pipe_mutex_init(svgascreen->tex_mutex);
    pipe_mutex_init(svgascreen->swc_mutex);
 
-   LIST_INITHEAD(&svgascreen->cached_buffers);
-   
    svga_screen_cache_init(svgascreen);
 
    return screen;
diff --git a/src/gallium/drivers/svga/svga_screen.h b/src/gallium/drivers/svga/svga_screen.h
index b94ca7fc1c..a009b60720 100644
--- a/src/gallium/drivers/svga/svga_screen.h
+++ b/src/gallium/drivers/svga/svga_screen.h
@@ -68,12 +68,6 @@ struct svga_screen
    pipe_mutex tex_mutex; 
    pipe_mutex swc_mutex; /* Protects the use of swc and dirty_buffers */
    
-   /** 
-    * List of buffers with cached GMR. Ordered from the most recently used to
-    * the least recently used 
-    */
-   struct list_head cached_buffers;
-   
    struct svga_host_surface_cache cache;
 };
 
diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index fa1fd1367e..cc4109dfe8 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -113,11 +113,6 @@ svga_buffer_destroy_hw_storage(struct svga_screen *ss, struct svga_buffer *sbuf)
    if(sbuf->hw.buf) {
       sws->buffer_destroy(sws, sbuf->hw.buf);
       sbuf->hw.buf = NULL;
-      assert(sbuf->head.prev && sbuf->head.next);
-      LIST_DEL(&sbuf->head);
-#ifdef DEBUG
-      sbuf->head.next = sbuf->head.prev = NULL; 
-#endif
    }
 }
 
@@ -166,8 +161,6 @@ svga_buffer_create_hw_storage(struct svga_screen *ss,
          return PIPE_ERROR_OUT_OF_MEMORY;
       
       assert(!sbuf->needs_flush);
-      assert(!sbuf->head.prev && !sbuf->head.next);
-      LIST_ADD(&sbuf->head, &ss->cached_buffers);
    }
    
    return PIPE_OK;
@@ -288,9 +281,10 @@ svga_buffer_upload_flush(struct svga_context *svga,
 
    assert(sbuf->head.prev && sbuf->head.next);
    LIST_DEL(&sbuf->head);
+#ifdef DEBUG
+   sbuf->head.next = sbuf->head.prev = NULL; 
+#endif
    sbuf->needs_flush = FALSE;
-   /* XXX: do we care about cached_buffers any more ?*/
-   LIST_ADD(&sbuf->head, &ss->cached_buffers);
 
    sbuf->hw.svga = NULL;
    sbuf->hw.boxes = NULL;
@@ -419,17 +413,6 @@ svga_buffer_map_range( struct pipe_screen *screen,
             sws->fence_reference(sws, &fence, NULL);
          }
       }
-      else {
-         if(!(usage & PIPE_BUFFER_USAGE_DISCARD) && !sbuf->needs_flush) {
-            /* We already had the hardware storage but we would have to issue
-             * a download if we hadn't, so move the buffer to the begginning
-             * of the LRU list.
-             */
-            assert(sbuf->head.prev && sbuf->head.next);
-            LIST_DEL(&sbuf->head);
-            LIST_ADD(&sbuf->head, &ss->cached_buffers);
-         }
-      }
          
       map = sws->buffer_map(sws, sbuf->hw.buf, usage);
    }
@@ -699,8 +682,7 @@ svga_buffer_handle(struct svga_context *svga,
       assert(sbuf->hw.svga == svga);
 
       sbuf->needs_flush = TRUE;
-      assert(sbuf->head.prev && sbuf->head.next);
-      LIST_DEL(&sbuf->head);
+      assert(!sbuf->head.prev && !sbuf->head.next);
       LIST_ADDTAIL(&sbuf->head, &svga->dirty_buffers);
    }
 
-- 
cgit v1.2.3


From 31e49dc54ee2d8b2c50ed1d19ab973f72a60f8d8 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Thu, 21 Jan 2010 12:12:47 -0800
Subject: svga: Remove duplicate code.

---
 src/gallium/drivers/svga/svga_screen_buffer.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index cc4109dfe8..727fd180c1 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -496,10 +496,8 @@ svga_buffer_destroy( struct pipe_buffer *buf )
    
    assert(!sbuf->needs_flush);
 
-   if(sbuf->handle) {
-      SVGA_DBG(DEBUG_DMA, "release sid %p sz %d\n", sbuf->handle, sbuf->base.size);
-      svga_screen_surface_destroy(ss, &sbuf->key, &sbuf->handle);
-   }
+   if(sbuf->handle)
+      svga_buffer_destroy_host_surface(ss, sbuf);
    
    if(sbuf->hw.buf)
       svga_buffer_destroy_hw_storage(ss, sbuf);
-- 
cgit v1.2.3


From e8b5203fbb7657980a4ba421e14d656b42d2a35f Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 21 Jan 2010 17:24:07 -0800
Subject: svga: Remove unused variable.

---
 src/gallium/drivers/svga/svga_screen_buffer.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_screen_buffer.c b/src/gallium/drivers/svga/svga_screen_buffer.c
index 727fd180c1..719adde27e 100644
--- a/src/gallium/drivers/svga/svga_screen_buffer.c
+++ b/src/gallium/drivers/svga/svga_screen_buffer.c
@@ -244,7 +244,6 @@ static void
 svga_buffer_upload_flush(struct svga_context *svga,
                          struct svga_buffer *sbuf)
 {
-   struct svga_screen *ss = svga_screen(svga->pipe.screen);
    SVGA3dCopyBox *boxes;
    unsigned i;
 
-- 
cgit v1.2.3


From a1c5d8f61df9d9e22bcb0fc37a602ab712958151 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 21 Jan 2010 23:09:06 -0800
Subject: i915g: Remove unnecessary header.

---
 src/gallium/drivers/i915/i915_surface.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i915/i915_surface.c b/src/gallium/drivers/i915/i915_surface.c
index 9b4366daff..793613be7b 100644
--- a/src/gallium/drivers/i915/i915_surface.c
+++ b/src/gallium/drivers/i915/i915_surface.c
@@ -28,7 +28,6 @@
 #include "i915_context.h"
 #include "i915_blit.h"
 #include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
 
 
 /* Assumes all values are within bounds -- no checking at this level -
-- 
cgit v1.2.3


From 76bc66b664c7e135aa9c06cedd94ee1961d68c70 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Sat, 23 Jan 2010 19:35:42 +1000
Subject: r300g: fix up buffer emission ordering.

This fixes the code space checking VBOs, then resetting the space list
and space checking other buffers. it fixes demos/ipers

Signed-off-by: Dave Airlie <airlied@redhat.com>
---
 src/gallium/drivers/r300/r300_emit.c   | 50 +++++++++++++++++++---------------
 src/gallium/drivers/r300/r300_emit.h   |  2 ++
 src/gallium/drivers/r300/r300_render.c |  4 +++
 3 files changed, 34 insertions(+), 22 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 921170aef1..cc40abf168 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -1014,32 +1014,12 @@ static void r300_flush_pvs(struct r300_context* r300)
     END_CS;
 }
 
-/* Emit all dirty state. */
-void r300_emit_dirty_state(struct r300_context* r300)
+void r300_emit_buffer_validate(struct r300_context *r300)
 {
-    struct r300_screen* r300screen = r300_screen(r300->context.screen);
     struct r300_texture* tex;
-    struct r300_atom* atom;
-    unsigned i, dwords = 1024;
-    int dirty_tex = 0;
+    unsigned i;
     boolean invalid = FALSE;
 
-    /* Check the required number of dwords against the space remaining in the
-     * current CS object. If we need more, then flush. */
-
-    foreach(atom, &r300->atom_list) {
-        if (atom->dirty || atom->always_dirty) {
-            dwords += atom->size;
-        }
-    }
-
-    /* Make sure we have at least 2*1024 spare dwords. */
-    /* XXX It would be nice to know the number of dwords we really need to
-     * XXX emit. */
-    if (!r300->winsys->check_cs(r300->winsys, dwords)) {
-        r300->context.flush(&r300->context, 0, NULL);
-    }
-
     /* Clean out BOs. */
     r300->winsys->reset_bos(r300->winsys);
 
@@ -1103,6 +1083,32 @@ validate:
         invalid = TRUE;
         goto validate;
     }
+}
+
+/* Emit all dirty state. */
+void r300_emit_dirty_state(struct r300_context* r300)
+{
+    struct r300_screen* r300screen = r300_screen(r300->context.screen);
+    struct r300_atom* atom;
+    unsigned i, dwords = 1024;
+    int dirty_tex = 0;
+
+    /* Check the required number of dwords against the space remaining in the
+     * current CS object. If we need more, then flush. */
+
+    foreach(atom, &r300->atom_list) {
+        if (atom->dirty || atom->always_dirty) {
+            dwords += atom->size;
+        }
+    }
+
+    /* Make sure we have at least 2*1024 spare dwords. */
+    /* XXX It would be nice to know the number of dwords we really need to
+     * XXX emit. */
+    if (!r300->winsys->check_cs(r300->winsys, dwords)) {
+        r300->context.flush(&r300->context, 0, NULL);
+	r300_emit_buffer_validate(r300);
+    }
 
     if (r300->dirty_state & R300_NEW_QUERY) {
         r300_emit_query_start(r300);
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index 05a6bfeae8..2f3d013f5e 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -95,4 +95,6 @@ void r300_flush_textures(struct r300_context* r300);
 /* Emit all dirty state. */
 void r300_emit_dirty_state(struct r300_context* r300);
 
+void r300_emit_buffer_validate(struct r300_context *r300);
+
 #endif /* R300_EMIT_H */
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 90de062bcd..68abfbe208 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -311,6 +311,8 @@ void r300_draw_range_elements(struct pipe_context* pipe,
         indexSize = 2;
     }
 
+    r300_emit_buffer_validate(r300->winsys);
+
     if (!r300->winsys->add_buffer(r300->winsys, indexBuffer,
                                   RADEON_GEM_DOMAIN_GTT, 0)) {
         goto cleanup;
@@ -361,6 +363,8 @@ void r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
 
     r300_update_derived_state(r300);
 
+    r300_emit_buffer_validate(r300);
+
     if (!r300_setup_vertex_buffers(r300)) {
         return;
     }
-- 
cgit v1.2.3


From 3b002c9f75dd76e41552306516545080719fc065 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Sat, 23 Jan 2010 20:07:08 +1000
Subject: r300: fix wrong call in last commit

---
 src/gallium/drivers/r300/r300_render.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 68abfbe208..cb2d8d4537 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -311,7 +311,7 @@ void r300_draw_range_elements(struct pipe_context* pipe,
         indexSize = 2;
     }
 
-    r300_emit_buffer_validate(r300->winsys);
+    r300_emit_buffer_validate(r300);
 
     if (!r300->winsys->add_buffer(r300->winsys, indexBuffer,
                                   RADEON_GEM_DOMAIN_GTT, 0)) {
-- 
cgit v1.2.3


From 232ddfec24d2e707c4e374d817c559cc6415d609 Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Sun, 24 Jan 2010 14:58:45 +1000
Subject: r300g: fix flushing harder.

I swear ipers runs here after this commit.

After we flush we need to redo the space check
---
 src/gallium/drivers/r300/r300_emit.c   | 3 ++-
 src/gallium/drivers/r300/r300_render.c | 4 ++--
 2 files changed, 4 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index cc40abf168..e08e6c2292 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -1105,9 +1105,10 @@ void r300_emit_dirty_state(struct r300_context* r300)
     /* Make sure we have at least 2*1024 spare dwords. */
     /* XXX It would be nice to know the number of dwords we really need to
      * XXX emit. */
+again:
     if (!r300->winsys->check_cs(r300->winsys, dwords)) {
         r300->context.flush(&r300->context, 0, NULL);
-	r300_emit_buffer_validate(r300);
+	goto again;
     }
 
     if (r300->dirty_state & R300_NEW_QUERY) {
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index cb2d8d4537..677031ef04 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -302,6 +302,8 @@ void r300_draw_range_elements(struct pipe_context* pipe,
 
     r300_update_derived_state(r300);
 
+    r300_emit_buffer_validate(r300);
+
     if (!r300_setup_vertex_buffers(r300)) {
         return;
     }
@@ -311,8 +313,6 @@ void r300_draw_range_elements(struct pipe_context* pipe,
         indexSize = 2;
     }
 
-    r300_emit_buffer_validate(r300);
-
     if (!r300->winsys->add_buffer(r300->winsys, indexBuffer,
                                   RADEON_GEM_DOMAIN_GTT, 0)) {
         goto cleanup;
-- 
cgit v1.2.3


From e9bb06441eb4a50e9927dcc09289ec63b0e2cf0a Mon Sep 17 00:00:00 2001
From: Dave Airlie <airlied@redhat.com>
Date: Sun, 24 Jan 2010 15:37:23 +1000
Subject: r300g: avoid the use of goto

I was being lazy on Sunday morning, Droste on irc called me on it ;-)
---
 src/gallium/drivers/r300/r300_emit.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index e08e6c2292..e07185b778 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -1105,10 +1105,8 @@ void r300_emit_dirty_state(struct r300_context* r300)
     /* Make sure we have at least 2*1024 spare dwords. */
     /* XXX It would be nice to know the number of dwords we really need to
      * XXX emit. */
-again:
-    if (!r300->winsys->check_cs(r300->winsys, dwords)) {
+    while (!r300->winsys->check_cs(r300->winsys, dwords)) {
         r300->context.flush(&r300->context, 0, NULL);
-	goto again;
     }
 
     if (r300->dirty_state & R300_NEW_QUERY) {
-- 
cgit v1.2.3


From d6fd31a8d47b6cc70b5e64757e1272ccc6594350 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Sun, 24 Jan 2010 15:13:17 +0100
Subject: nv50: implement render_condition

Still not handling queries on multiple contexts though.
---
 src/gallium/drivers/nv50/nv50_query.c | 40 +++++++++++++++++++++++++++++++----
 1 file changed, 36 insertions(+), 4 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c
index 5a4ab3508b..9eba4c9611 100644
--- a/src/gallium/drivers/nv50/nv50_query.c
+++ b/src/gallium/drivers/nv50/nv50_query.c
@@ -48,7 +48,7 @@ nv50_query_create(struct pipe_context *pipe, unsigned type)
 	assert (q->type == PIPE_QUERY_OCCLUSION_COUNTER);
 	q->type = type;
 
-	ret = nouveau_bo_new(dev, NOUVEAU_BO_VRAM | NOUVEAU_BO_MAP, 256,
+	ret = nouveau_bo_new(dev, NOUVEAU_BO_GART | NOUVEAU_BO_MAP, 256,
 			     16, &q->bo);
 	if (ret) {
 		FREE(q);
@@ -95,11 +95,13 @@ nv50_query_end(struct pipe_context *pipe, struct pipe_query *pq)
 
 	MARK_RING (chan, 5, 2); /* flush on lack of space or relocs */
 	BEGIN_RING(chan, tesla, NV50TCL_QUERY_ADDRESS_HIGH, 4);
-	OUT_RELOCh(chan, q->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
-	OUT_RELOCl(chan, q->bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	OUT_RELOCh(chan, q->bo, 0, NOUVEAU_BO_GART | NOUVEAU_BO_WR);
+	OUT_RELOCl(chan, q->bo, 0, NOUVEAU_BO_GART | NOUVEAU_BO_WR);
 	OUT_RING  (chan, 0x00000000);
 	OUT_RING  (chan, 0x0100f002);
-	FIRE_RING (chan);
+
+	BEGIN_RING(chan, tesla, NV50TCL_SAMPLECNT_ENABLE, 1);
+	OUT_RING  (chan, 0);
 }
 
 static boolean
@@ -123,6 +125,35 @@ nv50_query_result(struct pipe_context *pipe, struct pipe_query *pq,
 	return q->ready;
 }
 
+static void
+nv50_render_condition(struct pipe_context *pipe,
+		      struct pipe_query *pq, uint mode)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct nouveau_channel *chan = nv50->screen->base.channel;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_query *q;
+
+	if (!pq) {
+		BEGIN_RING(chan, tesla, NV50TCL_COND_MODE, 1);
+		OUT_RING  (chan, NV50TCL_COND_MODE_ALWAYS);
+		return;
+	}
+	q = nv50_query(pq);
+
+	if (mode == PIPE_RENDER_COND_WAIT ||
+	    mode == PIPE_RENDER_COND_BY_REGION_WAIT) {
+		/* XXX: big fence, FIFO semaphore might be better */
+		BEGIN_RING(chan, tesla, 0x0110, 1);
+		OUT_RING  (chan, 0);
+	}
+
+	BEGIN_RING(chan, tesla, NV50TCL_COND_ADDRESS_HIGH, 3);
+	OUT_RELOCh(chan, q->bo, 0, NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RELOCl(chan, q->bo, 0, NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RING  (chan, NV50TCL_COND_MODE_RES);
+}
+
 void
 nv50_init_query_functions(struct nv50_context *nv50)
 {
@@ -131,4 +162,5 @@ nv50_init_query_functions(struct nv50_context *nv50)
 	nv50->pipe.begin_query = nv50_query_begin;
 	nv50->pipe.end_query = nv50_query_end;
 	nv50->pipe.get_query_result = nv50_query_result;
+	nv50->pipe.render_condition = nv50_render_condition;
 }
-- 
cgit v1.2.3


From 49dde55b96f717171a9dc6facd800e1c4be6eedf Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Sat, 23 Jan 2010 18:35:27 +0100
Subject: nv50: only flush texture caches on st request

---
 src/gallium/drivers/nv50/nv50_context.c | 5 +++++
 src/gallium/drivers/nv50/nv50_tex.c     | 3 ---
 2 files changed, 5 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c
index 1e69746322..e2198b189d 100644
--- a/src/gallium/drivers/nv50/nv50_context.c
+++ b/src/gallium/drivers/nv50/nv50_context.c
@@ -34,6 +34,11 @@ nv50_flush(struct pipe_context *pipe, unsigned flags,
 	struct nv50_context *nv50 = nv50_context(pipe);
 	struct nouveau_channel *chan = nv50->screen->base.channel;
 
+	if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
+		BEGIN_RING(chan, nv50->screen->tesla, 0x1338, 1);
+		OUT_RING  (chan, 0x20);
+	}
+
 	if (flags & PIPE_FLUSH_FRAME)
 		FIRE_RING(chan);
 }
diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c
index 871536dca9..9f1a171303 100644
--- a/src/gallium/drivers/nv50/nv50_tex.c
+++ b/src/gallium/drivers/nv50/nv50_tex.c
@@ -220,11 +220,8 @@ nv50_tex_validate(struct nv50_context *nv50)
 		return;
 	}
 
-	/* not sure if the following really do what I think: */
 	so_method(so, tesla, 0x1330, 1); /* flush TIC */
 	so_data  (so, 0);
-	so_method(so, tesla, 0x1338, 1); /* flush texture caches */
-	so_data  (so, 0x20);
 
 	so_ref(so, &nv50->state.tic_upload);
 	so_ref(NULL, &so);
-- 
cgit v1.2.3


From 833acea8f6bcbed8b477eb1a1d897127afb0e73e Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Sun, 24 Jan 2010 14:22:41 +0100
Subject: nv50: fix handling of FragCoord input

---
 src/gallium/drivers/nv50/nv50_program.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index 20db51070f..cd2fd0a076 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -3728,13 +3728,21 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 		copy_semantic_info(p);
 	} else
 	if (p->type == PIPE_SHADER_FRAGMENT) {
-		int rid, aid, base;
+		int rid, aid;
 		unsigned n = 0, m = pc->attr_nr - flat_nr;
 
 		pc->allow32 = TRUE;
 
-		base = (TGSI_SEMANTIC_POSITION ==
-			p->info.input_semantic_name[0]) ? 0 : 1;
+		/* do we read FragCoord ? */
+		if (pc->attr_nr &&
+		    p->info.input_semantic_name[0] == TGSI_SEMANTIC_POSITION) {
+			/* select FCRD components we want accessible */
+			for (c = 0; c < 4; ++c)
+				if (pc->attr[c].acc)
+					p->cfg.regs[1] |= 1 << (24 + c);
+			aid = 0;
+		} else /* offset by 1 if FCRD.w is needed for pinterp */
+			aid = popcnt4(p->cfg.regs[1] >> 24);
 
 		/* non-flat interpolants have to be mapped to
 		 * the lower hardware IDs, so sort them:
@@ -3750,12 +3758,6 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 		}
 		copy_semantic_info(p);
 
-		if (!base) /* set w-coordinate mask from perspective interp */
-			p->cfg.in[0].mask |= p->cfg.regs[1] >> 24;
-
-		aid = popcnt4( /* if fcrd isn't contained in cfg.io */
-			base ? (p->cfg.regs[1] >> 24) : p->cfg.in[0].mask);
-
 		for (n = 0; n < pc->attr_nr; ++n) {
 			p->cfg.in[n].hw = rid = aid;
 			i = p->cfg.in[n].id;
@@ -3777,9 +3779,6 @@ nv50_program_tx_prep(struct nv50_pc *pc)
 			aid += popcnt4(p->cfg.in[n].mask);
 		}
 
-		if (!base)
-			p->cfg.regs[1] |= p->cfg.in[0].mask << 24;
-
 		m = popcnt4(p->cfg.regs[1] >> 24);
 
 		/* set count of non-position inputs and of non-flat
-- 
cgit v1.2.3


From 283b21d02bff3d4bd6127f6b642331e06db41ace Mon Sep 17 00:00:00 2001
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Date: Sun, 17 Jan 2010 23:50:50 +0100
Subject: nouveau: fix winsys object leak

---
 src/gallium/drivers/nouveau/nouveau_screen.c         | 2 ++
 src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c | 9 +++++++++
 2 files changed, 11 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nouveau/nouveau_screen.c b/src/gallium/drivers/nouveau/nouveau_screen.c
index 7ebc94ed6c..1ad539d285 100644
--- a/src/gallium/drivers/nouveau/nouveau_screen.c
+++ b/src/gallium/drivers/nouveau/nouveau_screen.c
@@ -260,6 +260,8 @@ nouveau_screen_init(struct nouveau_screen *screen, struct nouveau_device *dev)
 void
 nouveau_screen_fini(struct nouveau_screen *screen)
 {
+	struct pipe_winsys *ws = screen->base.winsys;
 	nouveau_channel_free(&screen->channel);
+	ws->destroy(ws);
 }
 
diff --git a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
index c9f39d815d..29afff6faf 100644
--- a/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
+++ b/src/gallium/winsys/drm/nouveau/drm/nouveau_drm_api.c
@@ -54,6 +54,14 @@ static struct dri1_api nouveau_dri1_api = {
 	nouveau_dri1_front_surface,
 };
 
+static void
+nouveau_drm_destroy_winsys(struct pipe_winsys *s)
+{
+	struct nouveau_winsys *nv_winsys = nouveau_winsys(s);
+	FREE(nv_winsys->pctx);
+	FREE(nv_winsys);
+}
+
 static struct pipe_screen *
 nouveau_drm_create_screen(struct drm_api *api, int fd,
 			  struct drm_create_screen_arg *arg)
@@ -105,6 +113,7 @@ nouveau_drm_create_screen(struct drm_api *api, int fd,
 		return NULL;
 	}
 	ws = &nvws->base;
+	ws->destroy = nouveau_drm_destroy_winsys;
 
 	nvws->pscreen = init(ws, dev);
 	if (!nvws->pscreen) {
-- 
cgit v1.2.3


From 0ef781c1ae04451ac4f184187e900f872cad5f94 Mon Sep 17 00:00:00 2001
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Date: Wed, 20 Jan 2010 23:27:07 +0100
Subject: nv50: fix crash in nv50_pre_pipebuffer_map (nv50_screen->cur_ctx)

nv50_pre_pipebuffer_map references screen->cur_ctx which points
to freed memory after the context is destroyed.
This crash is easily triggerable by progs/xdemos/glxcontexts.
---
 src/gallium/drivers/nv50/nv50_context.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c
index e2198b189d..5c705ccc8f 100644
--- a/src/gallium/drivers/nv50/nv50_context.c
+++ b/src/gallium/drivers/nv50/nv50_context.c
@@ -86,6 +86,10 @@ nv50_destroy(struct pipe_context *pipe)
 		so_ref(NULL, &nv50->state.vtxattr);
 
 	draw_destroy(nv50->draw);
+
+	if (nv50->screen->cur_ctx == nv50)
+		nv50->screen->cur_ctx = NULL;
+
 	FREE(nv50);
 }
 
-- 
cgit v1.2.3


From 9e1550dbafe1b28b81dc15822762e6cfe53e3a4b Mon Sep 17 00:00:00 2001
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Date: Sun, 24 Jan 2010 19:27:59 +0100
Subject: nv50: fix uninitialized variable in nv50_revdep_reorder

"unsafe" is never initialized, but used
(found by valgrind)
---
 src/gallium/drivers/nv50/nv50_program.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index cd2fd0a076..e74a526c62 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -3283,7 +3283,7 @@ prep_inspect_insn(struct nv50_pc *pc, const struct tgsi_full_instruction *insn)
 static unsigned
 nv50_revdep_reorder(unsigned m[4], unsigned rdep[4])
 {
-	unsigned i, c, x, unsafe;
+	unsigned i, c, x, unsafe = 0;
 
 	for (c = 0; c < 4; c++)
 		m[c] = c;
-- 
cgit v1.2.3


From b7a002df1e06504ff1ed0dc005c9b10aa4be54f7 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 00:57:04 +0100
Subject: r300g,radeong: set tiling flags in libdrm

---
 src/gallium/drivers/r300/r300_texture.c            |  7 +++++++
 src/gallium/winsys/drm/radeon/core/radeon_buffer.c | 22 ++++++++++++++++++++++
 src/gallium/winsys/drm/radeon/core/radeon_winsys.h |  6 ++++++
 3 files changed, 35 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c
index 1f73f74c26..67bf8ce13f 100644
--- a/src/gallium/drivers/r300/r300_texture.c
+++ b/src/gallium/drivers/r300/r300_texture.c
@@ -30,6 +30,8 @@
 #include "r300_texture.h"
 #include "r300_screen.h"
 
+#include "radeon_winsys.h"
+
 #define TILE_WIDTH 0
 #define TILE_HEIGHT 1
 
@@ -209,6 +211,7 @@ static struct pipe_texture*
 {
     struct r300_texture* tex = CALLOC_STRUCT(r300_texture);
     struct r300_screen* rscreen = r300_screen(screen);
+    struct radeon_winsys* winsys = (struct radeon_winsys*)screen->winsys;
 
     if (!tex) {
         return NULL;
@@ -225,6 +228,10 @@ static struct pipe_texture*
     tex->buffer = screen->buffer_create(screen, 2048,
                                         PIPE_BUFFER_USAGE_PIXEL,
                                         tex->size);
+    winsys->buffer_set_tiling(winsys, tex->buffer,
+                              tex->pitch[0],
+                              tex->microtile != R300_BUFFER_LINEAR,
+                              tex->macrotile != R300_BUFFER_LINEAR);
 
     if (!tex->buffer) {
         FREE(tex);
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
index b020ff38fa..25e1cdcdb6 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
@@ -202,6 +202,26 @@ static void radeon_buffer_unmap(struct pipe_winsys *ws,
     }
 }
 
+static void radeon_buffer_set_tiling(struct radeon_winsys *ws,
+                                     struct pipe_buffer *buffer,
+                                     uint32_t pitch,
+                                     boolean microtiled,
+                                     boolean macrotiled)
+{
+    struct radeon_pipe_buffer *radeon_buffer =
+        (struct radeon_pipe_buffer*)buffer;
+    uint32_t flags = 0;
+
+    if (microtiled) {
+        flags |= RADEON_BO_FLAGS_MICRO_TILE;
+    }
+    if (macrotiled) {
+        flags |= RADEON_BO_FLAGS_MACRO_TILE;
+    }
+
+    radeon_bo_set_tiling(radeon_buffer->bo, flags, pitch);
+}
+
 static void radeon_fence_reference(struct pipe_winsys *ws,
                                    struct pipe_fence_handle **ptr,
                                    struct pipe_fence_handle *pfence)
@@ -304,5 +324,7 @@ struct radeon_winsys* radeon_pipe_winsys(int fd)
 
     radeon_ws->base.get_name = radeon_get_name;
 
+    radeon_ws->buffer_set_tiling = radeon_buffer_set_tiling;
+
     return radeon_ws;
 }
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
index 9edc9e038c..864082b99b 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
@@ -100,6 +100,12 @@ struct radeon_winsys {
 			 void (*flush_cb)(void *), void *data);
 
     void (*reset_bos)(struct radeon_winsys *winsys);
+
+    void (*buffer_set_tiling)(struct radeon_winsys* winsys,
+                              struct pipe_buffer* buffer,
+                              uint32_t pitch,
+                              boolean microtiled,
+                              boolean macrotiled);
 };
 
 #endif
-- 
cgit v1.2.3


From 9f8ec533123f07f29b084e8a46fc35c498b3a670 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Sun, 17 Jan 2010 23:13:51 +0100
Subject: r300g: fix zero-area scissor

---
 src/gallium/drivers/r300/r300_emit.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index e07185b778..36d2c64b58 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -686,6 +686,22 @@ void r300_emit_scissor_state(struct r300_context* r300, void* state)
         maxy = MIN2(maxy, scissor->maxy);
     }
 
+    /* Special case for zero-area scissor.
+     *
+     * We can't allow the variables maxx and maxy to be zero because they are
+     * subtracted from later in the code, which would cause emitting ~0 and
+     * making the kernel checker angry.
+     *
+     * Let's consider we change maxx and maxy to 1, which is effectively
+     * a one-pixel area. We must then change minx and miny to a number which is
+     * greater than 1 to get the zero area back. */
+    if (!maxx || !maxy) {
+        minx = 2;
+        miny = 2;
+        maxx = 1;
+        maxy = 1;
+    }
+
     if (r300screen->caps->is_r500) {
         top_left =
             (minx << R300_SCISSORS_X_SHIFT) |
-- 
cgit v1.2.3


From 112239e9a66a155d36fe2ad0ab130e6f26eff298 Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Mon, 18 Jan 2010 00:15:52 +0100
Subject: r300g,radeong: finish and enable the immediate mode

Nearly 100% performance increase in glxgears.
---
 src/gallium/drivers/r300/r300_emit.c               |  22 ----
 src/gallium/drivers/r300/r300_render.c             | 141 ++++++++++++++++-----
 src/gallium/drivers/r300/r300_state.c              |  22 ++++
 src/gallium/winsys/drm/radeon/core/radeon_buffer.c |  83 ++++++++++--
 src/gallium/winsys/drm/radeon/core/radeon_buffer.h |   2 +
 src/gallium/winsys/drm/radeon/core/radeon_drm.h    |   2 +-
 src/gallium/winsys/drm/radeon/core/radeon_r300.c   |   8 +-
 src/gallium/winsys/drm/radeon/core/radeon_winsys.h |   6 +
 8 files changed, 219 insertions(+), 67 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 36d2c64b58..badbf3715c 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -772,22 +772,6 @@ void r300_emit_texture(struct r300_context* r300,
     END_CS;
 }
 
-static boolean r300_validate_aos(struct r300_context *r300)
-{
-    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
-    struct pipe_vertex_element *velem = r300->vertex_element;
-    int i;
-
-    /* Check if formats and strides are aligned to the size of DWORD. */
-    for (i = 0; i < r300->vertex_element_count; i++) {
-        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
-            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
-            return FALSE;
-        }
-    }
-    return TRUE;
-}
-
 void r300_emit_aos(struct r300_context* r300, unsigned offset)
 {
     struct pipe_vertex_buffer *vb1, *vb2, *vbuf = r300->vertex_buffer;
@@ -797,12 +781,6 @@ void r300_emit_aos(struct r300_context* r300, unsigned offset)
     unsigned packet_size = (aos_count * 3 + 1) / 2;
     CS_LOCALS(r300);
 
-    /* XXX Move this checking to a more approriate place. */
-    if (!r300_validate_aos(r300)) {
-        /* XXX We should fallback using Draw. */
-        assert(0);
-    }
-
     BEGIN_CS(2 + packet_size + aos_count * 2);
     OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size);
     OUT_CS(aos_count);
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 677031ef04..7f095bffe7 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -28,6 +28,7 @@
 
 #include "pipe/p_inlines.h"
 
+#include "util/u_format.h"
 #include "util/u_memory.h"
 #include "util/u_prim.h"
 
@@ -114,20 +115,53 @@ static uint32_t r300_provoking_vertex_fixes(struct r300_context *r300,
     return color_control;
 }
 
-static void r300_emit_draw_immediate(struct r300_context *r300,
-                                     unsigned mode,
-                                     unsigned start,
-                                     unsigned count)
+
+static void r300_emit_draw_arrays_immediate(struct r300_context *r300,
+                                            unsigned mode,
+                                            unsigned start,
+                                            unsigned count)
 {
-    struct pipe_buffer* vbo = r300->vertex_buffer[0].buffer;
-    unsigned vertex_size = r300->vertex_buffer[0].stride / sizeof(float);
-    unsigned i;
-    uint32_t* map;
+    struct pipe_vertex_element* velem;
+    struct pipe_vertex_buffer* vbuf;
+    unsigned vertex_element_count = r300->vertex_element_count;
+    unsigned i, v, vbi, dw, elem_offset;
+
+    /* Size of the vertex, in dwords. */
+    unsigned vertex_size = 0;
+
+    /* Offsets of the attribute, in dwords, from the start of the vertex. */
+    unsigned offset[PIPE_MAX_ATTRIBS];
+
+    /* Size of the vertex element, in dwords. */
+    unsigned size[PIPE_MAX_ATTRIBS];
+
+    /* Stride to the same attrib in the next vertex in the vertex buffer,
+     * in dwords. */
+    unsigned stride[PIPE_MAX_ATTRIBS];
+
+    /* Mapped vertex buffers. */
+    uint32_t* map[PIPE_MAX_ATTRIBS] = {0};
+
     CS_LOCALS(r300);
 
-    map = (uint32_t*)pipe_buffer_map_range(r300->context.screen, vbo,
-            start * vertex_size, count * vertex_size,
-            PIPE_BUFFER_USAGE_CPU_READ);
+    /* Calculate the vertex size, offsets, strides etc. and map the buffers. */
+    for (i = 0; i < vertex_element_count; i++) {
+        velem = &r300->vertex_element[i];
+        offset[i] = velem->src_offset >> 2;
+        size[i] = util_format_get_blocksize(velem->src_format) >> 2;
+        vertex_size += size[i];
+        vbi = velem->vertex_buffer_index;
+
+        /* Map the buffer. */
+        if (!map[vbi]) {
+            vbuf = &r300->vertex_buffer[vbi];
+            map[vbi] = (uint32_t*)pipe_buffer_map(r300->context.screen,
+                                                  vbuf->buffer,
+                                                  PIPE_BUFFER_USAGE_CPU_READ);
+            map[vbi] += vbuf->buffer_offset >> 2;
+            stride[vbi] = vbuf->stride >> 2;
+        }
+    }
 
     BEGIN_CS(10 + count * vertex_size);
     OUT_CS_REG(R300_GA_COLOR_CONTROL,
@@ -138,18 +172,31 @@ static void r300_emit_draw_immediate(struct r300_context *r300,
     OUT_CS_PKT3(R300_PACKET3_3D_DRAW_IMMD_2, count * vertex_size);
     OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED | (count << 16) |
             r300_translate_primitive(mode));
-    //debug_printf("r300: Immd %d verts, %d attrs\n", count, vertex_size);
-    for (i = 0; i < count * vertex_size; i++) {
-        if (i % vertex_size == 0) {
-            //debug_printf("r300: -- vert --\n");
+
+    /* Emit vertices. */
+    for (v = 0; v < count; v++) {
+        for (i = 0; i < vertex_element_count; i++) {
+            velem = &r300->vertex_element[i];
+            vbi = velem->vertex_buffer_index;
+            elem_offset = offset[i] + stride[vbi] * (v + start);
+
+            for (dw = 0; dw < size[i]; dw++) {
+                OUT_CS(map[vbi][elem_offset + dw]);
+            }
         }
-        //debug_printf("r300: 0x%08x\n", *map);
-        OUT_CS(*map);
-        map++;
     }
     END_CS;
 
-    pipe_buffer_unmap(r300->context.screen, vbo);
+    /* Unmap buffers. */
+    for (i = 0; i < vertex_element_count; i++) {
+        vbi = r300->vertex_element[i].vertex_buffer_index;
+
+        if (map[vbi]) {
+            vbuf = &r300->vertex_buffer[vbi];
+            pipe_buffer_unmap(r300->context.screen, vbuf->buffer);
+            map[vbi] = 0;
+        }
+    }
 }
 
 static void r300_emit_draw_arrays(struct r300_context *r300,
@@ -222,16 +269,49 @@ static void r300_emit_draw_elements(struct r300_context *r300,
 }
 
 
+static boolean r300_setup_local_vertex_buffers(struct r300_context *r300)
+{
+    struct pipe_vertex_buffer *vb;
+    boolean found_local_bo = FALSE, found_managed_bo = FALSE;
+    unsigned i;
+
+    /* See what buffers we got. */
+    for (i = 0; i < r300->vertex_element_count; i++) {
+        vb = &r300->vertex_buffer[r300->vertex_element[i].vertex_buffer_index];
+        if (r300->winsys->buffer_is_local(r300->winsys, vb->buffer)) {
+            found_local_bo = TRUE;
+        } else {
+            found_managed_bo = TRUE;
+        }
+    }
+
+    /* If we found both local and managed buffers, make local buffers managed
+     * because we shouldn't use the immediate mode in case a managed buffer is
+     * present, due to performance reasons. */
+    if (found_local_bo && found_managed_bo) {
+        for (i = 0; i < r300->vertex_element_count; i++) {
+            vb = &r300->vertex_buffer[r300->vertex_element[i].vertex_buffer_index];
+            if (r300->winsys->buffer_is_local(r300->winsys, vb->buffer)) {
+                r300->winsys->buffer_make_managed(r300->winsys, vb->buffer);
+            }
+        }
+    }
+
+    return !found_managed_bo;
+}
+
 static boolean r300_setup_vertex_buffers(struct r300_context *r300)
 {
     struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
     struct pipe_vertex_element *velem = r300->vertex_element;
+    struct pipe_buffer *pbuf;
 
 validate:
     for (int i = 0; i < r300->vertex_element_count; i++) {
-        if (!r300->winsys->add_buffer(r300->winsys,
-                vbuf[velem[i].vertex_buffer_index].buffer,
-            RADEON_GEM_DOMAIN_GTT, 0)) {
+        pbuf = vbuf[velem[i].vertex_buffer_index].buffer;
+
+        if (!r300->winsys->add_buffer(r300->winsys, pbuf,
+                                      RADEON_GEM_DOMAIN_GTT, 0)) {
             r300->context.flush(&r300->context, 0, NULL);
             goto validate;
         }
@@ -245,6 +325,7 @@ validate:
     return TRUE;
 }
 
+
 static void r300_shorten_ubyte_elts(struct r300_context* r300,
                                     struct pipe_buffer** elts,
                                     unsigned count)
@@ -365,15 +446,15 @@ void r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
 
     r300_emit_buffer_validate(r300);
 
-    if (!r300_setup_vertex_buffers(r300)) {
-        return;
-    }
-
-    r300_emit_dirty_state(r300);
-
-    if (FALSE && count <= 4 && r300->vertex_buffer_count == 1) {
-        r300_emit_draw_immediate(r300, mode, start, count);
+    if (r300_setup_local_vertex_buffers(r300)) {
+        r300_emit_dirty_state(r300);
+        r300_emit_draw_arrays_immediate(r300, mode, start, count);
     } else {
+        if (!r300_setup_vertex_buffers(r300)) {
+            return;
+        }
+
+        r300_emit_dirty_state(r300);
         r300_emit_aos(r300, start);
         r300_emit_draw_arrays(r300, mode, count);
     }
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index e2ec0bc5bd..641e95e7fc 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -924,6 +924,22 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
     r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
 }
 
+static boolean r300_validate_aos(struct r300_context *r300)
+{
+    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
+    struct pipe_vertex_element *velem = r300->vertex_element;
+    int i;
+
+    /* Check if formats and strides are aligned to the size of DWORD. */
+    for (i = 0; i < r300->vertex_element_count; i++) {
+        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
+            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
+            return FALSE;
+        }
+    }
+    return TRUE;
+}
+
 static void r300_set_vertex_elements(struct pipe_context* pipe,
                                     unsigned count,
                                     const struct pipe_vertex_element* elements)
@@ -939,6 +955,12 @@ static void r300_set_vertex_elements(struct pipe_context* pipe,
         draw_flush(r300->draw);
         draw_set_vertex_elements(r300->draw, count, elements);
     }
+
+    if (!r300_validate_aos(r300)) {
+        /* XXX We should fallback using draw. */
+        assert(0);
+        abort();
+    }
 }
 
 static void* r300_create_vs_state(struct pipe_context* pipe,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
index 25e1cdcdb6..5214b6d8bc 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
@@ -51,6 +51,23 @@ static const char *radeon_get_name(struct pipe_winsys *ws)
     return "Radeon/GEM+KMS";
 }
 
+uint32_t radeon_domain_from_usage(unsigned usage)
+{
+    uint32_t domain = 0;
+
+    if (usage & PIPE_BUFFER_USAGE_PIXEL) {
+        domain |= RADEON_GEM_DOMAIN_VRAM;
+    }
+    if (usage & PIPE_BUFFER_USAGE_VERTEX) {
+        domain |= RADEON_GEM_DOMAIN_GTT;
+    }
+    if (usage & PIPE_BUFFER_USAGE_INDEX) {
+        domain |= RADEON_GEM_DOMAIN_GTT;
+    }
+
+    return domain;
+}
+
 static struct pipe_buffer *radeon_buffer_create(struct pipe_winsys *ws,
                                                 unsigned alignment,
                                                 unsigned usage,
@@ -71,25 +88,17 @@ static struct pipe_buffer *radeon_buffer_create(struct pipe_winsys *ws,
     radeon_buffer->base.usage = usage;
     radeon_buffer->base.size = size;
 
-    if (usage == PIPE_BUFFER_USAGE_CONSTANT && is_r3xx(radeon_ws->pci_id)) {
+    if ((usage == PIPE_BUFFER_USAGE_CONSTANT && is_r3xx(radeon_ws->pci_id)) ||
+        (usage == PIPE_BUFFER_USAGE_VERTEX && size < 512)) {
         /* Don't bother allocating a BO, as it'll never get to the card. */
+        /* Also, create small vertex buffers in RAM. */
         desc.alignment = alignment;
         desc.usage = usage;
         radeon_buffer->pb = pb_malloc_buffer_create(size, &desc);
         return &radeon_buffer->base;
     }
 
-    domain = 0;
-
-    if (usage & PIPE_BUFFER_USAGE_PIXEL) {
-        domain |= RADEON_GEM_DOMAIN_VRAM;
-    }
-    if (usage & PIPE_BUFFER_USAGE_VERTEX) {
-        domain |= RADEON_GEM_DOMAIN_GTT;
-    }
-    if (usage & PIPE_BUFFER_USAGE_INDEX) {
-        domain |= RADEON_GEM_DOMAIN_GTT;
-    }
+    domain = radeon_domain_from_usage(usage);
 
     radeon_buffer->bo = radeon_bo_open(radeon_ws->priv->bom, 0, size,
             alignment, domain, 0);
@@ -222,6 +231,54 @@ static void radeon_buffer_set_tiling(struct radeon_winsys *ws,
     radeon_bo_set_tiling(radeon_buffer->bo, flags, pitch);
 }
 
+static boolean radeon_buffer_is_local(struct radeon_winsys *ws,
+                                      struct pipe_buffer *buffer)
+{
+    struct radeon_pipe_buffer *radeon_buffer =
+        (struct radeon_pipe_buffer*)buffer;
+
+    return radeon_buffer->pb != NULL;
+}
+
+static void radeon_buffer_make_managed(struct radeon_winsys *ws,
+                                       struct pipe_buffer *buffer)
+{
+    struct radeon_pipe_buffer* radeon_buffer =
+        (struct radeon_pipe_buffer*)buffer;
+    uint32_t domain;
+    void *map;
+
+    if (radeon_buffer->pb) {
+        domain = radeon_domain_from_usage(buffer->usage);
+
+        /* Create a managed buffer. */
+        radeon_buffer->bo = radeon_bo_open(ws->priv->bom, 0,
+                                           buffer->size, buffer->alignment,
+                                           domain, 0);
+        if (radeon_buffer->bo == NULL) {
+            /* XXX What now? */
+            fprintf(stderr, "radeon: cannot create a buffer in function %s\n",
+                    __FUNCTION__);
+            assert(0);
+            abort();
+        }
+
+        /* Move data. */
+        radeon_bo_map(radeon_buffer->bo, 1);
+        map = pb_map(radeon_buffer->pb, PIPE_BUFFER_USAGE_CPU_READ);
+
+        memcpy(radeon_buffer->bo->ptr, map, buffer->size);
+
+        pb_unmap(radeon_buffer->pb);
+        radeon_bo_unmap(radeon_buffer->bo);
+
+        /* Release the locally-created buffer. */
+        pipe_reference_init(&radeon_buffer->pb->base.reference, 0);
+        pb_destroy(radeon_buffer->pb);
+        radeon_buffer->pb = 0;
+    }
+}
+
 static void radeon_fence_reference(struct pipe_winsys *ws,
                                    struct pipe_fence_handle **ptr,
                                    struct pipe_fence_handle *pfence)
@@ -325,6 +382,8 @@ struct radeon_winsys* radeon_pipe_winsys(int fd)
     radeon_ws->base.get_name = radeon_get_name;
 
     radeon_ws->buffer_set_tiling = radeon_buffer_set_tiling;
+    radeon_ws->buffer_is_local = radeon_buffer_is_local;
+    radeon_ws->buffer_make_managed = radeon_buffer_make_managed;
 
     return radeon_ws;
 }
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.h b/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
index de71cb2f42..c46abff793 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
@@ -77,6 +77,8 @@ struct radeon_winsys_priv {
     void *flush_data;
 };
 
+uint32_t radeon_domain_from_usage(unsigned usage);
+
 struct radeon_winsys* radeon_pipe_winsys(int fb);
 #if 0
 struct pipe_surface *radeon_surface_from_handle(struct radeon_context *radeon_context,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.h b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
index ddd7983824..077388ee02 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
@@ -81,7 +81,7 @@ void radeon_destroy_drm_api(struct drm_api* api);
 /* Guess at whether this chipset should use r300g.
  *
  * I believe that this check is valid, but I haven't been exhaustive. */
-static boolean is_r3xx(int pciid)
+static INLINE boolean is_r3xx(int pciid)
 {
     return (pciid > 0x3150) && (pciid < 0x796f);
 }
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_r300.c b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
index 0253bc2527..d759beaba1 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_r300.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
@@ -81,9 +81,13 @@ static void radeon_write_cs_reloc(struct radeon_winsys* winsys,
                                   uint32_t flags)
 {
     int retval = 0;
+    struct radeon_pipe_buffer* radeon_buffer =
+        (struct radeon_pipe_buffer*)pbuffer;
 
-    retval = radeon_cs_write_reloc(winsys->priv->cs,
-            ((struct radeon_pipe_buffer*)pbuffer)->bo, rd, wd, flags);
+    assert(!radeon_buffer->pb);
+
+    retval = radeon_cs_write_reloc(winsys->priv->cs, radeon_buffer->bo,
+                                   rd, wd, flags);
 
     if (retval) {
         debug_printf("radeon: Relocation of %p (%d, %d, %d) failed!\n",
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
index 864082b99b..462fba844e 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
@@ -106,6 +106,12 @@ struct radeon_winsys {
                               uint32_t pitch,
                               boolean microtiled,
                               boolean macrotiled);
+
+    boolean (*buffer_is_local)(struct radeon_winsys* winsys,
+                               struct pipe_buffer* buffer);
+
+    void (*buffer_make_managed)(struct radeon_winsys* winsys,
+                                struct pipe_buffer* buffer);
 };
 
 #endif
-- 
cgit v1.2.3


From 4c082faa7e0887205fe9158320884c177f416b80 Mon Sep 17 00:00:00 2001
From: José Fonseca <jfonseca@vmware.com>
Date: Mon, 25 Jan 2010 11:54:54 +0000
Subject: llvmpipe: Remove lp_tex_sample_c.c again.

Popped back to life in some merge. Not necessary even as a reference,
since it is identical to softpipe's sp_tex_sample.c with renamed symbols.
---
 src/gallium/drivers/llvmpipe/lp_tex_sample_c.c | 1712 ------------------------
 1 file changed, 1712 deletions(-)
 delete mode 100644 src/gallium/drivers/llvmpipe/lp_tex_sample_c.c

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
deleted file mode 100644
index d1f5d9505d..0000000000
--- a/src/gallium/drivers/llvmpipe/lp_tex_sample_c.c
+++ /dev/null
@@ -1,1712 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * Copyright 2008 VMware, Inc.  All rights reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/**
- * Texture sampling
- *
- * Authors:
- *   Brian Paul
- */
-
-#include "lp_context.h"
-#include "lp_quad.h"
-#include "lp_texture.h"
-#include "lp_tex_sample.h"
-#include "lp_tex_cache.h"
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_shader_tokens.h"
-#include "util/u_math.h"
-#include "util/u_memory.h"
-
-
-
-/*
- * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
- * see 1-pixel bands of improperly weighted linear-filtered textures.
- * The tests/texwrap.c demo is a good test.
- * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
- * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
- */
-#define FRAC(f)  ((f) - util_ifloor(f))
-
-
-/**
- * Linear interpolation macro
- */
-static INLINE float
-lerp(float a, float v0, float v1)
-{
-   return v0 + a * (v1 - v0);
-}
-
-
-/**
- * Do 2D/biliner interpolation of float values.
- * v00, v10, v01 and v11 are typically four texture samples in a square/box.
- * a and b are the horizontal and vertical interpolants.
- * It's important that this function is inlined when compiled with
- * optimization!  If we find that's not true on some systems, convert
- * to a macro.
- */
-static INLINE float
-lerp_2d(float a, float b,
-        float v00, float v10, float v01, float v11)
-{
-   const float temp0 = lerp(a, v00, v10);
-   const float temp1 = lerp(a, v01, v11);
-   return lerp(b, temp0, temp1);
-}
-
-
-/**
- * As above, but 3D interpolation of 8 values.
- */
-static INLINE float
-lerp_3d(float a, float b, float c,
-        float v000, float v100, float v010, float v110,
-        float v001, float v101, float v011, float v111)
-{
-   const float temp0 = lerp_2d(a, b, v000, v100, v010, v110);
-   const float temp1 = lerp_2d(a, b, v001, v101, v011, v111);
-   return lerp(c, temp0, temp1);
-}
-
-
-
-/**
- * If A is a signed integer, A % B doesn't give the right value for A < 0
- * (in terms of texture repeat).  Just casting to unsigned fixes that.
- */
-#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B))
-
-
-/**
- * Apply texture coord wrapping mode and return integer texture indexes
- * for a vector of four texcoords (S or T or P).
- * \param wrapMode  PIPE_TEX_WRAP_x
- * \param s  the incoming texcoords
- * \param size  the texture image size
- * \param icoord  returns the integer texcoords
- * \return  integer texture index
- */
-static INLINE void
-nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
-                   int icoord[4])
-{
-   uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_REPEAT:
-      /* s limited to [0,1) */
-      /* i limited to [0,size-1] */
-      for (ch = 0; ch < 4; ch++) {
-         int i = util_ifloor(s[ch] * size);
-         icoord[ch] = REMAINDER(i, size);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP:
-      /* s limited to [0,1] */
-      /* i limited to [0,size-1] */
-      for (ch = 0; ch < 4; ch++) {
-         if (s[ch] <= 0.0F)
-            icoord[ch] = 0;
-         else if (s[ch] >= 1.0F)
-            icoord[ch] = size - 1;
-         else
-            icoord[ch] = util_ifloor(s[ch] * size);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            if (s[ch] < min)
-               icoord[ch] = 0;
-            else if (s[ch] > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(s[ch] * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [-1, size] */
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            if (s[ch] <= min)
-               icoord[ch] = -1;
-            else if (s[ch] >= max)
-               icoord[ch] = size;
-            else
-               icoord[ch] = util_ifloor(s[ch] * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      {
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const int flr = util_ifloor(s[ch]);
-            float u;
-            if (flr & 1)
-               u = 1.0F - (s[ch] - (float) flr);
-            else
-               u = s[ch] - (float) flr;
-            if (u < min)
-               icoord[ch] = 0;
-            else if (u > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         /* s limited to [0,1] */
-         /* i limited to [0,size-1] */
-         const float u = fabsf(s[ch]);
-         if (u <= 0.0F)
-            icoord[ch] = 0;
-         else if (u >= 1.0F)
-            icoord[ch] = size - 1;
-         else
-            icoord[ch] = util_ifloor(u * size);
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = 1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const float u = fabsf(s[ch]);
-            if (u < min)
-               icoord[ch] = 0;
-            else if (u > max)
-               icoord[ch] = size - 1;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            const float u = fabsf(s[ch]);
-            if (u < min)
-               icoord[ch] = -1;
-            else if (u > max)
-               icoord[ch] = size;
-            else
-               icoord[ch] = util_ifloor(u * size);
-         }
-      }
-      return;
-   default:
-      assert(0);
-   }
-}
-
-
-/**
- * Used to compute texel locations for linear sampling for four texcoords.
- * \param wrapMode  PIPE_TEX_WRAP_x
- * \param s  the texcoords
- * \param size  the texture image size
- * \param icoord0  returns first texture indexes
- * \param icoord1  returns second texture indexes (usually icoord0 + 1)
- * \param w  returns blend factor/weight between texture indexes
- * \param icoord  returns the computed integer texture coords
- */
-static INLINE void
-linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
-                  int icoord0[4], int icoord1[4], float w[4])
-{
-   uint ch;
-
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_REPEAT:
-      for (ch = 0; ch < 4; ch++) {
-         float u = s[ch] * size - 0.5F;
-         icoord0[ch] = REMAINDER(util_ifloor(u), size);
-         icoord1[ch] = REMAINDER(icoord0[ch] + 1, size);
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.0F, 1.0F);
-         u = u * size - 0.5f;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.0F, 1.0F);
-         u = u * size - 0.5f;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      {
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            float u = CLAMP(s[ch], min, max);
-            u = u * size - 0.5f;
-            icoord0[ch] = util_ifloor(u);
-            icoord1[ch] = icoord0[ch] + 1;
-            w[ch] = FRAC(u);
-         }
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      for (ch = 0; ch < 4; ch++) {
-         const int flr = util_ifloor(s[ch]);
-         float u;
-         if (flr & 1)
-            u = 1.0F - (s[ch] - (float) flr);
-         else
-            u = s[ch] - (float) flr;
-         u = u * size - 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         float u = fabsf(s[ch]);
-         if (u >= 1.0F)
-            u = (float) size;
-         else
-            u *= size;
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-      for (ch = 0; ch < 4; ch++) {
-         float u = fabsf(s[ch]);
-         if (u >= 1.0F)
-            u = (float) size;
-         else
-            u *= size;
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord0[ch] < 0)
-            icoord0[ch] = 0;
-         if (icoord1[ch] >= (int) size)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;;
-   case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
-      {
-         const float min = -1.0F / (2.0F * size);
-         const float max = 1.0F - min;
-         for (ch = 0; ch < 4; ch++) {
-            float u = fabsf(s[ch]);
-            if (u <= min)
-               u = min * size;
-            else if (u >= max)
-               u = max * size;
-            else
-               u *= size;
-            u -= 0.5F;
-            icoord0[ch] = util_ifloor(u);
-            icoord1[ch] = icoord0[ch] + 1;
-            w[ch] = FRAC(u);
-         }
-      }
-      break;;
-   default:
-      assert(0);
-   }
-}
-
-
-/**
- * For RECT textures / unnormalized texcoords
- * Only a subset of wrap modes supported.
- */
-static INLINE void
-nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
-                          int icoord[4])
-{
-   uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         int i = util_ifloor(s[ch]);
-         icoord[ch]= CLAMP(i, 0, (int) size-1);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      /* fall-through */
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      for (ch = 0; ch < 4; ch++) {
-         icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) );
-      }
-      return;
-   default:
-      assert(0);
-   }
-}
-
-
-/**
- * For RECT textures / unnormalized texcoords.
- * Only a subset of wrap modes supported.
- */
-static INLINE void
-linear_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
-                         int icoord0[4], int icoord1[4], float w[4])
-{
-   uint ch;
-   switch (wrapMode) {
-   case PIPE_TEX_WRAP_CLAMP:
-      for (ch = 0; ch < 4; ch++) {
-         /* Not exactly what the spec says, but it matches NVIDIA output */
-         float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f);
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         w[ch] = FRAC(u);
-      }
-      return;
-   case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      /* fall-through */
-   case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      for (ch = 0; ch < 4; ch++) {
-         float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F);
-         u -= 0.5F;
-         icoord0[ch] = util_ifloor(u);
-         icoord1[ch] = icoord0[ch] + 1;
-         if (icoord1[ch] > (int) size - 1)
-            icoord1[ch] = size - 1;
-         w[ch] = FRAC(u);
-      }
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-static unsigned
-choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
-{
-   /*
-      major axis
-      direction     target                             sc     tc    ma
-      ----------    -------------------------------    ---    ---   ---
-       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
-       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
-       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
-       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
-       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
-       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
-   */
-   const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz);
-   unsigned face;
-   float sc, tc, ma;
-
-   if (arx > ary && arx > arz) {
-      if (rx >= 0.0F) {
-         face = PIPE_TEX_FACE_POS_X;
-         sc = -rz;
-         tc = -ry;
-         ma = arx;
-      }
-      else {
-         face = PIPE_TEX_FACE_NEG_X;
-         sc = rz;
-         tc = -ry;
-         ma = arx;
-      }
-   }
-   else if (ary > arx && ary > arz) {
-      if (ry >= 0.0F) {
-         face = PIPE_TEX_FACE_POS_Y;
-         sc = rx;
-         tc = rz;
-         ma = ary;
-      }
-      else {
-         face = PIPE_TEX_FACE_NEG_Y;
-         sc = rx;
-         tc = -rz;
-         ma = ary;
-      }
-   }
-   else {
-      if (rz > 0.0F) {
-         face = PIPE_TEX_FACE_POS_Z;
-         sc = rx;
-         tc = -ry;
-         ma = arz;
-      }
-      else {
-         face = PIPE_TEX_FACE_NEG_Z;
-         sc = -rx;
-         tc = -ry;
-         ma = arz;
-      }
-   }
-
-   *newS = ( sc / ma + 1.0F ) * 0.5F;
-   *newT = ( tc / ma + 1.0F ) * 0.5F;
-
-   return face;
-}
-
-
-/**
- * Examine the quad's texture coordinates to compute the partial
- * derivatives w.r.t X and Y, then compute lambda (level of detail).
- *
- * This is only done for fragment shaders, not vertex shaders.
- */
-static float
-compute_lambda(struct tgsi_sampler *tgsi_sampler,
-               const float s[QUAD_SIZE],
-               const float t[QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               float lodbias)
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-   float rho, lambda;
-
-   if (samp->processor == TGSI_PROCESSOR_VERTEX)
-      return lodbias;
-
-   assert(sampler->normalized_coords);
-
-   assert(s);
-   {
-      float dsdx = s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT];
-      float dsdy = s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT];
-      dsdx = fabsf(dsdx);
-      dsdy = fabsf(dsdy);
-      rho = MAX2(dsdx, dsdy) * texture->width[0];
-   }
-   if (t) {
-      float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT];
-      float dtdy = t[QUAD_TOP_LEFT]     - t[QUAD_BOTTOM_LEFT];
-      float max;
-      dtdx = fabsf(dtdx);
-      dtdy = fabsf(dtdy);
-      max = MAX2(dtdx, dtdy) * texture->height[0];
-      rho = MAX2(rho, max);
-   }
-   if (p) {
-      float dpdx = p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT];
-      float dpdy = p[QUAD_TOP_LEFT]     - p[QUAD_BOTTOM_LEFT];
-      float max;
-      dpdx = fabsf(dpdx);
-      dpdy = fabsf(dpdy);
-      max = MAX2(dpdx, dpdy) * texture->depth[0];
-      rho = MAX2(rho, max);
-   }
-
-   lambda = util_fast_log2(rho);
-   lambda += lodbias + sampler->lod_bias;
-   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
-
-   return lambda;
-}
-
-
-/**
- * Do several things here:
- * 1. Compute lambda from the texcoords, if needed
- * 2. Determine if we're minifying or magnifying
- * 3. If minifying, choose mipmap levels
- * 4. Return image filter to use within mipmap images
- * \param level0  Returns first mipmap level to sample from
- * \param level1  Returns second mipmap level to sample from
- * \param levelBlend  Returns blend factor between levels, in [0,1]
- * \param imgFilter  Returns either the min or mag filter, depending on lambda
- */
-static void
-choose_mipmap_levels(struct tgsi_sampler *tgsi_sampler,
-                     const float s[QUAD_SIZE],
-                     const float t[QUAD_SIZE],
-                     const float p[QUAD_SIZE],
-                     float lodbias,
-                     unsigned *level0, unsigned *level1, float *levelBlend,
-                     unsigned *imgFilter)
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-
-   if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
-      /* no mipmap selection needed */
-      *level0 = *level1 = CLAMP((int) sampler->min_lod,
-                                0, (int) texture->last_level);
-
-      if (sampler->min_img_filter != sampler->mag_img_filter) {
-         /* non-mipmapped texture, but still need to determine if doing
-          * minification or magnification.
-          */
-         float lambda = compute_lambda(tgsi_sampler, s, t, p, lodbias);
-         if (lambda <= 0.0) {
-            *imgFilter = sampler->mag_img_filter;
-         }
-         else {
-            *imgFilter = sampler->min_img_filter;
-         }
-      }
-      else {
-         *imgFilter = sampler->mag_img_filter;
-      }
-   }
-   else {
-      float lambda = compute_lambda(tgsi_sampler, s, t, p, lodbias);
-
-      if (lambda <= 0.0) { /* XXX threshold depends on the filter */
-         /* magnifying */
-         *imgFilter = sampler->mag_img_filter;
-         *level0 = *level1 = 0;
-      }
-      else {
-         /* minifying */
-         *imgFilter = sampler->min_img_filter;
-
-         /* choose mipmap level(s) and compute the blend factor between them */
-         if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) {
-            /* Nearest mipmap level */
-            const int lvl = (int) (lambda + 0.5);
-            *level0 =
-            *level1 = CLAMP(lvl, 0, (int) texture->last_level);
-         }
-         else {
-            /* Linear interpolation between mipmap levels */
-            const int lvl = (int) lambda;
-            *level0 = CLAMP(lvl,     0, (int) texture->last_level);
-            *level1 = CLAMP(lvl + 1, 0, (int) texture->last_level);
-            *levelBlend = FRAC(lambda);  /* blending weight between levels */
-         }
-      }
-   }
-}
-
-
-/**
- * Get a texel from a texture, using the texture tile cache.
- *
- * \param face  the cube face in 0..5
- * \param level  the mipmap level
- * \param x  the x coord of texel within 2D image
- * \param y  the y coord of texel within 2D image
- * \param z  which slice of a 3D texture
- * \param rgba  the quad to put the texel/color into
- * \param j  which element of the rgba quad to write to
- *
- * XXX maybe move this into lp_tile_cache.c and merge with the
- * lp_get_cached_tile_tex() function.  Also, get 4 texels instead of 1...
- */
-static void
-get_texel_quad_2d(const struct tgsi_sampler *tgsi_sampler,
-                  unsigned face, unsigned level, int x, int y, 
-                  const uint8_t *out[4])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-
-   const struct llvmpipe_cached_tex_tile *tile
-      = lp_get_cached_tex_tile(samp->cache,
-                               tex_tile_address(x, y, 0, face, level));
-
-   y %= TEX_TILE_SIZE;
-   x %= TEX_TILE_SIZE;
-      
-   out[0] = &tile->color[y  ][x  ][0];
-   out[1] = &tile->color[y  ][x+1][0];
-   out[2] = &tile->color[y+1][x  ][0];
-   out[3] = &tile->color[y+1][x+1][0];
-}
-
-static INLINE const uint8_t *
-get_texel_2d_ptr(const struct tgsi_sampler *tgsi_sampler,
-                 unsigned face, unsigned level, int x, int y)
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-
-   const struct llvmpipe_cached_tex_tile *tile
-      = lp_get_cached_tex_tile(samp->cache,
-                               tex_tile_address(x, y, 0, face, level));
-
-   y %= TEX_TILE_SIZE;
-   x %= TEX_TILE_SIZE;
-
-   return &tile->color[y][x][0];
-}
-
-
-static void
-get_texel_quad_2d_mt(const struct tgsi_sampler *tgsi_sampler,
-                     unsigned face, unsigned level, 
-                     int x0, int y0, 
-                     int x1, int y1,
-                     const uint8_t *out[4])
-{
-   unsigned i;
-
-   for (i = 0; i < 4; i++) {
-      unsigned tx = (i & 1) ? x1 : x0;
-      unsigned ty = (i >> 1) ? y1 : y0;
-
-      out[i] = get_texel_2d_ptr( tgsi_sampler, face, level, tx, ty );
-   }
-}
-
-static void
-get_texel(const struct tgsi_sampler *tgsi_sampler,
-                 unsigned face, unsigned level, int x, int y, int z,
-                 float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j)
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-
-   if (x < 0 || x >= (int) texture->width[level] ||
-       y < 0 || y >= (int) texture->height[level] ||
-       z < 0 || z >= (int) texture->depth[level]) {
-      rgba[0][j] = sampler->border_color[0];
-      rgba[1][j] = sampler->border_color[1];
-      rgba[2][j] = sampler->border_color[2];
-      rgba[3][j] = sampler->border_color[3];
-   }
-   else {
-      const unsigned tx = x % TEX_TILE_SIZE;
-      const unsigned ty = y % TEX_TILE_SIZE;
-      const struct llvmpipe_cached_tex_tile *tile;
-
-      tile = lp_get_cached_tex_tile(samp->cache,
-                                    tex_tile_address(x, y, z, face, level));
-
-      rgba[0][j] = ubyte_to_float(tile->color[ty][tx][0]);
-      rgba[1][j] = ubyte_to_float(tile->color[ty][tx][1]);
-      rgba[2][j] = ubyte_to_float(tile->color[ty][tx][2]);
-      rgba[3][j] = ubyte_to_float(tile->color[ty][tx][3]);
-      if (0)
-      {
-         debug_printf("Get texel %f %f %f %f from %s\n",
-                      rgba[0][j], rgba[1][j], rgba[2][j], rgba[3][j],
-                      pf_name(texture->format));
-      }
-   }
-}
-
-
-/**
- * Compare texcoord 'p' (aka R) against texture value 'rgba[0]'
- * When we sampled the depth texture, the depth value was put into all
- * RGBA channels.  We look at the red channel here.
- * \param rgba  quad of (depth) texel values
- * \param p  texture 'P' components for four pixels in quad
- * \param j  which pixel in the quad to test [0..3]
- */
-static INLINE void
-shadow_compare(const struct pipe_sampler_state *sampler,
-               float rgba[NUM_CHANNELS][QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               uint j)
-{
-   int k;
-   switch (sampler->compare_func) {
-   case PIPE_FUNC_LESS:
-      k = p[j] < rgba[0][j];
-      break;
-   case PIPE_FUNC_LEQUAL:
-      k = p[j] <= rgba[0][j];
-      break;
-   case PIPE_FUNC_GREATER:
-      k = p[j] > rgba[0][j];
-      break;
-   case PIPE_FUNC_GEQUAL:
-      k = p[j] >= rgba[0][j];
-      break;
-   case PIPE_FUNC_EQUAL:
-      k = p[j] == rgba[0][j];
-      break;
-   case PIPE_FUNC_NOTEQUAL:
-      k = p[j] != rgba[0][j];
-      break;
-   case PIPE_FUNC_ALWAYS:
-      k = 1;
-      break;
-   case PIPE_FUNC_NEVER:
-      k = 0;
-      break;
-   default:
-      k = 0;
-      assert(0);
-      break;
-   }
-
-   /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */
-   rgba[0][j] = rgba[1][j] = rgba[2][j] = (float) k;
-   rgba[3][j] = 1.0F;
-}
-
-
-/**
- * As above, but do four z/texture comparisons.
- */
-static INLINE void
-shadow_compare4(const struct pipe_sampler_state *sampler,
-                float rgba[NUM_CHANNELS][QUAD_SIZE],
-                const float p[QUAD_SIZE])
-{
-   int j, k0, k1, k2, k3;
-   float val;
-
-   /* compare four texcoords vs. four texture samples */
-   switch (sampler->compare_func) {
-   case PIPE_FUNC_LESS:
-      k0 = p[0] < rgba[0][0];
-      k1 = p[1] < rgba[0][1];
-      k2 = p[2] < rgba[0][2];
-      k3 = p[3] < rgba[0][3];
-      break;
-   case PIPE_FUNC_LEQUAL:
-      k0 = p[0] <= rgba[0][0];
-      k1 = p[1] <= rgba[0][1];
-      k2 = p[2] <= rgba[0][2];
-      k3 = p[3] <= rgba[0][3];
-      break;
-   case PIPE_FUNC_GREATER:
-      k0 = p[0] > rgba[0][0];
-      k1 = p[1] > rgba[0][1];
-      k2 = p[2] > rgba[0][2];
-      k3 = p[3] > rgba[0][3];
-      break;
-   case PIPE_FUNC_GEQUAL:
-      k0 = p[0] >= rgba[0][0];
-      k1 = p[1] >= rgba[0][1];
-      k2 = p[2] >= rgba[0][2];
-      k3 = p[3] >= rgba[0][3];
-      break;
-   case PIPE_FUNC_EQUAL:
-      k0 = p[0] == rgba[0][0];
-      k1 = p[1] == rgba[0][1];
-      k2 = p[2] == rgba[0][2];
-      k3 = p[3] == rgba[0][3];
-      break;
-   case PIPE_FUNC_NOTEQUAL:
-      k0 = p[0] != rgba[0][0];
-      k1 = p[1] != rgba[0][1];
-      k2 = p[2] != rgba[0][2];
-      k3 = p[3] != rgba[0][3];
-      break;
-   case PIPE_FUNC_ALWAYS:
-      k0 = k1 = k2 = k3 = 1;
-      break;
-   case PIPE_FUNC_NEVER:
-      k0 = k1 = k2 = k3 = 0;
-      break;
-   default:
-      k0 = k1 = k2 = k3 = 0;
-      assert(0);
-      break;
-   }
-
-   /* convert four pass/fail values to an intensity in [0,1] */
-   val = 0.25F * (k0 + k1 + k2 + k3);
-
-   /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */
-   for (j = 0; j < 4; j++) {
-      rgba[0][j] = rgba[1][j] = rgba[2][j] = val;
-      rgba[3][j] = 1.0F;
-   }
-}
-
-
-
-static void
-lp_get_samples_2d_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler,
-                                    const float s[QUAD_SIZE],
-                                    const float t[QUAD_SIZE],
-                                    const float p[QUAD_SIZE],
-                                    float lodbias,
-                                    float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   unsigned  j;
-   unsigned level = samp->level;
-   unsigned xpot = 1 << (samp->xpot - level);
-   unsigned ypot = 1 << (samp->ypot - level);
-   unsigned xmax = (xpot - 1) & (TEX_TILE_SIZE - 1); /* MIN2(TEX_TILE_SIZE, xpot) - 1; */
-   unsigned ymax = (ypot - 1) & (TEX_TILE_SIZE - 1); /* MIN2(TEX_TILE_SIZE, ypot) - 1; */
-      
-   for (j = 0; j < QUAD_SIZE; j++) {
-      int c;
-
-      float u = s[j] * xpot - 0.5F;
-      float v = t[j] * ypot - 0.5F;
-
-      int uflr = util_ifloor(u);
-      int vflr = util_ifloor(v);
-
-      float xw = u - (float)uflr;
-      float yw = v - (float)vflr;
-
-      int x0 = uflr & (xpot - 1);
-      int y0 = vflr & (ypot - 1);
-
-      const uint8_t *tx[4];
-      
-
-      /* Can we fetch all four at once:
-       */
-      if (x0 < xmax && y0 < ymax)
-      {
-         get_texel_quad_2d(tgsi_sampler, 0, level, x0, y0, tx);
-      }
-      else 
-      {
-         unsigned x1 = (x0 + 1) & (xpot - 1);
-         unsigned y1 = (y0 + 1) & (ypot - 1);
-         get_texel_quad_2d_mt(tgsi_sampler, 0, level, 
-                              x0, y0, x1, y1, tx);
-      }
-
-
-      /* interpolate R, G, B, A */
-      for (c = 0; c < 4; c++) {
-         rgba[c][j] = lerp_2d(xw, yw, 
-                              ubyte_to_float(tx[0][c]), ubyte_to_float(tx[1][c]),
-                              ubyte_to_float(tx[2][c]), ubyte_to_float(tx[3][c]));
-      }
-   }
-}
-
-
-static void
-lp_get_samples_2d_nearest_repeat_POT(struct tgsi_sampler *tgsi_sampler,
-                                     const float s[QUAD_SIZE],
-                                     const float t[QUAD_SIZE],
-                                     const float p[QUAD_SIZE],
-                                     float lodbias,
-                                     float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   unsigned  j;
-   unsigned level = samp->level;
-   unsigned xpot = 1 << (samp->xpot - level);
-   unsigned ypot = 1 << (samp->ypot - level);
-
-   for (j = 0; j < QUAD_SIZE; j++) {
-      int c;
-
-      float u = s[j] * xpot;
-      float v = t[j] * ypot;
-
-      int uflr = util_ifloor(u);
-      int vflr = util_ifloor(v);
-
-      int x0 = uflr & (xpot - 1);
-      int y0 = vflr & (ypot - 1);
-
-      const uint8_t *out = get_texel_2d_ptr(tgsi_sampler, 0, level, x0, y0);
-
-      for (c = 0; c < 4; c++) {
-         rgba[c][j] = ubyte_to_float(out[c]);
-      }
-   }
-}
-
-
-static void
-lp_get_samples_2d_nearest_clamp_POT(struct tgsi_sampler *tgsi_sampler,
-                                     const float s[QUAD_SIZE],
-                                     const float t[QUAD_SIZE],
-                                     const float p[QUAD_SIZE],
-                                     float lodbias,
-                                     float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   unsigned  j;
-   unsigned level = samp->level;
-   unsigned xpot = 1 << (samp->xpot - level);
-   unsigned ypot = 1 << (samp->ypot - level);
-
-   for (j = 0; j < QUAD_SIZE; j++) {
-      int c;
-
-      float u = s[j] * xpot;
-      float v = t[j] * ypot;
-
-      int x0, y0;
-      const uint8_t *out;
-
-      x0 = util_ifloor(u);
-      if (x0 < 0) 
-         x0 = 0;
-      else if (x0 > xpot - 1)
-         x0 = xpot - 1;
-
-      y0 = util_ifloor(v);
-      if (y0 < 0) 
-         y0 = 0;
-      else if (y0 > ypot - 1)
-         y0 = ypot - 1;
-      
-      out = get_texel_2d_ptr(tgsi_sampler, 0, level, x0, y0);
-
-      for (c = 0; c < 4; c++) {
-         rgba[c][j] = ubyte_to_float(out[c]);
-      }
-   }
-}
-
-
-static void
-lp_get_samples_2d_linear_mip_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler,
-                                               const float s[QUAD_SIZE],
-                                               const float t[QUAD_SIZE],
-                                               const float p[QUAD_SIZE],
-                                               float lodbias,
-                                               float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   int level0;
-   float lambda;
-
-   lambda = compute_lambda(tgsi_sampler, s, t, p, lodbias);
-   level0 = (int)lambda;
-
-   if (lambda < 0.0) { 
-      samp->level = 0;
-      lp_get_samples_2d_linear_repeat_POT( tgsi_sampler,
-                                           s, t, p, 0, rgba );
-   }
-   else if (level0 >= texture->last_level) {
-      samp->level = texture->last_level;
-      lp_get_samples_2d_linear_repeat_POT( tgsi_sampler,
-                                           s, t, p, 0, rgba );
-   }
-   else {
-      float levelBlend = lambda - level0;
-      float rgba0[4][4];
-      float rgba1[4][4];
-      int c,j;
-
-      samp->level = level0;
-      lp_get_samples_2d_linear_repeat_POT( tgsi_sampler,
-                                           s, t, p, 0, rgba0 );
-
-      samp->level = level0+1;
-      lp_get_samples_2d_linear_repeat_POT( tgsi_sampler,
-                                           s, t, p, 0, rgba1 );
-
-      for (j = 0; j < QUAD_SIZE; j++) {
-         for (c = 0; c < 4; c++) {
-            rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]);
-         }
-      }
-   }
-}
-
-/**
- * Common code for sampling 1D/2D/cube textures.
- * Could probably extend for 3D...
- */
-static void
-lp_get_samples_2d_common(struct tgsi_sampler *tgsi_sampler,
-                         const float s[QUAD_SIZE],
-                         const float t[QUAD_SIZE],
-                         const float p[QUAD_SIZE],
-                         float lodbias,
-                         float rgba[NUM_CHANNELS][QUAD_SIZE],
-                         const unsigned faces[4])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-   unsigned level0, level1, j, imgFilter;
-   int width, height;
-   float levelBlend = 0.0F;
-
-   choose_mipmap_levels(tgsi_sampler, s, t, p, 
-                        lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   assert(sampler->normalized_coords);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-
-   assert(width > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4];
-         nearest_texcoord_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_4(sampler->wrap_t, t, height, y);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, faces[j], level0, x[j], y[j], 0, rgba, j);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare(sampler, rgba, p, j);
-            }
-
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               unsigned c;
-               x[j] /= 2;
-               y[j] /= 2;
-               get_texel(tgsi_sampler, faces[j], level1, x[j], y[j], 0,
-                         rgba2, j);
-               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
-                  shadow_compare(sampler, rgba2, p, j);
-               }
-
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
-         }
-      }
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
-      {
-         int x0[4], y0[4], x1[4], y1[4];
-         float xw[4], yw[4]; /* weights */
-
-         linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw);
-         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            float tx[4][4]; /* texels */
-            int c;
-            get_texel(tgsi_sampler, faces[j], level0, x0[j], y0[j], 0, tx, 0);
-            get_texel(tgsi_sampler, faces[j], level0, x1[j], y0[j], 0, tx, 1);
-            get_texel(tgsi_sampler, faces[j], level0, x0[j], y1[j], 0, tx, 2);
-            get_texel(tgsi_sampler, faces[j], level0, x1[j], y1[j], 0, tx, 3);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare4(sampler, tx, p);
-            }
-
-            /* interpolate R, G, B, A */
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_2d(xw[j], yw[j],
-                                    tx[c][0], tx[c][1],
-                                    tx[c][2], tx[c][3]);
-            }
-
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-
-               /* XXX: This is incorrect -- will often end up with (x0
-                *  == x1 && y0 == y1), meaning that we fetch the same
-                *  texel four times and linearly interpolate between
-                *  identical values.  The correct approach would be to
-                *  call linear_texcoord again for the second level.
-                */
-               x0[j] /= 2;
-               y0[j] /= 2;
-               x1[j] /= 2;
-               y1[j] /= 2;
-               get_texel(tgsi_sampler, faces[j], level1, x0[j], y0[j], 0, tx, 0);
-               get_texel(tgsi_sampler, faces[j], level1, x1[j], y0[j], 0, tx, 1);
-               get_texel(tgsi_sampler, faces[j], level1, x0[j], y1[j], 0, tx, 2);
-               get_texel(tgsi_sampler, faces[j], level1, x1[j], y1[j], 0, tx, 3);
-               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
-                  shadow_compare4(sampler, tx, p);
-               }
-
-               /* interpolate R, G, B, A */
-               for (c = 0; c < 4; c++) {
-                  rgba2[c][j] = lerp_2d(xw[j], yw[j],
-                                        tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
-               }
-
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
-         }
-      }
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-static INLINE void
-lp_get_samples_1d(struct tgsi_sampler *sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   static const unsigned faces[4] = {0, 0, 0, 0};
-   static const float tzero[4] = {0, 0, 0, 0};
-   lp_get_samples_2d_common(sampler, s, tzero, NULL,
-                            lodbias, rgba, faces);
-}
-
-
-static INLINE void
-lp_get_samples_2d(struct tgsi_sampler *sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   static const unsigned faces[4] = {0, 0, 0, 0};
-   lp_get_samples_2d_common(sampler, s, t, p,
-                            lodbias, rgba, faces);
-}
-
-
-static INLINE void
-lp_get_samples_3d(struct tgsi_sampler *tgsi_sampler,
-                  const float s[QUAD_SIZE],
-                  const float t[QUAD_SIZE],
-                  const float p[QUAD_SIZE],
-                  float lodbias,
-                  float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-   /* get/map pipe_surfaces corresponding to 3D tex slices */
-   unsigned level0, level1, j, imgFilter;
-   int width, height, depth;
-   float levelBlend;
-   const uint face = 0;
-
-   choose_mipmap_levels(tgsi_sampler, s, t, p, 
-                        lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   assert(sampler->normalized_coords);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-   depth = texture->depth[level0];
-
-   assert(width > 0);
-   assert(height > 0);
-   assert(depth > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4], z[4];
-         nearest_texcoord_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_4(sampler->wrap_t, t, height, y);
-         nearest_texcoord_4(sampler->wrap_r, p, depth, z);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, face, level0, x[j], y[j], z[j], rgba, j);
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               unsigned c;
-               x[j] /= 2;
-               y[j] /= 2;
-               z[j] /= 2;
-               get_texel(tgsi_sampler, face, level1, x[j], y[j], z[j], rgba2, j);
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba2[c][j], rgba[c][j]);
-               }
-            }
-         }
-      }
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
-      {
-         int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4];
-         float xw[4], yw[4], zw[4]; /* interpolation weights */
-         linear_texcoord_4(sampler->wrap_s, s, width,  x0, x1, xw);
-         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
-         linear_texcoord_4(sampler->wrap_r, p, depth,  z0, z1, zw);
-
-         for (j = 0; j < QUAD_SIZE; j++) {
-            int c;
-            float tx0[4][4], tx1[4][4];
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z0[j], tx0, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z0[j], tx0, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z0[j], tx0, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z0[j], tx0, 3);
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z1[j], tx1, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z1[j], tx1, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z1[j], tx1, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z1[j], tx1, 3);
-
-            /* interpolate R, G, B, A */
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j],
-                                    tx0[c][0], tx0[c][1],
-                                    tx0[c][2], tx0[c][3],
-                                    tx1[c][0], tx1[c][1],
-                                    tx1[c][2], tx1[c][3]);
-            }
-
-            if (level0 != level1) {
-               /* get texels from second mipmap level and blend */
-               float rgba2[4][4];
-               x0[j] /= 2;
-               y0[j] /= 2;
-               z0[j] /= 2;
-               x1[j] /= 2;
-               y1[j] /= 2;
-               z1[j] /= 2;
-               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z0[j], tx0, 0);
-               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z0[j], tx0, 1);
-               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z0[j], tx0, 2);
-               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z0[j], tx0, 3);
-               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z1[j], tx1, 0);
-               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z1[j], tx1, 1);
-               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z1[j], tx1, 2);
-               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z1[j], tx1, 3);
-
-               /* interpolate R, G, B, A */
-               for (c = 0; c < 4; c++) {
-                  rgba2[c][j] = lerp_3d(xw[j], yw[j], zw[j],
-                                        tx0[c][0], tx0[c][1],
-                                        tx0[c][2], tx0[c][3],
-                                        tx1[c][0], tx1[c][1],
-                                        tx1[c][2], tx1[c][3]);
-               }
-
-               /* blend mipmap levels */
-               for (c = 0; c < NUM_CHANNELS; c++) {
-                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
-               }
-            }
-         }
-      }
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-static void
-lp_get_samples_cube(struct tgsi_sampler *sampler,
-                    const float s[QUAD_SIZE],
-                    const float t[QUAD_SIZE],
-                    const float p[QUAD_SIZE],
-                    float lodbias,
-                    float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   unsigned faces[QUAD_SIZE], j;
-   float ssss[4], tttt[4];
-   for (j = 0; j < QUAD_SIZE; j++) {
-      faces[j] = choose_cube_face(s[j], t[j], p[j], ssss + j, tttt + j);
-   }
-   lp_get_samples_2d_common(sampler, ssss, tttt, NULL,
-                            lodbias, rgba, faces);
-}
-
-
-static void
-lp_get_samples_rect(struct tgsi_sampler *tgsi_sampler,
-                    const float s[QUAD_SIZE],
-                    const float t[QUAD_SIZE],
-                    const float p[QUAD_SIZE],
-                    float lodbias,
-                    float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   const struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-   const uint face = 0;
-   unsigned level0, level1, j, imgFilter;
-   int width, height;
-   float levelBlend;
-
-   choose_mipmap_levels(tgsi_sampler, s, t, p, 
-                        lodbias,
-                        &level0, &level1, &levelBlend, &imgFilter);
-
-   /* texture RECTS cannot be mipmapped */
-   assert(level0 == level1);
-
-   width = texture->width[level0];
-   height = texture->height[level0];
-
-   assert(width > 0);
-
-   switch (imgFilter) {
-   case PIPE_TEX_FILTER_NEAREST:
-      {
-         int x[4], y[4];
-         nearest_texcoord_unnorm_4(sampler->wrap_s, s, width, x);
-         nearest_texcoord_unnorm_4(sampler->wrap_t, t, height, y);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            get_texel(tgsi_sampler, face, level0, x[j], y[j], 0, rgba, j);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare(sampler, rgba, p, j);
-            }
-         }
-      }
-      break;
-   case PIPE_TEX_FILTER_LINEAR:
-   case PIPE_TEX_FILTER_ANISO:
-      {
-         int x0[4], y0[4], x1[4], y1[4];
-         float xw[4], yw[4]; /* weights */
-         linear_texcoord_unnorm_4(sampler->wrap_s, s, width,  x0, x1, xw);
-         linear_texcoord_unnorm_4(sampler->wrap_t, t, height, y0, y1, yw);
-         for (j = 0; j < QUAD_SIZE; j++) {
-            float tx[4][4]; /* texels */
-            int c;
-            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], 0, tx, 0);
-            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], 0, tx, 1);
-            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], 0, tx, 2);
-            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], 0, tx, 3);
-            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-               shadow_compare4(sampler, tx, p);
-            }
-            for (c = 0; c < 4; c++) {
-               rgba[c][j] = lerp_2d(xw[j], yw[j],
-                                    tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
-            }
-         }
-      }
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-/**
- * Error condition handler
- */
-static INLINE void
-lp_get_samples_null(struct tgsi_sampler *tgsi_sampler,
-                    const float s[QUAD_SIZE],
-                    const float t[QUAD_SIZE],
-                    const float p[QUAD_SIZE],
-                    float lodbias,
-                    float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   int i,j;
-
-   for (i = 0; i < 4; i++)
-      for (j = 0; j < 4; j++)
-         rgba[i][j] = 1.0;
-}
-
-/**
- * Called via tgsi_sampler::get_samples() when using a sampler for the
- * first time.  Determine the actual sampler function, link it in and
- * call it.
- */
-void
-lp_get_samples(struct tgsi_sampler *tgsi_sampler,
-               const float s[QUAD_SIZE],
-               const float t[QUAD_SIZE],
-               const float p[QUAD_SIZE],
-               float lodbias,
-               float rgba[NUM_CHANNELS][QUAD_SIZE])
-{
-   struct lp_shader_sampler *samp = lp_shader_sampler(tgsi_sampler);
-   const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
-
-   /* Default to the 'undefined' case:
-    */
-   tgsi_sampler->get_samples = lp_get_samples_null;
-
-   if (!texture) {
-      assert(0);                /* is this legal?? */
-      goto out;
-   }
-
-   if (!sampler->normalized_coords) {
-      assert (texture->target == PIPE_TEXTURE_2D);
-      tgsi_sampler->get_samples = lp_get_samples_rect;
-      goto out;
-   }
-
-   switch (texture->target) {
-   case PIPE_TEXTURE_1D:
-      tgsi_sampler->get_samples = lp_get_samples_1d;
-      break;
-   case PIPE_TEXTURE_2D:
-      tgsi_sampler->get_samples = lp_get_samples_2d;
-      break;
-   case PIPE_TEXTURE_3D:
-      tgsi_sampler->get_samples = lp_get_samples_3d;
-      break;
-   case PIPE_TEXTURE_CUBE:
-      tgsi_sampler->get_samples = lp_get_samples_cube;
-      break;
-   default:
-      assert(0);
-      break;
-   }
-
-   /* Do this elsewhere: 
-    */
-   samp->xpot = util_unsigned_logbase2( samp->texture->width[0] );
-   samp->ypot = util_unsigned_logbase2( samp->texture->height[0] );
-
-   /* Try to hook in a faster sampler.  Ultimately we'll have to
-    * code-generate these.  Luckily most of this looks like it is
-    * orthogonal state within the sampler.
-    */
-   if (texture->target == PIPE_TEXTURE_2D &&
-       sampler->min_img_filter == sampler->mag_img_filter &&
-       sampler->wrap_s == sampler->wrap_t &&
-       sampler->compare_mode == FALSE &&
-       sampler->normalized_coords) 
-   {
-      if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
-         samp->level = CLAMP((int) sampler->min_lod,
-                             0, (int) texture->last_level);
-
-         if (sampler->wrap_s == PIPE_TEX_WRAP_REPEAT) {
-            switch (sampler->min_img_filter) {
-            case PIPE_TEX_FILTER_NEAREST:
-               tgsi_sampler->get_samples = lp_get_samples_2d_nearest_repeat_POT;
-               break;
-            case PIPE_TEX_FILTER_LINEAR:
-               tgsi_sampler->get_samples = lp_get_samples_2d_linear_repeat_POT;
-               break;
-            default:
-               break;
-            }
-         } 
-         else if (sampler->wrap_s == PIPE_TEX_WRAP_CLAMP) {
-            switch (sampler->min_img_filter) {
-            case PIPE_TEX_FILTER_NEAREST:
-               tgsi_sampler->get_samples = lp_get_samples_2d_nearest_clamp_POT;
-               break;
-            default:
-               break;
-            }
-         }
-      }
-      else if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_LINEAR) {
-         if (sampler->wrap_s == PIPE_TEX_WRAP_REPEAT) {
-            switch (sampler->min_img_filter) {
-            case PIPE_TEX_FILTER_LINEAR:
-               tgsi_sampler->get_samples = lp_get_samples_2d_linear_mip_linear_repeat_POT;
-               break;
-            default:
-               break;
-            }
-         } 
-      }
-   }
-   else if (0) {
-      _debug_printf("target %d/%d min_mip %d/%d min_img %d/%d wrap %d/%d compare %d/%d norm %d/%d\n",
-                    texture->target, PIPE_TEXTURE_2D,
-                    sampler->min_mip_filter, PIPE_TEX_MIPFILTER_NONE,
-                    sampler->min_img_filter, sampler->mag_img_filter,
-                    sampler->wrap_s, sampler->wrap_t,
-                    sampler->compare_mode, FALSE,
-                    sampler->normalized_coords, TRUE);
-   }
-
-out:
-   tgsi_sampler->get_samples( tgsi_sampler, s, t, p, lodbias, rgba );
-}
-
-
-void PIPE_CDECL
-lp_fetch_texel_soa( struct tgsi_sampler **samplers,
-                    uint32_t unit,
-                    float *store )
-{
-   struct tgsi_sampler *sampler = samplers[unit];
-
-#if 0
-   uint j;
-
-   debug_printf("%s sampler: %p (%p) store: %p\n",
-                __FUNCTION__,
-                sampler, *sampler,
-                store );
-
-   debug_printf("lodbias %f\n", store[12]);
-
-   for (j = 0; j < 4; j++)
-      debug_printf("sample %d texcoord %f %f\n",
-                   j,
-                   store[0+j],
-                   store[4+j]);
-#endif
-
-   {
-      float rgba[NUM_CHANNELS][QUAD_SIZE];
-      sampler->get_samples(sampler,
-                           &store[0],
-                           &store[4],
-                           &store[8],
-                           0.0f, /*store[12],  lodbias */
-                           rgba);
-      memcpy(store, rgba, sizeof rgba);
-   }
-
-#if 0
-   for (j = 0; j < 4; j++)
-      debug_printf("sample %d result %f %f %f %f\n",
-                   j,
-                   store[0+j],
-                   store[4+j],
-                   store[8+j],
-                   store[12+j]);
-#endif
-}
-
-
-#include "lp_bld_type.h"
-#include "lp_bld_intr.h"
-#include "lp_bld_tgsi.h"
-
-
-struct lp_c_sampler_soa
-{
-   struct lp_build_sampler_soa base;
-
-   LLVMValueRef context_ptr;
-
-   LLVMValueRef samplers_ptr;
-
-   /** Coords/texels store */
-   LLVMValueRef store_ptr;
-};
-
-
-static void
-lp_c_sampler_soa_destroy(struct lp_build_sampler_soa *sampler)
-{
-   FREE(sampler);
-}
-
-
-static void
-lp_c_sampler_soa_emit_fetch_texel(struct lp_build_sampler_soa *_sampler,
-                                  LLVMBuilderRef builder,
-                                  struct lp_type type,
-                                  unsigned unit,
-                                  unsigned num_coords,
-                                  const LLVMValueRef *coords,
-                                  LLVMValueRef lodbias,
-                                  LLVMValueRef *texel)
-{
-   struct lp_c_sampler_soa *sampler = (struct lp_c_sampler_soa *)_sampler;
-   LLVMTypeRef vec_type = LLVMTypeOf(coords[0]);
-   LLVMValueRef args[3];
-   unsigned i;
-
-   if(!sampler->samplers_ptr)
-      sampler->samplers_ptr = lp_jit_context_samplers(builder, sampler->context_ptr);
-
-   if(!sampler->store_ptr)
-      sampler->store_ptr = LLVMBuildArrayAlloca(builder,
-                                            vec_type,
-                                            LLVMConstInt(LLVMInt32Type(), 4, 0),
-                                            "texel_store");
-
-   for (i = 0; i < num_coords; i++) {
-      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
-      LLVMValueRef coord_ptr = LLVMBuildGEP(builder, sampler->store_ptr, &index, 1, "");
-      LLVMBuildStore(builder, coords[i], coord_ptr);
-   }
-
-   args[0] = sampler->samplers_ptr;
-   args[1] = LLVMConstInt(LLVMInt32Type(), unit, 0);
-   args[2] = sampler->store_ptr;
-
-   lp_build_intrinsic(builder, "fetch_texel", LLVMVoidType(), args, 3);
-
-   for (i = 0; i < NUM_CHANNELS; ++i) {
-      LLVMValueRef index = LLVMConstInt(LLVMInt32Type(), i, 0);
-      LLVMValueRef texel_ptr = LLVMBuildGEP(builder, sampler->store_ptr, &index, 1, "");
-      texel[i] = LLVMBuildLoad(builder, texel_ptr, "");
-   }
-}
-
-
-struct lp_build_sampler_soa *
-lp_c_sampler_soa_create(LLVMValueRef context_ptr)
-{
-   struct lp_c_sampler_soa *sampler;
-
-   sampler = CALLOC_STRUCT(lp_c_sampler_soa);
-   if(!sampler)
-      return NULL;
-
-   sampler->base.destroy = lp_c_sampler_soa_destroy;
-   sampler->base.emit_fetch_texel = lp_c_sampler_soa_emit_fetch_texel;
-   sampler->context_ptr = context_ptr;
-
-   return &sampler->base;
-}
-
-- 
cgit v1.2.3


From 99e28d4ee3ce995845d9290b735b1fbe1b96886d Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:35:56 +0100
Subject: llvmpipe: adapt to blend changes

---
 src/gallium/drivers/llvmpipe/lp_test_blend.c | 72 ++++++++++++++--------------
 1 file changed, 37 insertions(+), 35 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_test_blend.c b/src/gallium/drivers/llvmpipe/lp_test_blend.c
index 6c29e8d8ac..07f636f8e7 100644
--- a/src/gallium/drivers/llvmpipe/lp_test_blend.c
+++ b/src/gallium/drivers/llvmpipe/lp_test_blend.c
@@ -104,18 +104,18 @@ write_tsv_row(FILE *fp,
 
    fprintf(fp,
            "%s\t%s\t%s\t",
-           blend->rgb_func != blend->alpha_func ? "true" : "false",
-           blend->rgb_src_factor != blend->alpha_src_factor ? "true" : "false",
-           blend->rgb_dst_factor != blend->alpha_dst_factor ? "true" : "false");
+           blend->rt[0].rgb_func != blend->rt[0].alpha_func ? "true" : "false",
+           blend->rt[0].rgb_src_factor != blend->rt[0].alpha_src_factor ? "true" : "false",
+           blend->rt[0].rgb_dst_factor != blend->rt[0].alpha_dst_factor ? "true" : "false");
 
    fprintf(fp,
            "%s\t%s\t%s\t%s\t%s\t%s\n",
-           debug_dump_blend_func(blend->rgb_func, TRUE),
-           debug_dump_blend_factor(blend->rgb_src_factor, TRUE),
-           debug_dump_blend_factor(blend->rgb_dst_factor, TRUE),
-           debug_dump_blend_func(blend->alpha_func, TRUE),
-           debug_dump_blend_factor(blend->alpha_src_factor, TRUE),
-           debug_dump_blend_factor(blend->alpha_dst_factor, TRUE));
+           debug_dump_blend_func(blend->rt[0].rgb_func, TRUE),
+           debug_dump_blend_factor(blend->rt[0].rgb_src_factor, TRUE),
+           debug_dump_blend_factor(blend->rt[0].rgb_dst_factor, TRUE),
+           debug_dump_blend_func(blend->rt[0].alpha_func, TRUE),
+           debug_dump_blend_factor(blend->rt[0].alpha_src_factor, TRUE),
+           debug_dump_blend_factor(blend->rt[0].alpha_dst_factor, TRUE));
 
    fflush(fp);
 }
@@ -137,12 +137,12 @@ dump_blend_type(FILE *fp,
 
    fprintf(fp,
            " %s=%s %s=%s %s=%s %s=%s %s=%s %s=%s",
-           "rgb_func",         debug_dump_blend_func(blend->rgb_func, TRUE),
-           "rgb_src_factor",   debug_dump_blend_factor(blend->rgb_src_factor, TRUE),
-           "rgb_dst_factor",   debug_dump_blend_factor(blend->rgb_dst_factor, TRUE),
-           "alpha_func",       debug_dump_blend_func(blend->alpha_func, TRUE),
-           "alpha_src_factor", debug_dump_blend_factor(blend->alpha_src_factor, TRUE),
-           "alpha_dst_factor", debug_dump_blend_factor(blend->alpha_dst_factor, TRUE));
+           "rgb_func",         debug_dump_blend_func(blend->rt[0].rgb_func, TRUE),
+           "rgb_src_factor",   debug_dump_blend_factor(blend->rt[0].rgb_src_factor, TRUE),
+           "rgb_dst_factor",   debug_dump_blend_factor(blend->rt[0].rgb_dst_factor, TRUE),
+           "alpha_func",       debug_dump_blend_func(blend->rt[0].alpha_func, TRUE),
+           "alpha_src_factor", debug_dump_blend_factor(blend->rt[0].alpha_src_factor, TRUE),
+           "alpha_dst_factor", debug_dump_blend_factor(blend->rt[0].alpha_dst_factor, TRUE));
 
    fprintf(fp, " ...\n");
    fflush(fp);
@@ -401,13 +401,15 @@ compute_blend_ref(const struct pipe_blend_state *blend,
    double src_term[4];
    double dst_term[4];
 
-   compute_blend_ref_term(blend->rgb_src_factor, blend->alpha_src_factor, src, src, dst, con, src_term);
-   compute_blend_ref_term(blend->rgb_dst_factor, blend->alpha_dst_factor, dst, src, dst, con, dst_term);
+   compute_blend_ref_term(blend->rt[0].rgb_src_factor, blend->rt[0].alpha_src_factor,
+                          src, src, dst, con, src_term);
+   compute_blend_ref_term(blend->rt[0].rgb_dst_factor, blend->rt[0].alpha_dst_factor,
+                          dst, src, dst, con, dst_term);
 
    /*
     * Combine RGB terms
     */
-   switch (blend->rgb_func) {
+   switch (blend->rt[0].rgb_func) {
    case PIPE_BLEND_ADD:
       ADD_SAT(res[0], src_term[0], dst_term[0]); /* R */
       ADD_SAT(res[1], src_term[1], dst_term[1]); /* G */
@@ -440,7 +442,7 @@ compute_blend_ref(const struct pipe_blend_state *blend,
    /*
     * Combine A terms
     */
-   switch (blend->alpha_func) {
+   switch (blend->rt[0].alpha_func) {
    case PIPE_BLEND_ADD:
       ADD_SAT(res[3], src_term[3], dst_term[3]); /* A */
       break;
@@ -806,14 +808,14 @@ test_all(unsigned verbose, FILE *fp)
                               continue;
 
                            memset(&blend, 0, sizeof blend);
-                           blend.blend_enable      = 1;
-                           blend.rgb_func          = *rgb_func;
-                           blend.rgb_src_factor    = *rgb_src_factor;
-                           blend.rgb_dst_factor    = *rgb_dst_factor;
-                           blend.alpha_func        = *alpha_func;
-                           blend.alpha_src_factor  = *alpha_src_factor;
-                           blend.alpha_dst_factor  = *alpha_dst_factor;
-                           blend.colormask         = PIPE_MASK_RGBA;
+                           blend.rt[0].blend_enable      = 1;
+                           blend.rt[0].rgb_func          = *rgb_func;
+                           blend.rt[0].rgb_src_factor    = *rgb_src_factor;
+                           blend.rt[0].rgb_dst_factor    = *rgb_dst_factor;
+                           blend.rt[0].alpha_func        = *alpha_func;
+                           blend.rt[0].alpha_src_factor  = *alpha_src_factor;
+                           blend.rt[0].alpha_dst_factor  = *alpha_dst_factor;
+                           blend.rt[0].colormask         = PIPE_MASK_RGBA;
 
                            if(!test_one(verbose, fp, &blend, mode, *type))
                              success = FALSE;
@@ -865,14 +867,14 @@ test_some(unsigned verbose, FILE *fp, unsigned long n)
       type = &blend_types[rand() % num_types];
 
       memset(&blend, 0, sizeof blend);
-      blend.blend_enable      = 1;
-      blend.rgb_func          = *rgb_func;
-      blend.rgb_src_factor    = *rgb_src_factor;
-      blend.rgb_dst_factor    = *rgb_dst_factor;
-      blend.alpha_func        = *alpha_func;
-      blend.alpha_src_factor  = *alpha_src_factor;
-      blend.alpha_dst_factor  = *alpha_dst_factor;
-      blend.colormask         = PIPE_MASK_RGBA;
+      blend.rt[0].blend_enable      = 1;
+      blend.rt[0].rgb_func          = *rgb_func;
+      blend.rt[0].rgb_src_factor    = *rgb_src_factor;
+      blend.rt[0].rgb_dst_factor    = *rgb_dst_factor;
+      blend.rt[0].alpha_func        = *alpha_func;
+      blend.rt[0].alpha_src_factor  = *alpha_src_factor;
+      blend.rt[0].alpha_dst_factor  = *alpha_dst_factor;
+      blend.rt[0].colormask         = PIPE_MASK_RGBA;
 
       if(!test_one(verbose, fp, &blend, mode, *type))
         success = FALSE;
-- 
cgit v1.2.3


From 88388b99fbde4652d39f4876a2fc0235de415a3f Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:36:37 +0100
Subject: cell: adapt to blend changes

---
 src/gallium/drivers/cell/ppu/cell_gen_fragment.c   | 20 ++++-----
 .../drivers/cell/ppu/cell_state_per_fragment.c     | 48 +++++++++++-----------
 2 files changed, 34 insertions(+), 34 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
index 66d4b3b6a3..0dab34075d 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
@@ -408,7 +408,7 @@ gen_blend(const struct pipe_blend_state *blend,
    int one_reg = -1;
    int constR_reg = -1, constG_reg = -1, constB_reg = -1, constA_reg = -1;
 
-   ASSERT(blend->blend_enable);
+   ASSERT(blend->rt[0].blend_enable);
 
    /* packed RGBA -> float colors */
    unpack_colors(f, color_format, fbRGBA_reg,
@@ -420,7 +420,7 @@ gen_blend(const struct pipe_blend_state *blend,
     * because in some cases (like PIPE_BLENDFACTOR_ONE and 
     * PIPE_BLENDFACTOR_ZERO) we can avoid doing unnecessary math.
     */
-   switch (blend->rgb_src_factor) {
+   switch (blend->rt[0].rgb_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* factors = (1,1,1), so term = (R,G,B) */
       spe_move(f, term1R_reg, fragR_reg);
@@ -574,7 +574,7 @@ gen_blend(const struct pipe_blend_state *blend,
     * the full term A*factor, not just the factor itself, because
     * in many cases we can avoid doing unnecessary multiplies.
     */
-   switch (blend->alpha_src_factor) {
+   switch (blend->rt[0].alpha_src_factor) {
    case PIPE_BLENDFACTOR_ZERO:
       /* factor = 0, so term = 0 */
       spe_load_float(f, term1A_reg, 0.0f);
@@ -648,7 +648,7 @@ gen_blend(const struct pipe_blend_state *blend,
     * the full term (Rfb,Gfb,Bfb)*(factor), not just the factor itself, because
     * in many cases we can avoid doing unnecessary multiplies.
     */
-   switch (blend->rgb_dst_factor) {
+   switch (blend->rt[0].rgb_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* factors = (1,1,1), so term = (Rfb,Gfb,Bfb) */
       spe_move(f, term2R_reg, fbR_reg);
@@ -786,7 +786,7 @@ gen_blend(const struct pipe_blend_state *blend,
     * the full term Afb*factor, not just the factor itself, because
     * in many cases we can avoid doing unnecessary multiplies.
     */
-   switch (blend->alpha_dst_factor) {
+   switch (blend->rt[0].alpha_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* factor = 1, so term = Afb */
       spe_move(f, term2A_reg, fbA_reg);
@@ -858,7 +858,7 @@ gen_blend(const struct pipe_blend_state *blend,
    /*
     * Combine Src/Dest RGB terms as per the blend equation.
     */
-   switch (blend->rgb_func) {
+   switch (blend->rt[0].rgb_func) {
    case PIPE_BLEND_ADD:
       spe_fa(f, fragR_reg, term1R_reg, term2R_reg);
       spe_fa(f, fragG_reg, term1G_reg, term2G_reg);
@@ -891,7 +891,7 @@ gen_blend(const struct pipe_blend_state *blend,
    /*
     * Combine Src/Dest A term
     */
-   switch (blend->alpha_func) {
+   switch (blend->rt[0].alpha_func) {
    case PIPE_BLEND_ADD:
       spe_fa(f, fragA_reg, term1A_reg, term2A_reg);
       break;
@@ -2118,7 +2118,7 @@ cell_gen_fragment_function(struct cell_context *cell,
    spe_comment(f, 0, "Fetch quad colors from tile");
    spe_lqx(f, fbRGBA_reg, color_tile_reg, quad_offset_reg);
 
-   if (blend->blend_enable) {
+   if (blend->rt[0].blend_enable) {
       spe_comment(f, 0, "Perform blending");
       gen_blend(blend, blend_color, f, color_format,
                 fragR_reg, fragG_reg, fragB_reg, fragA_reg, fbRGBA_reg);
@@ -2143,9 +2143,9 @@ cell_gen_fragment_function(struct cell_context *cell,
          gen_logicop(blend, f, rgba_reg, fbRGBA_reg);
       }
 
-      if (blend->colormask != PIPE_MASK_RGBA) {
+      if (blend->rt[0].colormask != PIPE_MASK_RGBA) {
          spe_comment(f, 0, "Compute color mask");
-         gen_colormask(f, blend->colormask, color_format, rgba_reg, fbRGBA_reg);
+         gen_colormask(f, blend->rt[0].colormask, color_format, rgba_reg, fbRGBA_reg);
       }
 
       /* Mix fragment colors with framebuffer colors using the quad/pixel mask:
diff --git a/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c b/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
index d97c22b2ef..21af7ed1c3 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_per_fragment.c
@@ -999,23 +999,23 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
    /* Does the selected blend mode make use of the source / destination
     * color (RGB) blend factors?
     */
-   boolean need_color_factor = b->blend_enable
-       && (b->rgb_func != PIPE_BLEND_MIN)
-       && (b->rgb_func != PIPE_BLEND_MAX);
+   boolean need_color_factor = b->rt[0].blend_enable
+       && (b->rt[0].rgb_func != PIPE_BLEND_MIN)
+       && (b->rt[0].rgb_func != PIPE_BLEND_MAX);
 
    /* Does the selected blend mode make use of the source / destination
     * alpha blend factors?
     */
-   boolean need_alpha_factor = b->blend_enable
-       && (b->alpha_func != PIPE_BLEND_MIN)
-       && (b->alpha_func != PIPE_BLEND_MAX);
+   boolean need_alpha_factor = b->rt[0].blend_enable
+       && (b->rt[0].alpha_func != PIPE_BLEND_MIN)
+       && (b->rt[0].alpha_func != PIPE_BLEND_MAX);
 
 
-   if (b->blend_enable) {
-      sF[0] = b->rgb_src_factor;
+   if (b->rt[0].blend_enable) {
+      sF[0] = b->rt[0].rgb_src_factor;
       sF[1] = sF[0];
       sF[2] = sF[0];
-      switch (b->alpha_src_factor & 0x0f) {
+      switch (b->rt[0].alpha_src_factor & 0x0f) {
       case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
          sF[3] = PIPE_BLENDFACTOR_ONE;
          break;
@@ -1023,30 +1023,30 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
       case PIPE_BLENDFACTOR_DST_COLOR:
       case PIPE_BLENDFACTOR_CONST_COLOR:
       case PIPE_BLENDFACTOR_SRC1_COLOR:
-         sF[3] = b->alpha_src_factor + 1;
+         sF[3] = b->rt[0].alpha_src_factor + 1;
          break;
       default:
-         sF[3] = b->alpha_src_factor;
+         sF[3] = b->rt[0].alpha_src_factor;
       }
 
-      dF[0] = b->rgb_dst_factor;
+      dF[0] = b->rt[0].rgb_dst_factor;
       dF[1] = dF[0];
       dF[2] = dF[0];
-      switch (b->alpha_dst_factor & 0x0f) {
+      switch (b->rt[0].alpha_dst_factor & 0x0f) {
       case PIPE_BLENDFACTOR_SRC_COLOR:
       case PIPE_BLENDFACTOR_DST_COLOR:
       case PIPE_BLENDFACTOR_CONST_COLOR:
       case PIPE_BLENDFACTOR_SRC1_COLOR:
-         dF[3] = b->alpha_dst_factor + 1;
+         dF[3] = b->rt[0].alpha_dst_factor + 1;
          break;
       default:
-         dF[3] = b->alpha_dst_factor;
+         dF[3] = b->rt[0].alpha_dst_factor;
       }
 
-      func[0] = b->rgb_func;
+      func[0] = b->rt[0].rgb_func;
       func[1] = func[0];
       func[2] = func[0];
-      func[3] = b->alpha_func;
+      func[3] = b->rt[0].alpha_func;
    } else {
       sF[0] = PIPE_BLENDFACTOR_ONE;
       sF[1] = PIPE_BLENDFACTOR_ONE;
@@ -1067,7 +1067,7 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
    /* If alpha writing is enabled and the alpha blend mode requires use of
     * the alpha factor, calculate the alpha factor.
     */
-   if (((b->colormask & 8) != 0) && need_alpha_factor) {
+   if (((b->rt[0].colormask & 8) != 0) && need_alpha_factor) {
       src_factor[3] = emit_alpha_factor_calculation(f, sF[3], const_color[3],
                                                     frag[3], pixel[3]);
 
@@ -1091,8 +1091,8 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
       src_factor[2] = dst_factor[3];
    } else if (need_color_factor) {
       emit_color_factor_calculation(f,
-                                    b->rgb_src_factor,
-                                    b->colormask,
+                                    b->rt[0].rgb_src_factor,
+                                    b->rt[0].colormask,
                                     frag, pixel, const_color, src_factor);
    }
 
@@ -1111,15 +1111,15 @@ cell_generate_alpha_blend(struct cell_blend_state *cb)
       dst_factor[2] = src_factor[2];
    } else if (need_color_factor) {
       emit_color_factor_calculation(f,
-                                    b->rgb_dst_factor,
-                                    b->colormask,
+                                    b->rt[0].rgb_dst_factor,
+                                    b->rt[0].colormask,
                                     frag, pixel, const_color, dst_factor);
    }
 
 
 
    for (i = 0; i < 4; ++i) {
-      if ((b->colormask & (1U << i)) != 0) {
+      if ((b->rt[0].colormask & (1U << i)) != 0) {
          emit_blend_calculation(f,
                                 func[i], sF[i], dF[i],
                                 frag[i], src_factor[i],
@@ -1216,7 +1216,7 @@ cell_generate_logic_op(struct spe_function *f,
 
    /* Short-circuit the noop and invert cases.
     */
-   if ((logic_op == PIPE_LOGICOP_NOOP) || (blend->colormask == 0)) {
+   if ((logic_op == PIPE_LOGICOP_NOOP) || (blend->rt[0].colormask == 0)) {
       spe_bi(f, 0, 0, 0);
       return;
    } else if (logic_op == PIPE_LOGICOP_INVERT) {
-- 
cgit v1.2.3


From bc160fc9b293f155d5ed48912dc09c111b97ec74 Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:37:09 +0100
Subject: i915g: adapt to blend changes

---
 src/gallium/drivers/i915/i915_state.c | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i915/i915_state.c b/src/gallium/drivers/i915/i915_state.c
index 0fab6e1bc3..db98256d2a 100644
--- a/src/gallium/drivers/i915/i915_state.c
+++ b/src/gallium/drivers/i915/i915_state.c
@@ -105,13 +105,13 @@ i915_create_blend_state(struct pipe_context *pipe,
    struct i915_blend_state *cso_data = CALLOC_STRUCT( i915_blend_state );
 
    {
-      unsigned eqRGB  = blend->rgb_func;
-      unsigned srcRGB = blend->rgb_src_factor;
-      unsigned dstRGB = blend->rgb_dst_factor;
+      unsigned eqRGB  = blend->rt[0].rgb_func;
+      unsigned srcRGB = blend->rt[0].rgb_src_factor;
+      unsigned dstRGB = blend->rt[0].rgb_dst_factor;
 
-      unsigned eqA    = blend->alpha_func;
-      unsigned srcA   = blend->alpha_src_factor;
-      unsigned dstA   = blend->alpha_dst_factor;
+      unsigned eqA    = blend->rt[0].alpha_func;
+      unsigned srcA   = blend->rt[0].alpha_src_factor;
+      unsigned dstA   = blend->rt[0].alpha_dst_factor;
 
       /* Special handling for MIN/MAX filter modes handled at
        * state_tracker level.
@@ -148,22 +148,22 @@ i915_create_blend_state(struct pipe_context *pipe,
    if (blend->dither)
       cso_data->LIS5 |= S5_COLOR_DITHER_ENABLE;
 
-   if ((blend->colormask & PIPE_MASK_R) == 0)
+   if ((blend->rt[0].colormask & PIPE_MASK_R) == 0)
       cso_data->LIS5 |= S5_WRITEDISABLE_RED;
 
-   if ((blend->colormask & PIPE_MASK_G) == 0)
+   if ((blend->rt[0].colormask & PIPE_MASK_G) == 0)
       cso_data->LIS5 |= S5_WRITEDISABLE_GREEN;
 
-   if ((blend->colormask & PIPE_MASK_B) == 0)
+   if ((blend->rt[0].colormask & PIPE_MASK_B) == 0)
       cso_data->LIS5 |= S5_WRITEDISABLE_BLUE;
 
-   if ((blend->colormask & PIPE_MASK_A) == 0)
+   if ((blend->rt[0].colormask & PIPE_MASK_A) == 0)
       cso_data->LIS5 |= S5_WRITEDISABLE_ALPHA;
 
-   if (blend->blend_enable) {
-      unsigned funcRGB = blend->rgb_func;
-      unsigned srcRGB  = blend->rgb_src_factor;
-      unsigned dstRGB  = blend->rgb_dst_factor;
+   if (blend->rt[0].blend_enable) {
+      unsigned funcRGB = blend->rt[0].rgb_func;
+      unsigned srcRGB  = blend->rt[0].rgb_src_factor;
+      unsigned dstRGB  = blend->rt[0].rgb_dst_factor;
 
       cso_data->LIS6 |= (S6_CBUF_BLEND_ENABLE |
                          SRC_BLND_FACT(i915_translate_blend_factor(srcRGB)) |
-- 
cgit v1.2.3


From 4e0c825a799b9982ad62125df1aa876b3b0c110d Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:37:27 +0100
Subject: i965g: adapt to blend changes

---
 src/gallium/drivers/i965/brw_pipe_blend.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i965/brw_pipe_blend.c b/src/gallium/drivers/i965/brw_pipe_blend.c
index b759a910b6..21f786f871 100644
--- a/src/gallium/drivers/i965/brw_pipe_blend.c
+++ b/src/gallium/drivers/i965/brw_pipe_blend.c
@@ -118,14 +118,14 @@ static void *brw_create_blend_state( struct pipe_context *pipe,
       blend->cc2.logicop_enable = 1;
       blend->cc5.logicop_func = translate_logicop(templ->logicop_func);
    } 
-   else if (templ->blend_enable) {
-      blend->cc6.dest_blend_factor = translate_blend_factor(templ->rgb_dst_factor);
-      blend->cc6.src_blend_factor = translate_blend_factor(templ->rgb_src_factor);
-      blend->cc6.blend_function = translate_blend_equation(templ->rgb_func);
+   else if (templ->rt[0].blend_enable) {
+      blend->cc6.dest_blend_factor = translate_blend_factor(templ->rt[0].rgb_dst_factor);
+      blend->cc6.src_blend_factor = translate_blend_factor(templ->rt[0].rgb_src_factor);
+      blend->cc6.blend_function = translate_blend_equation(templ->rt[0].rgb_func);
 
-      blend->cc5.ia_dest_blend_factor = translate_blend_factor(templ->alpha_dst_factor);
-      blend->cc5.ia_src_blend_factor = translate_blend_factor(templ->alpha_src_factor);
-      blend->cc5.ia_blend_function = translate_blend_equation(templ->alpha_func);
+      blend->cc5.ia_dest_blend_factor = translate_blend_factor(templ->rt[0].alpha_dst_factor);
+      blend->cc5.ia_src_blend_factor = translate_blend_factor(templ->rt[0].alpha_src_factor);
+      blend->cc5.ia_blend_function = translate_blend_equation(templ->rt[0].alpha_func);
 
       blend->cc3.blend_enable = 1;
       blend->cc3.ia_blend_enable = 
@@ -146,10 +146,10 @@ static void *brw_create_blend_state( struct pipe_context *pipe,
 
    /* Per-surface color mask -- just follow global state:
     */
-   blend->ss0.writedisable_red   = (templ->colormask & PIPE_MASK_R) ? 0 : 1;
-   blend->ss0.writedisable_green = (templ->colormask & PIPE_MASK_G) ? 0 : 1;
-   blend->ss0.writedisable_blue  = (templ->colormask & PIPE_MASK_B) ? 0 : 1;
-   blend->ss0.writedisable_alpha = (templ->colormask & PIPE_MASK_A) ? 0 : 1;
+   blend->ss0.writedisable_red   = (templ->rt[0].colormask & PIPE_MASK_R) ? 0 : 1;
+   blend->ss0.writedisable_green = (templ->rt[0].colormask & PIPE_MASK_G) ? 0 : 1;
+   blend->ss0.writedisable_blue  = (templ->rt[0].colormask & PIPE_MASK_B) ? 0 : 1;
+   blend->ss0.writedisable_alpha = (templ->rt[0].colormask & PIPE_MASK_A) ? 0 : 1;
 
    return (void *)blend;
 }
-- 
cgit v1.2.3


From d241964e47ac5576e754d6e6ae19fece8bf8752c Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:37:51 +0100
Subject: r300g: adapt to blend changes

---
 src/gallium/drivers/r300/r300_state.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index e2ec0bc5bd..c5e313ce43 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -164,15 +164,15 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
 {
     struct r300_blend_state* blend = CALLOC_STRUCT(r300_blend_state);
 
-    if (state->blend_enable)
+    if (state->rt[0].blend_enable)
     {
-        unsigned eqRGB = state->rgb_func;
-        unsigned srcRGB = state->rgb_src_factor;
-        unsigned dstRGB = state->rgb_dst_factor;
+        unsigned eqRGB = state->rt[0].rgb_func;
+        unsigned srcRGB = state->rt[0].rgb_src_factor;
+        unsigned dstRGB = state->rt[0].rgb_dst_factor;
 
-        unsigned eqA = state->alpha_func;
-        unsigned srcA = state->alpha_src_factor;
-        unsigned dstA = state->alpha_dst_factor;
+        unsigned eqA = state->rt[0].alpha_func;
+        unsigned srcA = state->rt[0].alpha_src_factor;
+        unsigned dstA = state->rt[0].alpha_dst_factor;
 
         /* despite the name, ALPHA_BLEND_ENABLE has nothing to do with alpha,
          * this is just the crappy D3D naming */
@@ -290,16 +290,16 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
     }
 
     /* Color Channel Mask */
-    if (state->colormask & PIPE_MASK_R) {
+    if (state->rt[0].colormask & PIPE_MASK_R) {
         blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_RED_MASK0;
     }
-    if (state->colormask & PIPE_MASK_G) {
+    if (state->rt[0].colormask & PIPE_MASK_G) {
         blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0;
     }
-    if (state->colormask & PIPE_MASK_B) {
+    if (state->rt[0].colormask & PIPE_MASK_B) {
         blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0;
     }
-    if (state->colormask & PIPE_MASK_A) {
+    if (state->rt[0].colormask & PIPE_MASK_A) {
         blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0;
     }
 
-- 
cgit v1.2.3


From e908abf1740dce1a3b391ce5bcfda4f2a0b480de Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:38:32 +0100
Subject: nouveau: adapt drviers to blend changes

---
 src/gallium/drivers/nv04/nv04_state.c | 10 +++++-----
 src/gallium/drivers/nv10/nv10_state.c | 20 ++++++++++----------
 src/gallium/drivers/nv20/nv20_state.c | 20 ++++++++++----------
 src/gallium/drivers/nv30/nv30_state.c | 20 ++++++++++----------
 src/gallium/drivers/nv40/nv40_state.c | 22 +++++++++++-----------
 src/gallium/drivers/nv50/nv50_state.c | 22 +++++++++++-----------
 6 files changed, 57 insertions(+), 57 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c
index b67f1e16b1..1fdb764b53 100644
--- a/src/gallium/drivers/nv04/nv04_state.c
+++ b/src/gallium/drivers/nv04/nv04_state.c
@@ -17,11 +17,11 @@ nv04_blend_state_create(struct pipe_context *pipe,
 
 	cb = MALLOC(sizeof(struct nv04_blend_state));
 
-	cb->b_enable = cso->blend_enable ? 1 : 0;
-	cb->b_src = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
-			 (nvgl_blend_func(cso->rgb_src_factor)));
-	cb->b_dst = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
-			 (nvgl_blend_func(cso->rgb_dst_factor)));
+	cb->b_enable = cso->rt[0].blend_enable ? 1 : 0;
+	cb->b_src = ((nvgl_blend_func(cso->rt[0].alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_src_factor)));
+	cb->b_dst = ((nvgl_blend_func(cso->rt[0].alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_dst_factor)));
 	
 
 	return (void *)cb;
diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c
index ad7def53b1..9aafc60411 100644
--- a/src/gallium/drivers/nv10/nv10_state.c
+++ b/src/gallium/drivers/nv10/nv10_state.c
@@ -17,16 +17,16 @@ nv10_blend_state_create(struct pipe_context *pipe,
 
 	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->b_enable = cso->rt[0].blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->rt[0].alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->rt[0].alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_dst_factor)));
+
+	cb->c_mask = (((cso->rt[0].colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
 
 	cb->d_enable = cso->dither ? 1 : 0;
 
diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c
index 45697a60ef..83335c790a 100644
--- a/src/gallium/drivers/nv20/nv20_state.c
+++ b/src/gallium/drivers/nv20/nv20_state.c
@@ -17,16 +17,16 @@ nv20_blend_state_create(struct pipe_context *pipe,
 
 	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->b_enable = cso->rt[0].blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->rt[0].alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->rt[0].alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rt[0].rgb_dst_factor)));
+
+	cb->c_mask = (((cso->rt[0].colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->rt[0].colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
 
 	cb->d_enable = cso->dither ? 1 : 0;
 
diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c
index 065c927a10..1f4c964270 100644
--- a/src/gallium/drivers/nv30/nv30_state.c
+++ b/src/gallium/drivers/nv30/nv30_state.c
@@ -16,27 +16,27 @@ nv30_blend_state_create(struct pipe_context *pipe,
 	struct nv30_blend_state *bso = CALLOC(1, sizeof(*bso));
 	struct nouveau_stateobj *so = so_new(5, 8, 0);
 
-	if (cso->blend_enable) {
+	if (cso->rt[0].blend_enable) {
 		so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 3);
 		so_data  (so, 1);
-		so_data  (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) |
-			       nvgl_blend_func(cso->rgb_src_factor));
-		so_data  (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 |
-			      nvgl_blend_func(cso->rgb_dst_factor));
+		so_data  (so, (nvgl_blend_func(cso->rt[0].alpha_src_factor) << 16) |
+			       nvgl_blend_func(cso->rt[0].rgb_src_factor));
+		so_data  (so, nvgl_blend_func(cso->rt[0].alpha_dst_factor) << 16 |
+			      nvgl_blend_func(cso->rt[0].rgb_dst_factor));
 		/* FIXME: Gallium assumes GL_EXT_blend_func_separate.
 		   It is not the case for NV30 */
 		so_method(so, rankine, NV34TCL_BLEND_EQUATION, 1);
-		so_data  (so, nvgl_blend_eqn(cso->rgb_func));
+		so_data  (so, nvgl_blend_eqn(cso->rt[0].rgb_func));
 	} else {
 		so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 1);
 		so_data  (so, 0);
 	}
 
 	so_method(so, rankine, NV34TCL_COLOR_MASK, 1);
-	so_data  (so, (((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)));
+	so_data  (so, (((cso->rt[0].colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_G) ? (0x01 <<  8) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_B) ? (0x01 <<  0) : 0)));
 
 	if (cso->logicop_enable) {
 		so_method(so, rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2);
diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c
index 7d990f7d56..7a63fba18c 100644
--- a/src/gallium/drivers/nv40/nv40_state.c
+++ b/src/gallium/drivers/nv40/nv40_state.c
@@ -18,26 +18,26 @@ nv40_blend_state_create(struct pipe_context *pipe,
 	struct nv40_blend_state *bso = CALLOC(1, sizeof(*bso));
 	struct nouveau_stateobj *so = so_new(5, 8, 0);
 
-	if (cso->blend_enable) {
+	if (cso->rt[0].blend_enable) {
 		so_method(so, curie, NV40TCL_BLEND_ENABLE, 3);
 		so_data  (so, 1);
-		so_data  (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) |
-			       nvgl_blend_func(cso->rgb_src_factor));
-		so_data  (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 |
-			      nvgl_blend_func(cso->rgb_dst_factor));
+		so_data  (so, (nvgl_blend_func(cso->rt[0].alpha_src_factor) << 16) |
+			       nvgl_blend_func(cso->rt[0].rgb_src_factor));
+		so_data  (so, nvgl_blend_func(cso->rt[0].alpha_dst_factor) << 16 |
+			      nvgl_blend_func(cso->rt[0].rgb_dst_factor));
 		so_method(so, curie, NV40TCL_BLEND_EQUATION, 1);
-		so_data  (so, nvgl_blend_eqn(cso->alpha_func) << 16 |
-			      nvgl_blend_eqn(cso->rgb_func));
+		so_data  (so, nvgl_blend_eqn(cso->rt[0].alpha_func) << 16 |
+			      nvgl_blend_eqn(cso->rt[0].rgb_func));
 	} else {
 		so_method(so, curie, NV40TCL_BLEND_ENABLE, 1);
 		so_data  (so, 0);
 	}
 
 	so_method(so, curie, NV40TCL_COLOR_MASK, 1);
-	so_data  (so, (((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)));
+	so_data  (so, (((cso->rt[0].colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_G) ? (0x01 <<  8) : 0) |
+		       ((cso->rt[0].colormask & PIPE_MASK_B) ? (0x01 <<  0) : 0)));
 
 	if (cso->logicop_enable) {
 		so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2);
diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c
index 6ab33be663..815dfa4f6f 100644
--- a/src/gallium/drivers/nv50/nv50_state.c
+++ b/src/gallium/drivers/nv50/nv50_state.c
@@ -44,7 +44,7 @@ nv50_blend_state_create(struct pipe_context *pipe,
 	 * 	- dither
 	 */
 
-	if (cso->blend_enable == 0) {
+	if (cso->rt[0].blend_enable == 0) {
 		so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
 		for (i = 0; i < 8; i++)
 			so_data(so, 0);
@@ -53,13 +53,13 @@ nv50_blend_state_create(struct pipe_context *pipe,
 		for (i = 0; i < 8; i++)
 			so_data(so, 1);
 		so_method(so, tesla, NV50TCL_BLEND_EQUATION_RGB, 5);
-		so_data  (so, nvgl_blend_eqn(cso->rgb_func));
-		so_data  (so, 0x4000 | nvgl_blend_func(cso->rgb_src_factor));
-		so_data  (so, 0x4000 | nvgl_blend_func(cso->rgb_dst_factor));
-		so_data  (so, nvgl_blend_eqn(cso->alpha_func));
-		so_data  (so, 0x4000 | nvgl_blend_func(cso->alpha_src_factor));
+		so_data  (so, nvgl_blend_eqn(cso->rt[0].rgb_func));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rt[0].rgb_src_factor));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rt[0].rgb_dst_factor));
+		so_data  (so, nvgl_blend_eqn(cso->rt[0].alpha_func));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rt[0].alpha_src_factor));
 		so_method(so, tesla, NV50TCL_BLEND_FUNC_DST_ALPHA, 1);
-		so_data  (so, 0x4000 | nvgl_blend_func(cso->alpha_dst_factor));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rt[0].alpha_dst_factor));
 	}
 
 	if (cso->logicop_enable == 0 ) {
@@ -71,13 +71,13 @@ nv50_blend_state_create(struct pipe_context *pipe,
 		so_data  (so, nvgl_logicop_func(cso->logicop_func));
 	}
 
-	if (cso->colormask & PIPE_MASK_R)
+	if (cso->rt[0].colormask & PIPE_MASK_R)
 		cmask |= (1 << 0);
-	if (cso->colormask & PIPE_MASK_G)
+	if (cso->rt[0].colormask & PIPE_MASK_G)
 		cmask |= (1 << 4);
-	if (cso->colormask & PIPE_MASK_B)
+	if (cso->rt[0].colormask & PIPE_MASK_B)
 		cmask |= (1 << 8);
-	if (cso->colormask & PIPE_MASK_A)
+	if (cso->rt[0].colormask & PIPE_MASK_A)
 		cmask |= (1 << 12);
 	so_method(so, tesla, NV50TCL_COLOR_MASK(0), 8);
 	for (i = 0; i < 8; i++)
-- 
cgit v1.2.3


From 4e6bfa6de1940f1da018aace444ae7ca16792d28 Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:39:16 +0100
Subject: softpipe: adapt to blend changes

---
 src/gallium/drivers/softpipe/sp_quad_blend.c    | 44 ++++++++++++-------------
 src/gallium/drivers/softpipe/sp_video_context.c | 17 +++++-----
 2 files changed, 31 insertions(+), 30 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c
index d9babe81da..abca323314 100644
--- a/src/gallium/drivers/softpipe/sp_quad_blend.c
+++ b/src/gallium/drivers/softpipe/sp_quad_blend.c
@@ -234,7 +234,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute src/first term RGB
     */
-   switch (softpipe->blend->rgb_src_factor) {
+   switch (softpipe->blend->rt[0].rgb_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
       VEC4_COPY(source[0], quadColor[0]); /* R */
       VEC4_COPY(source[1], quadColor[1]); /* G */
@@ -384,7 +384,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute src/first term A
     */
-   switch (softpipe->blend->alpha_src_factor) {
+   switch (softpipe->blend->rt[0].alpha_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
       VEC4_COPY(source[3], quadColor[3]); /* A */
       break;
@@ -453,7 +453,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute dest/second term RGB
     */
-   switch (softpipe->blend->rgb_dst_factor) {
+   switch (softpipe->blend->rt[0].rgb_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* dest = dest * 1   NO-OP, leave dest as-is */
       break;
@@ -593,7 +593,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute dest/second term A
     */
-   switch (softpipe->blend->alpha_dst_factor) {
+   switch (softpipe->blend->rt[0].alpha_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* dest = dest * 1   NO-OP, leave dest as-is */
       break;
@@ -656,7 +656,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Combine RGB terms
     */
-   switch (softpipe->blend->rgb_func) {
+   switch (softpipe->blend->rt[0].rgb_func) {
    case PIPE_BLEND_ADD:
       VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */
       VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */
@@ -689,7 +689,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Combine A terms
     */
-   switch (softpipe->blend->alpha_func) {
+   switch (softpipe->blend->rt[0].alpha_func) {
    case PIPE_BLEND_ADD:
       VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */
       break;
@@ -718,19 +718,19 @@ colormask_quad(struct quad_stage *qs,
    struct softpipe_context *softpipe = qs->softpipe;
 
    /* R */
-   if (!(softpipe->blend->colormask & PIPE_MASK_R))
+   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_R))
       COPY_4V(quadColor[0], dest[0]);
 
    /* G */
-   if (!(softpipe->blend->colormask & PIPE_MASK_G))
+   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_G))
       COPY_4V(quadColor[1], dest[1]);
 
    /* B */
-   if (!(softpipe->blend->colormask & PIPE_MASK_B))
+   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_B))
       COPY_4V(quadColor[2], dest[2]);
 
    /* A */
-   if (!(softpipe->blend->colormask & PIPE_MASK_A))
+   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_A))
       COPY_4V(quadColor[3], dest[3]);
 }
 
@@ -773,11 +773,11 @@ blend_fallback(struct quad_stage *qs,
          if (blend->logicop_enable) {
             logicop_quad( qs, quadColor, dest );
          }
-         else if (blend->blend_enable) {
+         else if (blend->rt[0].blend_enable) {
             blend_quad( qs, quadColor, dest );
          }
 
-         if (blend->colormask != 0xf)
+         if (blend->rt[0].colormask != 0xf)
             colormask_quad( qs, quadColor, dest );
    
          /* Output color values
@@ -954,23 +954,23 @@ choose_blend_quad(struct quad_stage *qs,
       qs->run = blend_noop;
    }
    else if (!softpipe->blend->logicop_enable &&
-            softpipe->blend->colormask == 0xf &&
+            softpipe->blend->rt[0].colormask == 0xf &&
             softpipe->framebuffer.nr_cbufs == 1)
    {
-      if (!blend->blend_enable) {
+      if (!blend->rt[0].blend_enable) {
          qs->run = single_output_color;
       }
-      else if (blend->rgb_src_factor == blend->alpha_src_factor &&
-               blend->rgb_dst_factor == blend->alpha_dst_factor &&
-               blend->rgb_func == blend->alpha_func)
+      else if (blend->rt[0].rgb_src_factor == blend->rt[0].alpha_src_factor &&
+               blend->rt[0].rgb_dst_factor == blend->rt[0].alpha_dst_factor &&
+               blend->rt[0].rgb_func == blend->rt[0].alpha_func)
       {
-         if (blend->alpha_func == PIPE_BLEND_ADD) {
-            if (blend->rgb_src_factor == PIPE_BLENDFACTOR_ONE &&
-                blend->rgb_dst_factor == PIPE_BLENDFACTOR_ONE) {
+         if (blend->rt[0].alpha_func == PIPE_BLEND_ADD) {
+            if (blend->rt[0].rgb_src_factor == PIPE_BLENDFACTOR_ONE &&
+                blend->rt[0].rgb_dst_factor == PIPE_BLENDFACTOR_ONE) {
                qs->run = blend_single_add_one_one;
             }
-            else if (blend->rgb_src_factor == PIPE_BLENDFACTOR_SRC_ALPHA &&
-                blend->rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA)
+            else if (blend->rt[0].rgb_src_factor == PIPE_BLENDFACTOR_SRC_ALPHA &&
+                blend->rt[0].rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA)
                qs->run = blend_single_add_src_alpha_inv_src_alpha;
 
          }
diff --git a/src/gallium/drivers/softpipe/sp_video_context.c b/src/gallium/drivers/softpipe/sp_video_context.c
index cae2d3efc5..d0723ebe9e 100644
--- a/src/gallium/drivers/softpipe/sp_video_context.c
+++ b/src/gallium/drivers/softpipe/sp_video_context.c
@@ -185,17 +185,18 @@ init_pipe_state(struct sp_mpeg12_context *ctx)
    ctx->rast = ctx->pipe->create_rasterizer_state(ctx->pipe, &rast);
    ctx->pipe->bind_rasterizer_state(ctx->pipe, ctx->rast);
 
-   blend.blend_enable = 0;
-   blend.rgb_func = PIPE_BLEND_ADD;
-   blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE;
-   blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
-   blend.alpha_func = PIPE_BLEND_ADD;
-   blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE;
-   blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
+   blend.independent_blend_enable = 0;
+   blend.rt[0].blend_enable = 0;
+   blend.rt[0].rgb_func = PIPE_BLEND_ADD;
+   blend.rt[0].rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+   blend.rt[0].rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
+   blend.rt[0].alpha_func = PIPE_BLEND_ADD;
+   blend.rt[0].alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+   blend.rt[0].alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
    blend.logicop_enable = 0;
    blend.logicop_func = PIPE_LOGICOP_CLEAR;
    /* Needed to allow color writes to FB, even if blending disabled */
-   blend.colormask = PIPE_MASK_RGBA;
+   blend.rt[0].colormask = PIPE_MASK_RGBA;
    blend.dither = 0;
    ctx->blend = ctx->pipe->create_blend_state(ctx->pipe, &blend);
    ctx->pipe->bind_blend_state(ctx->pipe, ctx->blend);
-- 
cgit v1.2.3


From 6893948d482ab69a41290d1b0172834197ee48f2 Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:39:39 +0100
Subject: svga: adapt to blend changes

---
 src/gallium/drivers/svga/svga_pipe_blend.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_pipe_blend.c b/src/gallium/drivers/svga/svga_pipe_blend.c
index 855d228755..4029299311 100644
--- a/src/gallium/drivers/svga/svga_pipe_blend.c
+++ b/src/gallium/drivers/svga/svga_pipe_blend.c
@@ -182,15 +182,15 @@ svga_create_blend_state(struct pipe_context *pipe,
          }
       }
       else {
-         blend->rt[i].blend_enable   = templ->blend_enable;
+         blend->rt[i].blend_enable   = templ->rt[0].blend_enable;
 
-         if (templ->blend_enable) {
-            blend->rt[i].srcblend       = svga_translate_blend_factor(templ->rgb_src_factor);
-            blend->rt[i].dstblend       = svga_translate_blend_factor(templ->rgb_dst_factor);
-            blend->rt[i].blendeq        = svga_translate_blend_func(templ->rgb_func);
-            blend->rt[i].srcblend_alpha = svga_translate_blend_factor(templ->alpha_src_factor);
-            blend->rt[i].dstblend_alpha = svga_translate_blend_factor(templ->alpha_dst_factor);
-            blend->rt[i].blendeq_alpha  = svga_translate_blend_func(templ->alpha_func);
+         if (templ->rt[0].blend_enable) {
+            blend->rt[i].srcblend       = svga_translate_blend_factor(templ->rt[0].rgb_src_factor);
+            blend->rt[i].dstblend       = svga_translate_blend_factor(templ->rt[0].rgb_dst_factor);
+            blend->rt[i].blendeq        = svga_translate_blend_func(templ->rt[0].rgb_func);
+            blend->rt[i].srcblend_alpha = svga_translate_blend_factor(templ->rt[0].alpha_src_factor);
+            blend->rt[i].dstblend_alpha = svga_translate_blend_factor(templ->rt[0].alpha_dst_factor);
+            blend->rt[i].blendeq_alpha  = svga_translate_blend_func(templ->rt[0].alpha_func);
 
             if (blend->rt[i].srcblend_alpha != blend->rt[i].srcblend ||
                 blend->rt[i].dstblend_alpha != blend->rt[i].dstblend ||
@@ -201,7 +201,7 @@ svga_create_blend_state(struct pipe_context *pipe,
          }
       }
 
-      blend->rt[i].writemask = templ->colormask;
+      blend->rt[i].writemask = templ->rt[0].colormask;
    }
 
    return blend;
-- 
cgit v1.2.3


From 13b74b585a013f0f089ea14c9512e3d709bdca67 Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 16:41:57 +0100
Subject: trace: adapt to blend changes

---
 src/gallium/drivers/trace/tr_dump_state.c | 32 ++++++++++++++++++++-----------
 1 file changed, 21 insertions(+), 11 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/trace/tr_dump_state.c b/src/gallium/drivers/trace/tr_dump_state.c
index 32f61f8c94..dbca5de0a5 100644
--- a/src/gallium/drivers/trace/tr_dump_state.c
+++ b/src/gallium/drivers/trace/tr_dump_state.c
@@ -323,9 +323,23 @@ void trace_dump_depth_stencil_alpha_state(const struct pipe_depth_stencil_alpha_
    trace_dump_struct_end();
 }
 
+static void trace_dump_rt_blend_state(const struct pipe_rt_blend_state *state)
+{
+   trace_dump_member(uint, state, rgb_func);
+   trace_dump_member(uint, state, rgb_src_factor);
+   trace_dump_member(uint, state, rgb_dst_factor);
+
+   trace_dump_member(uint, state, alpha_func);
+   trace_dump_member(uint, state, alpha_src_factor);
+   trace_dump_member(uint, state, alpha_dst_factor);
+
+   trace_dump_member(uint, state, colormask);
+
+}
 
 void trace_dump_blend_state(const struct pipe_blend_state *state)
 {
+   unsigned valid_entries = 1;
    if (!trace_dumping_enabled_locked())
       return;
 
@@ -336,21 +350,17 @@ void trace_dump_blend_state(const struct pipe_blend_state *state)
 
    trace_dump_struct_begin("pipe_blend_state");
 
-   trace_dump_member(bool, state, blend_enable);
-
-   trace_dump_member(uint, state, rgb_func);
-   trace_dump_member(uint, state, rgb_src_factor);
-   trace_dump_member(uint, state, rgb_dst_factor);
-
-   trace_dump_member(uint, state, alpha_func);
-   trace_dump_member(uint, state, alpha_src_factor);
-   trace_dump_member(uint, state, alpha_dst_factor);
+   trace_dump_member(bool, state, dither);
 
    trace_dump_member(bool, state, logicop_enable);
    trace_dump_member(uint, state, logicop_func);
 
-   trace_dump_member(uint, state, colormask);
-   trace_dump_member(bool, state, dither);
+   trace_dump_member(bool, state, independent_blend_enable);
+
+   if (state->independent_blend_enable)
+      valid_entries = PIPE_MAX_COLOR_BUFS;
+
+   trace_dump_struct_array(rt_blend_state, state->rt, valid_entries);
 
    trace_dump_struct_end();
 }
-- 
cgit v1.2.3


From df4395198ce433251c8bd2d497543c7b389e5aca Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Mon, 25 Jan 2010 20:20:52 +0100
Subject: softpipe: enable new blend functionality

works with tests/drawbuffers2
---
 src/gallium/drivers/softpipe/sp_quad_blend.c | 35 ++++++++++++++--------------
 src/gallium/drivers/softpipe/sp_screen.c     |  4 ++++
 2 files changed, 21 insertions(+), 18 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c
index abca323314..a1fe5192b0 100644
--- a/src/gallium/drivers/softpipe/sp_quad_blend.c
+++ b/src/gallium/drivers/softpipe/sp_quad_blend.c
@@ -224,7 +224,8 @@ logicop_quad(struct quad_stage *qs,
 static void
 blend_quad(struct quad_stage *qs, 
            float (*quadColor)[4],
-           float (*dest)[4])
+           float (*dest)[4],
+           unsigned cbuf)
 {
    static const float zero[4] = { 0, 0, 0, 0 };
    static const float one[4] = { 1, 1, 1, 1 };
@@ -234,7 +235,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute src/first term RGB
     */
-   switch (softpipe->blend->rt[0].rgb_src_factor) {
+   switch (softpipe->blend->rt[cbuf].rgb_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
       VEC4_COPY(source[0], quadColor[0]); /* R */
       VEC4_COPY(source[1], quadColor[1]); /* G */
@@ -384,7 +385,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute src/first term A
     */
-   switch (softpipe->blend->rt[0].alpha_src_factor) {
+   switch (softpipe->blend->rt[cbuf].alpha_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
       VEC4_COPY(source[3], quadColor[3]); /* A */
       break;
@@ -453,7 +454,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute dest/second term RGB
     */
-   switch (softpipe->blend->rt[0].rgb_dst_factor) {
+   switch (softpipe->blend->rt[cbuf].rgb_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* dest = dest * 1   NO-OP, leave dest as-is */
       break;
@@ -593,7 +594,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Compute dest/second term A
     */
-   switch (softpipe->blend->rt[0].alpha_dst_factor) {
+   switch (softpipe->blend->rt[cbuf].alpha_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
       /* dest = dest * 1   NO-OP, leave dest as-is */
       break;
@@ -656,7 +657,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Combine RGB terms
     */
-   switch (softpipe->blend->rt[0].rgb_func) {
+   switch (softpipe->blend->rt[cbuf].rgb_func) {
    case PIPE_BLEND_ADD:
       VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */
       VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */
@@ -689,7 +690,7 @@ blend_quad(struct quad_stage *qs,
    /*
     * Combine A terms
     */
-   switch (softpipe->blend->rt[0].alpha_func) {
+   switch (softpipe->blend->rt[cbuf].alpha_func) {
    case PIPE_BLEND_ADD:
       VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */
       break;
@@ -711,26 +712,24 @@ blend_quad(struct quad_stage *qs,
 }
 
 static void
-colormask_quad(struct quad_stage *qs,
+colormask_quad(unsigned colormask,
                float (*quadColor)[4],
                float (*dest)[4])
 {
-   struct softpipe_context *softpipe = qs->softpipe;
-
    /* R */
-   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_R))
+   if (!(colormask & PIPE_MASK_R))
       COPY_4V(quadColor[0], dest[0]);
 
    /* G */
-   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_G))
+   if (!(colormask & PIPE_MASK_G))
       COPY_4V(quadColor[1], dest[1]);
 
    /* B */
-   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_B))
+   if (!(colormask & PIPE_MASK_B))
       COPY_4V(quadColor[2], dest[2]);
 
    /* A */
-   if (!(softpipe->blend->rt[0].colormask & PIPE_MASK_A))
+   if (!(colormask & PIPE_MASK_A))
       COPY_4V(quadColor[3], dest[3]);
 }
 
@@ -773,12 +772,12 @@ blend_fallback(struct quad_stage *qs,
          if (blend->logicop_enable) {
             logicop_quad( qs, quadColor, dest );
          }
-         else if (blend->rt[0].blend_enable) {
-            blend_quad( qs, quadColor, dest );
+         else if (blend->rt[cbuf].blend_enable) {
+            blend_quad( qs, quadColor, dest, cbuf );
          }
 
-         if (blend->rt[0].colormask != 0xf)
-            colormask_quad( qs, quadColor, dest );
+         if (blend->rt[cbuf].colormask != 0xf)
+            colormask_quad( blend->rt[cbuf].colormask, quadColor, dest);
    
          /* Output color values
           */
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index bd3532de4f..e36f9ec5da 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -91,6 +91,10 @@ softpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_BLEND_EQUATION_SEPARATE:
       return 1;
+   case PIPE_CAP_INDEP_BLEND_ENABLE:
+      return 1;
+   case PIPE_CAP_INDEP_BLEND_FUNC:
+      return 1;
    default:
       return 0;
    }
-- 
cgit v1.2.3


From 72778a9d254f6c9c63d86413936ee7f3e5a6e56e Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Mon, 25 Jan 2010 10:52:19 -0800
Subject: Revert "r300g,radeong: finish and enable the immediate mode"

This reverts commit 112239e9a66a155d36fe2ad0ab130e6f26eff298.
---
 src/gallium/drivers/r300/r300_emit.c               |  22 ++++
 src/gallium/drivers/r300/r300_render.c             | 141 +++++----------------
 src/gallium/drivers/r300/r300_state.c              |  22 ----
 src/gallium/winsys/drm/radeon/core/radeon_buffer.c |  83 ++----------
 src/gallium/winsys/drm/radeon/core/radeon_buffer.h |   2 -
 src/gallium/winsys/drm/radeon/core/radeon_drm.h    |   2 +-
 src/gallium/winsys/drm/radeon/core/radeon_r300.c   |   8 +-
 src/gallium/winsys/drm/radeon/core/radeon_winsys.h |   6 -
 8 files changed, 67 insertions(+), 219 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index badbf3715c..36d2c64b58 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -772,6 +772,22 @@ void r300_emit_texture(struct r300_context* r300,
     END_CS;
 }
 
+static boolean r300_validate_aos(struct r300_context *r300)
+{
+    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
+    struct pipe_vertex_element *velem = r300->vertex_element;
+    int i;
+
+    /* Check if formats and strides are aligned to the size of DWORD. */
+    for (i = 0; i < r300->vertex_element_count; i++) {
+        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
+            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
+            return FALSE;
+        }
+    }
+    return TRUE;
+}
+
 void r300_emit_aos(struct r300_context* r300, unsigned offset)
 {
     struct pipe_vertex_buffer *vb1, *vb2, *vbuf = r300->vertex_buffer;
@@ -781,6 +797,12 @@ void r300_emit_aos(struct r300_context* r300, unsigned offset)
     unsigned packet_size = (aos_count * 3 + 1) / 2;
     CS_LOCALS(r300);
 
+    /* XXX Move this checking to a more approriate place. */
+    if (!r300_validate_aos(r300)) {
+        /* XXX We should fallback using Draw. */
+        assert(0);
+    }
+
     BEGIN_CS(2 + packet_size + aos_count * 2);
     OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size);
     OUT_CS(aos_count);
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 7f095bffe7..677031ef04 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -28,7 +28,6 @@
 
 #include "pipe/p_inlines.h"
 
-#include "util/u_format.h"
 #include "util/u_memory.h"
 #include "util/u_prim.h"
 
@@ -115,53 +114,20 @@ static uint32_t r300_provoking_vertex_fixes(struct r300_context *r300,
     return color_control;
 }
 
-
-static void r300_emit_draw_arrays_immediate(struct r300_context *r300,
-                                            unsigned mode,
-                                            unsigned start,
-                                            unsigned count)
+static void r300_emit_draw_immediate(struct r300_context *r300,
+                                     unsigned mode,
+                                     unsigned start,
+                                     unsigned count)
 {
-    struct pipe_vertex_element* velem;
-    struct pipe_vertex_buffer* vbuf;
-    unsigned vertex_element_count = r300->vertex_element_count;
-    unsigned i, v, vbi, dw, elem_offset;
-
-    /* Size of the vertex, in dwords. */
-    unsigned vertex_size = 0;
-
-    /* Offsets of the attribute, in dwords, from the start of the vertex. */
-    unsigned offset[PIPE_MAX_ATTRIBS];
-
-    /* Size of the vertex element, in dwords. */
-    unsigned size[PIPE_MAX_ATTRIBS];
-
-    /* Stride to the same attrib in the next vertex in the vertex buffer,
-     * in dwords. */
-    unsigned stride[PIPE_MAX_ATTRIBS];
-
-    /* Mapped vertex buffers. */
-    uint32_t* map[PIPE_MAX_ATTRIBS] = {0};
-
+    struct pipe_buffer* vbo = r300->vertex_buffer[0].buffer;
+    unsigned vertex_size = r300->vertex_buffer[0].stride / sizeof(float);
+    unsigned i;
+    uint32_t* map;
     CS_LOCALS(r300);
 
-    /* Calculate the vertex size, offsets, strides etc. and map the buffers. */
-    for (i = 0; i < vertex_element_count; i++) {
-        velem = &r300->vertex_element[i];
-        offset[i] = velem->src_offset >> 2;
-        size[i] = util_format_get_blocksize(velem->src_format) >> 2;
-        vertex_size += size[i];
-        vbi = velem->vertex_buffer_index;
-
-        /* Map the buffer. */
-        if (!map[vbi]) {
-            vbuf = &r300->vertex_buffer[vbi];
-            map[vbi] = (uint32_t*)pipe_buffer_map(r300->context.screen,
-                                                  vbuf->buffer,
-                                                  PIPE_BUFFER_USAGE_CPU_READ);
-            map[vbi] += vbuf->buffer_offset >> 2;
-            stride[vbi] = vbuf->stride >> 2;
-        }
-    }
+    map = (uint32_t*)pipe_buffer_map_range(r300->context.screen, vbo,
+            start * vertex_size, count * vertex_size,
+            PIPE_BUFFER_USAGE_CPU_READ);
 
     BEGIN_CS(10 + count * vertex_size);
     OUT_CS_REG(R300_GA_COLOR_CONTROL,
@@ -172,31 +138,18 @@ static void r300_emit_draw_arrays_immediate(struct r300_context *r300,
     OUT_CS_PKT3(R300_PACKET3_3D_DRAW_IMMD_2, count * vertex_size);
     OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED | (count << 16) |
             r300_translate_primitive(mode));
-
-    /* Emit vertices. */
-    for (v = 0; v < count; v++) {
-        for (i = 0; i < vertex_element_count; i++) {
-            velem = &r300->vertex_element[i];
-            vbi = velem->vertex_buffer_index;
-            elem_offset = offset[i] + stride[vbi] * (v + start);
-
-            for (dw = 0; dw < size[i]; dw++) {
-                OUT_CS(map[vbi][elem_offset + dw]);
-            }
+    //debug_printf("r300: Immd %d verts, %d attrs\n", count, vertex_size);
+    for (i = 0; i < count * vertex_size; i++) {
+        if (i % vertex_size == 0) {
+            //debug_printf("r300: -- vert --\n");
         }
+        //debug_printf("r300: 0x%08x\n", *map);
+        OUT_CS(*map);
+        map++;
     }
     END_CS;
 
-    /* Unmap buffers. */
-    for (i = 0; i < vertex_element_count; i++) {
-        vbi = r300->vertex_element[i].vertex_buffer_index;
-
-        if (map[vbi]) {
-            vbuf = &r300->vertex_buffer[vbi];
-            pipe_buffer_unmap(r300->context.screen, vbuf->buffer);
-            map[vbi] = 0;
-        }
-    }
+    pipe_buffer_unmap(r300->context.screen, vbo);
 }
 
 static void r300_emit_draw_arrays(struct r300_context *r300,
@@ -269,49 +222,16 @@ static void r300_emit_draw_elements(struct r300_context *r300,
 }
 
 
-static boolean r300_setup_local_vertex_buffers(struct r300_context *r300)
-{
-    struct pipe_vertex_buffer *vb;
-    boolean found_local_bo = FALSE, found_managed_bo = FALSE;
-    unsigned i;
-
-    /* See what buffers we got. */
-    for (i = 0; i < r300->vertex_element_count; i++) {
-        vb = &r300->vertex_buffer[r300->vertex_element[i].vertex_buffer_index];
-        if (r300->winsys->buffer_is_local(r300->winsys, vb->buffer)) {
-            found_local_bo = TRUE;
-        } else {
-            found_managed_bo = TRUE;
-        }
-    }
-
-    /* If we found both local and managed buffers, make local buffers managed
-     * because we shouldn't use the immediate mode in case a managed buffer is
-     * present, due to performance reasons. */
-    if (found_local_bo && found_managed_bo) {
-        for (i = 0; i < r300->vertex_element_count; i++) {
-            vb = &r300->vertex_buffer[r300->vertex_element[i].vertex_buffer_index];
-            if (r300->winsys->buffer_is_local(r300->winsys, vb->buffer)) {
-                r300->winsys->buffer_make_managed(r300->winsys, vb->buffer);
-            }
-        }
-    }
-
-    return !found_managed_bo;
-}
-
 static boolean r300_setup_vertex_buffers(struct r300_context *r300)
 {
     struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
     struct pipe_vertex_element *velem = r300->vertex_element;
-    struct pipe_buffer *pbuf;
 
 validate:
     for (int i = 0; i < r300->vertex_element_count; i++) {
-        pbuf = vbuf[velem[i].vertex_buffer_index].buffer;
-
-        if (!r300->winsys->add_buffer(r300->winsys, pbuf,
-                                      RADEON_GEM_DOMAIN_GTT, 0)) {
+        if (!r300->winsys->add_buffer(r300->winsys,
+                vbuf[velem[i].vertex_buffer_index].buffer,
+            RADEON_GEM_DOMAIN_GTT, 0)) {
             r300->context.flush(&r300->context, 0, NULL);
             goto validate;
         }
@@ -325,7 +245,6 @@ validate:
     return TRUE;
 }
 
-
 static void r300_shorten_ubyte_elts(struct r300_context* r300,
                                     struct pipe_buffer** elts,
                                     unsigned count)
@@ -446,15 +365,15 @@ void r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
 
     r300_emit_buffer_validate(r300);
 
-    if (r300_setup_local_vertex_buffers(r300)) {
-        r300_emit_dirty_state(r300);
-        r300_emit_draw_arrays_immediate(r300, mode, start, count);
-    } else {
-        if (!r300_setup_vertex_buffers(r300)) {
-            return;
-        }
+    if (!r300_setup_vertex_buffers(r300)) {
+        return;
+    }
+
+    r300_emit_dirty_state(r300);
 
-        r300_emit_dirty_state(r300);
+    if (FALSE && count <= 4 && r300->vertex_buffer_count == 1) {
+        r300_emit_draw_immediate(r300, mode, start, count);
+    } else {
         r300_emit_aos(r300, start);
         r300_emit_draw_arrays(r300, mode, count);
     }
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 641e95e7fc..e2ec0bc5bd 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -924,22 +924,6 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
     r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
 }
 
-static boolean r300_validate_aos(struct r300_context *r300)
-{
-    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
-    struct pipe_vertex_element *velem = r300->vertex_element;
-    int i;
-
-    /* Check if formats and strides are aligned to the size of DWORD. */
-    for (i = 0; i < r300->vertex_element_count; i++) {
-        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
-            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
-            return FALSE;
-        }
-    }
-    return TRUE;
-}
-
 static void r300_set_vertex_elements(struct pipe_context* pipe,
                                     unsigned count,
                                     const struct pipe_vertex_element* elements)
@@ -955,12 +939,6 @@ static void r300_set_vertex_elements(struct pipe_context* pipe,
         draw_flush(r300->draw);
         draw_set_vertex_elements(r300->draw, count, elements);
     }
-
-    if (!r300_validate_aos(r300)) {
-        /* XXX We should fallback using draw. */
-        assert(0);
-        abort();
-    }
 }
 
 static void* r300_create_vs_state(struct pipe_context* pipe,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
index 5214b6d8bc..25e1cdcdb6 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.c
@@ -51,23 +51,6 @@ static const char *radeon_get_name(struct pipe_winsys *ws)
     return "Radeon/GEM+KMS";
 }
 
-uint32_t radeon_domain_from_usage(unsigned usage)
-{
-    uint32_t domain = 0;
-
-    if (usage & PIPE_BUFFER_USAGE_PIXEL) {
-        domain |= RADEON_GEM_DOMAIN_VRAM;
-    }
-    if (usage & PIPE_BUFFER_USAGE_VERTEX) {
-        domain |= RADEON_GEM_DOMAIN_GTT;
-    }
-    if (usage & PIPE_BUFFER_USAGE_INDEX) {
-        domain |= RADEON_GEM_DOMAIN_GTT;
-    }
-
-    return domain;
-}
-
 static struct pipe_buffer *radeon_buffer_create(struct pipe_winsys *ws,
                                                 unsigned alignment,
                                                 unsigned usage,
@@ -88,17 +71,25 @@ static struct pipe_buffer *radeon_buffer_create(struct pipe_winsys *ws,
     radeon_buffer->base.usage = usage;
     radeon_buffer->base.size = size;
 
-    if ((usage == PIPE_BUFFER_USAGE_CONSTANT && is_r3xx(radeon_ws->pci_id)) ||
-        (usage == PIPE_BUFFER_USAGE_VERTEX && size < 512)) {
+    if (usage == PIPE_BUFFER_USAGE_CONSTANT && is_r3xx(radeon_ws->pci_id)) {
         /* Don't bother allocating a BO, as it'll never get to the card. */
-        /* Also, create small vertex buffers in RAM. */
         desc.alignment = alignment;
         desc.usage = usage;
         radeon_buffer->pb = pb_malloc_buffer_create(size, &desc);
         return &radeon_buffer->base;
     }
 
-    domain = radeon_domain_from_usage(usage);
+    domain = 0;
+
+    if (usage & PIPE_BUFFER_USAGE_PIXEL) {
+        domain |= RADEON_GEM_DOMAIN_VRAM;
+    }
+    if (usage & PIPE_BUFFER_USAGE_VERTEX) {
+        domain |= RADEON_GEM_DOMAIN_GTT;
+    }
+    if (usage & PIPE_BUFFER_USAGE_INDEX) {
+        domain |= RADEON_GEM_DOMAIN_GTT;
+    }
 
     radeon_buffer->bo = radeon_bo_open(radeon_ws->priv->bom, 0, size,
             alignment, domain, 0);
@@ -231,54 +222,6 @@ static void radeon_buffer_set_tiling(struct radeon_winsys *ws,
     radeon_bo_set_tiling(radeon_buffer->bo, flags, pitch);
 }
 
-static boolean radeon_buffer_is_local(struct radeon_winsys *ws,
-                                      struct pipe_buffer *buffer)
-{
-    struct radeon_pipe_buffer *radeon_buffer =
-        (struct radeon_pipe_buffer*)buffer;
-
-    return radeon_buffer->pb != NULL;
-}
-
-static void radeon_buffer_make_managed(struct radeon_winsys *ws,
-                                       struct pipe_buffer *buffer)
-{
-    struct radeon_pipe_buffer* radeon_buffer =
-        (struct radeon_pipe_buffer*)buffer;
-    uint32_t domain;
-    void *map;
-
-    if (radeon_buffer->pb) {
-        domain = radeon_domain_from_usage(buffer->usage);
-
-        /* Create a managed buffer. */
-        radeon_buffer->bo = radeon_bo_open(ws->priv->bom, 0,
-                                           buffer->size, buffer->alignment,
-                                           domain, 0);
-        if (radeon_buffer->bo == NULL) {
-            /* XXX What now? */
-            fprintf(stderr, "radeon: cannot create a buffer in function %s\n",
-                    __FUNCTION__);
-            assert(0);
-            abort();
-        }
-
-        /* Move data. */
-        radeon_bo_map(radeon_buffer->bo, 1);
-        map = pb_map(radeon_buffer->pb, PIPE_BUFFER_USAGE_CPU_READ);
-
-        memcpy(radeon_buffer->bo->ptr, map, buffer->size);
-
-        pb_unmap(radeon_buffer->pb);
-        radeon_bo_unmap(radeon_buffer->bo);
-
-        /* Release the locally-created buffer. */
-        pipe_reference_init(&radeon_buffer->pb->base.reference, 0);
-        pb_destroy(radeon_buffer->pb);
-        radeon_buffer->pb = 0;
-    }
-}
-
 static void radeon_fence_reference(struct pipe_winsys *ws,
                                    struct pipe_fence_handle **ptr,
                                    struct pipe_fence_handle *pfence)
@@ -382,8 +325,6 @@ struct radeon_winsys* radeon_pipe_winsys(int fd)
     radeon_ws->base.get_name = radeon_get_name;
 
     radeon_ws->buffer_set_tiling = radeon_buffer_set_tiling;
-    radeon_ws->buffer_is_local = radeon_buffer_is_local;
-    radeon_ws->buffer_make_managed = radeon_buffer_make_managed;
 
     return radeon_ws;
 }
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_buffer.h b/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
index c46abff793..de71cb2f42 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_buffer.h
@@ -77,8 +77,6 @@ struct radeon_winsys_priv {
     void *flush_data;
 };
 
-uint32_t radeon_domain_from_usage(unsigned usage);
-
 struct radeon_winsys* radeon_pipe_winsys(int fb);
 #if 0
 struct pipe_surface *radeon_surface_from_handle(struct radeon_context *radeon_context,
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.h b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
index 077388ee02..ddd7983824 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_drm.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.h
@@ -81,7 +81,7 @@ void radeon_destroy_drm_api(struct drm_api* api);
 /* Guess at whether this chipset should use r300g.
  *
  * I believe that this check is valid, but I haven't been exhaustive. */
-static INLINE boolean is_r3xx(int pciid)
+static boolean is_r3xx(int pciid)
 {
     return (pciid > 0x3150) && (pciid < 0x796f);
 }
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_r300.c b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
index d759beaba1..0253bc2527 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_r300.c
+++ b/src/gallium/winsys/drm/radeon/core/radeon_r300.c
@@ -81,13 +81,9 @@ static void radeon_write_cs_reloc(struct radeon_winsys* winsys,
                                   uint32_t flags)
 {
     int retval = 0;
-    struct radeon_pipe_buffer* radeon_buffer =
-        (struct radeon_pipe_buffer*)pbuffer;
 
-    assert(!radeon_buffer->pb);
-
-    retval = radeon_cs_write_reloc(winsys->priv->cs, radeon_buffer->bo,
-                                   rd, wd, flags);
+    retval = radeon_cs_write_reloc(winsys->priv->cs,
+            ((struct radeon_pipe_buffer*)pbuffer)->bo, rd, wd, flags);
 
     if (retval) {
         debug_printf("radeon: Relocation of %p (%d, %d, %d) failed!\n",
diff --git a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
index 462fba844e..864082b99b 100644
--- a/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
+++ b/src/gallium/winsys/drm/radeon/core/radeon_winsys.h
@@ -106,12 +106,6 @@ struct radeon_winsys {
                               uint32_t pitch,
                               boolean microtiled,
                               boolean macrotiled);
-
-    boolean (*buffer_is_local)(struct radeon_winsys* winsys,
-                               struct pipe_buffer* buffer);
-
-    void (*buffer_make_managed)(struct radeon_winsys* winsys,
-                                struct pipe_buffer* buffer);
 };
 
 #endif
-- 
cgit v1.2.3


From 5fe71949d9e3790ce38ab616aee29079c6b67721 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Mon, 25 Jan 2010 11:27:50 -0800
Subject: r300g: Finish and enable immediate mode.

Based on Marek's immd-mode patch.
---
 src/gallium/drivers/r300/r300_emit.c   |  22 -------
 src/gallium/drivers/r300/r300_render.c | 115 ++++++++++++++++++++++++---------
 src/gallium/drivers/r300/r300_state.c  |  22 +++++++
 3 files changed, 106 insertions(+), 53 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 36d2c64b58..badbf3715c 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -772,22 +772,6 @@ void r300_emit_texture(struct r300_context* r300,
     END_CS;
 }
 
-static boolean r300_validate_aos(struct r300_context *r300)
-{
-    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
-    struct pipe_vertex_element *velem = r300->vertex_element;
-    int i;
-
-    /* Check if formats and strides are aligned to the size of DWORD. */
-    for (i = 0; i < r300->vertex_element_count; i++) {
-        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
-            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
-            return FALSE;
-        }
-    }
-    return TRUE;
-}
-
 void r300_emit_aos(struct r300_context* r300, unsigned offset)
 {
     struct pipe_vertex_buffer *vb1, *vb2, *vbuf = r300->vertex_buffer;
@@ -797,12 +781,6 @@ void r300_emit_aos(struct r300_context* r300, unsigned offset)
     unsigned packet_size = (aos_count * 3 + 1) / 2;
     CS_LOCALS(r300);
 
-    /* XXX Move this checking to a more approriate place. */
-    if (!r300_validate_aos(r300)) {
-        /* XXX We should fallback using Draw. */
-        assert(0);
-    }
-
     BEGIN_CS(2 + packet_size + aos_count * 2);
     OUT_CS_PKT3(R300_PACKET3_3D_LOAD_VBPNTR, packet_size);
     OUT_CS(aos_count);
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 677031ef04..dcd0761944 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -28,6 +28,7 @@
 
 #include "pipe/p_inlines.h"
 
+#include "util/u_format.h"
 #include "util/u_memory.h"
 #include "util/u_prim.h"
 
@@ -114,20 +115,58 @@ static uint32_t r300_provoking_vertex_fixes(struct r300_context *r300,
     return color_control;
 }
 
-static void r300_emit_draw_immediate(struct r300_context *r300,
-                                     unsigned mode,
-                                     unsigned start,
-                                     unsigned count)
+static boolean immd_is_good_idea(struct r300_context *r300,
+                                      unsigned count)
 {
-    struct pipe_buffer* vbo = r300->vertex_buffer[0].buffer;
-    unsigned vertex_size = r300->vertex_buffer[0].stride / sizeof(float);
-    unsigned i;
-    uint32_t* map;
+    return count <= 4;
+}
+
+static void r300_emit_draw_arrays_immediate(struct r300_context *r300,
+                                            unsigned mode,
+                                            unsigned start,
+                                            unsigned count)
+{
+    struct pipe_vertex_element* velem;
+    struct pipe_vertex_buffer* vbuf;
+    unsigned vertex_element_count = r300->vertex_element_count;
+    unsigned i, v, vbi, dw, elem_offset;
+
+    /* Size of the vertex, in dwords. */
+    unsigned vertex_size = 0;
+
+    /* Offsets of the attribute, in dwords, from the start of the vertex. */
+    unsigned offset[PIPE_MAX_ATTRIBS];
+
+    /* Size of the vertex element, in dwords. */
+    unsigned size[PIPE_MAX_ATTRIBS];
+
+    /* Stride to the same attrib in the next vertex in the vertex buffer,
+     * in dwords. */
+    unsigned stride[PIPE_MAX_ATTRIBS];
+
+    /* Mapped vertex buffers. */
+    uint32_t* map[PIPE_MAX_ATTRIBS] = {0};
+
     CS_LOCALS(r300);
 
-    map = (uint32_t*)pipe_buffer_map_range(r300->context.screen, vbo,
-            start * vertex_size, count * vertex_size,
-            PIPE_BUFFER_USAGE_CPU_READ);
+    /* Calculate the vertex size, offsets, strides etc. and map the buffers. */
+    for (i = 0; i < vertex_element_count; i++) {
+        velem = &r300->vertex_element[i];
+        offset[i] = velem->src_offset / 4;
+        size[i] = util_format_get_blocksize(velem->src_format) / 4;
+        vertex_size += size[i];
+        vbi = velem->vertex_buffer_index;
+
+        /* Map the buffer. */
+        if (!map[vbi]) {
+            vbuf = &r300->vertex_buffer[vbi];
+            map[vbi] = (uint32_t*)pipe_buffer_map(r300->context.screen,
+                                                  vbuf->buffer,
+                                                  PIPE_BUFFER_USAGE_CPU_READ);
+            map[vbi] += vbuf->buffer_offset / 4;
+            stride[vbi] = vbuf->stride / 4;
+        }
+    }
 
     BEGIN_CS(10 + count * vertex_size);
     OUT_CS_REG(R300_GA_COLOR_CONTROL,
@@ -138,18 +177,31 @@ static void r300_emit_draw_immediate(struct r300_context *r300,
     OUT_CS_PKT3(R300_PACKET3_3D_DRAW_IMMD_2, count * vertex_size);
     OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED | (count << 16) |
             r300_translate_primitive(mode));
-    //debug_printf("r300: Immd %d verts, %d attrs\n", count, vertex_size);
-    for (i = 0; i < count * vertex_size; i++) {
-        if (i % vertex_size == 0) {
-            //debug_printf("r300: -- vert --\n");
+
+    /* Emit vertices. */
+    for (v = 0; v < count; v++) {
+        for (i = 0; i < vertex_element_count; i++) {
+            velem = &r300->vertex_element[i];
+            vbi = velem->vertex_buffer_index;
+            elem_offset = offset[i] + stride[vbi] * (v + start);
+
+            for (dw = 0; dw < size[i]; dw++) {
+                OUT_CS(map[vbi][elem_offset + dw]);
+            }
         }
-        //debug_printf("r300: 0x%08x\n", *map);
-        OUT_CS(*map);
-        map++;
     }
     END_CS;
 
-    pipe_buffer_unmap(r300->context.screen, vbo);
+    /* Unmap buffers. */
+    for (i = 0; i < vertex_element_count; i++) {
+        vbi = r300->vertex_element[i].vertex_buffer_index;
+
+        if (map[vbi]) {
+            vbuf = &r300->vertex_buffer[vbi];
+            pipe_buffer_unmap(r300->context.screen, vbuf->buffer);
+            map[vbi] = NULL;
+        }
+    }
 }
 
 static void r300_emit_draw_arrays(struct r300_context *r300,
@@ -221,17 +273,18 @@ static void r300_emit_draw_elements(struct r300_context *r300,
     END_CS;
 }
 
-
 static boolean r300_setup_vertex_buffers(struct r300_context *r300)
 {
     struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
     struct pipe_vertex_element *velem = r300->vertex_element;
+    struct pipe_buffer *pbuf;
 
 validate:
     for (int i = 0; i < r300->vertex_element_count; i++) {
-        if (!r300->winsys->add_buffer(r300->winsys,
-                vbuf[velem[i].vertex_buffer_index].buffer,
-            RADEON_GEM_DOMAIN_GTT, 0)) {
+        pbuf = vbuf[velem[i].vertex_buffer_index].buffer;
+
+        if (!r300->winsys->add_buffer(r300->winsys, pbuf,
+                                      RADEON_GEM_DOMAIN_GTT, 0)) {
             r300->context.flush(&r300->context, 0, NULL);
             goto validate;
         }
@@ -365,15 +418,15 @@ void r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
 
     r300_emit_buffer_validate(r300);
 
-    if (!r300_setup_vertex_buffers(r300)) {
-        return;
-    }
-
-    r300_emit_dirty_state(r300);
-
-    if (FALSE && count <= 4 && r300->vertex_buffer_count == 1) {
-        r300_emit_draw_immediate(r300, mode, start, count);
+    if (immd_is_good_idea(r300, count)) {
+        r300_emit_dirty_state(r300);
+        r300_emit_draw_arrays_immediate(r300, mode, start, count);
     } else {
+        if (!r300_setup_vertex_buffers(r300)) {
+            return;
+        }
+
+        r300_emit_dirty_state(r300);
         r300_emit_aos(r300, start);
         r300_emit_draw_arrays(r300, mode, count);
     }
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index e2ec0bc5bd..641e95e7fc 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -924,6 +924,22 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
     r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
 }
 
+static boolean r300_validate_aos(struct r300_context *r300)
+{
+    struct pipe_vertex_buffer *vbuf = r300->vertex_buffer;
+    struct pipe_vertex_element *velem = r300->vertex_element;
+    int i;
+
+    /* Check if formats and strides are aligned to the size of DWORD. */
+    for (i = 0; i < r300->vertex_element_count; i++) {
+        if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 ||
+            util_format_get_blocksize(velem[i].src_format) % 4 != 0) {
+            return FALSE;
+        }
+    }
+    return TRUE;
+}
+
 static void r300_set_vertex_elements(struct pipe_context* pipe,
                                     unsigned count,
                                     const struct pipe_vertex_element* elements)
@@ -939,6 +955,12 @@ static void r300_set_vertex_elements(struct pipe_context* pipe,
         draw_flush(r300->draw);
         draw_set_vertex_elements(r300->draw, count, elements);
     }
+
+    if (!r300_validate_aos(r300)) {
+        /* XXX We should fallback using draw. */
+        assert(0);
+        abort();
+    }
 }
 
 static void* r300_create_vs_state(struct pipe_context* pipe,
-- 
cgit v1.2.3


From 786bd78be4a412e58ded6eed1d6d3e4a61b1ba2b Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 26 Jan 2010 07:21:40 -0800
Subject: r300g: Per-MRT blending is not supported.

---
 src/gallium/drivers/r300/r300_screen.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 67325c6b80..ebe8793486 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -149,6 +149,9 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
             } else {
                 return 0;
             }
+        case PIPE_CAP_INDEP_BLEND_ENABLE:
+        case PIPE_CAP_INDEP_BLEND_FUNC:
+            return 0;
         default:
             debug_printf("r300: Implementation error: Bad param %d\n",
                 param);
-- 
cgit v1.2.3


From 1a9d01dcde26aee4db662deb7069fa14069062de Mon Sep 17 00:00:00 2001
From: Roland Scheidegger <sroland@vmware.com>
Date: Tue, 26 Jan 2010 17:32:57 +0100
Subject: gallium: add the new PIPE_CAP blend bits to more drivers

some drivers will complain if they don't know the cap bit
---
 src/gallium/drivers/llvmpipe/lp_screen.c | 4 ++++
 src/gallium/drivers/nv04/nv04_screen.c   | 4 ++++
 src/gallium/drivers/nv10/nv10_screen.c   | 4 ++++
 src/gallium/drivers/nv20/nv20_screen.c   | 4 ++++
 src/gallium/drivers/nv30/nv30_screen.c   | 4 ++++
 src/gallium/drivers/nv40/nv40_screen.c   | 4 ++++
 src/gallium/drivers/nv50/nv50_screen.c   | 4 ++++
 7 files changed, 28 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index 9b47415f00..f04c0320ba 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -110,6 +110,10 @@ llvmpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_BLEND_EQUATION_SEPARATE:
       return 1;
+   case PIPE_CAP_INDEP_BLEND_ENABLE:
+      return 0;
+   case PIPE_CAP_INDEP_BLEND_FUNC:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c
index 7c5b6e8229..4eba772575 100644
--- a/src/gallium/drivers/nv04/nv04_screen.c
+++ b/src/gallium/drivers/nv04/nv04_screen.c
@@ -45,6 +45,10 @@ nv04_screen_get_param(struct pipe_screen *screen, int param)
 	case NOUVEAU_CAP_HW_VTXBUF:
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 0;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c
index 69a6dab866..2dabc4509f 100644
--- a/src/gallium/drivers/nv10/nv10_screen.c
+++ b/src/gallium/drivers/nv10/nv10_screen.c
@@ -40,6 +40,10 @@ nv10_screen_get_param(struct pipe_screen *screen, int param)
 	case NOUVEAU_CAP_HW_VTXBUF:
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 0;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
index d091335063..7782cbe3a8 100644
--- a/src/gallium/drivers/nv20/nv20_screen.c
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -40,6 +40,10 @@ nv20_screen_get_param(struct pipe_screen *screen, int param)
 	case NOUVEAU_CAP_HW_VTXBUF:
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 0;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
index 9ed48178dc..8f8f3ea1a9 100644
--- a/src/gallium/drivers/nv30/nv30_screen.c
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -67,6 +67,10 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param)
 	case NOUVEAU_CAP_HW_VTXBUF:
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 1;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c
index 9e55e5a089..4eab5daa80 100644
--- a/src/gallium/drivers/nv40/nv40_screen.c
+++ b/src/gallium/drivers/nv40/nv40_screen.c
@@ -52,6 +52,10 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param)
 		if (screen->curie->grclass == NV40TCL)
 			return 1;
 		return 0;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index 9d58f3c965..3992469383 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -135,6 +135,10 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 1;
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 0;
+	case PIPE_CAP_INDEP_BLEND_ENABLE:
+		return 0;
+	case PIPE_CAP_INDEP_BLEND_FUNC:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
-- 
cgit v1.2.3


From a5d67dc0119035c518d9778854948d13f0ce2d69 Mon Sep 17 00:00:00 2001
From: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Date: Tue, 26 Jan 2010 17:58:50 +0100
Subject: nv50: supports independent blend enables

---
 src/gallium/drivers/nv50/nv50_screen.c |  2 +-
 src/gallium/drivers/nv50/nv50_state.c  | 56 ++++++++++++++++++++++++----------
 2 files changed, 41 insertions(+), 17 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index 3992469383..6af9700c1a 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -136,7 +136,7 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param)
 	case NOUVEAU_CAP_HW_IDXBUF:
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_ENABLE:
-		return 0;
+		return 1;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
 	default:
diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c
index 815dfa4f6f..da8dc933b0 100644
--- a/src/gallium/drivers/nv50/nv50_state.c
+++ b/src/gallium/drivers/nv50/nv50_state.c
@@ -31,6 +31,23 @@
 
 #include "nouveau/nouveau_stateobj.h"
 
+static INLINE uint32_t
+nv50_colormask(unsigned mask)
+{
+	uint32_t cmask = 0;
+
+	if (mask & PIPE_MASK_R)
+		cmask |= 0x0001;
+	if (mask & PIPE_MASK_G)
+		cmask |= 0x0010;
+	if (mask & PIPE_MASK_B)
+		cmask |= 0x0100;
+	if (mask & PIPE_MASK_A)
+		cmask |= 0x1000;
+
+	return cmask;
+}
+
 static void *
 nv50_blend_state_create(struct pipe_context *pipe,
 			const struct pipe_blend_state *cso)
@@ -38,20 +55,29 @@ nv50_blend_state_create(struct pipe_context *pipe,
 	struct nouveau_stateobj *so = so_new(5, 24, 0);
 	struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla;
 	struct nv50_blend_stateobj *bso = CALLOC_STRUCT(nv50_blend_stateobj);
-	unsigned cmask = 0, i;
+	unsigned i, blend_enabled = 0;
 
 	/*XXX ignored:
 	 * 	- dither
 	 */
 
-	if (cso->rt[0].blend_enable == 0) {
-		so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
+	so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
+	if (cso->independent_blend_enable) {
+		for (i = 0; i < 8; ++i) {
+			so_data(so, cso->rt[i].blend_enable);
+			if (cso->rt[i].blend_enable)
+				blend_enabled = 1;
+		}
+	} else
+	if (cso->rt[0].blend_enable) {
+		blend_enabled = 1;
 		for (i = 0; i < 8; i++)
-			so_data(so, 0);
+			so_data(so, 1);
 	} else {
-		so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
 		for (i = 0; i < 8; i++)
-			so_data(so, 1);
+			so_data(so, 0);
+	}
+	if (blend_enabled) {
 		so_method(so, tesla, NV50TCL_BLEND_EQUATION_RGB, 5);
 		so_data  (so, nvgl_blend_eqn(cso->rt[0].rgb_func));
 		so_data  (so, 0x4000 | nvgl_blend_func(cso->rt[0].rgb_src_factor));
@@ -71,17 +97,15 @@ nv50_blend_state_create(struct pipe_context *pipe,
 		so_data  (so, nvgl_logicop_func(cso->logicop_func));
 	}
 
-	if (cso->rt[0].colormask & PIPE_MASK_R)
-		cmask |= (1 << 0);
-	if (cso->rt[0].colormask & PIPE_MASK_G)
-		cmask |= (1 << 4);
-	if (cso->rt[0].colormask & PIPE_MASK_B)
-		cmask |= (1 << 8);
-	if (cso->rt[0].colormask & PIPE_MASK_A)
-		cmask |= (1 << 12);
 	so_method(so, tesla, NV50TCL_COLOR_MASK(0), 8);
-	for (i = 0; i < 8; i++)
-		so_data(so, cmask);
+	if (cso->independent_blend_enable)
+		for (i = 0; i < 8; ++i)
+			so_data(so, nv50_colormask(cso->rt[i].colormask));
+	else {
+		uint32_t cmask = nv50_colormask(cso->rt[0].colormask);
+		for (i = 0; i < 8; i++)
+			so_data(so, cmask);
+	}
 
 	bso->pipe = *cso;
 	so_ref(so, &bso->so);
-- 
cgit v1.2.3


From 6ab9740b7a4f2be55533b5229519ae9563bd88ee Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Tue, 26 Jan 2010 16:45:32 -0800
Subject: softpipe: Remove unused variables.

---
 src/gallium/drivers/softpipe/sp_tex_sample.c | 3 ---
 1 file changed, 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c
index 1ae8fecacf..473ec3e150 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.c
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.c
@@ -517,7 +517,6 @@ compute_lambda_1d(const struct sp_sampler_varient *samp,
                   const float p[QUAD_SIZE])
 {
    const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
    float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
    float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
    float rho = MAX2(dsdx, dsdy) * texture->width0;
@@ -533,7 +532,6 @@ compute_lambda_2d(const struct sp_sampler_varient *samp,
                   const float p[QUAD_SIZE])
 {
    const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
    float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
    float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
    float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
@@ -553,7 +551,6 @@ compute_lambda_3d(const struct sp_sampler_varient *samp,
                   const float p[QUAD_SIZE])
 {
    const struct pipe_texture *texture = samp->texture;
-   const struct pipe_sampler_state *sampler = samp->sampler;
    float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]);
    float dsdy = fabsf(s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT]);
    float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]);
-- 
cgit v1.2.3


From 25b8ce3a6aa67db96550946aab94bd3416e5b10a Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Tue, 26 Jan 2010 17:00:15 -0800
Subject: i965g: Silence uninitialized variable warning.

---
 src/gallium/drivers/i965/brw_screen.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i965/brw_screen.c b/src/gallium/drivers/i965/brw_screen.c
index 0ecacac9a3..a3728fb16e 100644
--- a/src/gallium/drivers/i965/brw_screen.c
+++ b/src/gallium/drivers/i965/brw_screen.c
@@ -138,6 +138,9 @@ brw_get_name(struct pipe_screen *screen)
    case PCI_CHIP_ILM_G:
       chipset = "ILM_G";
       break;
+   default:
+      chipset = "unknown";
+      break;
    }
 
    util_snprintf(buffer, sizeof(buffer), "i965 (chipset: %s)", chipset);
-- 
cgit v1.2.3


From 15e283669aa8832347b4733f9556efdb5692640e Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 26 Jan 2010 20:11:39 -0800
Subject: r300g: Add A8_UNORM texture format.

Keeps teeworld happy.
---
 src/gallium/drivers/r300/r300_screen.c  | 1 +
 src/gallium/drivers/r300/r300_texture.h | 1 +
 2 files changed, 2 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 67325c6b80..a4a0e52815 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -219,6 +219,7 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
         case PIPE_FORMAT_DXT5_RGBA:
         case PIPE_FORMAT_YCBCR:
         case PIPE_FORMAT_L8_UNORM:
+        case PIPE_FORMAT_A8_UNORM:
         case PIPE_FORMAT_A8L8_UNORM:
             retval = usage & PIPE_TEXTURE_USAGE_SAMPLER;
             break;
diff --git a/src/gallium/drivers/r300/r300_texture.h b/src/gallium/drivers/r300/r300_texture.h
index 1be1e6843c..453fb1accc 100644
--- a/src/gallium/drivers/r300/r300_texture.h
+++ b/src/gallium/drivers/r300/r300_texture.h
@@ -42,6 +42,7 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format)
 {
     switch (format) {
         /* X8 */
+        case PIPE_FORMAT_A8_UNORM:
         case PIPE_FORMAT_I8_UNORM:
             return R300_EASY_TX_FORMAT(X, X, X, X, X8);
         case PIPE_FORMAT_L8_UNORM:
-- 
cgit v1.2.3


From e62520de70731b39bb34411372eb3d30638cf5a6 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 26 Jan 2010 21:41:44 -0800
Subject: r300g: Kill dead code for hashing custom state.

It just wasn't meant to be.
---
 src/gallium/drivers/r300/r300_context.c       | 16 ---------
 src/gallium/drivers/r300/r300_context.h       |  3 --
 src/gallium/drivers/r300/r300_state_derived.c | 52 ---------------------------
 3 files changed, 71 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 3bf9c8b18d..5a689e0bd7 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -40,25 +40,12 @@
 #include "r300_texture.h"
 #include "r300_winsys.h"
 
-static enum pipe_error r300_clear_hash_table(void* key, void* value,
-                                             void* data)
-{
-    FREE(key);
-    FREE(value);
-    return PIPE_OK;
-}
-
 static void r300_destroy_context(struct pipe_context* context)
 {
     struct r300_context* r300 = r300_context(context);
     struct r300_query* query, * temp;
 
     util_blitter_destroy(r300->blitter);
-
-    util_hash_table_foreach(r300->shader_hash_table, r300_clear_hash_table,
-        NULL);
-    util_hash_table_destroy(r300->shader_hash_table);
-
     draw_destroy(r300->draw);
 
     /* Free the OQ BO. */
@@ -183,9 +170,6 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     r300->context.is_texture_referenced = r300_is_texture_referenced;
     r300->context.is_buffer_referenced = r300_is_buffer_referenced;
 
-    r300->shader_hash_table = util_hash_table_create(r300_shader_key_hash,
-        r300_shader_key_compare);
-
     r300_setup_atoms(r300);
 
     r300->blend_color_state.state = CALLOC_STRUCT(r300_blend_color_state);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index 48c86fdad7..a25e46afaa 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -272,9 +272,6 @@ struct r300_context {
     struct r300_query *query_current;
     struct r300_query query_list;
 
-    /* Shader hash table. Used to store vertex formatting information, which
-     * depends on the combination of both currently loaded shaders. */
-    struct util_hash_table* shader_hash_table;
     /* Vertex formatting information. */
     struct r300_vertex_info* vertex_info;
 
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 99c2720897..45a3e93df1 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -37,32 +37,6 @@
 /* r300_state_derived: Various bits of state which are dependent upon
  * currently bound CSO data. */
 
-struct r300_shader_key {
-    struct r300_vertex_shader* vs;
-    struct r300_fragment_shader* fs;
-};
-
-struct r300_shader_derived_value {
-    struct r300_vertex_format* vformat;
-    struct r300_rs_block* rs_block;
-};
-
-unsigned r300_shader_key_hash(void* key) {
-    struct r300_shader_key* shader_key = (struct r300_shader_key*)key;
-    unsigned vs = (intptr_t)shader_key->vs;
-    unsigned fs = (intptr_t)shader_key->fs;
-
-    return (vs << 16) | (fs & 0xffff);
-}
-
-int r300_shader_key_compare(void* key1, void* key2) {
-    struct r300_shader_key* shader_key1 = (struct r300_shader_key*)key1;
-    struct r300_shader_key* shader_key2 = (struct r300_shader_key*)key2;
-
-    return (shader_key1->vs == shader_key2->vs) &&
-        (shader_key1->fs == shader_key2->fs);
-}
-
 static void r300_draw_emit_attrib(struct r300_context* r300,
                                   enum attrib_emit emit,
                                   enum interp_mode interp,
@@ -440,32 +414,6 @@ static void r300_update_derived_shader_state(struct r300_context* r300)
 {
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
 
-    /*
-    struct r300_shader_key* key;
-    struct r300_shader_derived_value* value;
-    key = CALLOC_STRUCT(r300_shader_key);
-    key->vs = r300->vs;
-    key->fs = r300->fs;
-
-    value = (struct r300_shader_derived_value*)
-        util_hash_table_get(r300->shader_hash_table, (void*)key);
-    if (value) {
-        //vformat = value->vformat;
-        rs_block = value->rs_block;
-
-        FREE(key);
-    } else {
-        rs_block = CALLOC_STRUCT(r300_rs_block);
-        value = CALLOC_STRUCT(r300_shader_derived_value);
-
-        r300_update_rs_block(r300, rs_block);
-
-        //value->vformat = vformat;
-        value->rs_block = rs_block;
-        util_hash_table_set(r300->shader_hash_table,
-            (void*)key, (void*)value);
-    } */
-
     /* Reset structures */
     memset(r300->rs_block, 0, sizeof(struct r300_rs_block));
     memset(r300->vertex_info, 0, sizeof(struct r300_vertex_info));
-- 
cgit v1.2.3


From 60aa7f1f8042a9291339a3b337a2f1adacf12841 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 26 Jan 2010 21:49:02 -0800
Subject: r300g: Make vertex_format into an atom.

Some delicious hax here.
---
 src/gallium/drivers/r300/r300_context.c       |  5 +++--
 src/gallium/drivers/r300/r300_context.h       |  3 +--
 src/gallium/drivers/r300/r300_emit.c          | 30 ++++++++++++---------------
 src/gallium/drivers/r300/r300_emit.h          |  2 +-
 src/gallium/drivers/r300/r300_render.c        |  2 +-
 src/gallium/drivers/r300/r300_state.c         |  9 ++++----
 src/gallium/drivers/r300/r300_state_derived.c | 25 +++++++++++++++-------
 7 files changed, 41 insertions(+), 35 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 5a689e0bd7..1460778ece 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -61,7 +61,7 @@ static void r300_destroy_context(struct pipe_context* context)
     FREE(r300->clip_state.state);
     FREE(r300->rs_block);
     FREE(r300->scissor_state.state);
-    FREE(r300->vertex_info);
+    FREE(r300->vertex_format_state.state);
     FREE(r300->viewport_state.state);
     FREE(r300->ztop_state.state);
     FREE(r300);
@@ -125,6 +125,7 @@ static void r300_setup_atoms(struct r300_context* r300)
     R300_INIT_ATOM(rs, 25);
     R300_INIT_ATOM(scissor, 3);
     R300_INIT_ATOM(viewport, 9);
+    R300_INIT_ATOM(vertex_format, 26);
 }
 
 struct pipe_context* r300_create_context(struct pipe_screen* screen,
@@ -176,7 +177,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     r300->clip_state.state = CALLOC_STRUCT(pipe_clip_state);
     r300->rs_block = CALLOC_STRUCT(r300_rs_block);
     r300->scissor_state.state = CALLOC_STRUCT(pipe_scissor_state);
-    r300->vertex_info = CALLOC_STRUCT(r300_vertex_info);
+    r300->vertex_format_state.state = CALLOC_STRUCT(r300_vertex_info);
     r300->viewport_state.state = CALLOC_STRUCT(r300_viewport_state);
     r300->ztop_state.state = CALLOC_STRUCT(r300_ztop_state);
 
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index a25e46afaa..407891e175 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -147,7 +147,6 @@ struct r300_ztop_state {
 #define R300_ANY_NEW_SAMPLERS    0x0001fe00
 #define R300_NEW_TEXTURE         0x00040000
 #define R300_ANY_NEW_TEXTURES    0x03fc0000
-#define R300_NEW_VERTEX_FORMAT   0x04000000
 #define R300_NEW_VERTEX_SHADER   0x08000000
 #define R300_NEW_VERTEX_SHADER_CONSTANTS    0x10000000
 #define R300_NEW_QUERY           0x40000000
@@ -273,7 +272,7 @@ struct r300_context {
     struct r300_query query_list;
 
     /* Vertex formatting information. */
-    struct r300_vertex_info* vertex_info;
+    struct r300_atom vertex_format_state;
 
     /* Various CSO state objects. */
     /* Beginning of atom list. */
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index badbf3715c..8081f1d956 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -812,38 +812,39 @@ void r300_emit_aos(struct r300_context* r300, unsigned offset)
     END_CS;
 }
 
-void r300_emit_vertex_format_state(struct r300_context* r300)
+void r300_emit_vertex_format_state(struct r300_context* r300, void* state)
 {
-    int i;
+    struct r300_vertex_info* vertex_info = (struct r300_vertex_info*)state;
+    unsigned i;
     CS_LOCALS(r300);
 
     DBG(r300, DBG_DRAW, "r300: VAP/PSC emit:\n");
 
     BEGIN_CS(26);
-    OUT_CS_REG(R300_VAP_VTX_SIZE, r300->vertex_info->vinfo.size);
+    OUT_CS_REG(R300_VAP_VTX_SIZE, vertex_info->vinfo.size);
 
     OUT_CS_REG_SEQ(R300_VAP_VTX_STATE_CNTL, 2);
-    OUT_CS(r300->vertex_info->vinfo.hwfmt[0]);
-    OUT_CS(r300->vertex_info->vinfo.hwfmt[1]);
+    OUT_CS(vertex_info->vinfo.hwfmt[0]);
+    OUT_CS(vertex_info->vinfo.hwfmt[1]);
     OUT_CS_REG_SEQ(R300_VAP_OUTPUT_VTX_FMT_0, 2);
-    OUT_CS(r300->vertex_info->vinfo.hwfmt[2]);
-    OUT_CS(r300->vertex_info->vinfo.hwfmt[3]);
+    OUT_CS(vertex_info->vinfo.hwfmt[2]);
+    OUT_CS(vertex_info->vinfo.hwfmt[3]);
     for (i = 0; i < 4; i++) {
        DBG(r300, DBG_DRAW, "    : hwfmt%d: 0x%08x\n", i,
-               r300->vertex_info->vinfo.hwfmt[i]);
+               vertex_info->vinfo.hwfmt[i]);
     }
 
     OUT_CS_REG_SEQ(R300_VAP_PROG_STREAM_CNTL_0, 8);
     for (i = 0; i < 8; i++) {
-        OUT_CS(r300->vertex_info->vap_prog_stream_cntl[i]);
+        OUT_CS(vertex_info->vap_prog_stream_cntl[i]);
         DBG(r300, DBG_DRAW, "    : prog_stream_cntl%d: 0x%08x\n", i,
-               r300->vertex_info->vap_prog_stream_cntl[i]);
+               vertex_info->vap_prog_stream_cntl[i]);
     }
     OUT_CS_REG_SEQ(R300_VAP_PROG_STREAM_CNTL_EXT_0, 8);
     for (i = 0; i < 8; i++) {
-        OUT_CS(r300->vertex_info->vap_prog_stream_cntl_ext[i]);
+        OUT_CS(vertex_info->vap_prog_stream_cntl_ext[i]);
         DBG(r300, DBG_DRAW, "    : prog_stream_cntl_ext%d: 0x%08x\n", i,
-               r300->vertex_info->vap_prog_stream_cntl_ext[i]);
+               vertex_info->vap_prog_stream_cntl_ext[i]);
     }
     END_CS;
 }
@@ -1171,11 +1172,6 @@ void r300_emit_dirty_state(struct r300_context* r300)
         r300_flush_textures(r300);
     }
 
-    if (r300->dirty_state & R300_NEW_VERTEX_FORMAT) {
-        r300_emit_vertex_format_state(r300);
-        r300->dirty_state &= ~R300_NEW_VERTEX_FORMAT;
-    }
-
     if (r300->dirty_state & (R300_NEW_VERTEX_SHADER | R300_NEW_VERTEX_SHADER_CONSTANTS)) {
         r300_flush_pvs(r300);
     }
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index 2f3d013f5e..6788e3d655 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -73,7 +73,7 @@ void r300_emit_texture(struct r300_context* r300,
 
 void r300_emit_vertex_buffer(struct r300_context* r300);
 
-void r300_emit_vertex_format_state(struct r300_context* r300);
+void r300_emit_vertex_format_state(struct r300_context* r300, void* state);
 
 void r300_emit_vertex_program_code(struct r300_context* r300,
                                    struct r300_vertex_program_code* code);
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index dcd0761944..354f3bb921 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -556,7 +556,7 @@ r300_render_get_vertex_info(struct vbuf_render* render)
 
     r300_update_derived_state(r300);
 
-    return &r300->vertex_info->vinfo;
+    return (struct vertex_info*)r300->vertex_format_state.state;
 }
 
 static boolean r300_render_allocate_vertices(struct vbuf_render* render,
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 641e95e7fc..7068a5a4ec 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -552,7 +552,7 @@ static void r300_bind_fs_state(struct pipe_context* pipe, void* shader)
     r300_pick_fragment_shader(r300);
 
     if (r300->vs && r300_vertex_shader_setup_wpos(r300)) {
-        r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
+        r300->vertex_format_state.dirty = TRUE;
     }
 
     r300->dirty_state |= R300_NEW_FRAGMENT_SHADER | R300_NEW_FRAGMENT_SHADER_CONSTANTS;
@@ -921,7 +921,7 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
         draw_set_vertex_buffers(r300->draw, count, buffers);
     }
 
-    r300->dirty_state |= R300_NEW_VERTEX_FORMAT;
+    r300->vertex_format_state.dirty = TRUE;
 }
 
 static boolean r300_validate_aos(struct r300_context *r300)
@@ -1001,9 +1001,10 @@ static void r300_bind_vs_state(struct pipe_context* pipe, void* shader)
             r300_vertex_shader_setup_wpos(r300);
         }
 
+        r300->vertex_format_state.dirty = TRUE;
+
         r300->dirty_state |=
-            R300_NEW_VERTEX_SHADER | R300_NEW_VERTEX_SHADER_CONSTANTS |
-            R300_NEW_VERTEX_FORMAT;
+            R300_NEW_VERTEX_SHADER | R300_NEW_VERTEX_SHADER_CONSTANTS;
     } else {
         draw_flush(r300->draw);
         draw_bind_vertex_shader(r300->draw,
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 45a3e93df1..1a1eabe5fb 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -48,7 +48,9 @@ static void r300_draw_emit_attrib(struct r300_context* r300,
     output = draw_find_shader_output(r300->draw,
                                      info->output_semantic_name[index],
                                      info->output_semantic_index[index]);
-    draw_emit_vertex_attr(&r300->vertex_info->vinfo, emit, interp, output);
+    draw_emit_vertex_attr(
+        (struct vertex_info*)r300->vertex_format_state.state,
+        emit, interp, output);
 }
 
 static void r300_draw_emit_all_attribs(struct r300_context* r300)
@@ -104,7 +106,8 @@ static void r300_draw_emit_all_attribs(struct r300_context* r300)
 /* Update the PSC tables. */
 static void r300_vertex_psc(struct r300_context* r300)
 {
-    struct r300_vertex_info *vformat = r300->vertex_info;
+    struct r300_vertex_info *vformat =
+        (struct r300_vertex_info*)r300->vertex_format_state.state;
     uint16_t type, swizzle;
     enum pipe_format format;
     unsigned i;
@@ -156,7 +159,8 @@ static void r300_vertex_psc(struct r300_context* r300)
 /* Update the PSC tables for SW TCL, using Draw. */
 static void r300_swtcl_vertex_psc(struct r300_context* r300)
 {
-    struct r300_vertex_info *vformat = r300->vertex_info;
+    struct r300_vertex_info *vformat =
+        (struct r300_vertex_info*)r300->vertex_format_state.state;
     struct vertex_info* vinfo = &vformat->vinfo;
     uint16_t type, swizzle;
     enum pipe_format format;
@@ -413,11 +417,15 @@ static void r300_update_rs_block(struct r300_context* r300,
 static void r300_update_derived_shader_state(struct r300_context* r300)
 {
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
+    struct r300_vertex_info *vformat =
+        (struct r300_vertex_info*)r300->vertex_format_state.state;
+    struct vertex_info* vinfo = &vformat->vinfo;
 
     /* Reset structures */
     memset(r300->rs_block, 0, sizeof(struct r300_rs_block));
-    memset(r300->vertex_info, 0, sizeof(struct r300_vertex_info));
-    memcpy(r300->vertex_info->vinfo.hwfmt, r300->vs->hwfmt, sizeof(uint)*4);
+    /* Mmm, delicious hax */
+    memset(r300->vertex_format_state.state, 0, sizeof(struct r300_vertex_info));
+    memcpy(vinfo->hwfmt, r300->vs->hwfmt, sizeof(uint)*4);
 
     r300_update_rs_block(r300, &r300->vs->outputs, &r300->fs->inputs);
 
@@ -425,7 +433,8 @@ static void r300_update_derived_shader_state(struct r300_context* r300)
         r300_vertex_psc(r300);
     } else {
         r300_draw_emit_all_attribs(r300);
-        draw_compute_vertex_size(&r300->vertex_info->vinfo);
+        draw_compute_vertex_size(
+            (struct vertex_info*)r300->vertex_format_state.state);
         r300_swtcl_vertex_psc(r300);
     }
 
@@ -507,8 +516,8 @@ void r300_update_derived_state(struct r300_context* r300)
 {
     /* XXX */
     if (r300->dirty_state &
-        (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER |
-         R300_NEW_VERTEX_FORMAT) || r300->rs_state.dirty) {
+        (R300_NEW_FRAGMENT_SHADER | R300_NEW_VERTEX_SHADER) ||
+        r300->vertex_format_state.dirty || r300->rs_state.dirty) {
         r300_update_derived_shader_state(r300);
     }
 
-- 
cgit v1.2.3


From fc20efe8e511bb9ec15c3d70e28b348ddaa4ad37 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Tue, 26 Jan 2010 22:08:11 -0800
Subject: r300g: Turn the RS block into an atom.

At least one extraneous dirty was eliminated, as well as the chance for
avoiding dirty on shader change.
---
 src/gallium/drivers/r300/r300_context.c       |  5 ++--
 src/gallium/drivers/r300/r300_context.h       |  3 +--
 src/gallium/drivers/r300/r300_emit.c          | 11 +++-----
 src/gallium/drivers/r300/r300_emit.h          |  3 +--
 src/gallium/drivers/r300/r300_state.c         |  1 -
 src/gallium/drivers/r300/r300_state_derived.c | 36 ++++++++++++++-------------
 6 files changed, 27 insertions(+), 32 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 1460778ece..052cc23a77 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -59,7 +59,7 @@ static void r300_destroy_context(struct pipe_context* context)
 
     FREE(r300->blend_color_state.state);
     FREE(r300->clip_state.state);
-    FREE(r300->rs_block);
+    FREE(r300->rs_block_state.state);
     FREE(r300->scissor_state.state);
     FREE(r300->vertex_format_state.state);
     FREE(r300->viewport_state.state);
@@ -125,6 +125,7 @@ static void r300_setup_atoms(struct r300_context* r300)
     R300_INIT_ATOM(rs, 25);
     R300_INIT_ATOM(scissor, 3);
     R300_INIT_ATOM(viewport, 9);
+    R300_INIT_ATOM(rs_block, 21);
     R300_INIT_ATOM(vertex_format, 26);
 }
 
@@ -175,7 +176,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
 
     r300->blend_color_state.state = CALLOC_STRUCT(r300_blend_color_state);
     r300->clip_state.state = CALLOC_STRUCT(pipe_clip_state);
-    r300->rs_block = CALLOC_STRUCT(r300_rs_block);
+    r300->rs_block_state.state = CALLOC_STRUCT(r300_rs_block);
     r300->scissor_state.state = CALLOC_STRUCT(pipe_scissor_state);
     r300->vertex_format_state.state = CALLOC_STRUCT(r300_vertex_info);
     r300->viewport_state.state = CALLOC_STRUCT(r300_viewport_state);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index 407891e175..c4c137d6f8 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -142,7 +142,6 @@ struct r300_ztop_state {
 #define R300_NEW_FRAMEBUFFERS    0x00000010
 #define R300_NEW_FRAGMENT_SHADER 0x00000020
 #define R300_NEW_FRAGMENT_SHADER_CONSTANTS    0x00000040
-#define R300_NEW_RS_BLOCK        0x00000100
 #define R300_NEW_SAMPLER         0x00000200
 #define R300_ANY_NEW_SAMPLERS    0x0001fe00
 #define R300_NEW_TEXTURE         0x00040000
@@ -294,7 +293,7 @@ struct r300_context {
     /* Rasterizer state. */
     struct r300_atom rs_state;
     /* RS block state. */
-    struct r300_rs_block* rs_block;
+    struct r300_atom rs_block_state;
     /* Sampler states. */
     struct r300_sampler_state* sampler_states[8];
     int sampler_count;
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 8081f1d956..8bc9da9361 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -627,10 +627,10 @@ void r300_emit_rs_state(struct r300_context* r300, void* state)
     END_CS;
 }
 
-void r300_emit_rs_block_state(struct r300_context* r300,
-                              struct r300_rs_block* rs)
+void r300_emit_rs_block_state(struct r300_context* r300, void* state)
 {
-    int i;
+    struct r300_rs_block* rs = (struct r300_rs_block*)state;
+    unsigned i;
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
     CS_LOCALS(r300);
 
@@ -1142,11 +1142,6 @@ void r300_emit_dirty_state(struct r300_context* r300)
         r300->dirty_state &= ~R300_NEW_FRAMEBUFFERS;
     }
 
-    if (r300->dirty_state & R300_NEW_RS_BLOCK) {
-        r300_emit_rs_block_state(r300, r300->rs_block);
-        r300->dirty_state &= ~R300_NEW_RS_BLOCK;
-    }
-
     /* Samplers and textures are tracked separately but emitted together. */
     if (r300->dirty_state &
             (R300_ANY_NEW_SAMPLERS | R300_ANY_NEW_TEXTURES)) {
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index 6788e3d655..c6dbc5af1e 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -61,8 +61,7 @@ void r300_emit_query_end(struct r300_context* r300);
 
 void r300_emit_rs_state(struct r300_context* r300, void* state);
 
-void r300_emit_rs_block_state(struct r300_context* r300,
-                              struct r300_rs_block* rs);
+void r300_emit_rs_block_state(struct r300_context* r300, void* state);
 
 void r300_emit_scissor_state(struct r300_context* r300, void* state);
 
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 7068a5a4ec..f3acd16579 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -720,7 +720,6 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state)
     r300->viewport_state.dirty = TRUE;
 
     /* XXX Clean these up when we move to atom emits */
-    r300->dirty_state |= R300_NEW_RS_BLOCK;
     if (r300->fs && r300->fs->inputs.wpos != ATTR_UNUSED) {
         r300->dirty_state |= R300_NEW_FRAGMENT_SHADER_CONSTANTS;
     }
diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index 1a1eabe5fb..a351c9d01b 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -305,7 +305,7 @@ static void r300_update_rs_block(struct r300_context* r300,
                                  struct r300_shader_semantics* vs_outputs,
                                  struct r300_shader_semantics* fs_inputs)
 {
-    struct r300_rs_block* rs = r300->rs_block;
+    struct r300_rs_block rs = { 0 };
     int i, col_count = 0, tex_count = 0, fp_offset = 0;
     void (*rX00_rs_col)(struct r300_rs_block*, int, int, boolean);
     void (*rX00_rs_col_write)(struct r300_rs_block*, int, int);
@@ -332,11 +332,11 @@ static void r300_update_rs_block(struct r300_context* r300,
             vs_outputs->color[1] != ATTR_UNUSED) {
             /* Always rasterize if it's written by the VS,
              * otherwise it locks up. */
-            rX00_rs_col(rs, col_count, i, FALSE);
+            rX00_rs_col(&rs, col_count, i, FALSE);
 
             /* Write it to the FS input register if it's used by the FS. */
             if (fs_inputs->color[i] != ATTR_UNUSED) {
-                rX00_rs_col_write(rs, col_count, fp_offset);
+                rX00_rs_col_write(&rs, col_count, fp_offset);
                 fp_offset++;
             }
             col_count++;
@@ -354,11 +354,11 @@ static void r300_update_rs_block(struct r300_context* r300,
         if (vs_outputs->generic[i] != ATTR_UNUSED) {
             /* Always rasterize if it's written by the VS,
              * otherwise it locks up. */
-            rX00_rs_tex(rs, tex_count, tex_count, FALSE);
+            rX00_rs_tex(&rs, tex_count, tex_count, FALSE);
 
             /* Write it to the FS input register if it's used by the FS. */
             if (fs_inputs->generic[i] != ATTR_UNUSED) {
-                rX00_rs_tex_write(rs, tex_count, fp_offset);
+                rX00_rs_tex_write(&rs, tex_count, fp_offset);
                 fp_offset++;
             }
             tex_count++;
@@ -375,11 +375,11 @@ static void r300_update_rs_block(struct r300_context* r300,
     if (vs_outputs->fog != ATTR_UNUSED) {
         /* Always rasterize if it's written by the VS,
          * otherwise it locks up. */
-        rX00_rs_tex(rs, tex_count, tex_count, TRUE);
+        rX00_rs_tex(&rs, tex_count, tex_count, TRUE);
 
         /* Write it to the FS input register if it's used by the FS. */
         if (fs_inputs->fog != ATTR_UNUSED) {
-            rX00_rs_tex_write(rs, tex_count, fp_offset);
+            rX00_rs_tex_write(&rs, tex_count, fp_offset);
             fp_offset++;
         }
         tex_count++;
@@ -394,8 +394,8 @@ static void r300_update_rs_block(struct r300_context* r300,
     /* Rasterize WPOS. */
     /* If the FS doesn't need it, it's not written by the VS. */
     if (fs_inputs->wpos != ATTR_UNUSED) {
-        rX00_rs_tex(rs, tex_count, tex_count, FALSE);
-        rX00_rs_tex_write(rs, tex_count, fp_offset);
+        rX00_rs_tex(&rs, tex_count, tex_count, FALSE);
+        rX00_rs_tex_write(&rs, tex_count, fp_offset);
 
         fp_offset++;
         tex_count++;
@@ -403,17 +403,23 @@ static void r300_update_rs_block(struct r300_context* r300,
 
     /* Rasterize at least one color, or bad things happen. */
     if (col_count == 0 && tex_count == 0) {
-        rX00_rs_col(rs, 0, 0, TRUE);
+        rX00_rs_col(&rs, 0, 0, TRUE);
         col_count++;
     }
 
-    rs->count = (tex_count*4) | (col_count << R300_IC_COUNT_SHIFT) |
+    rs.count = (tex_count*4) | (col_count << R300_IC_COUNT_SHIFT) |
         R300_HIRES_EN;
 
-    rs->inst_count = MAX3(col_count - 1, tex_count - 1, 0);
+    rs.inst_count = MAX3(col_count - 1, tex_count - 1, 0);
+
+    /* Now, after all that, see if we actually need to update the state. */
+    if (memcmp(r300->rs_block_state.state, &rs, sizeof(struct r300_rs_block))) {
+        memcpy(r300->rs_block_state.state, &rs, sizeof(struct r300_rs_block));
+        r300->rs_block_state.dirty = TRUE;
+    }
 }
 
-/* Update the vertex format. */
+/* Update the shader-dependant states. */
 static void r300_update_derived_shader_state(struct r300_context* r300)
 {
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
@@ -421,8 +427,6 @@ static void r300_update_derived_shader_state(struct r300_context* r300)
         (struct r300_vertex_info*)r300->vertex_format_state.state;
     struct vertex_info* vinfo = &vformat->vinfo;
 
-    /* Reset structures */
-    memset(r300->rs_block, 0, sizeof(struct r300_rs_block));
     /* Mmm, delicious hax */
     memset(r300->vertex_format_state.state, 0, sizeof(struct r300_vertex_info));
     memcpy(vinfo->hwfmt, r300->vs->hwfmt, sizeof(uint)*4);
@@ -437,8 +441,6 @@ static void r300_update_derived_shader_state(struct r300_context* r300)
             (struct vertex_info*)r300->vertex_format_state.state);
         r300_swtcl_vertex_psc(r300);
     }
-
-    r300->dirty_state |= R300_NEW_RS_BLOCK;
 }
 
 static boolean r300_dsa_writes_depth_stencil(struct r300_dsa_state* dsa)
-- 
cgit v1.2.3


From aa2423b5b32735b5d71a71546b8b0300aab3c1b8 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Tue, 26 Jan 2010 23:26:41 -0800
Subject: i915g: Remove unnecessary headers.

---
 src/gallium/drivers/i915/i915_surface.c | 2 --
 src/gallium/drivers/i915/i915_texture.c | 1 -
 2 files changed, 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i915/i915_surface.c b/src/gallium/drivers/i915/i915_surface.c
index 13684aa59c..1ff6b9f4c6 100644
--- a/src/gallium/drivers/i915/i915_surface.c
+++ b/src/gallium/drivers/i915/i915_surface.c
@@ -28,8 +28,6 @@
 #include "i915_context.h"
 #include "i915_blit.h"
 #include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_format.h"
 
 
diff --git a/src/gallium/drivers/i915/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c
index 441bc4f193..612e5c1cdd 100644
--- a/src/gallium/drivers/i915/i915_texture.c
+++ b/src/gallium/drivers/i915/i915_texture.c
@@ -34,7 +34,6 @@
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 #include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
-- 
cgit v1.2.3


From 2a0622182281bfa8402d3c64345c698583c383d7 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Tue, 26 Jan 2010 23:28:23 -0800
Subject: softpipe: Remove unnecessary header.

---
 src/gallium/drivers/softpipe/sp_winsys.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_winsys.c b/src/gallium/drivers/softpipe/sp_winsys.c
index 8169071dc9..38bcd64c6a 100644
--- a/src/gallium/drivers/softpipe/sp_winsys.c
+++ b/src/gallium/drivers/softpipe/sp_winsys.c
@@ -39,7 +39,6 @@
 #include "pipe/internal/p_winsys_screen.h"/* port to just p_screen */
 #include "pipe/p_format.h"
 #include "pipe/p_context.h"
-#include "pipe/p_inlines.h"
 #include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
-- 
cgit v1.2.3


From b8656c4825b9e054f05258773ba012e41d4fcdee Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 27 Jan 2010 09:21:57 -0700
Subject: cell: remove commas from structs wrapped in PIPE_ALIGN_TYPE macro

This avoids the need to make PIPE_ALIGN_TYPE a variadic macro.
---
 src/gallium/drivers/cell/spu/spu_main.h | 25 ++++++++++++++++++-------
 1 file changed, 18 insertions(+), 7 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/cell/spu/spu_main.h b/src/gallium/drivers/cell/spu/spu_main.h
index b18f4c22ef..a9d72f84d5 100644
--- a/src/gallium/drivers/cell/spu/spu_main.h
+++ b/src/gallium/drivers/cell/spu/spu_main.h
@@ -100,8 +100,10 @@ struct spu_framebuffer
    void *depth_start;              /**< addr of depth surface in main memory */
    enum pipe_format color_format;
    enum pipe_format depth_format;
-   uint width, height;             /**< size in pixels */
-   uint width_tiles, height_tiles; /**< width and height in tiles */
+   uint width;                     /**< width in pixels */
+   uint height;                    /**< height in pixels */
+   uint width_tiles;               /**< width in tiles */
+   uint height_tiles;              /**< width in tiles */
 
    uint color_clear_value;
    uint depth_clear_value;
@@ -116,15 +118,23 @@ PIPE_ALIGN_TYPE(16,
 struct spu_texture_level
 {
    void *start;
-   ushort width, height, depth;
+   ushort width;
+   ushort height;
+   ushort depth;
    ushort tiles_per_row;
    uint bytes_per_image;
    /** texcoord scale factors */
-   vector float scale_s, scale_t, scale_r;
+   vector float scale_s;
+   vector float scale_t;
+   vector float scale_r;
    /** texcoord masks (if REPEAT then size-1, else ~0) */
-   vector signed int mask_s, mask_t, mask_r;
+   vector signed int mask_s;
+   vector signed int mask_t;
+   vector signed int mask_r;
    /** texcoord clamp limits */
-   vector signed int max_s, max_t, max_r;
+   vector signed int max_s;
+   vector signed int max_t;
+   vector signed int max_r;
 });
 
 
@@ -166,7 +176,8 @@ struct spu_global
    boolean read_depth_stencil;
 
    /** Current tiles' status */
-   ubyte cur_ctile_status, cur_ztile_status;
+   ubyte cur_ctile_status;
+   ubyte cur_ztile_status;
 
    /** Status of all tiles in framebuffer */
    PIPE_ALIGN_VAR(16) ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE];
-- 
cgit v1.2.3


From 01ed5c6ef5395038add9eae2735fa32e40d39895 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Wed, 27 Jan 2010 14:04:53 -0800
Subject: r300g: Silence GCC "missing braces around initializer" warning.

---
 src/gallium/drivers/r300/r300_state_derived.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c
index a351c9d01b..bad9e76067 100644
--- a/src/gallium/drivers/r300/r300_state_derived.c
+++ b/src/gallium/drivers/r300/r300_state_derived.c
@@ -305,7 +305,7 @@ static void r300_update_rs_block(struct r300_context* r300,
                                  struct r300_shader_semantics* vs_outputs,
                                  struct r300_shader_semantics* fs_inputs)
 {
-    struct r300_rs_block rs = { 0 };
+    struct r300_rs_block rs = { { 0 } };
     int i, col_count = 0, tex_count = 0, fp_offset = 0;
     void (*rX00_rs_col)(struct r300_rs_block*, int, int, boolean);
     void (*rX00_rs_col_write)(struct r300_rs_block*, int, int);
-- 
cgit v1.2.3


From 83e6ba564c82ddf6403223588e8155f8fcb570e6 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Wed, 27 Jan 2010 15:54:21 -0800
Subject: i965g: Remove unnecessary headers.

---
 src/gallium/drivers/i965/brw_cc.c               | 1 -
 src/gallium/drivers/i965/brw_clip.c             | 1 -
 src/gallium/drivers/i965/brw_clip_line.c        | 1 -
 src/gallium/drivers/i965/brw_clip_point.c       | 1 -
 src/gallium/drivers/i965/brw_clip_tri.c         | 1 -
 src/gallium/drivers/i965/brw_clip_util.c        | 1 -
 src/gallium/drivers/i965/brw_context.c          | 1 -
 src/gallium/drivers/i965/brw_curbe.c            | 2 --
 src/gallium/drivers/i965/brw_draw.c             | 1 -
 src/gallium/drivers/i965/brw_gs.c               | 1 -
 src/gallium/drivers/i965/brw_gs_emit.c          | 1 -
 src/gallium/drivers/i965/brw_pipe_fb.c          | 1 -
 src/gallium/drivers/i965/brw_pipe_sampler.c     | 1 -
 src/gallium/drivers/i965/brw_sf.c               | 1 -
 src/gallium/drivers/i965/brw_sf_emit.c          | 1 -
 src/gallium/drivers/i965/brw_state_cache.c      | 1 -
 src/gallium/drivers/i965/brw_util.c             | 2 --
 src/gallium/drivers/i965/brw_vs.c               | 2 --
 src/gallium/drivers/i965/brw_vs_surface_state.c | 1 -
 src/gallium/drivers/i965/brw_wm_fp.c            | 1 -
 src/gallium/drivers/i965/brw_wm_surface_state.c | 1 -
 21 files changed, 24 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/i965/brw_cc.c b/src/gallium/drivers/i965/brw_cc.c
index 3e070f5591..4a543276f5 100644
--- a/src/gallium/drivers/i965/brw_cc.c
+++ b/src/gallium/drivers/i965/brw_cc.c
@@ -32,7 +32,6 @@
 
 #include "brw_context.h"
 #include "brw_state.h"
-#include "brw_defines.h"
 
 
 static enum pipe_error prepare_cc_vp( struct brw_context *brw )
diff --git a/src/gallium/drivers/i965/brw_clip.c b/src/gallium/drivers/i965/brw_clip.c
index d67a1a6263..ccba205e8c 100644
--- a/src/gallium/drivers/i965/brw_clip.c
+++ b/src/gallium/drivers/i965/brw_clip.c
@@ -38,7 +38,6 @@
 #include "brw_defines.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_state.h"
 #include "brw_pipe_rast.h"
 #include "brw_clip.h"
diff --git a/src/gallium/drivers/i965/brw_clip_line.c b/src/gallium/drivers/i965/brw_clip_line.c
index 54282d975e..66caadc4d5 100644
--- a/src/gallium/drivers/i965/brw_clip_line.c
+++ b/src/gallium/drivers/i965/brw_clip_line.c
@@ -33,7 +33,6 @@
 
 #include "brw_defines.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_clip.h"
 
 
diff --git a/src/gallium/drivers/i965/brw_clip_point.c b/src/gallium/drivers/i965/brw_clip_point.c
index e0a5330556..124156c1b5 100644
--- a/src/gallium/drivers/i965/brw_clip_point.c
+++ b/src/gallium/drivers/i965/brw_clip_point.c
@@ -31,7 +31,6 @@
 
 #include "brw_defines.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_clip.h"
 
 
diff --git a/src/gallium/drivers/i965/brw_clip_tri.c b/src/gallium/drivers/i965/brw_clip_tri.c
index 4cde7294ea..069524bc14 100644
--- a/src/gallium/drivers/i965/brw_clip_tri.c
+++ b/src/gallium/drivers/i965/brw_clip_tri.c
@@ -31,7 +31,6 @@
 
 #include "brw_defines.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_clip.h"
 
 static void release_tmps( struct brw_clip_compile *c )
diff --git a/src/gallium/drivers/i965/brw_clip_util.c b/src/gallium/drivers/i965/brw_clip_util.c
index 97a5710310..23e51ee9bc 100644
--- a/src/gallium/drivers/i965/brw_clip_util.c
+++ b/src/gallium/drivers/i965/brw_clip_util.c
@@ -32,7 +32,6 @@
 
 #include "brw_defines.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_clip.h"
 
 
diff --git a/src/gallium/drivers/i965/brw_context.c b/src/gallium/drivers/i965/brw_context.c
index e67551882d..d60b7b99b6 100644
--- a/src/gallium/drivers/i965/brw_context.c
+++ b/src/gallium/drivers/i965/brw_context.c
@@ -34,7 +34,6 @@
 #include "util/u_simple_list.h"
 
 #include "brw_context.h"
-#include "brw_defines.h"
 #include "brw_draw.h"
 #include "brw_state.h"
 #include "brw_batchbuffer.h"
diff --git a/src/gallium/drivers/i965/brw_curbe.c b/src/gallium/drivers/i965/brw_curbe.c
index 3f031577d5..4b215a001c 100644
--- a/src/gallium/drivers/i965/brw_curbe.c
+++ b/src/gallium/drivers/i965/brw_curbe.c
@@ -36,9 +36,7 @@
 #include "brw_context.h"
 #include "brw_defines.h"
 #include "brw_state.h"
-#include "brw_util.h"
 #include "brw_debug.h"
-#include "brw_screen.h"
 
 
 /**
diff --git a/src/gallium/drivers/i965/brw_draw.c b/src/gallium/drivers/i965/brw_draw.c
index ea8d39adaf..1b5cd23995 100644
--- a/src/gallium/drivers/i965/brw_draw.c
+++ b/src/gallium/drivers/i965/brw_draw.c
@@ -34,7 +34,6 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_debug.h"
-#include "brw_screen.h"
 
 #include "brw_batchbuffer.h"
 
diff --git a/src/gallium/drivers/i965/brw_gs.c b/src/gallium/drivers/i965/brw_gs.c
index 921b201bae..06826635a8 100644
--- a/src/gallium/drivers/i965/brw_gs.c
+++ b/src/gallium/drivers/i965/brw_gs.c
@@ -34,7 +34,6 @@
 #include "brw_defines.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_state.h"
 #include "brw_gs.h"
 
diff --git a/src/gallium/drivers/i965/brw_gs_emit.c b/src/gallium/drivers/i965/brw_gs_emit.c
index fd8e2acced..9b58773b3b 100644
--- a/src/gallium/drivers/i965/brw_gs_emit.c
+++ b/src/gallium/drivers/i965/brw_gs_emit.c
@@ -35,7 +35,6 @@
 #include "brw_defines.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_gs.h"
 
 static void brw_gs_alloc_regs( struct brw_gs_compile *c,
diff --git a/src/gallium/drivers/i965/brw_pipe_fb.c b/src/gallium/drivers/i965/brw_pipe_fb.c
index 5d4e5025f9..c1f049272a 100644
--- a/src/gallium/drivers/i965/brw_pipe_fb.c
+++ b/src/gallium/drivers/i965/brw_pipe_fb.c
@@ -3,7 +3,6 @@
 #include "pipe/p_state.h"
 
 #include "brw_context.h"
-#include "brw_debug.h"
 
 /**
  * called from intelDrawBuffer()
diff --git a/src/gallium/drivers/i965/brw_pipe_sampler.c b/src/gallium/drivers/i965/brw_pipe_sampler.c
index 81712798a5..ef6c1bb315 100644
--- a/src/gallium/drivers/i965/brw_pipe_sampler.c
+++ b/src/gallium/drivers/i965/brw_pipe_sampler.c
@@ -7,7 +7,6 @@
 
 #include "brw_context.h"
 #include "brw_defines.h"
-#include "brw_debug.h"
 
 
 
diff --git a/src/gallium/drivers/i965/brw_sf.c b/src/gallium/drivers/i965/brw_sf.c
index e1986a9dbb..6e99eaa09d 100644
--- a/src/gallium/drivers/i965/brw_sf.c
+++ b/src/gallium/drivers/i965/brw_sf.c
@@ -36,7 +36,6 @@
 #include "brw_context.h"
 #include "brw_pipe_rast.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_sf.h"
 #include "brw_state.h"
 
diff --git a/src/gallium/drivers/i965/brw_sf_emit.c b/src/gallium/drivers/i965/brw_sf_emit.c
index 3b85725e36..497634ec9e 100644
--- a/src/gallium/drivers/i965/brw_sf_emit.c
+++ b/src/gallium/drivers/i965/brw_sf_emit.c
@@ -35,7 +35,6 @@
 #include "brw_defines.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "brw_util.h"
 #include "brw_sf.h"
 
 
diff --git a/src/gallium/drivers/i965/brw_state_cache.c b/src/gallium/drivers/i965/brw_state_cache.c
index 16b643ceb2..85c20076fb 100644
--- a/src/gallium/drivers/i965/brw_state_cache.c
+++ b/src/gallium/drivers/i965/brw_state_cache.c
@@ -59,7 +59,6 @@
 
 #include "brw_debug.h"
 #include "brw_state.h"
-#include "brw_batchbuffer.h"
 
 /* XXX: Fixme - have to include these to get the sizes of the prog_key
  * structs:
diff --git a/src/gallium/drivers/i965/brw_util.c b/src/gallium/drivers/i965/brw_util.c
index 458058d668..1fd2e29713 100644
--- a/src/gallium/drivers/i965/brw_util.c
+++ b/src/gallium/drivers/i965/brw_util.c
@@ -30,8 +30,6 @@
   */
          
 
-#include "brw_util.h"
-#include "brw_defines.h"
 
 
 
diff --git a/src/gallium/drivers/i965/brw_vs.c b/src/gallium/drivers/i965/brw_vs.c
index e3ea5a3a13..ca8ee79550 100644
--- a/src/gallium/drivers/i965/brw_vs.c
+++ b/src/gallium/drivers/i965/brw_vs.c
@@ -33,9 +33,7 @@
 
 #include "brw_context.h"
 #include "brw_vs.h"
-#include "brw_util.h"
 #include "brw_state.h"
-#include "brw_pipe_rast.h"
 
 
 
diff --git a/src/gallium/drivers/i965/brw_vs_surface_state.c b/src/gallium/drivers/i965/brw_vs_surface_state.c
index 177a5170d2..004e3cb4e6 100644
--- a/src/gallium/drivers/i965/brw_vs_surface_state.c
+++ b/src/gallium/drivers/i965/brw_vs_surface_state.c
@@ -31,7 +31,6 @@
 
 #include "brw_context.h"
 #include "brw_state.h"
-#include "brw_defines.h"
 #include "brw_winsys.h"
 
 /* XXX: disabled true constant buffer functionality
diff --git a/src/gallium/drivers/i965/brw_wm_fp.c b/src/gallium/drivers/i965/brw_wm_fp.c
index 9c5b527f89..9c67759ad0 100644
--- a/src/gallium/drivers/i965/brw_wm_fp.c
+++ b/src/gallium/drivers/i965/brw_wm_fp.c
@@ -41,7 +41,6 @@
 #include "tgsi/tgsi_util.h"
 
 #include "brw_wm.h"
-#include "brw_util.h"
 #include "brw_debug.h"
 
 
diff --git a/src/gallium/drivers/i965/brw_wm_surface_state.c b/src/gallium/drivers/i965/brw_wm_surface_state.c
index f92b8198ed..b01a7f194b 100644
--- a/src/gallium/drivers/i965/brw_wm_surface_state.c
+++ b/src/gallium/drivers/i965/brw_wm_surface_state.c
@@ -34,7 +34,6 @@
 #include "brw_batchbuffer.h"
 #include "brw_context.h"
 #include "brw_state.h"
-#include "brw_defines.h"
 #include "brw_screen.h"
 
 
-- 
cgit v1.2.3


From 978568c647844693f602364bd9e1041d1cecea4f Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Wed, 27 Jan 2010 16:11:23 -0800
Subject: r300g: Remove unnecessary headers.

---
 src/gallium/drivers/r300/r300_blit.c    | 2 --
 src/gallium/drivers/r300/r300_context.c | 4 ----
 src/gallium/drivers/r300/r300_debug.c   | 2 --
 src/gallium/drivers/r300/r300_emit.c    | 2 --
 src/gallium/drivers/r300/r300_flush.c   | 1 -
 src/gallium/drivers/r300/r300_state.c   | 1 -
 6 files changed, 12 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_blit.c b/src/gallium/drivers/r300/r300_blit.c
index c14414fff6..11c21e071b 100644
--- a/src/gallium/drivers/r300/r300_blit.c
+++ b/src/gallium/drivers/r300/r300_blit.c
@@ -23,8 +23,6 @@
 #include "r300_blit.h"
 #include "r300_context.h"
 
-#include "util/u_rect.h"
-
 static void r300_blitter_save_states(struct r300_context* r300)
 {
     util_blitter_save_blend(r300->blitter, r300->blend_state.state);
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 052cc23a77..694452a389 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -22,9 +22,6 @@
 
 #include "draw/draw_context.h"
 
-#include "tgsi/tgsi_scan.h"
-
-#include "util/u_hash_table.h"
 #include "util/u_memory.h"
 #include "util/u_simple_list.h"
 
@@ -35,7 +32,6 @@
 #include "r300_query.h"
 #include "r300_render.h"
 #include "r300_screen.h"
-#include "r300_state_derived.h"
 #include "r300_state_invariant.h"
 #include "r300_texture.h"
 #include "r300_winsys.h"
diff --git a/src/gallium/drivers/r300/r300_debug.c b/src/gallium/drivers/r300/r300_debug.c
index 00d4f31c2b..b881730848 100644
--- a/src/gallium/drivers/r300/r300_debug.c
+++ b/src/gallium/drivers/r300/r300_debug.c
@@ -22,8 +22,6 @@
 
 #include "r300_context.h"
 
-#include <ctype.h>
-
 
 struct debug_option {
     const char * name;
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 8bc9da9361..8c5b1607d0 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -32,9 +32,7 @@
 #include "r300_emit.h"
 #include "r300_fs.h"
 #include "r300_screen.h"
-#include "r300_state_derived.h"
 #include "r300_state_inlines.h"
-#include "r300_texture.h"
 #include "r300_vs.h"
 
 void r300_emit_blend_state(struct r300_context* r300, void* state)
diff --git a/src/gallium/drivers/r300/r300_flush.c b/src/gallium/drivers/r300/r300_flush.c
index 15e612d8a6..e37d309270 100644
--- a/src/gallium/drivers/r300/r300_flush.c
+++ b/src/gallium/drivers/r300/r300_flush.c
@@ -29,7 +29,6 @@
 #include "r300_cs.h"
 #include "r300_emit.h"
 #include "r300_flush.h"
-#include "r300_state_invariant.h"
 
 static void r300_flush(struct pipe_context* pipe,
                        unsigned flags,
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index f3acd16579..b1c9aebd28 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -30,7 +30,6 @@
 #include "tgsi/tgsi_parse.h"
 
 #include "pipe/p_config.h"
-#include "pipe/internal/p_winsys_screen.h"
 
 #include "r300_context.h"
 #include "r300_reg.h"
-- 
cgit v1.2.3


From c2bd1657a80cec2099e648ec62424d0d06cd6b9b Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Wed, 27 Jan 2010 18:08:37 -0800
Subject: llvmpipe: Remove unnecessary headers.

---
 src/gallium/drivers/llvmpipe/lp_setup.c         | 1 -
 src/gallium/drivers/llvmpipe/lp_state_derived.c | 1 -
 src/gallium/drivers/llvmpipe/lp_texture.c       | 1 -
 3 files changed, 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c
index 0b2d3a2801..50b07308c7 100644
--- a/src/gallium/drivers/llvmpipe/lp_setup.c
+++ b/src/gallium/drivers/llvmpipe/lp_setup.c
@@ -40,7 +40,6 @@
 #include "draw/draw_private.h"
 #include "draw/draw_vertex.h"
 #include "pipe/p_shader_tokens.h"
-#include "pipe/p_thread.h"
 #include "util/u_format.h"
 #include "util/u_math.h"
 #include "util/u_memory.h"
diff --git a/src/gallium/drivers/llvmpipe/lp_state_derived.c b/src/gallium/drivers/llvmpipe/lp_state_derived.c
index 6c1ef6bc42..0155b9be50 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_derived.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_derived.c
@@ -33,7 +33,6 @@
 #include "draw/draw_private.h"
 #include "lp_context.h"
 #include "lp_screen.h"
-#include "lp_tex_cache.h"
 #include "lp_state.h"
 
 
diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c
index 039539d682..3701219dcf 100644
--- a/src/gallium/drivers/llvmpipe/lp_texture.c
+++ b/src/gallium/drivers/llvmpipe/lp_texture.c
@@ -33,7 +33,6 @@
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
 #include "pipe/p_inlines.h"
-#include "pipe/internal/p_winsys_screen.h"
 
 #include "util/u_format.h"
 #include "util/u_math.h"
-- 
cgit v1.2.3


From b9ae570bab5d0fefef93a9f6808603971304a74d Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Wed, 27 Jan 2010 21:31:36 -0800
Subject: r300g: Atomize framebuffers.

There might be some optimizations possible here...
---
 src/gallium/drivers/r300/r300_blit.c    | 12 ++++++-----
 src/gallium/drivers/r300/r300_context.c |  1 +
 src/gallium/drivers/r300/r300_context.h |  5 ++---
 src/gallium/drivers/r300/r300_emit.c    | 36 ++++++++++++++++-----------------
 src/gallium/drivers/r300/r300_emit.h    |  3 +--
 src/gallium/drivers/r300/r300_state.c   | 18 +++++++++++++----
 6 files changed, 43 insertions(+), 32 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_blit.c b/src/gallium/drivers/r300/r300_blit.c
index 11c21e071b..cdedb30220 100644
--- a/src/gallium/drivers/r300/r300_blit.c
+++ b/src/gallium/drivers/r300/r300_blit.c
@@ -73,13 +73,15 @@ void r300_clear(struct pipe_context* pipe,
      */
 
     struct r300_context* r300 = r300_context(pipe);
+    struct pipe_framebuffer_state* fb =
+        (struct pipe_framebuffer_state*)r300->fb_state.state;
 
     r300_blitter_save_states(r300);
 
     util_blitter_clear(r300->blitter,
-                       r300->framebuffer_state.width,
-                       r300->framebuffer_state.height,
-                       r300->framebuffer_state.nr_cbufs,
+                       fb->width,
+                       fb->height,
+                       fb->nr_cbufs,
                        buffers, rgba, depth, stencil);
 }
 
@@ -97,7 +99,7 @@ void r300_surface_copy(struct pipe_context* pipe,
      * is really transparent. The states will be restored by the blitter once
      * copying is done. */
     r300_blitter_save_states(r300);
-    util_blitter_save_framebuffer(r300->blitter, &r300->framebuffer_state);
+    util_blitter_save_framebuffer(r300->blitter, r300->fb_state.state);
 
     util_blitter_save_fragment_sampler_states(
         r300->blitter, r300->sampler_count, (void**)r300->sampler_states);
@@ -121,7 +123,7 @@ void r300_surface_fill(struct pipe_context* pipe,
     struct r300_context* r300 = r300_context(pipe);
 
     r300_blitter_save_states(r300);
-    util_blitter_save_framebuffer(r300->blitter, &r300->framebuffer_state);
+    util_blitter_save_framebuffer(r300->blitter, r300->fb_state.state);
 
     util_blitter_fill(r300->blitter,
                       dst, dstx, dsty, width, height, value);
diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 694452a389..9837530a7a 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -118,6 +118,7 @@ static void r300_setup_atoms(struct r300_context* r300)
     R300_INIT_ATOM(blend_color, 3);
     R300_INIT_ATOM(clip, 29);
     R300_INIT_ATOM(dsa, 8);
+    R300_INIT_ATOM(fb, 56);
     R300_INIT_ATOM(rs, 25);
     R300_INIT_ATOM(scissor, 3);
     R300_INIT_ATOM(viewport, 9);
diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h
index c4c137d6f8..3f461640a0 100644
--- a/src/gallium/drivers/r300/r300_context.h
+++ b/src/gallium/drivers/r300/r300_context.h
@@ -139,7 +139,6 @@ struct r300_ztop_state {
     uint32_t z_buffer_top;      /* R300_ZB_ZTOP: 0x4f14 */
 };
 
-#define R300_NEW_FRAMEBUFFERS    0x00000010
 #define R300_NEW_FRAGMENT_SHADER 0x00000020
 #define R300_NEW_FRAGMENT_SHADER_CONSTANTS    0x00000040
 #define R300_NEW_SAMPLER         0x00000200
@@ -288,8 +287,8 @@ struct r300_context {
     struct r300_atom dsa_state;
     /* Fragment shader. */
     struct r300_fragment_shader* fs;
-    /* Framebuffer state. We currently don't need our own version of this. */
-    struct pipe_framebuffer_state framebuffer_state;
+    /* Framebuffer state. */
+    struct r300_atom fb_state;
     /* Rasterizer state. */
     struct r300_atom rs_state;
     /* RS block state. */
diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 8c5b1607d0..f1d32764fd 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -38,12 +38,14 @@
 void r300_emit_blend_state(struct r300_context* r300, void* state)
 {
     struct r300_blend_state* blend = (struct r300_blend_state*)state;
+    struct pipe_framebuffer_state* fb =
+        (struct pipe_framebuffer_state*)r300->fb_state.state;
     CS_LOCALS(r300);
 
     BEGIN_CS(8);
     OUT_CS_REG(R300_RB3D_ROPCNTL, blend->rop);
     OUT_CS_REG_SEQ(R300_RB3D_CBLEND, 3);
-    if (r300->framebuffer_state.nr_cbufs) {
+    if (fb->nr_cbufs) {
         OUT_CS(blend->blend_control);
         OUT_CS(blend->alpha_blend_control);
         OUT_CS(blend->color_channel_mask);
@@ -110,6 +112,8 @@ void r300_emit_dsa_state(struct r300_context* r300, void* state)
 {
     struct r300_dsa_state* dsa = (struct r300_dsa_state*)state;
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
+    struct pipe_framebuffer_state* fb =
+        (struct pipe_framebuffer_state*)r300->fb_state.state;
     CS_LOCALS(r300);
 
     BEGIN_CS(r300screen->caps->is_r500 ? 8 : 6);
@@ -122,7 +126,7 @@ void r300_emit_dsa_state(struct r300_context* r300, void* state)
 
     OUT_CS_REG_SEQ(R300_ZB_CNTL, 3);
 
-    if (r300->framebuffer_state.zsbuf) {
+    if (fb->zsbuf) {
         OUT_CS(dsa->z_buffer_control);
         OUT_CS(dsa->z_stencil_control);
     } else {
@@ -382,17 +386,14 @@ void r500_emit_fs_constant_buffer(struct r300_context* r300,
     END_CS;
 }
 
-void r300_emit_fb_state(struct r300_context* r300,
-                        struct pipe_framebuffer_state* fb)
+void r300_emit_fb_state(struct r300_context* r300, void* state)
 {
+    struct pipe_framebuffer_state* fb = (struct pipe_framebuffer_state*)state;
     struct r300_texture* tex;
     struct pipe_surface* surf;
     int i;
     CS_LOCALS(r300);
 
-    /* Shouldn't fail unless there is a bug in the state tracker. */
-    assert(fb->nr_cbufs <= 4);
-
     BEGIN_CS((10 * fb->nr_cbufs) + (2 * (4 - fb->nr_cbufs)) +
              (fb->zsbuf ? 10 : 0) + 6);
 
@@ -671,11 +672,13 @@ void r300_emit_scissor_state(struct r300_context* r300, void* state)
     uint32_t top_left, bottom_right;
     struct r300_screen* r300screen = r300_screen(r300->context.screen);
     struct pipe_scissor_state* scissor = (struct pipe_scissor_state*)state;
+    struct pipe_framebuffer_state* fb =
+        (struct pipe_framebuffer_state*)r300->fb_state.state;
     CS_LOCALS(r300);
 
     minx = miny = 0;
-    maxx = r300->framebuffer_state.width;
-    maxy = r300->framebuffer_state.height;
+    maxx = fb->width;
+    maxy = fb->height;
 
     if (((struct r300_rs_state*)r300->rs_state.state)->rs.scissor) {
         minx = MAX2(minx, scissor->minx);
@@ -1009,6 +1012,8 @@ static void r300_flush_pvs(struct r300_context* r300)
 
 void r300_emit_buffer_validate(struct r300_context *r300)
 {
+    struct pipe_framebuffer_state* fb =
+        (struct pipe_framebuffer_state*)r300->fb_state.state;
     struct r300_texture* tex;
     unsigned i;
     boolean invalid = FALSE;
@@ -1018,8 +1023,8 @@ void r300_emit_buffer_validate(struct r300_context *r300)
 
 validate:
     /* Color buffers... */
-    for (i = 0; i < r300->framebuffer_state.nr_cbufs; i++) {
-        tex = (struct r300_texture*)r300->framebuffer_state.cbufs[i]->texture;
+    for (i = 0; i < fb->nr_cbufs; i++) {
+        tex = (struct r300_texture*)fb->cbufs[i]->texture;
         assert(tex && tex->buffer && "cbuf is marked, but NULL!");
         if (!r300->winsys->add_buffer(r300->winsys, tex->buffer,
                     0, RADEON_GEM_DOMAIN_VRAM)) {
@@ -1028,8 +1033,8 @@ validate:
         }
     }
     /* ...depth buffer... */
-    if (r300->framebuffer_state.zsbuf) {
-        tex = (struct r300_texture*)r300->framebuffer_state.zsbuf->texture;
+    if (fb->zsbuf) {
+        tex = (struct r300_texture*)fb->zsbuf->texture;
         assert(tex && tex->buffer && "zsbuf is marked, but NULL!");
         if (!r300->winsys->add_buffer(r300->winsys, tex->buffer,
                     0, RADEON_GEM_DOMAIN_VRAM)) {
@@ -1135,11 +1140,6 @@ void r300_emit_dirty_state(struct r300_context* r300)
         r300->dirty_state &= ~R300_NEW_FRAGMENT_SHADER_CONSTANTS;
     }
 
-    if (r300->dirty_state & R300_NEW_FRAMEBUFFERS) {
-        r300_emit_fb_state(r300, &r300->framebuffer_state);
-        r300->dirty_state &= ~R300_NEW_FRAMEBUFFERS;
-    }
-
     /* Samplers and textures are tracked separately but emitted together. */
     if (r300->dirty_state &
             (R300_ANY_NEW_SAMPLERS | R300_ANY_NEW_TEXTURES)) {
diff --git a/src/gallium/drivers/r300/r300_emit.h b/src/gallium/drivers/r300/r300_emit.h
index c6dbc5af1e..6b96d9b57c 100644
--- a/src/gallium/drivers/r300/r300_emit.h
+++ b/src/gallium/drivers/r300/r300_emit.h
@@ -51,8 +51,7 @@ void r500_emit_fragment_program_code(struct r300_context* r300,
 void r500_emit_fs_constant_buffer(struct r300_context* r300,
                                   struct rc_constant_list* constants);
 
-void r300_emit_fb_state(struct r300_context* r300,
-                        struct pipe_framebuffer_state* fb);
+void r300_emit_fb_state(struct r300_context* r300, void* state);
 
 void r300_emit_query_begin(struct r300_context* r300,
                            struct r300_query* query);
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index b1c9aebd28..1b920563a5 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -487,20 +487,30 @@ static void
     struct r300_context* r300 = r300_context(pipe);
     uint32_t zbuffer_bpp = 0;
 
+    r300->fb_state.size = (10 * state->nr_cbufs) +
+        (2 * (4 - state->nr_cbufs)) +
+        (state->zsbuf ? 10 : 0) + 6;
+
+    if (state->nr_cbufs > 4) {
+        debug_printf("r300: Implementation error: Too many MRTs in %s, "
+            "refusing to bind framebuffer state!\n", __FUNCTION__);
+        return;
+    }
+
     if (r300->draw) {
         draw_flush(r300->draw);
     }
 
-    r300->framebuffer_state = *state;
+    r300->fb_state.state = state;
 
     /* Don't rely on the order of states being set for the first time. */
-    r300->dirty_state |= R300_NEW_FRAMEBUFFERS;
-
+    /* XXX wait what */
     r300->blend_state.dirty = TRUE;
     r300->dsa_state.dirty = TRUE;
+    r300->fb_state.dirty = TRUE;
     r300->scissor_state.dirty = TRUE;
 
-    /* Polyfon offset depends on the zbuffer bit depth. */
+    /* Polygon offset depends on the zbuffer bit depth. */
     if (state->zsbuf && r300->polygon_offset_enabled) {
         switch (util_format_get_blocksize(state->zsbuf->texture->format)) {
             case 2:
-- 
cgit v1.2.3


From 12f97d9c09dee7fffe073c943fd8428db4362e0c Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Thu, 28 Jan 2010 01:28:59 -0800
Subject: r300g: Fix bizarre half-flushes with immd mode.

---
 src/gallium/drivers/r300/r300_render.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index 354f3bb921..bcb3707149 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -168,6 +168,8 @@ static void r300_emit_draw_arrays_immediate(struct r300_context *r300,
         }
     }
 
+    r300_emit_dirty_state(r300);
+
     BEGIN_CS(10 + count * vertex_size);
     OUT_CS_REG(R300_GA_COLOR_CONTROL,
             r300_provoking_vertex_fixes(r300, mode));
@@ -419,7 +421,6 @@ void r300_draw_arrays(struct pipe_context* pipe, unsigned mode,
     r300_emit_buffer_validate(r300);
 
     if (immd_is_good_idea(r300, count)) {
-        r300_emit_dirty_state(r300);
         r300_emit_draw_arrays_immediate(r300, mode, start, count);
     } else {
         if (!r300_setup_vertex_buffers(r300)) {
-- 
cgit v1.2.3


From 835bab0143e11ab98551a061043f944fd6eab456 Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@vmware.com>
Date: Tue, 19 Jan 2010 13:20:15 +0100
Subject: gallium: Implement 2D constant buffers for fragment shader in
 softpipe.

---
 src/gallium/auxiliary/draw/draw_gs.c          |   2 +-
 src/gallium/auxiliary/draw/draw_vs_exec.c     |   2 +-
 src/gallium/auxiliary/tgsi/tgsi_exec.c        | 267 +++++++++++---------------
 src/gallium/auxiliary/tgsi/tgsi_exec.h        |   2 +-
 src/gallium/drivers/softpipe/sp_context.c     |  10 +-
 src/gallium/drivers/softpipe/sp_context.h     |   4 +-
 src/gallium/drivers/softpipe/sp_draw_arrays.c |  35 ++--
 src/gallium/drivers/softpipe/sp_fs_sse.c      |   2 +-
 src/gallium/drivers/softpipe/sp_quad_fs.c     |   7 +-
 src/gallium/drivers/softpipe/sp_state_fs.c    |   4 +-
 10 files changed, 158 insertions(+), 177 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/auxiliary/draw/draw_gs.c b/src/gallium/auxiliary/draw/draw_gs.c
index daf8d071f1..c3cc365a8d 100644
--- a/src/gallium/auxiliary/draw/draw_gs.c
+++ b/src/gallium/auxiliary/draw/draw_gs.c
@@ -302,7 +302,7 @@ void draw_geometry_shader_run(struct draw_geometry_shader *shader,
    unsigned num_primitives = count/num_vertices;
    unsigned inputs_from_vs = 0;
 
-   machine->Consts = constants;
+   machine->Consts[0] = constants;
 
    for (i = 0; i < shader->info.num_inputs; ++i) {
       if (shader->info.input_semantic_name[i] != TGSI_SEMANTIC_PRIMID)
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 41cc802613..3f7a5ca14b 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -95,7 +95,7 @@ vs_exec_run_linear( struct draw_vertex_shader *shader,
    unsigned int i, j;
    unsigned slot;
 
-   machine->Consts = constants;
+   machine->Consts[0] = constants;
 
    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
       unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 83646b73c1..74e7e637cc 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -953,107 +953,90 @@ micro_sub(
 }
 
 static void
-fetch_src_file_channel(
-   const struct tgsi_exec_machine *mach,
-   const uint file,
-   const uint swizzle,
-   const union tgsi_exec_channel *index,
-   union tgsi_exec_channel *chan )
-{
-   switch( swizzle ) {
-   case TGSI_SWIZZLE_X:
-   case TGSI_SWIZZLE_Y:
-   case TGSI_SWIZZLE_Z:
-   case TGSI_SWIZZLE_W:
-      switch( file ) {
-      case TGSI_FILE_CONSTANT:
-         assert(mach->Consts);
-         if (index->i[0] < 0)
-            chan->f[0] = 0.0f;
-         else
-            chan->f[0] = mach->Consts[index->i[0]][swizzle];
-         if (index->i[1] < 0)
-            chan->f[1] = 0.0f;
-         else
-            chan->f[1] = mach->Consts[index->i[1]][swizzle];
-         if (index->i[2] < 0)
-            chan->f[2] = 0.0f;
-         else
-            chan->f[2] = mach->Consts[index->i[2]][swizzle];
-         if (index->i[3] < 0)
-            chan->f[3] = 0.0f;
-         else
-            chan->f[3] = mach->Consts[index->i[3]][swizzle];
-         break;
+fetch_src_file_channel(const struct tgsi_exec_machine *mach,
+                       const uint file,
+                       const uint swizzle,
+                       const union tgsi_exec_channel *index,
+                       const union tgsi_exec_channel *index2D,
+                       union tgsi_exec_channel *chan)
+{
+   uint i;
 
-      case TGSI_FILE_INPUT:
-      case TGSI_FILE_SYSTEM_VALUE:
-         chan->u[0] = mach->Inputs[index->i[0]].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Inputs[index->i[1]].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Inputs[index->i[2]].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Inputs[index->i[3]].xyzw[swizzle].u[3];
-         break;
+   switch (file) {
+   case TGSI_FILE_CONSTANT:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index2D->i[i] >= 0 && index2D->i[i] < PIPE_MAX_CONSTANT);
+         assert(mach->Consts[index2D->i[i]]);
 
-      case TGSI_FILE_TEMPORARY:
-         assert(index->i[0] < TGSI_EXEC_NUM_TEMPS);
-         chan->u[0] = mach->Temps[index->i[0]].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Temps[index->i[1]].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Temps[index->i[2]].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Temps[index->i[3]].xyzw[swizzle].u[3];
-         break;
+         if (index->i[i] < 0) {
+            chan->u[i] = 0;
+         } else {
+            const uint *p = (const uint *)mach->Consts[index2D->i[i]];
 
-      case TGSI_FILE_IMMEDIATE:
-         assert( index->i[0] < (int) mach->ImmLimit );
-         chan->f[0] = mach->Imms[index->i[0]][swizzle];
-         assert( index->i[1] < (int) mach->ImmLimit );
-         chan->f[1] = mach->Imms[index->i[1]][swizzle];
-         assert( index->i[2] < (int) mach->ImmLimit );
-         chan->f[2] = mach->Imms[index->i[2]][swizzle];
-         assert( index->i[3] < (int) mach->ImmLimit );
-         chan->f[3] = mach->Imms[index->i[3]][swizzle];
-         break;
+            chan->u[i] = p[index->i[i] * 4 + swizzle];
+         }
+      }
+      break;
 
-      case TGSI_FILE_ADDRESS:
-         chan->u[0] = mach->Addrs[index->i[0]].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Addrs[index->i[1]].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Addrs[index->i[2]].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Addrs[index->i[3]].xyzw[swizzle].u[3];
-         break;
+   case TGSI_FILE_INPUT:
+   case TGSI_FILE_SYSTEM_VALUE:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         /* XXX: 2D indexing */
+         chan->u[i] = mach->Inputs[index2D->i[i] * TGSI_EXEC_MAX_INPUT_ATTRIBS + index->i[i]].xyzw[swizzle].u[i];
+      }
+      break;
 
-      case TGSI_FILE_PREDICATE:
-         assert(index->i[0] < TGSI_EXEC_NUM_PREDS);
-         assert(index->i[1] < TGSI_EXEC_NUM_PREDS);
-         assert(index->i[2] < TGSI_EXEC_NUM_PREDS);
-         assert(index->i[3] < TGSI_EXEC_NUM_PREDS);
-         chan->u[0] = mach->Predicates[0].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Predicates[0].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Predicates[0].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Predicates[0].xyzw[swizzle].u[3];
-         break;
+   case TGSI_FILE_TEMPORARY:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index->i[i] < TGSI_EXEC_NUM_TEMPS);
+         assert(index2D->i[i] == 0);
 
-      case TGSI_FILE_OUTPUT:
-         /* vertex/fragment output vars can be read too */
-         chan->u[0] = mach->Outputs[index->i[0]].xyzw[swizzle].u[0];
-         chan->u[1] = mach->Outputs[index->i[1]].xyzw[swizzle].u[1];
-         chan->u[2] = mach->Outputs[index->i[2]].xyzw[swizzle].u[2];
-         chan->u[3] = mach->Outputs[index->i[3]].xyzw[swizzle].u[3];
-         break;
+         chan->u[i] = mach->Temps[index->i[i]].xyzw[swizzle].u[i];
+      }
+      break;
 
-      default:
-         assert( 0 );
-         chan->u[0] = 0;
-         chan->u[1] = 0;
-         chan->u[2] = 0;
-         chan->u[3] = 0;
+   case TGSI_FILE_IMMEDIATE:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index->i[i] >= 0 && index->i[i] < (int)mach->ImmLimit);
+         assert(index2D->i[i] == 0);
+
+         chan->f[i] = mach->Imms[index->i[i]][swizzle];
+      }
+      break;
+
+   case TGSI_FILE_ADDRESS:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index->i[i] >= 0);
+         assert(index2D->i[i] == 0);
+
+         chan->u[i] = mach->Addrs[index->i[i]].xyzw[swizzle].u[i];
+      }
+      break;
+
+   case TGSI_FILE_PREDICATE:
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index->i[i] >= 0 && index->i[i] < TGSI_EXEC_NUM_PREDS);
+         assert(index2D->i[i] == 0);
+
+         chan->u[i] = mach->Predicates[0].xyzw[swizzle].u[i];
+      }
+      break;
+
+   case TGSI_FILE_OUTPUT:
+      /* vertex/fragment output vars can be read too */
+      for (i = 0; i < QUAD_SIZE; i++) {
+         assert(index->i[i] >= 0);
+         assert(index2D->i[i] == 0);
+
+         chan->u[i] = mach->Outputs[index->i[i]].xyzw[swizzle].u[i];
       }
       break;
 
    default:
-      assert( 0 );
-      chan->u[0] = 0;
-      chan->u[1] = 0;
-      chan->u[2] = 0;
-      chan->u[3] = 0;
+      assert(0);
+      for (i = 0; i < QUAD_SIZE; i++) {
+         chan->u[i] = 0;
+      }
    }
 }
 
@@ -1065,6 +1048,7 @@ fetch_source(const struct tgsi_exec_machine *mach,
              enum tgsi_exec_datatype src_datatype)
 {
    union tgsi_exec_channel index;
+   union tgsi_exec_channel index2D;
    uint swizzle;
 
    /* We start with a direct index into a register file.
@@ -1103,12 +1087,12 @@ fetch_source(const struct tgsi_exec_machine *mach,
 
       /* get current value of address register[swizzle] */
       swizzle = tgsi_util_get_src_register_swizzle( &reg->Indirect, CHAN_X );
-      fetch_src_file_channel(
-         mach,
-         reg->Indirect.File,
-         swizzle,
-         &index2,
-         &indir_index );
+      fetch_src_file_channel(mach,
+                             reg->Indirect.File,
+                             swizzle,
+                             &index2,
+                             &ZeroVec,
+                             &indir_index);
 
       /* add value of address register to the offset */
       index.i[0] += indir_index.i[0];
@@ -1129,35 +1113,15 @@ fetch_source(const struct tgsi_exec_machine *mach,
     * subscript to a register file. Effectively it means that
     * the register file is actually a 2D array of registers.
     *
-    *    file[3][1] == file[3*sizeof(file[1])+1],
+    *    file[3][1],
     *    where:
     *       [3] = Dimension.Index
     */
    if (reg->Register.Dimension) {
-      int array_size;
-      union tgsi_exec_channel dim_index;
-
-      /* The size of the first-order array depends on the register file type.
-       * We need to multiply the index to the first array to get an effective,
-       * "flat" index that points to the beginning of the second-order array.
-       */
-      switch (reg->Register.File) {
-      case TGSI_FILE_INPUT:
-      case TGSI_FILE_SYSTEM_VALUE:
-         array_size = TGSI_EXEC_MAX_INPUT_ATTRIBS;
-         break;
-      case TGSI_FILE_CONSTANT:
-         array_size = TGSI_EXEC_MAX_CONST_BUFFER;
-         break;
-      default:
-         assert( 0 );
-         array_size = 0;
-      }
-
-      dim_index.i[0] =
-      dim_index.i[1] =
-      dim_index.i[2] =
-      dim_index.i[3] = reg->Dimension.Index;
+      index2D.i[0] =
+      index2D.i[1] =
+      index2D.i[2] =
+      index2D.i[3] = reg->Dimension.Index;
 
       /* Again, the second subscript index can be addressed indirectly
        * identically to the first one.
@@ -1182,45 +1146,46 @@ fetch_source(const struct tgsi_exec_machine *mach,
          index2.i[3] = reg->DimIndirect.Index;
 
          swizzle = tgsi_util_get_src_register_swizzle( &reg->DimIndirect, CHAN_X );
-         fetch_src_file_channel(
-            mach,
-            reg->DimIndirect.File,
-            swizzle,
-            &index2,
-            &indir_index );
-
-         dim_index.i[0] += indir_index.i[0];
-         dim_index.i[1] += indir_index.i[1];
-         dim_index.i[2] += indir_index.i[2];
-         dim_index.i[3] += indir_index.i[3];
+         fetch_src_file_channel(mach,
+                                reg->DimIndirect.File,
+                                swizzle,
+                                &index2,
+                                &ZeroVec,
+                                &indir_index);
+
+         index2D.i[0] += indir_index.i[0];
+         index2D.i[1] += indir_index.i[1];
+         index2D.i[2] += indir_index.i[2];
+         index2D.i[3] += indir_index.i[3];
 
          /* for disabled execution channels, zero-out the index to
           * avoid using a potential garbage value.
           */
          for (i = 0; i < QUAD_SIZE; i++) {
-            if ((execmask & (1 << i)) == 0)
-               dim_index.i[i] = 0;
+            if ((execmask & (1 << i)) == 0) {
+               index2D.i[i] = 0;
+            }
          }
       }
 
-      index.i[0] += dim_index.i[0] * array_size;
-      index.i[1] += dim_index.i[1] * array_size;
-      index.i[2] += dim_index.i[2] * array_size;
-      index.i[3] += dim_index.i[3] * array_size;
-
       /* If by any chance there was a need for a 3D array of register
        * files, we would have to check whether Dimension is followed
        * by a dimension register and continue the saga.
        */
+   } else {
+      index2D.i[0] =
+      index2D.i[1] =
+      index2D.i[2] =
+      index2D.i[3] = 0;
    }
 
    swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index );
-   fetch_src_file_channel(
-      mach,
-      reg->Register.File,
-      swizzle,
-      &index,
-      chan );
+   fetch_src_file_channel(mach,
+                          reg->Register.File,
+                          swizzle,
+                          &index,
+                          &index2D,
+                          chan);
 
    if (reg->Register.Absolute) {
       if (src_datatype == TGSI_EXEC_DATA_FLOAT) {
@@ -1283,12 +1248,12 @@ store_dest(struct tgsi_exec_machine *mach,
       swizzle = tgsi_util_get_src_register_swizzle( &reg->Indirect, CHAN_X );
 
       /* fetch values from the address/indirection register */
-      fetch_src_file_channel(
-         mach,
-         reg->Indirect.File,
-         swizzle,
-         &index,
-         &indir_index );
+      fetch_src_file_channel(mach,
+                             reg->Indirect.File,
+                             swizzle,
+                             &index,
+                             &ZeroVec,
+                             &indir_index);
 
       /* save indirection offset */
       offset = indir_index.i[0];
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index 59e3b445cc..1f3e85fb62 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -260,7 +260,7 @@ struct tgsi_exec_machine
    struct tgsi_sampler           **Samplers;
 
    unsigned                      ImmLimit;
-   const float                   (*Consts)[4];
+   const void *Consts[PIPE_MAX_CONSTANT];
    const struct tgsi_token       *Tokens;   /**< Declarations, instructions */
    unsigned                      Processor; /**< TGSI_PROCESSOR_x */
 
diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index 73e075f0d8..a3a7825aa1 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -111,9 +111,13 @@ softpipe_destroy( struct pipe_context *pipe )
       pipe_texture_reference(&softpipe->vertex_textures[i], NULL);
    }
 
-   for (i = 0; i < Elements(softpipe->constants); i++) {
-      if (softpipe->constants[i]) {
-         pipe_buffer_reference(&softpipe->constants[i], NULL);
+   for (i = 0; i < PIPE_SHADER_TYPES; i++) {
+      uint j;
+
+      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+         if (softpipe->constants[i][j]) {
+            pipe_buffer_reference(&softpipe->constants[i][j], NULL);
+         }
       }
    }
 
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index da673c57ad..f19b3cd5be 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -63,7 +63,7 @@ struct softpipe_context {
    /** Other rendering state */
    struct pipe_blend_color blend_color;
    struct pipe_clip_state clip;
-   struct pipe_buffer *constants[PIPE_SHADER_TYPES];
+   struct pipe_buffer *constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT];
    struct pipe_framebuffer_state framebuffer;
    struct pipe_poly_stipple poly_stipple;
    struct pipe_scissor_state scissor;
@@ -92,7 +92,7 @@ struct softpipe_context {
    ubyte *mapped_vbuffer[PIPE_MAX_ATTRIBS];
 
    /** Mapped constant buffers */
-   void *mapped_constants[PIPE_SHADER_TYPES];
+   void *mapped_constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT];
 
    /** Vertex format */
    struct vertex_info vertex_info;
diff --git a/src/gallium/drivers/softpipe/sp_draw_arrays.c b/src/gallium/drivers/softpipe/sp_draw_arrays.c
index 03b58d2fb7..cbb9631fa5 100644
--- a/src/gallium/drivers/softpipe/sp_draw_arrays.c
+++ b/src/gallium/drivers/softpipe/sp_draw_arrays.c
@@ -52,26 +52,32 @@ softpipe_map_constant_buffers(struct softpipe_context *sp)
    uint i, vssize, gssize;
 
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
-      if (sp->constants[i] && sp->constants[i]->size)
-         sp->mapped_constants[i] = ws->buffer_map(ws, sp->constants[i],
-                                                  PIPE_BUFFER_USAGE_CPU_READ);
+      uint j;
+
+      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+         if (sp->constants[i][j] && sp->constants[i][j]->size) {
+            sp->mapped_constants[i][j] = ws->buffer_map(ws,
+                                                        sp->constants[i][j],
+                                                        PIPE_BUFFER_USAGE_CPU_READ);
+         }
+      }
    }
 
-   if (sp->constants[PIPE_SHADER_VERTEX])
-      vssize = sp->constants[PIPE_SHADER_VERTEX]->size;
+   if (sp->constants[PIPE_SHADER_VERTEX][0])
+      vssize = sp->constants[PIPE_SHADER_VERTEX][0]->size;
    else
       vssize = 0;
 
-   if (sp->constants[PIPE_SHADER_GEOMETRY])
-      gssize = sp->constants[PIPE_SHADER_GEOMETRY]->size;
+   if (sp->constants[PIPE_SHADER_GEOMETRY][0])
+      gssize = sp->constants[PIPE_SHADER_GEOMETRY][0]->size;
    else
       gssize = 0;
 
    draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX,
-                                   sp->mapped_constants[PIPE_SHADER_VERTEX],
+                                   sp->mapped_constants[PIPE_SHADER_VERTEX][0],
                                    vssize);
    draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_GEOMETRY,
-                                   sp->mapped_constants[PIPE_SHADER_GEOMETRY],
+                                   sp->mapped_constants[PIPE_SHADER_GEOMETRY][0],
                                    gssize);
 }
 
@@ -91,9 +97,14 @@ softpipe_unmap_constant_buffers(struct softpipe_context *sp)
    draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_GEOMETRY, NULL, 0);
 
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
-      if (sp->constants[i] && sp->constants[i]->size)
-         ws->buffer_unmap(ws, sp->constants[i]);
-      sp->mapped_constants[i] = NULL;
+      uint j;
+
+      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+         if (sp->constants[i][j] && sp->constants[i][j]->size) {
+            ws->buffer_unmap(ws, sp->constants[i][j]);
+         }
+         sp->mapped_constants[i][j] = NULL;
+      }
    }
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c
index f912950658..acee213670 100644
--- a/src/gallium/drivers/softpipe/sp_fs_sse.c
+++ b/src/gallium/drivers/softpipe/sp_fs_sse.c
@@ -135,7 +135,7 @@ fs_sse_run( const struct sp_fragment_shader *base,
    tgsi_set_exec_mask(machine, 1, 1, 1, 1);
 
    shader->func( machine,
-		 machine->Consts,
+                 (const float (*)[4])machine->Consts[0],
                  (const float (*)[4])shader->immediates,
 		 machine->InterpCoefs
 		 /*, &machine->QuadPos*/
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index e799df136e..9c497073c2 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -107,10 +107,11 @@ shade_quads(struct quad_stage *qs,
    struct quad_shade_stage *qss = quad_shade_stage( qs );
    struct softpipe_context *softpipe = qs->softpipe;
    struct tgsi_exec_machine *machine = qss->machine;
-
    unsigned i, pass = 0;
-   
-   machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];
+
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      machine->Consts[i] = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT][i];
+   }
    machine->InterpCoefs = quads[0]->coef;
 
    for (i = 0; i < nr; i++) {
diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c
index b7ed4441b4..50ed51661a 100644
--- a/src/gallium/drivers/softpipe/sp_state_fs.c
+++ b/src/gallium/drivers/softpipe/sp_state_fs.c
@@ -164,12 +164,12 @@ softpipe_set_constant_buffer(struct pipe_context *pipe,
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
    assert(shader < PIPE_SHADER_TYPES);
-   assert(index == 0);
+   assert(index < PIPE_MAX_CONSTANT);
 
    draw_flush(softpipe->draw);
 
    /* note: reference counting */
-   pipe_buffer_reference(&softpipe->constants[shader], buf);
+   pipe_buffer_reference(&softpipe->constants[shader][index], buf);
 
    softpipe->dirty |= SP_NEW_CONSTANTS;
 }
-- 
cgit v1.2.3


From 9851644435f991a1a1bbb145333a97601627b37d Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@vmware.com>
Date: Mon, 25 Jan 2010 12:36:50 +0100
Subject: gallium: Enable multiple constant buffers for vertex and geometry
 shaders.

---
 src/gallium/auxiliary/draw/draw_context.c          | 11 +++--
 src/gallium/auxiliary/draw/draw_context.h          | 10 +++--
 src/gallium/auxiliary/draw/draw_gs.c               | 14 ++++---
 src/gallium/auxiliary/draw/draw_gs.h               |  2 +-
 src/gallium/auxiliary/draw/draw_private.h          | 28 ++++++++-----
 .../auxiliary/draw/draw_pt_fetch_shade_pipeline.c  | 12 +++---
 src/gallium/auxiliary/draw/draw_vs.c               | 40 ++++++++++--------
 src/gallium/auxiliary/draw/draw_vs.h               |  8 ++--
 src/gallium/auxiliary/draw/draw_vs_aos.c           | 10 ++++-
 src/gallium/auxiliary/draw/draw_vs_aos.h           |  2 +-
 src/gallium/auxiliary/draw/draw_vs_aos_machine.c   | 14 ++++---
 src/gallium/auxiliary/draw/draw_vs_exec.c          |  6 ++-
 src/gallium/auxiliary/draw/draw_vs_llvm.c          |  5 ++-
 src/gallium/auxiliary/draw/draw_vs_ppc.c           |  4 +-
 src/gallium/auxiliary/draw/draw_vs_sse.c           |  4 +-
 src/gallium/auxiliary/draw/draw_vs_varient.c       |  4 +-
 src/gallium/drivers/cell/ppu/cell_draw_arrays.c    |  2 +-
 src/gallium/drivers/i915/i915_context.c            |  2 +-
 src/gallium/drivers/llvmpipe/lp_state_fs.c         |  2 +-
 src/gallium/drivers/nv04/nv04_vbo.c                |  2 +-
 src/gallium/drivers/nv10/nv10_vbo.c                |  1 +
 src/gallium/drivers/nv20/nv20_vbo.c                |  2 +-
 src/gallium/drivers/nv40/nv40_draw.c               |  2 +-
 src/gallium/drivers/r300/r300_render.c             |  2 +
 src/gallium/drivers/softpipe/sp_draw_arrays.c      | 48 +++++++++++++---------
 src/gallium/drivers/svga/svga_swtnl_draw.c         |  2 +-
 src/mesa/state_tracker/st_draw_feedback.c          |  2 +-
 27 files changed, 146 insertions(+), 95 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index 1639043832..a52b93e122 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -234,17 +234,20 @@ draw_set_mapped_vertex_buffer(struct draw_context *draw,
 void
 draw_set_mapped_constant_buffer(struct draw_context *draw,
                                 unsigned shader_type,
+                                unsigned slot,
                                 const void *buffer,
                                 unsigned size )
 {
    debug_assert(shader_type == PIPE_SHADER_VERTEX ||
                 shader_type == PIPE_SHADER_GEOMETRY);
+   debug_assert(slot < PIPE_MAX_CONSTANT);
+
    if (shader_type == PIPE_SHADER_VERTEX) {
-      draw->pt.user.vs_constants = buffer;
-      draw_vs_set_constants( draw, (const float (*)[4])buffer, size );
+      draw->pt.user.vs_constants[slot] = buffer;
+      draw_vs_set_constants(draw, slot, buffer, size);
    } else if (shader_type == PIPE_SHADER_GEOMETRY) {
-      draw->pt.user.gs_constants = buffer;
-      draw_gs_set_constants( draw, (const float (*)[4])buffer, size );
+      draw->pt.user.gs_constants[slot] = buffer;
+      draw_gs_set_constants(draw, slot, buffer, size);
    }
 }
 
diff --git a/src/gallium/auxiliary/draw/draw_context.h b/src/gallium/auxiliary/draw/draw_context.h
index 8a64c06efc..acd81b9712 100644
--- a/src/gallium/auxiliary/draw/draw_context.h
+++ b/src/gallium/auxiliary/draw/draw_context.h
@@ -151,10 +151,12 @@ void draw_set_mapped_element_buffer( struct draw_context *draw,
 void draw_set_mapped_vertex_buffer(struct draw_context *draw,
                                    unsigned attr, const void *buffer);
 
-void draw_set_mapped_constant_buffer(struct draw_context *draw,
-                                     unsigned shader_type,
-                                     const void *buffer,
-                                     unsigned size );
+void
+draw_set_mapped_constant_buffer(struct draw_context *draw,
+                                unsigned shader_type,
+                                unsigned slot,
+                                const void *buffer,
+                                unsigned size);
 
 
 /***********************************************************************
diff --git a/src/gallium/auxiliary/draw/draw_gs.c b/src/gallium/auxiliary/draw/draw_gs.c
index c3cc365a8d..96be5dba0b 100644
--- a/src/gallium/auxiliary/draw/draw_gs.c
+++ b/src/gallium/auxiliary/draw/draw_gs.c
@@ -69,9 +69,11 @@ void draw_gs_destroy( struct draw_context *draw )
    tgsi_exec_machine_destroy(draw->gs.machine);
 }
 
-void draw_gs_set_constants( struct draw_context *draw,
-                            const float (*constants)[4],
-                            unsigned size )
+void
+draw_gs_set_constants(struct draw_context *draw,
+                      unsigned slot,
+                      const void *constants,
+                      unsigned size)
 {
 }
 
@@ -291,7 +293,7 @@ draw_geometry_fetch_outputs(struct draw_geometry_shader *shader,
 void draw_geometry_shader_run(struct draw_geometry_shader *shader,
                               const float (*input)[4],
                               float (*output)[4],
-                              const float (*constants)[4],
+                              const void *constants[PIPE_MAX_CONSTANT],
                               unsigned count,
                               unsigned input_stride,
                               unsigned vertex_size)
@@ -302,7 +304,9 @@ void draw_geometry_shader_run(struct draw_geometry_shader *shader,
    unsigned num_primitives = count/num_vertices;
    unsigned inputs_from_vs = 0;
 
-   machine->Consts[0] = constants;
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      machine->Consts[i] = constants[i];
+   }
 
    for (i = 0; i < shader->info.num_inputs; ++i) {
       if (shader->info.input_semantic_name[i] != TGSI_SEMANTIC_PRIMID)
diff --git a/src/gallium/auxiliary/draw/draw_gs.h b/src/gallium/auxiliary/draw/draw_gs.h
index d6a97d9c4e..e055c8a05c 100644
--- a/src/gallium/auxiliary/draw/draw_gs.h
+++ b/src/gallium/auxiliary/draw/draw_gs.h
@@ -62,7 +62,7 @@ struct draw_geometry_shader {
 void draw_geometry_shader_run(struct draw_geometry_shader *shader,
                               const float (*input)[4],
                               float (*output)[4],
-                              const float (*constants)[4],
+                              const void *constants[PIPE_MAX_CONSTANT],
                               unsigned count,
                               unsigned input_stride,
                               unsigned output_stride);
diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h
index ef49e57536..ecb17cbbe2 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gallium/auxiliary/draw/draw_private.h
@@ -153,8 +153,8 @@ struct draw_context
          const void *vbuffer[PIPE_MAX_ATTRIBS];
          
          /** constant buffer (for vertex/geometry shader) */
-         const void *vs_constants;
-         const void *gs_constants;
+         const void *vs_constants[PIPE_MAX_CONSTANT];
+         const void *gs_constants[PIPE_MAX_CONSTANT];
       } user;
 
       boolean test_fse;         /* enable FSE even though its not correct (eg for softpipe) */
@@ -202,10 +202,10 @@ struct draw_context
       struct aos_machine *aos_machine; 
 
 
-      const float (*aligned_constants)[4];
+      const void *aligned_constants[PIPE_MAX_CONSTANT];
 
-      const float (*aligned_constant_storage)[4];
-      unsigned const_storage_size;
+      const void *aligned_constant_storage[PIPE_MAX_CONSTANT];
+      unsigned const_storage_size[PIPE_MAX_CONSTANT];
 
 
       struct translate *fetch;
@@ -256,9 +256,11 @@ void draw_vs_destroy( struct draw_context *draw );
 void draw_vs_set_viewport( struct draw_context *, 
                            const struct pipe_viewport_state * );
 
-void draw_vs_set_constants( struct draw_context *,
-                            const float (*constants)[4],
-                            unsigned size );
+void
+draw_vs_set_constants(struct draw_context *,
+                      unsigned slot,
+                      const void *constants,
+                      unsigned size);
 
 
 
@@ -266,9 +268,13 @@ void draw_vs_set_constants( struct draw_context *,
  * Geometry shading code:
  */
 boolean draw_gs_init( struct draw_context *draw );
-void draw_gs_set_constants( struct draw_context *,
-                            const float (*constants)[4],
-                            unsigned size );
+
+void
+draw_gs_set_constants(struct draw_context *,
+                      unsigned slot,
+                      const void *constants,
+                      unsigned size);
+
 void draw_gs_destroy( struct draw_context *draw );
 
 /*******************************************************************************
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
index 2c996e7308..56b69354b2 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
@@ -163,7 +163,7 @@ static void fetch_pipeline_run( struct draw_pt_middle_end *middle,
       vshader->run_linear(vshader,
                           (const float (*)[4])pipeline_verts->data,
                           (      float (*)[4])pipeline_verts->data,
-                          (const float (*)[4])draw->pt.user.vs_constants,
+                          draw->pt.user.vs_constants,
                           fetch_count,
                           fpme->vertex_size,
                           fpme->vertex_size);
@@ -171,7 +171,7 @@ static void fetch_pipeline_run( struct draw_pt_middle_end *middle,
          draw_geometry_shader_run(gshader,
                                   (const float (*)[4])pipeline_verts->data,
                                   (      float (*)[4])pipeline_verts->data,
-                                  (const float (*)[4])draw->pt.user.gs_constants,
+                                  draw->pt.user.gs_constants,
                                   fetch_count,
                                   fpme->vertex_size,
                                   fpme->vertex_size);
@@ -248,7 +248,7 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
       shader->run_linear(shader,
 			 (const float (*)[4])pipeline_verts->data,
 			 (      float (*)[4])pipeline_verts->data,
-			 (const float (*)[4])draw->pt.user.vs_constants,
+                         draw->pt.user.vs_constants,
 			 count,
 			 fpme->vertex_size,
 			 fpme->vertex_size);
@@ -257,7 +257,7 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
          draw_geometry_shader_run(geometry_shader,
                                   (const float (*)[4])pipeline_verts->data,
                                   (      float (*)[4])pipeline_verts->data,
-                                  (const float (*)[4])draw->pt.user.gs_constants,
+                                  draw->pt.user.gs_constants,
                                   count,
                                   fpme->vertex_size,
                                   fpme->vertex_size);
@@ -328,7 +328,7 @@ static boolean fetch_pipeline_linear_run_elts( struct draw_pt_middle_end *middle
       shader->run_linear(shader,
 			 (const float (*)[4])pipeline_verts->data,
 			 (      float (*)[4])pipeline_verts->data,
-			 (const float (*)[4])draw->pt.user.vs_constants,
+                         draw->pt.user.vs_constants,
 			 count,
 			 fpme->vertex_size,
 			 fpme->vertex_size);
@@ -337,7 +337,7 @@ static boolean fetch_pipeline_linear_run_elts( struct draw_pt_middle_end *middle
          draw_geometry_shader_run(geometry_shader,
                                   (const float (*)[4])pipeline_verts->data,
                                   (      float (*)[4])pipeline_verts->data,
-                                  (const float (*)[4])draw->pt.user.gs_constants,
+                                  draw->pt.user.gs_constants,
                                   count,
                                   fpme->vertex_size,
                                   fpme->vertex_size);
diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c
index e03ac8c229..120b976d9f 100644
--- a/src/gallium/auxiliary/draw/draw_vs.c
+++ b/src/gallium/auxiliary/draw/draw_vs.c
@@ -48,24 +48,27 @@
 
 
 
-void draw_vs_set_constants( struct draw_context *draw,
-                            const float (*constants)[4],
-                            unsigned size )
+void
+draw_vs_set_constants(struct draw_context *draw,
+                      unsigned slot,
+                      const void *constants,
+                      unsigned size)
 {
    if (((uintptr_t)constants) & 0xf) {
-      if (size > draw->vs.const_storage_size) {
-         if (draw->vs.aligned_constant_storage)
-            align_free((void *)draw->vs.aligned_constant_storage);
-         draw->vs.aligned_constant_storage = align_malloc( size, 16 );
+      if (size > draw->vs.const_storage_size[slot]) {
+         if (draw->vs.aligned_constant_storage[slot]) {
+            align_free((void *)draw->vs.aligned_constant_storage[slot]);
+         }
+         draw->vs.aligned_constant_storage[slot] = align_malloc(size, 16);
       }
-      memcpy( (void*)draw->vs.aligned_constant_storage,
-              constants, 
-              size );
-      constants = draw->vs.aligned_constant_storage;
+      memcpy((void *)draw->vs.aligned_constant_storage[slot],
+             constants,
+             size);
+      constants = draw->vs.aligned_constant_storage[slot];
    }
-      
-   draw->vs.aligned_constants = constants;
-   draw_vs_aos_machine_constants( draw->vs.aos_machine, constants );
+
+   draw->vs.aligned_constants[slot] = constants;
+   draw_vs_aos_machine_constants(draw->vs.aos_machine, slot, constants);
 }
 
 
@@ -182,6 +185,8 @@ draw_vs_init( struct draw_context *draw )
 void
 draw_vs_destroy( struct draw_context *draw )
 {
+   uint i;
+
    if (draw->vs.fetch_cache)
       translate_cache_destroy(draw->vs.fetch_cache);
 
@@ -191,8 +196,11 @@ draw_vs_destroy( struct draw_context *draw )
    if (draw->vs.aos_machine)
       draw_vs_aos_machine_destroy(draw->vs.aos_machine);
 
-   if (draw->vs.aligned_constant_storage)
-      align_free((void*)draw->vs.aligned_constant_storage);
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      if (draw->vs.aligned_constant_storage[i]) {
+         align_free((void *)draw->vs.aligned_constant_storage[i]);
+      }
+   }
 
    tgsi_exec_machine_destroy(draw->vs.machine);
 }
diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h
index 00036cfe68..6bd3683d39 100644
--- a/src/gallium/auxiliary/draw/draw_vs.h
+++ b/src/gallium/auxiliary/draw/draw_vs.h
@@ -132,7 +132,7 @@ struct draw_vertex_shader {
    void (*run_linear)( struct draw_vertex_shader *shader,
 		       const float (*input)[4],
 		       float (*output)[4],
-		       const float (*constants)[4],
+                      const void *constants[PIPE_MAX_CONSTANT],
 		       unsigned count,
 		       unsigned input_stride,
 		       unsigned output_stride );
@@ -212,8 +212,10 @@ static INLINE int draw_vs_varient_key_compare( const struct draw_vs_varient_key
 struct aos_machine *draw_vs_aos_machine( void );
 void draw_vs_aos_machine_destroy( struct aos_machine *machine );
 
-void draw_vs_aos_machine_constants( struct aos_machine *machine,
-                                    const float (*constants)[4] );
+void
+draw_vs_aos_machine_constants(struct aos_machine *machine,
+                              unsigned slot,
+                              const void *constants);
 
 void draw_vs_aos_machine_viewport( struct aos_machine *machine,
                                    const struct pipe_viewport_state *viewport );
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index 1aaae4ab7a..4d73de413d 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -2114,11 +2114,14 @@ static void PIPE_CDECL vaos_run_elts( struct draw_vs_varient *varient,
 {
    struct draw_vs_varient_aos_sse *vaos = (struct draw_vs_varient_aos_sse *)varient;
    struct aos_machine *machine = vaos->draw->vs.aos_machine;
+   unsigned i;
 
    if (0) debug_printf("%s %d\n", __FUNCTION__, count);
 
    machine->internal[IMM_PSIZE][0] = vaos->draw->rasterizer->point_size;
-   machine->constants = vaos->draw->vs.aligned_constants;
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      machine->constants[i] = vaos->draw->vs.aligned_constants[i];
+   }
    machine->immediates = vaos->base.vs->immediates;
    machine->buffer = vaos->buffer;
 
@@ -2135,12 +2138,15 @@ static void PIPE_CDECL vaos_run_linear( struct draw_vs_varient *varient,
 {
    struct draw_vs_varient_aos_sse *vaos = (struct draw_vs_varient_aos_sse *)varient;
    struct aos_machine *machine = vaos->draw->vs.aos_machine;
+   unsigned i;
 
    if (0) debug_printf("%s %d %d const: %x\n", __FUNCTION__, start, count, 
                        vaos->base.key.const_vbuffers);
 
    machine->internal[IMM_PSIZE][0] = vaos->draw->rasterizer->point_size;
-   machine->constants = vaos->draw->vs.aligned_constants;
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      machine->constants[i] = vaos->draw->vs.aligned_constants[i];
+   }
    machine->immediates = vaos->base.vs->immediates;
    machine->buffer = vaos->buffer;
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.h b/src/gallium/auxiliary/draw/draw_vs_aos.h
index 2cf72ddf7b..eefc097464 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.h
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.h
@@ -122,7 +122,7 @@ struct aos_machine {
    ushort fpucntl;              /* one of FPU_* above */
 
    const float (*immediates)[4];     /* points to shader data */
-   const float (*constants)[4];      /* points to draw data */
+   const void *constants[PIPE_MAX_CONSTANT]; /* points to draw data */
 
    const struct aos_buffer *buffer; /* points to ? */
 };
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos_machine.c b/src/gallium/auxiliary/draw/draw_vs_aos_machine.c
index 3240e3745d..0eda414ee6 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos_machine.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos_machine.c
@@ -219,10 +219,12 @@ static void PIPE_CDECL populate_lut( struct aos_machine *machine,
 }
 
 
-void draw_vs_aos_machine_constants( struct aos_machine *machine,
-                                    const float (*constants)[4] )
+void
+draw_vs_aos_machine_constants(struct aos_machine *machine,
+                              unsigned slot,
+                              const void *constants)
 {
-   machine->constants = constants;
+   machine->constants[slot] = constants;
 
    {
       unsigned i;
@@ -307,8 +309,10 @@ void draw_vs_aos_machine_viewport( struct aos_machine *machine,
 {
 }
 
-void draw_vs_aos_machine_constants( struct aos_machine *machine,
-                                    const float (*constants)[4] )
+void
+draw_vs_aos_machine_constants(struct aos_machine *machine,
+                              unsigned slot,
+                              const void *constants)
 {
 }
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 3f7a5ca14b..652c844654 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -85,7 +85,7 @@ static void
 vs_exec_run_linear( struct draw_vertex_shader *shader,
 		    const float (*input)[4],
 		    float (*output)[4],
-		    const float (*constants)[4],
+                   const void *constants[PIPE_MAX_CONSTANT],
 		    unsigned count,
 		    unsigned input_stride,
 		    unsigned output_stride )
@@ -95,7 +95,9 @@ vs_exec_run_linear( struct draw_vertex_shader *shader,
    unsigned int i, j;
    unsigned slot;
 
-   machine->Consts[0] = constants;
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      machine->Consts[i] = constants[i];
+   }
 
    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
       unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c
index b3535c0e48..e8e72f06ec 100644
--- a/src/gallium/auxiliary/draw/draw_vs_llvm.c
+++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c
@@ -64,7 +64,7 @@ static void
 vs_llvm_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-		   const float (*constants)[4],
+                   const void *constants[PIPE_MAX_CONSTANT],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
@@ -74,7 +74,8 @@ vs_llvm_run_linear( struct draw_vertex_shader *base,
 
    gallivm_cpu_vs_exec(shader->llvm_prog, shader->machine,
                        input, base->info.num_inputs, output, base->info.num_outputs,
-                       constants, count, input_stride, output_stride);
+                       (const float (*)[4])constants[0],
+                       count, input_stride, output_stride);
 }
 
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c
index da9f3e3d35..9952b40fcc 100644
--- a/src/gallium/auxiliary/draw/draw_vs_ppc.c
+++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c
@@ -85,7 +85,7 @@ static void
 vs_ppc_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-		   const float (*constants)[4],
+                  const void *constants[PIPE_MAX_CONSTANT],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
@@ -125,7 +125,7 @@ vs_ppc_run_linear( struct draw_vertex_shader *base,
        */
       shader->func(inputs_soa, outputs_soa, temps_soa,
 		   (float (*)[4]) shader->base.immediates,
-		   (float (*)[4]) constants,
+                   (const float (*)[4])constants[0],
                    ppc_builtin_constants);
 
       /* convert (up to) four output verts from SoA back to AoS format */
diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c
index 702051387a..297013dc74 100644
--- a/src/gallium/auxiliary/draw/draw_vs_sse.c
+++ b/src/gallium/auxiliary/draw/draw_vs_sse.c
@@ -83,7 +83,7 @@ static void
 vs_sse_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-		   const float (*constants)[4],
+                  const void *constants[PIPE_MAX_CONSTANT],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
@@ -112,7 +112,7 @@ vs_sse_run_linear( struct draw_vertex_shader *base,
       /* run compiled shader
        */
       shader->func(machine,
-		   constants,
+                   (const float (*)[4])constants[0],
 		   shader->base.immediates,
                    input,
                    base->info.num_inputs,
diff --git a/src/gallium/auxiliary/draw/draw_vs_varient.c b/src/gallium/auxiliary/draw/draw_vs_varient.c
index b87a465f6b..5ed706cb4f 100644
--- a/src/gallium/auxiliary/draw/draw_vs_varient.c
+++ b/src/gallium/auxiliary/draw/draw_vs_varient.c
@@ -147,7 +147,7 @@ static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient,
    vsvg->base.vs->run_linear( vsvg->base.vs, 
                               temp_buffer,
                               temp_buffer,
-                              (const float (*)[4])vsvg->base.vs->draw->pt.user.vs_constants,
+                             vsvg->base.vs->draw->pt.user.vs_constants,
                               count,
                               temp_vertex_stride, 
                               temp_vertex_stride);
@@ -210,7 +210,7 @@ static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient,
    vsvg->base.vs->run_linear( vsvg->base.vs, 
                               temp_buffer,
                               temp_buffer,
-                              (const float (*)[4])vsvg->base.vs->draw->pt.user.vs_constants,
+                             vsvg->base.vs->draw->pt.user.vs_constants,
                               count,
                               temp_vertex_stride, 
                               temp_vertex_stride);
diff --git a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
index 0a4da8ecc8..c674d0be63 100644
--- a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
+++ b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c
@@ -59,7 +59,7 @@ cell_map_constant_buffers(struct cell_context *sp)
       }
    }
 
-   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX,
+   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX, 0,
                                    sp->mapped_constants[PIPE_SHADER_VERTEX],
                                    sp->constants[PIPE_SHADER_VERTEX]->size);
 }
diff --git a/src/gallium/drivers/i915/i915_context.c b/src/gallium/drivers/i915/i915_context.c
index 4c5ff37ca8..a0c80d0228 100644
--- a/src/gallium/drivers/i915/i915_context.c
+++ b/src/gallium/drivers/i915/i915_context.c
@@ -81,7 +81,7 @@ i915_draw_range_elements(struct pipe_context *pipe,
    }
 
 
-   draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX,
+   draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX, 0,
                                    i915->current.constants[PIPE_SHADER_VERTEX],
                                    (i915->current.num_user_constants[PIPE_SHADER_VERTEX] * 
                                       4 * sizeof(float)));
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index 70e2a082cc..0cb335fe27 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -731,7 +731,7 @@ llvmpipe_set_constant_buffer(struct pipe_context *pipe,
    }
 
    if(shader == PIPE_SHADER_VERTEX) {
-      draw_set_mapped_constant_buffer(llvmpipe->draw, PIPE_SHADER_VERTEX,
+      draw_set_mapped_constant_buffer(llvmpipe->draw, PIPE_SHADER_VERTEX, 0,
                                       data, size);
    }
 
diff --git a/src/gallium/drivers/nv04/nv04_vbo.c b/src/gallium/drivers/nv04/nv04_vbo.c
index 3484771814..2db61d8941 100644
--- a/src/gallium/drivers/nv04/nv04_vbo.c
+++ b/src/gallium/drivers/nv04/nv04_vbo.c
@@ -45,7 +45,7 @@ void nv04_draw_elements( struct pipe_context *pipe,
 		draw_set_mapped_element_buffer(draw, 0, NULL);
 	}
 
-	draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX,
+	draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX, 0,
 					nv04->constbuf[PIPE_SHADER_VERTEX],
 					nv04->constbuf_nr[PIPE_SHADER_VERTEX]);
 
diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c
index 9180c72c9b..da5c93f081 100644
--- a/src/gallium/drivers/nv10/nv10_vbo.c
+++ b/src/gallium/drivers/nv10/nv10_vbo.c
@@ -46,6 +46,7 @@ void nv10_draw_elements( struct pipe_context *pipe,
 
 	draw_set_mapped_constant_buffer(draw,
                                         PIPE_SHADER_VERTEX,
+                                        0,
 					nv10->constbuf[PIPE_SHADER_VERTEX],
 					nv10->constbuf_nr[PIPE_SHADER_VERTEX]);
 
diff --git a/src/gallium/drivers/nv20/nv20_vbo.c b/src/gallium/drivers/nv20/nv20_vbo.c
index 52991a0d85..6dc9538901 100644
--- a/src/gallium/drivers/nv20/nv20_vbo.c
+++ b/src/gallium/drivers/nv20/nv20_vbo.c
@@ -45,7 +45,7 @@ void nv20_draw_elements( struct pipe_context *pipe,
 		draw_set_mapped_element_buffer(draw, 0, NULL);
 	}
 
-	draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX,
+	draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX, 0,
 					nv20->constbuf[PIPE_SHADER_VERTEX],
 					nv20->constbuf_nr[PIPE_SHADER_VERTEX]);
 
diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c
index d826f8c2f5..a2fd5f6c02 100644
--- a/src/gallium/drivers/nv40/nv40_draw.c
+++ b/src/gallium/drivers/nv40/nv40_draw.c
@@ -271,7 +271,7 @@ nv40_draw_elements_swtnl(struct pipe_context *pipe,
 		map = pipe_buffer_map(pscreen,
 				      nv40->constbuf[PIPE_SHADER_VERTEX],
 				      PIPE_BUFFER_USAGE_CPU_READ);
-		draw_set_mapped_constant_buffer(nv40->draw, PIPE_SHADER_VERTEX,
+		draw_set_mapped_constant_buffer(nv40->draw, PIPE_SHADER_VERTEX, 0,
                                                 map, nr);
 	}
 
diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c
index bcb3707149..95b7cb5fa4 100644
--- a/src/gallium/drivers/r300/r300_render.c
+++ b/src/gallium/drivers/r300/r300_render.c
@@ -462,6 +462,7 @@ void r300_swtcl_draw_arrays(struct pipe_context* pipe,
 
     draw_set_mapped_constant_buffer(r300->draw,
 				    PIPE_SHADER_VERTEX,
+                                    0,
 				    r300->shader_constants[PIPE_SHADER_VERTEX].constants,
 				    r300->shader_constants[PIPE_SHADER_VERTEX].count *
                 (sizeof(float) * 4));
@@ -506,6 +507,7 @@ void r300_swtcl_draw_range_elements(struct pipe_context* pipe,
 
     draw_set_mapped_constant_buffer(r300->draw,
 				    PIPE_SHADER_VERTEX,
+                                    0,
             r300->shader_constants[PIPE_SHADER_VERTEX].constants,
             r300->shader_constants[PIPE_SHADER_VERTEX].count *
                 (sizeof(float) * 4));
diff --git a/src/gallium/drivers/softpipe/sp_draw_arrays.c b/src/gallium/drivers/softpipe/sp_draw_arrays.c
index cbb9631fa5..06b97ab7c9 100644
--- a/src/gallium/drivers/softpipe/sp_draw_arrays.c
+++ b/src/gallium/drivers/softpipe/sp_draw_arrays.c
@@ -49,7 +49,7 @@ static void
 softpipe_map_constant_buffers(struct softpipe_context *sp)
 {
    struct pipe_winsys *ws = sp->pipe.winsys;
-   uint i, vssize, gssize;
+   uint i;
 
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
       uint j;
@@ -63,22 +63,22 @@ softpipe_map_constant_buffers(struct softpipe_context *sp)
       }
    }
 
-   if (sp->constants[PIPE_SHADER_VERTEX][0])
-      vssize = sp->constants[PIPE_SHADER_VERTEX][0]->size;
-   else
-      vssize = 0;
-
-   if (sp->constants[PIPE_SHADER_GEOMETRY][0])
-      gssize = sp->constants[PIPE_SHADER_GEOMETRY][0]->size;
-   else
-      gssize = 0;
-
-   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX,
-                                   sp->mapped_constants[PIPE_SHADER_VERTEX][0],
-                                   vssize);
-   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_GEOMETRY,
-                                   sp->mapped_constants[PIPE_SHADER_GEOMETRY][0],
-                                   gssize);
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      if (sp->constants[PIPE_SHADER_VERTEX][i]) {
+         draw_set_mapped_constant_buffer(sp->draw,
+                                         PIPE_SHADER_VERTEX,
+                                         i,
+                                         sp->mapped_constants[PIPE_SHADER_VERTEX][i],
+                                         sp->constants[PIPE_SHADER_VERTEX][i]->size);
+      }
+      if (sp->constants[PIPE_SHADER_GEOMETRY][i]) {
+         draw_set_mapped_constant_buffer(sp->draw,
+                                         PIPE_SHADER_GEOMETRY,
+                                         i,
+                                         sp->mapped_constants[PIPE_SHADER_GEOMETRY][i],
+                                         sp->constants[PIPE_SHADER_GEOMETRY][i]->size);
+      }
+   }
 }
 
 
@@ -93,8 +93,18 @@ softpipe_unmap_constant_buffers(struct softpipe_context *sp)
     */
    draw_flush(sp->draw);
 
-   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX, NULL, 0);
-   draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_GEOMETRY, NULL, 0);
+   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+      draw_set_mapped_constant_buffer(sp->draw,
+                                      PIPE_SHADER_VERTEX,
+                                      i,
+                                      NULL,
+                                      0);
+      draw_set_mapped_constant_buffer(sp->draw,
+                                      PIPE_SHADER_GEOMETRY,
+                                      i,
+                                      NULL,
+                                      0);
+   }
 
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
       uint j;
diff --git a/src/gallium/drivers/svga/svga_swtnl_draw.c b/src/gallium/drivers/svga/svga_swtnl_draw.c
index 0ae58c7d9d..5e6e30c7df 100644
--- a/src/gallium/drivers/svga/svga_swtnl_draw.c
+++ b/src/gallium/drivers/svga/svga_swtnl_draw.c
@@ -89,7 +89,7 @@ svga_swtnl_draw_range_elements(struct svga_context *svga,
                             PIPE_BUFFER_USAGE_CPU_READ);
       assert(map);
       draw_set_mapped_constant_buffer(
-         draw, PIPE_SHADER_VERTEX,
+         draw, PIPE_SHADER_VERTEX, 0,
          map,
          svga->curr.cb[PIPE_SHADER_VERTEX]->size);
    }
diff --git a/src/mesa/state_tracker/st_draw_feedback.c b/src/mesa/state_tracker/st_draw_feedback.c
index dd27a1e3a8..cdaee2a353 100644
--- a/src/mesa/state_tracker/st_draw_feedback.c
+++ b/src/mesa/state_tracker/st_draw_feedback.c
@@ -241,7 +241,7 @@ st_feedback_draw_vbo(GLcontext *ctx,
    mapped_constants = pipe_buffer_map(pipe->screen,
                                       st->state.constants[PIPE_SHADER_VERTEX],
                                       PIPE_BUFFER_USAGE_CPU_READ);
-   draw_set_mapped_constant_buffer(st->draw, PIPE_SHADER_VERTEX,
+   draw_set_mapped_constant_buffer(st->draw, PIPE_SHADER_VERTEX, 0,
                                    mapped_constants,
                                    st->state.constants[PIPE_SHADER_VERTEX]->size);
 
-- 
cgit v1.2.3


From 7c5f255201f42303188137f56ea8acc030444f0e Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@vmware.com>
Date: Mon, 25 Jan 2010 13:29:33 +0100
Subject: gallium: Rename PIPE_MAX_CONSTANT to PIPE_MAX_CONSTANT_BUFFERS.

---
 src/gallium/auxiliary/draw/draw_context.c     |  2 +-
 src/gallium/auxiliary/draw/draw_gs.c          |  4 ++--
 src/gallium/auxiliary/draw/draw_gs.h          |  2 +-
 src/gallium/auxiliary/draw/draw_private.h     | 10 +++++-----
 src/gallium/auxiliary/draw/draw_vs.c          |  2 +-
 src/gallium/auxiliary/draw/draw_vs.h          |  2 +-
 src/gallium/auxiliary/draw/draw_vs_aos.c      |  4 ++--
 src/gallium/auxiliary/draw/draw_vs_aos.h      |  2 +-
 src/gallium/auxiliary/draw/draw_vs_exec.c     |  4 ++--
 src/gallium/auxiliary/draw/draw_vs_llvm.c     |  2 +-
 src/gallium/auxiliary/draw/draw_vs_ppc.c      |  2 +-
 src/gallium/auxiliary/draw/draw_vs_sse.c      |  2 +-
 src/gallium/auxiliary/tgsi/tgsi_exec.c        |  2 +-
 src/gallium/auxiliary/tgsi/tgsi_exec.h        |  2 +-
 src/gallium/auxiliary/tgsi/tgsi_ureg.c        |  6 +++---
 src/gallium/drivers/softpipe/sp_context.c     |  2 +-
 src/gallium/drivers/softpipe/sp_context.h     |  4 ++--
 src/gallium/drivers/softpipe/sp_draw_arrays.c |  8 ++++----
 src/gallium/drivers/softpipe/sp_quad_fs.c     |  2 +-
 src/gallium/drivers/softpipe/sp_state_fs.c    |  2 +-
 src/gallium/include/pipe/p_state.h            |  2 +-
 21 files changed, 34 insertions(+), 34 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index a52b93e122..d3084fd428 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -240,7 +240,7 @@ draw_set_mapped_constant_buffer(struct draw_context *draw,
 {
    debug_assert(shader_type == PIPE_SHADER_VERTEX ||
                 shader_type == PIPE_SHADER_GEOMETRY);
-   debug_assert(slot < PIPE_MAX_CONSTANT);
+   debug_assert(slot < PIPE_MAX_CONSTANT_BUFFERS);
 
    if (shader_type == PIPE_SHADER_VERTEX) {
       draw->pt.user.vs_constants[slot] = buffer;
diff --git a/src/gallium/auxiliary/draw/draw_gs.c b/src/gallium/auxiliary/draw/draw_gs.c
index 96be5dba0b..7069aa6b18 100644
--- a/src/gallium/auxiliary/draw/draw_gs.c
+++ b/src/gallium/auxiliary/draw/draw_gs.c
@@ -293,7 +293,7 @@ draw_geometry_fetch_outputs(struct draw_geometry_shader *shader,
 void draw_geometry_shader_run(struct draw_geometry_shader *shader,
                               const float (*input)[4],
                               float (*output)[4],
-                              const void *constants[PIPE_MAX_CONSTANT],
+                              const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
                               unsigned count,
                               unsigned input_stride,
                               unsigned vertex_size)
@@ -304,7 +304,7 @@ void draw_geometry_shader_run(struct draw_geometry_shader *shader,
    unsigned num_primitives = count/num_vertices;
    unsigned inputs_from_vs = 0;
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       machine->Consts[i] = constants[i];
    }
 
diff --git a/src/gallium/auxiliary/draw/draw_gs.h b/src/gallium/auxiliary/draw/draw_gs.h
index e055c8a05c..d8eb210343 100644
--- a/src/gallium/auxiliary/draw/draw_gs.h
+++ b/src/gallium/auxiliary/draw/draw_gs.h
@@ -62,7 +62,7 @@ struct draw_geometry_shader {
 void draw_geometry_shader_run(struct draw_geometry_shader *shader,
                               const float (*input)[4],
                               float (*output)[4],
-                              const void *constants[PIPE_MAX_CONSTANT],
+                              const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
                               unsigned count,
                               unsigned input_stride,
                               unsigned output_stride);
diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h
index ecb17cbbe2..6a7190e975 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gallium/auxiliary/draw/draw_private.h
@@ -153,8 +153,8 @@ struct draw_context
          const void *vbuffer[PIPE_MAX_ATTRIBS];
          
          /** constant buffer (for vertex/geometry shader) */
-         const void *vs_constants[PIPE_MAX_CONSTANT];
-         const void *gs_constants[PIPE_MAX_CONSTANT];
+         const void *vs_constants[PIPE_MAX_CONSTANT_BUFFERS];
+         const void *gs_constants[PIPE_MAX_CONSTANT_BUFFERS];
       } user;
 
       boolean test_fse;         /* enable FSE even though its not correct (eg for softpipe) */
@@ -202,10 +202,10 @@ struct draw_context
       struct aos_machine *aos_machine; 
 
 
-      const void *aligned_constants[PIPE_MAX_CONSTANT];
+      const void *aligned_constants[PIPE_MAX_CONSTANT_BUFFERS];
 
-      const void *aligned_constant_storage[PIPE_MAX_CONSTANT];
-      unsigned const_storage_size[PIPE_MAX_CONSTANT];
+      const void *aligned_constant_storage[PIPE_MAX_CONSTANT_BUFFERS];
+      unsigned const_storage_size[PIPE_MAX_CONSTANT_BUFFERS];
 
 
       struct translate *fetch;
diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c
index 120b976d9f..6bdd612e6f 100644
--- a/src/gallium/auxiliary/draw/draw_vs.c
+++ b/src/gallium/auxiliary/draw/draw_vs.c
@@ -196,7 +196,7 @@ draw_vs_destroy( struct draw_context *draw )
    if (draw->vs.aos_machine)
       draw_vs_aos_machine_destroy(draw->vs.aos_machine);
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       if (draw->vs.aligned_constant_storage[i]) {
          align_free((void *)draw->vs.aligned_constant_storage[i]);
       }
diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h
index 6bd3683d39..d095c9bad1 100644
--- a/src/gallium/auxiliary/draw/draw_vs.h
+++ b/src/gallium/auxiliary/draw/draw_vs.h
@@ -132,7 +132,7 @@ struct draw_vertex_shader {
    void (*run_linear)( struct draw_vertex_shader *shader,
 		       const float (*input)[4],
 		       float (*output)[4],
-                      const void *constants[PIPE_MAX_CONSTANT],
+                      const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
 		       unsigned count,
 		       unsigned input_stride,
 		       unsigned output_stride );
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index 4d73de413d..e7121f3654 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -2119,7 +2119,7 @@ static void PIPE_CDECL vaos_run_elts( struct draw_vs_varient *varient,
    if (0) debug_printf("%s %d\n", __FUNCTION__, count);
 
    machine->internal[IMM_PSIZE][0] = vaos->draw->rasterizer->point_size;
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       machine->constants[i] = vaos->draw->vs.aligned_constants[i];
    }
    machine->immediates = vaos->base.vs->immediates;
@@ -2144,7 +2144,7 @@ static void PIPE_CDECL vaos_run_linear( struct draw_vs_varient *varient,
                        vaos->base.key.const_vbuffers);
 
    machine->internal[IMM_PSIZE][0] = vaos->draw->rasterizer->point_size;
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       machine->constants[i] = vaos->draw->vs.aligned_constants[i];
    }
    machine->immediates = vaos->base.vs->immediates;
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.h b/src/gallium/auxiliary/draw/draw_vs_aos.h
index eefc097464..1911242f82 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.h
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.h
@@ -122,7 +122,7 @@ struct aos_machine {
    ushort fpucntl;              /* one of FPU_* above */
 
    const float (*immediates)[4];     /* points to shader data */
-   const void *constants[PIPE_MAX_CONSTANT]; /* points to draw data */
+   const void *constants[PIPE_MAX_CONSTANT_BUFFERS]; /* points to draw data */
 
    const struct aos_buffer *buffer; /* points to ? */
 };
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 652c844654..7deca2b69d 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -85,7 +85,7 @@ static void
 vs_exec_run_linear( struct draw_vertex_shader *shader,
 		    const float (*input)[4],
 		    float (*output)[4],
-                   const void *constants[PIPE_MAX_CONSTANT],
+                   const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
 		    unsigned count,
 		    unsigned input_stride,
 		    unsigned output_stride )
@@ -95,7 +95,7 @@ vs_exec_run_linear( struct draw_vertex_shader *shader,
    unsigned int i, j;
    unsigned slot;
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       machine->Consts[i] = constants[i];
    }
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c
index e8e72f06ec..fd9166fda5 100644
--- a/src/gallium/auxiliary/draw/draw_vs_llvm.c
+++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c
@@ -64,7 +64,7 @@ static void
 vs_llvm_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-                   const void *constants[PIPE_MAX_CONSTANT],
+                   const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c
index 9952b40fcc..d869eecec5 100644
--- a/src/gallium/auxiliary/draw/draw_vs_ppc.c
+++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c
@@ -85,7 +85,7 @@ static void
 vs_ppc_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-                  const void *constants[PIPE_MAX_CONSTANT],
+                  const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c
index 297013dc74..54e6423388 100644
--- a/src/gallium/auxiliary/draw/draw_vs_sse.c
+++ b/src/gallium/auxiliary/draw/draw_vs_sse.c
@@ -83,7 +83,7 @@ static void
 vs_sse_run_linear( struct draw_vertex_shader *base,
 		   const float (*input)[4],
 		   float (*output)[4],
-                  const void *constants[PIPE_MAX_CONSTANT],
+                  const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
 		   unsigned count,
 		   unsigned input_stride,
 		   unsigned output_stride )
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 74e7e637cc..fbb9aa0e63 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -965,7 +965,7 @@ fetch_src_file_channel(const struct tgsi_exec_machine *mach,
    switch (file) {
    case TGSI_FILE_CONSTANT:
       for (i = 0; i < QUAD_SIZE; i++) {
-         assert(index2D->i[i] >= 0 && index2D->i[i] < PIPE_MAX_CONSTANT);
+         assert(index2D->i[i] >= 0 && index2D->i[i] < PIPE_MAX_CONSTANT_BUFFERS);
          assert(mach->Consts[index2D->i[i]]);
 
          if (index->i[i] < 0) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index 1f3e85fb62..a22873e4c2 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -260,7 +260,7 @@ struct tgsi_exec_machine
    struct tgsi_sampler           **Samplers;
 
    unsigned                      ImmLimit;
-   const void *Consts[PIPE_MAX_CONSTANT];
+   const void *Consts[PIPE_MAX_CONSTANT_BUFFERS];
    const struct tgsi_token       *Tokens;   /**< Declarations, instructions */
    unsigned                      Processor; /**< TGSI_PROCESSOR_x */
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
index c18db2e5b7..df0173928c 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -136,7 +136,7 @@ struct ureg_program
    unsigned temps_active[UREG_MAX_TEMP / 32];
    unsigned nr_temps;
 
-   struct const_decl const_decls[PIPE_MAX_CONSTANT];
+   struct const_decl const_decls[PIPE_MAX_CONSTANT_BUFFERS];
 
    unsigned property_gs_input_prim;
 
@@ -367,7 +367,7 @@ ureg_DECL_constant2D(struct ureg_program *ureg,
 {
    struct const_decl *decl = &ureg->const_decls[index2D];
 
-   assert(index2D < PIPE_MAX_CONSTANT);
+   assert(index2D < PIPE_MAX_CONSTANT_BUFFERS);
 
    if (decl->nr_constant_ranges < UREG_MAX_CONSTANT_RANGE) {
       uint i = decl->nr_constant_ranges++;
@@ -1209,7 +1209,7 @@ static void emit_decls( struct ureg_program *ureg )
                        ureg->sampler[i].Index, 1 );
    }
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       struct const_decl *decl = &ureg->const_decls[i];
 
       if (decl->nr_constant_ranges) {
diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index a3a7825aa1..53133d2888 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -114,7 +114,7 @@ softpipe_destroy( struct pipe_context *pipe )
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
       uint j;
 
-      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+      for (j = 0; j < PIPE_MAX_CONSTANT_BUFFERS; j++) {
          if (softpipe->constants[i][j]) {
             pipe_buffer_reference(&softpipe->constants[i][j], NULL);
          }
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index f19b3cd5be..be4613b622 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -63,7 +63,7 @@ struct softpipe_context {
    /** Other rendering state */
    struct pipe_blend_color blend_color;
    struct pipe_clip_state clip;
-   struct pipe_buffer *constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT];
+   struct pipe_buffer *constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT_BUFFERS];
    struct pipe_framebuffer_state framebuffer;
    struct pipe_poly_stipple poly_stipple;
    struct pipe_scissor_state scissor;
@@ -92,7 +92,7 @@ struct softpipe_context {
    ubyte *mapped_vbuffer[PIPE_MAX_ATTRIBS];
 
    /** Mapped constant buffers */
-   void *mapped_constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT];
+   void *mapped_constants[PIPE_SHADER_TYPES][PIPE_MAX_CONSTANT_BUFFERS];
 
    /** Vertex format */
    struct vertex_info vertex_info;
diff --git a/src/gallium/drivers/softpipe/sp_draw_arrays.c b/src/gallium/drivers/softpipe/sp_draw_arrays.c
index 06b97ab7c9..2b82427d54 100644
--- a/src/gallium/drivers/softpipe/sp_draw_arrays.c
+++ b/src/gallium/drivers/softpipe/sp_draw_arrays.c
@@ -54,7 +54,7 @@ softpipe_map_constant_buffers(struct softpipe_context *sp)
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
       uint j;
 
-      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+      for (j = 0; j < PIPE_MAX_CONSTANT_BUFFERS; j++) {
          if (sp->constants[i][j] && sp->constants[i][j]->size) {
             sp->mapped_constants[i][j] = ws->buffer_map(ws,
                                                         sp->constants[i][j],
@@ -63,7 +63,7 @@ softpipe_map_constant_buffers(struct softpipe_context *sp)
       }
    }
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       if (sp->constants[PIPE_SHADER_VERTEX][i]) {
          draw_set_mapped_constant_buffer(sp->draw,
                                          PIPE_SHADER_VERTEX,
@@ -93,7 +93,7 @@ softpipe_unmap_constant_buffers(struct softpipe_context *sp)
     */
    draw_flush(sp->draw);
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       draw_set_mapped_constant_buffer(sp->draw,
                                       PIPE_SHADER_VERTEX,
                                       i,
@@ -109,7 +109,7 @@ softpipe_unmap_constant_buffers(struct softpipe_context *sp)
    for (i = 0; i < PIPE_SHADER_TYPES; i++) {
       uint j;
 
-      for (j = 0; j < PIPE_MAX_CONSTANT; j++) {
+      for (j = 0; j < PIPE_MAX_CONSTANT_BUFFERS; j++) {
          if (sp->constants[i][j] && sp->constants[i][j]->size) {
             ws->buffer_unmap(ws, sp->constants[i][j]);
          }
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 9c497073c2..ad04dc2afc 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -109,7 +109,7 @@ shade_quads(struct quad_stage *qs,
    struct tgsi_exec_machine *machine = qss->machine;
    unsigned i, pass = 0;
 
-   for (i = 0; i < PIPE_MAX_CONSTANT; i++) {
+   for (i = 0; i < PIPE_MAX_CONSTANT_BUFFERS; i++) {
       machine->Consts[i] = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT][i];
    }
    machine->InterpCoefs = quads[0]->coef;
diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c
index 50ed51661a..04bdcaacc2 100644
--- a/src/gallium/drivers/softpipe/sp_state_fs.c
+++ b/src/gallium/drivers/softpipe/sp_state_fs.c
@@ -164,7 +164,7 @@ softpipe_set_constant_buffer(struct pipe_context *pipe,
    struct softpipe_context *softpipe = softpipe_context(pipe);
 
    assert(shader < PIPE_SHADER_TYPES);
-   assert(index < PIPE_MAX_CONSTANT);
+   assert(index < PIPE_MAX_CONSTANT_BUFFERS);
 
    draw_flush(softpipe->draw);
 
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index fdd29ed449..b250ed9440 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -58,7 +58,7 @@ extern "C" {
 #define PIPE_MAX_ATTRIBS          32
 #define PIPE_MAX_CLIP_PLANES       6
 #define PIPE_MAX_COLOR_BUFS        8
-#define PIPE_MAX_CONSTANT         32
+#define PIPE_MAX_CONSTANT_BUFFERS 32
 #define PIPE_MAX_SAMPLERS         16
 #define PIPE_MAX_VERTEX_SAMPLERS  16
 #define PIPE_MAX_SHADER_INPUTS    16
-- 
cgit v1.2.3


From 3320acb67dfa9330a80ab38e0c164f5d9c0a074a Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@vmware.com>
Date: Mon, 25 Jan 2010 14:31:32 +0100
Subject: softpipe: Report some sane values for constant buffer CAP queries.

---
 src/gallium/drivers/softpipe/sp_screen.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index bd3532de4f..69f40dc7b6 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -91,6 +91,10 @@ softpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_BLEND_EQUATION_SEPARATE:
       return 1;
+   case PIPE_CAP_MAX_CONST_BUFFERS:
+      return PIPE_MAX_CONSTANT_BUFFERS;
+   case PIPE_CAP_MAX_CONST_BUFFER_SIZE:
+      return 4096 * 4 * sizeof(float);
    default:
       return 0;
    }
-- 
cgit v1.2.3


From 00873227292df458b28249d7c3736d633e619dab Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Thu, 28 Jan 2010 13:39:13 -0800
Subject: r300g: Clean up blend masks, change caps.

progs/tests/drawbuffers should work, except I can't test it. GG new
shader parser. GG.
---
 src/gallium/drivers/r300/r300_emit.c   |  9 ++++++++-
 src/gallium/drivers/r300/r300_screen.c |  4 ++++
 src/gallium/drivers/r300/r300_state.c  | 18 ++++++------------
 3 files changed, 18 insertions(+), 13 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index f1d32764fd..9e813569e3 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -406,7 +406,14 @@ void r300_emit_fb_state(struct r300_context* r300, void* state)
         R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE);
 
     /* Set the number of colorbuffers. */
-    OUT_CS_REG(R300_RB3D_CCTL, R300_RB3D_CCTL_NUM_MULTIWRITES(fb->nr_cbufs));
+    if (fb->nr_cbufs > 1) {
+        OUT_CS_REG(R300_RB3D_CCTL,
+            R300_RB3D_CCTL_NUM_MULTIWRITES(fb->nr_cbufs) |
+            R300_RB3D_CCTL_INDEPENDENT_COLOR_CHANNEL_MASK_ENABLE |
+            R300_RB3D_CCTL_INDEPENDENT_COLORFORMAT_ENABLE_ENABLE);
+    } else {
+        OUT_CS_REG(R300_RB3D_CCTL, 0x0);
+    }
 
     /* Set up colorbuffers. */
     for (i = 0; i < fb->nr_cbufs; i++) {
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index b8fbf03d3a..679a5644af 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -113,6 +113,9 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
              * ~ C.
              */
             return 1;
+        case PIPE_CAP_DUAL_SOURCE_BLEND:
+            /* XXX need to figure out how to do it */
+            return 0;
         case PIPE_CAP_ANISOTROPIC_FILTER:
             return 1;
         case PIPE_CAP_POINT_SPRITE:
@@ -150,6 +153,7 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
                 return 0;
             }
         case PIPE_CAP_INDEP_BLEND_ENABLE:
+            return 1;
         case PIPE_CAP_INDEP_BLEND_FUNC:
             return 0;
         default:
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 3cdcf94274..8295799553 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -288,18 +288,12 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
                 (state->logicop_func) << R300_RB3D_ROPCNTL_ROP_SHIFT;
     }
 
-    /* Color Channel Mask */
-    if (state->rt[0].colormask & PIPE_MASK_R) {
-        blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_RED_MASK0;
-    }
-    if (state->rt[0].colormask & PIPE_MASK_G) {
-        blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0;
-    }
-    if (state->rt[0].colormask & PIPE_MASK_B) {
-        blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0;
-    }
-    if (state->rt[0].colormask & PIPE_MASK_A) {
-        blend->color_channel_mask |= RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0;
+    /* Color channel masks for all MRTs. */
+    blend->color_channel_mask = state->rt[0].colormask;
+    if (state->independent_blend_enable) {
+        blend->color_channel_mask |= (state->rt[1].colormask << 4);
+        blend->color_channel_mask |= (state->rt[2].colormask << 8);
+        blend->color_channel_mask |= (state->rt[3].colormask << 12);
     }
 
     if (state->dither) {
-- 
cgit v1.2.3


From 1a2f9d722e9492bad1f3d8c7fa99ff19af829b56 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 28 Jan 2010 15:54:57 -0800
Subject: llvmpipe: Remove unnecessary header.

---
 src/gallium/drivers/llvmpipe/lp_setup.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c
index 50b07308c7..92baa980bc 100644
--- a/src/gallium/drivers/llvmpipe/lp_setup.c
+++ b/src/gallium/drivers/llvmpipe/lp_setup.c
@@ -37,7 +37,6 @@
 #include "lp_setup.h"
 #include "lp_state.h"
 #include "draw/draw_context.h"
-#include "draw/draw_private.h"
 #include "draw/draw_vertex.h"
 #include "pipe/p_shader_tokens.h"
 #include "util/u_format.h"
-- 
cgit v1.2.3


From 7280c4c9fead8fac6bfc2fe765b8f434fd2a0ed0 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 28 Jan 2010 15:56:50 -0800
Subject: softpipe: Remove unnecessary header.

---
 src/gallium/drivers/softpipe/sp_setup.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c
index f6c3a2b38b..ba1f0f0b2e 100644
--- a/src/gallium/drivers/softpipe/sp_setup.c
+++ b/src/gallium/drivers/softpipe/sp_setup.c
@@ -38,7 +38,6 @@
 #include "sp_setup.h"
 #include "sp_state.h"
 #include "draw/draw_context.h"
-#include "draw/draw_private.h"
 #include "draw/draw_vertex.h"
 #include "pipe/p_shader_tokens.h"
 #include "util/u_math.h"
-- 
cgit v1.2.3


From 548b9c4ece5acd747b42241dce9db51d9c84dc11 Mon Sep 17 00:00:00 2001
From: Vinson Lee <vlee@vmware.com>
Date: Thu, 28 Jan 2010 17:32:33 -0800
Subject: svga: Remove unnecessary headers.

---
 src/gallium/drivers/svga/svga_pipe_constants.c | 1 -
 src/gallium/drivers/svga/svga_pipe_vertex.c    | 1 -
 2 files changed, 2 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/svga/svga_pipe_constants.c b/src/gallium/drivers/svga/svga_pipe_constants.c
index 93022f3cc5..9022f7c439 100644
--- a/src/gallium/drivers/svga/svga_pipe_constants.c
+++ b/src/gallium/drivers/svga/svga_pipe_constants.c
@@ -26,7 +26,6 @@
 #include "pipe/p_inlines.h"
 #include "pipe/p_defines.h"
 #include "util/u_math.h"
-#include "util/u_memory.h"
 #include "tgsi/tgsi_parse.h"
 
 #include "svga_context.h"
diff --git a/src/gallium/drivers/svga/svga_pipe_vertex.c b/src/gallium/drivers/svga/svga_pipe_vertex.c
index 0bf43fa900..724166a335 100644
--- a/src/gallium/drivers/svga/svga_pipe_vertex.c
+++ b/src/gallium/drivers/svga/svga_pipe_vertex.c
@@ -26,7 +26,6 @@
 #include "pipe/p_inlines.h"
 #include "pipe/p_defines.h"
 #include "util/u_math.h"
-#include "util/u_memory.h"
 #include "tgsi/tgsi_parse.h"
 
 #include "svga_screen.h"
-- 
cgit v1.2.3


From 6c403750d6515f8b2f333d995c87676016e1883d Mon Sep 17 00:00:00 2001
From: Luca Barbieri <luca@luca-barbieri.com>
Date: Thu, 21 Jan 2010 05:36:14 +0100
Subject: tgsi: add caps for fragment coord conventions (v3)

Changes in v3:
- Renumber caps to accomodate caps to add to master in the meantime
- Document caps
- Add unsupported caps to *_screen.c too

Changes in v2:
- Split for properties patch
- Use positive caps instead of negative caps

This adds 4 caps to indicate support of each of the fragment coord
conventions.

All drivers are also modifed to add the appropriate caps.

Some drivers were incorrectly using non-Gallium-default conventions,
and caps for them have them set so that they will behave correctly
after the later state tracker patches.

This drivers are softpipe/llvmpipe (uses integer rather than half
integer) and pre-nv50 Nouveau (uses lower left rather than upper left).

Other drivers might be broken. With this patchset, fixing them is
only a matter of exposing the appropriate caps that match the behavior
of the existing code.

Drivers are encouraged to support all conventions themselves for better
performance, and this feature is added to softpipe in a later patch.
---
 src/gallium/docs/source/screen.rst         | 8 ++++++++
 src/gallium/drivers/cell/ppu/cell_screen.c | 6 ++++++
 src/gallium/drivers/i915/i915_screen.c     | 6 ++++++
 src/gallium/drivers/i965/brw_screen.c      | 6 ++++++
 src/gallium/drivers/llvmpipe/lp_screen.c   | 6 ++++++
 src/gallium/drivers/nv04/nv04_screen.c     | 6 ++++++
 src/gallium/drivers/nv10/nv10_screen.c     | 6 ++++++
 src/gallium/drivers/nv20/nv20_screen.c     | 6 ++++++
 src/gallium/drivers/nv30/nv30_screen.c     | 6 ++++++
 src/gallium/drivers/nv40/nv40_screen.c     | 6 ++++++
 src/gallium/drivers/nv50/nv50_screen.c     | 6 ++++++
 src/gallium/drivers/r300/r300_screen.c     | 6 ++++++
 src/gallium/drivers/softpipe/sp_screen.c   | 6 ++++++
 src/gallium/drivers/svga/svga_screen.c     | 7 +++++++
 src/gallium/include/pipe/p_defines.h       | 4 ++++
 15 files changed, 91 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/docs/source/screen.rst b/src/gallium/docs/source/screen.rst
index f2421c74ae..0507b81627 100644
--- a/src/gallium/docs/source/screen.rst
+++ b/src/gallium/docs/source/screen.rst
@@ -57,6 +57,14 @@ The integer capabilities:
   only permit binding one constant buffer per shader, and the shaders will
   not permit two-dimensional access to constants.
 * ``MAX_CONST_BUFFER_SIZE``: Maximum byte size of a single constant buffer.
+* ``PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT``: Whether the TGSI property
+  FS_COORD_ORIGIN with value UPPER_LEFT is supported
+* ``PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT``: Whether the TGSI property
+  FS_COORD_ORIGIN with value LOWER_LEFT is supported
+* ``PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER``: Whether the TGSI
+  property FS_COORD_PIXEL_CENTER with value HALF_INTEGER is supported
+* ``PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER``: Whether the TGSI
+  property FS_COORD_PIXEL_CENTER with value INTEGER is supported
 
 The floating-point capabilities:
 
diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c
index d185c6b849..37b04396b3 100644
--- a/src/gallium/drivers/cell/ppu/cell_screen.c
+++ b/src/gallium/drivers/cell/ppu/cell_screen.c
@@ -86,6 +86,12 @@ cell_get_param(struct pipe_screen *screen, int param)
       return 0; /* XXX to do */
    case PIPE_CAP_TGSI_CONT_SUPPORTED:
       return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/i915/i915_screen.c b/src/gallium/drivers/i915/i915_screen.c
index d4ee8f5339..7b8d66b8aa 100644
--- a/src/gallium/drivers/i915/i915_screen.c
+++ b/src/gallium/drivers/i915/i915_screen.c
@@ -117,6 +117,12 @@ i915_get_param(struct pipe_screen *screen, int param)
       return 8;  /* max 128x128x128 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
       return 11; /* max 1024x1024 */
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/i965/brw_screen.c b/src/gallium/drivers/i965/brw_screen.c
index a3728fb16e..a8f9892d71 100644
--- a/src/gallium/drivers/i965/brw_screen.c
+++ b/src/gallium/drivers/i965/brw_screen.c
@@ -175,6 +175,12 @@ brw_get_param(struct pipe_screen *screen, int param)
       return 8;  /* max 128x128x128 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
       return 11; /* max 1024x1024 */
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c
index f04c0320ba..902009d90b 100644
--- a/src/gallium/drivers/llvmpipe/lp_screen.c
+++ b/src/gallium/drivers/llvmpipe/lp_screen.c
@@ -114,6 +114,12 @@ llvmpipe_get_param(struct pipe_screen *screen, int param)
       return 0;
    case PIPE_CAP_INDEP_BLEND_FUNC:
       return 0;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c
index 4eba772575..97538964af 100644
--- a/src/gallium/drivers/nv04/nv04_screen.c
+++ b/src/gallium/drivers/nv04/nv04_screen.c
@@ -49,6 +49,12 @@ nv04_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c
index 2dabc4509f..4e6bc12994 100644
--- a/src/gallium/drivers/nv10/nv10_screen.c
+++ b/src/gallium/drivers/nv10/nv10_screen.c
@@ -44,6 +44,12 @@ nv10_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
index 7782cbe3a8..df9401bcce 100644
--- a/src/gallium/drivers/nv20/nv20_screen.c
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -44,6 +44,12 @@ nv20_screen_get_param(struct pipe_screen *screen, int param)
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
index 8f8f3ea1a9..ee83ec5676 100644
--- a/src/gallium/drivers/nv30/nv30_screen.c
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -71,6 +71,12 @@ nv30_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c
index 4eab5daa80..b1a7343b40 100644
--- a/src/gallium/drivers/nv40/nv40_screen.c
+++ b/src/gallium/drivers/nv40/nv40_screen.c
@@ -56,6 +56,12 @@ nv40_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 0;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
index 6af9700c1a..c4465a228c 100644
--- a/src/gallium/drivers/nv50/nv50_screen.c
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -139,6 +139,12 @@ nv50_screen_get_param(struct pipe_screen *pscreen, int param)
 		return 1;
 	case PIPE_CAP_INDEP_BLEND_FUNC:
 		return 0;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+		return 1;
+	case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+	case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+		return 0;
 	default:
 		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
 		return 0;
diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 679a5644af..f311284d15 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -156,6 +156,12 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
             return 1;
         case PIPE_CAP_INDEP_BLEND_FUNC:
             return 0;
+        case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+        case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+	    return 1;
+        case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+        case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+            return 0;
         default:
             debug_printf("r300: Implementation error: Bad param %d\n",
                 param);
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index b2841f4103..410fabf490 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -99,6 +99,12 @@ softpipe_get_param(struct pipe_screen *screen, int param)
       return 1;
    case PIPE_CAP_INDEP_BLEND_FUNC:
       return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 0;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index cd1ed7bac8..c9db9ac874 100644
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -144,6 +144,13 @@ svga_get_paramf(struct pipe_screen *screen, int param)
    case PIPE_CAP_BLEND_EQUATION_SEPARATE: /* req. for GL 1.5 */
       return 1;
 
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
+      return 1;
+   case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 0;
+
    default:
       return 0;
    }
diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h
index b28441dca9..63ba3119b8 100644
--- a/src/gallium/include/pipe/p_defines.h
+++ b/src/gallium/include/pipe/p_defines.h
@@ -409,6 +409,10 @@ enum pipe_transfer_usage {
 #define PIPE_CAP_MAX_CONST_BUFFER_SIZE   33  /*< In bytes */
 #define PIPE_CAP_INDEP_BLEND_ENABLE      34  /*< blend enables and write masks per rendertarget */
 #define PIPE_CAP_INDEP_BLEND_FUNC        35  /*< different blend funcs per rendertarget */
+#define PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT 36
+#define PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT 37
+#define PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER 38
+#define PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER 39
 
 
 /**
-- 
cgit v1.2.3


From b2299d80b4278b8b6553d4e4da4d40d37881d76e Mon Sep 17 00:00:00 2001
From: Luca Barbieri <luca@luca-barbieri.com>
Date: Thu, 21 Jan 2010 05:38:45 +0100
Subject: softpipe: support all TGSI fragment coord conventions (v4)

Changes in v4:
- Rebase and modify for changes in previous patches

Changes in v3:
- Use positive caps instead of negative caps

Changes in v2:
- Now takes the fragment convention directly from the fragment shader

Adds internal support for all fragment coord conventions to softpipe.

This patch is not required for use with the current state trackers, but it
allows softpipe to run any TGSI program and enhances performance.
---
 src/gallium/drivers/softpipe/sp_screen.c   |  5 ++---
 src/gallium/drivers/softpipe/sp_setup.c    | 11 +++++++----
 src/gallium/drivers/softpipe/sp_state.h    |  3 +++
 src/gallium/drivers/softpipe/sp_state_fs.c |  8 ++++++++
 4 files changed, 20 insertions(+), 7 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index 410fabf490..714a1cf534 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -100,11 +100,10 @@ softpipe_get_param(struct pipe_screen *screen, int param)
    case PIPE_CAP_INDEP_BLEND_FUNC:
       return 1;
    case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
-   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
-      return 1;
    case PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT:
    case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER:
-      return 0;
+   case PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER:
+      return 1;
    default:
       return 0;
    }
diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c
index ba1f0f0b2e..bb1bff581c 100644
--- a/src/gallium/drivers/softpipe/sp_setup.c
+++ b/src/gallium/drivers/softpipe/sp_setup.c
@@ -504,21 +504,24 @@ static void tri_persp_coeff( struct setup_context *setup,
 
 /**
  * Special coefficient setup for gl_FragCoord.
- * X and Y are trivial, though Y has to be inverted for OpenGL.
+ * X and Y are trivial, though Y may have to be inverted for OpenGL.
  * Z and W are copied from posCoef which should have already been computed.
  * We could do a bit less work if we'd examine gl_FragCoord's swizzle mask.
  */
 static void
 setup_fragcoord_coeff(struct setup_context *setup, uint slot)
 {
+   struct sp_fragment_shader* spfs = setup->softpipe->fs;
    /*X*/
-   setup->coef[slot].a0[0] = 0;
+   setup->coef[slot].a0[0] = spfs->pixel_center_integer ? 0.0 : 0.5;
    setup->coef[slot].dadx[0] = 1.0;
    setup->coef[slot].dady[0] = 0.0;
    /*Y*/
-   setup->coef[slot].a0[1] = 0.0;
+   setup->coef[slot].a0[1] =
+		   (spfs->origin_lower_left ? setup->softpipe->framebuffer.height : 0)
+		   + (spfs->pixel_center_integer ? 0.0 : 0.5);
    setup->coef[slot].dadx[1] = 0.0;
-   setup->coef[slot].dady[1] = 1.0;
+   setup->coef[slot].dady[1] = spfs->origin_lower_left ? -1.0 : 1.0;
    /*Z*/
    setup->coef[slot].a0[2] = setup->posCoef.a0[2];
    setup->coef[slot].dadx[2] = setup->posCoef.dadx[2];
diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h
index 7f244c4fd4..a83cae7361 100644
--- a/src/gallium/drivers/softpipe/sp_state.h
+++ b/src/gallium/drivers/softpipe/sp_state.h
@@ -68,6 +68,9 @@ struct sp_fragment_shader {
 
    struct tgsi_shader_info info;
 
+   boolean origin_lower_left; /**< fragment shader uses lower left position origin? */
+   boolean pixel_center_integer; /**< fragment shader uses integer pixel center? */
+
    void (*prepare)( const struct sp_fragment_shader *shader,
 		    struct tgsi_exec_machine *machine,
 		    struct tgsi_sampler **samplers);
diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c
index 04bdcaacc2..de3edde976 100644
--- a/src/gallium/drivers/softpipe/sp_state_fs.c
+++ b/src/gallium/drivers/softpipe/sp_state_fs.c
@@ -44,6 +44,7 @@ softpipe_create_fs_state(struct pipe_context *pipe,
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
    struct sp_fragment_shader *state;
+   unsigned i;
 
    /* debug */
    if (softpipe->dump_fs) 
@@ -60,6 +61,13 @@ softpipe_create_fs_state(struct pipe_context *pipe,
    /* get/save the summary info for this shader */
    tgsi_scan_shader(templ->tokens, &state->info);
 
+   for (i = 0; i < state->info.num_properties; ++i) {
+      if (state->info.properties[i].name == TGSI_PROPERTY_FS_COORD_ORIGIN)
+         state->origin_lower_left = state->info.properties[i].data[0];
+      else if (state->info.properties[i].name == TGSI_PROPERTY_FS_COORD_PIXEL_CENTER)
+	 state->pixel_center_integer = state->info.properties[i].data[0];
+   }
+
    return state;
 }
 
-- 
cgit v1.2.3


From 5a64bdaa0150c0d3883953237b7b8a910c98e5b3 Mon Sep 17 00:00:00 2001
From: Mathias Fröhlich <Mathias.Froehlich@gmx.net>
Date: Sat, 30 Jan 2010 11:38:21 -0800
Subject: r300g: Add alpha render targets.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Signed-off-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Tested-by: Mathias Fröhlich <Mathias.Froehlich@gmx.net>
---
 src/gallium/drivers/r300/r300_screen.c        |  4 ++--
 src/gallium/drivers/r300/r300_state_inlines.h | 13 +++++++++++++
 2 files changed, 15 insertions(+), 2 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index f311284d15..7df3ad02c8 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -217,6 +217,8 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
         case PIPE_FORMAT_A4R4G4B4_UNORM:
         case PIPE_FORMAT_R5G6B5_UNORM:
         case PIPE_FORMAT_A1R5G5B5_UNORM:
+        case PIPE_FORMAT_A8_UNORM:
+        case PIPE_FORMAT_L8_UNORM:
             retval = usage &
                 (PIPE_TEXTURE_USAGE_RENDER_TARGET |
                  PIPE_TEXTURE_USAGE_DISPLAY_TARGET |
@@ -231,8 +233,6 @@ static boolean r300_is_format_supported(struct pipe_screen* screen,
         case PIPE_FORMAT_DXT3_RGBA:
         case PIPE_FORMAT_DXT5_RGBA:
         case PIPE_FORMAT_YCBCR:
-        case PIPE_FORMAT_L8_UNORM:
-        case PIPE_FORMAT_A8_UNORM:
         case PIPE_FORMAT_A8L8_UNORM:
             retval = usage & PIPE_TEXTURE_USAGE_SAMPLER;
             break;
diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index e2180b33b7..faa049492f 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -331,7 +331,10 @@ static INLINE uint32_t r300_translate_colorformat(enum pipe_format format)
 {
     switch (format) {
         /* 8-bit buffers */
+        case PIPE_FORMAT_A8_UNORM:
         case PIPE_FORMAT_I8_UNORM:
+        case PIPE_FORMAT_L8_UNORM:
+        /* case PIPE_FORMAT_S8_UNORM: ??? */
             return R300_COLOR_FORMAT_I8;
         /* 16-bit buffers */
         case PIPE_FORMAT_R5G6B5_UNORM:
@@ -408,6 +411,16 @@ static INLINE uint32_t r300_translate_out_fmt(enum pipe_format format)
             return R300_US_OUT_FMT_C4_8 |
                 R300_C0_SEL_A | R300_C1_SEL_B |
                 R300_C2_SEL_G | R300_C3_SEL_R;
+
+        /* 8-bit outputs */
+        case PIPE_FORMAT_A8_UNORM:
+            return R300_US_OUT_FMT_C4_8 |
+                R300_C0_SEL_A;
+        case PIPE_FORMAT_I8_UNORM:
+        case PIPE_FORMAT_L8_UNORM:
+            return R300_US_OUT_FMT_C4_8 |
+                R300_C0_SEL_R;
+ /* R300_OUT_SIGN(x) */
         default:
             debug_printf("r300: Implementation error: "
                 "Got unsupported output format %s in %s\n",
-- 
cgit v1.2.3


From bfbec0818f0e7ab9358809a1af732a1366245507 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 30 Jan 2010 12:03:20 -0800
Subject: r300g: Dual-source blending isn't possible, period.

---
 src/gallium/drivers/r300/r300_screen.c        |  1 -
 src/gallium/drivers/r300/r300_state_inlines.h | 14 +++++++++-----
 2 files changed, 9 insertions(+), 6 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index 7df3ad02c8..ae3155c0ed 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -114,7 +114,6 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
              */
             return 1;
         case PIPE_CAP_DUAL_SOURCE_BLEND:
-            /* XXX need to figure out how to do it */
             return 0;
         case PIPE_CAP_ANISOTROPIC_FILTER:
             return 1;
diff --git a/src/gallium/drivers/r300/r300_state_inlines.h b/src/gallium/drivers/r300/r300_state_inlines.h
index faa049492f..5df6815221 100644
--- a/src/gallium/drivers/r300/r300_state_inlines.h
+++ b/src/gallium/drivers/r300/r300_state_inlines.h
@@ -81,9 +81,6 @@ static INLINE uint32_t r300_translate_blend_factor(int blend_fact)
             return R300_BLEND_GL_CONST_COLOR;
         case PIPE_BLENDFACTOR_CONST_ALPHA:
             return R300_BLEND_GL_CONST_ALPHA;
-        /* XXX WTF are these?
-        case PIPE_BLENDFACTOR_SRC1_COLOR:
-        case PIPE_BLENDFACTOR_SRC1_ALPHA: */
         case PIPE_BLENDFACTOR_ZERO:
             return R300_BLEND_GL_ZERO;
         case PIPE_BLENDFACTOR_INV_SRC_COLOR:
@@ -98,9 +95,16 @@ static INLINE uint32_t r300_translate_blend_factor(int blend_fact)
             return R300_BLEND_GL_ONE_MINUS_CONST_COLOR;
         case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
             return R300_BLEND_GL_ONE_MINUS_CONST_ALPHA;
-        /* XXX see above
+
+        case PIPE_BLENDFACTOR_SRC1_COLOR:
+        case PIPE_BLENDFACTOR_SRC1_ALPHA:
         case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
-        case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: */
+        case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
+            debug_printf("r300: Implementation error: "
+                "Bad blend factor %d not supported!\n", blend_fact);
+            assert(0);
+            break;
+
         default:
             debug_printf("r300: Unknown blend factor %d\n", blend_fact);
             assert(0);
-- 
cgit v1.2.3


From f7467f7437af57c1b1c9686c05ea5f1be25176f1 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 30 Jan 2010 12:23:24 -0800
Subject: r300g: Fix comment on SHADOW_AMBIENT.

---
 src/gallium/drivers/r300/r300_emit.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 9e813569e3..a429457625 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -170,9 +170,8 @@ static const float * get_shader_constant(
                     vec[1] = 1.0 / tex->height0;
                     break;
 
-                /* Texture compare-fail value. */
-                /* XXX Since Gallium doesn't support GL_ARB_shadow_ambient,
-                 * this is always (0,0,0,0), right? */
+                /* Texture compare-fail value. Shouldn't ever show up, but if
+                 * it does, we'll be ready. */
                 case RC_STATE_SHADOW_AMBIENT:
                     vec[3] = 0;
                     break;
-- 
cgit v1.2.3


From 627a45848faf9e6d19d386b9eb30d36bf18d0998 Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 30 Jan 2010 12:23:43 -0800
Subject: r300g: Don't handle SHADOW_AMBIENT.

Bisected so that only this commit has to be reverted if I'm wrong.

...Pfft, like I'm ever wrong. :3
---
 src/gallium/drivers/r300/r300_emit.c | 6 ------
 1 file changed, 6 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index a429457625..04001ee129 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -170,12 +170,6 @@ static const float * get_shader_constant(
                     vec[1] = 1.0 / tex->height0;
                     break;
 
-                /* Texture compare-fail value. Shouldn't ever show up, but if
-                 * it does, we'll be ready. */
-                case RC_STATE_SHADOW_AMBIENT:
-                    vec[3] = 0;
-                    break;
-
                 case RC_STATE_R300_VIEWPORT_SCALE:
                     if (r300->tcl_bypass) {
                         vec[0] = 1;
-- 
cgit v1.2.3


From b6ac2403aca240c72c8622ddbc3a4e123d5aa37f Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 30 Jan 2010 12:33:44 -0800
Subject: r300g: Independent blend enables only work on r500.

---
 src/gallium/drivers/r300/r300_screen.c |  6 +++++-
 src/gallium/drivers/r300/r300_state.c  | 15 +++++++++++----
 2 files changed, 16 insertions(+), 5 deletions(-)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c
index ae3155c0ed..fa8ed581da 100644
--- a/src/gallium/drivers/r300/r300_screen.c
+++ b/src/gallium/drivers/r300/r300_screen.c
@@ -152,7 +152,11 @@ static int r300_get_param(struct pipe_screen* pscreen, int param)
                 return 0;
             }
         case PIPE_CAP_INDEP_BLEND_ENABLE:
-            return 1;
+            if (r300screen->caps->is_r500) {
+                return 1;
+            } else {
+                return 0;
+            }
         case PIPE_CAP_INDEP_BLEND_FUNC:
             return 0;
         case PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT:
diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c
index 8295799553..a5c0869066 100644
--- a/src/gallium/drivers/r300/r300_state.c
+++ b/src/gallium/drivers/r300/r300_state.c
@@ -161,6 +161,7 @@ static boolean blend_discard_if_src_alpha_color_1(unsigned srcRGB, unsigned srcA
 static void* r300_create_blend_state(struct pipe_context* pipe,
                                      const struct pipe_blend_state* state)
 {
+    struct r300_screen* r300screen = r300_screen(pipe->screen);
     struct r300_blend_state* blend = CALLOC_STRUCT(r300_blend_state);
 
     if (state->rt[0].blend_enable)
@@ -290,10 +291,16 @@ static void* r300_create_blend_state(struct pipe_context* pipe,
 
     /* Color channel masks for all MRTs. */
     blend->color_channel_mask = state->rt[0].colormask;
-    if (state->independent_blend_enable) {
-        blend->color_channel_mask |= (state->rt[1].colormask << 4);
-        blend->color_channel_mask |= (state->rt[2].colormask << 8);
-        blend->color_channel_mask |= (state->rt[3].colormask << 12);
+    if (r300screen->caps->is_r500 && state->independent_blend_enable) {
+        if (state->rt[1].blend_enable) {
+            blend->color_channel_mask |= (state->rt[1].colormask << 4);
+        }
+        if (state->rt[2].blend_enable) {
+            blend->color_channel_mask |= (state->rt[2].colormask << 8);
+        }
+        if (state->rt[3].blend_enable) {
+            blend->color_channel_mask |= (state->rt[3].colormask << 12);
+        }
     }
 
     if (state->dither) {
-- 
cgit v1.2.3


From a794c660d30b04ed50482cf2f26e0a7f9687319e Mon Sep 17 00:00:00 2001
From: Corbin Simpson <MostAwesomeDude@gmail.com>
Date: Sat, 30 Jan 2010 14:09:04 -0800
Subject: Revert "r300g: Don't handle SHADOW_AMBIENT."

This reverts commit 627a45848faf9e6d19d386b9eb30d36bf18d0998.

I was wrong. Who could have known? :3
---
 src/gallium/drivers/r300/r300_emit.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'src/gallium/drivers')

diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c
index 04001ee129..a429457625 100644
--- a/src/gallium/drivers/r300/r300_emit.c
+++ b/src/gallium/drivers/r300/r300_emit.c
@@ -170,6 +170,12 @@ static const float * get_shader_constant(
                     vec[1] = 1.0 / tex->height0;
                     break;
 
+                /* Texture compare-fail value. Shouldn't ever show up, but if
+                 * it does, we'll be ready. */
+                case RC_STATE_SHADOW_AMBIENT:
+                    vec[3] = 0;
+                    break;
+
                 case RC_STATE_R300_VIEWPORT_SCALE:
                     if (r300->tcl_bypass) {
                         vec[0] = 1;
-- 
cgit v1.2.3